Thursday, June 27, 9:00 AM - 10:30 AM
TA01
Pyrmont Theatre
Special Session MASS: The Past, Present and Future of Marketing Analytics Sponsored by Credamo
Invited Session
Special Sessions
Chair: Jennifer Killian, XPON
1 - Keynote - Pradeep Chintagunta
Pradeep Chintagunta, University of Chicago, Chicago, Illinois, United States
Keynote
2 - Panelist - Kusum Ailawadi
Kusum Ailawadi, Tuck School of Business at Dartmouth College, Hanover, New Hampshire, United States
Panelist
3 - Panelist - Pradeep Chintagunta
Pradeep Chintagunta, University of Chicago, Chicago, Illinois, United States
Panelist
4 - Panelist - Jadanne Dare
Jadanne Dare, KPMG, Sydney, Australia
Panelist
TA02
Meeting Room C2.1
Pricing: Economics-based Approaches
Contributed Session
Chair: Madhu Viswanathan, Indian School of Business, Telangana, India
1 - Advertising and Leader-follower Pricing between an Off-patent Drug and its Generic Counterpart - A Dynamic Model Estimated
Abhik Roy, Quinnipiac University, Hamden, CT, United States

A dynamic model of price competition between an off-patent drug and its generic version is developed. Following the static model of Amaldoss and He (2009), the off-patent brand acts as a price leader while the generic drug is its price follower. Furthermore, the branded drug uses direct-to-consumer advertising, while the generic drug does not. The advertising can be either brand specific or beneficial to the entire drug category. A system of price and sales is estimated using data from three drug categories – hypothyroid, hypertension, and hyperlipidemia. Constraints are imposed on the parameters of the system of equations, based on a solution to the dynamic price and advertising game. Off label usage of the branded drug is controlled for and so is the existence of a newer generation of drugs for the therapeutic condition. It is found that the existence of a Stackelberg system of price competition coincides with significant effects of advertising on category demand. The implication is that an off-patent brand is willing to act as a price leader if its advertising is effective in stimulating demand for the drug category. Another way to explain the strategy is that direct-to-consumer advertising allows a branded drug to be a price leader. While recent empirical studies have found evidence that leader-follower pricing between competing brands of prescription drugs coincides with advertising effectiveness, this is the first example of a similar strategic interaction occurring between an off-patent drug and its generic counterpart.

2 - Optimal Bundling Design: The Case of Gym Training Sessions
Qinxin Chen, Washington University in St. Louis, St. Louis, MO, United States, Yijun Chen, Tat Chan

We study the optimal bundle design, involving the number of bundles offered to consumers and the size and price of each bundle, that can maximize the profit of service provider firms, as well as the impact of the bundle design on consumer usage level and welfare. As a second-degree price discrimination, one critical distinguishing feature in the service-bundle setting is purchase-first-and-consume-later, which can cause over- or under-usage for bundle users. Utilizing data from a premium gym, we develop a structural model of consumer bundle choices and consumption decisions. Our model allows for a rich heterogeneity in consumers' needs and preferences, price sensitivities and demand uncertainties, using a Gaussian Mixture Model. Based on the estimation results, we run a series of counterfactuals to explore the optimal bundle design of the gym, where we leverage the innovative power of neural networks for computational efficiency when there is a huge computational burden due to rich consumer heterogeneity and a large number of state variables. Our research extends the previous theoretical works on bundling, providing empirical insights and practical implications for service firms seeking to capitalize on the full potential of their bundle offerings.

Keywords: Second-Degree Price Discrimination, Optimal Bundles, Consumer Behavior, Structural Modeling, Neural Networks, Gaussian Mixture Distribution, Consumption Uncertainty

3 - Renting Regret
Madhu Viswanathan, Indian School of Business, Hyderabad, India, Manish Gangwar, Pranav Jindal, Abhinav Uppal
Three-part tariffs are extremely common for a number of products and services such as cell phone plans, car rentals, online newspaper subscription etc. The profitability of these plans relative to other plans such as flat-rate pricing, two-part tariffs etc. depends not only on consumer preferences, but also on the behavioral biases they suffer from. In this paper, we study the role of regret aversion in driving not only consumer plan choice, but also how much they drive a rental car. As compared to cell phone plans and newspaper subscription where the marginal cost of usage is close to zero, any excess driving due to regret aversion is costly for the firm if it does not generate additional revenue, and thereby could impact the relative profitability of three-part tariffs. Using data from a large car rental company in India where the firm randomly varied the tariff choices, we document evidence of regret aversion. We show that the marginal effect of regret aversion is comparable in magnitude to consumer’s price sensitivity. Based on this, we estimate a structural model of consumer demand for three-part tariffs in this market and quantify the importance of accounting for regret aversion. This paper contributes to the existing literature on the role of behavioral biases in three-part tariffs and pricing which shows that consumer behavior is often plagued by flat-rate bias and pay-per-use bias (Lambrecht and Skiera 2006), and how much they value free minutes (Ascarza et al 2012).
4 - Postponement Sludge: How Much Does It Cost to Cancel at a Later Date?
Matthew Fisher, Southern Methodist University, Dallas, TX, United States

In their seminal work about ‘nudging’ consumers toward particular choices, Thaler and Sunstein used the term ‘sludge’ to capture frictions that dissuade consumers from making beneficial choices. When the consumer’s choice involves postponing an action until later, there is an interesting complication of sludge. Examples of these postponed actions include unsubscribing from a service or avoiding autorenewal, stopping automatic payments, and changing privacy settings or permissions in the future. Such actions are generally postponed so that consumers can extract the benefits of a trial, discount, or other favorable limited-time offer. Yet consumers are subjected to temporal biases and biases in risk assessment which may lead them less beneficial decisions.

In this paper, we identify the costs of sludge, both economic and psychological, incorporating them into a probabilistic framework in order to capture the risk of taking the postponed action. Then, we conduct a series of experiments to assess the framework, including measuring the different costs and the probabilities of incurring them. Finally, we compare the measured costs with the expected costs before choosing to postpone.
TA03
Meeting Room C2.2
Salesforce Management and Personal Selling: Theory and Models
Contributed Session
Chair: Xiangyin Kong, University of Science and Technology of China, Hefei
1 - Income Tax and Salesforce Performance: A Micro Perspective
Yi Xiang, China Europe International Business School, Shanghai, China, People's Republic of, Yuxin Chen, Fang Yu

How does the change in income tax affect sales performance? Our paper explores the link between economic policy and salesforce management at the transactional level, using data from a large fashion retailer in China. The analysis shows that the implementation of a nationwide personal income tax cut in October 2018 improves sales performance more among those salespersons who benefited from the policy, compared to others. The performance gain is observed after the tax cut and persisted in future months, and is particularly significant in low-income regions. We also find that the performance gain is largely due to an increase in sales of pricier items, rather than more reliance on discounts to boost demand. Finally, the research shows that the net effect of the tax cut in the context of the study is an increase in the government's revenue due to the firm's higher sales and the corresponding increase in corporate tax paid.

2 - Exploring the Impact of Psychological Distance on the Ethical Behavior of Sales Managers
Tobias Heußler, Wiesbaden Business School, Wiesbaden, Germany, Lea Schwehn

Management attention to counteract unethical sales force behavior is growing as the business opportunities for sustainable strategies, products, and services become more promising. A growing body of ethics literature applies construal level theory (CLT) to explain ethical behavior and moral judgment. Based on the rationale of the psychological distance the theory provides an adequate principle for how context shapes individual decision-making processes. Although, the intersection of sales and ethics research is well developed, few management scholars have built on CLT.

This research adds to the existing literature by transferring CLT into a sales context and an associated experimental design. We conducted two studies with a total of 1,340 salespeople. ANOVA results indicate that psychological distance is a predictor of unethical behavior and moderates the relationship between a salesperson's financial benefits and the manager's ethical judgments of the salesperson's behavior. We manipulated psychological distance for two dimensions: (1) the individual perceived distance of each sales manager to the decision facts and (2) the distance of the individual position to the ethical conditions imposed by the company on its employees. In both cases, distal target stimuli are judged more abstractly. Thus, managers' ethical judgment of a salesperson's personal enrichment will be judged more positively as abstract mental construals excuse unethical behavior. By influencing perceived psychological distance, we derive managerial implications even though unethical behavior is a phenomenon that managers cannot easily control. Precise policies provide guidance, avoid the stress of role ambiguity, and result in low levels of construal.

3 - Under What Circumstances Does a Convex Sales Compensation Plan Make Sense?
James Lattin, Stanford University, Stanford, CA, United States

Early theoretical research in quantitative marketing (e.g., Basu et al 1985) drew on agency theory to suggest that under certain circumstances, the optimal sales compensation function might be written as a convex function of sales output. Piecewise-linear functional forms (e.g., flat salary with accelerating commission rates) were justified because they provided an incentive for high-performing sales reps to continue to invest in sales effort. In Silicon Valley in particular, such plans have been de rigueur for decades.

More recently, the pendulum seems to have swung back. In a complex sales environment, Holmstrom and Milgrom (1987) argued for a more robust linear compensation plan. In his study of enterprise sales reps with leveraged plans, Larkin(2014) found evidence of widespread gaming (i.e., pulling or pushing the timing of a deal in ways not favorable to the company). Misra and Nair (2011) found that replacing a relatively complex, output-based compensation plan (involving quotas, commissions, caps) with a simple linear plan based on salary and commission led to an increase in profitability. The implication is that highly-leveraged plans perhaps may be more trouble than they are worth.

Zoltners et al (2012) have argued that in sales environments that are more complex, relying heavily on compensation incentives to motivate desired sales behaviors has become too challenging. In this paper, we discuss when it is possible to impose management controls to achieve the same results as nonlinear compensation plans, and examine the circumstances under which a convex compensation function might still make sense.

4 - Optimal Quota-Bonus Contract with Bayesian Learning: Early Success Bias and Late Success Encouragement
Xiangyin Kong, University of Science and Technology of China, Hefei, China, People's Republic of, Lina Bao, Yimin Yu
Incomplete information is common for new products, or new sales agents. We investigate the optimal quota-bonus contracts with Bayesian learning, where both the firm and the sales agent are uncertain about the market condition/agent ability. The firm offers a long-term contract to incentivize the sales agent to always exert high effort. The agent is subject to the limited liability. To capture learning, we build a simple repeated moral hazard model with two periods. We show that the optimal contract in each period is a quota-bonus contract and explicitly characterize the impacts of learning on the optimal bonuses. We show that a low outcome leads to the underestimate of the market condition/agent ability, and the sales agent has incentive to shirk privately to manipulate the firm’s belief. We quantify the impact of the belief manipulation through the belief manipulation rent. As a result, the optimal bonus exhibits an early success bias and a late success encouragement. We further investigate the role of probation contracts and strategic disincentives in mitigating the belief manipulation rent. Our results provide new explanations for the popularity of the early success bias and late success encouragement and the probation contracts in practice
TA04
Meeting Room C2.3
Measuring and Modeling Emotions
Contributed Session
Chair: Mohammad Hosein Tavakoli, University of Warwick, Coventry, United Kingdom
1 - Consumers' Motivations and Preferences in Entertainment Media Consumption: Love vs. Hate and Suspense vs. Surprise
Ian Weaver, National University of Singapore, SIngapore, Singapore, Jiwoong Shin, Vanya Klenovskiy
This paper aims to gain insight into the choices made by consumers regarding their media and entertainment product consumption. We propose that individuals are driven to view and derive enjoyment from different motives – supporting their own team's success (“love”) and their rival team's failure (“hate”). We explore these concepts of "love" and "hate" as underlying motivations and their interplay with specific content characteristics such as suspense and surprise. We collected YouTube viewing data at the state level, focusing on full game highlights from an American Major Sports League between 2015 and 2022. Our analysis is conducted on two levels: aggregate viewership and individual video watching. At the aggregate level, we examine game characteristics such as scores, score differences, team standings, winning percentages, and the betting odds for each team from major gambling sites. On the individual video watching level, we analyze each play within the video content and calculate surprise and suspense scores for each event. Videos comprise multiple events that shape viewers' beliefs about which team will emerge victorious, generating measures of "suspense" and "surprise" experienced by viewers during entertainment consumption. We identify the causal effects of different motives (love and hate) and their interactions with suspense and surprise measures based on viewers' engagement, particularly in terms of video-watching durations. We find that consumers derive enjoyment from supporting their own team's success and their rival team's failure. Furthermore, the effects of surprise and suspense differ depending on whether viewers are motivated by "love" or "hate" during game-watching behavior.
2 - Beyond Positivity: Investigating the Contagiousness of Mixed Emotions on Facebook Brand Pages
Chedia Dhaoui, UNSW Business School, Sydney, Australia, Cynthia Webster

Emotionality plays a key role in consumers’ purchase decision. Although researchers tend to believe that posting positive emotional content conveying happiness generates positive outcomes, we demonstrate that happy emotions displayed in brand posts can backfire and elicit negative emotions when consumers interpret them as inauthentic. Combining happiness with negative emotions like sadness and anger can be beneficial as it amplifies positive emotional reactions among consumers on Facebook brand pages. Building on emotional contagion theory and social information theory, this research examines the contagious effect of mixed emotions of opposite valence. We investigate how conflictual emotions (e.g. happy and sad; happy and angry) in brand posts is more beneficial than single positive emotions in amplifying positive emotional reactions. We collected unstructured social media data using Facebook Graph API from 942 Facebook brand pages among the most talked about brands on Facebook, spanning 16 industry categories. We analysed 83,310,772 consumers' emotional reactions to 317,357 brand posts (including 141,665 images, 102,588 videos, and 73,104 text-only brand posts and captions associated with image/video brand posts) using Computer Vision AI and Natural Language Processing. The findings suggest brand posts conveying happiness generate more happy consumer emotional reactions than brand posts with no emotions. Brand posts conveying happiness have a significant contagion effect for all emotions studied; however, happiness also triggers sad and angry emotional reactions. The results provide evidence that happiness in brand posts becomes more contagious, elicits more happiness reactions, when combined with negative emotions such as sadness and anger.

3 - Political Parties' Stance on Different Topics Impacting Parties' Anticipated Future Success
Mohammad Hosein Tavakoli, University of Warwick, Coventry, United Kingdom, Thomas Gruca, Vahid Karimi Motahhar

Political parties share platforms, positions, and plans with potential voters to gain their support on election day. While polls ask potential voters about the issues most important to them, exit polls try to determine what topics impacted supporters on election day. Therefore, the party's political positions reflect the self-interests of its voters. In this study, we consider alternative measures of these self-interests from political prediction markets, wherein traders aim to make money by buying and selling contracts associated with the success or failure of political parties.

The setting for our study is in the context of the 2015 UK general election, particularly a set of 6 prediction markets designed to predict the number of seats won by members of the Conservative, Green, Liberal Democratic, UKIP, SNP, and Labour Parties. The six markets around the 2015 UK General Election yield insights, especially since the outcome – the expansion of the Conservative majority – was unexpected in both the polls and prediction markets.

Our data comes from MediaPredict, a commercial prediction market company. A unique aspect of these prediction markets is that each trader is asked to provide a written justification for each trade.

We apply text analysis techniques, i.e., BERT and topic modeling, to investigate the reasoning traders use (and emotions they express) when betting on a given party's future success (or failure). To categorize the diversity of political topics driving trader behavior, we examine party-specific and general topics that influence trader predictions of the outcomes across all the markets.

TA05
Meeting Room C2.4
Marketing Strategy: Advertising
Contributed Session
Chair: Jacqueline Chang, 50 Stamford Road, 178899
1 - Firm Life Cycle, Managerial Ability and Accounting Conservatism: Implications for Advertising Spend and Shareholder Returns
Steven D'Alessandro, Edith Cowan University, Perth, Australia, Domenic Gasbarro, Hong Miao, Kenton Zumwalt
Managing advertising spend is a crucial responsibility for firms, but the factors influenc-ing the advertising budget vary across industries. This study examines how firms’ life cycles, managerial ability, and accounting conservatism impact advertising spend and shareholder returns. To address the industry effect, we introduce the median-adjusted advertising spend ratio (MAASR), by subtracting the industry’s median advertising spend ratio from the firm’s ratio. We show the impact of the mean adjusted advertising spend ratio on firm performance depends on the firm’s life cycle, managerial ability, and accounting conservatism. Fortunately, firms can adjust their advertising spend based on their managerial ability and accounting conservatism, enabling more effective budget management. Our findings reveal that during the growth and maturity phases of the life cycle the spending is focused on the lower deciles of MAASR. Interestingly, the spending is focussed on the lower deciles of MAASR. The paper contributes to the determinants and efficacy of advertising spend and provides managers and researchers with important guidelines for when to increase or manage advertising budgets. Share-holder wealth from advertising expenditure can be increased by judiciously considering the firm life cycle, managerial ability, and accounting conservatism.
2 - Trade Secret Protection and its Impact on Advertising Spending
Jacqueline Chang, Singapore Management University, Singapore, Singapore, Sungkyun Moon, Kapil Tuli

Exploiting the staggered state-wide adoption of trade secret protection, this study examines how the plausibly exogenous changes in trade secret protection are likely to have a significant impact on the advertising spending of firms headquartered in these states. The adoption of legal doctrines offers greater protection for a firm’s trade secrets because it restricts the ability of a firm’s current employees to work for its competitors. However, at the same time, it can also restrict the flow of information across firms and therefore exert a negative impact on innovation in general. Building on recent work that outlines the employee, CEO, and board-level impact of these legal doctrines, we argue that firms exposed to these exogenous shocks are likely to increase their advertising spending due to the rising investor and analyst pressure to boost sales. Utilizing a staggered difference-in-differences model, we find that treated firms are significantly more likely to increase their advertising spending as compared to firms in the control group. Importantly, the observed effects are robust across specification choices and the use of alternative estimation methods. In addition, we find significant heterogeneity in the firm response to this exogenous shock in their legislative environment.

3 - CANCELLED Advertising and the IPO Price Revision Process
Yi-Lin Tsai, University of Melbourne, Carlton, Victoria, Australia, Michal Herzenstein, Dan Horsky, Tzachi Zach

The Initial Public Offering (IPO) is a crucial event for firms — it is their opportunity to raise substantial funds to expand their operations. Therefore, of utmost importance for managers are the IPO proceeds which are determined by the offer price and number of shares sold. That price is set through negotiations among the issuing firm, its underwriters, and institutional investors. The resulting offer price reflects the issuer’s valuation and expectation of the share price on the first day of trading (the aftermarket). In this paper we examine the effect of the intensity of product advertising just before the IPO on the change in the offer price between the initial one stated in the prospectus and the final one. We analyze data on 1,115 IPOs where some of the issuers increased their advertising spending and some did not, and mitigate the endogeneity problem with a non-parametric inverse-probability weighting method. We find that increasing the amount spent on product advertising in the months leading to the IPO is associated with an increase in the offer price and hence the proceeds. Using Trade and Quote data we show that the increased advertising spending is associated with more retail investors’ participation on the first day of trading, which contributes to the success of the IPO. We further find that non-consumer facing issuers benefit more from increased advertising likely because consumers are less aware of them. Lastly, we discuss how market structure curbs issuers from abusing the relation between advertising and the price revision.

TA06
Meeting Room C2.5
Analytical Models for Digital Platform Interactions
Contributed Session
Chair: Di Yuan, University of Illinois Urbana-Champaign, 1206 S Sixth St., Champaign, IL, 61820, United States
1 - Quality Signaling in Rewards-Based Crowdfunding: The Role of Deferred Payments
Xiaobei Shen, University of Science and Technology of China, Hefei, China, People's Republic of, Jiancheng Lyu, Xiaoting Xie

Crowdfunding has emerged as a popular fundraising method in recent years. However, information asymmetry brings a great challenge to backers as they have no assurance of receiving the product/service they are supporting. To mitigate this problem, crowdfunding platforms have introduced deferred payment mechanisms, in which a small fraction of the funds raised by a campaign is held back as the insurance for backers. We explore how deferred payments affect crowdfunding in the presence of information asymmetry. We develop a stylized game-theoretic signaling model that captures the strategic behavior of creators in leveraging deferred payments to signal high-quality projects. We show that, when all creators are required to accept deferred payments, high-quality creators shall signal quality through a low reward price but keep the funding target at the complete information funding level, which is in sharp contrast to the existing literature. We also reveal that exempting low-quality creators from deferred payments is in fact more beneficial to both the platforms and the high-quality creators. To implement deferred payments, the reserve ratio shall be determined with caution. In particular, we find that deferring payments for only high-quality creators can align the benefits of the platforms and the high-quality creators, i.e., there exists a unique optimal reserve ratio that maximizes the high-quality creators' profit and the platform's commission simultaneously. However, if low-quality creators have to follow the rule of deferred payments, then the optimal reserve ratio of high-quality creators may be different from that of the platforms.

2 - Creative Marketplaces for Humans and Machines
Lijuan Luo, Renmin University of China, Beijing, China, People's Republic of, Emaad Manzoor, Nathan Yang
Generative artificial intelligence (AI) offers a new source of content creation in creative marketplaces (e.g., digital art, images, music and video). While generative AI relies on original human-created content for training, its outputs can compete with those of human creators. This competition raises important considerations for platforms in effectively utilizing AI while also providing incentives to human creators to share their work. In this research, we present an analytical model for platform design that serves content produced by a combination of human creators and generative AI. We aim to answer the following questions: (1) How should platforms price content, distinguishing between original human-created and AI-generated synthetic content? (2) How can platforms devise revenue-sharing mechanisms that effectively motivate human creators to share their content? (3) What are the consumer welfare implications of these platform design choices? Our analysis yields a few substantive insights. First, we illustrate the potential lost profits associated with pricing that ignores the alignment costs incurred by consumers when prompting AI for more desired content. Second, we demonstrate welfare improvements from AI-generated content for customers that face imperfect matches with original content. Finally, we describe conditions that might lead to a collapse of the marketplace. Our results provide valuable managerial implications for online platforms, particularly for those targeting the generative AI for content creation purposes.
3 - Generative AI Adoption by Creator Platforms
Di Yuan, University of Illinois Urbana-Champaign, Champaign, IL, United States, Manmohan Aseri, Vibhanshu Abhishek, Kartik Hosanagar

The recent advancements in generative AI, e.g., ChatGPT, have raised several questions about the implications of such technology. As the name suggests, generative AI (GenAI) generates content. Therefore, an important question is how this technology will affect user-generated content (UGC) platforms such as Spotify, YouTube, etc. A UGC platform can adopt GenAI to help its existing content creators. While GenAI is expected to improve content creation productivity, it could impact content creators differently. In this paper, we analyze these strategic interactions between content creators and the platform using a game-theoretic model and show that the provision of GenAI may lead to an exodus of high-quality content creators. Interestingly, due to this exodus of high-quality creators, the adoption of GenAI can lead to overall lower content quality and lower profit for the platform.

We also analyze methods of mitigating this problem. In our analysis, we find that charging a price for using GenAI (instead of providing it for free) not only leads to an additional source of revenue for the platform but also prevents the exodus of high-quality content creators. This is because the price charged for using GenAI reduces its usage by low-quality creators. This mitigates their competition with high-quality creators and keeps high-quality creators on the platform. Thus, a priced GenAI is more likely to be adopted by a UGC platform than a free GenAI.

TA07
Meeting Room C2.6
Charitable Giving / Donations: Empirical Analysis
Contributed Session
Chair: Daniel Winkler, University of Economics and Business-Vienna, Welthandelsplatz 1, Vienna, 1020, Austria
1 - Words Matter: When and Why Direct Speech Appeal Promotes Charity Crowdfunding Project Success
Ziqi Zhang, Renmin University of China, Beijing, China, People's Republic of

Direct speech is a prevalent rhetorical tactic in advertising, which helps to construct persuasive appeals. Nevertheless, limited research has examined the impact of direct speech on persuasion and intended actions. In the current study, we empirically investigate the influence of direct speech appeals on the performance of online charity crowdfunding initiatives. Prior research suggests that direct speech promotes viewers to shift to the speaker’s perspective, thereby increasing their empathy level with the speaker. Since prosocial research has highlighted the pivotal role of empathy in helping behavior, we propose that direct speech appeals can increase the performance of donation projects by promoting potential donors’ perspective-taking and empathy. Leveraging a dataset with over 50,000 online charity crowdfunding projects, we examine the effect of direct speech appeals on the success of donation projects. Our findings reveal a significant positive relationship between the use of direct speech appeals and project success, including generating more donations and attracting more donors. Furthermore, our experimental results support the mediating roles of perspective-taking and empathy in this relationship. This research makes significant contributions to both advertising and charity fundraising literature by systematically examining the impact of direct speech appeals on donation project success. Additionally, our findings offer valuable guidance to practitioners seeking to enhance the efficacy of online charity appeals by incorporating direct speech contents.

2 - Be More Generous on Computers or Smartphones? A Device Effect on Donations
Shengru Ren, Renmin University of China, Beijing, China, People's Republic of, Jun Pang

In recent years, a growing number of people use mobile devices (e.g., smartphone and tablets), instead of traditional computers, to make online donations. This shift raises an intriguing question about how devices influence donation behavior, which has been largely underexplored in extant literature. In this research, we collected data from six donation projects hosted by China Charity Federation, one of the leading non-profit charity platforms in China, between June 2015 and July 2023. The data consisted of 363149 donation records, each including the project name, donation time, donor ID and comments, donation amount, payment method, and access platform (smartphone APP or computer-based official website). Using Ordinary Least Squares regression analyses, we found that for the same donation project, computer users donated more than smartphone users. Drawing on the construal level theory, we proposed that the observed device effect on donations is driven by the different construal levels consumers employ when making donations on computers versus smartphones. Specifically, computers (vs. smartphones) evoke a more abstract construal level, directing people to focus more on a project’s desirability rather than its feasibility. The attentional focus on desirability then increases donations. We conducted two lab experiments to examine these propositions and found supporting evidence. Our research advances knowledge of how devices shape people’s donation behavior and provides valuable managerial implications for online charity platforms.

3 - Echoes from the Past: Too Costly or Worth Gaining to Be Nice Again?
Guangxin Yang, Peking University, Beijing, China, People's Republic of, Qiaowei Shen

Existing psychological literature in prosocial domain suggests previous moral behavior sometimes leads people to do more of the same, while sometimes it liberates people to do the opposite in lab settings. We reconcile this debate by providing different types of historical donation information and conducting a large-scale field experiment on a major charitable platform. The real-world study involving millions of participants identifies effects similar to both consistency-seeking and licensing phenomena, with frequency VS. amount information of prior moral behavior being primed, respectively. Further analysis suggests that while amount information primes people the cost to do good deeds, frequency information relates more to possible gains. Additionally, such historical moral information also works as a decent peripheral cue for opt-put donators, accelerating their decision processes of not donating. Given the large heterogeneity of the population, we finally develop a personalized targeting policy based on causal machine learning algorithms. The best policy allocates the optimal treatment to each potential donator and demonstrates a sizable improvement in the donation ratio compared to the status quo. Our results shed lights on both psychological mechanisms in charitable research and managerial practice in platform design.

4 - B-DiDi: A Novel Bayesian Approach to Difference-in-differences Estimation with an Application to Organ Donation Policy Changes
Daniel Winkler, University of Economics and Business-Vienna, Vienna, Vienna, Austria, Pascal Güntürkün
Recent difference-in-differences literature has focused on two main extensions for the case of multiple available periods. First, the canonical difference-in-differences model (two periods and two groups, one of which receives treatment between the first and second period) has been extended to staggard adoption of the treatment. Second, methodology to inidcate robustness of the treatment effects to violations of the parallel trend assumption has been developed.

In this project I present a Bayesian extension of the "staggard adoption" difference-in-differences estimation based on Gaussian state-space models and show that it has multiple benefits. First, the lack of pre-treatment parallel trends violations can be shown naturally in a Bayesian framework using the Savage-Dickey density ratio. Second, I propose a more robust way of showing violations by allowing them to accumulate over time. Third, the robustness to a hypothesized violation in the parallel trends asumption of treatment effects can be easily assessed by shifting the posterior distribution of the treatment effects. Finally, modern shrinkage priors can be used to obtain confidence in the presence of treatment effects (or lack thereof) and their possible time variation.

The benefits of the proposed model are shown in the context of organ donation policy. In recent years multiple countries have changed the default policy from opt-in to opt-out for deceased organ donations, hoping to increase the number of available organs for transplantation. In this context robust identification of the causal effects of the policy change on both living and deceased donations is crucial to guide policy makers.
TA08
Meeting Room C3.3
Customer Relationship Management: Targeting Dormant/Churned Customers
Contributed Session
Chair: YuXuan Zhao, Guoshun Road No.670, ShangHai,China
1 - Engaging the Already (dis)Engaged: A Customer Relationship Management Field Study in an Emerging Market
Peren Ozturan, Copenhagen Business School, Copenhagen, Denmark, Aysegul Ozsomer, Gurkan Saman, Fernaz Alimoglu Sar, Aytac Yapici, Erman Adali, Gokce Yavuz

Customer relationship management campaigns in competitive service industries rely on costly segmentation efforts (Lemmens and Gupta 2020), and their effectiveness is altered by some external factors such as the social network of customers (Ascarza, Ebbes, Netzer, and Danielson 2017). Dynamic nature of growing markets hardens this task of resource allocation across different segments (Min, Zhang, Kim, and Srivastava 2016). Therefore, it is important to examine the overall effectiveness of these campaigns as well as differences across segmentation approaches (Brusco, Cradit, and Tashchian 2003). In this study, we investigate the existing segmentation efforts of a telecommunication company operating in an oligopolistic market structure in a large emerging market and offer latent profiling as an alternative segmentation approach. We utilize data spanning across 100,000 individual customers of the company over a 12-month period. We employ psychographic, behavioral, and loyalty-based variables (Lemmens and Croux 2006, Lemmens and Gupta 2020) and find out four main segments in each of the post-paid and pre-paid lines varying mainly in their engagement with the firm’s free and paid marketing efforts. A follow-up study with four field experiments shows that the proposed latent profiling-based segmentation approach enables higher marketing effectiveness, that is increased adoption of marketing campaigns. We thereby extend previous work on segmentation in emerging markets (Schlager and Maas 2013, Tanusondjaja, Greenacre, Banelis, Truong, and Andrews 2015) and guide managers therein with more specific insights.

2 - Big Data or Big Brother: For Whom does Personalized Recontacting Work?
Yin Bai, Tianjin University, Tianjin , China, People's Republic of, Min Zhang, Qingmei Tan

Extending retargeting literature focusing on inferring the decision stage through fragmented specific actions of a purchase journey, we apply a human-machine collaboration approach to uncover consumers’ purchase concerns from a series of actions in their purchase journey holistically. Furthermore, through a randomized field experiment, we find that the impact of personalized retargeting on purchase conversion varies depending on the dominant purchase concern. Specifically, personalization is effective for consumers with product-related concerns, such as product fit or price, leading to improved purchase conversion. However, for consumers with privacy concern dominance, personalization can backfire, resulting in decreased purchase conversion. Substantiating these contrasting effects highlights the importance of establishing a connection between two research streams (i.e., retargeting audience and retargeting content) from the perspective of identifying and addressing the dominant purchase concern. Insights from the studies can help firms make better choices about whom to retarget and what to say to them when developing retargeting strategies.

3 - How to Kiss the Sleeping Beauty: Dynamic Retargeting Strategy in Non-contractual Service Setting
Zhen Fang, Fudan University, Shanghai, China, People's Republic of, YuXuan Zhao, Chen Lin

Understanding consumer dynamics has been formidable important for firms to formulate appropriate reactivation marketing actions. However, in today’s digital, high-frequency service context, consumers’ strategic forward-looking behavior, which can highly influence the reactivation outcome, has rarely been studied. Using a large-scale panel data and from a leading coffee brand, we propose a Forward-looking Hidden Markov Model (FHMM) to examine personalized retargeting strategy by (1) inferring the hidden states of each customer, (2) discovering the effectiveness of multi-period contact timing and communication strategies on state transition, and (3) investigating the causal impact of reactivation strategies on future purchase across different hidden stages. Contrary to previous belief of the “recency trap” (i.e., recency increases for customers who do not purchase in a given period, making it even less likely that they will purchase in the next period), our model shows that communication messages (new product notification vs. coupons), leads to a 40% reactivation rate, and more loyal behavior in the next period. Our research provides insight on managing customer lifecycles in non-contractual service setting. When dealing with lapsed customers, marketers realizing customers’ “dynamic mindset” can also better adjust its marketing mix that converts potential customer churns into high value opportunities. Our research enhances the reactivation strategy in customer relationship management literature by examining customers' future-oriented behavior leveraging a novel FHMM model.

Key Words: Service Marketing, Forward-looking Hidden Markov model, Customer Relationship Management, Reactivation Strategy

4 - Sensory Stimulation in Customer Churn Management: an Empirical Study of Reactivation Effectiveness
YuXuan Zhao, Fudan University, Shanghai, China, People's Republic of, Chen Lin, YunLu Yin

Research across the realms of medical, psychology, and more recently in marketing, underscores a strong correlation between sensory perception and individuals' cognitive processes, emotions, judgments, and memory. Moreover, sensory-encoded information, particularly scent-encoded, has been verified to endure for prolonged durations compared to information encoded by other cues. Nevertheless, there was limited understanding of how sensory elements could serve as a subconscious trigger to enhance consumers' memory recall and facilitate future purchases. Using extensive panel data from a prominent coffee brand, we introduce a novel retargeting method that leverages sensory cues to re-engage consumers with extended purchase intervals. In this research, we advance sensory marketing by (1) identifying the most effective sensory for eliciting consumers’ memories, (2) exploring the causal impact of reactivation strategies on future purchases across various sensory cues, and (3) scrutinizing the cooperation and conflicts in sensory perceptions during the retargeting process. Our findings indicate that nearly half of consumers respond positively to a retargeting strategy involving sensory cues combined with communication messages, such as new product notifications or coupons. This positive response translates into increased activity and engagement during the subsequent period. We offer valuable insights for real-world marketing practices, suggesting that sensory marketing can be effectively integrated with the marketing mix to address lapsed customers. This integration proves instrumental in aiding customers in recalling products and, consequently, igniting heightened engagement.

Key Words: Sensory Marketing, Consumer Memory Recall, Customer Relationship Management

TA10
Meeting Room C3.5
Music Marketing
Contributed Session
Chair: Alexander Zhou, No. 5, Seventh street Hong Lok Yuen, NT
1 - Reinventing Music Consumption: A Comparative Study of the Impact of Metaverse Concerts and Physical Concerts on Piracy and Legal Downloading
Zhenyu Wang, University of Sydney, Sydney, Australia, Qiang Lu, Ranjit Voola, Jiang Qian

Abstract:

In recent years, the popularity of Metaverse concerts has surged. They provide inclusive, immersive experiences for audiences unable to attend physical events, potentially presenting a solution to the persistent challenge of music piracy. This has profound implications for artists and the music industry. This research aims to address the pivotal issue of the impact of Metaverse concerts on legitimate and illegal music downloads.

Utilizing a comprehensive dataset that encompasses both pirated and legitimate music downloads, our analysis employs causal inference methods and machine learning techniques. The findings unveil a compelling dichotomy: physical concerts correlate with an increase in music piracy, while Metaverse concerts contribute to its reduction. These insights not only underscore the contrasting effects of various concert formats on music consumption but also offer significant theoretical and practical contributions.

This study highlights the Metaverse's role in reshaping the landscape of music distribution and consumption, providing a strategic avenue to mitigate piracy while enhancing the global reach and concert experience. The research is instrumental in informing industry stakeholders, guiding them to adapt and thrive in the rapidly evolving digital music era, and paving the way for future explorations in virtual entertainment platforms.

Key words: Music Piracy, Metaverse Concerts, Physical Concerts, Legitimate Music Downloads, Causal Inference.

All authors contributed equally.


2 - User Engagement on Digital Media Platforms: The Dynamic Interplay between Discovery and Consumption
Qi Xie, University of Minnesota, Minneapolis, MN, United States, Linli Xu, George John

On digital media platforms, two major types of user interfaces have been widely used to facilitate the process of digital content discovery and consumption. The first type is often referred to as “content discovery page,” where users are presented with a list of options to browse and choose from. The second type allows users to continuously consume content without the need to browse by automatically presenting a new item when they are done with a particular content. An emerging trend in practice is to utilize both interfaces to improve user engagement. This study examines how the users’ dynamic experience on two interfaces shapes their behavioral patterns on such digital media platforms. We develop an empirical model framework capturing each user’s sequence of decisions in discovering and viewing content and allowing for individual heterogeneity. We then estimate the model in a Bayesian framework using granular data from a music mobile platform in China. We find that: 1) A broad variety of content helps mitigate satiation, lengthening the duration of user engagement on the platform. 2) On average, consumers prefer to browse more options on the discovery page, but a greater variety level during the browsing process is likely to make a user switch to consuming content. 3) Our examination of user heterogeneity reveals two types of users: “Binge-viewers” who favor extended viewing within a single session and waits longer to return to the platform; and “Browsers”, who prefer exploring a wide set of content and returns to the platform sooner.

3 - Optimal Playlist Design Strategies over the Product Life Cycle
Zhaokun Li, University of New South Wales Sydney NSW, Sydney, New South Wales, Australia, Junbum Kwon, Jack Cadeaux
With the evolution of the music streaming industry, various stakeholders now use playlists directly or indirectly to promote songs. Curators create and update playlists. Despite the growing importance of playlists as a marketing tool, we know little about how playlist design strategies influence their ability to gain followers. This paper investigates (1) the impact of playlist design strategies on follower gain and (2) how this effect is moderated by the product life cycle (PLC) stage of the playlist. Utilizing weekly follower data from 974 playlists curated by 501 curators over 329 weeks, encompassing 445,647 tracks, the paper examines how five design variables—cover, serendipity, diversity, popularity, and “untrendiness”—affect follower gain speed and how these effects vary across different PLC stages of a playlist. The analysis identifies optimal designs for an average playlist based on these variables and highlights significant variations in optimal design strategies at different PLC stages. Attributes beneficial at one stage may prove detrimental at another. This study contributes to the field of music streaming by suggesting some optimal design strategies for this emerging marketing tool. The findings also offer some specific managerial recommendations to allow curators to move beyond the introduction stage and extend the growth stage for their playlists.
4 - Does Diversity Lead to Thrive? Exploring the Impact of Acoustic Differences on Music Success
Alexander Zhou, The University of Hong Kong, Hong Kong, Hong Kong, Tingting Fan

In the digital music era, where around 120,000 tracks debut daily (Luminate, 2023), success proves a rare challenge for artists, producers, and marketers. To stand out, creators often attempt to deviate from the norm. However, being too different may distance a creation from prevailing trends, potentially hindering success. Prior research on the acoustic elements influencing music success has focused on the impact of specific music features (e.g., tempo) on consumer response or examined broad acoustic elements across albums, often neglecting their nuanced role in different songs and genres. Our study examines a comprehensive set of acoustic features (e.g., loudness, key, mode, tempo, vocal, duration) and constructs a Deviation Index for each feature in each song. Focusing on the influence of deviation in acoustic features on consumer responses (e.g., streams and ratings), our multimedia dataset spans popular genres (e.g., pop, rock, jazz) and includes Spotify audio tracks, music tags, ratings, releases, and artist information from social media and music rating websites. Using machine learning, we analyze the acoustic features of all songs. Employing a mixed-method approach (i.e., recursive mixed-process models) with controls for individual artist heterogeneity and music genre diversity, we find that unique acoustic features enhance positive consumer responses to music. However, the optimal degree of deviation varies across features and genres. Our findings offer crucial implications for artists and producers in the music industry and contribute to the literature on variety-seeking, optimal stimulation levels, product portfolio strategies, and recommendations beyond the music domain.

TA11
Meeting Room C3.6
Online Reviews: Factors Driving Persuasivenes and Helpfullness 1
Contributed Session
Chair: Charan Bagga, University of Calgary
1 - Does Rating Distribution Polarization Matter: The Interaction of Rating Distributions,Average Ratings, and the Number of Reviews on Consumer Decision-making
Jialei Wang, Southwestern University of Finance and Economics, Chengdu, China, People's Republic of, Xiaorong Fu, Tianshu Chu

Consumer product reviews are often displayed in both graphical and numerical form on webpages. These reviews offer prospective consumers a range of rating distributions to consider when evaluating a product. Using graphical rating distributions is a convenient way for consumers to visually analyze data dispersion, mode, and skewness. Although scholars have studied the impact of rating distributions on consumer perceptions and decision-making extensively, a limited body of research is available on how polarized rating distributions and numerical reviews affect consumer choices based on graphical depictions. This study investigates how polarized and non-polarized graphical rating distributions impact average ratings and the number of reviews involved in consumer decision-making. A polarized rating distribution has a relatively high number of 5-star or 1-star reviews and a small number of intermediate ratings. Based on data obtained from the most widely used review sites, a polarized distribution rating diminishes the influence of average ratings on consumers’ purchase intention while simultaneously increasing the significance of the number of reviews. Although the impact of polarized rating graphs on the importance of average ratings in consumer decision-making is not significant when the average rating is unfavorable, its influence on the number of reviews remains significant. A polarized rating distribution graph weakens the power of the number of reviews on consumers’ purchase intention. This study advances our understanding of consumer decision-making and several useful practical implications.

2 - CANCELLED An Empirical Investigation of Factors Affecting a Consumer’s Decision to Read More Reviews and to Make a Purchase
Stanislav Stakhovych, Monash University, Caulfield East, Australia, Aakash Mayur Shah

Online reviews have become a norm in website design and form a crucial aspect of the consumer decision-making process. While prior research has examined the influence of online reviews on a customer’s purchase decision, it has done so by focusing only on the aggregate measures of reviews, overlooking the actual text. In this study, we investigate the influence of both aggregate and textual aspects of online reviews on a customer’s decision to purchase using individual-level click stream data. Furthermore, we incorporate in our analysis the factors that lead consumers to read the reviews in the first place. In other words, we study a consumer’s process of extracting information from reviews and the influence it has on the subsequent purchase decision. We apply a hierarchical Bayesian framework, controlling for customer heterogeneity and addressing potential endogeneity concerns. We find that consumers choosing to read reviews focus more on the information gained from the individual reviews rather than aggregate measures such as the average rating and volume. Specifically, customers reading reviews focus on the message of the reviews, not how they are presented, that is, their language. Besides, customers reading reviews are also likely to read reviews for a product more than once, that is, across multiple sessions, and are likely to spend considerable time reading reviews. The findings of this study will provide marketers the means to better understand the impact of online reviews on consumers’ decision making and, as such, improve customer engagement and conversion rates.

3 - Is This Review Really Useful?: Using Topic Modeling and Opinion Mining Modules to Measure Movie Review Usefulness
Yeeun Lee, Dongguk University, Seoul, Korea, Republic of, Sunnyoung Lee, Dongyoon Lee

Consumers use online WOM to reduce perceived risk in their decision-making (Hussain et al. 2017). With the growth of social media platforms, consumers are more likely to express their opinions, and in fact, they read reviews before visiting a movie theater (Statista 2018).

We study what review features consumers find useful. We collected 138,733 reviews for 106 movies in 9 genres on the movie review platform IMDb. We propose to analyze review usefulness for spoiler and non-spoiler groups by focusing on user information, user preferences, and genre.

To analyze review usefulness, we first used an opinion mining module from a previous study to investigate user preferences (Cheng, L. C. et al, 2020), and found out their preferences for 6 features : OA (movie), ST (plot), SE (effects), MS (music), PDR (director), and PAC (actor) (Li Zhuang et al, 2006), and the ratio of upvote votes to total votes was calculated to derive review usefulness (Korfiatis et al., 2012).

As a result of the topic modeling of 9 genres, 6 spoiler topics about the plot of each movie and 13 non-spoiler topics about the movie features were extracted for SF. Analyzing through the opinion mining module is expected to provide a more objective interpretation.

Our study analyzed the significant effects of genre, user information, user preference, and review text among the features of reviews, which can be used as a basis for future research. Online review platforms can positively consider disclosing user's information to readers.

4 - The Impact of Lack of Reviewer Anonymity on the Quantity and Quality of Product Reviews: Evidence from a Mobile Application Market
Charan Bagga, University of Calgary, Calgary, Alberta, Canada, Sandeep Arora

There is ample evidence in academic literature that online user evaluations of products impact their sales (Chevalier and Mayzlin 2006) and revenues (Liu 2006). Scholars also show that the impact of user evaluations on product sales becomes even stronger when descriptive information on the reviewer’s identity is disclosed (Dou et al. 2012). Hence, it is not surprising that Google started linking user reviews of Apps sold through its Google Play store to the public profiles of reviewers. Even though reviewer identity disclosure is expected to increase the credibility of reviews, it is not clear if it will have a positive or negative impact on the number of reviews written for Apps. Scholars show that it is the volume of user evaluations and not their valence that influences customer purchase decisions (Duan, Gu, and Whinston 2008). Hence, a lack of reviewer anonymity may it may affect adoption speed of Apps positively or negatively. Scholars have also shown that a lack of anonymity encourages ‘group-think’ and forces individuals to follow group norm, thus hampering originality of opinions. As the variance of ratings of a product has been shown to impact its sales (Sun 2012) a decrease in diversity of user reviews is also likely to impact the adoption speed of Apps. Additionally, it is not clear whether the nature of reviews will also change once customers are not able to write reviews anonymously. We propose to answer these questions by using in-lab experiments and secondary data.

TA12
Meeting Room C4.1
Marketing and Social Impact
Invited Session
Special Sessions
Chair: Adithya Pattabhiramaiah, Georgia Institute of Technology, Atlanta, GA, United States
Co-Chair: Vamsi Kanuri, University of Notre Dame, Notre Dame, IN, United States
1 - Gender Stereotyping and Advertising Effectiveness
Martina Pocchiari, National University of Singapore, Singapore, Singapore, Clement Bellet

Gender-based targeting can resort to stereotyped gender representations to communicate marketing messages in a fast and effective way. However, these strategies can backfire when they rely on outdated clichés or perpetuate harmful beliefs about vulnerable social groups. While the detrimental role of gender stereotypes in education and in the labor market is well-documented, the effects of gender stereotyping within consumer markets remain understudied.

In our research, we initially assess the prevalence of gender-based targeting in American TV advertising using a comprehensive dataset of TV ads aired between 2010 and 2020, encompassing numerous brands across various product categories. We then combine this data with a consumer panel and store-level sales data to investigate to what extent brands benefit from gender-based targeting strategies, and the extent to which they suffer when these strategies hinge on stereotypical content. We integrate unsupervised and supervised machine learning approaches to measure gender-stereotypical and non-stereotypical ad content from unstructured text. Based on these measurements, we categorize brands based on the extent to which they rely on gender-stereotypical content. Drawing from Shapiro et al. (2021), we estimate the effectiveness of gender-based targeting strategies on store-level sales at the brand level, discuss the role of stereotypical content as a potential driver of heterogeneity in effect size, and contemplate the implications of our findings for the future of gender-based targeting.

2 - Do Digital Payments Increase the Consumption of Vice Goods? Evidence from the Introduction of UPI Payments in India
S Sriram, University of Michigan, Ann Arbor, MI, United States, Varad Deolankar

The Government of India launched UPI, an instant real-time cashless digital payment system in August 2016. This gave more than 260 million consumers in India the experience of making purchases digitally in a cashless manner for the first time in their lives. In addition, by enabling money transfer between the bank account of a consumer and that of the merchant using a single unified interface, UPI increased both the transparency of transactions and budget salience at the point of transaction. Thus, on the one hand, increased transaction transparency and budget salience suggest that consumers would curb their expenditure while using such a cashless digital mode of payment. On the other hand, prior research on consumer purchase behavior has shown that plastic-based cashless payments result in consumers experiencing weaker pain of payment while making purchases, which in turn, increases their shopping expenditure. We investigate the association between adoption of UPI and consumer expenditures, with special emphasis on vice goods. We find a positive association between the amount of household shopping expenses and the increased proliferation of the UPI-based cashless digital mode of payment. Interestingly, we also find a positive association between the share of expenses on vice goods and the increased proliferation of the UPI-based cashless digital mode of payment. This finding provides suggestive evidence in favor of the pain-of-payment account. Our research highlights the importance of considering the unintended consequences of policies aimed at improving access to financial services.

3 - The Role of Carbon Taxes in the Airline Industry
Madhu Viswanathan, Indian School of Business, Hyderabad, India, Mauricio Varela
The airline industry contributes 2% of global CO2 emissions and these are expected to grow substantially over the next 25 years. We study the effect of a carbon tax on promoting efficiency and curbing consumption in the US domestic sector. We find no effects of shifts in long-run fuel prices on fuel efficiency and estimate an elasticity of long-run fuel prices on air service consumption of -0.06. Together, these findings imply a $40/mT of CO2 tax would generate a 1% reduction in CO2 emissions in this industry. We estimate the elasticity of long-run fuel prices on consumption using two approaches, a top-down approach that uses national accounts and a bottom-up approach that models demand and pricing decisions at the route level. Both approaches suggest elasticities between -0.2 and -0.06. The route-level analysis allows for welfare calculations, from which we estimate a welfare loss per abated ton of CO2, exclusive of the social benefits of reduced pollution, of $937/mT. A smarter implementation of the tax, taxing cities differently, could reduce this welfare loss to $786/mT while reducing pollution by the same amount.
Keywords: carbon tax, US airlines, jet fuel tax
4 - School Shootings and Community Economic Activity
Adithya Pattabhiramaiah, Georgia Institute of Technology, Smyrna, GA, United States, Muzeeb Shaik, Mike Palazzolo, Shrihari Sridhar
School shootings in the United States have become depressingly common. While the psychological effects of these tragedies are well-documented in the literature, their economic impact has received less attention. Combining data from multiple sources, we study the effects of fatal school shootings on three barometers of economic activity: grocery purchases, vice goods purchases, and home values. We find that economic activity declines across all three areas, but in ways that are asymmetric across counties' political orientations. Specifically, we find a roughly 1% reduction in grocery purchases in liberal-leaning counties and a 6.5% decline in home values in conservative counties, following a school shooting. Interestingly, we find no change in alcohol spending, but do find a 13% drop in cigarette spending in conservative-leaning counties -- welcome news in light of the tendency of vices to be used as a coping mechanism. Our findings show that the impact of these acts of violence is far-reaching, harming the surrounding community economically. This reduction in economic activity suggests a need for community leaders to support economic recovery efforts for affected communities, in addition to the usual remedial measures directed at victims and/or their immediate families. Moreover, the reduction in home values suggests that fatal school shootings harm a community's desirability, suggesting a need for the types of "community marketing" that community leaders often engage in to attract new and retain existing residents.
5 - A Novel Approach for Responsible Targeted Advertising
Vamsi Kanuri, University of Notre Dame, Notre Dame, IN, United States, Sriram Somanchi, Shrihari Sridhar
More than 60% of today’s half-a-trillion-dollar worldwide digital advertising spend is allocated toward programmatic advertising, where users are micro-targeted on various observed user characteristics in real-time. In theory, the micro-targeting ability of programmatic advertising should simultaneously fulfill the goals of the ad platform to identify the best advertiser for an ad slot as well as the advertiser to reach the right target customer. Yet, emergent research shows that micro-targeting could propagate unnoticed societal biases in advertising exposures, where it can inadvertently favor users in some demographic segments, simply by virtue of being correlated to the advertiser’s targeting criteria. In this research, we leverage advances in computational machine learning to propose a novel approach called responsible targeted advertising (RTA) to mitigate such algorithmic bias. Through simulations, we demonstrate the tradeoffs between economic and social returns while using RTA and discuss the implications of giving advertisers the ability to choose the level of equitability in real time.
TA13
Meeting Room C4.2
Game Theory
Invited Session
Special Sessions
Chair: Yi Zhu, University of Minnesota, Minneapolis, MN, United States
Co-Chair: Krista Li, Indiana University, Bloomington, IN, United States
1 - Content Provision on UGC Platforms
Woochoel Shin, University of Florida, Gainesville, FL, United States, Wilfred Amaldoss
Consumers visiting platforms that host user-generated content (UGC) not only consume content but also generate content by investing time and effort. We seek to examine a UGC platform's content provision strategy: how a UGC platform can motivate consumers to generate UGC and how it can manage the balance between UGC and platform's own content. As UGC and the platform's own content perform the same function, one may be inclined to think that the two types of content are substitutes. Our analysis shows that they could function as strategic complements. This is because increasing the platform's own content provision raises the quality of content on the platform, motivates more consumers to join the platform, and increases the total UGC provision on the platform. The fact that consumers dislike advertising could lead us to believe that consumers will be less motivated to generate UGC if ad space increases. On the contrary, we find that consumers may be motivated to increase UGC provision to make up for the loss in enjoyment and increase the overall quality of contents on the platform. The public good characteristics of UGC could prompt us to think that UGC provision on the platform will be less than the socially optimal level. Our analysis identifies conditions when the total provision of UGC can be more than the social optimum. One may wonder whether it is profitable for a UGC platform to completely dispense with its own content.
2 - Optimal News Search and Reporting
Yi Zhu, University of Minnesota, Minneapolis, MN, United States, Weize Yin
One consumer’s belief of an event may frequently fluctuate as she encounters different news reports during her news search on a news medium. In this paper, we investigate how a medium strategically chooses its news reporting slant to maximize consumer engagement and, consequently, advertising revenue. In our model, the consumer not only values information acquired through the news search—leading to more assured beliefs—but also prefers news that aligns with her inherent attitudes. By establishing a micro foundation for the consumer’s sequential news search, we model a Bayesian consumer’s belief update based on a stochastic Brownian motion. We explicitly outline the expected duration of the consumer’s news search, allowing us to examine the optimal news slant for the medium. We find that a consumer stops her news search if her belief becomes sufficiently certain. A stronger attitude tends to prolong a consumer's search time, even if the medium's slant opposes the consumer's attitude, provided the consumer's attitude is sufficiently strong. Counterintuitively, we find that the medium tends to be more slanted if the consumer values information learned from news search more. This is because the medium can utilize news slant to obfuscate the consumer’s information learning, thereby retaining the consumer for a longer duration.





3 - Paying for Effort Not Luck: When Should Firms use Compensation Plan Decelerators?
Rob Waiser, University of Kansas, Lawrence, KS, United States, David Soberman
One of the key decisions in designing an incentive compensation plan for a salesperson is the shape of the payout curve (i.e. the relationship between sales and the salesperson’s incentive payout). Existing studies provide a number of intuitive explanations for the common use of payout ‘accelerators’ – increasing marginal payouts for sales above a specified threshold. On the other hand, the literature provides little justification for the use of ‘decelerators’, which are also commonly observed. We use a game-theoretic model to better understand why firms use payout decelerators and when it is optimal to do so. Our model introduces two factors that are not commonly incorporated in compensation design models. First, market demand for a product can be uncertain when incentives are designed, revealing itself later and allowing the salesperson to adjust her effort accordingly. Second, different sales outcomes can vary in sensitivity to sales effort. We find that the optimality of payout decelerators depends on both of these factors. For example, if a high-demand product ‘sells itself’ with little salesperson effort while a low-demand product requires strong effort to achieve more modest sales, then decelerating payouts are optimal when demand is uncertain.
4 - Retention or Acquisition? Behavior-Based Quality Disclosure
Krista Li, Indiana University, Bloomington, IN, United States, Jianqiang Zhang
Firms frequently introduce new products and offer free samples, demonstrations, or trial versions for customers to learn about the quality of new products before purchasing. Based on customers' purchase history data, firms face two customer segments: their own customers and the competitor's customers. Firms practice behavior-based pricing (BBP) by offering discounts to poach the competitor's customers. Moreover, firms can practice behavior-based quality disclosure (BBD) by disclosing quality information to the two customer segments differently. In this paper, we use a two-period game-theoretic model to examine how firms practice BBD in addition to BBP. We find that in contrast to the acquisition-focused BBP, firms perform retention-focused BBD by disclosing quality information more likely to their own customers than to the competitor's customers. In the second period, BBP intensifies competition and reduces profit, whereas BBD improves disclosure efficiency and increases quality differentiation to increase profit. In the first period, BBP increases profit, whereas BBD reduces profit when the disclosure cost is low while increasing it when the disclosure cost is high. BBD increases firms’ total profits if and only if the cost of disclosure is high. Firms endogenously choose BBD, resulting in a win-win outcome when the disclosure cost is high and a prisoner’s dilemma when the disclosure cost is low.
TA14
Meeting Room C4.3
Consumer Behavior: Advertising 1
Contributed Session
Chair: Matthew Tingchi Liu, University of Macau
1 - Conveying Advertising Affection with Virtual Endorsers: The Interactive Impact of Affection Appeals and Endorser Types on Advertising Effectiveness
Chenying Hai, Huazhong University of Science and Technology,China, Wuhan, China, People's Republic of, Jun Yan, Shaowei Chai
The realistic appearance and manipulable nature of virtual endorsers aid enterprises in shaping brand image and enhancing brand awareness. While virtual endorsers are frequently used for affectional marketing with affection appeals, their overall effectiveness is inconclusive among scholars. Drawing on mind perception theory and construct level theory, we analyze consumer perceptions of the psychological distance between virtual endorsers and human endorsers. Accordingly, we propose distinct advertising affection appeal strategies based on the language expectation theory for different endorser types. Six experiments reveal a matching effect between endorser type and advertising affection appeal type on effectiveness. Specifically, virtual endorsers excel with abstract affection appeals, while real endorsers perform better with concrete affection appeals. Empathy acts as a mediator and the belief in virtual endorsers' affection competence serves as a moderator in the above effect. This study focuses on the effectiveness of virtual endorsers in conveying advertisement affections, offering them an innovative advertising appeals strategy. It complements theoretical research on virtual endorsers' advertisements, stimulating further consideration of virtual humans' roles and functions. The exploration of affection appeals enriches related research on advertising appeals. Lastly, this research provides valuable insights for companies planning their advertising campaigns.
2 - Effectiveness of Necro-Advertising in Ethical Fashion Engagement
Fuxuan Jia, University of New South Wales, Sydney, Australia, Ka Wing Chan, Chris Baumann

The fashion industry faces ethical criticism due to substantial resource consumption. Yet, ethical fashion has been praised for its environmental benefits and minimal resource demands, representing a more sustainable alternative. Still, ethical fashion brands face the challenge of being less successful in engaging audiences on social media. To overcome this challenge, our study proposes necro-advertising —the use of deceased celebrities in advertising and an under researched area— as a novel strategy to increase social media engagement. Despite necro-advertising's potential to positively influence consumers and increase sales, existing studies often focus on evaluating ethical aspects, neglecting focusing on its effectiveness. This study addresses this important gap by examining whether, when and why necro-advertising impacts social media engagement for ethical fashion brands, shedding light on the effectiveness of necro-advertising.

Analyzing 12,581 social media posts from 11 ethical fashion accounts, we find that featuring necro celebrities significantly boosts social media likes and comments. In fact, the longer the time passed since the necro celebrity’s death, the stronger the demonstrated effect. To further deepen our understanding of necro-advertising, we investigate the moderating effects of the colorfulness of images used, and the underlying mechanism of vintage anemoia, i.e., a nostalgic longing for bygone eras, through experiments.

This study offers an important theoretical understanding of the effectiveness of necro-advertising in generating engagement, an area far neglected in the literature. We shed light on the influence of color features and underlying mechanisms how such effect customers, offering actionable guidance for marketers to create more effective advertising content.

3 - Nativeness Backfire: Unraveling the Dynamics of Design and Content Nativeness in In-Feed Ad
Shaowei Chai, Huazhong University of Science and Technology,China, Wuhan, China, People's Republic of, Jun Yan, Yaping Chang, Chenying Hai

Advertisers widely adopt in-feed ads for their native format in response to the ad avoidance. Yet, a crucial question remains: does increased nativeness in in-feed ads consistently enhance advertising effectiveness? This study delves into this query by examining the interactive effects of design and content nativeness in in-feed ads, aiming to uncover potential pitfalls associated with heightened nativeness. While prior research acknowledged the dual impact of in-feed ads, the authors primarily focused on isolated nativeness dimensions, i.e. design or content. This study addresses this gap by exploring the combined influence of design and content nativeness on advertising effectiveness. Conducting online experiments (including an eye tracking) on a virtual news website, we find that low (vs. high) design nativeness catches more advertising attention and subsequently fosters more positive advertising attitude only when content nativeness is high (vs. low). Utilizing the Statistics plug-in of WordPress.com, we illustrate that in-feed ad with low design nativeness matching high content nativeness (low-high) can achieve the highest click-through rate (CTR) at 5.3%, surpassing other combinations (high-high: 4.6%; high-low: 4.4%; low-low: 0.8%). The eye-tracking experiment verifies the positive relationship between attention and click-through rate. This research significantly contributes to our knowledge of in-feed ad nativeness by examining the interaction of different dimensions of nativeness, providing advertisers with cost-efficient advice to mitigate nativeness backfire through proper matching of the design and content of in-feed ad.

4 - The Interplay Between Color Temperature and Psychological Distance
Matthew Tingchi Liu, University of Macau, Macau, Macau, Mingxia Zhu, Xi Song

Keywords: Psychological distance, Color temperature

As previous research intensively investigates how image affect consumers’ behavioral consequences, we focus on one novel aspect of image: color temperature. Particularly, we connect the color temperature with psychological distance. According to theory of color, warm regions appear to advance toward the viewer while cold regions appear to recede. We hypothesize that this visual perception may activate a sense of psychological distance that is more proximal or distal. To identify the effect of color temperature (warm vs. cold) on consumers’ perceived temporal and spatial distance, an experimental between-group study was employed. 230 participants were recruited (Mage = 30.83 years, SD = 7.96; 66% female). The participants view a poster of a newly open coffee shop in either warm or cold color. Then, they were asked to estimate when the coffee shop open (temporal distance) and how far the coffee shop locate. People estimated that the coffee shop will open sooner after viewing the poster in warm color (Mwarm = 7.25 days, SD = 5.92) than in cold color (Mcold = 9.46 days, SD = 10.37; t(228) = 1.98, p < .01). In addition, people thought that the coffee shop would be closer in proximity after viewing the poster in warm color(Mwarm = 6.58 km, SD = 6.66) than in cold color (Mcold = 8.19 km, SD = 8.18; t(228) = 1.62, p < .05). We contribute to the existing literature on construal level theory and provide insights for marketers in designing effective marketing strategies.

TA15
Meeting Room C4.4
Retailing: Location
Contributed Session
Chair: YUTIAN LIU, University of Wisconsin - Madison, 975 University Ave, Madison, WI, 53706, United States
1 - Electric Vehicle Charging Availability and Retail Store Performance
Ka Wing Chan, University of New South Wales, Sydney, Australia, Hauke Wetzel, Ljubomir Pupovac

Insufficient charging infrastructure slows down the widespread adoption of Electric Vehicles (EVs)—critical to achieving the United Nations’ Sustainable Development Goals. To overcome this problem, EV charging availability in proximity to retail stores has been spotlighted as a critical element. However, retailers lack clarity regarding the impact of EV charging availability on their performance. While some retailers view EV charging availability as an opportunity to increase sales, for instance, by increasing sales to EV drivers, others are afraid of possible negative consequences due to parking lots blocked for non-EV drivers and disfigured car parks. The goal of this research is to examine whether, why and when EV charging availability affects retail store performance. To analyze the effect of EV charging availability on retailer performance, we collaborate with one of Europe’s largest retailers. We use a dataset that contains monthly observations of EV-charging availability and store performance for 348 retailer locations, resulting in 17,748 store-month observations spanning from 2019 to 2023. Employing a quasi-experimental design and novel staggered difference-in-differences methods, we reveal that EV charging availability increases retail store performance in terms of dollar sales, and number of transactions. The findings further suggest that the increases in retail store performance can be explained by impulse buying effects and crowd avoidance effects. Overall, the study offers important new insights into the effect of EV charging availability on retailer performance. Equipped with this knowledge, retailers, policymakers, and investors can make more sustainable decisions.

2 - An Experiential Multipurpose Shopping Model to Assist Retail Location Decisions
Harmen Oppewal, Monash University, Caulfield East, Australia, Ari Pramono, Benedict Dellaert

Shopping centers increasingly become experientially rich, multipurpose destinations that attract consumers for a variety of reasons beyond only wanting to shop. Shopping center managers therefore need to know how experiential activities add value to a consumer’s visit to a shopping center. This paper proposes three ways in which consumer experiential activity choices differ from store choices as traditionally conceived. First, experiential activities tend to be discretionary, which means that consumers can choose to add more activities or to end their visit at any time. Second, spending time on an experiential activity will represent a benefit instead only a cost. Third, consumers will prefer to undertake experiential activities in particular, asymmetric orders. After theorizing the nature of experiential activities, the paper presents an experiential shopping activity model that accommodates these three aspects. The empirical part of the study consists of an experiment that asks participants to schedule activities for a specific visit to their downtown area using a map interface. The empirical results support the predicted impact of the three experiential characteristics activities and how they influence consumer visitation patterns. Simulations next demonstrate the managerial consequences of various experiential activity mix and location decisions.


3 - Physical Store Visits and Virtual Brand Perceptions: Market Structure Analysis with Customer Trajectories in Shopping Malls
Dai Yao, The Hong Kong Polytechnic University, Hong Kong, Hong Kong, Yulin Hao
Customers’ trajectories in shopping malls capture the sequences of the store visits and the time spent on each store. The trajectories at the aggregate level could uncover the collective customer perceptions of the various stores/brands within a mall. A unique feature of customer’s trajectories, as opposed to other data commonly used for market structure analysis, is that they are influenced by the physical locations of stores. In this research, the authors assemble customers’ trajectories from two shopping malls of a large mall group, and compress the data into a lower-dimensional space using a deep embedding model. They validate this approach by conducting the analysis with each of the two shopping malls, using an alternative baseline lift measure, and using customer surveys. They visualize the market structure of malls using the learned representations of stores. More importantly, they investigate the fusion of the physical mall space and the virtual market structure, and find: First, the physical distance between two stores within a mall plausibly decreases their conceptual similarity; Second, while conceptually similar (dissimilar) stores are usually located in close physical proximity (apart from each other), it is also not rare that there is mismatch between the two spaces.
4 - Unveiling Store Network Dynamics: A Graph Perspective of Offline Retailing Activities
Yutian Liu, University of Wisconsin - Madison, Madison, WI, United States, Qing Liu, Cheng He

Despite the continuous growth of e-commerce, a substantial 85% of retail sales in the U.S. still occur at physical stores in 2022 and 2023. Among these, shopping malls and business plazas play crucial roles as hubs for brick-and-mortar stores. Unlike previous studies that treat individual stores in isolation, our research adopts a social network perspective, viewing stores as interconnected nodes in dynamic graphs shaped by customers' shopping paths. Utilizing a unique dataset gathered from five shopping malls in China over a year, we construct daily dynamic store networks, capturing the connections between stores. Our analysis further delves into the impact of offline retail activities on these network structures. Our work offers a new perspective for both researchers and practitioners examining the dynamics of the offline retail environment.

TA16
Meeting Room C4.5
Livestreaming 1
Contributed Session
Chair: Wenyu Jiao, University of Science and Technology of China, Hefei
1 - Cool Thinking or Hot Feeling: The Role of Cognitive Empathy and Affective Empathy in Philanthropic Live-streaming Sales
Ying Yu, University of Macau, Macau, China, People's Republic of, Ying Ho
Philanthropic live-streaming sales gain importance nowadays due to its considerable economic value and impact on social well-being. During such live-streaming, anchors provide information regarding social issues or charitable causes to viewers, hoping to stimulate viewer’s empathy and generate positive consumer behaviors. While existing literature regards empathy as an important factor in shaping customer behaviors, relatively less research pays attention to the distinction between cognitive empathy and affective empathy. This study explores how quality of philanthropic information induces viewer’s cognitive empathy and affective empathy, which in turn influence their behavioral intentions (including purchase intention, word-of-mouth intention, willingness to pay price premium, and consumer interest). In particular, we highlight the different mediating roles played by cognitive empathy and affective empathy in the relationship between information quality and consumer behavioral intentions. Data was collected from 500 Chinese participants who viewed and/or purchased products during philanthropic live-streaming. Results show that when perceived homophily is controlled, information quality generates higher purchase intention and word-of-mouth intention via cognitive empathy, whereas it affects viewer’s willingness to pay price premium and interest in the philanthropic event via affective empathy. This suggests that recognizing the beneficiaries’ problems (i.e., cognitive empathy) is adequate to generate purchase and word-of-mouth intentions. However, being able to feel the beneficiary’s emotions (i.e., affective empathy) is required for viewers to pay price premium and demonstrate higher interests. Our results provide valuable insights to charity organizations, governments and live-streaming platforms on how to optimize philanthropic live-streaming sales operations and encourage positive consumer behaviors.
2 - The Effect of Speed Volatility on Consumer's Engagement in Live-streaming: Evidence from TikTok Game Live-streaming
Xinyue Xia, Beihang University, Beijing, China, People's Republic of, Chenxi Li, Anghong Wu, Yixun Lin

The global live streaming market, growing steadily at a rate of 30% and projected to reach $145.8 billion by 2024, is capturing significant attention from both industry and academia due to its substantial development potential. Considering that the voice of streamers is one of the most important stimuli in a live-streaming context, prior studies have explored the impact of various static voice characteristics (e.g., average speech rate, average pitch, tone) on consumer behavior. However, research addressing how dynamic voice features (such as speed volatility) of streamers influence consumer behavior is limited. We scrutinized how streamers' speed volatility affects consumer engagement using field game live-streaming data from one of the most popular game live streaming platforms in Asia. Our results reveal that streamers' speed volatility can positively affect consumer engagement (e.g., comments and gifting) and also uncover a novel synergistic interaction between speed volatility and two language styles: affective/cognitive and sensory language. Furthermore, a follow-up online experiment replicated the causal impact between speed volatility and consumer engagement, uncovering the underlying affective process mechanism. Our findings offer valuable managerial insights for the live streaming industry, suggesting that by merely adjusting their speech rate and language style, streamers can significantly enhance consumer engagement.

3 - Product Search and Sourcing in Live-Commerce: Evidence from a Quasi-Experiment
Jialu Liu, Shanghai Jiaotong University, Shanghai, China, People's Republic of, Chu (Ivy) Dang

Live-commerce has emerged as a major shopping channel in recent years. Those livestream platforms usually integrate external shopping channels (i.e., external e-commerce platforms) into their livestream sessions, allowing influencers and customers to access a broader selection of products. Leveraging the policy shock of banning external shopping channels on TikTok China in 2020, we investigate how reducing the market thickness of product sourcing pool impacts influencers, consumers, and the platform. Surprisingly, we find that reducing product sourcing pools significantly improves live-commerce performance, especially benefitting the mid- and low-tier influencers. To identify the mechanism behind these results, we conduct an in-depth analysis of various live-commerce-related marketing mix variables, such as product promotion, broadcast duration, product pricing and product choices. We find that the improved live-commerce performance is driven by the fact that the influencers are able to select better products from a smaller (vs. larger) product sourcing market. This surprising result occurs because limiting influencers to search and source products from a single platform has an unintended consequence, that is, the influencers are now able to learn the distribution of product quality within the sourcing channel more accurately. As a result, they can select higher-quality products for their livestream sessions. We rule out several alternative explanations and conduct various robustness checks. Our research highlights the significance of platform governance and regulations in shaping the live-commerce ecosystem and enhancing consumer welfare.

4 - Unlocking Sales Potential: The Impact of Livestreaming Hosts and Brand Origin on Consumer Choices in E-commerce
Wenyu Jiao, University of Science and Technology of China, Hefei, China, People's Republic of

With the popularity of livestreaming e-commerce, leveraging this platform to empower brands and boost sales has become critical for marketing managers. This study addresses the following research questions: (1) What types of livestreaming hosts should brands choose for livestreaming e-commerce? (2) How do brand and livestreaming hosts impact consumers' purchase decisions and product sales? We categorize brands into Chinese and non-Chinese based on their country of origin. Livestreaming hosts are classified as enterprise hosts and celebrity hosts. Utilizing data from Taobao and online random experiments, this research reveals that: (1) For Chinese brands, celebrity hosts drive more sales than enterprise hosts; (2) Product type (hedonic vs functional) moderates the effects of host type—the positive impact of celebrity hosts on sales is stronger in hedonic product categories; (3) Brand origin moderates the moderating effects of product type and host type. For Chinese brands, the moderating effect of host type and product type is positive. However, for non-Chinese brands, the moderating effect is not significant. These findings provide valuable insights into livestreaming and branding literature, offering strategic considerations for brands in choosing the 'right' host for livestreaming.

TA17
Meeting Room C4.6
Health Care Marketing: Patient Outcomes 1
Contributed Session
Chair: Fang Fang, CAL STATE LOS ANGELES, San Gabriel, CA, United States
1 - Enhancing Cancer Outreach Effectiveness through Monetary vs. Time Incentives: Insights from a Randomized Field Experiment
Xiuping Li, National University of Singapore, Singapore, Singapore, Siqi Dai, Zheshuai Yang

Cancer is a leading cause of death worldwide and accounts for nearly 10 million deaths in 2020. Despite the effectiveness of cancer screening in reducing mortality, the participation rates remain low in developing countries. How to incentivize the poor in developing countries to participate in free cancer screening programs has long been of interest to marketing scholars. However, to the best of our knowledge, there has been no consistent conclusion about the effectiveness of monetary incentives in promoting participation in cancer screening, and no research has explored the effectiveness of time incentives. Therefore, this research aims to examine whether time and monetary incentive could promote individuals’ cancer screening participation. Based on semi-structured interviews with 43 farmers and 10 doctors, we decided to examine the effectiveness of several incentives (20 cash vs. rice worth 20 yuan vs. Spending 20 yuan to send the test tube to participants’ home). A survey among 105 medical students found that 61.9% and 11.4% of participants predicted rice and delivering tubes to the door would be the most effective incentive respectively. We then conducted a large-scale randomized field experiment (N = 911) to measure farmers’ real behavior. Compared to no incentive, cash and rice significantly increased participation rates (31.7% for control condition; 49.4% for cash condition, p = .001; 50.4% for the rice condition, p < .001). Surprisingly, delivering the tube is the most effective incentive (63.9%, ps < .006). Our research has obvious theoretical and managerial implications for the government’s introduction of free cancer screening.

2 - Were Consumers Eating Healthier During the Covid Pandemic? It Was a Mixed Bag
Min Kim, NUS Business School, National University of Singapore, SIngapore, Singapore, Huong Nguyen, Sara Rosengren, Jie Zhang
The Covid-19 pandemic has brought about unprecedented disruptions to all aspects of work and life for people around the world for more than two years. In times of crises, consumer behaviors often exhibit major shifts. On one hand, consumers had been urged by public health officials and experts to adopt healthier lifestyles and eating habits in order to strengthen immunity and health conditions. On the other hand, many people might have resorted to various coping strategies to deal with pandemic-related disruptions and stress, including indulgence food that was unhealthy. Behavioral changes in a drawn-out public health crisis like the Covid pandemic can have profound impact that is long-lasting. Yet little is known about whether consumers had adopted healthier eating habits or become worse-off in their food health.

In this study, we investigate whether the healthiness of consumers’ food consumptions changed during the Covid pandemic, and if so, how the evolving patterns might be driven by their attitudes toward the pandemic and officials’ abilities to handle the health crisis. We analyze household purchase history data of a comprehensive range of food and beverage products from a large European grocery retailer, spanning from 2019 to 2022. We intend to make the following contributions in this research: 1) to discover detailed insights on consumers’ food consumptions and evolving patterns during the pandemic, 2) to offer insights to governments and public health experts on how to guide consumers toward adopting healthier lifestyles and to avoid unhealthy traps during a major public health crisis.
3 - Marketing for Good: Fostering Positive Doctor-Patient Relationships through Social Media Interactions between Doctors and Patients
Deli Kong, Shanghai International Studies University, Shanghai, China, People's Republic of, Jiayin Qi

In the age of social media, healthcare professionals leverage platforms like Douyin (Chinese TikTok) to engage with patients, providing health information and building trust. Drawing from the Professional-Client Interaction Theory, this study categorizes social media interactions into instrumental and affective interactions, examining the potential impact on perceived doctor-patient relationships through satisfaction and trust towards doctors. A questionnaire survey with 325/398 valid responses collected and structural equation modeling (SEM) used for this study.

The findings reveal that instrumental interaction did not significantly impact patient satisfaction (H1a, p > 0.05), while affective interaction significantly influenced patient satisfaction (H1b, p < 0.001, β=0.394 ). Instrumental interaction positively affected trust towards doctors (H2a, p < 0.001, β=0.33), and affective interaction also played a role in establishing trust towards doctors (H2b, p < 0.001, β=0.369). Patient satisfaction positively influenced trust towards doctors (H3, p < 0.001, β=0.247). However, there is non-significant relationship between patient satisfaction and perceived doctor-patient relationship (H4, p > 0.05). Examining trust towards doctors and its impact on perceived doctor-patient relationship, we observed a highly significant and positive effect (H5, p < 0.001, β=0.622). This suggests the pivotal role of trust in shaping perceived doctor-patient relationship on social media.

This study provides a low-cost and efficient approach to better conduct health marketing and improve the perceived doctor-patient relationships. Such strategies enables healthcare professionals to engage in effective health communication and potentially mitigate the rising trend of malicious attacks on healthcare professionals in China in recent decades.

4 - Nudging Patients’ Choices to Cost-Effective Providers to Reduce Healthcare Expenditures
Fang Fang, California State University, Los Angeles, Los Angeles, CA, United States, Mili Mehrotra, Hari Natarajan

Misalignments between patients’ choices of providers and those of the health insurance company (HIC) can result in significant costs. Misalignments may occur either because enrollees are unaware of their options or because they do not have an incentive to choose the cost-effective provider. Motivated by emerging mechanisms in the industry, we examine how an insurer can exert effort and/or offer cash rewards to nudge patients towards cost-effective providers. We build an analytical model that captures the salient aspects of an HIC’s decision problem while incorporating how enrollees choose providers. With this versatile framework, we analyze the HIC’s optimal effort and reward, individually and jointly, under different cost-share structures (i.e., copayment and coinsurance). Comparing the HIC’s savings with the effort and cash reward-based approaches, we find that when coinsurance is high, the HIC prefers the effort-based approach. Conversely, the cash reward-based approach is better when coinsurance is low and the price difference between the two providers is high. With copayment, the HIC prefers to use a cash reward when the price difference is high; otherwise, it prefers to exert effort. Thus, neither a reward-only nor an effort-only approach uniformly outperforms the other. The two approaches can serve as tactical complements as indicated by the superiority of the joint approach in some cases. This work provides a framework for the HIC to tailor the nudge (effort or reward or both) for different procedures and geographies based on the cost-share structure and the relative magnitude of related costs.

TA18
Meeting Room C4.7
Consumer Behavior: Promoting Broad Sustainable Behavior
Contributed Session
Chair: Lawrence Sze Lok Choi, WLB 501, 5/F, The Wing Lung Bank Building for Business Studies, HKBU, Kowloon tong, 000000
1 - From Limited Control to Green Empowerment: The Motivational Synergy of Fresh Starts in Predicting Environmental Collective Actions
Seon Min Lee, Korea University, Seoul, Korea, Republic of

This research explores the impact of the fresh start concept on individual motivation in collective environmental initiatives. It proposes that the fresh start effect significantly enhances the motivation of those with low personal control to participate in actions such as energy saving. The hypothesis was tested through two studies. A panel data analysis revealed that the fresh start effect is particularly influential for individuals with low personal control, boosting their engagement in collective environmental actions. Additionally, a survey demonstrated that the influence of global identity on the relationship between the desire for personal control and environmental consciousness is more pronounced in individuals with a strong belief in the fresh start mindset. These findings suggest that striving for collective goals can act as an extended form of primary control, which can help individuals with a low sense of personal control to restore a sense of agency through group participation, particularly when a fresh start mindset is salient. This research broadens the understanding of the fresh start effect beyond an individualistic framework, showcasing its significance in collective endeavors. It offers novel insights into the interplay between personal motivation and collective environmental initiatives, underscoring the role of fresh starts in enhancing individual commitment towards shared environmental goals.

2 - Regulatory Focus and Eco-Friendly Product Preferences in Mountain and Ocean Environments
Hyunji Lee, Hongik University, Seoul, Korea, Republic of, Shengmin Si , Nara Youn

This research aims to explore the interaction between individuals' regulatory focus (promotion vs. prevention) and their preferences for eco-friendly products associated with specific natural environments (mountain vs. ocean). We hypothesize that individuals with a promotion focus will respond more positively to eco-friendly products linked to mountains, while those with a prevention focus will favor eco-friendly products related to oceans. Additionally, we investigate the mediating role of environmental self-efficacy, defined as the belief in one's ability to achieve specific eco-friendly goals, in these relationships.

Central to our study is the metaphorical association of regulatory focus theory, categorizing individual tendencies into promotion focus (related to growth and improvement) and prevention focus (associated with protection and risk avoidance), with mountains and oceans, respectively. We view mountains as symbolizing ascending actions, success, growth, and challenge, while oceans represent descending actions, uncertainty, and tranquility due to their vastness and fluidity.

Study 1 examines how individuals' regulatory focus (promotion vs. prevention) and the type of natural environment (mountain vs. ocean) affect advertising effectiveness. Promotion-focused individuals respond more favorably to mountain-associated eco-friendly products, while prevention-focused individuals show a preference for ocean-related eco-friendly products. Study 2 further investigates these relationships by manipulating regulatory focus. The results indicate that promotion-focused advertisements are more effective in mountain-related eco-friendly products, whereas prevention-focused advertisements perform better in ocean settings. Additionally, the study confirms that environmental self-efficacy mediates the interaction between mountain versus ocean natural environment type and regulatory focus on advertising effectiveness.
3 - Responsibilizing the Net-Zero Hero? Creation and Implications of a Tragic Subject Position
Tom van Laer, The University of Sydney, Sydney, Australia, Morgan Smith
Marketing relevance: The energy sector is the largest contributor to carbon dioxide emissions. To address the current climate emergency however, energy market stakeholders (e.g., energy providers, nongovernmental organizations, policy makers) try to make individual consumers take responsibility for achieving an overall net balance of zero greenhouse gas emissions.
Purpose, Method, and Findings: The purpose of this research is to understand this process and its implications. The research method is a qualitative narrative discourse analysis of hundreds of public documents by energy market actors. The findings show that market actors try to shape ordinary consumers into “net-zero heroes” with responsibility for emissions reduction but end up creating a tragedy when they translate their collective agenda.
Contributions: These findings have implications for consumer responsibilization specifically and the conversion of agendas into action more generally. Theoretically, this research shows (1) the influence of the translation stage in the agenda-to-action chain, (2) the way market actors form net-zero heroes, and (3) the limited usefulness of the hero narrative. Practically, the research explains the implications of making consumers solely responsible for the emissions reduction problem.
Keywords: consumer responsibilization, metanarrative, net zero, subject position, translation
4 - Exploring Generation Z’s Socially Responsible Consumption: A Revisit of the KAB
Lawrence Choi, Hong Kong Baptist University, Hong Kong, Hong Kong, Kimmy Chan, Henry Fock, Howard Ling

As an expanding segment in the consumer market, Generation Z (Gen Z) is featured by its rising environmental concerns. Utilizing the knowledge-attitude-behavior (KAB) framework, this research investigates and elaborates on the relationships between environmental knowledge, attitude toward socially responsible consumption (SRC), and SRC behaviors through surveys and field experiments across three distinct SRC contexts (Fast fashion, ugly food and food waste, and plastic product consumption) with both cross sectional and longitudinal research design. Our research initially generalizes the baseline KAB model to the context of Gen Z and SRC. More importantly, we reveal three notable extensions of the KAB model. First, we introduce perceived consumer effectiveness (PCE) as an additional antecedent of this model, which helps increase the explanatory power of the current model. Second, we uncover the spillover effects of environmental knowledge, where Gen Z’s prior environmental knowledge positively impacts their present knowledge and subsequently strengthens their attitude toward SRC and SRC behaviors. Lastly, we distinguish between general and specific environmental knowledge, identifying that specific environmental knowledge mitigates the influence of general environmental knowledge on Gen Z’s attitude toward SRC and SRC behaviors. Overall, our research provides insights for marketers who strive to develop strategies to foster SRC with Gen Z.

Keywords: Socially responsible consumption (SRC), Generation Z, KAB framework, General and specific environmental knowledge, Perceived consumer effectiveness

TA19
Meeting Room C4.8
Platforms/Two-Sided Markets: Insights from Structural Models and Experiments 1
Contributed Session
Chair: Levent Uyar, University of Hamburg, Hamburg, Germany
1 - Optimizing Human Capital Signals in Online Contract Markets
Ernan Haruvy, McGill University, Cote St. Luc, QC, Canada, Kim Strunk, Marina Fiedler

Online contract markets enable workers to bid for jobs and firms to access skilled labor. Workers and firms differentiate themselves with price and quality signals, which complement the bids. The purpose of this work is to identify the optimal weights on different communication elements (signals) employed by firms and workers. We do so by conceptualizing the market as consisting of networks with different response functions to signals, and where workers and firms’ decisions pertain to the communication mix that would maximize their match return in these networks. Network analysis reveals that there are indeed overlapping skill networks that respond differentially to the communication mix. The fitted model provides concrete recommendation to platforms on network identification and the mix of weights on signals to recommend to both firms and workers to maximize their match return.

2 - Bidding or Allocation? The Design of Dispatch Systems in the Ride-Hailing Market
Xueli Zhang, Xiamen University, Xiamen, China, People's Republic of, Wei Miao, Junhong Chu
This paper investigates the efficiency of two prevalent centralized dispatch systems in the ride-hailing market: a bidding system where drivers bid for booking requests, and an allocation system where riders and drivers are algorithmically matched. We develop a structural model that incorporates a finite-horizon dynamic oligopoly game for drivers, who make bidding and relocation decisions, and a stochastic arrival pattern for riders, who make discrete choices over the means of transportation. Applying this model to Singapore's taxi market and using the oblivious equilibrium concept to solve drivers' dynamic spatial oligopoly game, we find that the allocation system outperforms the bidding system in terms of both riders' welfare and drivers' earnings. However, this improvement leads to unintended consequences: a reduction in the number of drivers available for street hails and an increase in unmet demand for both street-hailing and booking trips. The study provides valuable insights for optimizing dispatch systems in the ride-hailing industry.
3 - When Consumers Reflect Business Models: The Influence of Digital Platform Remuneration Models on Consumers' Decision Making and Market Outcomes
Levent Uyar, University of Hamburg, Hamburg, Germany, Rouven Seifert, Michel Clement, Sonke Albers, Patrik Wikstrom, François Moreau, Frederik Juul Jensen, Ola Haampland
Many online platforms (e.g., Spotify) charge consumers a monthly flat rate that is then allocated to the suppliers (artists) based on remuneration business models. This research focuses on consumers’ reactions and subsequent market outcomes when consumers reflect on remuneration business models. Online and social media make these models transparent and increase the likelihood that consumers want to know what happens with their subscription fees. Focusing on subscription platforms in the music market, we investigate consumers’ response to two fundamental remuneration models in the market (pro rata versus user-centric remuneration model). We study the impact of remuneration models on consumers’ underlying psychological factors, their choice behavior, and the monetary consequences for right holders (e.g., artists) in three studies. We find that (1) perceived process transparency, (2) user centricity, and (3) distribution fairness, influence consumers’ preferences (Study 1). Additionally, we find that consumers’ choice behavior is substantially influenced by the implemented remuneration model of subscription platforms (Study 2). Finally, we study the respective impact on demand using market data from Deezer and simulate payout scenarios to right holders under different switching conditions (Study 3). Overall, we contribute to the literature with a comprehensive behavioral framework to study the monetary outcomes of consumers’ reactions when different remuneration models are offered. We find that platform managers can use the remuneration model as a strategic instrument to attract and retain customers – with substantial consequences for the artists’ right holders.
TA20
Meeting Room C4.9
Consumer Behavior: Branding 1
Contributed Session
Chair: Sami Al Battashi, RMIT University, 480 Elizabeth Street, Melbourne, 3000, Australia
1 - Shape My Choice: How Iconicity and Nostalgia Marketing Influence Brand Preference
Ashish Sadh, Indian Institute of Management, Indore, Indore, India, Mayukh Mukhopadhyay

Nostalgia, which is the sentimental longing for the past (Huang et al., 2016) is found to influence consumers’ brand connectedness (Tilburg et al., 2019) and their purchase intentions (Xia et al., 2021). The sound-symbolic effect of brand names has been found to trigger various associations in consumers' minds (Motoki, & Pathak, 2022). By tapping into nostalgia and aligning brand names with consumer preferences, companies can foster sustainable connections with their target audience, potentially leading to increased customer loyalty and satisfaction. This study aims to investigate the effects of nostalgia marketing and the Kiki Bouba effect in shaping consumer perceptions of brand preference by providing insights into the interaction of nostalgia that a brand communication triggers and other brand elements. Through a 2x2 experimental design with nostalgia marketing (nostalgic branding vs. contemporary branding) and brand name iconicity (round-sounding vs. sharp-sounding) as independent variables, this study attempts to gauge the brand preferences of consumers. This research will contribute to the field by examining the joint effects of nostalgia marketing and the iconicity effect on brand preference, providing new insights into their interplay and influence on consumer behavior. The findings will have practical implications for marketers seeking to leverage nostalgia marketing and brand names to enhance consumer perceptions of brand preference. Understanding the influence of the Kiki Bouba effect can help marketers craft effective branding strategies that resonate with consumers on an emotional level.

2 - Deciphering the Role of Anthropomorphic Marketing in Consumer-Brand Relationships
Mingjing Zhang, BNU-HKBU United International College , Guangdong, China, People's Republic of, Huimin Wang, Peng Wang

As relationship marketing evolves, businesses are interested in establishing emotional connections with their consumers. Anthropomorphic marketing is thus being increasingly employed to foster brand attachment among consumers. The essence of anthropomorphic marketing lies in imbuing non-human brands with human characteristics, treating non-human brands as entities with emotions and intentions akin to human beings. Albeit a few recent studies have explored the effectiveness of anthropomorphic marketing, it is still in an exploratory stage and the underlying mechanism remains unclear. To address this gap, we categorize anthropomorphic marketing into two types based on the stereotype content model: warm vs. competent anthropomorphic marketing. Meanwhile, we also introduce a pair of dependent variables: brand credibility and brand attachment, to deeply distinguish the relative advantage of warm and competent anthropomorphic marketing, respectively. We hypothesize that warm anthropomorphic marketing is more likely to foster brand attachment, while competent anthropomorphic marketing is more likely to enhance brand credibility. Based on a between-group experimental approach, we randomly assign participants into three groups (i.e., non-anthropomorphism, warm anthropomorphism, vs. competent anthropomorphism) to verify our hypotheses. We contribute to the domain of relationship marketing by exploring the underlying mechanism of anthropomorphic marketing through comparing the differential effects of warm and competent anthropomorphic marketing on brand attachment and brand credibility. Additionally, we aim to provide managerial implications for brand marketing practices. This classification and framework can potentially guide businesses in tailoring their marketing strategies to strengthen brand-consumer relationships.

Keywords: Anthropomorphic Marketing; Warm anthropomorphism; Competent anthropomorphism; Brand Attachment; Brand Credibility

3 - Consumer Responses to Brand Ethicality: The Roles of Brand Credibility, Brand Trust and Emotional Attachment
Tawsif Abrar Faiyaz, North South University, Dhaka, Bangladesh, Mahmud Hassan
In this paper we wanted to understand how brand ethicality influences consumers' purchase intention. In addition, we have looked at the mediating roles of emotional attachment and brand trust and the moderating role of brand credibility. A 2x2 between -subjects experiment was conducted in Bangladesh using student samples (N =252). Results support the proposed direct, indirect and interaction effects.
4 - CANCELLED Brand Coolness: Exploring Consumer's Psychological Needs and Motives
Sami Al Battashi, RMIT, Melbourne, Australia, Ashish Kumar, Kaleel Rahman

Brands strive to stay current with trends and win customers by being “cool”. Recent studies suggest that coolness has become an indicator of brands’ success; however, a deeper understanding of why, how, and when coolness adds value to brands is missing. Consumers seek cool brands not only for their utilitarian values, but also for their symbolic meaning and intangible attributes. For example, consumers gravitate towards Apple to fulfil their need for uniqueness. Yet, what remains overlooked is the role of consumers’ psychological needs and motives in driving cool brand engagement. This study aims to fill this gap by addressing the following research question: what motivate consumers to seek out cool brands? To answer this question, this study uses a Qualtrics-based questionnaire to collect data from a sample of (n= 860) Australian consumers recruited through Amazon’s Mechanical Turk (MTurk). Data are statistically analysed using Structural Equation Modeling (SEM). The preliminary findings suggest that consumers’ needs (i.e., uniqueness and status) and motives (i.e., self-verification, self-enhancement, susceptibility to interpersonal influence, and fear of missing out) have a positive effect on perceived brand coolness, brand love, and purchase intention. These findings offer valuable insights for fostering cool brand engagement.

Keywords: brand coolness, cool brand engagement, branding

*Presenting and corresponding author: Sami Al Battashi, a Ph.D. Candidate at the School of Economics, Finance, and Marketing, College of Business and Law, RMIT University, Melbourne, Australia. Email: s3837900@student.rmit.edu.au

TA21
Meeting Room C4.10
Consumer Behavior: Response to AI 1
Contributed Session
Chair: Fangyu Zhong, Business School, Sichuan University, Chengdu
1 - Human-AI Moral Relation: When Would People Regard AI as Moral Patient?
Mingxia Zhu, University of Macau, Macau, Macau, Matthew Tingchi Liu, Xi Song

Keywords: AI ethics, Human-AI interaction

Artificial intelligence (AI) is rapidly changing the way service encounters take place and transforming the consumers’ overall experience. Although interactions between humans and AI are gradually becoming part of individuals’ everyday life, people’s moral consideration about AI is still unclear. To find out whether people generate moral judgment and ethical obligations when they are conducting immoral behaviors toward AI (e.g., physical violence, dishonesty), we review the influential literatures that directly or indirectly addresses the ethics for AI adoption in the domains of marketing, management and human-AI interaction. Extending prior research, we outline a novel conceptual typology based on the extent to which AI would be considered as an adequate moral patient, that is, an entity with capacity to be a target of others’ right or wrong action as beneficiary or victim. The more people envision an AI as a moral patient, the more likely people would be to project human rights onto AI moral patients and to obey general ethical rules during the interaction with the AI. Therefore, we classified AI into four types of moral patients and demonstrated the behavioral guidelines for each type of AI moral patient, respectively. In addition, we posit a framework encompassing the mechanisms of misbehavior toward an AI from the view of extrinsic cost and psychological intrinsic cost of unethical behavior. Our research offers theoretical and practical implications for regulators, organizations, and firms about developing norms and guidelines for proper management for AI interaction.

2 - Evolution of Service Robots from Mechanical to Humanoid Forms: An Examination between Utilitarian and Hedonic Service Settings
Sisi Lei, Shanghai University of International Business and Economics, Shanghai, China, People's Republic of, Yunqi (Milly) Su, Yiping (Amy) Song, Peng Zuo

In today’s service marketing, a revolution is underway, spearheaded by the application of robotics. This revolution has seen robots evolve from basic mechanical to sophisticated humanoid forms, paving the way for unprecedented service experiences. The shift holds particular relevance in marketing, where understanding the impact of humanoid service robots on customer perception and the consequent business outcomes is of paramount interest. The crux of the issue lies in deciphering how the transition from mechanical to humanoid robots influences customer perceptions, especially how consumer response varies when they have utilitarian versus hedonic motives in service consumption. To address these questions, our research conducted a survey and followed it with a series of controlled experiments, meticulously designed to mirror real-life service scenarios. The results revealed that the degree of “humanness” in a robot significantly alters customer perceptions, with humanoid robots perceived as more likable and intelligent. This perception, in turn, has a direct impact on customers’ purchase intention. Interestingly, our findings suggest that the likability of a service robot’s form contributes more to purchase in hedonic consumption, whereas the robot’s perceived intelligence takes precedence to drive purchase in utilitarian settings. The research aims to fill a crucial theoretical gap by exploring how the evolution of service robots from basic mechanical form to advanced humanoid form impacts customer experiences, especially in different types of consumption. Conclusively, this research provides actionable strategies for marketing practitioners to leverage humanoid robotics to boost customer purchases in services.

Keywords: Service robots, Robot humanness, Hedonic versus utilitarian consumption

3 - How do Consumers Respond to Chatbot Anthropomorphism on a Cosmetic EC Website
Mirai Igarashi, Osaka University, Toyonaka-shi, Oosaka, Japan, Shohei Hasegawa, Hiroshi Onishi, Osamu Sakai

Chatbots have become popular for interacting with consumers visiting websites. Especially, chatbots on e-commerce websites are a useful tool for saving companies’ labors and costs. However, consumers expect chatbots to be functional as conversation partners and want to feel like that they are being treated with care and talking with a person. The review paper by Miao et al. (2021) presented a 2x2 taxonomy of anthropomorphic chatbot design based on its form realism and behavioral realism, and proposed a hypothesis that cartoonish chatbots (low form realism) with human-like fluent interactions (high behavioral realism), such as AI chatbots, will perform better than the other types. Although they expected that the differences between the chatbot’s form and behavioral realism lead customers experience positive disconfirmation and can increase their satisfaction and purchase intentions, the previous studies found mixture of results due to their diverse context and different setting of chatbot taxonomies.

We conduct a field experiment with an e-commerce website of a Japanese cosmetics brand and investigate the effectiveness of the above four types of anthropomorphic chatbots (human/cartoon avatars x courteous/concise interactions) which randomly appear for each customer during a two-month long campaign period. The results show that customers responded differently to the four anthropomorphic chatbots depending on their characteristics. The chatbots with high form and low behavioral realism (human avatar x concise interaction) lead to purchases of campaign products as the short-term effect, but they become diminishing at the long-run. We then also discuss managerial implications for effective usages of anthropomorphic chatbots.

4 - Preference for AI Avatar Versus Human Interpreters in Digital Interpretation Platforms
Fangyu Zhong, SiChuan University, Chengdu, China, People's Republic of

Tourism digitalization leads to the emergence of digital interpretation platforms, reforming traditional interpretation services and marketing strategies. Meanwhile, AI-generated content and avatars also emerge to gradually replace traditional human voice interpretation in such platforms. However, it is not clear how consumers perceive, prefer, and react differently to AI-generated interpretation in digital interpretation platforms.

Through five scenario-based experiments, this study verifies why and how the identity of the digital interpreter (human vs. AI avatar) influences customers' preferences and choices of digital interpretation. Findings from Study 1 confirm consumers' general preference for humans (vs. AI avatars) at the current state of the art. Study 2 further uncovers the mediating mechanism of stereotype perception on this preference, i.e., higher perceived competence and perceived warmth towards human (vs. AI avatar). Study 3 finds a moderated mediating effect of a sense of power on stereotype perception. Study 4 investigated the stereotype reversal effect by manipulating stereotypes and found that it becomes more popular when the AI avatar has a warm image and when the human interpreter has a competent image; After manipulating the stereotype, study 5 further revealed that individuals with high sense of power turn to appreciate AI interpreters.

These findings enhance understanding of AI-generated content and avatars, as well as digital interpretation platforms, and also provide insights for travel companies and digital service providers to design a warm image for AI avatars, rather than just emphasizing powerful capabilities. Besides, marketers should have a more granular understanding of the psychological characteristics of tourists.

TA22
Meeting Room C4.11
Consumer Finance and Lending
Invited Session
Special Sessions
Chair: Pranav Jindal, Indian School of Business, Indian School of Business, 140306, India
1 - Credit Failures and Entrepreneurial Risk Aversion
Przemyslaw Jeziorski, University of California-Berkeley, Berkeley, CA, United States, Ganesh Iyer
This paper examines how adverse experiences of credit use arising post the adoption of a new credit technology impacts entrepreneurial risk aversion. In a randomized controlled trial which deployed a new credit technology to small retail entrepreneurs in Kenya we show that the experience of a failed use of a new credit line significantly increases their risk aversion. By separating out the causal effect of credit from selection effects we show that accounting for selection, especially selection into credit, matters -- those who adopt credit have substantially lower ex-ante risk aversion. Thus, their post adoption treatment effect which increases risk aversion is even more pronounced. Consequently, the more risk loving entrepreneurs in the population adopt the credit, but they also end up becoming substantially more risk averse upon experiencing a credit failure. A heterogeneous treatment effects analysis identifies important demographic moderators and shows that male entrepreneurs who are younger and who run smaller businesses are more likely to adopt the new credit product, but they are also particularly susceptible to credit failures experiences and the resulting increase in their risk aversion. We also show that the increased risk aversion leads to a counterfactual credit adoption of only 29% in comparison to the model's adoption prediction of 42%.
2 - Designing Dealer Compensation in the Auto Loan Market: Implications from a Policy Change
Tat Chan, Washington University, St. Louis, MO, United States, Zhenling Jiang, Yanhao Wei
- We study dealer compensation in the indirect auto lending market, where most lenders give dealers the discretion to mark up interest rates and the markup constitutes a dealer’s compensation. To protect consumers from potential discrimination by this dealer discretion, several banks adopted a policy that removes dealer discretion and compensates dealers by a fixed percentage of the loan amount. We document that this policy decreased (increased) the interest rates for low-credit (high-credit) consumers. However, the market share of these banks decreased (increased) for low-credit (high-credit) consumers — a reversal of the usual demand curve. This reversal highlights the influence of dealers on consumers when choosing which bank to finance a loan. Accordingly, we develop an empirical model that features dealer–consumer bargaining. Our estimation results show discrimination without the policy, as reflected in the systematically lower bargaining power of certain consumer groups. We use the model to explore alternative compensation schemes that remove dealer discretion. We find that a lump-sum compensation scheme obtains the most market share. In particular, this holds true for low-credit consumers, where the lump-sum scheme allows more of them to access lower interest rates. Our study highlights the importance of accounting for the incentives and bargaining power of intermediaries
3 - Costs and Benefits of Buy Now, Pay Later (BNPL)
Preyas Desai, Duke University, Durham, NC, United States, Pranav Jindal
Buy Now, Pay Later (BNPL) is a relatively new payment mechanism which, at the point of sale, allows consumers to pay for a product in interest-free installments spread out over a few weeks/months. We examine costs and benefits of BNPL for retailers and consumers. For consumers, BNPL offers the benefit of additional liquidity for a specific purchase from a specific retailer and that too with a 0% interest rate for the first few installments. In addition, BNPL credit is usually not reported to credit bureaus. On the other hand, the additional credit might lead to suboptimal decisions by the consumers if they underestimate the cost of BNPL. In particular, we find that the reduction in consumers' welfare stems from three different sources - (i) higher product prices, (ii) excessive purchase, and (ii) excessive upgrades to the higher quality product.

For retailers, offering BNPL expands the market by enabling purchases by the consumers who do not have the liquidity to purchase its product. Therefore, a monopolist is always better-off providing BNPL to its consumers. However, ina competitive environment, offering BNPL is a more complex strategic decision for the retailers because they also need to consider strategic reactions from their competitors. We find that a duopoly either of the two retailers might refrain from offering BNPL. This is because when one retailer offers BNPL, the other firm not offering BNPL also benefits from competitive spillovers.

In addition to asymmetric equilibria, we find that there is a symmetric equilibrium which both retailers offer BNPL.

4 - Buy Now, Ponder Later: The Impact of the Textual Analysis of BNPL on Consumer Purchase Decisions
Moonjung Kim, Dongguk University, Seoul, Korea, Republic of, Sunnyoung Lee, Minjun Choi

Consumers today are increasingly opting for Buy-Now-Pay-Later (BNPL) over traditional payment methods like cash and checks. BNPL, a fintech innovation, offers a deferred payment scheme without interest or fees (Di Maggio et al., 2022). However, existing research has primarily focused on specific countries, failing to capture broader market trends and consumer dynamics. Additionally, previous studies have often relied on surveys, overlooking insights from online reviews that directly reflect user experiences. This study aims to explore how BNPL services influence consumer purchasing decisions and psychology, with a focus on online reviews and comprehensive analysis.

This study analyzes online reviews of five BNPL services from Trustpilot.com. The dataset includes 37,240 reviews, collected from 2017 to 2024. Latent Dirichlet Allocation (LDA) is applied to extract topics from these reviews, with correspondence score and perplexity used for topic visualization. The study progresses by employing machine learning models and regression analysis to identify the relationship between topics and sentiment scores in consumer behavior.

This comprehensive analysis, encompassing both LDA and sentiment analysis, has identified four topics with predominantly positive evaluations. Through topic analysis, we aim to determine the impact of specific topics and positive evaluations on BNPL service utilization, particularly in terms of stimulating impulsive purchasing behavior. This study enhances the understanding of BNPL service impacts and facilitates data-driven approaches in future consumer dynamics research. This can assist Pay Later businesses in shaping their strategies for service improvement and influencing consumer behavior.

Thursday, June 27, 11:00 AM - 12:30 PM
TB01
Pyrmont Theatre
Special Session MASS: Generative AI in Marketing Sponsored by Apteco
Invited Session
Special Sessions
Chair: Natalia Stanbury, IAB Australia, Surry Hill, N/A, Australia
1 - Keynote - Koen Pauwels
Koen Pauwels, Northeastern University, Newton, Massachusetts, United States
Keynote
2 - Fireside Chat - Sameer Chopra
Sameer Chopra, Benhamou Global Ventures, Menlo Park, CA, United States
Fireside Chat
5 - Panelist - Sameer Chopra
Sameer Chopra, Benhamou Global Ventures, Menlo Park, CA, United States
Panelist
6 - Panelist - Lan Luo
Lan Luo, University of Southern California, Los Angeles, California, United States
Panelist
7 - Panelist - Koen Pauwels
Koen Pauwels, Northeastern University, Newton, Massachusetts, United States
Panelist
TB02
Meeting Room C2.1
Analytical Pricing Models
Contributed Session
Chair: Maxim Sinitsyn, Northwestern University, 60208
1 - Price Cues vs. Service Cues; Drivers of Retailer's Strategic Choice and Their Impact
Preethika Sainam, Thunderbird School of Global Management at Arizona State University, Phoenix, AZ, United States, Sajeesh Sajeesh

Retailers have many strategic levers to enhance consumers’ perceptions of value for their products/services. Two such levers are providing price cues and service cues. Prior research has shown that displaying a price cue (i.e., a reference price) can create consumer value through the perception of a “deal.” Similarly, retailers can enhance customer value by providing service cues, such as exciting store atmospherics. While the incentives for different retailers to charge different prices are well understood in the literature, little is known about why and when they may differ in utilizing price versus service cues. Additionally, while the prior literature has demonstrated the positive effect of providing service cues and price cues independently to consumers, comparing the two value drivers as a strategic choice for the retailer has not been studied previously.

Using a game-theoretic model and a duopolistic framework, we examine which policy may be optimal for retailers in the presence of (a) competition, (b) consumer heterogeneity, and (c) service costs. Specifically, we identify conditions in which retailers might adopt symmetric strategies and those in which two ex-ante symmetric firms might prefer asymmetric strategies. The analysis suggests that the relative value that consumers place on historical versus current period price cues, the differential effect of transaction utility versus acquisition utility, the value of service to customers, and costs related to service provision represent essential determinants of the choice of retailers’ price versus service cue strategy. Our results can help managers design optimal service levels and pricing in many product markets.

2 - NFT Royalties and Art Market Efficiency
Jieteng Chen, Chinese University of Hong Kong, Hong Kong, Hong Kong, Xinyu Cao, Tony Ke
A non-fungible token (NFT) is a uniquely identified digital asset that utilizes blockchain technology for verification of ownership and authenticity. NFTs have revolutionized the art world by providing a secure and transparent way to authenticate and trade digital artworks. In contrast to the traditional art market, a distinctive feature of the NFT market is that artists are enabled to receive royalties each time their artwork is resold in a secondary market. In this paper, we study the impact of NFT royalties on artists' pricing decisions and the overall efficiency of the art market. We find that if the popularity of artworks is publicly known in the market, royalties diminish the resale value of artworks. This leads to lower transaction volume in both the primary and resale markets, making all market participants worse off. If artists possess superior knowledge about the popularity of their artworks than buyers, a popular artist may set an inefficiently high price to signal his popularity. Royalties benefit the popular artist by mitigating the price distortion in the primary market, but still hurt the transaction volume in the resale markets. As a result, the popular artist's profit first increases and then decreases with the royalty rate, and there exists a unique positive royalty rate that maximizes his profit. The total social welfare may increase or decrease with royalties.
3 - Customer Segmentation and Priority Queues
Qian Wang, Lingnan University, Kowloon, China, People's Republic of

In this study, we examine the impact of customers’ heterogeneous valuation for service quality on a service provider that offers a priority service option with premium quality. We assume customers’ valuation for quality follows a uniform distribution and is unknown to the service provider. The system offers two strategies: (1) a single queue with normal-quality service, where customers are charged a single price, and (2) two queues with different levels of service quality, prioritizing customers in the high-quality queue and charging two prices accordingly.

Customers make their decision on which queue to join based on the publicly announced queue terms before entering. In the system with priority queues, the service provider can charge a higher price to customers with a higher valuation for quality, compensating for their reduced waiting time. By encouraging self-selection behaviour, the service provider can effectively segment customers based on quality and price differentiation. However, in the system with only normal-quality service, customers with a high valuation for quality will experience waiting costs in the queue along with customers who value quality less. The pricing decisions also impact the effective arrival rates of customers joining the queues.

The objective of this study is to investigate the conditions under which quality differentiation with priority queues is more profitable compared to a single price service. By exploring these conditions, we aim to provide insights and guidance to service providers on the optimal strategy for service provision, considering customer heterogeneity in quality valuation.

4 - Licensing Mechanisms for Product Lines
Maxim Sinitsyn, Northwestern University, Cook County, IL, United States
This paper examines the effect of different types of licensing payments on the strategies and profits of contracting parties when a licensee pays royalties on technology it uses to develop a product line. I compare three types of royalties: 1) per-unit royalties; 2) ad valorem royalties; and 3) ad valorem royalties with a cap, where the licensor uses an ad valorem rate but places a cap on price to which this rate is applied. Consumers differ in their willingness to pay for quality, and the licensee can not distinguish among them. I find that per-unit contracts leave the quality levels the same as in the case without royalties, but a larger per-unit rate leads to a decrease in market coverage and higher prices. In contrast, ad valorem contracts lead to lower quality levels. Ad valorem contracts with a cap further deteriorate the lower quality products, but keep the higher quality products at the same quality level as in the case without royalties. For an intermediate dispersion in quality preferences between the consumer groups and sufficient number of low-type consumers, all parties prefer per-unit contracts. For a small dispersion in quality preferences, the licensee prefers an ad valorem contract while the licensor prefers a per-unit contract. These preferences also hold for a large dispersion in quality preferences if there are only two consumer types. However, if there is a continuum of consumer types, the licensor prefers an ad valorem rate with a cap while the licensee prefers a per-unit rate.
TB03
Meeting Room C2.2
Salesforce Management and Personal Selling: AI, Digital Channels and Training
Contributed Session
Chair: Siddharth Singh, Indian School of Business, Hyderabad and Mohali, India
1 - Sales Skill Development and Coaching in a Post Generative AI World
Howard Dover, University of Texas at Dallas, Richardson, Texas, United States

Skill Development and coaching is perceived as costly, yet sorely needed in professional sales. The industry staple of classic skill drills (Role plays, elevator pitches, etc.) has been augmented by the arrival of conversation intelligence platforms which provide analysis and recordings of more sales interactions. Evaluations of these drills and calls could be vital to skill development, but such evaluations often contain various forms of bias (experience, methodology, etc.) as well as lack of focus due to the tediousness of the task. Prior to the advent of Generative Artificial Intelligence (AI) and Large Language Models (LLM), what were the limiting factors of A.I. tools to provide evaluation and coaching? Can generative A.I. overcome some of these limiting factors and increase the efficacy of skill development and coaching? A review of existing technologies and models will be identified. A summary of emerging methods and models will be presented, including testing some models to compare and contrast human evaluations versus A.I. evaluations.

2 - How does Feedback Design Motivate the Next Generation of Salespeople? Theory and Evidence from an Experimental Study
Ying Yang, University of Washington Bothell, Bothell, WA, United States, Xiaofei Pan, Christine Lai-Bennejean

Generation Z, often characterized by their craving for instant gratification, seeks immediate and comprehensive feedback—a challenge that managers may find difficult to meet promptly. How can we effectively address this challenge? Should we strive to provide immediate feedback to meet their expectations? What types of feedback would be most effective in motivating the next generation of salespeople? To answer these research questions, we employed a comprehensive, mixed-method approach to investigate how various feedback designs impact the effort and performance of the younger generation in sales tasks. Drawing insights from interviews with managers, we developed a theoretical model that encompasses three dimensions of performance feedback: 1) timing (immediate vs. delayed); 2) assessment form (summative score without explanation vs. formative with explanation); and 3) comparison (with comparison vs. without comparison). To test theoretical predictions, we desigend laboratory experiements using real-effort sales tasks. Our primary finding reveals that participants exhibit greater effort and improved performance when exposed to comparative information, particularly when the timing of summative and formative feedback is inconsistent. This effect is most pronounced when participants receive their summative scores immediately with a delayed explanation. On a continuum where the timing of summative and formative feedback ranges from most immediate, inconsistent to most delayed, we observe a hump-back shaped pattern in both effort and performance. Our research suggests that a strategic blend of immediate quantitative feedback and delayed comprehensive insights, coupled with intermittent relative performance information, would yield the most motivating and productive outcomes for the new generation of salespeople.

3 - Effect of Product and Sales Skills Training, Engagement with Training, and Training Spillover on Sales Performance
Siddharth Singh, Indian School of Business, Hyderabad and Mohali, India, Raghu Bommaraju, Sharad Borle
The authors use salesperson-level data from an online modular training program of the Indian subsidiary of a global electronics manufacturer to investigate the following: (1) Effect of product-related training compared to sales skill training, (2) effect of engagement with the training program modules, and (3) spillover effects of the overall training and product-related training done by peers, on the number of units of products sold by a salesperson. The authors also analyze the heterogeneity in the above effects. The results show that Product-related training has a stronger effect on sales performance than sales skills training. The engagement with the training (measured via quiz score at the end of each training module) positively impacts sales performance. We find a higher spillover effect on sales when the proportion of product training by peers in the overall training mix is higher. We find that these effects vary significantly across salespersons based on their characteristics, with some salespersons showing a decrease in sales performance after the training, although the average impact of training on sales is positive.
4 - Introducing a Deep Hypergraph-Learning Approach for Modeling Salesperson Networks
Qianyin Xia, Georgia State University, Atlanta, GA, United States, Yi Zhao, Sarang Sunder

In the contemporary business landscape, effective team formation and composition are pivotal in shaping overall team performance, particularly in domains like sales. While prior research has predominantly concentrated on pairwise interpersonal relationships of teamwork, there exists a knowledge gap in understanding how the projects and peers influence an individual agent's ability and subsequently impact team outcomes. This study introduces an innovative deep learning model that predicts team-level outcomes, especially in sales, by considering complex agent-agent and agent-project interactions, while also addressing potential endogeneity. Our approach employs a hypergraph structure to represent relationships between agents and projects, enabling higher-order relationships where a single edge can link multiple nodes. To achieve unbiased sales predictions, we extend the Heckman two-stage framework by integrating a hypergraph representation structure. The main model uses historical data, including agent, project, and network information, to generate informative hyperedge embeddings through hypergraph convolution. These embeddings, combined with the latest year's project data and network structure, serve as the basis for predicting future project outcomes (sales). The selection model leverages historical data to model team formation, determining agent collaboration probabilities based on the latest year's network structure and calculating the inverse Mills Ratio to address endogeneity concerns in the main model. Through counterfactual analysis and simulation studies, our research validates the effectiveness of this model, providing valuable managerial insights.

TB04
Meeting Room C2.3
Frontiers of Preference Measurement with Unstructured Data
Invited Session
Special Sessions
Chair: Jia Liu, Hong Kong University of Science and Technology, Hong Kong, Hong Kong
Co-Chair: Ryan Dew, Wharton, Philadelphia, PA, United States
1 - Adaptive Preference Measurement with Unstructured Data
Ryan Dew, Wharton, Philadelphia, PA, United States
Many products are most meaningfully described using unstructured data like text or images. Unstructured data are also common in e-commerce, where products are often described by photos and text, but not with standardized sets of attributes. While much is known about how to efficiently measure consumer preferences when products can be meaningfully described by structured attributes, there is scant research on doing the same for unstructured data. This paper introduces a real-time, adaptive survey design framework for measuring preferences over unstructured data that fuses representation learning and Bayesian optimization. By adaptively choosing items to display based on uncertainty around a nonparametric utility model, the proposed method maximizes information gain per question, enabling quick estimation of individual-level preferences. The approach operates directly on raw, unstructured data, thereby eliminating the requirement for manual coding of product attributes. We apply the method to measuring preferences over clothing, and highlight its potential both for the general task of marketing research, and for the specific task of designing customer onboarding surveys to mitigate the cold-start recommendation problem. We also develop methods for interpreting the nonparametric utility functions, which allow us to reconstruct consumer valuations of discrete attributes, even for attributes that were not considered or available a priori.
2 - Understanding Consumers Fast: Meta-learned Temporal Processes for Modeling Consumption Dynamics
Mingzhang Yin, University of Florida, Gainesville, FL, United States, Khaled Boughanmi, Asim Ansari
Understanding user preferences from a small amount of individual data is essential for enhancing customer engagement on a digital platform. For a recently acquired customer, the observed interactions with the platform are often limited, which poses challenges to adapting to personal needs and providing targeted products. This is the few-shot learning problem for the customer journey studied in this paper. We introduce the Meta-Temporal Process (MetaTP), a novel approach to rapidly understand and predict new consumers' behaviors. This neural network-based model mimics human learning strategy -- solving new problems by learning to learn from past experiences. Utilizing meta-learning and a Transformer model, MetaTP maximizes the information gains from the initial period of the customer journey and is able to continuously benefit from a stream of incoming interactions. This framework excels at utilizing small-sized data and producing interpretable decision processes. Empirically, we validate MetaTP in the context of digital platforms with music products. MetaTP outperforms a variety of linear and nonlinear, static and adaptive methods in predicting future behaviors from minimal observations. Managerially, it reveals fine-grained, time-varying individual preferences and generates tailored listening sessions on the fly.
3 - The Impact of Facial Femininity and Gender Identity on Perceptions and Behavior: Using AI for Controllable Stimuli Generation
Lan Luo, Columbia Business School, New York, NY, United States, Olivier Toubia
Natural stimuli involving unstructured data like images can be decomposed into many latent features, which may be highly correlated and partially unobserved to the researcher. This makes it difficult to identify the impact of any single feature on outcomes of interest. To resolve this issue, we leverage disentangled representations from generative AI to manipulate realistic stimuli on just a single attribute at a time—an instance of what we call controllable stimuli generation. This paper applies this novel approach to address unique questions in gender discrimination with faces. We use a Generative Adversarial Network (or GAN) to create realistic photos of men and women and then manipulate their facial femininity independent of other features like pose, expression, background, hair style, and clothes. Then, we devise a novel experimental design that allows us to separately identify the impact of facial femininity and gender identity on perceptions (of trustworthiness and competence) and behavior (in an online tutoring setting).
4 - User Preferences for Influencer Debunking
Zijun Tian, Washington University in St. Louis, St. Louis, MO, United States, Boya Xu, Tong Guo
The prevalence of misinformation on social media platforms has underscored the critical role of crowdsourcing in propagating corrections, with influencers increasingly recognized as the pivotal agents in the battle against misinformation. In this project, we explore user preferences for influencer-generated videos regarding controversial topics on TikTok. Specifically, we ask: is it beneficial for influencers to debunk misinformation, and if so what content strategies prove most effective for varying influencer profiles? We present a case study on #greenwash, an emerging topic around the deceptive practice of conveying a false (or exaggerated) sense of environmental responsibility by a company or brand. We document an increasing number of videos on Tiktok related to greenwashing over the past 4 years, and measure audience engagement via major KPI metrics such as views, likes and shares. We further analyze audience responsiveness to the video content by mining the full text of user comments. The results can help various stakeholders (e.g., policy makers, competing firms) understand the role of social media and in particular influencers in debunking misinformation. The study also yields actionable recommendations for content strategies that effectively counter false claims while make the influencers more successful.
5 - Segmenting Consumer Location-Product Preferences for Assortment Localization
Jia Liu, Hong Kong University of Science and Technology, Hong Kong, Hong Kong, Kohei Kawaguchi, Tao Li

When managing multiple stores in the same marketplace, retailers need to select store locations and localize product assortments to reflect the heterogeneous demand preferences across communities. This paper develops a dual Poisson Dynamic System with Multilayer Factorization (dPDS-MF) for panel data on product assortments and individual consumers’ purchases across store/vending locations. The dPDS-MF can help retailers automatically profile different consumer segments driven by store visiting preferences, measure the relationships across store locations, and estimate the product preferences for each consumer segment simultaneously. The dPDS-MF relies on a Bayesian nonparametric prior and can be efficiently trained for large-scale transactional data across hundreds of stores and SKUs, using our proposed MCMC inference algorithm. We apply the dPDS-MF in the retail vending market in major train stations in Japan. We demonstrate the face validity of the direct outputs from the dPDS-MF for improving vending location decisions as well as location-specific assortments. More importantly, we showcase how the dPDS-MF can be combined with a choice model to solve the optimal localized assortments efficiently and effectively. We show that compared with several benchmark strategies, including the nested-logit choice model, our proposed assortment strategy not only improves the expected revenue up-to 30% but also gives more meaningful localized assortment decisions.

TB05
Meeting Room C2.4
Marketing Strategy: Disruptive Technology
Contributed Session
Chair: Hongyu DUAN
1 - The Relationship Between IT Capabilities, Agility, and Organizational Performance: A Meta-Analysis
Jerry Jin, The University of New South Wales, UNSW Sydney, Australia, Ting Yu
The fast-changing environment, especially post-COVID, has impacted customers in changing their demands of products and services fast to meet the new environment dynamics, which in turn places increasing emphasis on agility—a firm’s ability to sense and respond to changes quickly. Despite extensive information technology (IT) resources to pursue agility, firms continue to puzzle over how IT can enhance agility more effectively and efficiently across various organizational levels. Analysis from a dataset of 427 effect sizes from 193 studies (N= 46,494) published between 2003 and 2023 suggested that IT has a positive and significant impact on agility across organizational levels and organizational performance. Specifically, IT integration significantly boosts agility at organizational and functional levels, while IT competency is most impactful at the employee level. Moreover, data analytics technologies strengthen these relationships more than social media technologies, and between-team collaboration is more effective than within-team collaboration in strengthening these relationships. This research offers a comprehensive analysis of how IT impacts agility across various organizational levels and how that agility, in turn, impacts organizational performance by synthesizing existing empirical studies. It extends beyond prior empirical studies by uncovering previously overlooked boundary conditions in the above relationships. This integrated understanding moves scholarship forward. Additionally, the study provides practical insights for firms, their departments, and employees. It also suggests avenues for future scholarly research by connecting the findings to emerging technological trends, such as service robots and Generative AI.
2 - Technological Disruption in the Fashion, High Tech and Packaged Goods Industries: A Quasi-Experimental Analysis
Sajeev Nair, University of Kansas, Lawrence, Kansas, United States, Francesca Bonetti, Gerard Tellis

AI-driven technological innovation is fundamental to growth and success of firms. Yet incumbents repeatedly stumble and fall while new entrants rise. Given that the global fashion industry is worth more than $1.7 trillion in 2023, we examine whether this pattern occurs in fashion markets as it does in high tech markets. We exploit a quasi-experiment, comparing the US Los Angeles market vs. the UK London market, before vs. after shocks (i.e., recessions, Covid) with a long-term horizon, and across industries (i.e., fashion vs. high-tech vs. packaged goods). We employ multiple methods, including quasi-experiment analysis of market data and a cultural survey of managers. We explore returns to R&D vs. advertising over the long run. Our preliminary results show that entrants are gaining on incumbents more sharply in the US than in the UK and after Covid than before. Data suggest that the internal culture of the firm and investment in R&D (vs. traditional advertising) are key drivers. We find only modest support for firm size and bureaucracy. Brand category is a moderator in fashion: apparel and sportswear gain in the US while luxury gains in the UK. The difference is probably due to differences in brand heritage, desirability, and customer loyalty between these two markets. Our work contributes an examination of different and contrasting theories for the fall of incumbents across markets.

3 - The Spillover Effect of Supply Chain Related Blockchain Announcements on Supplier's Market Value
Atya Zeb, University of Adelaide, Adelaide, Australia, Dimitri Simonin, Ljubomir Pupovac, Manjunath Padigar, Mahima Hada

In recent years, there has been an increase in the number of companies engaging in supply chain-related blockchain adoption. However, despite the growing interest and implementation by companies, the impact of such adoption on suppliers remain elusive. This study aims to fill this gap by examining the impact of supply chain-related blockchain adoption announcements (SCBA) by customer firms on the market value of their suppliers. Grounded in signaling, screening, and power-dependence theories, we explore how customer-supplier dependencies characteristics as well as SCBA characteristics influence investor perceptions and the market valuation of supplier firms. Based on the event study analysis of 1234 SCBAs from 2016 to 2022, we find that suppliers’ investors react negatively to such announcements. Based on the insights from the power dependency framework, our analysis suggests that suppliers' dependency amplifies, while customers’ dependency mitigates investors' negative reaction to SCBA announcements. Furthermore, SCBA announcement characteristics, such as blockchain use of sensitive data sharing and partner involvement in the initiative attenuate the negative effects of SCBA. This study contributes to the literature on technology adoption in supply chains by providing nuanced insights into the interplay between signaling innovations and supplier firm valuation, highlighting the importance of supplier power and relational dynamics in the B2B setting. Furthermore, our study offers insights to supplier firm managers on making more informed decisions related to the adjustments of the blockchain related innovation of their customers.

4 - Impact of Disruptive Digital Technologies on Long-Term Firm Performance
Hongyu Duan, Hong Kong Baptist University, Hong Kong, Hong Kong, Danny T. Wang, John Qi Dong, Mengxiang Li

This study delves into the escalating emphasis on disruptive digital technologies such as cloud computing, big data, AI, and blockchain within management practices. Despite their potential in revolutionizing business models and operational frameworks, a substantial 70% of digital transformation initiatives fail (Tabrizi et al. 2019). The research aims to dissect the multifaceted impacts of these technologies on firm performance, particularly focusing on Chinese listed companies. Using text-mining techniques, we analyze longitudinal data (2010–2022) extracted from annual reports to measure the evolution in developing disruptive digital technologies.

Utilizing a mediator model that accounts for endogeneity in a firm’s organizational change, the study reveals that while disruptive digital technologies offer structural integration and revenue diversification, these aspects appear to have negative impacts on a firm’s long-term performance. We show that developing these technologies pays off only in specific competitive environments. This nuanced understanding challenges the notion of a one-size-fits-all approach to digital transformation.

The research contributes to the digital transformation literature by providing an empirical investigation into the performance implications of disruptive digital technologies and detailing the complex nature of their impacts on organizations. It provides a practical framework for managers, aiding in the development and implementation of effective digital strategies, tailored to specific competitive environments. This study underscores the importance of strategic application and understanding of disruptive technologies in maintaining competitiveness and capturing strategic advantages in the dynamic digital era.

TB06
Meeting Room C2.5
Analytical and Empirical Models for CSR
Contributed Session
Chair: Sumitro Banerjee, Grenoble Ecole de Management, 35 avenue des Cottages, GRENOBLE, 38000, France
1 - Stock Market Responses to Negative Word of Mouth: Moderating Roles of Corporate Social Responsibility Initiatives
Jaihyun Jeon, Syracuse University, Syracuse, NY, United States, Guiyang Xiong

Negative word-of-mouth (NWOM) from consumers can have detrimental effects on firms, making it crucial for companies to minimize its negative impact. However, the ways to mitigate the influence of NWOM are underexplored. While firms have the option to directly address NWOM on functional product attributes by improving or modifying related products, we investigate how strategies unrelated to the products themselves indirectly mitigate the negative effects of NWOM. Specifically, this study examines the link between functional NWOM and stock returns, and the moderating effects of different types of CSR initiatives such as environmental sustainability, philanthropy, and sponsorship. Our findings reveal that the moderating effects of CSR initiatives on the impact of NWOM on stock returns are asymmetric and depend on the type of CSR initiative. Specifically, while some CSR initiatives can effectively mitigate the negative effects of NWOM, others may be ineffective or even backfire. Additionally, our study suggests that the moderating role of CSR initiatives can be contingent on firm and industry-level factors. Our study contributes to the literature on word-of-mouth (WOM) and CSR by empirically examining how various types of CSR initiatives moderate the impact of NWOM on abnormal stock returns, offering new insights for marketers and investors.

2 - The Strategic Failure of Climate Targets
Jin Huang, NYU Shanghai, Shanghai, China, People's Republic of, Yuxin Chen, Zheng Gong
In recent years, numerous companies have publicly announced their commitment to reducing greenhouse gas emissions, including pledges to achieve net-zero emissions by specified future dates. However, the latest findings reported in various media outlets have raised concerns that a significant number of these firms are unlikely to meet their stated targets. This paper aims to provide economic insights into the strategic considerations that companies may have when announcing climate targets. We propose a game-theoretical model where companies compete for consumers in environmental sustainability and product price. Surprisingly, when there is a cost associated with failing to meet pre-announced climate targets, in equilibrium, all firms set overly ambitious targets and ultimately fail to meet them. Despite not achieving these targets, the final greenhouse gas emissions are lower in equilibrium compared to the case where companies cannot make pledges. Building on this theoretical framework, we also examine various regulatory policies regarding firms' climate claims.
3 - Consumer Learning of Corporate Social Responsibility: The Hyped and the Understated
Sumitro Banerjee, Grenoble Ecole de Management, GRENOBLE, France
The paper conceptualizes corporate social responsibility (CSR) as a supplementary credence good to products sold by firms in consumer markets. It analyzes a rational learning model of CSR where consumers receive CSR information from two senders: a) independent observers of CSR, and b) the firm. We show that if the independent reports lead to a lower posterior belief of consumers, the firm has incentives to report truthfully but a selection of the private signals it receives; but if the independent reports lead to a higher posterior beliefs consumers, the firm has incentives to allow or add noise to the private signals before reporting them to consumers. In case CSR assumes binary states of nature, we find that a pessimistic firm has incentives to withhold information to prevent a "crash" or sharp drop in posterior beliefs of consumers when the firm re the information is negative. Conversely, optimistic firms have incentives to transmit all the information they have leading to a "boom" or a sharp increase in consumer posterior beliefs about CSR. Hence, despite a rational learning based on observed signals, consumer posterior beliefs of CSR may vary depending on firm incentives to selectively report or withhold private information due to considerations arising from the sale of their products in markets.
TB07
Meeting Room C2.6
Consumer Behavior: Charitable Giving and Donations
Contributed Session
Chair: Natalina Zlatevska, University of Technology Sydney
1 - Ending Temporal Landmarks Increase Charitable Donation
Yuxuan Liu, Tsinghua University, Beijing, China, People's Republic of, Rong Chen

Abstract: Timing is crucial for the effectiveness of marketing campaigns. An increasingly enriched body of literature has investigated how temporal landmarks (e.g., morning, the end of a year) influence consumers' mindset and behaviors. However, limited research has explored how temporal landmarks influence consumers' donation behavior. The authors propose that consumers have a higher donation intention when priming a temporal landmark as the end of a time period than when priming a temporal landmark as the start of a time period. Three empirical studies tested the ending temporal landmark-donation intention effect. In Study 1, an analysis of timing and donation data from over 394,000 records of a charitable organization demonstrated a naturally occurring relationship between ending temporal landmarks and increased donation amounts. Both Study 2 and Study 3 employed a two-group (temporal landmark: ending vs. starting) between-subjects design. In Study 2, the authors used a scenario imagination task to manipulate the temporal landmark framing and demonstrated that ending (vs. starting) temporal landmark increases prosocial value orientation. Study 3 was conducted on May 31st, 2023, framed as either the end of May (ending temporal landmark) or Wednesday (starting temporal landmark). Study 3 demonstrated that ending (vs. starting) temporal landmark increases donation willingness. Our findings contribute to the literature on donation by exploring temporal landmarks as situational factors that influence donation intention. Our research also sheds light on effective charitable communication by providing practical suggestions on choosing the right time to conduct charitable campaigns.

Keywords: Temporal Landmark, Donation, Prosocial Behavior

2 - Too Painful to Donate: Why Don't People Donate to Critical Illness Causes?
Yuanjie Zhao, Peking University, Beijing, China, People's Republic of, Jingjing Ma, Yitian Liang

With the rapid rise of digitalization and social media, charitable donations are moving from traditional channels (e.g., door-to-door fundraising and charitable events) to online fundraising platforms. Unlike traditional channels, these platforms offer NPOs and individuals a place to list their causes next to one another, with limited information displayed for each. Consequently, attracting online traffic to causes on these platforms is challenging, and so guidance for better design of information display is much needed both in the field and academia. Among others, both prior research and practitioners point to the importance of titles, which play a critical role in driving donations. This research examines the impact of titles on online medical fundraising. Specifically, while it is common practice for fundraisers to include the names of critical illnesses in the titles (i.e., CI-title), this research finds such practice can sometimes hurt donation performance on online fundraising platforms. A series of five studies combining secondary, field, and experimental data in both the United States and China provide convergent evidence for this undesirable CI-title effect. Moreover, two studies offer mediation and moderation evidence that this effect is driven by anticipated distress, showing that donors tend to avoid CI-title causes by opting for others that are just one click away. The present research contributes to past research on prosocial behavior and the title effect and offers important practical implications for fundraisers.

Keywords: Fundraising Platform; Title Effect; Critical Illness; Donation; Anticipated Distress

3 - CANCELLED Helping Type and Recipients' Facial Expressions Jointly Impact Charitable Responses
Phyllis Wang, Renmin University of China, Beijing, China, People's Republic of, Michael Jia, Sara Kim, Ce Liang
Recognized as a salient and powerful component of donation appeals, recipients’ facial expressions have been found to significantly influence donors’ charitable responses. However, previous studies have yielded mixed findings regarding the effectiveness of different types of facial expressions in soliciting donations. To enhance our comprehension of this issue and provide actionable recommendations for charitable solicitation strategies, the present research establishes a framework proposing that the impact of facial expressions on charitable responses depends on the specific type of helping (immediate helping vs. autonomous helping). Seven main studies and four supplementary studies (N = 7,239) consistently demonstrate that compared to happy facial expressions, sad facial expressions decrease donors’ perception of the recipient’s resilience. Consequently, this decline in perceived resilience increases donors’ responses to requests for immediate helping but decreases donors’ responses to requests for autonomous helping. The current research contributes to the literature on charitable giving, facial expression, and resilience and offers valuable managerial implications for the effective design of donation appeals.
4 - Understanding the Effect of Psychological Distance in Prosocial Cause Appeals
Natalina Zlatevska, University of Technology Sydney, New South Wales, Australia, Aimee Smith

Recent global events have intensified the pressure on non-for-profit organisations to secure a sustainable future by prompting increased prosocial responses from consumers towards distant causes—those removed from the consumer's immediate context. Despite over 16 years of research, it is unclear whether consumers are more inclined to engage in prosocial behaviour for distant or proximal causes. We investigate how marketers can effectively utilise marketing appeals and position causes to maximise consumer prosocial responses. We conduct a comprehensive meta-analysis, in which we review 216 effect sizes obtained from 116 experimental studies to evaluate the effectiveness of various elements included in these cause appeals. Our findings suggest that neither distant nor proximal cause appeals inherently more effective. Instead, the effectiveness varies based on how the cause appeal's psychological distance from the consumer is established. Specifically, cause appeals depicting temporally distant (future) events prove to be more effective. Additionally, distant appeals are more effective when the cause is related to animals, homelessness, or when the organisation or groups of victims/beneficiaries are highlighted as focal stakeholders. Conversely, proximal cause appeals are more effective when they focus on international issues, utilise branding, suggest specific donation amounts, or explicitly describe how the donation will be used. These findings provide valuable insights for practitioners, guiding them on how best to enhance the effectiveness of prosocial cause appeals.

TB08
Meeting Room C3.3
Customer Relationship Management: Advanced Methods
Contributed Session
Chair: Akihiro Inoue, Keio University, 4-1-1 Hiyoshi, Yokohama, 223-8526, Japan
1 - Large N, Small T, Multiple P: A Causal Matrix Completion Method for CRM Panel Data
Zhongming Jiang, Harvard Business School, Boston, MA, United States, Longxiu Tian, Fred Feinberg

The prototypical customer relationship management (CRM) panel structure is composed of many customers (large N), with short histories (small T), and multiple outcome metrics (multiple P). Our paper aims to tackle the challenges of causal inference that firms face in such CRM settings, which are additionally characterized by unobserved heterogeneity, time dynamics, and staggered adoption. Despite the success of synthetic control methods (SCM) in contemporary marketing applications, extant variants typically necessitate “small N, large T” data regimes to be performant -- e.g., handful of firm- or jurisdiction- level donor units, each with long time-series.

To extend to the “large N, small T, multiple P” setting, we bridge SCM to the broader causal matrix completion paradigm and leverage the “multiple P” ubiquitous to contemporary CRM: the presence of multiple outcomes enables a shared matrix singular value decomposition (cf. SCM's factorization), which helps jointly identify individual-level latent factors to establish conditional ignorability, compensating for overall short time-series at the customer-level. We employ a Bayesian causal inference approach, specifying a joint posterior of the nonrandom missingness of potential outcomes, together with the likelihood of the observed outcomes. The proposed model is estimated via a Hamiltonian Monte Carlo -based data augmentation procedure. We empirically illustrate our approach through a comprehensive customer-level database of gift card purchases and redemptions from a U.S. hospitality startup. We compare the effectiveness with extant SCM and devise a generalized framework for marketing researchers applicable to a wide range of CRM panel structures.

2 - Enhanced Model Coordination Through Multi-Task Learning: An Application to Customer Base Analysis
Kyeongbin Kim, Emory University, Atlanta, Georgia, United States, Daniel McCarthy, Dokyun Lee
A canonical problem within the customer relationship management (CRM) literature is how to predict customer adoption, repeat purchase, and spending given purchase, with customer-based corporate valuation (CBCV) generalizing this problem across behavioral processes to model overall revenue. However, extant work within these streams has several limitations. Existing CBCV literature almost exclusively assumes highly aggregated data are available and for this reason, specifies simple parametric models for customer behavior. Conversely, while existing empirical CRM literature often assumes more granular data are available, they typically model one behavior of interest for a single company or customer cohort. Finally, both literature streams solely focus on point predictions, and validate their proposed models upon a limited number of companies/cohorts, obscuring their empirically generalizable performance. We bridge these gaps through a Transformer-based multi-task learning approach. The proposed method enables information transfer across companies and behavioral processes using the Transformer model as a shared representation, obtaining both point estimates and predictive intervals. Second, through a comprehensive meta-analysis across 1,000 companies, we evaluate the performance of our method against existing probabilistic benchmark models. We achieve better predictive accuracy at the individual process level and in terms of how these processes come together into overall revenue prediction. Lastly, we uncover the source of outperformance for the proposed model by analyzing its predictive accuracy as a function of contextual factors, such as forecasting horizon, industry sector, and other company-specific characteristics, as part of a broader theoretical discussion of the merits of deep learning models for these use cases.
3 - Expanding B2B Customer Churn Model - Identifying Dormant and Churned Customers
Hoying Chiang, National Chung Hsing University, Taichung City, Taiwan, Shihyu Chou, Ming Chih Tsai

Previous research highlights the advantages of employing data mining techniques, especially within customer relationship management, to unveil valuable insights that contribute to customer retention. Traditional churn prediction models often oversimplify the customer status to a binary outcome—active or churned. In reality, companies should also address another critical status: dormant. Dormant customers, though not entirely lost, require strategic intervention, such as reminders for repurchasing. Identifying and engaging with dormant customers allows companies to prevent them from transitioning to the churned status.

This study utilizes data from a Taiwanese manufacturing company. With guidance from the company's experts, customers are categorized into active, dormant, or churned based on predefined rules. The recency, frequency, and monetary (RFM) variables are derived from transaction data and recoded into 5-level factors to represent consumption behavior. To tailor the RFM model to the business-to-business (B-to-B) context, a fourth variable—customization level—is introduced.

Various models, including decision tree, random forest, and multiple logistics regression, are trained to link customer status with the expanded RFM variables. The study compares the classification performance across models, focusing on the identification of dormant and churned customers.

Based on the comparison of classification performance, the most effective model will be suggested to the company for accurately identifying dormant and churned customers. This strategic approach can significantly enhance the company's ability to proactively mitigate potential churn risks.

Index Terms: Churn prediction, dormant customers, machine learning, customer relationship management

4 - An Approach to Develop Scales Simultaneously for Multiple Objects
Akihiro Inoue, Keio University, Yokohama, Japan, Hanako Imai, Naoki Ogawa, Kozue Ogawa
Marketing scales are important in many marketing studies and applications. Most of marketing scales are developed regarding a single object and the method to develop scales is designed for this regard (e.g., Bearden, et al. 1993). However, in some marketing context, we need to build scales simultaneously for multiple objects, such as the attitude toward ads, brands, companies, and so on. In this research, we propose an approach to develop marketing scales simultaneously for multiple objects, based upon the extension of structural equation modeling. Another benefit of the proposed approach is to allow a statistical test on the selection of scale items, either to delete, include, or remain.
We applied the proposed approach to "Hometown Tax Donation Program", a Japanese government policy, of which the purpose is to help the rejuvenation of local areas via co-creation among consumers, local companies, and local government. We derive several important theoretical constructs about relationship management and build the candidate of the scale items for each construct via the past studies and the interviews with consumers, some local companies and local government officers. We run a preliminary survey and the nationwide survey so as to develop the scales of several key theoretical constructs simultaneously for them.
Using the developed scales, we run several field experiments in several areas. We found the significant role of commitment and harmony among consumers, local companies, and local governments and the value of the measures simultaneously developed among them. We will show the results in detail in our presentation.
TB09
Meeting Room C3.4
Privacy
Contributed Session
Chair: Yuan Yuan, Carnegie Mellon University, Pittsburgh, PA, United States
1 - Profiling, Personalisation, Data Analytics and Martech: Developing Frames of Consumer Expectations and the Lagging Response of Privacy Regulation
Peter Leonard, University of New South Wales, Sydney, Australia

The well regarded Australian Community Attitudes to Privacy Survey indicates: (1) consensus for privacy law reform and for tighter rules as to algorithmically based differentiation in marketing offers made to consumers, (2) most citizens believe that ‘my personal information (PI) must not be misused or used against me’, (3) most citizens expect use of their PI as an incident of free online services, and (4) year on year growth in consumer acceptance of targeted marketing.

These findings do not answer four key questions:

Q1: Which practices in collection, use and sharing of PI are a reasonable incident of algorithmic (martech enabled) targeting?

Q2: How does a marketer determine when algorithmic targeting is too opaque, or too targeted, for broad consumer acceptance?

Q3: Should use of effective anonymisation or other privacy enhancing technologies in targeted marketing make a difference? If a marketer doesn’t know who you are, but is able to infer interests or preferences of ‘unidentifiable you', should privacy regulation restrict this targeting, or should non-PI based targeting be addressed by regulation at all?

Q4: Most consumers want transparency and choice, but also want free services and accept that targeted marketing enables free services. Are there genuine choices (e.g., defaults, toggles in targeted advertising settings) that can and should be offered to consumers?

This paper will explore a principled methodological approach to answering these questions, provide examples of how marketing organisations apply those principles, and discuss how changes in regulation of uses of martech might promote adoption of these principles.

2 - Pricing with Differential Privacy
Vanessa Alwan, University of Chicago Booth School of Business, Chicago, IL, United States

With growing privacy concerns among consumers and policy makers, many organizations have started to implement differential privacy (DP) in order to protect individual's personal data. However, adding privacy measures is a concern for businesses given this could affect how they target consumers and reduce profits. Therefore, it is of interest to know if we are able to still price discriminate even in presence of privacy. Firms, including the Census and Google, have concentrated on applying DP to their data to give their consumers privacy guarantees. This motivates us to study how to price with DP data. We propose a method to extract accurate willingness-to-pay estimates using deep neural nets and show how to make inferences even when varying the level of privacy. We additionally show how varying levels of privacy affects different forms of price discrimination. We show that it is possible to still perform some forms of price discrimination and, thus, obtain incremental profits. We demonstrate this with both simulations and empirical applications.

3 - Privacy Risk and Competitive Data Protection Investment with Behavior-Based Pricing
Weining Bao, University of Connecticut, Storrs, CT, United States, Yuxin Chen, Ping Xiao
This paper explores firms’ incentives to invest in data protection to mitigate the risk of compromising consumer privacy, in the context of duopoly competition with behavior-based pricing. We highlight the unique nature of privacy protection as a product characteristic, which distinguishes a firm’s investment in privacy protection from standard improvement in product quality. The analysis uncovers an inverted U-shaped relationship between firms’ provision of data protection and privacy leak risk. We show that firms can be worse-off with a decrease in the privacy leak risk. Our analysis of consumer welfare suggests a free-market solution to consumer privacy protection if the privacy leak risk is low and supports privacy regulations otherwise. An important policy implication is that a moderate regulatory effort to improve consumer privacy protection may backfire and reduce consumer welfare in markets with a high risk of privacy leaks. In the extensions, we show that increase in consumer loyalty or growth of consumer demand may result in reduced investment in firms’ data protection. Therefore, policymakers aiming at enhancing consumer privacy protection should watch closely in such situations.
4 - Clean Up the Act: Impact of Privacy Regulation on Live Streaming Influencers
Yuan Yuan, Carnegie Mellon University, Pittsburgh, Pennsylvania, United States, Nikhil Malik, Wen Wang, Kannan Srinivasan

In the digital age, live-streaming influencers on platforms like TikTok and Douyu.com have seen a surge in popularity. These influencers often strategically leverage a portion of their "privacy assets", such as by revealing private details, to garner rewards like engagement and intimacy. However, this strategy has led to significant privacy concerns and safety issues, prompting policy regulations in recent years. This study empirically investigates such phenomena and quantifies the impact of privacy regulations on influencers. Our analysis, using observational data from Douyu.com and multimodal video analytics, shows that self-disclosure is associated with higher audience engagement for influencers. Leveraging a natural policy shock, we find that regulatory policies on such behavior have led to a notable decrease in audience engagement, with heterogeneous effects across influencers creating different styles of content, suggesting widespread economic losses due to these regulations. Interestingly, midst the general negative effects on the platform, some influencers have proactively adapted by switching to different styles of content creation, resulting in increased engagement. We leverage a game theoretical model to explore the underlying mechanisms and examine the reasons behind influencers’ decisions to change content categories in response to policy shocks. This study provides vital insights into the dynamics of the digital economy, especially in the live-streaming domain. It highlights the significant implications for future policy development, platform management, and the evolution of content creation strategies, balancing live-streaming costs with the pursuit of gains on online platforms.

TB10
Meeting Room C3.5
Movie Marketing
Contributed Session
Chair: Khaled Boughanmi, Cornell University, ITHACA, NY, United States
1 - The Price of Streaming
Franco Berbeglia, Purdue University, Lafayette, Indiana, United States, Timothy Derdenger, Sridhar Tayur
We study the impact of streaming strategies on total movie revenues. Using panel data from 2014-2021 on US box office ticket sales, transactional video-on-demand (TVOD) rentals, digital sales, DVD/Blu-ray sales, and rentals, as well as streaming video-on-demand (SVOD) availability information, we estimate a dynamic structural model of demand across channels. We first analyze a direct-to-streaming strategy implemented by Warner Bros. (WB) and HBO Max: Project Popcorn. We determine that WB was not incorrect with its streaming strategy, notwithstanding the popular view in the media and the industry. We provide quantitative support and qualitative insights on the performance of hybrid strategies: simultaneous release in theaters and streaming, and delayed streaming within the theatrical window. Through ``what-if" analyses our structural model, updating as new data become available, can aid studio executives in crafting their streaming strategy by analyzing the pertinent trade-offs.
2 - Global Cinema Dynamics: Analyzing Non-Tariff Barriers in Film Distribution
Hongye Sun, University of Utah, Salt Lake City, UT, United States, Tianyu Gu

Navigating the complexities of releasing movies in international markets presents unique challenges for filmmakers and marketers. This paper delves into the impact of two predominant non-tariff trade barriers (NTBs) on global box office performance: policy-driven NTBs and cultural NTBs. Our comparative study of Hong Kong and mainland China highlights that release date delays adversely affect box office outcomes in markets with import restrictions (e.g., annual quotas), but not in unrestricted markets. This disparity underscores the influence of policy NTBs within the film industry: extended delays imply escalated financial, relational, and opportunistic costs associated with censorship compliance, which in turn restricts marketing agility and diminishes box office success.

Additionally, we explore the effect of cultural NTBs by examining the 'sentimental distance' between English and Chinese movie reviews for the same films. Our findings reveal a negative correlation between this distance and a movie's global box office success. Essentially, films that resonate similarly across culturally diverse audiences tend to achieve better worldwide performance. Based on these insights, we offer actionable strategies for movie marketers aiming to optimize international film distribution amidst these complex trade barriers.

3 - The Impact of Review Bombing on Film Revenue Across Channels
Paul Parker, University of Kansas, Lawrence, KS, United States, Francisco Zuloaga Cosme, Jonathan Beck

Review bombing, a strategy used by deceptive audiences to manipulate public opinion by flooding online platforms with often ideological motivated reviews, seeks to impact the reputation and financial success of newly released movies. In this research, the authors formally conceptualize and investigate the effects of review bombing on both the domestic box office and video-on-demand revenue of movies. To assess the extent of review bombing on movie performance, this paper utilizes box office, video-on-demand, production budget, and review data from all theatrically released films between 2017 and 2022. Moreover, to identify the movies that experienced a review bomb, the authors develop a measure based on Google search results which circumvents endogeneity concerns related to review bombing counter-strategies employed by platforms. Empirical analysis reveals a significant and positive relationship between review bombing and domestic box office. This positive correlation suggests that review bombing may work as a catalyzer for online word-of-mouth, increasing public attention on a movie; this aligns with the notion that "there is no bad publicity." This review bombing effect, however, does not extend to video-on-demand revenue. Finally, analysis of audience review scores shows that, when controlling for objective entertainment value, ratings for movies affected by review bombing are more skewed toward both the negative and positive ends of the scoring scale. This finding underscores the distinct impact that review bombing can have on public evaluations of films, revealing a noteworthy divergence in audience reactions for movies that have experienced this phenomenon.

4 - Partially Interpretable AI for Market Research
Khaled Boughanmi, Cornell University, ITHACA, New York, United States, Kamel Jedidi
Music, as a highly sought-after product with over five trillion songs streamed in the past year, offers a unique experiential dimension, often tied to emotions and subjective evaluations. Music is also typically consumed in bundled formats like albums or playlists, curated by experts around an overarching theme, making production, sales, and marketing processes costly and prone to failure. Our research employs generative AI (e.g., ChatGPT) to extract experiential features from expert reviews of Rock albums. Through prompt engineering, we identify and score albums on attributes consistently mentioned across all reviews that are relevant to describing album success such as cohesion, emotional impact, and chord progression. We address the inherent uncertainty of the generative AI by employing a diverse range of prompts, which incorporate both human-derived and AI-discovered attributes. Our prompting procedure results in a large number of experiential features. We therefore employ statistical dimensionality reduction techniques to uncover meaningful musical factors that reveal correlations among artistic attributes across various albums. Finally, by regressing the ratings of albums on these experiential dimensions, we show how these musical factors contribute to album success, offering valuable insights to benefit artists and record label firms in designing and delivering creative outputs more effectively. Our initial investigations have yielded encouraging results, with our approach accounting for up to 70% of the variance in album ratings in hold-out samples. We are currently validating our results and comparing the predictive performance of our approach against state-of-the-art natural language processing models.
TB11
Meeting Room C3.6
Online Reviews: Factors Driving Persuasivenes and Helpfullness 2
Contributed Session
Chair: Badr-Eddine Lahsini, University Mohammed VI Polytechnic, Rabat, Morocco
1 - Fine-Grained Attribute-Level Approach for Consumer Reviews: Investigating the Effectiveness of Hotel Service Attributes on Consumer Experience
Vladislava Iakovleva, Pusan National University, Busan, Korea, Republic of, Jiwon Baek, Tae Ho Song

This study proposes a fine-grained attribute level analysis model for evaluating the effectiveness of service attributes on consumer evaluation. We employ machine learning and natural language processing techniques to extract service attributes from online reviews and propose a framework for identifying fine-grained attribute-level sentiment analysis. Attribute-level sentiment scoring is defined as a method of extracting a specific attribute to an associated satisfaction rating. Although numerous marketing scholars have investigated the relationship between integrated sentiment scoring and consumer evaluation, fine-grained attribute level sentiment scoring is still evolving and requires further research. By segmenting each service attribute that constructs the integrated consumer experience, this addresses an opportunity for understanding how each independent service attribute effects consumer sentiment and evaluation. Furthermore, this framework suggests a practical method for analyzing complex online reviews in an easy-to-follow, intuitive manner that allows firms to investigate their service strategy. To illustrate this framework, we provide an empirical analysis by using Yelp hotel review data from January 1, 2018, to December 31, 2021. We adopt a natural language processing method to refine the online review data for analysis. After pre-processing the data, we extract the specific service attributes of online reviews by employing Latent Dirichlet Allocation (LDA) Topic Modeling. Subsequently, each fine-grained attributes are measured based on sentiment scoring. Lastly, an ordered regression model is employed to analyze the relationship between the attribute-level sentiment score and consumer's evaluation on the Yelp platform.

Keywords: Machine Learning, Online Reviews, Text Mining, Consumer Satisfaction

2 - Research on the Characteristics and Differences of Consumer Reviews on Different Network Platforms
Jingying Wu, Fudan University, Shanghai, China, People's Republic of

[Background/Significance] With the development of Internet, the competition among enterprises is becoming increasingly fierce. In order to improve the recognition of consumers, online word-of-mouth marketing has been paid more and more attention by enterprises. According to the characteristics of network platforms, it can be divided into e-commerce, social media and question-and-answer categories. Different network platforms have their own unique positioning and user groups. This study aims to explore the characteristics of consumer comments on different network platforms and whether there are differences? [Method/Process] Take JD.com, Weibo, Wukong platform as the representative platforms, choose Huawei mate40pro as the research object, collect consumer review data, and complete frequency analysis, sentiment analysis, semantic network analysis, and LDA topic model. [Results/Conclusions] The experimental results show that e-commerce, social media, and Q&A review sites have their own characteristics, and there are differences in four aspects: consumer concern, emotional orientation, topic development method, and discussion topic. After comparative analysis, corporate marketers can more accurately grasp the pain points of users, carry out precision marketing for different platforms, so as to further improve the consumer satisfaction of the products.

Key words: Sentiment Analysis LDA Consumer Review Platform

3 - Unravelling the Helpfulness of Online Consumer Reviews: The Moderating Roles of Product Development Stages and Reviewers' Characteristics
Ava Yu, The University of Western Australia, Perth, Australia, Pei Yu Chien, Jacky Mo, Daren Smith

Online review helpfulness has attracted great attention as consumers are increasingly relying on online reviews for decision-making. While it has been established that the helpfulness of a review is influenced by different stimuli, i.e., peripheral and central cues, that are embedded in the message, their respective impact remains equivocal. To shed light on the inconsistent findings reported in the literature, this study investigates product development stages and reviewers’ characteristics as two boundary conditions to explain when and how the effects of peripheral and central cues vary in relation to review helpfulness.

Based on a large, comprehensive dataset that comprises 3,222,761 online reviews for 24 games from a leading gaming platform, this study provides empirical evidence on the impact of peripheral and central cues on review helpfulness. Specifically, the results suggest that peripheral (vs. central) cues have a stronger impact on review helpfulness in the full release (vs. early access) product stage. Moreover, the impact of central cues on review helpfulness in both stages is contingent on the level of reviewer expertise and experience.

Theoretically, this paper contributes to the review helpfulness literature by considering the nuances of product development stages to explain the impact of peripheral and central cues. It further clarifies the differences between two important reviewer characteristics (i.e., expertise vs. experience), and empirically demonstrates their varying effects on review helpfulness. Managerially, the findings are beneficial to both game developers in terms of review site design and game players in terms of better decision-making based on the reviews.

4 - Does Receiving Managerial Responses Influence Travellers' Subsequent Engagement? A Textual Analysis Perspective
Badr-Eddine Lahsini, University Mohammed VI Polytechnic , Ben Guerir, Morocco, Ravi Prakash Ranjan

Travel review platforms have popularised electronic word-of-mouth (eWOM), underscoring the need for hotels to manage and respond to customer reviews for reputation enhancement. Our hypothesis posits that managerial responses (MRs) can shape the subsequent review of returning customers. Utilising TripAdvisor hotel reviews and response data from six Moroccan cities, our study delves into the impact of response characteristics, including response length, speed, content similarity to the review, and degree of topic consistency, on travellers' review change, including review length and ratings. In contrast to prior research that often overlooks the role of reviewer expertise, our study measures the influence of MRs attributes while considering the moderating effect of reviewers' expertise in writing online reviews. To initiate data analysis, we aim to assess whether receiving a response influences future review engagement by categorising returning visitors into those who received a response to their previous review and those who did not. Subsequently, we intend to gauge the impact of MRs on review change using a diff-in-diff design. Furthermore, we employ Text Analysis to quantify response attributes and measure their influence on travellers' review change. At present, we are in the process of analysing the data, with our initial dataset comprising 756 hotels and 9763 returning customers. This research provides a unique contribution to the field of MRs by exploring how returning customers' future actions change with response characteristics to their previous review. Also, the study offers valuable insights into how experienced customers react to manager responses, contributing to the existing knowledge.

TB12
Meeting Room C4.1
Marketing and Policy 1: Grocery Retail
Invited Session
Special Sessions
Chair: Mike Palazzolo, University of California - Davis, Davis, CA, United States
1 - The Effects of Temporary Grocery Store Closures
Julia Levine, Carey School of Business at Johns Hopkins University, Baltimroe, MD, United States, Sylvia Hristakeva
There has been a large uptick in the number of strikes and labor protests in the U.S., with 2023 seeing the most workdays lost due to labor stoppages in the past two decades. In this project, we use retail and household scanner data to study strikes at two large Northeastern grocery chains, affecting just under three hundred stores in total. We study the time-varying effects of the strikes on demand for the striking stores and the spillover effects on demand for non-striking stores of the striking retailer and for the nearby stores of non-striking retailers. We find that even after the strikes have been resolved, sales remain five percent lower for the striking stores of the affected retailer. Although we find effects of the strikes on non-striking stores of the affected retailer during the strike, there are no long-term effects. Persistent effects of strikes on demand for the striking stores (and/or retailers) have important implications for companies and unions entering into contract negotiations.
2 - Retiring the Store Flyer? Effects of Dropping Print Store Flyers on Household Grocery Shopping Behavior
Jonne Guyt, University of Amsterdam Business School, Amsterdam, Netherlands, Arjen van Lin, Kristopher Keller

Store flyers help retailers communicate about the availability, price, and promotions of the products in their assortment. Even in the digital age, many grocery retailers continue to invest in delivering physical store flyers to shoppers’ homes, and store flyers often account for the largest share of grocery retailers’ marketing budgets. However, digital engagement among consumers is constantly increasing, and costs for store flyers have skyrocketed, prompting retailers to re-evaluate their flyer distribution. Some retailers have stopped distributing them altogether, but there is disagreement on the effects on consumer behavior. As such, this study focuses on: (i) What is the effect of a retailer retiring the store flyer on household grocery shopping behavior? (ii) Can a digital alternative (e.g., an app) serve as a replacement?, and (iii) How does the effect differ across primary vs. non-primary shoppers of the retailer? To answer these pertinent questions, this paper looks at the Dutch retail market, where hard-discount chain Lidl ceased delivery of its store flyers in the province of Utrecht at the beginning of 2023. Using household panel scanner data and the recently proposed synthetic difference-in-differences approach, the authors assess the change in grocery shopping behavior of households living in the province of Utrecht along seven relevant grocery shopping dimensions such as the number of shopping trips or total grocery expenditure. With these insights, the paper provides valuable insights to retailers that re-evaluate the distribution of their store flyers.

3 - Taste and Nutrition Information
Beatriz Pereira, Monash University, Melbourne, Australia, Mike Palazzolo
Despite the efforts of policymakers and growing usage of front-of-package and on-shelf evaluative nutrition labels, these simplified nutritional information systems have led to only modest improvements in the healthiness of grocery baskets. A key reason may be that consumers prioritize taste over nutrition. We hypothesize that providing consumers with information about both taste and nutrition can help them satisfy their first priority—taste—and help them make healthier choices from among the alternatives they think they might like. Through a combination of online an in-person experiments, we examine whether presenting nutrition labels with taste information can lead to healthier choices than nutrition labels alone. Importantly, taste information need not be mandated by government policy, as retailers (particularly online grocers) are already providing recommendations to consumers based on their past purchase histories, from which consumer preferences about previously unpurchased products can be inferred.

4 - Price Variation and Store Choice: The Welfare Implications of EDLP-HiLo Pricing
Avery Haviv, University of Rochester, Rochester, NY, United States, Kristina Brecko, Paul Ellickson

This paper investigates the distributional effects and implications on firm profits of two common store pricing strategies: everyday low price (EDLP) and high average prices with deep discounts (HiLo). High variation in prices may disuade low-income households from visiting HiLo stores because they cannot stockpile and take advantage of a low prices, while the possibility of higher prices may result in the household leaving some needs unmet. This leads to low-income households preferring everyday low price (EDLP) stores to those that occasionally provide deeper discounts (HiLo). We document several trends in relation to these pricing trends, and how they relate to store choice for different income groups. We find that consumers mistiming their purchases within a month increases their grocery bill by roughly 9%, and this effect is more pronounced for low-income households.

Moreover, preliminary results show that the product assortment at EDLP stores is less healthful than in HiLo stores, suggesting the correlation between firm’s pricing strategies and assortment healthfulness as one potential source of nutritional inequality widely documented in the literature. Finally, we consider the implications of this sorting on the store’s ability to price discriminate and alternative pricing policies that may improve both firm profits and consumer welfare. Our paper contributes to the rich literatures on store pricing strategies and consumer store choice as well as literature on the causes of nutritional inequality.

TB13
Meeting Room C4.2
Game Theory for Product Offerings and Choice Sets
Contributed Session
Chair: Zhe Yang, Tianjin University, Tianjin
1 - Optimal Design of Recommended Choice Sets
Chi-Ying Wang, Yale School of Management, New Haven, Connecticut, United States

A consumer visits a firm and makes a product query that potentially relates to two product categories. Besides the information provided by the consumer’s endogenous query choice, the firm may have an additional informative signal about the consumer’s true category preference. Facing this dual information source, the firm then chooses between diversifying products across both categories (product breadth) and selecting products primarily from the category predicted by its signal (product depth) in its recommended choice set. We demonstrate that the equilibrium outcome hinges on several key factors: the firm’s predictive accuracy, the strength of the consumer’s outside option, the importance of category preference for the consumer, and the variance in product match values. Specifically, we find that the firm puts more emphasis on product breadth in recommended choice sets under lower predictive accuracy, weaker consumer outside options, and higher importance of category preference for the consumer. The influence of higher variance in product match values on the firm’s optimal recommendation strategy de- pends on the strength of the consumer’s outside options, yielding differential implications for firms with distinct market power levels.

2 - Vertical Line Extensions and Counterfeits in a Status Goods Market
Yuetao Gao, Xiamen University, Xiamen, China, People's Republic of
Consumers may purchase counterfeits instead of authentic products. This is especially true given that consumption of counterfeits can be deceptive as consumers find it difficult to tell counterfeits apart from authentic products based on their similar product appearance. To cope with counterfeits, one of the strategies for an authentic firm is to engage in vertical line extensions by additionally introducing a new upgraded (high-quality) product such that counterfeiters are unable to copy the new product. We develop a game-theoretic model to study the interaction of vertical line extensions and counterfeits in a status goods market. We find that cheaper counterfeits can boost the price and the demand for the new product, which in turn makes the authentic firm earn more profits.
3 - Multistage Joint Selling Strategy for Probabilistic and Deterministic Products Considering Irrational Consumers
Bolin Wang, Shanghai Business School, Shanghai, China, China, People's Republic of, Yi Bu

Existing research on probabilistic product sales focuses solely on decisions related to consumers' rational consumption, neglecting situations in which retailers promote both traditional and probabilistic sales in the context of consumers' irrational consumption. This study integrates consumers' irrational behaviors into utility functions based on game theory. It constructs a two-period model of retailers' probabilistic sales alongside deterministic product sales, exploring dynamic sales strategies. This study also explores the impact of consumers' risk-averse and regret-averse behaviors on retailers' selling strategies. It reveals that retailers are more inclined to use probabilistic product sales when adopting a multiple-product-format sales strategy, and selling both probabilistic and deterministic products simultaneously can significantly increase retailers' profitability in specific situations. Consumers' risk-averse and regret-averse tendencies, however, reduce retailers' motivations to sell probabilistic products and decrease profits. Retailers may then attempt to increase profitability by promoting direct deterministic products, achieved not by raising the probabilistic product's winning rate, but by lowering it to incentivize consumers to choose deterministic products. This study provides a research basis for retailers to develop and implement precise sales strategies, particularly focusing on optimising sales approaches for both probabilistic and deterministic products. It assists retailers in avoiding the risk of cannibalization when implementing diversified sales strategies.

4 - The Dark Side of Personalized Product Recommendation in Online Marketplaces with Sponsored Advertising
Zhe Yang, Tianjin University, Tianjin, China, People's Republic of, Zhaofang Mao, Yunchuan Liu, Jie Miao
In this paper, we examine strategic personalized product recommendation tailored for individual consumers of a platform who has the ability to manipulate ranking lists on organic slots; sellers on this platform can compete for sponsored advertising slots through bidding. To investigate whether and why a platform has incentives to introduce biased personalized product recommendation, particularly when faced with consumers exhibiting horizontal preferences across products, we design a game-theoretic model in which two competing sellers sell products through a platform to heterogeneous consumers with different tastes. The platform applies one recommendation strategy among three strategies, random product recommendation, (not-biased) personalized product recommendation and biased personalized product recommendation, to strategically influence bidding of sellers and products shown to consumers, and finally charges sellers a commission after sales. Our analysis of the interplay between the platform, sellers, and consumers reveals that a platform may strategically benefit from deliberately biasing product recommendation in order to induce sellers to bid higher compared to other recommendation strategies, when the commission rate is low or when the product substitution rate is high.
TB14
Meeting Room C4.3
Consumer Behavior: Advertising 2
Contributed Session
Chair: Yeqing Bao, 301 Sparkman Dr, 35899
1 - ChatGPT Recommended: Product Placement Advertising in AI Knowledge Search
Justin Huang, University of Michigan Ross School of Business, Ann Arbor, MI, United States, Aradhna Krishna
Large language models such as ChatGPT have the potential to massively disrupt consumer knowledge search, threatening to replace search rankings and websites by providing unique, highly contextual answers to user questions and raising questions of how advertisers will reach consumers in this new paradigm. This research examines one avenue for advertising on AI knowledge search: product placement. We present a series of randomized studies assessing its effectiveness, mechanism, and moderating effects of advertisement disclosure and availability of targeting information. Our results show that product placements outperform traditional display advertising in motivating purchase intention, through a mechanism of increased customer awareness rather than shifting brand sentiment. However, inserted product placements come at the cost of lower user satisfaction with the AI knowledge search experience, particularly when the advertising is disclosed.
2 - Open-Mindedness and Consumer's Response to Religious Cues in Advertising: Solving the Paradox
Yeqing Bao, UAH, Huntsville, AL, United States, David Allen, Shi Zhang, Sa'arah Alhouti

The topic of religion in the advertising domain remains largely under-researched, and more work is encouraged (Waller and Casidy 2021). Accordingly, this paper examines and aims to resolve an intriguing paradox found by Minton (2020) who investigated the roles of religious affiliation and open-mindedness in consumer evaluation of advertisements.

Specifically, Minton proposed that “consumers that are high, as opposed to low, in open‐mindedness will have higher product evaluations when exposed to marketing communications with a religious cue (p. 372).” However, two studies consistently revealed the opposite, i.e., consumers high in open‐mindedness responded less positively to advertisements with religious cues. This left a lingering paradox.

The current study reexamines the literature and proposes that the open-mindedness construct actually contains two dimensions, open to persuasion and tolerance of difference. An econometric application model is developed, leading to two research hypotheses:

H1: Consumers that are high, as opposed to low, in openness to persuasion, have lower brand evaluations when exposed to marketing communications with a religious cue.

H2: Consumers that are high, as opposed to low, in tolerance of difference, have higher brand evaluations when exposed to marketing communications with a religious cue.

Empirical data collected via MTurk supported the two-dimensional structure of the open-mindedness construct, and further supported the two research hypotheses. Thus, the paradox is resolved. This finding provides important implications to companies planning to target consumers and promote products via marketing messages with religious cues.

3 - White Noise Makes You Yourself: The Effect of White Noise on Preference
Song Ha Lee, Yonsei Univ, Seoul, Korea, Republic of, Jae Young Lee, Tae-Hyung Pyo

White noise refers to an audible sound comprising a broad spectrum of frequencies at an equal volume, low-pitched sounds, subtle and repetitive sounds. Its continuous and monotonous sound calms the mind, enhancing concentration. Autonomous Sensory Meridian Response(ASMR) would be an example of white noise that are used to induce relaxation in the body and mind. As such, many companies are increasingly utilizing white noise to appeal to consumers. Beyond the feeling of calm and relaxation, we conjecture that white noise can further shift consumer preference. To be specific, the exposure to subtle and repetitive sounds of white noise is expected to heighten the listener's sense of isolation, which subsequently increases the listener's preference towards uniqueness over popularity. To confirm our theorization, we test our hypothesis using the data set collected from an online experiment. The analysis revealed that white noise shift consumer preference such that individuality or uniqueness is prioritized over popularity or agreement among the majority. And white noise heighten the sense of isolation, which in turn drives the shift in consumer preference. The demonstrated white noise effect provides important theoretical and practical implications.

KEYWORD

white noise, ASMR, sense of isolation, unique product, popular product, consumer preference

TB15
Meeting Room C4.4
Reshaping Retail Analytics: Habits, Patterns and Narratives Impacting Consumer Choice
Invited Session
Special Sessions
Chair: Tirtha Dhar, University of Guelph, Guelph, ON, Canada
1 - Changing Patterns: (In)consistent Purchasing of Food Items
Demetrios Vakratsas, McGill University, Montreal, QC, Canada, Wei-Lin Wang
Consistency is an important characteristic of both favorable and unfavorable behaviors with managerial and public policy implications. The authors introduce a methodology, based on statistical pattern recognition and the Bhattacharyya Coefficient, for testing whether households exhibit consistent (single pattern) or inconsistent (dual pattern) purchasing behavior. They then assess the economic value of behavioral consistency via an application focused on the purchasing of convenience foods, which represents both favorable (from a retailer perspective) and unfavorable (from a public policy perspective) behaviors. The method classifies households into three segments characterized by their purchasing pattern: a) consistent; exhibiting a single pattern, b) delayed; exhibiting a dual pattern with delayed purchases, and c) accelerated; exhibiting a dual pattern with accelerated purchases. Consistent behavior is associated with higher customer value; specifically consistent households can be up to two times more valuable than the rest of the households and account for the majority of customer equity. The implications of the findings for both marketers and policy makers are discussed.
2 - Bad vs Good Habits: Investigating the Role of Different Types of Habits on Consumers' Response to Healthy Eating Nudges
Fares Belkhiria, McGill University, Montreal, QC, Canada, Necati Ertekin, Mehmet Gumus, Laurette Dube
An important question in consumer research and public health is why persuasive interventions aimed to improve people’s food choices (e.g. healthy-eating nudges) continue to report modest success, resulting in significant losses in taxpayers money, and an ever increasing obesity epidemic. We argue that this is due, at least in part, to persistent shopping habits that obstruct shoppers deliberation, which results in shoppers' resistance to those persuasive interventions. Using longitudinal loyalty program purchase data obtained from one of largest retailers in Canada, we draw from neuroscience literature to characterize 71,749 transactions on item-level habit strength based on time stability, location stability, and package and price stability. We then leverage a retailer’s nudge implementation as a natural experiment to show that habit strength reduces habitual shoppers' response to the nudge. First, we document particularly strong stickiness in purchases that are primarily unhealthy, i.e. Beverages, Fats and Oils, Snacks and Sugars and Sweets. Then, using a triple-difference (DiDiD) design, we demonstrate that shoppers who are habitual have lower responsiveness to nudges, and spend less on promotional products than non-habitual shoppers. Our results remain statistically significant in several robustness tests. These preliminary findings support our hypothesis that most habitual customers are conducting their shopping episodes with minimal deliberation. They further indicate that unhealthful purchases are especially sticky and habitual. As public health interventions continue to fail in making impactful improvement in people’s diet, our results suggest that shoppers are in minimal deliberation when conducting such purchases and could therefore be “immune” to such interventions.
3 - Word of Mouth as A Network of Narratives: A New Framework to Study Public Discourse’s Impact on Product Performance
Tirtha Dhar, University of Guelph, Guelph, ON, Canada, Poorya Selkghafari
Since the introduction of the word of mouth (WOM) concept to marketing literature in 1957, this field of inquiry has developed into an important topic among scholars, with most of the focus on micro-level analysis and WOM’s psychological effects. Less attention has been paid to dynamic impacts, as scholars consider WOM as an isolated event, overlooking the context that it happens in, and usually track its intensity by downloading the product or brand’s search volume from Google Trends, or counting the number of tweets, comments and blog posts with product’s name. In light of Shiller's (2017) narrative economics, that focuses dynamic relationships between social discourse and market outcomes, this article assumes that WOM comprises a set of narratives, each having a(n) direct/indirect causal interdependencies with product development and performance. To develop the causal interlinkages between narratives and products development and performance, in the context recent explosive growth of gluten free products, we employ convergence cross-mapping combined with a nonparametric Bayesian approach called Gaussian processes to simultaneously account for inherent nonlinearity and endogeneity in the WOM and product performance interaction. Our approach as described can help academic researchers and practitioners generate insights based on dynamic interdependencies between narrative or WOM and product development and performance. Practitioners can also benefit from the framework’s capability to identify search queries, keywords, and narratives with the highest impact on their product which can aid them in making more informed decisions for marketing campaigns.
4 - Modelling Commercial and Societal Drivers and Impacts of Consumer Food Choice within a Multi-layered Digital Ecosystem
Catherine Paquet, Université Laval, Quebac, Quebec, Canada, Antonia Gieschen, Fares Belkhiria, Raja Sengupta, Laurette Dubé
Consumer food choice is a complex phenomenon influenced by several factors, including individual, social, commercial and environmental elements. To better understand consumer decisions in context, we have developed a multi-layered digital ecosystem, Syntheco, which creates a statistically representative synthetic population in which individuals are characterized with respect to socio-demographic, financial, behavioral, psychosocial and environmental attributes using a range of data sources including cohorts and cross-sectional surveys, census, loyalty programs and geolocated business information. The synthetic population, created through iterative proportional fitting from census data, is geospatially located which allows to consider the local availability, diversity, and accessibility to a range of food outlets and their behavioral and health correlates, while considering a variety of moderating influences. Other examples of applications include capturing variations in healthfulness of local food supply to identify areas with limited healthy options (food deserts) or saturated with unhealthy options (food swamps), simulating the impact of food retailing interventions, assessing the ecological footprint of local food purchasing patterns, and evaluating economic disparities in healthy food availability through the use of national financial capability surveys. The platform therefore represents a comprehensive tool enabling the application of computational modelling to study food choice behavior, offering valuable insights for policymakers, researchers, and industry stakeholders interested in promoting affordable, sustainable, equitable, and healthier food choices and addressing food-related health disparities.
TB16
Meeting Room C4.5
Livestreaming 2
Contributed Session
Chair: Ivy C. Dang, University of Hong Kong, Hong Kong
1 - Value of the Seller Agency: Evidence from the Live Streaming Platform
Shuwen Jiang, University of International Business and Economics, Beijing, China, People's Republic of, Hongju Liu, Qiaowei Shen
Seller agencies, which are traditional organizations that represent and support individual sellers, play an important role in service industries such as real estate, insurance, and entertainment. With the emergence of social commerce and live streaming, platforms have increasingly adopted online seller agencies to help individual sellers establish connections through more flexible forms, such as seller groups or communities. Notable examples include Facebook Groups and Etsy Teams. On one hand, the agency could provide valuable resources, such as skills and technics to improve products, peer learning, and the ability to reach potential customers. On the other hand, sellers within the same agency face intensified competition as they share similar resources. To investigate the net impact of joining a seller agency on streaming performance, we employ a difference-in-differences methodology utilizing the unique data from a leading live streaming platform. Our empirical results show that joining an agency leads to improved performance in terms of income-related metrics, popularity, new customer acquisition, and longer work durations. Importantly, we find that the net positive effect is primarily driven by accessing a more matched customer base through the agency, rather than improvements in streaming content quality resulting from knowledge spillover. Furthermore, we explore heterogeneous effects to determine the optimal combination of individual sellers and agencies that yield the greatest benefits. This research contributes to the existing literature on seller/firm networks and provides managerial implications for online small businesses and live streaming platforms.
2 - A Deep Dive into Live-Streaming Sales Pitch
Yanwen Wang, University of British Columbia, Vancouver, BC, Canada, Zining Wang, Shuai Yang, Hongju Liu

This research aims to unravel the mystique of live shopping. While live-streaming technology has reshaped the retail industry, the dynamics within live channels remain largely unobserved, presenting a challenge in accurately evaluating the performance of live-streamers. To bridge this knowledge gap, we leverage a unique live-streaming dataset to characterize and investigate the sales impact of communication strategies employed by live-streamers. Specifically, we focus on three types of sales pitches: product, promotion, and customer relationship management (CRM). We develop an automated, deep-learning-based framework to extract granular textual, visual, and vocal features from 3,057 live product videos broadcasted by 111 live-streamers on a leading online marketplace in 2020. Our findings indicate that the usage of product sales pitches significantly impacts live-streaming sales, such that on average, a 10% increase in the ratio of product-related content to CRM-related content triggers an 8.9% increase in live-streaming sales. This sales lift of product-related content, however, wanes as the follower size of live-streamers decreases. In contrast, the sales lift of promotion-related content is contingent on discount depth, such that promotion sales pitches boost sales only given the presence of substantial discounts. Our research takes the initiative to evaluate the communication strategies utilized in live commerce.

3 - Consumers' Visual Behaviors on Livestreaming Shopping Videos through Eye-Tracking and Machine Learning
Ming Chen, University of North Carolina at Charlotte, Charlotte, NC, United States, Keran Zhao, Wael Jabr, Shi Wang, MiHyun Lee

Live streaming videos have gained popularity as a prevalent online shopping platform for consumers. In a typical live stream, influencers synchronously showcase and endorse featured products while engaging with the audience in a shared virtual space. Although live streaming appears to capture more consumer attention, certain retail platforms still offer asynchronous videos (e.g., Amazon’s recorded videos). An important empirical question remains unanswered: what is the role of synchronization of online shopping video design in shaping consumers’ attention and interest? This study draws on data from a unique platform, Amazon Live, to examine potential differences in consumers’ shopping interest and purchase intention when they are exposed to two different settings: synchronous vs. asynchronous videos using eye-tracking experiment. Specifically, the authors apply machine learning object detection method (YOLO) to identify and extract consumers’ real-time visual fixation counts (FC) and fixation duration (FD) on various ad elements including influencer, featured product, price and chat boxes. With a set of control on the video content, the preliminary results reveal counterintuitive findings: (1) consumers tend to show higher interest in product from the asynchronous video rather than the synchronous video, (2) consumers’ time spent on watching synchronous video content amplifies the purchase intention compared to asynchronous videos, and (3) the chat box for the synchronous videos may distract users, leading to relatively less interests toward products. This study contributes to the design of live streaming content and offers managerial implications for streamers and platforms.

4 - A Structural Model of Shoppertainment Salesforce in Live-Commerce
Chu (Ivy) Dang, University of Hong Kong, Hong Kong, Hong Kong, Arwen Zhou

Live-commerce, also known as “shoppertainment commerce”, combines shopping and entertainment by integrating live streaming, social media, and e-commerce. At the core of “shoppertainment” are influencers who showcase and promote products during live shopping sessions, balancing between “selling effort” and “entertaining effort.” As influencers play a critical role in engaging and converting viewers, it is essential to understand their effort allocation strategies. We develop a structural model based on the salesforce literature which models the effort allocation decisions of the influencers, who are also the salesforce in the live-commerce business model. The model captures the multitasking nature of “shoppertainment.” Moreover, we conduct counterfactual analysis to examine how sales incentives and engagement metrics impact the effort allocation strategies, and how to design optimal incentive and compensation schemes for influencers. The findings shed light on the optimal design of incentive and compensation structures for influencers in the shoppertainment industry.

TB17
Meeting Room C4.6
Health Care Marketing: Patient Outcomes 2
Contributed Session
Chair: Xiumei Tang, West China Hospital, Sichuan University, 37 Guoxue Road, Chengdu, 610041, People's Republic of China., Chengdu, 610041
1 - Engagement and Well-Being as Outcomes of Perceived Social Support in Digital Health Services: A Meta-Analysis
Karen Tian, The University of New South Wales, UNSW Sydney, Australia, Ting Yu, Ko de Ruyter, Debbie Isobel Keeling

Rapid advances in digital health options as potential sources of socio-emotional support are reshaping consumers’ healthcare experiences. In response, a growing body of literature addresses how healthcare consumers’ perceptions of the social support provided by these digital health options affects their engagement with these services and their own subjective well-being. However, to date, it offers fragmented and sometimes contradictory findings. In an attempt to establish a systematic, unified understanding of consumers’ perceived social support in digital health services (DHPSS), this meta-analysis reviews 270 effect sizes drawn from 51 primary studies across multiple disciplines, such as marketing, information systems, medicine, and clinical psychology. The results suggest that DHPSS has a significant and positive impact on healthcare consumers’ engagement with digital health services, which in turn has significant and positive impacts on their subjective well-being. Specifically, the impact on engagement is more pronounced when the perceived social support is specific (i.e., informational, emotional, or esteem) as opposed to general. In addition, DHPSS more strongly improves digital health engagement when the digital health service allows consumers to interact with the healthcare professionals directly (vs. not). The synthesis of existing research findings quantifies the overall impact of DHPSS on healthcare consumers’ engagement with digital health services, and then the impact of that engagement on consumers’ subjective well-being. An integrated understanding that spans all these disciplines moves scholarship forward and enables better decision-making by healthcare providers.

2 - Enhancing Patient Engagement and Adherence in Healthcare: The Role of AI-Driven Pre-Consultation Interactions
Hamid Shirdastian, Bishop’s University, Sherbrooke, QC, Canada, Behzad Seyfi

The interplay between Large Language Models (LLMs) and the doctor-patient relationship is an evolving domain of inquiry. However, there has been insufficient research exploring its potential to revolutionize healthcare delivery and improve patient outcomes. Through a rigorous investigation, this study aims to contribute to the healthcare marketing literature by establishing a theoretical framework for AI-mediated communication in healthcare settings. More specifically, we investigate how consistency (vs. inconsistency) of chatbot responses with the physician treatment plan as well as chatbot's response depth (detailed vs. shallow) changes patients’ intention to engage in treatment and their adherence over time. In a 2×2 experimental design, participants engage in a chat with DocSplain, an AI Doctor. Subsequently, they visit an online doctor to receive the doctor's diagnosis and treatment plan. We find that the combination of high consistency and detailed information yields maximum patient engagement and treatment adherence. On the other hand, shallow information coupled with high consistency serves as the minimal threshold for effective patient interaction with AI in healthcare. It also shows the mediation effect of affective and cognitive responses. We shed light on the role of tailored, AI-driven pre-consultation interactions impact on priming patients for more effective healthcare encounters, potentially reducing the cognitive load and emotional stress associated with medical visits. By uncovering the underlying mechanisms of AI's impact on patient behavior, the findings are poised to inform the integration of AI solutions in healthcare systems and more effective marketing strategies in healthcare industry to intricate human communication and trust-building.

3 - Source Credibility and the Adoption and Valuation of Risk-Mitigating Services
Vedha Ponnappan, Indian Institute of Management Udaipur, Udaipur, India, Anima Nivsarkar, Prakash Satyavageeswaran, Sundar Bharadwaj

Cervical cancer (CC) claimed 342,000 lives among women in 2020. Despite its preventability and curability, CC screening (CCS) remains underutilized. In this study, the researchers conducted a field experiment on CCS to assess the comparative effectiveness of two approaches to conveying risk-related information: one delivered by a medical expert (viewed as an authoritative, informative source) and the other by a peer (considered similar and conforming to societal norms) relative to a control group exposed to an infographic video. The findings from this field experiment in India indicate that both the consumer adoption of CCS and its valuation, measured as incentive-compatible willingness to pay (WTP), are highest when the message is conveyed by a medical expert. A parallel mediation analysis finds two key aspects of risk perception, namely, the perceived susceptibility to and severity of the risk, play essential roles in influencing individuals’ WTP. A causal forests-based analysis identifies heterogeneity in treatment effects due to demographic factors and perceptual beliefs. Employing credible experts doubles adoption (from 4.35% to 9.83%), potentially translating to 25 million more women undergoing CCS in India alone and increasing valuation by INR 534 enhancing the sustained viability of CCS testing for an economically constrained public healthcare entity.

4 - Influence of Marketing Communication on Awareness and Engagement of Patients in Online Health Communities: Online Field Experiment
Myrthe Kuipers, University of Amsterdam, Amsterdam, Netherlands, Umut Konus

The health care market significantly changed over the past years with many disruptive actors, such as online health communities (OHCs), entering the market. The complexity of understanding how marketing communication affects patient health behavior has led to diverse research outcomes, highlighting the need to identify effective marketing communication strategies to raise awareness and engage patients within OHCs. This online field experiment investigates the causal impact of marketing communication elements (emotions, topics, appeal and linguistic style) on awareness and engagement of users on Facebook, Instagram and an online health collaboration platform for heart patients. Various predictive models analyze online user behavior, focusing on awareness (Click-Through-Rate) and engagement (average engagement time and community subscriptions), while accounting for situational and temporal factors. These models utilize datasets that include 2,513,911 impressions and 20,123 platform visitors. The findings highlight both the independent impacts of communication elements and their interaction effects, revealing that effective strategies for awareness may not necessarily translate into successful engagement. Notably, fear-inducing content and topics associated with self-protection are effective for generating awareness, whereas topics like affiliation and kin care emerge as the most influential predictors for average engagement time on the platform and increased community subscriptions. Our study contributes to the literature by investigating the real-time interaction between marketing communication and online patient awareness and engagement, providing both theoretical and practical implications for effectively reaching and engaging patient audiences in OHCs.

5 - A Study on Marketing Strategies for Comprehensive Lung Cancer Management Services Based on Smart Healthcare
Xiumei Tang, West China Hospital, Chengdu, China, People's Republic of
In recent years, smart healthcare technology has made significant strides, presenting promising opportunities for enhancing the management of lung cancer.
Methods:
This study adopts a mixed-methods approach, integrating literature review, service model design, market research, and marketing strategy formulation. Data collection involves systematic literature review, interviews with healthcare professionals, and surveys targeting potential users and medical institutions.
Results:
Smart healthcare technology, encompassing remote monitoring devices, mobile applications, and artificial intelligence algorithms, has notably enhanced the effectiveness and efficiency of lung cancer management. Specifically, it has facilitated early detection, personalized treatment planning, and remote patient monitoring. We have developed comprehensive lung cancer management services leveraging smart healthcare technology and assessed their utilization and effectiveness across the entire lung cancer management process (diagnosis, treatment, rehabilitation). Furthermore, tailored marketing strategies have been devised to promote adoption and gather feedback. These strategies focus on promotional activities, pricing policies, and channel selection to attract more patients and medical institutions to adopt and utilize the comprehensive lung cancer management services.
Conclusions:
Overall, the research underscores the significant impact of smart healthcare technology on improving lung cancer management and emphasizes the importance of tailored marketing strategies in fostering the adoption of comprehensive management services. The findings underscore the necessity for continuous evaluation and refinement to ensure the provision of high-quality care to lung cancer patients.
TB18
Meeting Room C4.7
Consumer Behavior: Promoting Specific Sustainable Behavior
Contributed Session
Chair: Jiming Wu, NEOMA Business School
1 - The Effect of Package Size on Eating of Restrained Eaters
Haruka Kozuka, Seikei University, Tokyo, Japan

The purpose of this study is to determine how the package size of the food offered influences the consumption amount of food by restrained eaters.

Consumers believe that smaller packages containing smaller amounts of food help them to reduce their eating. In particular, the tendency is stronger among restrained eaters, who also believe that they should avoid eating from large packages containing large amounts of food in order to control their eating. However, the amount of food consumed by restrained eaters is smaller when they eat from a large package than when they eat from a small package of the same total amount of food, suggesting that the small package is not useful for their restraint.

To understand why this phenomenon occurs, this study was tested using data collected in laboratory experiments. A mediation analysis revealed that package size affects the actual amount of food consumed by restrained eaters by influencing the amount of food they consider acceptable to eat. The results also indicated that the larger packages contributed to reducing the amount of food consumed by restrained eaters by making them evaluate the amount of food they could tolerate as smaller than the smaller packages. The results of this study provide suggestions on how to provide food products that enhance the wellbeing of restrained eaters.

2 - New Nutritional Label Effectiveness: The Case of the Inclusion of Added Sugar Daily Value on the Nutritional Facts Panel
Hamzeh Rayej, Concordia University, Montreal, QC, Canada, Kamila Sobol, SunAh Kim

Added sugars are linked to significant health issues, including obesity, type 2 diabetes, and cardiovascular disease. Despite health risks, the intake of added sugars still surpasses recommended levels (USDA, 2020). In 2016, the FDA mandated a nutritional label change, separating added sugar details – i.e., grams and %daily value (DV), from total sugar content on packaged foods to raise awareness about the prevalence of added sugars in foods. This study uses a multi-method approach to examine the intervention’s impact on soft drink sales. In study 1, using a difference-in-difference design based on the staggered adoption of the new nutritional facts label among established soft drink brands, our findings suggest that the effectiveness of the labeling intervention is contingent on the extent to which the product’s added sugar content exceeds 100%DV. Notably, significant sales drops were evident only for larger-size soda bottles that contain more than the recommended daily amount of added sugar. In study 2, an experimental investigation replicated the observed findings, confirming that added sugar information significantly decreased product attitude and purchase intention only when %DV on the label exceeds 100%. In study 3, the underlying mechanism was examined, centering on anticipated guilt. Specifically, surpassing the recommended daily allowance of added sugar raises awareness of acting inconsistently with health goals, prompting anticipated guilt and thereby motivating healthier choices to alleviate psychological discomfort. The findings suggest that adjusting %DV by increasing serving size on labels could be a compelling intervention to reduce overall added sugar consumption.

3 - Guilt-Driven Persuasion in Sustainable Fashion: The Roles of Empathic Trust
Hanna Shin, Hongik University, Seoul, Korea, Republic of, Yan Li, Nara Youn

This study investigates how trust in ethical commitments, specifically related to animal rights, amplifies the persuasive impact of advertising by evoking guilt. This effect is particularly prominent among consumers with high levels of empathic concern and those with prior contributions to animal welfare causes.

Our five studies offer robust empirical support for our hypotheses. In Study 1, we investigated how an increase in consumers' perceived trust in a brand's commitment to animal welfare leads to a corresponding increase in their feelings of guilt. This heightened guilt, in turn, enhances the effectiveness of advertisements. We consistently replicated these findings across all three different stimuli used in Study 1, demonstrating their robustness. In Study 2, we applied the insights gained from Study 1 by manipulating brand trust in the stimulus. Furthermore, we confirmed the moderating role of empathic concern. Participants encountering an advertisement emphasizing a strong commitment to animal welfare exhibited higher levels of guilt, particularly when their empathic concern was elevated. Subsequently, this increased guilt contributed to greater ad effectiveness. In Study 3, we shifted our focus to actual user engagement, specifically examining the number of "Likes" received by brand trust message posts on Instagram, as a realistic experiment in the field. The results revealed that trust-based advertisements evoked more guilt, which, in turn, prompted consumers to click the "Like" button on Instagram. Additionally, consumers' past experiences with donations positively moderated the relationship between guilt and "Likes" for advertisements that emphasized trust in a brand's commitment to animal welfare.
4 - Beyond Beauty: Exploring the Primacy of Freshness in Consumer Selection of Natural Fresh Foods
Jiming Wu, NEOMA Business School, Reims, France, Yiping (Amy) Song, Peng Zuo, Kexin Yan

A healthy diet is essential for good health and nutrition. According to Health Halo Effect, people often associate visually appealing foods with healthiness, freshness, and higher quality. Moreover, previous research has proved that consumers tend to perceive prettier foods as healthier as they possess classical aesthetical features that make them look more natural. However, for natural fresh foods, we propose that consumers prioritize freshness before prettiness in their decision-making process. In this research, we investigate the intricate interplay of food freshness, prettiness, health consciousness, and perceived human care in shaping consumer choices for natural fresh foods. Consumer judgments for appearance are often intuitive, influenced by immediate visual cues and aesthetic appeal. On the contrary, freshness is a multi-faceted concept that encompasses factors such as taste, texture, and nutrition value, which requires a more deliberate evaluation. It tends to be more rational and cognitive process when it comes to assess food freshness. Findings from a series of studies reveal that consumers prioritize freshness over prettiness, a preference heightened by increased health consciousness. Prettier foods, often associated with greater human care, paradoxically lead to perfections of reduced freshness. The study underscores the importance of better product positioning and recognizing the nuanced role of human care in shaping consumer perceptions of natural fresh foods. Moreover, it emphasizes the distinct processing pathways that consumers engage with when evaluating food prettiness and freshness.

Keywords

Food and health decision making, perceived aesthetics, health consciousness, human care

TB19
Meeting Room C4.8
Platforms/Two-Sided Markets: Insights from Structural Models and Experiments 2
Contributed Session
Chair: Xinbo Wang, 36 College Avenue East, 138600
1 - Participation Inequality in Online Content Contribution and Platform Incentives
Shibo Li, Indiana University Bloomington, Bloomington, IN, United States, Amy Wenxuan Ding

Digital technologies enable the flourishing of the creator economy. However, there exists a common issue of the Rule of Participation Inequality such that on most creator economy platforms over 90% of users never contribute content, which poses a big threat to the platform’s survival and growth. Creators may also exhibit different unobserved engagement states and be forward-looking in content contribution over time. However, little is known about creators’ such dual dynamics and how they impact content contribution strategies and platform incentive policies. This paper presents a novel dynamic structural model to examine these issues and provide guidance on how the platform incentivizes creators to reduce participation inequality. Using online literary markets as the context, we show that a creator exhibits multiple engagement states and is forward looking when producing a product serially. The impact of various incentives on content contribution depends on the creator’s engagement states. We find that ignoring either of the dual dynamics leads to biased estimates of product performance. Counterfactual analysis shows that a personalized engagement-state-based incentive design increases the platform’s revenue by 310%. Our study provides important implications for platform managers and creators to mitigate participation inequality and improve their performance.

2 - Catch Me if You Can: Punishing Disintermediation in the Gig-Economy Grey Market
Yingkang Xie, Washington University in St. Louis, Sant Louis, MO, United States

Punishment is commonly used as a threat to prevent fraud and opportunistic behavior (e.g., tax avoidance, informal economy, disintermediation). However, its true efficacy is not well understood because cheaters could strategically hide their traces from the platform. Using a field experiment on an Uber-like platform, I investigate how likely different types of drivers are to re-engage in under-the-table transactions after being caught and informed about their violation (e.g., side deals with customers). I compare the effectiveness of temporarily suspending drivers from acquiring new jobs to that of issuing a simple warning, finding that the latter is just as effective in reducing side deals. Moreover, warning triggers less strategic behavior, such as drivers disabling GPS to evade future detection. The effectiveness of warning and account suspension is moderated by the drivers' operating system, stakes in a transaction, and rapport with the platform. Our findings have implications for platform governance and customer relationship management on whom to target and what to communicate when they engage in grey market activities.

3 - Feedback Design in Content Market
Xinbo Wang, Department of Economics, National University of Singapore, Singapore, Singapore, Nan Chen

Feedback systems are widely used in content markets. This paper empirically examines how both positive and negative feedback impact the user contribution and its implication on the design of the feedback system in the context of a knowledge-sharing platform. Utilizing variations in the timing of feedback, we find that users respond significantly to both positive and negative feedback, but their response to the latter is greater. Interestingly, feedback tends to affect how much a user contributes, in terms of quantity, rather than improving or diminishing the quality of their contributions. We develop a structural equilibrium model of knowledge sharing in the content market where the supply side responses to feedback and faces competition. Through model estimation and counterfactual simulations, we contrast the efficacy of the existing feedback system with that of alternative designs. Our study offers insights into optimizing feedback mechanisms to foster high-quality content and active user participation.

Keywords: Content market; Reputation system; Upvotes and downvotes; Negative feedback; Platform design

TB20
Meeting Room C4.9
Consumer Behavior: Branding 2
Contributed Session
Chair: Nader Tavassoli, London Business School, NW1 4SA
1 - How to Promote Local Specialty that Encourages Travel to the Production Area: Focusing on the Effects of Presenting the Product's Production process as a Story
Takanori Suda, Toyo Gakuen University, Tokyo, Japan, Hiroaki Ishii, Ikumi Hiraki, Naoto Onzo
Attracting tourists with inbound tourism has become a significant issue in regional destination marketing. This study focuses on the effect of promoting a local specialty as a story, which lacks research. Subsequently, we identify the effect of narrative information in presenting the production process of a local specialty on the intention to visit the production area. Previous research has explored various production-related information affecting product perception and destination image, such as the Country of Origin (Dichter, 1962). However, few discussions on the effects of showing the entire production process on consumers have been made. Previous studies have confirmed that chronological information, such as the production process, may induce consumers to process narrative information (e.g., Van Laer et al., 2014). Moreover, narrative information processing can increase the destination image and visit intention through the consumer’s immersive imagination of the story. Therefore, we conducted an online survey, under two conditions (participants are shown six images of the production process of a product vs. six images of the finished product) single-factor randomized between-subjects design. Participants who viewed the production process had increased intention to visit the production area compared with those who viewed the finished product. Furthermore, this effect was sequentially mediated by narrative transformation and evaluation of the destination image. Findings indicate that presenting the entire production process of a local specialty to consumers is an effective way to increase their intention to visit the production area.
2 - Beyond Hashtags: Do Social Media Callouts and Shoutouts Echo in Brand Performance?
Mithila Guha, San Jose State University, San Jose, CA, United States, Daniel Korschun, Trina Andras

This research investigates the impact of consumer activism on a brand's social media engagement and stock market performance, specifically examining the sociopolitical context of the ‘2022 Russian invasion and humanitarian crisis in Ukraine.' Analyzing daily tweets from 159 publicly traded global brands between February and March 2022, the study identifies brands actively participating in the discourse. The investigation includes 1,322 daily tweets, 39,578 user replies, and 41,358 activist tweets from consumers directed at these brands. Through classifying consumer social media activism into "shoutouts" and "callouts," representing expressions of approval and criticism, respectively, the study explores the nuanced tones in these interactions. Sentiment analysis extends to users' engagement with activist brands' other social media content on the same day as their response to the sociopolitical issue. Additionally, daily stock returns data are examined to assess the impact of consumer activism on brand performance. The findings suggest a positive association between brand activism and both callouts and shoutouts. Brands with higher valuations may receive more shoutouts. Importantly, the study provides insights into how sociopolitical issues influence brand performance through brand-consumer dialogues. Notably, shoutouts positively correlate with a brand's stock market performance, while callouts negatively impact both users' engagement and daily stock returns. This research underscores the significance of understanding and navigating consumer activism for brands, highlighting its implications on public perception and stock market performance of the target brands.

Keywords: Brand-Consumer Dialogues, Consumer Activism, Sociopolitical Stance, Stock Market Performance, Social Media.

3 - How Warmth and Competence Language Cues Impact Human Brands: The Case of Business Leaders
Valeria Noguti, University of Technology-Sydney, Sydney, Australia, Valentyna Melnyk

This research investigates how written language—i.e., competent and warm language cues that business leaders use—drives consumer responses on social media. The empirical investigation employs automated text analyses (i.e., through LIWC—Linguistic Inquiry and Word Count) using a unique dataset of 16,000+ posts from two-way communications between business leaders and consumers on social media, more specifically Reddit. Results demonstrate that competent and warm language cues affect visibility and WOM valence toward business leaders differently. Both types of cues enhance WOM valence. However, in the case of visibility, warm language cues decrease visibility, whereas competent language cues have no overall effect. The relationships are moderated by business leaders’ personal popularity and company popularity. The research offers guidelines for business leaders on how to improve their language use to reach desirable consumer outcomes.

4 - Categorization and Horizontal Versus Vertical Brand Differentiation
Nader Tavassoli, London Business School, London, United Kingdom

Is Red Bull perceived differently when categorized as a soft drink versus an energy drink? Our research builds on literature demonstrating that brand perceptions, attitudes, and choices depend on the competitive context. We specifically examine how categorization affects horizontal and vertical brand differentiation. Our studies reveal that brands are perceived as more horizontally differentiated (e.g., more unique) within broader supracategories (e.g., Red Bull as a soft drink) and as more vertically differentiated (e.g., higher quality) within narrower subcategories (e.g., Red Bull as an energy drink). Theoretically, we test the roles of perceived entitativity and attribute diagnosticity in driving horizontal differentiation, and the 'frog-pond effect' and contrast effects in influencing vertical differentiation. Practically, this research offers marketers valuable guidance for developing positioning strategies, including the use of competitive set labels such as 'energy drink', and tactics for effective brand placement in both online and physical retail spaces. Additionally, our research underscores the potential unintended consequences in market research. We highlight the need for caution regarding brand-order effects in questionnaires (e.g., Red Bull rated after Pepsi versus Monster) and the implications of response formats in tools like Brand Asset Valuator © for assessing horizontal and vertical differentiation.

TB21
Meeting Room C4.10
Consumer Behavior: Response to AI 2
Contributed Session
Chair: Joyce Chua, Singapore Management University
1 - Bias in Generative AI
Vibhanshu Abhishek, UCI Paul Merage School of Business, Irvine, CA, United States, Mi Zhou, Kannan Srinivasan

This study analyzed images generated by three popular generative artificial intelligence (AI) tools representing various occupations to investigate potential bias in AI generators. Our analysis revealed two overarching areas of concern in these AI generators, including (1) systematic gender and racial biases, and (2) subtle biases in facial expressions and appearances. Firstly, we found that all three AI generators exhibited bias against women and African Americans. Moreover, we found that the evident gender and racial biases uncovered in our analysis were even more pronounced than the status quo when compared to labor force statistics or Google images, intensifying the harmful biases we are actively striving to rectify in today’s society. Secondly, our study uncovered more nuanced prejudices in the portrayal of emotions and appearances. For example, women were depicted as younger with more smiles and happiness, while men were depicted as older with more neutral expressions and anger, posing a risk that generative AI models may unintentionally depict women as more submissive and less competent than men. Such nuanced biases, by their less overt nature, might be more problematic as they can permeate perceptions unconsciously and may be more difficult to rectify. Although the extent of bias varied depending on the model, the direction of bias remained consistent in both commercial and open-source AI generators. As these tools become commonplace, our study highlights the urgency to identify and mitigate various biases in generative AI, reinforcing the commitment to ensuring that AI technologies lead to a more inclusive future.

2 - Say "Hi" to AI Colleague: How AI Influence Organizational Decision Making and Knowledge Transformation
Ru Yu, Fudan University, Shanghai, China, People's Republic of, YuXuan Zhao

Abstract: The capability of knowledge transfer within an organization is a source of its competitive edge. Newcomers serve as a crucial conduit for such knowledge transfer. Technological progress has equipped artificial intelligence with extensive and integrated business acumen. Meanwhile, the cost of deploying AI has been drastically reduced in many ways, leading AI become novel new labor force and “coworkers” in business. While there is a wealth of research on AI's influence on individual decision-making, little has known of how new AI employees impact organizational decision-making and knowledge transfer. Using a combination of filed and lab studies, this paper explores team decision-making process, and investigates the following questions: (1) What’s new when AI employees joining the team decision-making? (2) Compared to human labors, how AI’s knowledge transfer to the whole team? (3) How can managers coordinate the relationship between human employees and AI employees to ensure the long-term development of human-machine collaboration? The market potential of artificial intelligence is enormous. However, when AI enters the business world as a new type of workforce, organizations and managers are still unclear about the impact it brings. The author explores one scenario of commercial applications of artificial intelligence, show the impact of AI in the domain of organizational knowledge, and highlight future research directions.


Key Words: AI employee, Group Decision Making, Knowledge Transfer
3 - Complementarity Neglect: When People Select Disadvantageous Partners with Overlapping Mistakes
Meng Zhu, Virginia Tech, Baltimore, Maryland, United States, Moritz Jörling, Ye Li
Research has shown that teams, consisting of human and AI, perform better than teams of only human or only AI. Due to relying on different sources of information and different ways of processing that information, humans and AI have particular strengths but also particular weaknesses, creating complementarity. However, it remains unclear if consumers recognize complementary and strategically choose a collaborator that helps them achieve the best outcome. This is an important question in an environment where consumers have many options to collaborate and share tasks with intelligent technology. We approach this question with multiple incentivized classification task experiments. Several preregistered lab and online studies (N > 3,000) find that people have difficulties in selecting a complementary collaborator and instead choose a collaborator that makes the same mistakes that they make themselves. This finding is independent of whether the potential collaborators are labelled as AI or not. Different interventions were tested to improve individuals' choice of collaborators. The research demonstrates that asking people about perceived complementarity increases their sensitivity for it and improves their decision-making. Individuals decision-making can be additionally improved by providing performance summaries that underline individual strengths and weaknesses. While previous research on human-technology collaboration focused solely on accuracy, this research adds complementarity as an important additional component and demonstrates that accuracy in collaborations is a short-sighted assessment of collaboration partners.
4 - Emerging Technologies Affect the Use of Specific Language in Consumer Reviews
Joyce Chua, Singapore Management University, Singapore, Singapore, Hannah Chang
New technology has expanded the ways in which consumers can interact with businesses. We investigate how emerging technologies may influence consumer reviews on online platforms. Across three studies, we examine how the design of consumer review platform affects the use of emotional and prosocial language in reviews, assessing the potential effects of review agents (AI vs. human) and review modality (voice vs. written) on one’s use of prosocial language in their reviews. We find these design factors—who the expected review agents are (agent) and how the review is elicited (modality)—mediates consumer reviews. The findings show that the review modality has a consistent effect on the content of consumer review: Written reviews elicited more use of prosocial language and this process was mediated by specific positive emotions (e.g., elevation and admiration). Overall, we seek to understand the moral dimensions of consumer behavior in online review platforms and highlight the review platform design in shaping the use of emotional language in consumer reviews. Implications are discussed.
TB22
Meeting Room C4.11
Four Applications of Natural Experiments in Marketing: Online Gambling, Corporate Political Positioning, Food Shopping Carbon Footprint, and Open Innovation
Invited Session
Special Sessions
Chair: Wayne Taylor, Southern Methodist University, Dallas, TX, United States
1 - The Impact of Online Gambling on Risky Behavior and its Implications for Demarketing
Wayne Taylor, Southern Methodist University, Dallas, TX, United States, Daniel McCarthy, Kenneth Wilbur
Online sports betting and online casino gambling is experiencing rapid growth, with 24 states adding online sports betting since 2019 and 7 states currently permitting online casinos. This paper studies the effect of legalization of online betting on deposit and withdrawal activity from over 1.5 million gamblers at a variety of gambling merchants, including market leaders DraftKings and FanDuel. We use generalized synthetic control with treatments grouped by gambling penetration within the state at time of legalization: legalizing online sports betting with and without established retail sports betting outlets, and legalizing online casinos with and without established online sports betting. We augment the gambling transaction data with related measures including excessive gambling spells and calls to a gambling help hotline. Initial results suggest that the impact of legalizing online sports betting is much larger than the impact of legalizing online casino games in states with established online sports betting. We present the tradeoffs between projected increases in gambling revenue against the changes in risky gambling behavior and discuss the implications for potentially demarketing targeted gamblers.
2 - Corporate Political Positioning and Sales: Evidence from a Natural Experiment
Kitty Wang, University of Houston, Houston, TX, United States, Shijie Lu
We study if and how corporate political positioning affects retail sales. We use
data from a large U.S.-based lifestyle specialty retail brand and a similar control brand before
and after an involuntary revelation of the focal brand’s political position. We find that the
focal brand’s total sales dollar amount and quantity increased by 17.1% and 12.7% after the
event relative to the control brand. The change in the customer base rather than basket size
drives the measured effect, suggesting an informative function of corporate political
positioning by communicating the brand’s political ideology to consumers who share a
similar ideology, which in turn brings an influx of new customers and reminds existing
customers to repurchase. Further, sales increase more in places where the local political
preference aligns more with the focal brand’s position. This moderating effect of local
political preference mainly applies to conspicuous rather than inconspicuous products,
which provides suggestive evidence that supporting consumers use consumption as a
signaling device to communicate their political ideologies to others
3 - Household Food Carbon Footprint and Club Store Shopping
Lu Yan, Hong Kong Polytechnic University , Hong Kong, Hong Kong, Ting Zhu, Li Song
As food system contributes approximately 34% of global greenhouse gas (GHG)
emissions, individuals’ unintended decisions on what to eat and where to purchase food
have a consequential effect on the environment. With their fast growth and unique shopping
environment (e.g., lower prices, larger package sizes, membership fees), club stores play a
significant role in households’ food purchases. This study examines whether and why
shopping at warehouse club stores affects a household’s total food carbon footprint. Using
the process-based Life Cycle Assessment (LCA) model and Nielsen Consumer Panel data, we calculate the household-level food carbon footprint between 2007 and 2017. With both
staggered difference-in-differences and synthetic difference-in-differences approaches, we
find that households generate 7-11% more per capita food carbon emissions after starting
shopping at club stores. Moreover, we observe the heterogeneous treatment effect that a
larger increase for low-income and small households. We further investigate how retail
formats, package size and price at the club store relative to other local grocery stores, affect
household food carbon footprint increases and find that larger package sizes in club stores
play a more important role in the food carbon footprint increase.
4 - The Impact of Large Language Models on Open-source Innovation: Evidence from GitHub Copilot
Raveesh Mayya, NYU Stern School of Business, New York, NY, United States, Doron Yeverechyahu, Gal Oestreicher-Singer
Open innovation, particularly in open-source software, relies on voluntary contributions from subject matter experts, dedicating their time to enhance both the quantitative and qualitative aspects of projects. While existing research acknowledges that Large Language Models (LLMs) enhance productivity in programming tasks, the impact on voluntary contributions in community settings remains unclear. This is crucial in opensource communities, where innovation depends on self-motivated contributors with above average programming expertise. Examining this in the context of GitHub Copilot, a programming-focused LLM launched in October 2021, our study leverages a natural experiment by comparing supported languages (e.g., Python) with those omitted (e.g., R). Using classical Time-Weighted Fixed Effects (TWFE) with matching and modern estimation techniques like synthetic Difference-in-Differences (DiD), we reveal compelling insights. Python packages experience a significant increase in new updates compared to matched R packages, driven by a rise in commits per contributor. Notably, the surge is more pronounced in bug fixes than new features, indicating LLMs influence contributors' motivations toward resolving time-critical, incremental issues. Individual contributor-level analysis unveils a substantial increase in the count of packages to which an existing contributor contributes, demonstrating that GitHub Copilot enhances the mobility of voluntary contributors. Our findings underscore the potential of programming-focused LLMs to stimulate innovation in community settings, fostering knowledge spillover across projects. This has practical implications for firms encouraging employees to contribute to open-source projects.
Thursday, June 27, 1:30 PM - 3:00 PM
TC01
Pyrmont Theatre
Practice Prize Competition 1
Invited Session
Chair: Lan Luo, Columbia Business School
1 - Modelling Customer Lifetime Value in the Retail Banking Industry
Greig Cowan, NatWest Group , Edinburgh, United Kingdom, Salvatore Mercuri, Raad Khraishi
Understanding customer lifetime value is key to nurturing long-term customer relationships, however, estimating it is far from straightforward. In the banking industry, commonly used approaches rely on simple heuristics and do not take advantage of the high predictive ability of modern machine learning techniques. We present a general framework for modelling customer lifetime value which may be applied to industries with long-lasting contractual and product-centric customer relationships, of which retail banking is an example. This framework is novel in facilitating CLV predictions over arbitrary time horizons and product-based propensity models. We also detail an implementation of this model which is currently in production at a large UK lender. In testing, we estimate an 43% improvement in out-of-time CLV prediction error relative to a popular baseline approach. Propensity models derived from our CLV model have been used to support customer contact marketing campaigns. In testing, we saw that the top 10% of customers ranked by their propensity to take up investment products were 3.2 times more likely to take up an investment product in the next year than a customer chosen at random.
2 - Business Policy Experiments using Fractional Factorial Designs: Consumer Retention on DoorDash
Yixin Tang, DoorDash, San Francisco, CA, United States, Yicong Lin, Navdeep Sahni

This paper investigates an approach to both speed up business decision-making and lower the cost of learning through experimentation by factorizing business policies and employing fractional factorial experimental designs for their evaluation. We illustrate how this method integrates with advances in the estimation of heterogeneous treatment effects, elaborating on its advantages and foundational assumptions. We empirically demonstrate the implementation and benefits of our approach and assess its validity in evaluating consumer promotion policies at DoorDash, which is one of the largest delivery platforms in the US. Our approach discovers a policy with 5% incremental profit at 67% lower implementation cost.

TC02
Meeting Room C2.1
Algorithmic Pricing
Contributed Session
Chair: Hangcheng Zhao, Wharton School, University of Pennsylvania, Philadelphia, PA, United States
1 - Welfare Cost of Fair Prediction and Pricing in the Insurance Market
Fei Huang, University of New South Wales, Sydney, Australia, Hajime Shimao
While the fairness and interpretability in machine learning tasks have attracted attention from practitioners, regulators, and academics for many applications, their consequence in terms of stakeholders' welfare is under-explored, especially via empirical studies and in the context of insurance pricing. General insurance pricing is a complicated process that can involve cost modeling, demand modeling, and price optimization. Fairness and interpretability regulatory constraints in the insurance market can be applied at each stage of the insurers’ decision-making. What are the welfare implications of the existing and potential fairness and interpretation regulations on both consumers and firms in the insurance market? In this paper, we answer this question by developing an empirical framework covering the entire pricing process and applying it to a dataset of the French auto insurance market. Our main results show that fairness in machine learning does not lead to fairness in insurance pricing or consumer welfare, while they can harm the insurer’s profit and consumer welfare significantly and differently under different regulatory scenarios.
2 - Price Competition Under Information (Dis)Advantage
Nan Chen, National University of Singapore, Singapore, Singapore, Tiffany Tsai
We examine the impact of asymmetric data access within a vertically integrated e-commerce platform. Using a unique daily panel, we find that both the platform owner and third-party sellers base pricing on past sales, but only the owner exploits competitors' sales data. We estimate a price competition model with heterogeneous seller learning. We find that (1) eliminating the owner's access to competitors' data increases social welfare by 0.33%, negatively impacting the owner through reduced first-party sales and (2) providing third-party sellers with equal access to competitors' data as the owner increases social welfare by 1.65%, benefiting the owner through increased referral fees and yielding a Pareto improvement.
3 - Sponsored Products Ads, Algorithmic Pricing, and More Harmless Collusion
Hangcheng Zhao, Wharton School, University of Pennsylvania, Philadelphia, Pennsylvania, United States
Artificial Intelligence (AI) algorithms are becoming increasingly pervasive in helping firms to learn about their competitive environments and make business decisions. Recent research has shown that these algorithms raise concerns about learning to collude on high prices, which may benefit firms but hurt consumers. However, the majority of the analysis so far has looked at AI algorithms which make a single-dimensional pricing decision. This paper investigates whether AI algorithms can also benefit consumers when firms use these algorithms for another important decision - how to buy sponsored ads. We investigate the interaction between AI pricing algorithms and sponsored ads bidding algorithms. We utilize a game-theoretic model to analyze firms' competitive strategies and conduct extensive experiments simulating an AI-powered environment and interactions. Our findings show that sponsored ads can alleviate the concerns about collusion on high prices. When firms use AI algorithms (specifically Q-learning algorithms, which are commonly used in computer science) to inform their pricing and bidding strategies, algorithms can facilitate outcomes beneficial for both consumers and sellers. The intuition is that algorithms learn to coordinate on lower bids, lowering the advertising costs and resulting in lower prices. This paper contributes to the existing literature by shedding light on the impact of AI in advertising and providing insights that challenge the prevailing policy concern that algorithms always lead to harmful collusion.
TC03
Meeting Room C2.2
B2B Marketing 1
Contributed Session
Chair: Uyen Uyen Banh, University of Manitoba, Winnipeg, MB, Canada
1 - Effectiveness of Sales Visits over the Business Cycle
Christian Pescher, Universidad de los Andes, Chile, Las Condes, Region Metropolitana, Chile, Hernan Bruno, Shantanu Dutta
Sales visits are seen as viable cost cuts during difficult times. When the economy enters a downturn, marketing managers increasingly face pressure to justify and reduce their sales budgets. We have little evidence to justify this conventional wisdom and there is a paucity of research on the effectiveness of sales visits during contractions (or recessions) relative to expansions. Part of the reason is that business cycle fluctuations are hard to predict and transpire over years, making it next to impossible to plan randomized experiments. We analyze sales visit effectiveness (lift) over the business cycle using a causal framework, by creating placebo visits to the same customer within a year before and after the actual visit and applying a propensity weight to account for the likelihood of a visit at a given point in time. The dataset covers seven years of data including over 10,000 sales visits before, during and after the Great Recession from 2008/2009, which provides a strong contrast between expansion and contraction periods. We find that sales visits have a positive effect in both, expansion and contraction periods. Surprisingly, the percentage “lift” of a sales in the month after a visit is higher in contractions (+25%) than in expansions (+10%) periods. During expansion periods, we see evidence for an increase in average revenue per purchase driven by sales visits. In contraction periods, we find not only an increase in average revenue but also an increase in the number of items sold as a result of the visit.
2 - Buyer-driven Financial Aid and Technology Assistance on Marketing Capabilities Development of Global Suppliers in Emerging Markets
Tianjiao Qiu, California State University Long Beach, Long Beach, California, United States
The importance of supplier development in emerging markets through dedicated financial aid and non-financial assistance from the buying firms has long been posited. However, little is known about the influence of buyer-driven financial aid and non-financial assistance on global suppliers’ marketing capabilities development in emerging markets. This study examines how buyer-driven financial aid and technology assistance impact global suppliers’ marketing capabilities. The results from the mixed linear models with maximum likelihood methods on the survey data from 239 global suppliers in emerging markets indicate that buyer-driven financial aid to global suppliers is significantly related to buyer-driven technology assistance. While buyer-driven technology assistance positively impacts global suppliers’ marketing capabilities, buyer-driven financial aid negatively affects the development of global suppliers’ marketing capabilities. The moderation test shows that suppliers’ new product development capabilities moderate the effect of buyer-driven technology assistance on global suppliers’ marketing capabilities development. Interestingly, a further quadratic model shows that buyer-driven technology assistance has a nonlinear relationship with global suppliers’ improvement of marketing capabilities. High levels of buyer-driven technology assistance contribute to improved supplier marketing capabilities, while low levels dampen supplier marketing capabilities. The findings offer insights into the differential impacts of buyer-driven financial aid and technology assistance on developing global suppliers’ marketing capabilities in emerging markets.
3 - Exploring the Antecedents of Distributor Whistleblowing: A Cognitive Approach
Jing Zhou, Dalian University of Technology, Dalian, China, People's Republic of

Distributor whistleblowing, a distributor’s reporting of a peer’s wrongdoing to a manufacturer, can help the manufacturer to detect and deter distributors’ misconduct within the distribution channel. This research presents a cognitive process model of distributor whistleblowing. Building on social information processing theory, our model posits that a wrongdoing leads to distributor whistleblowing through two pathways. Specifically, the seriousness of a peer’s wrongdoing evokes a focal distributor’s evaluation of economic unfairness and moral responsibility, which in turn affect the focal distributor’s intention to blow the whistle. Our model also describes how the use of the two pathways is influenced by market uncertainty and the tie strength between the focal distributor and manufacturer.

To test our hypotheses, we collaborated with a top automobile manufacturer in China to collect data, including survey data from the manufacturer and its distributors, and secondary data from the manufacturer and publicly available sources. Analysis of the data generally supports our hypotheses. With these insights, we aim to offer three theoretical contributes. First, this study provides the first gleam of cognitive processes of distributor whistleblowing, which can be used to identify antecedents of whistleblowing in the buyer-supplier context. Second, uncovering the mediating effects of economic fairness and moral responsibility, this study addresses the underlying mechanisms between wrongdoing seriousness and whistleblowing. Third, this study offers a contextual perspective to explore the moderating role of exchange ties and market uncertainty. The findings also extend SIP theory by uncovering boundary conditions that affect the use of decision frames.

4 - Identifying the Drivers of Review Generation in Business-to-Business e-Commerce
Uyen Uyen Banh, University of Manitoba, Winnipeg, MB, Canada, Sandeep Arora
Although online reviews play an important role in the decision-making process for both business to consumer (B2C) and business to business (B2B) consumers, extant literature on online reviews has primarily focused on B2C. The differences between B2B and B2C are distinctive enough to justify dedicated investigations and distinct approaches. This large-scale empirical study contributes to the scarce literature on B2B online review sharing. The findings suggest that customer-initiated contacts with firms (CICs), relationship duration, free-tiered pricing, and competition are the main factors driving the behavior of B2B buyers’ online review sharing. Specifically, the authors find that relationship duration has a U-shaped relationship on online review sharing, recommending B2B marketing practitioners to have appropriate strategies to encourage buyers to share online reviews at either the early or later stages of the relationship. They also find that CICs have an inverted U-shaped effect on online review sharing, advising marketing practitioners to be cautious of the double-edged sword of this important variable. The authors hope this research will attract more academic work to offer insights on B2B online reviews.
TC04
Meeting Room C2.3
Choice Models 1
Contributed Session
Chair: Sicheng Song, The Chinese University of HK,Shatin, Jokey POSTGRADUATE HALL No.3, 999077
1 - The Role of Emotions in Consumer Decision Making Across Different Categories
John Roberts, University of New South Wales, Kirribilli, Australia, Songting Dong, Karen Hansen
Emotions have been shown to influence consumer evaluation and choice processes in ways that provide additional explanatory power to the measure of cognitions alone. However, very little is known how this influence varies depending on the category being purchased. After calibrating the characteristics of different categories we examine the role of emotions in each in terms of the level of the emotions elicited, the dispersion of emotions between brands in the category, the importance weight of emotions, and the overall impact of emotions on choice.
2 - Theory-driven Deep Learning-Based Consumer Choice Models
Dongcheng Zhang, Emory university, Atlanta, GA, United States, Kunpeng Zhang, David Schweidel

Consumer choice modeling is a fundamental task in marketing research and practice. To model consumer choices, on the one hand, machine learning-based models (e.g., neural networks) have demonstrated superior predictive performance. However, these models are completely non-parametric and lack micro-foundation and interpretability, preventing them from generating deeper marketing insights beyond prediction. On the other hand, well-established consumer choice models such as the Multinomial Logit Model, are well-known and commonly used for their simplicity, micro-foundation, and interpretability. However, these models rely on restrictive parametric assumptions, which limit their ability to capture the complex consumer heterogeneity and dynamics. Motivated by the drawbacks of these two streams of methods (i.e., machine learning models and traditional parametric models), we develop a theory-driven deep learning-based consumer choice modeling framework, in which we can simultaneously leverage the power of deep learning and the micro-foundation of consumer choice theories. On the one hand, compared to traditional parametric models, our model captures the complex heterogeneity and dynamics of the parameters of interest (e.g., price sensitivity and brand preference) in a non-parametric manner, which also leads to the improvement of predictive capability. On the other hand, compared to machine learning models, our approach achieves comparable or better predictive performance while keeping the micro-foundation, so that we have economic interpretations of consumer dynamics, which can also help us do better targeting, such as coupon assignment. Therefore, by developing a theory-driven deep learning framework, we are able to provide a more accurate and interpretable understanding of consumer choices.

3 - Modelling Individual and Group Travel Behavior for After-work Activities via Metro Data
Sicheng Song, Chinese University of Hong Kong, Shenzhen, Hong Kong

Non-commuting travel demand takes a larger share today while travel demand for after-work activities, which induce more expense and consumptions, is a significant part of it especially on weekdays. Meanwhile, group travel behavior lacks understanding compared with individual travel behavior. Here, we used smart card dataset to model location choice of commuters’ after-work activity on weekdays, involving a trade-off between station attractiveness and travel impedance. We defined a new metric of travel impedance and used observable proxy indicators, including commuting distance and the characteristics of commuters’ home and workplace stations, to capture interpersonal heterogeneity. This heterogeneity is predicted to explain the difference of commuters’ location choice behavior and to distinguish the population of individual traveler and group traveler. We applied this method to metro commuters in Shanghai, China. The model fits well and the new metrics of travel impedance result in a better fit than the current detour impedance. Moreover, the individual behavior patten is distinctly different from group travel behavior patten in terms of location choice, consumption intention, length of stay, and types of activity. The use of new metric of travel impedance and proxy variables gives an improvement in modelling location choice of consumers and in capturing interpersonal heterogeneity. Also, it helps service and retail business planners to understand where people would like to consume and where their customers come from. It will be good for them to select palace to open new branch stores or recreational centers, where can attract more consumers especially on weekdays.

TC05
Meeting Room C2.4
Marketing Strategy: New Products and Diffusion
Contributed Session
Chair: Yuhsiang Lin, Tian-Mu Campus: No.101, Sec. 2, Zhongcheng Rd., Shilin Dist., Taipei City , 111036
1 - Understanding the Effects of Uncertainty on NPD Speed: A Temporal Perspective
Qing Ye, Hunan University, Changsha, China, People's Republic of, Fue Zeng, Cheng Lu Wang

Existing research describes a broad range of adaptive strategies to attend to environmental uncertainty (i.e., market and technological uncertainty). This phenomenon is based partially on a paucity of requisite information with which to make accurate predictions. Another basic and important element of environmental uncertainty is rapid change. To date, this aspect of such incertitude remains unexamined. We suggest that one effective adaptive strategy to attend to environmental uncertainty is to increase the speed of new product development (NPD). Drawing from entrainment theory, we investigate the effects of environmental uncertainty on NPD speed in 248 firms across multiple industry sectors. We examine the direct effect, as well as how this effect is moderated by information acquired from external networks—trade associations and non-profit and non-government organizations. We postulate that, for market uncertainty, the effect is stronger with TAs and weaker with NPONGOs. For technological uncertainty, however, we anticipate that the effect is stronger with NPONGOs and weaker with TAs. Our work extends the uncertainty literature by refining the two natures of uncertainty, unpredictability and rapid change. And our efforts also contribute to the uncertainty literature also by using a temporal perspective as the theoretical lens to propose a novel strategy to respond to uncertainty.

2 - Does Breaking News Break Investors' Attention?
Reo Song, California State University Long Beach, Long Beach, California, United States, Eunho Park, Andre Martin, Tarun Kushwaha

People strategically allocate their limited resources by directing attention toward salient events, recognizing the scarcity of this valuable resource. While prior research has shown that the salience of an event related to the focal firm can influence investors' decisions, an underexplored area concerns the impact of media coverage of an event that is not specific to the focal firm on investors' focus. We explore how unrelated breaking news can dilute the positive effect of a company's announcement of new products on stock returns. Our sample included 1,705 new product announcements from 500 companies between 2016 and 2019 and variables such as stock prices and firm-level details. Daily news significance was constructed using daily traffic of ten major news sites. News articles were categorized into predictable and unpredictable news using topic modeling. Each new product was classified as either breakthrough or incremental innovation using the BART (Bidirectional and Autoregressive Transformers)-Large-MNLI. Additionally, we estimated the sentiment of new product announcement and daily news articles using DistilBERT (Bidirectional Encoder Representations from Transformers). Our analysis reveals that the positive effect of new product announcements on stock returns diminishes when investors are distracted by breaking news. This negative moderating effect of breaking news is more pronounced when the event was unpredictable, especially for firms with a higher proportion of retail investors. Our research shows that adjusting the timing of corporate event announcements can enhance firm value, offering a strategic approach to minimize the adverse effects of external distractions.

3 - To 'Make'; or 'Source' New Product Innovation Assets Integrated with Emerging General-purpose-technology? Evidence from Preannouncements of New Products Integrated with Artificial Intelligence
Manjunath Padigar, Macquarie University, Macquarie Park, Australia, Ljubomir Pupovac, Ashish Sinha, Abhishek Borah
Artificial Intelligence (AI) is increasingly seen as a General-Purpose Technology (GPT) of the digital era, with many market application domains integrating it in their new products and services. While previous new product preannouncement (NPP) literature offers insights on innovations based on Specific-Purpose Technologies (SPTs), no prior study has examined how NPPs of GPT-based innovations are valued by investors. We address this gap by first highlighting the key conceptual differences between GPT- and SPT- based NPPs and then emphasizing the criticality of the route to innovation (i.e., Make or Source innovation assets) choice in GPT-based innovations. Further, we build on the resource-based view to develop an GPT investment uncertainty resolution framework, and argue that ‘Make’ choice can be (a) counter effective for firms with a lower strategic flexibility (e.g., high financial leverage and low slack resources) as such firms may be unable to manage uncertainties and absorb new capabilities offered by GPTs, but (b) effective for firms with a higher strategic emphasis on value creation, as they likely possess prior experience and dynamic capabilities necessary to create superior value. An event study of 227 NPPs of new product innovations integrated with AI between 2009-2018 lends support to our predictions. The theoretical and managerial implications are discussed.
4 - Word-of-Mouth Life Cycle: Exploring Antecedents, Process, and Consequences
Yuhsiang Lin, University of Taipei, Taipei, Taiwan
Over the last two decades, myriad scholarly investigations have delved into the influence of diverse word-of-mouth (WOM) features on sales. Unfortunately, these studies have yielded inconsistent outcomes. Furthermore, there is a notable deficiency in comprehending the antecedents that influence the dynamics of WOM diffusion. In response, this study establishes a WOM lifecycle model to examine the turning points in WOM diffusion and its antecedents, as well as their impact on different sales stages. Additionally, this research explores the extent to which national cultures and product categories influence WOM and sales diffusion. Focusing on four distinct product categories, this study utilizes online reviews and sales data from Amazon and other official websites. The initial step involves mining WOM data to establish dimensions of WOM motivation and product attributes. Subsequently, the WOM lifecycle is segmented into various stages, leading to the development of the WOM lifecycle model using a hierarchical Bayesian structural equation approach. This model assesses the antecedents that affect the dynamics of diverse WOM features and their influence over different stages of the product life cycle (PLC) sales, in addition to evaluating the roles of national culture and product category. The results demonstrate that the proposed WOM lifecycle model effectively forecasts WOM diffusion patterns, indicating that various WOM motivations and innovative product attributes play a pivotal role in influencing sales at different PLC stages. In conclusion, the study highlights the significant influence of national culture and product category on the dissemination of product sales.
TC06
Meeting Room C2.5
Analytical Models for Consumer Behavior
Contributed Session
Chair: Alan Montgomery, Carnegie Mellon University, Tepper School of Business, Pittsburgh, PA, 15213, United States
1 - Consumer Search and Product Line Length: The Role of the Consumer-Product Fit Distribution
Mohammad Zia, Chapman University, Orange, United States, Dmitri Kuksov
More intense consumer search across firms may lead to both stronger price competition and a better match between customers and products. We show that the net result of these forces may lead to either shorter or longer product lines, and higher or lower prices and profits depending on the distribution of product valuations across consumers, even if there is no market expansion effect (i.e., if all consumers buy). We derive a general condition on the distribution (related to its hazard rate) under which lower search costs lead to longer product lines even when the market expansion effect is absent, thus highlighting the critical role of the value distribution in determining firms' optimal strategies in competitive markets. In particular, when consumer search costs decrease, product lines become longer if the distribution is Exponential or Log-Normal, become shorter if the distribution is Normal, Logistic, or Gumbel-Minimum, and do not change if the distribution is Gumbel-Maximum. With uniformly distributed product values, equilibrium prices and product lines exhibit an inverted-U-shaped relationship. Incorporating the market expansion effect, firms have an additional incentive to expand product lines when search costs decrease, and the relationship between search costs and equilibrium product line length may reverse direction.
2 - Product Reviews, Prices, and Profits in Markets Driven by Randomness: The Null Model
Man Xie, Arizona State University, Phoenix, AZ, United States, Steven Shugan

Myriad examples (e.g., Regression-to-the-Mean, Prosecutor’s fallacy, Gambler’s Fallacy, Multiple Comparisons Fallacy) illustrate how simple randomness evokes perilously wrong causal explanations and misguided implications. Judging the contribution of explanatory variables requires naïve null benchmark models that exclude those variables. We provide a novel null model where consumers lack preferences, bia­ses, and acquisition strategies. Like other null models, randomness influences outcomes. Information is randomly-distributed and the distribution depends on the product’s desirability. Consumers buy a product only when sufficiently positive information (reviews) reaches them. We show that (1) randomness creates market outcomes that are not obvious, uncomplicated, or trivial; (2) simple randomness produces outcomes that were previously explained with causal assumptions; (3) how markets behave when consumers use simple effort-minimizing heuristics (available information). Moreover, we provide optimal managerial strategies for markets resembling our null model. For example, we show when consumers use simple effort-minimizing heuristics, (1) profit-maximizing product prices can exceed the product’s expected worth implied by the same information that dictates that worth; (2) having more products is more profitable, even for almost identical products in direct competition; (3) more unbiased information always improves profits. Although explanatory theories are obviously indispensable, null models keep us from being fooled by randomness.

3 - Multitasking and Goal-Setting
Dong Wook Chae, Indiana University, Bloomington, IN, United States, Yao Yao, Krista Li

With technological advancement and increased task complexity, consumer multitasking has become a prominent phenomenon in modern society. However, existing research on the effect of multitasking on consumer outcomes is inconclusive. In this paper, we use an analytical model to examine how consumers with self-control problems allocate effort between multiple tasks over time to achieve certain goals. We examine the interplay between multitasking and goal-setting on the consumers' welfare to offer several insights and managerial applications. First, we find that multitasking benefits consumers by improving efficiency unless they face an unattainable goal. Second, general intuition suggests that multitasking distracts consumers from focusing on a focal task, resulting in a low-achieving goal and a lower payoff for the focal task. However, we find that consumers raise the difficulty of the goal for the focal task to mitigate self-control problems during multitasking, improving their performance and payoff for the focal task. Third, in contrast to the commonly held belief that more time could benefit outcomes, consumers could, in fact, set a lower goal with abundant time. Further, even with more time and higher goals, consumer welfare could be lower than a moderate goal and time. These results reflect novel non-monotonic and counterintuitive relationships between goal, time, and consumer welfare and provide novel insights for consumer-side and managerial applications.

4 - Modeling Mental Accounting with Two-stage Budgeting and Bounded Rationality
Alan Montgomery, Carnegie Mellon University, Pittsburgh, PA, United States, Nicholas Pretnar, Christopher Olivola
We construct a unifying theory of two-stage budgeting and bounded rationality to yield a model with mental accounting features. Mental accounting and rational inattention induce behavioral wedges between first-stage and second-stage expenditure budgets, reducing the fungibility of money across accounts. Because reviewing one's financial activities is cognitively costly, consumers might re-assess only a subset of their spending budgets every period. Over- or under-spending affects future budgeting and expenditure decisions. We apply latent Bayesian inference to agent-level weekly expenditure data in order to structurally estimate the degree to which low-income consumers appear rationally constrained with respect to budgeting. Our findings provide insight into how consumers may respond to interventions that encourage more disciplined budgeting behavior, like push notifications in budgeting apps. We show that if financial attentiveness constraints are relaxed, consumers who want but fail to discipline their expenditure may become financially worse off.
TC07
Meeting Room C2.6
Consumer Behavior: Gift Giving and Planning
Contributed Session
Chair: Jason Stornelli
1 - The Gift of Cultural Capital: How Culturally Mixed Gifts Strengthen Giver-recipient Bonds
Qianying Huang, School of Business, Sun Yat-Sen University, Guangzhou, China, People's Republic of, Zhuomin Shi

Gifts can serve as tangible expressions of social connections. Culturally mixed gifts (i.e., objects simultaneously displaying cultural symbols from various countries) are growing in popularity within the gift market. However, their symbolic significance in building relationships remained unexplored. This research, rooted in cultural capital theory, investigates how culturally mixed gifts impact giver-recipient relationships.

The research comprises five experiments, incorporating various gifting occasions and methods to assess relationship strength. In Study 1, we conducted a field experiment involving actual gift exchange within established personal relationships, demonstrating that from the perspective of recipients, culturally mixed gifts strengthen the giver-recipient relationship more than nonculturally mixed gifts (H1). Studies 2a and 2b aimed to test the underlying serial mediation of the proposed effect across different traditional festival gifting contexts, showing that this effect is motivated by recipients' views that culturally mixed gifts represent desirable cultural capital and enhance their perceived status (H2). Study 3 further examined the effect under the influence of gifting occasions, such that the beneficial effect of culturally mixed gifts is expected to exist in formal rather than informal settings (H3). Study 4 investigated the influence of the giver-recipient relative status, revealing that the positive impact of such gifts is greater when the recipient has a higher status than the giver (H4).

This paper contributes to the growing body of gift-giving research by examining how different gift types affect the development of giver-recipient relationships. Practically, our findings could guide gift selection strategies and cross-cultural marketing efforts.

2 - A Study on the Repulsion of Gift Givers to the Recommendation Function of Algorithms
Yueyan Zhang, Southwest Jiaotong University, Chengdu, China, People's Republic of

While AI algorithms can bring convenience and efficiency to recommendation services, consumers sometimes prefer time-consuming and laborious choices, such as in the case of buying gifts for others. This paper focuses on the special situation of artificial intelligence gift recommendation and explores the impact of algorithmic effort savings on people's perception and behavior in altruistic situations.The results of Four experiments found that participants who were personally involved in the gift selection thought they put in more effort and rated their gift-giver status more positively than those who used the AI recommendation feature. Guilt plays a mediating role in the influence of consumer effort on gift-givers' self-evaluation, and the size of choice set plays a moderating role in the relationship between consumer effort and givers' perceived guilt. The study also noted that the presence of positive gift messages (compared to no prompt messages) helped alleviate feelings of guilt about not making an effort and increased their self-evaluation of gift-giver status. This study enriches the explanation mechanism of algorithmic avoidance motivation in the academic field of marketing, and also provides new evidence for the positive consequences of consumer efforts in the altruistic context, which is helpful to change the long-held negative bias of academic circles about this variable. The results of this study help to provide practical advice from a consumer self-evaluation perspective for businesses that are using or planning to use AI recommendations.

3 - An Intertemporal Choice Model for Sustainable Consumption: Role of Connectedness to the Other's Future Self
Anas El Assadi, Mohammed VI Polytechnic University, Hay Moulay Rachid Ben Guerir, Morocco, Ravi Prakash Ranjan
Connectedness to the future self is a pivotal factor shaping intertemporal consumer decisions. While existing research primarily focuses on how this connection influences choices for oneself, the reality of intertemporal decisions often involves an individual acting on behalf of another. Delving into this unexplored territory, our research introduces a novel intertemporal choice model that scrutinizes the significance of connectedness to the other's future self in intertemporal decision-making scenarios involving multiple individuals. Specifically, we investigate its impact on the behavior of givers when faced with the dilemma of choosing between a present and a future gift. Our investigation unravels the profound influence of connectedness to the other's future self in shaping intertemporal choices made for others and contributes to advancing the connectedness to the future self-theory and resource optimization in sustainable consumption practices.
4 - The Multidimensional Planning Scale: Assessing Whether and Why Consumers Plan
Jason Stornelli, Oregon State University, Corvallis, OR, United States, Colleen Bee

We develop the Multidimensional Planning Scale (MPS) to measure both whether and why people do (not) plan for goal pursuit. We build on the Propensity to Plan Scale (Lynch et al. 2010; PTPS), which examines whether people plan but is largely silent on why they do so, even though different reasons may require different corrective action. Uncovering these reasons should help to design better tools to manage valued consumption goals such as eating well, exercising, and prudent spending, bolstering individual well-being along with organizational and societal outcomes.

The MPS contains four dimensions: (1) strategic thinking assesses comfort with adopting a “big picture” view of goal pursuit; (2) scheduling examines preferences on deciding when and where action will occur; (3) execution evaluates feelings on performing planned actions; and (4) achievement measures whether planning prompts feelings of control and progress.

Across seven studies, we show that the MPS demonstrates good reliability, consistency, and validity with an easy-to-administer format. It is not unduly influenced by social desirability or common method variance and it yields good test-retest reliability. Further, the MPS also adds predictive power over and above the PTPS on self-report outcomes. Finally, the MPS predicts results of a consequential consumer goal – finishing one’s holiday shopping – and the subjective experience of that goal pursuit.

We submit that the MPS fosters a better understanding of both whether and why consumers choose or avoid planning. We anticipate this work will help to maximize the effectiveness of planning tools and bolster consumer welfare.
TC08
Meeting Room C3.3
Customer Lifetime Value Models
Contributed Session
Chair: Sibo Zhang, University of New South Wales
1 - Revisiting “A Seasonal Model with Dropout …”: A Simple Model with Seasonal Effects for Noncontractual Settings
Bruce Hardie, London Business School, London, United Kingdom, Peter Fader, Daniel McCarthy

One problem with the classic models of repeat buying in noncontractual settings (e.g., Pareto/NBD, BG/NBD) is that they fail to account for seasonality in the flow of transactions – a critical problem that can lead to poor forecasts, biased parameter estimates, and misleading managerial inferences. Several recent papers have addressed this issue – either specifically (Wünderlich et al. 2022) or more generally (e.g., Bachmann et al. 2021) – but all are hampered by substantial computational burdens that greatly limit their applicability for full-scale commercial applications. We examine and exploit a nested variant of one of these models, i.e., the “seasonal model with dropout” (SMDO) that performs almost as well as Wünderlich et al.’s (2022) full HSMDO model. Its simplicity facilitates a valuable benefit, i.e., simple closed-form expressions for the key quantities of interest for any analyst working with such models. This means we can use standard maximum likelihood methods for parameter estimation and generate critical managerial inferences in near real time. We show how the model parameters can be estimated using simple cohort-level summaries of buying behavior, thereby further reducing the cost of implementation, and enabling the use of a model that allows for seasonal variation in the flow of transactions to a much broader audience.

2 - Predicting and Explaining Customer Purchases: A Joint Model of Attrition, Transaction, and Spending Behavior in Non-contractual Settings
Patrick Bachmann, ETH Zurich, Zurich, Zurich, Switzerland, Jeffrey Naef, Markus Meierer

Explaining and predicting customer purchases is a complex task. In non-contractual business settings, a popular approach to disentangle this complexity is to model three behavioral processes: customer attrition, transaction, and spending. Typically, the former two are estimated separately from the latter. This disjoint modeling approach has regularly raised methodological concerns in the literature. However, its versatility, scalability, and accessibility have not been matched by joint modeling approaches. With both rigor and applicability in mind, this study introduces a novel joint model. Generalizing previous work, this single probabilistic model captures all three processes with well-established statistical distributions. Optionally, time-invariant and time-varying covariates can be included for each process. Further, expressions to derive key managerial metrics are available. The model has a closed-form likelihood for scalable estimation and is made accessible to a broad audience through open-source software. The model's inferential capabilities are assessed through simulation studies, which demonstrate the identification of exogenous and endogenous covariate effects. The predictive capabilities are evaluated using real-world data sets, showing significant improvements in predictive performance compared to state-of-the-art methods. Given these results, this novel model fosters marketers' ability to explain and predict customer purchases.

3 - Effects of Recommendation Timing in Media Customer Lifecycle
Sibo Zhang, University of New South Wales, Sydney, Australia, Yu-Ting Lin, Maggie Dong

The positive effects of media firms’ active participation in customer lifecycle have been widely studied in customer relationship management. Previous research has highlighted the significance of timing in various marketing initiatives, with a primary focus on the pre-consumption context to influence consumer purchasing behavior. However, the temporal effects on post-consumption customer re-engagement, particularly in the context of media content consumption such as after watching a TV series on a media platform, remain underexplored. Specifically, this study investigates the impact of recommendation timing on customer retention (watching frequency) and customer value (video consumption). Utilizing a unique panel dataset from a leading media firm operating on a non-subscription business model, the research examines the effects of immediate versus delayed recommendations following the completion of a TV series, and whether recommending content of the same or different genre nudge customers for an improved experience with the watching platform. We employ Logistic regression and Pareto/NBD models to analyze how recommendation timing and genre choice moderate the effects of completing a TV series on the likelihood of customer retention. Our findings address the persistent question of how to effectively target the right customer with the appropriate genre and at the most opportune moment following media content consumption to proactively boost customer loyalty. Overall, this research contributes to the literature on the temporal effects of personalized recommendations and offers actionable insights for media service providers and marketers.

4 - Valuing Customers in the Insurance Industry: A Joint Model for Claims and Churn
Yumo Dong, Australian National University, Canberra, Australia
Customer lifetime value (CLV) quantifies the present value of projected profits generated from a customer throughout their engagement with a firm. Traditional CLV models focus on modeling revenues and churn, but often neglect modeling costs to serve customers. While in many industries these costs are low and/or steady, in the insurance industry (which contributes over 7% to the global economy) a major cost component is insurance claims, which are highly variable and can be extremely large in size. Existing CLV models are not suitable for the insurance industry because they do not capture costs adequately. Moreover, extant models do not consider the association between costs (claims) and the likelihood to churn - this is an oversight since a policyholder who has a high intrinsic claim risk may be less likely to churn. To bridge this gap, we introduce a cost-based CLV modeling framework. Utilizing a Tweedie distribution for claims along with shared random effects between the claims and churn processes, our model captures unobserved and correlated heterogeneity in claims and churn risks across customers. Applying our proposed model to commercial insurance data, we illustrate how this model can capture latent associations between claims and churn tendencies, enhance predictive CLV performance, and bring novel managerial insights into CLV analyses.
TC09
Meeting Room C3.4
User Generated Content: Insights from Field- and Quasi- Experiments
Contributed Session
Chair: Rupali Kaul, Stanford University
1 - Dynamics in Knowledge Contributions - Evidence from a Field Experiment
Yikun Jiang, University of California-Berkeley, Berkeley, California, United States

With the rise of the digital economy, online knowledge platforms have been gaining popularity. However, it is unclear what motivates individuals to contribute knowledge, whether platform strategies can facilitate individuals' long-term contributions, and how the quality of future contributions may be affected. I develop a stylized model of individual knowledge contribution decisions and generate several theoretical predictions. Based on the model, I design and conduct a large-scale field experiment on one of the largest online Q&A platforms by collecting and treating daily samples. I then track comprehensive data on individuals' subsequent daily posting behavior on the platform. I find that recognition by other users on the platform can increase an individual's probability of contributing by more than 10% of the baseline, and the quality of the contribution does not change on average. The effect is heterogeneous across individuals with different levels of contribution experience and current online reputation. I discuss implications for platform strategy.

2 - The Impact of Social Exclusion on User-Generated Content: Evidence from a Quasi-Experiment on a Mobile Karaoke Singing App
Xiaochi Sun, Nanjing University of Aeronautics and Astronautics, Nanjing, China, People's Republic of, Banggang Wu, Xuebin Cui, Andreas Eisingerich, Zhiling Yang
User-generated content (UGC) is of great value to many online platforms. Extant work has found that several social factors impact users’ UGC production. However, whether and how social exclusion, a ubiquitous phenomenon in people’s social lives, affects UGC production in online communities remains inconclusive. The authors explore these questions by utilizing unique individual-level behavioral data from a popular mobile karaoke singing platform and leveraging a quasi-experiment. The findings show that experiencing social exclusion in a social group on an online platform (i.e., a user is removed from a social group on a platform but can still engage in other activities on this platform) leads to an average 5.8% decrease in an excluded user’s subsequent UGC frequency on that platform, and reduces this user’s tipping and subscription behavior. Furthermore, this negative effect is stronger when a user is excluded from a larger group or a higher-level group. Nevertheless, this negative effect can be mitigated when the excluded user has received greater social support during her/his experiences on the online platform. These findings shed light on the critical role of social exclusion in shaping users’ UGC behavior and offer practical implications for online platforms to manage effective UGC production.
3 - The Causal Effect of Attention and Recognition on the Nature of User-Generated Content: Experimental Results from an Image-Sharing Social Network
Rupali Kaul, Stanford University, Stanford, CA, United States, Justin Huang, Sridhar Narayanan
Social networks rely on sharing engaging content for their users. Since continued production of user-generated content is critical to their success, they have constructed a variety of tools to motivate new content creation, to facilitate user discovery of new content, and to provide attention and recognition to the best user-generated content. Past research shows that such attention and recognition increases the volume of content shared on the networks. But how do these affect the nature of content shared on their platforms? Do they cause creators to share content similar to the ones that received attention and recognition? Or do creators take risks and create different content than the ones recognized? These are the questions we ask in this paper. Our empirical context is an image-sharing social network, where creators share digital art and photography. We leverage a randomized controlled experiment to induce exogenous variation in attention and recognition to specific content. Using a transfer learning-based machine learning algorithm we convert complex images into lower-level features. This allows us to analyze similarities and differences between images. Our main findings are that creators produce and share different content on the social network, than the ones that received attention and recognition. This result is robust to a variety of ways in which we classify image content. Our results illustrate the importance of tools aimed to induce attention and recognition to the creation and development of diverse content by social media creators, and give insights into factors that motivate content creators to create content.
TC10
Meeting Room C3.5
Tourism and Sport Marketing
Contributed Session
Chair: Peng Wang, Tianjin University, College of Management and Economics
1 - Improved Technology Acceptance Model (DL-TAM) for Measuring Tourist Satisfaction on Digitalised Facilities of the Hospitality & Tourism (H & T) Industry
Edward Mariyani-Squire, Western Sydney University, Sydney, Australia, Samanthi Konarasinghe Wiriththamulle Gamage, Francine Garlin

Customer satisfaction is crucial to the survival of tourist organisations and indeed national tourist industries. The emergence of digital technologies to facilitate tourist experiences is, historically speaking, a relatively new phenomenon that may impact on their customer satisfaction. The digitalisation of a tourist organisation is, however, is a time-consuming and costly process; it will be worthwhile for the organisation only if customers are willing to accept and derive satisfaction from it. The Technology Acceptance Model (TAM) and its revised versions – namely UTAUT, UTAUT-2, and UTAUT-3 – have been applied by some researchers to determine if people are prepared to use new technologies and whether they want to do so in the future. However, the importance of Digital Literacy (DL) has not been considered in these models. Studies have shown that the DL of tourists is a key element for the success of digital transformation in the tourism context. On one hand, if a tourist does not have basic skills in using digitalised facilities, they will not obtain the advantages offered by the organisation. On the other hand, a digitally illiterate tourist might consider digitalised facilities as unnecessary burdens to enjoy their travel. This study develops a conceptual model of technology acceptance that incorporates DL to determine the impact of tourists’ acceptance of digital technology on their ‘level’ of customer satisfaction. The model will be tested against survey data to check for confirmation.

Keywords: Digitalisation, Customer satisfaction, TAM, H&T, DL

2 - How do New Technologies Reshape the Travel Experience?
Chin-Ching Yin, National Taipei University of Technology, Tapei City, Taiwan

Metaverse offers unique opportunities for integrating virtual and physical tourism marketing, attracting a wide range of stakeholders. Metaverse not only amplifies social interactions among consumers-to-suppliers and consumers-to-consumers within the tourism industry ecosystem, but its potential also lies in extending the physical realm using augmented reality (AR) and virtual reality (VR) technologies, enabling seamless engagement in both real and virtual environments. Consequently, this broadened access to diverse virtual resources is expected to alter consumer behavior in the Metaverse. However, empirical research on how the Metaverse reshapes consumer behavior and marketing strategies is relatively scarce. Based on existing research on the Metaverse in the field of tourism, this study proposes four dimensions of Metaverse: immersiveness, vividness, interactivity, and interoperability. Since Intelligent Voice Assistants (IVAs) can comprehend human language and respond with intelligent speech, assisting in activities such as information search and schedule management, they are becoming increasingly crucial in enhancing consumer experiences. However, little research has explored the integration of intelligent voice assistants with Metaverse in tourism. To bridge these research gaps, this study adopts Stimulus-Organism-Response (SOR) theory and attitude theory, focusing on how Metaverse and IVAs impact tourists' emotional and cognitive aspects, which in turn affect their travel intentions. The findings will provide a generalizable view of how to integrate Metaverse technologies and IVAs to optimize consumer experiences and thus drive their intention to visit the destination.

3 - How Consumers Use Their Consumer Credit for Travel in Inflation Surge?: Intertemporal Consumption Perspective
Darren Kim, Macquarie University, Macquarie Park, Australia, Jenny Lee

Inflation has once again emerged as a critical concern in recent times, significantly eroding the purchasing power of consumers in both the present and future. This challenge is particularly pronounced of non-essential items, such as travel. Both contingent travelers and service providers face difficulties due to heightened price levels, resulting in an overall reduction in travel demand. Despite these challenges, consumers can smooth their consumption over time by leveraging their credit as a means of borrowing against future income. When consumers perceive that future inflation will be higher, the cost of travel at present is considered cheaper compared to traveling later with potentially higher price levels. The primary objective of this paper is to examine the interplay of consumer credit and inflation on travel decisions. Additionally, we explore how education levels influence this relationship. Utilizing over 10 years of longitudinal data from United States authorities, the findings of this study reveal that consumer credit directly increases the air travel demand, encompassing both outbound and inbound travel. Interestingly, the interaction between the consumer credit and perceived future inflation dramatically enhances the travel demand, particularly among groups with higher education levels. This demographic contributes to higher travel demand at present, benefiting from lower price levels compared to potentially higher prices in the future. Our research aims to gain an understanding of the roles played by consumer credit and inflation in shaping travel demand, as well as the influence of education levels.

4 - The Effect of Temporal Distance on Consumers’ Experiential Baskets in the Context of Trip Planning
Muru Chen, Monash University, Caulfield East, Australia, Harmen Oppewal, Bhoomija Ranjan, Jill Lei

Experiential products such as concerts, festivals or vacations require advance planning for an optimal consumption experience. Tourism as a representative industry offering experiential products is one of the major trade categories growing over last decade reaching USD 3.5 trillion in 2019. Companies hence need to communicate their experiential products with the right message at the right time, to maximise the opportunity of being chosen. Additionally, consumers increasingly personalize and co-create experiences. It is therefore important to understand consumers’ experiential basket planning decisions, and the role of temporal distance to the event in determining their experiential baskets.

Drawing from construal level theory, we propose that consumers planning at a greater temporal distance will select more experiential products into their basket. Construal level will mediate this the effect while it is moderated by information search and depth of processing such that the effect will be mitigated with greater information search and processing. To investigate this proposition we analyse a secondary dataset provided by a tourism platform. The dataset contains information of international trip itineraries consumers planned at different temporal distances to the departure time for trip. Preliminary results confirm the proposed effects that greater temporal distance enhances consumers’ preference for larger basket size. Two experimental studies will follow to test the mediating and moderating effects in more controlled settings. The findings provide insights into how managers should design the experiential offerings in terms of the number, variety and popularity of options when communicating to consumers at different temporal distance.

5 - Unexpected Impacts of Unexpected Games: The Implication of Game Outcomes on Sports Events Advertising
Peng Wang, Tianjin University, College of Management and Economics, Tianjin, China, People's Republic of, Yuxin Huang, Cheng Zhang

Major sports events such Super Bowl or FIFA World Cup are garnering heavy advertising attention. When discussing the effectiveness of sports event advertising, the uncertain nature of game outcomes has been largely overlooked. Our research highlights the uncertainty-of-outcome aspect of sports events and exploits the impact of unexpected game outcomes on the sales of brands that advertised on TV during the broadcasting of the 2014 World Cup FIFA games. We trace the in-store sales data of a major chain supermarket with over 40 branches in China. Our empirical analysis shows that, following World Cup games with unexpected outcomes, the sales of advertised brands drop in terms of both absolute volume and market share. For those brands who did not advertise on TV during the games, the effect varies across categories. When games with unexpected outcomes occur, the sales of unadvertised beers go up while the opposite is true for the category of soda. In addition, we look into the possible mechanisms for the above effects. We discuss the implications of our findings for advertisement design and retailing assortment management.

TC11
Meeting Room C3.6
Social Media Engagement 1
Contributed Session
Chair: Jason Weismueller
1 - Live vs. Post Comments: Empirical Analysis of Their Differential Effects on Consumers' Evaluation of Video Quality
Di Huo, Harbin Institute of Technology, Harbin, China, People's Republic of, Peng Zou, Ning Fu, Qi Wang

Live comments have emerged as a unique form of social interaction on online video platforms. In comparison to post comments, typically located below online videos and posted by viewers after they have completed video-watching, live comments can affect consumers’ video-watching experiences in real time, potentially leading to different influences on their evaluations of video quality. Despite their potential significance, limited research has investigated the differential impacts of live comments and post comments on consumers' evaluation of video quality. This study examines these differential impacts by analyzing a dataset consisting of 667 online videos collected from Bilibili, a prominent Chinese online video platform, between September 2019 and July 2020, along with their corresponding live and post comments, totaling in approximately 2 million textual data. Our findings demonstrate substantial differences in their main effects between live comments and post comments, and in the moderation effects of video types between informational videos and emotional videos. As our data span periods before and after the outburst of the Covid-19 pandemic, our results also reveal interesting differences in the impact of live comments and post comments before and after the pandemic. Overall, these results provide significant theoretical insights on the underlying mechanisms driving the diverse impacts of various forms of social interactions, as well as managerial implications regarding how online video platforms facilitate the engagement among viewers and between video creators and viewers.

Keywords: live comments; video user behavior; informational videos, emotional videos; COVID-19 pandemic.
2 - The U-shape Effect of Online Exposure to Nature on Customer Engagement
Jiaqi Wang, Harbin Institute of Technology, Harbin, China, People's Republic of, Peng Zou, Bo Yu, Hao Qin, Furong Wang

The increasing popularity of online information and communication technology (ICT) devices has shifted customers’ contact with nature from offline to online exposure. Social media includes elements such as bloggers’ social influence, image clarity, text sentiment, blog length, and online interactions with people that are inaccessible offline. The impact of blog content to customer engagement on social media has became an essential topic in marketing fields. This study examines the relationship between online exposure to nature and customer engagement on social media from the perspective of environmental psychology, using the data from Sina Weibo and the machine learning models. We find a significant U-shape correlation between online exposure to nature and customer engagement. Additionally, this research examines the moderating effects of social influence and image clarity on the relationship between online exposure to nature and customer engagement. The U-shape relationship is weakened by social influence and image clarity, bloggers with higher social influence and images with higher clarity have the potential to distract customers’ attention from online exposure to nature. Our study indicates that online exposure to nature significantly differs from offline exposure to nature, both in theoretically and in terms of results. Our paper enriches the literature on visual content and enhances the understanding of customer engagement on social media. The results also reveal the implementation on social media marketing strategies with online exposure to nature-related image.

Keywords: online exposure to nature, social media engagement, environmental psychology, attention restoration theory, stress reduction theory

3 - More Anthropomorphism, More Engagement? The Effect of Matching Anthropomorphic Cues and FGC Ad Appeal on Social Media Engagement
Zihui Mao, Southwestern University of Finance and Economics, Chengdu, China, People's Republic of, Xiaorong Fu, Junjie Kuang

Firms’ utilization of anthropomorphic strategies on social media has shown potential for enhancing consumer engagement. Brands typically deliver Firm-Generated Content (FGC) through two types of anthropomorphic cues: visual and linguistic. However, limited research explores the effect of these two anthropomorphic cues on consumer engagement under different ad appeal (information-focused vs. emotion-focused). Moreover, there are still conflicting conclusions regarding the comparative effectiveness of using two types of anthropomorphic cues versus using only one. This study aims to investigate the effect of matching anthropomorphic cues and FGC ad appeal on consumer engagement as well as the interactive effect of visual and linguistic anthropomorphic cues based on dual processing theory. By using Recurrent Neural Network (RNN) to analyze the real social media data set and conducting two experiments, the findings show that visual anthropomorphic cues matched with information-focused content and linguistic anthropomorphic cues matched with emotion-focused content led to higher consumer engagement. This effect is respectively mediated by information credibility and emotional arousal. Importantly, when the first anthropomorphic cue is already matched with the ad appeal of FGC, adding a second anthropomorphic cue does not lead to additional increases in consumer engagement. These findings provide practical guidance for companies to effectively utilize different anthropomorphic cues in social media marketing communications.

4 - How Value-expressive Language Shapes Brand Posts on Social Media
Jason Weismueller, The University of Western Australia, Perth, Australia, Jo Sneddon

Consumer expectations for brands to demonstrate sustainable practices has spurred a growing trend of brands posting Sustainable Development Goal (SDG) related content on social media. This research focuses on how brands should design their social media posts about SDG-related issues and initiatives to attract consumer engagement and improve brand attitudes. Drawing on Schwartz's Theory of Basic Human Values, we investigate whether value-expressive social media posts (i.e., posts using words aligned with specific value dimensions like universalism) result in higher engagement and better brand attitudes than less or no value-expressive ones.

We argue that value-expressive language increases authenticity, which, in turn, can boost engagement and improve brand attitudes. Moreover, focusing on a specific set of value dimensions may be the most effective approach depending on the SDG. For instance, language embodying universalism might be especially impactful in discussions related to environmental SDGs, such as Climate Action.

In Study 1, we use CrowdTangle and University of Auckland's SDG keyword mapping to collect SDG-related Facebook posts from over 100 brands across various industries. These posts are then analysed using natural language processing (NLP) and Ponizovskiy et al.’s values dictionary to differentiate between value-expressive and non-expressive content. In Study 2, using Prolific, we follow up with a controlled experiment (n = 300) in which we expose participants to brand posts that are either value-expressive or not before assessing their engagement levels and brand attitudes.

Our findings will provide several insights that make both practical and theoretical contributions concerning brands' social media communication.

TC12
Meeting Room C4.1
Marketing and Policy 2: Financial Decision Making
Invited Session
Special Sessions
Chair: Mike Palazzolo, University of California - Davis, Davis, CA, United States
1 - Modeling Scale Attraction Effects: Optimal Laddering in a Charity Panel Field Experiment
Fred Feinberg, University of Michigan, Ross Business School, Ann Arbor, MI, United States, Kee Yeun Lee
Multi-item numerical scales are ubiquitous in marketing. Yet firms and researchers utilizing such scales cannot assess how quantities comprising the scale jointly “pull” eventual response, in timing nor degree, or how these effects may differ across consumers. To this end, we develop a novel model of individual-level scale attraction, where effects accrue to sway consumers from a latent reference quantity. The model is estimated for a multi-year donation panel from a large national charity coupled to a field experiment where points on an “appeals scale” are manipulated. Temporal carryover to utility “shocks” is assessed via a Bayesian solution to a classic problem in panel data econometrics, estimating not only the contemporaneous error correlation of the standard Tobit II (“Heckman model”), but serial correlations for both donation incidence and amount, finding it only for the former. The model allows measurement of heterogeneity across donors in upward vs. downward scale point attraction and incidence patterns (e.g., seasonality), and accommodates multiple operationalizations of internal and external referents to proxy for donor history and scale points, respectively. Results reveal that scale points exert substantial attraction effects; that these vary markedly across respondents; that they are (non-uniformly) referent-influenced; that donors more readily adjust amounts downward than up; and that scale point influence wanes with distance. Importantly, we show how to use the model to assess optimal "laddering": how quickly to increase points across the scale, and the scale itself over time.
2 - Nonparametric Bayesian Counterfactual Inference for One-Shot Field Experiments: Evidence from a Large-Scale Donation Drive
Gwen Ahn, Leeds School of Business, Boulder, CO, United States, Kee-Yeun Lee, Arnaud De Bruyn, Fred Feinberg
“Appeals scales” (suggested amounts from which donors can choose) are standard solicitation tools used by charities. Designing superior, individual-level scales requires understanding their effects on donation incidence and amount. Because self-selection artifacts limit the usefulness of field data alone, charities sometimes supplement historical giving data with experiments. Here, we analyze a fully randomized large-scale field experiment that orthogonalized major components of the appeals scales: minima, maxima, and steepness.

Yet generating plausible counterfactuals requires overcoming the "one-shot" nature of the experiment. To that end, we first develop a multi-output Gaussian Process (GP) model that jointly accounts for donation incidence and amount, nonparametrically “smoothing” over donors' histories to yield joint posteriors for incidence and amount, and thereby overall expected donation outcomes. To evaluate the causal effects of appeals scales, we compare counterfactual donation outcomes in a generalized additive model (GAM) setting, for full nonparametric inference. By proposing optimal appeals scales while taking donor and temporal heterogeneity into account, charities can tailor their individual requests for subsequent donation drives.
3 - Enabling Consumer Financial Data Exchange in Marketing Research and Practice
Linda Salisbury, Boston College, Chestnut Hill, MA, United States, Simon Blanchard, Kelly Martin
Consumer financial data (e.g., transactions, account balances) are valuable for marketing research and practice, but they are also sensitive and difficult to obtain. Existing methods of collecting consumer financial data, such as self-reports or acquiring data from third-party data providers, have limitations in terms of accuracy, reliability, and privacy. To address these challenges, we develop an application that allows researchers to verify the eligibility of potential research participants based on financial information from their financial institutions. The app leverages the services of a digital finance platform (Plaid) to establish a secure connection between consumers’ financial accounts and researchers’ applications, without storing or sharing any sensitive data that would compromise consumer privacy. The app enables researchers to conduct studies with participants who meet specific financial criteria (e.g., having a student loan balance over $5000, purchasing from Uber twice in the past 30 days), which can improve data quality and enable novel research questions. We believe that the app has significant potential
to advance marketing research and practice. By providing a secure and efficient way to access verified consumer financial data, the app can enable marketing researchers to test hypotheses with higher data quality and validity, as well as explore new research topics and domains that rely on financial data. Ultimately, the app can benefit practitioners who seek to understand and target consumers based on their financial profiles, preferences, and behaviors.
4 - Game Over? Assessing the Impact of Gamification Discontinuation on Mobile Banking Behaviors
Mike Palazzolo, University of California - Davis, Davis, CA, United States, Simon Blanchard
Gamification (incorporating game design elements in non-game contexts) has gained significant attention in marketing. While previous research has primarily focused on short-term effects and customer acquisition, this study explores the discontinuation of gamification in financial services and how it affects previously incentivized behaviors. We leverage a dataset from a prominent Latin American bank that once introduced a reward module in its mobile app. The module incentivized behaviors through game elements such as prize wheels, raffles, and points. Unexpectedly, the reward module was temporarily stopped due to a government shutdown, providing a natural experiment to measure the causal relationship between the reward module's features and user behaviors. Using age-period-cohort models to identify the causal effect of the shutdown, we find that the reward module incentivized desired behaviors, but these behaviors declined after the discontinuation. Furthermore, different cohorts demonstrated the importance of maintenance for sustained effects, particularly in transactional activities (bill payments and loan repayments). We also find that both cash rewards and module engagement played distinct roles in influencing user behavior. Finally, we examine the impact of reward module outcomes on engagement, highlighting the effectiveness of different game-design elements. This study provides implications for the design and implementation of gamification strategies.
TC13
Meeting Room C4.2
Game Theory for (Online) Retailing
Contributed Session
Chair: Yuichiro Kanazawa, International Christian University, 1-4-20 Kasuga-cho, Nerima-Ku, 179-0074, Japan
1 - Designing Marketplace Analytics for Sellers: Platform-offered vs. Third-party Providers
Yi Liu, University of Wisconsin - Madison, Madison, WI, United States, Fei Long

The rapid growth of e-commerce giants and the abundance of data has spurred the development of AI-powered analytics services, such as competitive intelligence and automated pricing, which enables marketplace sellers to make informed, data-driven decisions. Third-party providers (e.g., Jungle Scout) compete with platforms themselves (such as Amazon's brand analytics) in offering marketplace analytics. Yet we are witnessing platforms adopt various strategies in sharing data with third-party analytics providers, ranging from restrictive to permissive (e.g., permitting data-scraping) with some even actively facilitating (e.g., API sharing). In this paper, we ask why and how an e-commerce platform may benefit from sharing data with third-party providers, despite the platform's inherent advantages in data access and the capability to design its own analytics services for better control over sellers' actions.

We find that the platform maximizes its own sales commissions by providing an over-optimistic analytics service (which means under-reports the market competition levels). When market competition is moderate, this may lead to sellers' reluctance to use the platform's analytics service, resulting in a lose-lose situation, and prompting the platform to share data with third-party providers. However, in highly competitive markets, over-optimistic predictions can actually benefit sellers. In this case, the platform adopts a more restrictive data-sharing approach. Furthermore, when the platform can control the accuracy of third-party analytics via API access, it shares data in a larger parameter space, and it shares broader data when market competition intensifies.

2 - Which Free Rider Would Be Better? The Strategies and Decision-Making of Multiple Stakeholders in Anti-Piracy Initiatives
Xiaotong Du, Tianjin University, Tianjin, China, People's Republic of

Problem definition: In the digital marketplace, retailers and online sellers relying on e-commerce platforms offer both physical and electronic versions of digital products, providing a breeding ground for e-piracy. Currently, sellers and platforms grapple with the escalating challenges posed by e-piracy. We compare the anti-piracy models employed by the online seller and platform, examining the diverse anti-piracy measures taken by various entities. Academic/practical relevance: Different anti-piracy subjects exhibit varying action efficiencies and impact retailers, online sellers, and platforms differently. Methodology: We employ a game-theoretic model that captures the retailer and seller’s sales and anti-piracy decisions. Results: We reveal the retailer as the primary victim of e-piracy, potentially gaining the most from anti-piracy efforts. The structural effect of anti-piracy measures on overall merchant profits remains robust, with all merchants favoring the online sellers' model only when the platform's commission rate is low. We further characterize necessary and sufficient conditions for the effectiveness of anti-piracy measures, which can be achieved by comparing the relationship between the two anti-piracy models in terms of anti-piracy investment and piracy market share. Managerial implications: These results provide crucial insights for platforms, online sellers, and governmental entities, guiding regulatory and business decisions to enhance the effectiveness of anti-piracy initiatives.

3 - The Influence of Copycat Appearance Period on the Brand-name Supply Chain with Different Sales Models
Jie Miao, Tianjin University, Tianjin, China, People's Republic of, Zhe Yang, Zhaofang Mao, Yunchuan Liu

Retail platform sales are booming. Many brand-name suppliers choose reselling or agency selling models to promote and sell products through the platform. In reselling models, the retail platform acts as a retailer. Therefore, in the non-single-period game scenario, to obtain lower wholesale prices and higher marginal revenue, the platform usually orders more goods than needed in the current period to hold strategic inventory and negotiate with the supplier. There is not only a battle for profits within the brand-name supply chain but also a threat from copycats. In various industries, these copycat products appear at different time points. This prompts us to examine the pricing and inventory incentives of the brand-name platform supply chain in response to the entry of counterfeiters. We construct a two-period game model in which a brand-name supplier and a copycat compete on a platform through different selling models. We anticipate the platform may resist or delay the appearance of copycat products, despite the potential benefits of a new revenue stream, market competition, and reduced wholesale prices. Besides, strategic inventory can still enhance the efficiency of the brand-name supply chain in the reselling model. However, we expect it may not necessarily be advantageous for consumers.

4 - Examining Bargaining Power in the Distribution Channel under Possible Price Pass-through Behaviors of Retailers
Yuichiro Kanazawa, International Christian University, Nerima-Ku, Tokyo, Japan, Tomoki Matsumoto, Tomohito Kamai
We investigate the determinants of relative power within a distribution channel by incorporating the price pass-through behavior of a retailer into Nash bargaining between a retailer and a manufacturer. We assume manufacturers can observe such behavior as well as the retail price itself. We first derive the retail margins from a retailer maximizing its profit, assuming that the retailer anticipates the manufacturer’s profit-maximizing response to the price pass-through behavior. The manufacturer margins are derived from the generalized Nash bargaining with the retailer, where the parties negotiate the wholesale price. Finally, we identify conditions under which the bargaining power parameter is well-defined based on the values of retailer and manufacturer margins and the parameter describing the degree of price pass-through. A toy Bayesian estimation example using daily scanner panel data for canned tuna at a single retail chain in western Tokyo, Japan, demonstrates the applicability of our model. We contrast this result with those obtained under the framework proposed by Draganska et al. (2010) without the price pass-through behavior and the retail price observability by manufacturers.
TC14
Meeting Room C4.3
Marketing Communication
Contributed Session
Chair: David Fang, University of Toronto, Toronto, ON, Canada
1 - Perception of Dissonance and Self-persuasion: A Study of the Impact of Technological Discomfort on Multidimensional Consumer Engagement in Computational Advertising
Zhonghuan Wu, South China University of Technology, Guangzhou, China, People's Republic of, Chunlin Duan
In terms of advertising content, consumers can benefit from customized advertising content. However, the accuracy of the content raises consumer concerns about data privacy. This psychological ambivalence on consumers may create distrust in advertisers and advertising platforms, making them more cautious about engaging in advertising campaigns. This is particularly relevant to Gen Z consumers, who are perceived as tech-savvy, narcissistic, and vulnerable when using SNSs. Firstly, sub-study one unravels the internal mechanisms of consumer technological dissonance generation, experience, and self-persuasion in consumers' perceptions of computational advertisements from the perspective of cognitive dissonance through exploratory analysis, employing in-depth interview method, big data analysis with k-means algorithm; and based on the results of the analysis, propose relevant propositions to construct the overall framework of this study. Second, sub-study two explores the impact of content customizability technology on the sense of technological dissonance from the perspective of technology acceptance, with perceived content usefulness and privacy issues playing a mediating role in this process, and validates the hypotheses of study two through big data analysis. Our empirical study in China suggests that technology customizability negatively affects multidimensional consumer engagement. When the critical role of the self in the dissonance of Generation Z SNS behaviors is recognized, fragile narcissism is found to display nuances of psychological dissonance caused by privacy concerns. Our findings strengthen the discussion on the customizability of AI technologies by pointing out how advertisers can improve targeted marketing and ensure the well-being of Gen Z consumers.
2 - Perception of Artificial Intelligence in Email Marketing
Herbert Endres, University of Regensburg, Regensburg, Germany, H G

This paper examines the perception of artificial intelligence in email marketing. The aim of the paper is to first understand which marketing strategies prevail in the digital environment and how these differ from marketing strategies in the non-digital environment. With the help of an e-mail experiment, an attempt is made to determine those factors that significantly influence the value of an online course after an AI-based e-mail communication compared to an e-mail communication carried out by a human. Significantly influencing the value of the course, only the intention to enroll in the course can be identified after an AI-based email response to student inquiries. Nevertheless, the present work helps to identify which other factors an AI-based technology should consider in the future when interacting with the target customer in order to positively influence the value of a product from the customer's point of view.

3 - Can Messaging Lead to More Effective Subscription Sign-ups? Insights from a Field Experiment
Kirthi Kalyanam, Santa Clara University, Santa Clara, CA, United States, Raphael Thomadsen, Nan Zhao
Subscriptions of physical goods that are also available through conventional retailers are becoming increasingly popular. The dominant strategy currently employed to gain sign-ups is to give price discounts for people who subscribe. However, price discounts may not be the most effective motivation – in either the short or long run – for getting consumers to sign up for a subscription, since the message does not speak to a motivation of repeat buying. We consider alternative non-price motivations firms could appeal to in order to sign up and stay with the subscription. We work with a major retailer that sells pet products to launch a four-week field experiment where we randomize messages in email advertising. We find that messages that highlight a reminder motivation consistently outperform price messages overall. We also find that price and non-price messages work differently with customers who have prior multi-channel versus brick-and-mortar experience with the firm. Price messages work better with customers who have no prior experience with the firm, highlighting that different messages are effective in acquiring customers for subscriptions versus modifying the approach (Subscriptions versus on time purchases) that they use to transact with the firm.
4 - Short and Sweet or Short and Sour: The Influence of Text Message Abbreviations on Relational Outcomes
David Fang, University of Toronto, Toronto, ON, Canada, Yiran Zhang, Sam Maglio

As social interactions increasingly move to digital platforms, deciphering factors that enhance or diminish these virtual interactions becomes vital. Texting abbreviations are pervasive in digital communication, yet its impact on interpersonal communication outcomes remains unclear. The present study presents a novel examination of the influence of texting abbreviation use on relational outcomes (e.g., does using texting abbreviations influence how sincere a texter is perceived to be?). We investigate the effects of texting abbreviations on perceived sincerity and likelihood to respond, testing two competing hypotheses: (a) abbreviations impede relationship development by reducing the perception of effort, and (b) abbreviations foster closeness due to heightened informality. The former hypothesis is consistently supported through ten pre-registered studies (N = 4984) using mixed methods (e.g., field experiment, lab experiments, field survey, and archival analysis of 686 Tinder users’ conversation history spanning 37 countries), revealing that abbreviations negatively impact relational outcomes through perceptions of less effort put into conversations. We also found that factors such as the level of familiarity between communicators, familiarity with texting abbreviations, the type of abbreviations used (i.e., phonetic vs. initialized), and the length of the text exchange did not attenuate our observed effects, providing a more nuanced understanding of the phenomenon. Marketing practitioners focused on digital impression management can gain valuable insights from this study.

TC15
Meeting Room C4.4
Research Based on AiMark Data: Unlocking Insights from International Household and Retail Panel Data
Invited Session
Special Sessions
Chair: Bernadette van Ewijk, University of Amsterdam Amsterdam Business School
1 - Why and When to Launch New Products During a Recession
Koen Pauwels, Northeastern University, Newton, MA, United States

Do new products launched during a recession perform better? Does the severity of the recession matter? Are products more successful when launched earlier or later in a recession? These are all questions of managerial importance that as yet remain unanswered in the extant marketing literature. The authors analyze 2 datasets: 1) 8,981 product launches in 20 United Kingdom fast-moving consumer goods categories over 18 years, provided by AI Mark and 2) 1,071 product launches in the United States automotive market over 63 years. The results reveal products launched (a) during a recession and (b) later rather than earlier in the recession survive longer, while more severe recessions are associated with shorter survival. This paper thus enriches marketing theory on recessions by conceptualizing and quantifying timing effects on new product launch success. For managers, the results demonstrate the benefits of countercyclical launching of new product during recessions and to market proactively in such economic conditions.

2 - Evolution of Marketing Mix Effectiveness over Time: A Global Investigation
Katrijn Gielens, University of North Carolina-Chapel Hill, Chapel Hill, NC, United States, Jan-Benedict Steenkamp
There is considerable evidence on the effectiveness of marketing mix instruments, especially price but we know very little about the temporal dynamics of the effectiveness of various marketing mix instruments. Is marketing mix effectiveness increasing or decreasing over time? Is this dependent on brand characteristics like price positioning and brand architecture? What is the role of brand internationalization? Can we observe systematic differences between countries? Which economic and cultural factors explain cross-national differences? We examine these and other questions adopting the “Empirics-First Approach to Relevant Knowledge Generation,” for three key marketing mix instruments—price, distribution, and assortment (line length). We draw empirical insights from the largest-scale econometric study on marketing mix effectiveness ever conducted, involving 15,000+ brands in 80 CPG categories over a period of 5-15 years in 34 countries in Asia (China, India, Indonesia, Thailand, Taiwan, Turkey, Vietnam), Western Europe (Austria, Belgium, Denmark, France, Germany, Ireland, Italy, Netherlands, Spain, Sweden, UK), Eastern Europe (Czech Republic, Hungary, Poland, Romania, Russia, Slovakia), North America (US, Mexico, Central American countries (combined)), and South America (Argentina, Bolivia, Brazil, Chile, Colombia, Ecuador, Peru). Our data include both the G7 and BRIC countries. Theoretical and managerial implications are drawn.
3 - How Loyalty Programs Influence Consumer Response to Marketing Crises
Samuel Staebler, Tilburg University, Tilburg, Netherlands, Hannes Datta, Marnik Dekimpe
While loyalty programs (LPs) remain popular among both retailers and consumers, research indicates that LPs often do not live up to managers’ expectations. But maybe there are also some more hidden benefits that only materialize when the going gets tough. Specifically, do LPs help retailers when they get caught up in a marketing crisis? For practitioners the answer is not apparent. Some business analysts argue that loyalty programs could indeed help firms when involved in crises: “Even when their image may be in bad light, loyalty programs help keep a customer around” (Danielle Brown, LP consultant, Bizreport 2016). Still, there are also recent industry reports (see, e.g., Wharton Retail Report 2021) that argue that these indirect benefits should not be taken for granted, and that LPs and LP membership may actually backfire when problems occur in the customer journey.
To address this question, the authors conduct, adopting an Empirics-First approach, a large-scale study in the Dutch retail industry. The study investigates how more than 11,000 households (who are either a member or non-member of various retailers’ LP) respond to close to 50 different crises that took place in a 9-year period. The results indicate that LP membership buffers against a crisis’ negative effects not only in terms of a smaller reduction in purchase volume, but also in terms of a higher effectiveness of various marketing-mix-related recovery efforts.
4 - The Impact of Product Strategies on Market Share and the Moderating Role of Country Characteristics
Bernadette van Ewijk, University of Amsterdam Amsterdam Business School, Amsterdam, Netherlands, Harald van Heerde, Marnik Dekimpe, Jan-Benedict Steenkamp
Companies worldwide invest substantial financial resources to implement marketing strategies to acquire or retain market share for their brands. Whilst existing academic literature has predominantly focused on the impact of price and advertising on brand performance (see e.g., the meta-analyses by Bijmolt et al (2005) and Sethuraman et al (2011)), research by Ataman et al (2010) highlights that the Product component is the most important element of the marketing strategy. While variations in line length elasticities may exist across countries (Datta et al 2022), it is imperative for brands to not only consider the size of their offering (line length strategy), but also to address the composition of their product line. This includes determining the proportion dedicated to new products (innovation strategy), varying attributes across products (specialization strategy), and differentiating their products’ attributes from competitors (competitive strategy). To formulate effective global product strategies, firms need insights on an international basis. This research adresses the following questions. First, what are the effects of the four product strategies on market share? Second, what is the relative effect of each strategy? Third, how do country characteristics influence the effectiveness of product strategies? Leveraging a large-scale, consistently collected data set from AiMark covering 50+ consumer electronics and appliances categories in 25+ countries, the authors apply a hierarchical market share attraction model to estimate the various product elasticities (controlling for other marketing mix instruments). Further, the study explores how these elasticities vary systematically with countries’ institutional (economic, regulative and cultural) context.
TC16
Meeting Room C4.5
Livestreaming 3
Contributed Session
Chair: Kane Koh, University of Melbourne
1 - Leveraging Influencer Networks for Successful Event Marketing
Yiwei Li, Lingnan University, Hong Kong, Hong Kong, Hong Kong, Jisu Kim, Robert Palmatier

With the exponential rise of influencer marketing over the past decade, marketers are gradually shifting from passive observers to active collaborators of influencers. This evolution gives rise to a modern influencer-event marketing strategy: leverage social media influencers to create events aimed at improving customer reach and engagement. Despite prevalence of influencer events in industry, our knowledge on this exciting field remains somewhat limited. By engaging with a multitude of influencers, firms can access broader follower network, and bolster the event’s credibility through consistent endorsements from influencers. As such, we posit that influencer-event is effective in improving customer spending. Yet the question remains: what specific strategies can an event employ to fully harness the power of influencers? The answer lies in a deeper comprehension of influencers' social network, as events essentially provide unique networking and relationship-exchanging opportunities. Drawing on social-network theory, we propose to delineate the effects of network, from nodes (network degree), ties (relational/structural embeddedness), to structure (network density), on influencer-event effectiveness. We employ quasi-experiments to provide proof of concept, utilizing broadcast data of 1,000 influencers from a major livestreaming company in South Korea. Our preliminary analyses suggest that influencer-event improves sales performance. Firms can enhance event effectiveness by strategically targeting influencers with high-level network degree, relational embeddedness, and density. Particularly, soliciting influencers based on high-level structural embeddedness is not always good, as it exerts an inverted U-shaped effect on influencer-event effectiveness. Eventually, this research will provide multi-disciplinary implications in fields including influencer, event, and social-network marketing.

2 - Modeling Creative Selling in the Streaming Economy: Leveraging Streamers' Verbal, Vocal, and Visual Features Through Predictive, Interpretable, and Generative AI
Yu (Eric) Kou, Temple University, Philadelphia, PA, United States, Xueming Luo, Zheng Fang

In the context of hyper competitive livestreaming platforms such as Twitch and TikTok, the creative selling ability of a streamer becomes of unprecedented significance. However, it is challenging to scientifically model streamer creative selling in an automatic, scalable, and theory-consistent manner. This research develops a method to measure streamer creative selling by customizing multimodal transformer-based deep learning algorithms for livestream video data. Our model outperforms a host of deep learning benchmarks and reveals that it is not only the multimodal representations of lower-level verbal, vocal, and visual features but also their crossmodal interactions (verbal-vocal, verbal-visual, vocal-visual, and verbal-vocal-visual) that capture creative selling. We validate our algorithm by showing that the predicted creative selling score is correlated with the higher-level four theoretical creativity constructs of verbal originality and appropriateness, vocal arousal, and visual body motion as well as their interaction terms in an upstream analysis. Further, in a downstream analysis, we validate that streamers who have higher creative selling scores tend to generate more product sales, and interestingly this relationship is amplified for high-priced products and hedonic goods. Moreover, a follow-up field experiment, using ChatGPT to generate stimuli videos with exogenous variations, corroborates the causal impact of creative selling on customer responses, providing more external validity of our algorithmic measure of creative selling. Platforms can leverage our algorithm to rank streamers based on creative selling scores for advertising and promotion activities, and streamers can benefit from it by improving their verbal, vocal, and visual cues of creative selling to increase sales.

3 - Diligent Teller or Skillful Teaser? An Empirical Analysis of Influencer Pre-announcement Strategies in Live-streaming
Zifei Li, Harbin Institute of Technology, Harbin, China, People's Republic of, Shaohui Wu

The live-streaming industry, relying on User Generated Content (UGC) platforms, has tremendously revolutionized marketing and consumer behavior. Two conventional types of live-streaming influencers are active on UGC platforms: the “just- chatting” influencer for entertainment or educational content, and the marketing influencer for real-time selling. An evident trend has emerged in the realm of marketing live-streaming influencers that sometimes they conduct “just-chatting” live-streaming sessions as pre-announcement or prelude to their subsequent marketing sessions for reasons, deviating from continuous selling sessions. Our research aims to elucidate how “just-chatting” sessions influence the performance of marketing sessions at sales and viewer traffic levels. The research was supported by transaction data from over 46000 live-streaming sessions by 482 influencers at Douyin platform within 90 days. To mitigate potential self-selection bias, we introduced propensity score and developed a comprehensive model. The model considers preannouncement, number of goods, session duration, and accumulated session amount, in relation to sales, viewer count, and Gross Profit Margin (GPM). We found that “just-chatting” live-streaming sessions have negative effect on live-streaming performance at levels. Furtherly, the mechanism analysis indicated noteworthy results when interacting with weekday, time of day, live-streaming frequency, and follower amount. In the past, influencer performance and live-streaming features have proved to be critical to viewer stickiness and purchase intention in real-time; however, our research extends these findings to trans-sessions dynamic, offering a more comprehensive understanding of influencer and viewer interaction over time.

Keywords: Live-Streaming industry, Influencer market, Propensity Score, Live-streaming strategy, Purchase intention, Viewer engagement

4 - Can Brands use Toxicity to their Advantage? Toxicity on Engagement in Live Streaming
Kane Koh, The University of Melbourne, Melbourne, Australia, Liliana Bove, Jill Lei, Fang-Chi Lu

The proliferation of online toxicity in recent years warrants novel moderation strategies. While the negative outcomes of toxicity are well-known, the entertainment value it can provide brands who harness it, and its impact to subsequent behavior, is lacking, despite successful real-world applications (e.g., Wendy’s). This research explores toxic interactions as an online community management strategy that optimizes paid (subscribing, tipping) and non-paid (commenting) behavior. Supported by machine learning (audio extraction, speech-to-text algorithms, natural language processing) on unstructured data (videos, chat logs), the authors estimate a panel vector autoregression (PVAR) model on 1,211 Twitch streams and 349,211 minutes. The findings show that streamer and viewer toxicity positively associate, with the effect of streamer (viewer) toxicity on viewer (streamer) toxicity persisting for 11 (17) minutes. In addition to highlighting the nature of toxicity exchange, this shows that streamer toxicity is more enduring than viewer toxicity. Additionally, streamer and viewer toxicity positively associate with tips, subscriptions and comments, with these effects persisting for 15 to 17 minutes, suggesting that toxicity can motivate positive subsequent behavior. Interestingly, differential effects to various types of subsequent behavior are observed, where streamer toxicity has a larger (smaller) effect on tips (subscriptions and comments). Further moderation analyses show that these results vary with streamer characteristics, including gender, genre, channel size and influence. This research challenges conventional wisdom on toxicity’s negative effects and contributes to brand community management knowledge by considering brands as another toxicity source, thus extending toxicity as more than mere customer misbehavior.

TC17
Meeting Room C4.6
Health Care Marketing: Prescription Drugs
Contributed Session
Chair: Seethu Seetharaman, Washington University in St. Louis
1 - Getting the Price Right: Heterogeneous Pricing of Over-the-Counter Drugs
Yeowoon Park, The University of Texas at Dallas, Richardson, TX, United States, Joonhwi Joo, Yufeng Huang

We study market power, competition, and pricing strategies in the Over-the-Counter (OTC) drug industry. Pharmaceutical companies and retailers chase the opportunity to sell non-prescription drugs while consumers are demanding new levels of innovation (e.g., formulations). Events such as Prescription to Over-the-Counter (Rx-to-OTC) switches or patent expiration provide plausibly exogenous timing and opportunities for entry by competitors. These events introduce new competitors to the category and potentially alter consumer behavior, creating opportunities for strategic pricing.

In this study, we mainly focus on pricing strategies in the OTC drug market, analyzing over 90 formulations across six drug classes using the Nielsen-Kilts Retail Scanner Database during 2006-2020. We broadly aim to answer two research questions: 1) How do exogenous events like Rx-to-OTC switches and generic entry impact pricing strategies in the OTC market? 2) What role do factors such as formulations, target consumer heterogeneity, and market concentration play in influencing price changes during these events?

Preliminary findings suggest that many branded drugs do not lower prices upon the entry of their generic competitors, contradicting the predictions from conventional theories of pricing strategies and market power. Pricing patterns differ substantially across branded-drug producers, major pharmaceutical companies that sell their own-branded generics, and pharmacy-branded private labels.

2 - Does Advertising Content Regulation Work? Evidence from Canadian DTCA
June Lee, McGill University, Montreal, QC, Canada, Demetrios Vakratsas

While many studies on advertising regulation focus on bans, not much is known about the effects of regulated restrictions on advertising content. Canadian regulations on Direct-to-Consumer Advertising (DTCA) provide a unique opportunity to evaluate such effects since they allow only two types of DTCA with different content: 1) help-seeking messages which discuss a disease and seek to inform patients who are not aware of the existence of a treatment, and 2) reminder advertisements which are allowed to mention only the name of the advertised drug. Using this context, the authors assess the effectiveness of advertising content regulation by examining whether the two types of ads can produce distinct effects as suggested by their respective content restrictions. The empirical analysis relies on a utility-based diffusion model, using data on one of the most extensive Canadian DTCA campaigns by a pioneering drug. The findings indicate that both help-seeking and reminder advertisements affect new as well as repeat adopters of the drug. This suggests that: a) the two types of ads fail to produce distinct effects, and b) regulations inadvertently encourage anti-competitive, persuasive advertising (Ackerberg 2001). Based on this assessment, the authors conclude that Canadian DTCA regulation is not effective and provide recommendations for regulators.

3 - IDN: Impeding Drug Newcomers? How Vertical Integration in Healthcare Industry Affects Physician New Drug Adoption
Siyi Yu, Purdue, West Lafayette, IN, United States, Hongju Liu, Qiang Liu

The consolidation of independent physicians into integrated (healthcare) delivery networks (IDNs) marks an institutional change in the U.S. healthcare system. Existing studies on the topic have documented the effect on healthcare outcomes and expenditure, but the mechanism of how joining an IDN influences physician behavior remains an open question. Leveraging a large-scale physician affiliation and prescription dataset during 2015 – 2016 for an anti-asthmatic drug, this paper investigates how consolidating physicians into IDNs influences physicians’ new drug adoption behaviors. The paper first shows reduced-form evidence with survival models that independent physicians are significantly more likely to adopt the new drug than their IDN-affiliated counterparts within 8 months after the first launch; and that independent and IDN physicians experience different effects within their cohort networks. The paper then applies structural model to examine the reason for IDN-affiliated physician’s inertia for adopting the new drug, and proposes that IDN-affiliated physicians’ expectation for peer knowledge sharing explains the initial inertia for new drug adoption.

4 - Metaphors for Dry Eyes: How the Attention to Metaphors in Patients’ Narratives on their Experiences with Dry Eyes Could Make Commercials for Eye Drops More Effective
Carina Rasse, University of Klagenfurt, Austria, Klagenfurt, Austria, Ralf Terlutter, Sandra Diehl

Metaphors are commonly used in health marketing campaigns and commercials (e.g., Lazard et al., 2016; Spina et al., 2017). However, metaphors do not always have the expected positive effects on the consumers as they may be misunderstood or misinterpreted due to their ambiguous nature (Landau et al., 2018; Fatehi et al., 2022). In this project, we compare the use of metaphors (1) in eye drop commercials for dry eyes to (2) metaphors that patients use when they talk about dry eye symptoms. Data was obtained by, first, analyzing 15 commercials (short video-clips and posters) for, according to MiamiHerald (2023), the five highest-rated eye drops of 2023 for dry eyes. To get the patients’ perspectives, we analyzed 15 blog posts and 300 short, free-form entries to a survey of people having dry eyes (data available at dryeyefoundation.org). The data was analyzed for metaphors by applying the metaphor identification procedure MIP (Pragglejaz, 2007) and Conceptual Metaphor Theory (Lakoff & Johnson, 1980). We will first present an overview of the most common metaphors used for dry eyes. Secondly, we compare how the metaphors used in the commercials for eye drops differ from the metaphors that patients use when they talk about dry eye symptoms. Thirdly, we present potential commercials for eye drops that we had created ourselves that include some of the previously identified metaphors. Eventually, we propose ways for how taking the patients’ perspectives into account could make health-related commercials more effective.

References available upon request.

5 - Tackling the US Opioid Crisis: Data-Driven Detection of Suspicious Retail Buyers
Seethu Seetharaman, Washington University in St. Louis, Saint Louis, MO, United States, Annie Shi, Antonio Sardella, Michael Wall, Abeyakaran Chenthuran

The opioid epidemic affects thousands of communities across the US. Recently, the Drug Enforcement Administration (DEA) launched an aggressive effort to monitor drug diversion. Our objective in this research is to assist the DEA’s effort to stop opioid shipments from reaching those at risk. We propose an anomaly detection algorithm to identify suspicious retail buyers of opioids. We implement our algorithm on the ARCOS database -- which tracks all opioid drug shipments across the US from 2006 to 2012. Our algorithm effectively identifies suspicious retail pharmacies and practitioners involved in drug diversion. It achieves 100 % precision and 100 % sensitivity, resulting in 100 % F-1 score for retail pharmacies. For practitioners, while sensitivity is 30 %, precision remains at 100 %, leading to 46 % F-1 score. By applying our algorithm, the DEA gains a powerful tool for promptly detecting suspicious retail buyers. This enables prevention of large opioid shipments by identifying potentially negligent or criminal drug retailers early. By doing so, we can safeguard vulnerable communities and save lives by ensuring that dangerous drugs do not easily reach them.

TC18
Meeting Room C4.7
Consumer Behavior: Generational and Individual Differences in Sustainable Behavior
Contributed Session
Chair: Ashish Sachdeva, Indian Business School, Indian School of Business, Hyderabad, 140306, India
1 - Adolescent's Sustainable Food Consumption Behavior: The Effect of Parental Influence, Peer Influence, Ethical Sensitivity and Consumer Effectiveness
Long-Chuan Lu, National Chung Cheng University, Chia-Yi, Taiwan, Yan-Jun Huang, Yih-Jen Lu, Hsi-Chieh Lin

The exploration of generational perspectives has gained prominence in the investigation of sustainable food consumption (SFC) behavior. This paradigm involves the adoption of consumption practices geared towards minimizing adverse environmental impacts, advancing social equity, and fortifying overall sustainability. Significantly, it has introduced novel dynamics and considerations, intensifying the imperative to address environmental and social challenges through SFC practices. This study posits that the SFC behaviors of adolescents are intricately influenced by their ethical sensitivity, shaped by pivotal factors such as parental influence, peer influence, perceived responsibility, and long-term orientation. Furthermore, the study posits that consumer effectiveness plays a moderating role in the relationship between ethical sensibility and SFC behaviors.

To empirically investigate these hypotheses, a quantitative research design will be employed, with data collection facilitated through an online survey targeting a diverse sample of adolescents in both the United States and Taiwan. The analytical framework will leverage Structural Equation Modeling (SEM) to rigorously test the proposed hypotheses. The anticipated findings aim to enrich existing literature by elucidating the nuanced interplay of factors influencing adolescents' SFC behavior.

Beyond academic contributions, this study aspires to offer practical insights for policymakers, educators, and marketing practitioners. By understanding the multifaceted determinants of adolescent SFC behaviors, targeted interventions and strategies can be devised to promote sustainable food choices among this demographic. Ultimately, these efforts may contribute to the cultivation of a more resilient and sustainable future.

Key words: Parental Influence, Peer Influence, Ethical Sensitivity, Sustainable FoodConsumption

2 - Factors Influencing Millennnials' Purchase Intentions Towards Vegan Products in South Africa
Siphiwe Dlamini, University of Cape Town, Cape Town, South Africa, Keina Mehta
The purpose of this study is to examine if personal reasons, environmental consciousness, health consciousness and animal welfare influence the purchase intention of vegan products amongst millennials in South Africa. Little research has been conducted on the influence of age and gender on veganism. Using data collected via online surveys from 156 respondents between the ages of 18-35 in South Africa, the study examined the factors that influence the purchase intention of vegan products, which included testing whether age and gender impacts the purchase intention of vegan products. It hypothesized that the factors analysed – ‘personal reasons’, ‘environmental consciousness’, ‘health consciousness’ and ‘animal welfare’ - have an impact on the success of vegan food companies in South Africa. It was also hypothesized that women are more likely to purchase vegan products than men. The results concluded that there is a positive relationship between the factors analysed and the purchase intention of vegan products. Animal welfare had the strongest positive relationship with the purchase intention, while there is no difference in the purchase intention of vegan products between South African men and women. This study provides insights into factors that contribute to millennials’ purchasing intention of vegan products in a South African context. Given that veganism is a growing trend in South Africa, especially amongst the millennial generation, it provides vegan product companies with a greater understanding of their target market and the factors that are influencing individuals to purchase vegan products which will assist them in implementing suitable strategies.
3 - Individual Sophistication and Learning
Ashish Sachdeva, Indian Business School, Hyderabad, India, Pranav Jindal
Several decisions such as getting a gym membership, magazine subscription renewals, choosing between different phone plans etc. are based on individual's beliefs about how they'll behave in the future, and their degree of sophistication in accurately predicting this behavior. From a firm's perspective, it is important to understand how sophisticated individuals are, and whether it can influence their degree of sophistication. In this paper, we answer the following questions: (1) How sophisticated are individuals? (2) Can we influence the degree of consumer sophistication?, and (3) Do individuals learn and become more sophisticated over time? As compared to previous research which estimates degree of sophistication, we elicit beliefs within an RCT which provides a direct individual-level measure of sophistication. Based on a field experiment which randomizes people’s incentives to complete higher number of steps daily, we find that on average individuals are sophisticated, a firm can manipulate sophistication, but individuals do not learn to get more sophisticated even after they receive feedback about their performance. Our findings have implications for firms selling subscriptions for goods and services.
TC19
Meeting Room C4.8
Emerging Research on Platforms, Service Arrangement, and Revenue Sharing Programs
Invited Session
Special Sessions
Chair: Mengze Shi, Hong Kong University of Science and Technology, Kowloon, Hong Kong
1 - Platform Competition with Multihoming on Both Sides: the Exclusive Dealing Agreements
Mantian Hu, The Chinese University of Hong Kong, Shatin, Hong Kong

Online platforms are playing increasingly important roles in people's daily lives. This research will exploit a unique data set on the online food delivery industries in Hong Kong, supplied by a third-party business analytics company. We use consumer transaction data to develop a structural model system for the decisions of consumers, merchants, and platforms and allow multihoming behavior on both sides of the platforms. We study the exclusive dealing agreements in Hong Kong's online food-delivery industry. An exclusive dealing agreement is a strategy often used by online platforms to gain competitive advantages over their competitors. It remains to be empirically determined whether the exclusive dealing agreements block competition and create monopoly power for the online food-delivery platforms in Hong Kong. The findings from this research will contribute to our understanding of the competition between online platforms and provide implementable and strategic recommendations for platforms and merchants. In addition, our findings will help policy-makers and governments to develop advanced policies to better manage the platform economy.

2 - The Impact of Teaming and Revenue Sharing on the Productivity of Financial Advisors in the Wealth Management Market
Bicheng Yang, Hong Kong Polytechnic University, Hong Kong, Hong Kong
This research empirically examines a common practice in the financial advisor (FA) industry: FAs team up to serve clients and split the revenue. We make use of a unique dataset that documents all the teaming and revenue-sharing activities as well as FA performances between January 2018 and December 2021 in one of the largest financial advisory companies in the US, aiming to answer several important managerial questions, including: 1. Should teaming be encouraged? What is the impact on the firm and on customers? 2. Should negotiation on share of revenue be encouraged? Will the firm be better or worse off by imposing the share? To answer these questions, we will use a structural model which allow financial advisors to team up with others to manage accounts, and conditional on teaming, bargain on revenue shares, and all decisions are made to maximize one’s expected utility. This research will contribute to both the salesforce management literature and the matching literature.
3 - Selecting Creators to Sign on a Content-Sharing Platform: A Deep-DiD Approach
Xinyu Cao, Chinese University of Hong Kong, New York University, Hong Kong, Hong Kong, Yan Cheng, Jingbo Wang, Zuo-Jun Shen, Yuhui Zhang

A large video-sharing platform introduced a “Creator Signing Program” aimed at signing creators and motivating them to generate more high-quality video content on the platform. Leveraging a matched dataset from the platform, we employ Difference-in-Difference (DiD) analyses to demonstrate the significant positive impact of the signing program on signed creators' performance, measured by the number of uploaded videos as well as the total user time and user engagement contributed by the creators’ videos. More importantly, we propose a novel Deep-DiD model that combines deep neural networks with DiD to estimate the individual-level heterogeneous treatment effects of the signing program. Based on the estimated individual-level treatment effects as a function of creators' pre-treatment characteristics, the platform can optimize creator selection by selecting creators with the highest estimated treatment effects. Comparing creators selected using our Deep-DiD model to those selected by the platform, we show that the former have significantly higher estimated treatment effects and experience substantially larger actual performance jumps. Lastly, we demonstrate the importance of incorporating unstructured data (visual and audio features) in the model.

4 - Incentivizing Mass Creativity: An Empirical Study of Online Publishing Market
Clarice Zhao, McGill University, Montreal, QC, Canada, Xiaolin Li, Mengze Shi
This paper examines the effects of incentive plans on the quantity and quality of creative production. We study a serial publishing platform which switched from a uniform commission plan to a quantity-based incentive plan offering higher commission rates if a writer’s production meets higher quantity brackets. Our analysis shows that, for a given book, the chapters published in the time periods when writers reached higher brackets of quantity (hence higher commission rates) had higher quality measured by chapter-to-chapter customer retention rates. Such a positive correlation is not significant in books published when the platform offered a uniform commission plan. We theorize that the quantity-based commission plan can enhance the quantity-quality complementarity in a writer’s payoff function. With the enhanced complementarity, the creators who reach a higher bracket of quantity will produce a higher quality under a quantity-based plan than under a uniform commission plan. Further empirical analysis indicates that the degree of enhanced complementarity is weaker for the writers who earn commissions from multiple books. Overall, our results underscore the importance of proper incentive design in improving the platform’s performance in managing mass creativity.
TC20
Meeting Room C4.9
Consumer Behavior: Brand Activism
Contributed Session
Chair: Sohyun Bae, Tainan, Taiwan
1 - Perceptions of Group Gatekeeping and Inclusivity
Evan Weingarten, Arizona State University, Phoenix, AZ, United States, Rachel Gershon, Amit Bhattacharjee

Who defines identity or group membership, and how do we perceive those who exclude (vs. include) outsiders from groups that promote veganism or climate change? Key to marketing research in the past few years has been how to build membership of social movements and causes and what tradeoffs these movements and causes face. Five preregistered studies (N=3,027), six preregistered replications (N=2,645) and ratings of field stimuli find that excluders (gatekeepers) are seen as less likeable but more committed to their group (vs. those who are inclusive to outsiders). These perceptions depend on candidate fit and group is defined by “sacred values.” While gatekeeping increases perceived commitment only when the applicant is a bad fit with the group’s values, it reduces likeability regardless of fit, albeit less so for applicants who are a bad fit with the group’s values. However, people who hold group values sacred favor gatekeeping more: as an individual perceived the group’s commitments to be more sacred, they perceive gatekeepers as more likeable and more committed to the cause (relative to inclusive leaders). Finally, members reward exclusive group leaders with increased support and actual donations in a consequential study.

2 - Corporate Activism and the Effects of Sociopolitical Debate Strategies on Customer Responses
Michael Paul, University of Augsburg, Augsburg, Germany, Janne Skamel
Polarization on sociopolitical issues is on the rise worldwide. While companies have been silent on sociopolitical issues in the past, many now act like activists and take a public stand on issues like sustainability, gender equality, or racial equality. Companies are in a dilemma as participating in debates as an activist can alienate customers who disagree with the company’s stand and increase polarization, but staying out of the debate can also have negative consequences as customers expect companies to take a stand. Resolving this corporate activism dilemma is further complicated by the fact that customer reactions to company behavior depend on further boundary conditions, such as the source of a stand. This research presents the mediator strategy as a way for companies to participate in debates without increasing polarization. In four experimental studies, the authors compare the effects of the activist versus mediator strategy on the polarization of customer responses, depending on customers’ stand in a debate (change vs. maintain status quo). They also examine moderating effects of stand source characteristics (e.g., spokesperson race and sex or past strategy), from which customers infer whether a company pursues a strategy seriously with perceived opportunism being the underlying mechanism. The results help research and practice to understand the complexities of polarization in customer reactions towards companies’ sociopolitical debate strategies and the boundary conditions influencing their effectiveness. This research ultimately helps practice to select sociopolitical debate strategies.
3 - The Effect of Political Ideology on Consumers' Reactions to Highly Competent but Politically Opposite Service Providers
Sohyun Bae, National Cheng Kung University, Tainan, Taiwan

Although it has become common for service providers to express their political stances to the public, little research has been conducted to investigate how consumers make trade-offs between service providers’ competence and political alignment. To close that gap, in our research we examined how consumers’ political ideology (i.e., conservatism vs. liberalism) influences their preferences between highly competent but politically opposite service providers and less competent but politically aligned ones. In three experiments, we found that conservative consumers preferred highly competent service providers regardless of whether the providers were politically aligned (i.e., conservative) or opposite (i.e., liberal) to them. By contrast, highly competent but politically opposite (i.e., conservative) service providers were less preferred among liberal consumers. Moreover, feelings of guilt about selecting highly competent but politically opposite service providers mediated the effect of political ideology on consumers’ choices between the two types of service providers. Those findings shed light on what segments of consumers are more sensitive to service providers’ expression of their political stances and why.

TC21
Meeting Room C4.10
Consumer Response to AI: Real-world Data
Contributed Session
Chair: Iqbal Ahmed, Virginia Tech, 2629 frontage road nw apt 313, roanoke, VA, United States
1 - Transforming Market Dynamics and Consumer Preferences: The Impacts of AIGC on Human-Made Counterparts in the Audiobook Industry
Ning Wan, Peking University, Beijing, China, People's Republic of, Yingjie Zhang

Nowadays, the swift progression of artificial intelligence (AI) has underscored the growing importance of AI-generated content (AIGC) in the business landscape. The progressively sophisticated AI methods have resulted in AI-generated products closely resembling those crafted by humans. Consequently, this not only brings about a notable transformation in the market framework but also has the potential to influence consumers’ preferences, particularly as distinguishing between AI-generated and human-made products becomes more challenging. Though extensive literatures have delved into consumers’ responses toward algorithms or related products, little is known about how the advent of AI-generated products alters consumer tastes and subsequently influences their consumption practices, especially on those incumbent human-made products. To address this gap, this study focuses on the audiobook industry, which has witnessed the increasing prevalence of AI-made products that have demonstrated satisfactory audio quality in the market. Leveraging diverse econometric and NLP techniques on a large-scale, fine-grained dataset collected from a prominent Chinese audiobook platform, we analyze the impacts of AI-made audiobooks on the consumption of human-made products. The findings reveal that the introduction of AI-made audiobooks has led to a notable 8.1% reduction in the consumption of human-made audiobooks. More importantly, consumers’ attitudes exhibited a heightened level of extremity, with no discernible shift in sentiment score means. Besides, consumers prioritize voice and emotions over their previous emphasis on audio content and updating frequency. Furthermore, we conducted multiple heterogeneity analyses with respect to review volumes, content genres, popularity, etc. Overall, our work provides non-trivial insights both theoretically and empirically.

2 - Blending and Winning: How Generative AI is Redefining Social Media Marketing with Innovative Content Strategies
Yuan Feng, Northeastern University, Boston, MA, United States, Junbum Kwon

In this era of digital content proliferation, our study investigates the impact of generative artificial intelligence (Gen AI) in social media marketing, with a focus on an innovative image blending technique. This technique, driven by Gen AI, synergistically combines elements from the most popular Instagram posts of various brands into a singular, compelling image.

The study spans across both utilitarian (e.g., Toyota, Dell) and hedonic (e.g., Mercedes, Starbucks) brands, offering insights into Gen AI's efficacy in different market contexts. Our experimental design compared original top posts and standard AI-generated social media posts (control group) with those created using Gen AI-blended images and AI-derived texts (treatment groups). We meticulously assessed both quantitative engagement metrics (likes, comments, shares) and qualitative aspects (appealing, creativity, inviting, nature, fascination, and perceived AI integration). The findings revealed a consistent and significant preference for Gen AI-enhanced content across all tested brands, with the Gen AI-blended posts achieving markedly higher engagement than the control. This trend was evident across brand categories, illustrating Gen AI's versatility in resonating with diverse audience groups. The Gen AI-blended content notably excelled in qualitative evaluations, being perceived as more appealing, creative, and inviting. This research underscores the power of content integration Gen AI in content creation, offering profound insights for integrating advanced AI in social media strategies. It highlights Gen AI's capacity to elevate audience engagement and interaction, providing a valuable paradigm for marketers and content creators in navigating the complexities of modern digital marketing.

3 - Exploring the Use of Generative AI in Ad Copy Design: A Pilot Study
Li Xiao, Fudan University, Shanghai, China, People's Republic of, Xinlan Li

The application of generative AI (GenAI) in the realm of advertising has seen a remarkable uptick, driven by its capacity to amplify the creative process. In the field of advertising, a plethora of GenAI tools are being deployed to craft ad copy, and create visually engaging graphics. In this research, we explore the capabilities of the GenAI tools, particularly OpenAI's GPT-4, released in September 2021, assessing their performance in ad copy generation. We synthesized insights from advertising literature and expert input to formulate a prompt for GPT-4, which then produced a series of ad drafts. These AI-crafted ads, alongside their human-originated counterparts, underwent evaluation by advertising professionals for creativity, and by consumers for advertising professionals, i.e. attitude towards ads, attitude toward brand, and purchase intention. The results indicate that while GenAI-generated ads were often distinguishable and rated lower than their human-originated counterparts—especially when the original ads were considered highly creative, such as incorporating humor rather than mere product and brand display—their creative and effectiveness performance was not on par. Nevertheless, GenAI tools can significantly aid the ad copy design process, especially in the initial stages by fostering divergent thinking and clarifying client objectives through the rapid, cost-effective generation of numerous ad concepts. Therefore, they hold substantial value for advertising agencies and could potentially supplant junior-level positions. However, the use of GenAI tools must be approached with caution, keeping in mind potential copyright infringement issues.

4 - Propensity to Use AI Agents as Companions: The Mediating Role of Social Stigma
Iqbal Ahmed, Virginia Tech, Washington, DC, United States, Dipankar Chakravarti

People possess social skills to different degrees and individuals with chronic deficiencies in social competence are often socially excluded. Other individuals, whom we term “social excluders,” are guided by their social preferences to actively avoid social interactions.

These two groups are similar in social interactions, but have pre-dispositions towards the experience of loneliness. For the socially excluded, loneliness reflects a gap between the desired quality of connections vs. what is actually in place. This gap has negative psychological and physiological effects on well-being (Hawkley & Cacioppo, 2010). Social excluders, by contrast may avoid general human companionship and feel comfortable only with specific qualities of interactions.

Technological advances are creating AI entities that can potentially provide social companionship experiences without the attendant costs. Yet research suggests that individuals may hesitate to adopt AI agents as companions due to an apprehension that others infer that they suffer from mental or personality deficiencies. Such stigma may differ by the type of AI entity and the usage context and mediate the propensity to use AI agents as companions.

Stigma associated with the adoption of AI agents as companions may stem from three source. Thus, individuals may see their need for AI companionship as a signal to themselves (self-stigma), or to others (anticipated stigma) of such deficiencies. Finally, they may themselves attach such stigma to others who use AI entities as companions. We examine the extent to which each of these sources of stigma influences the propensity to use AI agents for social companionship.

TC22
Meeting Room C4.11
Behavioral Economics and Marketing
Invited Session
Special Sessions
Chair: Zijun (June) Shi, HKUST, Clear Water Bay
Co-Chair: Tianyu Han, University of California-Berkeley, 747 Red Oak Ave, Berkeley, CA, 94706, United States
1 - Interest Rate Misperception in the Credit Card Market
Tianyu Han, University of California-Berkeley, Albany, CA, United States, Xiao Yin
Combining bank account data and surveys, we find that consumers have noisy perceptions of the interest costs of credit card debt. Underestimations in interest rates induce a higher debt, while overestimations do not affect borrowing. Undergoing an information treatment that informs the true costs of borrowing, consumers with underestimations upwardly revised their perceived interest rates and reduced debts. Despite a huge instantaneous response, consumers' misperceptions and debts reverted by more than 50\% six months post-treatment. To explain the short-living effects, we use consumers' banking app login behavior to show that the reversal of misperception is consistent with the ostrich effect where consumers selectively avoid unfavorable information when interest rates are high.
2 - Shopping Addiction
Zijun (June) Shi, HKUST, Clear Water Bay, Hong Kong, Chen Cao, Xuhang Fan, Xiao Liu

Shopping addiction is a behavioral disorder of global prevalence. In the United States alone, approximately 18 million adults have been affected by shopping addiction. Consumers with addiction persist in shopping to seek enjoyment from the purchasing process and cannot restrict their expenditure and time under budget. In this case, shopping can be seen as an addictive behavior and may result in severe consequences. In this research, we aim to present empirical evidence, formalize the concept, and explore the implications of shopping addiction behavior. We seek to make three primary contributions. First, we provided field evidence of addictive behaviors under the online shopping context. Second, we applied a structural approach to theorize shopping addiction as the combination of habit formation and self-control. We identified the structural parameters using data generated from large-scale randomized controlled experimental studies. Finally, the study examines the heterogeneity among consumers, product categories, and different shopping channels. Our research can deliver helpful implications to practitioners, consumers, and policymakers.

3 - Authenticity Matters: The Persuasive Power of Self- vs. Other-Focused Product Recommendations Posts on Social Media Platforms
Chong Yu, CUHK Business School, Hong Kong, Hong Kong, Leilei Gao, Xinyu Cao
Social media influencers often share product reviews and recommendations via online posts, blogs, vlogs, or streaming videos. Their communication style may lean toward a more self-focused style (incorporating self-reference cues without audience address) or a more other-focused style (incorporating other-reference cues by directly addressing the audience). The present research compares the persuasive effectiveness of self-focused versus other-focused social media posts. Specifically, we propose that a self-focused product recommendation post yields greater persuasion than an other-focused post, leading to increased audience interest in the recommended product and post engagement. We further propose that this effect is driven by authenticity perception: compared to self-focused posts, other-focused posts diminish the perceived authenticity of influencers. We test these predictions across a series of studies, including a large-scale secondary data analysis using scraped data from Weibo and four controlled experiments. Our results demonstrate the proposed effect and provide insights into the underlying mechanism. Our findings highlight the impact of self- versus other-reference cues on consumer perception and contribute to the research on authenticity. Our findings offer managerial guidance for social media influencers, brands, and MCN organizations in creating effective promotional content.

4 - Income Misreporting in the Credit Card Market
Xiao Yin, University College London, London, United Kingdom, Tianyu Han
In the process of acquiring credit cards, consumers often self-report their income levels, a practice that tends to be prone to unverified overstatements. We empirically investigate the existence of such income misrepresentation and assess whether financial institutions take this potential exaggeration into account. Collaborating with a leading commercial bank in China, we survey consumers on their income growth rates. By utilizing these reported growth rates and current incomes, we infer the consumers' actual income at the time of their credit card application. Our findings indicate a significant degree of income over-reporting among consumers, with an average exaggeration of approximately 30%. Further, we employ a quasi-experimental approach to determine the causal effect of this income misreporting on the allocation of credit limits. Our results suggest that the bank indeed takes into account such misreporting behaviors: income exaggerated by 10% decreases credit limit by around 100 US dollars.
Thursday, June 27, 3:30 PM - 5:00 PM
TD01
Pyrmont Theatre
Practice Prize Competition 2
Invited Session
Chair: Lan Luo, Columbia Business School
1 - Sales Automation
Saiquan Hu, Hunan University, Hunan, China, People's Republic of, Juanjuan Zhang, Yuting Zhu
Helping new salespeople succeed is critical in sales force management. We develop a deep-learning-based recommender system to help new salespeople recognize suitable customers, leveraging historical sales records of experienced salespeople. One challenge is how to learn from experienced salespeople’s own failures, which are prevalent but often do not show up in sales records. We develop a parsimonious model to capture these “missing by choice” sales records and incorporate the model into a neural network to form an augmented, deep-learning-based recommender system. We validate our method using sales force transaction data from a large insurance company. Our method outperforms common benchmarks in prediction accuracy and recommendation quality, while being simple, explainable, and flexible. We demonstrate the value of our method in improving sales force productivity.
2 - Motivating Sustainable Energy Consumption Within Organizations: The Role of Artificial Intelligence and Behavioral Insights
Christopher Amaral, University of Bath, Bath, United Kingdom, Ceren Kolsarici, Lina Ikonen, Nicole Robitaille

Despite a breadth of research showing how to reduce individuals’ energy consumption with techniques such as behavioral insights and dynamic pricing, we know relatively little about their effectiveness when targeted at organizations and over time. We partnered with an energy consulting company to develop and test a multidisciplinary approach to reduce organizations’ energy consumption within a demand pricing program, critical peak pricing (CPP), which involves charging high energy prices during demand peaks. Using a multi-phase longitudinal randomized field experiment, we test the effectiveness of improved demand forecasting using artificial intelligence and behaviorally informed emails leveraging planning prompts on organizations’ energy consumption.

Our results indicate that CPP effectively decreases organizational energy consumption. At baseline, before incorporating our interventions, we find that when an energy peak is predicted accurately, organizations, on average, decrease their hourly energy consumption by approximately 320 kWh, leading to an average annual energy cost savings of $162k per organization. More importantly, we find that CPP can be enhanced through improved forecasting and behaviorally informed emails. After our intervention, we show that neural network time series modeling allows for an 80% improvement in identifying peak demand, enabling energy curtailment at the right time, and decreasing an organization’s energy costs by an additional $41k on average. We also find that an email incorporating planning prompts improves an organization’s hourly energy curtailment during critical peaks by an additional 186.64 kWh on average, leading to additional savings of $121k per organization on average. Importantly, our interventions remained effective even after repeated exposures.
TD02
Meeting Room C2.1
Game Theory for Pricing
Contributed Session
Chair: Julie Wang, Carnegie Mellon University, 4765 Forbes Ave, pittsburgh, PA, 15213, United States
1 - Unveiling Optimal Trial-Based Pricing Strategies in Subscription Models: A Game Theory Analysis
Chicheng Wu, National Sun Yat-sen University, Kaohsiung, Taiwan

Subscription model is one of the prevailing contemporary business models aimed at establishing long-term relationships with consumers. For subscription-based companies, the trial period is a crucial phase because many consumers use trials to discern their preferences. The usage frequency during trial periods can significantly influence consumers' evaluations of the product or service. This study focuses on the trial period strategies of businesses and the corresponding consumer behaviors. Currently, there are three common pricing strategies in subscription models: free trials followed by long-term subscriptions, subscription with the freedom to cancel anytime after payment, and offering basic features for free while requiring payment for premium features. Prior academic literature on subscription models has relatively limited discussion regarding the effectiveness of subscription strategies. Moreover, literature employing game theory models to explore subscriptions fails to incorporate changes in consumer behavior induced by subscriptions (such as sunk cost effects or endowment effects), thus not fully capturing the true impact of subscription strategies. This study's primary goal is to employ game theory analysis to explore optimal subscription pricing, particularly trial strategies, for companies. It constructs a game model categorizing consumers as natural (displaying irrational behaviors such as sunk costs and endowment effects) and economic individuals making rational choices. Within this framework, the research assesses the most suitable trial (pricing) strategy for subscription-based firms. Through game equilibrium analysis, it proposes the most fitting trial (pricing) strategies for subscription-based companies, considering varied product types and target consumers’ characteristics.

2 - Clandestine Information Gathering about Competitors’ R&D: Good or Bad?
Yuanfang Lin, University of Guelph, Guelph, ON, Canada, Amit Pazgal, David Soberman
Typically, intelligence gathering on competitors has been portrayed as being underhanded or sneaky: a form of spying or skullduggery to acquire intellectual property (IP) or corporate secrets. There are certainly many examples of this and accusations and lawsuits regarding IP and general information theft are increasingly common. Here, we take a different perspective and characterize information gathering as effort devoted to learning about the status of a competitor’s R&D effort. We use a game theoretic model in which two firms compete for market demand by making decisions about R&D, information gathering, product launch, and pricing. We simplify the act of information gathering as a firm making a payment to learn, prior to its own launch decision, whether the rival’s R&D has been successful. When markets are characterized by a large segment of innovators (i.e., consumers who are excited and motivated to buy and use the most innovative product in the category), this motivates firms to engage in competitor information gathering. Analyses of the theoretical model indicate that information gathering on the competitor’s R&D has positive effects on a) the focal firm (that is why they do it), b) the competitor (they benefit even more than the focal firm due to a positive externality), c) industry-wide investment in R&D and d) the likelihood of innovation being introduced to the market. These benefits flow from the observation that better-informed players typically make more efficient decisions. In particular, the fundamental inefficiency of an R&D race is mitigated by such competitive information gathering activity.
3 - A Model of Two Learning Processes
Zheng Gong, Chinese University of Hong Kong - Shenzhen, Toronto, ON, Canada, Jin Huang
Many studies have shown that consumers, before deciding whether to purchase a new product or service, draw inferences from the choices of other consumers and actively acquire information from other sources. We propose a novel model that integrates two learning processes: observational learning and active learning. Building upon the classic observational learning framework, this model allows each consumer to make dynamic choices of information acquisition. We first analyze consumers' learning behavior in the presence of a given price, and then we endogenize the seller's dynamic pricing strategy in response to the two learning processes of consumers. We show that a forward-looking seller may find it optimal to sacrifice short-term profits by setting a higher price to induce active learning, thereby improving the information transmitted through observational learning and ultimately gaining higher expected future profits. We also investigate consumers' learning behavior, the seller's dynamic pricing strategy, and long-run market learning outcomes when the speed of information acquisition increases with sales.
4 - Impact of the Invisibles: Personalized Pricing on Platform with Anonymous Users
Julie Wang, Carnegie Mellon University, pittsburgh, PA, United States, Zoey Jiang, Param Singh
Many businesses are leveraging comprehensive consumer data to implement personalized pricing strategies. Yet, these practices are increasingly challenged by data privacy regulations, including the General Data Protection Regulation (GDPR). Such regulations have resulted in an increase in anonymous consumers who do not share their data. This study investigates the market implications of this reduced access to consumer data, particularly within e-commerce platforms. We develop a two-stage theoretical model where, in the first stage, the platform owner creates customer segmentation, and in the second stage, a duopoly competition unfolds between first-party and third-party sellers under segment-based differential pricing.

Contrary to the classical intuition that segmented pricing intensifies competition compared to uniform pricing, our findings reveal: The presence of anonymous consumers enables the platform to craft a "fuzzy" segmentation strategy (by combining anonymous customers with part of data-sharing consumers), which softens competition and allows both first-party and third-party sellers to achieve higher profits than uniform pricing. Under this pricing scheme, not only are anonymous consumers likely to pay a price premium due to their preference for privacy, data-sharing consumers may also experience a negative spillover effect, facing higher equilibrium prices. We analytically show the change in producer surplus and consumer surplus under different market assumptions. This study brings to light the potential adverse effects of privacy-protecting behaviors on consumer welfare under platform competition.
TD03
Meeting Room C2.2
B2B Marketing 2
Contributed Session
Chair: Adnan Yusuf, Deakin University, 70 Elgar Road, Melbourne, 3125, Australia
1 - Customer Pull for Eco-Design: How Do SMEs Respond?
Ralph Kauffman, University of Houston-Downtown, Frisco, TX, United States, Faiza Khoja, Jeffery Adams, Mikayel Yegiyan

Improvement in sustainability has continued to be an important objective of marketers in companies of all sizes. A few of the motivators behind this include: improving and/or maintaining a positive company image by being an environmentally responsible corporate citizen, and meeting legal and regulatory requirements. However, perhaps the strongest motivator is customer preference for products that have a minimal environmental impact in their production and use. Evidence of this preference is indicated in a relatively recent Nielsen study finding that, over a several year period, consumer sales of sustainable products grew at a compound average growth rate four times that of conventional products. This present study uses a sample of SMEs from diverse industries to examine the direct influence of customer preference and the potential moderating influence of laws and regulations and green purchasing practices on eco-design of products and services in small and medium-sized enterprises (SMEs). The results indicate that in addition to meeting expectations of customers and regulatory bodies, eco-design of products and services also improves firm performance. The analysis applies stakeholder and neo-institutional theories and uses survey data collected from a sample of SMEs in various industries. Regression path analysis is applied, bootstrapped with 0.95 CI (bias corrected and accelerated) and 2,000 samples on all regressions. Significant positive relationships were found between customer preference and eco design, moderated by impact of laws, and between eco-design and environmental and economic performance. Green purchasing was not found to have a significant effect on adoption of eco-design.

2 - Contagion of Failure: Paths and Curbing Mechanisms of Successive Store Failure Within Franchise Networks
Xu Zheng, City University of Hong Kong, Hong Kong, Hong Kong, Yajing Fan, Erik Mooi, Stephan Ludwig
Failures of franchised stores affect franchise systems in many ways, such as incurring significant losses on the franchisee, negatively impacting customers, and causing a jeopardy of franchise system reputation. In addition, failures have an important and hitherto undocumented “multiplier” effect—the potential of failure to create “contagion” where one store failure may spread to other same-brand stores. Anchoring on the emerging literature of the spillover effect of store closure, we theorize and test how and why such contagion may occur. We identify two related, yet distinctive, contagion paths—geographic proximity and customer sentiment proximity, based on cohesion and structural equivalence arguments from social contagion theory, and identify three distinct means from franchisors’ governance toolkit in curbing contagion. To have a more complete understanding of the phenomenon of interest, we collect data and analyze the contagion effect from both the firm- and consumer-angle. Two studies examine this phenomenon from a franchise system and customers’ standpoint, respectively. In Study 1, we employ a spatial econometrics model to examine the influence of preceding outlets’ failure through the two contagion paths covering 12,759 observations over the years 2016 to 2022. In Study 2, we further employ a difference-in-difference model and analyze the relative change in customer review comments from Google Reviews for the stores included in Study 1, and provide a refined understanding on how customers’ ratings, sentiment, and topics evolve pursuant to store closure in the franchise network.
3 - Effect of Digital Marketing Strategy on Franchisor and Franchisee Performance
Jun Wang, Monash Univesity, Caulfield North, Australia, Sudha Mani, Bhoomija Ranjan

Digital marketing has been shown to improve firm performance. However, little is known about how franchisor’s digital marketing strategy affects their franchisees. Digital marketing strategy can increase franchisor control and power over their franchisees, leading to greater conflict and even resulting in the dissolution of the franchisor-franchisee relationships. The objective of this study, thus, is to investigate the impact of digital marketing strategy on franchise performance, including both franchisor and franchisee performance. Drawing on the motivation-ability framework, we hypothesize that the influence of digital marketing strategy on franchisor performance depends on the franchisor’s motivation (market position) and ability (marketing capability). Similarly, the effect of digital marketing strategy on franchisee performance is contingent upon the franchisor-deployed governance mechanisms that motivate franchisees (fees-charged by franchisors) and enhance franchisee abilities (franchisor-provided support). We assemble a unique dataset of 241 publicly listed franchise brands in the United States from 2001 to 2022, yielding 1477 observations. We use a natural language processing approach (BERT) to measure digital marketing strategy and employ the control function approach to analyse the data. Our findings aim to offer quantitative insights into the effects of digital marketing strategy on franchise firms and provide guidance on how to effectively leverage franchisee relationships.

Keywords: digital marketing strategy, franchise performance, motivation, ability, governance

4 - Solution Orientation and B2B Firm Performance
Adnan Yusuf, Deakin University , Burwood, Australia, Ujwal Kayande

B2B researchers have explored characteristics of solution providers and solution customers that enhance solution quality and profitability growth. However, to our knowledge, there is no research that explores how B2B solution providers ensure their own profitability as well as effectiveness of customer solutions for their customers. In this research, we introduce the concept of ‘solution orientation’ and show how it positively impacts solution effectiveness and solution provider’s profitability. We define solution orientation as the degree to which a B2B solution provider is inclined to work on a solution with their customers, involving (i) trading value by sharing risks and returns, (ii) learning from customer firms, and (iii) relationship building focus. We conducted a qualitative study to develop measurement scales for solution orientation, followed by a pilot study to refine these scales. We tested our hypotheses in two separate surveys. In the first survey, we collected data from key account managers in solution provider firms. In a separate study, we surveyed key informants in a leading global solutions provider firm and their matching customers. Further, we used secondary data from company filings of annual 10K reports to measure solution orientation and to validate our research findings. Our research underscores the critical role of solution orientation in creating value for both solution providers and customers. The positive impact of solution orientation on provider firms’ profitability suggests that managers in solution providing firms should prioritize developing the required attitude to collaborate with customer firms.

Key words: B2B customer solutions, customer-based profitability, solution orientation.

TD04
Meeting Room C2.3
Choice Models 2
Contributed Session
Chair: I Hsuan Chiu, University of Wisconsin-Milwaukee, Milwaukee, WI, United States
1 - A Solution to the Problem of High Dimensionality in Market Structure Analysis
Leonard Coote, University of Queensland, Brisbane, Australia, Edward Wei
This paper offers an innovative solution to the problem of high dimensionality in market structure analysis. Consider a market structure analysis where in addition to many product attributes (A), the analyst confronts many brands (B). The market for wine offers an example where B is very large. Market structure analysis is of enduring importance to quantitative marketing and economics, and potential solutions to the problem of high dimensionality have great practical value. The contribution of the current paper is primarily empirical. The paper illustrates the factor-analytic form of mixed logit, which is arguably the most general and flexible form of logit model. The proposed solution is based on the specification of higher-order factor models. Many brands can be mapped into first-order factors and these factors and factors representing preference heterogeneity for product attributes can be mapped into factors of higher order. The illustration is based on a stated preference (SP) dataset that records the choices of approximately 800 consumers from among 32 brands of wine. The experimental space is large – with this many brands in addition to the product attributes studied. The model retrieves aggregate preferences under a specific representation of the structure of the heterogeneity. Illustrations of market structure analysis use datasets where B is small, including typical factor-analytic approaches. This paper offers a natural extension to this foundational work, as the model specification introduced here immediately generalises to settings where is B can be potentially very large.
2 - Duplicated Information and Price Sensitivity
Haosheng Fan, NEOMA Business School, Mont-Saint-Aignan, France, Ralf van der Lans
Managers and marketers spend significant effort to lower price sensitivity. They use marketing and advertising strategies to differentiate themselves from competitors and make consumers shift their focus from price to other factors. This research studies the effect of product information presentation on price sensitivity. Displaying product pictures is a common practice of online retailing platforms. Some attributes can be either displayed in a picture or described in text. In this research, we are interested in consumers’ choices in different information display, especially when some attributes are presented in both texts and pictures. Study 1 tests the main effect of duplicated information on price sensitivity. We find that consumers are more price sensitive when product attributes are duplicated. Their price sensitivity is lower when these attributes are presented in texts or in pictures only. We replicate the effect of duplicated information in Study 2 and explore the underlying mechanism. We develop a choice model that is calibrated with consumer choice data and eye tracking data. We find that the difference in prices can be explained by the number of revisits in a choice task. Consumers perceived alternatives to be more similar in a choice task when they make more revisits. When alternatives are difficult to compare, consumers rely more on prices in their decision-making and exhibit higher price sensitivity.
3 - Modeling Reference Dependence Effect and Regulatory Orientation
I Hsuan Chiu, University of Wisconsin-Milwaukee, Milwaukee, WI, United States, Gary Russell

One major implication of the value function in prospect theory is loss aversion: losses loom larger than gains of the same size. The empirical examinations of loss aversion from consumer brand choice decisions, however, provide mixed results. For example, Bell and Lattin (2002) show that loss aversion is reduced or disappears when taking into account consumer heterogeneity. In this research, we show that regulatory focus theory can account for the heterogeneity in the reference dependent effect: consumers in a promotion-oriented mindset show a weaker response to loss aversion comparing to those in a prevention-oriented mindset. We consider a decision context where consumers rely on external reference points. To capture the reference dependence effect, we use the random regret minimization (RRM) model developed by Chorus (2012). The RRM model describes a decision process that compares attributes of an alternative to that of all other alternatives in the consideration set. RRM allows for flexibility in evaluating reference dependence effects across all attributes and alternatives, and provides a framework for incorporating regulatory orientations into the model specification. Using data from a choice experiment, we demonstrate that prospect theory describes the within-individual differences with respect to losses and gains, while regulatory focus theory describes the between-individual differences.

TD05
Meeting Room C2.4
Marketing Strategy
Contributed Session
Chair: Eugene Pavlov, University of Miami, Miami, FL, United States
1 - From Words to Metrics: Measuring Digitalization from Corporate Reports
Mayank Nagpal, Indian Institute of Management, Bangalore, Bengaluru, India, Andrew Petersen, Arvind Rangaswamy

We propose a generalizable computer-aided text analysis using supervised Latent Dirichlet Allocation (CATA-sLDA) approach for measuring strategic orientations (SOs) using text analysis of firms' annual 10-K reports. This approach automates the creation of an SO-based dictionary of words weighted based on their relevance to the SO and to reflect the strategic intent of senior management. We assess the effectiveness and value of the CATA-sLDA approach by measuring the construct of digitalization from 2007 to 2022 for 245 firms across three industries. To validate our digitalization score, we compute its correlation with: (1) scores developed independently by industry sources, (2) data from a survey we conducted with industry executives, and (3) scores developed from alternate CATA-based approaches such as expert-based dictionaries and the Doc2Vec model. The proposed CATA-sLDA approach has a positive and significant correlation with the alternative scores in all these assessments, suggesting it captures the essence of the construct digitalization. Further, the digitalization scores from our CATA-sLDA approach also outperform the competing scores in predicting firm performance measured as Tobin’s Q. We conclude with observations about potential applications of our digitalization score in gaining a better understanding of the trajectory of digitalization in firms.

2 - Navigating Uncertainty: A Mixed-Methods Analysis of Marketing Strategies for Growth in Chinese Firms
Rui Wang, Peking University, Beijing, China, People's Republic of

This research addresses the challenge of achieving sustainable growth in emerging markets, characterized by uncertainties in demand, competition, institutional environment, and technology. Through a comprehensive analysis involving surveys and in-depth interviews with marketing executives from 181 Chinese firms, the study identifies pivotal marketing strategies that drive growth in these volatile environments. Utilizing Latent Dirichlet Allocation (LDA) models for advanced text analysis, the study explores the effective adoption of marketing strategies in the context of emerging market uncertainties.

Findings highlight that Chinese marketing executives predominantly employ four strategic approaches: marketing strategic planning, brand building, relationship management, and digital transformation. Each of these plays a crucial role in navigating the complexities of emerging markets. Digital transformation emerges as a unique component, with Chinese firms leveraging digital technology for establishing information networks, enabling platform-centric business transformations, and utilizing data analytics for strategic decision-making. This approach significantly contributes to market responsiveness and growth.

Moreover, the study reveals notable differences in the application of these strategies between Chinese firms and their counterparts in developed countries. Chinese companies exhibit a greater emphasis on agility in strategic planning, actively engaging with consumer demands in brand development, and maintaining a focused and coordinated approach in relationship management.

This research provides valuable insights into the strategic orientations that are instrumental for firms operating in the challenging landscape of emerging markets, highlighting the distinctive approaches adopted by Chinese companies to ensure sustainable growth.

3 - Brand Political Positioning: Implications of the 2016 US Presidential Election
Eugene Pavlov, University of Miami, Miami, Florida, United States, Natalie Mizik

The heightened polarization of the US political landscape and the increased emphasis on personal values and identity in politics create important externalities affecting the functioning of commercial entities in the marketplace. We discuss the construct of a brand’s political positioning—the extent to which the perceptual profile (brand image) of a commercial entity aligns with the perceptual profile of a major political party—and show its effects on firm valuation and sales in the aftermath of the 2016 US presidential election. We find a relative increase in valuation and sales for firms whose corporate brand image is close to that of the winning (vs. losing) political party. We propose a mechanism to explain the observed performance effects—the consumers’ shifting preferences toward (away from) the brands perceptually associated with the winning (losing) political party. We present evidence supporting this mechanism: the sales effects manifest immediately after the election (fourth quarter of 2016) and the valuation and sales effects we document are primarily driven by the consumer-facing firms. We replicate our analyses for the 2012 US presidential election and document a consistent pattern of results.

4 - Le Chatelier Revisited: The Dynamic Role of Consumers’ Satisfaction with Products on Their Sentiment about the Economy
Debanjan Mitra, University of Connecticut, Glastonbury, CT, United States, Shervin Tehrani, Sandip Roy

Economic theories contend consumers’ sentiment of current and future economic conditions and expectations of future economic conditions to be the central determinant of their consumption decisions. In contrast, marketing theory establishes the primacy of consumers’ prior satisfaction with products (i.e., goods and services) in driving consumption. On its face, the two consumer judgments of the economy and the satisfaction with its products pertain to different comparisons, and even refer to different time periods. We present a theory of dynamic equilibrium where consumer satisfaction is at the center of expectations of the future economy and perceptions of the current economy. Merging publicly available aggregate data from multiple sources spanning 30 years and applying an empirical dynamic model plus a laboratory experiment, we find strong support for our theory. Specifically, we find (i) higher satisfaction with offerings improves future perceptions of economic conditions, but (ii) higher expectation of economic conditions lowers future consumer satisfaction with offerings. Together with the fact that improved perceptions of current economic conditions co-occur with higher expectations of future economic conditions, our findings fulfill the Le Chatelier principle of a dynamic equilibrium. Most importantly, we find the interaction between satisfaction and the economic sentiment to be the key driver of economic growth, one that is significantly larger than the main effect of the individual constructs. Our study illustrates how marketing theory on consumer satisfaction complements macroeconomic theories of consumption and growth.

TD06
Meeting Room C2.5
Analytical Models for Retailing
Contributed Session
Chair: Nanda Kumar, University of Texas-Dallas, Richardson, TX, United States
1 - Bargaining in Live Streaming Commerce with Online Celebrity
Lin Liu, Beihang University, Beijing, China, People's Republic of, Qianqian Chen, Shouchang Chen, Yi Yang
Live streaming commerce is an innovative and immersive shopping format. This paper explores how a seller negotiates with a celebrity over the proportion of shared revenue of products sold through the live streaming channel. We build a game theoretical model to capture the bargaining process between the seller and the celebrity in which consumers choose to be followers or non-followers of the celebrity. We use a Markov chain framework to keep in check the transition process of celebrity's followers and non-followers. The key feature in our bargaining model is that the proportion of followers can affect the celebrity’s bargaining power which in turn influences the split of the shared revenue. Our analysis yields several new results. First, when the celebrity's consumer-stickiness is sufficiently high, the celebrity can set a price to ensure not only a maximal total profit but also a maximal portion of the shared revenue (i.e., a larger slice from a larger pie). However, when the consumer-stickiness is lower, the celebrity either maximizes the total profit or maximizes its portion of shared revenue. Second, the seller can achieve the coordination in the live streaming chain through bargaining when consumer-stickiness is not intermediate, but a higher consumer-stickiness may not increase sellers' profits. Third, the celebrity tends to set a price (weakly) lower than that set by the seller, and the level of consumer-stickiness and the equilibrium price have a non-monotonicity relationship. Fourth, a strategic live streaming platform may restrict consumer traffic to the celebrity when the consumer-stickiness is intermediate.
2 - On The Profitability of Loyalty
Ying Lei, New York University Shanghai, Shanghai, China, People's Republic of, Ji Shen, Ei Yang, Xin Zhai
This paper aims to contribute to the understanding of customer loyalty and its relationship with firm profits. Customer loyalty may be driven by various factors, such as favorable brand attitude and switching costs, and is often viewed as a strong indicator of a company's profits. However, empirical findings are conflicting on the profitability of customer loyalty, and yet a comprehensive framework to analyze the link between customer loyalty and firm profits is still lacking. This paper fills the gap in the literature and proposes an analytical framework to characterize how repeat-purchase loyalty is generated from consumer choice in a dynamic competition environment and examines the impact of brand preference and switching costs on firm profits and customer loyalty. The results indicate that favorable brand preference can result in profitable customer loyalty, while switching costs do not. We also extend the model to different scenarios of consumer preference change processes and demonstrates the robustness of the results.
3 - Points and Dollars: Strategic Competition in two Currencies
Nanda Kumar, The University of Texas at Dallas, Richardson, TX, United States, Anmol Madan, Chakravarthi Narasimhan
A rising number of companies are using reward points as part of their loyalty programs, transforming them from a simple marketing tactic to boost customer loyalty into a significant source of profit. This paper inquires into the strategic implications of reward points, focusing on their impact on a firm's pricing strategies, overall profitability, and influence on consumer behavior. Our research reveals that companies can implement a form of price discrimination by setting prices in two different currencies, enabling them to tailor costs for various consumer segments. By utilizing reward points, firms can reduce the perceived expense for their loyal customers and capture more value while simultaneously competing aggressively to attract customers from rivals. Additionally, companies may promote loyalty by charging higher prices during an initial period, leveraging reward points as an incentive. Consumers may be willing to pay more upfront to accumulate these points, leading to increased profits for the firm. The study also shows that reward points are especially effective for firms with relatively similar offerings, creating barriers to consumer switching. Finally, the research suggests that reward programs are most beneficial when the market is equally divided between different types of users, such as high- and low-value customers.
TD07
Meeting Room C2.6
Conspicuous Consumption, Luxury Marketing and Indulgence
Contributed Session
Chair: Damaris Gomez Calvo
1 - Conspicuous Consumption and Optimal Return Policies
John Park, Pepperdine University, Malibu, CA, United States, Changhwan Lee

This study investigates the controversial practice of luxury brands, epitomized by Burberry, destroying millions of pounds worth of unsold and returned products to uphold brand exclusivity. Focusing on the discerning characteristic of snobs, who highly value exclusivity, we examine how salvaging returned goods in the market can significantly influence their perception. Despite extensive research on conspicuous consumers' impact on firm decisions, the implications of return policies remain insufficiently explored.

We conduct a comprehensive examination of various return policies, including return/salvage, no refund, and return/price commitment policies. The return/salvage policy involves liquidating returned products at a discounted price, while the return/price commitment policy permits returns without salvaging, positively impacting snobs' utility by enhancing scarcity. The no-return policy addresses wardrobing issues by disallowing product returns. These diverse return policies distinctly affect the valuation for snobs, subsequently influencing the valuation of followers. Snobs' utility increases with product scarcity, while followers' utility rises with increased snob consumption. We determine the optimal return policy, along with its optimal pricing and production decisions under varying conditions. This research provides valuable insights into crafting effective return policies for luxury brands, considering the delicate balance between brand exclusivity, consumer perceptions, and market dynamics.

2 - An Indulgence Framework: Conceptualization, Operationalization, and Measurement
Connie Chang, Musashino University, Tokyo, Japan, Yu Hsu, Shane Mathews

Marketers and academics have long been interested in understanding what drives indulges, and explored factors that lead to why individuals indulge. Indulgence is a concept based in cultural and societal variation and has presented in various ways throughout history. The purpose of this study is to examine how external factors such as both pull (e.g., economic factors, price/cost and global inflation) and push (e.g., stress, social media, and government policy) factors influence people’s indulgence. Three years after pandemic, consumers are looking for ways to indulge themselves as a means of rejuvenation and reclaiming a sense of normality. This inclination toward indulgence reflects a deeper psychological need for self-care. While seeking self-indulgence, consumers consider tradeoffs that influence their indulgence, including a variety of factors including; social media usage, government policy, culture, economic conditions, price/cost, and levels of stress, and perceived risk. The primary emphasis of this study is to investigate these relationships and evaluate the role of these factors on indulgence consumption.

This study employs a mixed-method research approach. Data from Japan, Australia, Taiwan, and Malaysia based on country wealth (average GDP per capita) and the impact of COVID on both health and the economy during the pandemic will be evaluated. Since individuals in these four countries experienced different travel restrictions and lockdowns, individuals likely indulged by planning for a holiday to cope with the challenges of the pandemic/post pandemic, fulfilling their desire for leisure. This study seeks to investigate this form of indulgence.
3 - Too Luxurious for CSR? The Effect of Luxury Level and CSR Message Orientation on Customers' Perceptions of CSR Activities
Damaris Gomez Calvo, Pusan National University, Busan, Korea, Republic of, YongHee Kim, Da Eun Park

Corporate Social Responsibility (CSR) is renowned for its benefits, including enhancing a firm's image, fostering customer loyalty, and influencing purchase intentions. However, luxury firms may face difficulties promoting their CSR actions due to the conflicting values associated with luxury (egoistic) and those associated with CSR (altruistic). In this context, consumer skepticism arises due to conflicting self-indulgence inherent in luxury and CSR's self-transcendence values, potentially leading to negative attitudes toward luxury brands.Research suggests that for luxury firms, focusing on internal CSR activities may be more beneficial to decrease the dissonance caused by conflicting values. However, societal pressure drives companies, including luxury ones, to engage in external CSR, intensifying consumer confusion. To address these challenges, this study proposes effective communication as a means for luxury companies to mitigate consumer skepticism about their CSR initiatives. It explores whether messages highlighting benefits for the community (others-benefits) or individual customers (self-benefits) influence consumer perception of the reasons behind a luxury brand's CSR engagement. Additionally, the study investigates if the perceived ‘luxuriousness’ of a brand may influence how customers process the message and perceptions towards the CSR activity and towards the brand. Adopting a quasi-experimental design, it explores CSR message orientations (self-benefits vs. others-benefit) and luxury levels (prestige vs. masstige) using brands like Chanel and Coach. This study aims to create a deeper understanding of how luxury firms can navigate the dissonance created when promoting their CSR activities.

TD08
Meeting Room C3.3
Marketing-mix Adjustments in Times of Market Turbulence
Invited Session
Special Sessions
Chair: Marnik Dekimpe, Tilburg University, Tilburg, Netherlands
1 - Price Dynamics in the Wake of Store Exits: An In-Depth Analysis of Market Impact
Katrijn Gielens, University of North Carolina-Chapel Hill, Chapel Hill, NC, United States, Zhiling Bei
Retailers are grappling with an unprecedented surge in the closure of brick-and-mortar stores, exemplified by over 2800 store closures in the United States in 2023. This study aims to explore the repercussions of these closures on retail prices. Does the departure of these physical stores exert upward pressure on prices, potentially imposing higher costs on consumers? Through an examination of store closures within a major retail chain and employing a synthetic difference-in-difference estimator, our findings reveal that, on average, consumers experience elevated prices for both national-brand (NB) and private-label (PL) products post store closures although considerable heterogeneity can be observed across markets.
Further analysis indicates that the price increase for NB products is primarily driven by store closures in low-income communities, while prices for NB products decrease in middle-class and upper-class communities. As for PL products, the price surge is observed in both low-income and middle-class communities, with no discernible impact on upper-class communities. These results underscore the significance of considering socio-economic factors when evaluating the consumer implications of retail store closures.
2 - Don't Panic: Predicting the Likelihood and Duration of Retail Brand Crises
Kelly Hewett, University of Tennessee, Knoxville, TN, United States, Raoul Kubler, Koen Pauwels, Meike Eilert
A key challenge facing retailers is how to anticipate negative events in the media. Along with traditional media, social media creates a “echoverse,” a communication environment in which customers, stakeholders, the media and firms interact. As a result, negative events reported in traditional media may spillover into social media and vice versa. In addition, individuals on social media can take the megaphone to rapidly disseminate information which can change the course of a crisis, making it difficult for retailers to predict how it will play out. Given these challenges, how can retailers tell whether a publicized event will turn into a crisis? And which factors does the evolution of the event depend on? This study addresses investigates how retailers can anticipate the trajectory of such negative events across the communications echoverse and illuminates different retail crisis patterns as reflected in the degree to which the traditional media and consumer impressions (from surveys) and expressions (from social media) change throughout an event. Our data include information on identifiable negative brand events and their coverage in traditional media and social media, firms’ press releases, as well as attitudinal metrics regarding the retail brand and abnormal stock returns for five retail firms over a six-year period. We use time series cluster analysis to model the patterns of each variable prior to and following a crisis using each of the metrics identified above, and assess their influences on abnormal stock returns.
3 - Product Price Level and Retail Price Promotion: An Empirics-First Analysis of Promotion Effectiveness and Implications for Theory and Practice
Kristopher Keller, Kenan-Flagler Business School, Chapel Hill, NC, United States, Rajagopalan Sethuraman
Conventional wisdom states higher-priced products should be promoted more frequently and deeply than lower-priced products because they generate a greater promotional sales lift. However, such early theories and empirical evidence take a product (or brand) perspective by focusing on product sales and refer to a context in which NBs were high-priced and competed with low-priced SBs—a reasonable representation of the market several decades ago. Nowadays, the retailscape has changed and features high-priced store brands that we now have both high-priced and low-priced national and store brands. With these high-priced SBs, SBs no longer fall neatly into the dichotomy of “high-priced NBs” and “low-priced SBs”. This increased variation allows us to disentangle whether the previously documented lower promotional sales lift of SBs is due to their brand type (SB vs. NB) or due to their lower price Furthermore, conventional wisdom considered only brand switching and not incremental category sales – a key metric for retailers. Hence, two relevant questions arise: (i) Do price promotions and deeper discounts result in greater sales lifts and incremental category sales for higher-priced (cf. lower-priced) products? (ii) Are these effects thesame for NBs and SBs? Employing an empirics-first approach and analyzing weekly store-level scanner data of more than 80,000 products from 30grocery product categories, covering 100markets and more than 4,000 stores (resulting in >700m observations) during 2016–2019, this paper’s results reveal several empirical regularities that confirm, add nuance to, or challenge conventional wisdom. These results guide theory development and suggest a range of price promotion strategies.
4 - How to Market Investment Products to Micro-Investors: Should One Spend More or Less When the Stock Market is Booming?
Jake An, University of Technology Sydney, Sydney, Australia, Harald van, Marnik Dekimpe, Darren Kim
Fueled by easy access through digital platforms and the fear of missing out, consumers from all walks of life become micro-investors, flocking into investment products that were traditionally not accessible to them. Unlike consumption goods, investment products are characterized by the goal of making a positive return and by strong ups and downs in their value, as also reflected in the high volatility of many stock-market indices.

Given the fundamentally different nature of investment goods, the question arises how micro-investment firms should spend their advertising to attract micro-investors, especially the more valuable ones. By investigating a micro-investment firm with over 250,000 active customers, the authors provide the first empirical evidence that even though the overall demand for investment services increases during periods when the stock-market index is high, advertising effectiveness, customer conversion and the customer lifetime value of micro-investors acquired during such periods decrease. Based on the net effect of these opposing changes, they conclude that it is advisable to spend less (more) on advertising when the stock-market index is high (low).
TD09
Meeting Room C3.4
User Generated Content: Insights from Machine Learning
Contributed Session
Chair: Ning Zhong, Penn State University, University Park, PA, United States
1 - Optimizing Release Cadence of Successive Digital Content
Hanbing Xue, University of Science and Technology of China, Hefei, China, People's Republic of, Qianzhou Du, Natasha Foutz

The digital creator economy, fueled by platforms such as YouTube and TikTok, has surged to $250 billion in 2023. The 303 million creators have also captured an astounding 23% of the global population. These creators release their contents with a spontaneous cadence, i.e., varied number of days between success contents. While swifter cadence (i.e., faster releases) catalyzes audience fatigue, slower cadence risks forgetting or declining interest. Hence, optimizing release cadence is vital for sustaining audience engagement and boosting creator profitability. Analyzing 14,000 Bilibili videos by 623 creators, we quantify the relationship between the release cadence and audience viewership (and tipping). This relationship further varies by content type (sequel versus single), successive content overlap, and content category, highlighting the importance of nuanced personalization. We further explore the underlying mechanisms of the discovered relationship and propose optimal release strategies.

2 - Are Multi-Images Really Better than the Single Image? Investigating the Key User-generated Multi-image Features Influence on Customer Engagement
Lu Zhao, Academy of Mathematics and Systems Science Chinese Academy of Sciences, Beijing, China, People's Republic of, Shouyang Wang, He Zhu, Jialing Li, Mingli Zhang

Abstract: Multi-image can express abundant and subjective experience as customers visually storytelling their experience towards the brands. Hence, multi-image is crucial for brands to understand their customers in the social media context. However, the multi-image’s key feature-relevant studies are still limited in the marketing field. To address this gap, we constructed the key feature framework of the multi-image based on the visual storytelling theory, including the image content (centrality, richness, foreground and background relationship) and image content discourse (image similarity, image quality, image number). We crawled the data from Sina Weibo.com and gained 39,806 images from 14,743 posts. By using the deep learning models (YOLO models), we draw the major conclusions as follows: Image number has a reverted U-shaped relationship with the customer engagement (like and comment). Different product categories can moderate the U-shaped curve. Besides, image richness and image background can exert significant influence on customer engagement. Our paper enriches the visual marketing and customer engagement-related literature. In addition, our findings can be beneficial for brand marketers to timely comprehend their customers from the social media multi-image information.

3 - A Combined Topic Model for Unstructured and Structured Data
Sotaro Katsumata, Osaka University, Osaka, Japan, Kazuki Takahashi

The purpose of this study is to develop a model that integrates and analyzes both unstructured and structured data, in other words, open-ended data observed as natural language and selected answer data measured as numerical values. Specifically, this study extends the model to incorporate integer value based on LDA (Latent Dirichlet Allocation), which is one of the most popular methods for quantitative analysis of natural language. LDA is a classification model that incorporates tokenized text data as input and assumes a high-dimensional categorical distribution for the observed probability of the words. In addition, the Dirichlet distribution, which is conjugate to the categorical distribution, is assumed for the prior distribution. Therefore, in this study, we developed an integrated model by assuming the binomial and beta distribution, which are congruent to the categorical and Dirichlet distribution, for the structured selected answer data, and used it for the online review. In addition, we developed an explanatory model that further extends the model for exploratory classification purposes, assuming an objective variable observed as the structured data, which we also applied to the consumer review data and discussed the results.

4 - Visualizing the Competitive Market Structure: A Dynamic Latent Factor Model for Mapping Products and Brands using Online Reviews
Ning Zhong, Penn State University, University Park, PA, United States, Yifan Zhang
Online reviews have become an essential tool for businesses to learn about consumer perception of their brand and products. In this research, we develop a dynamic latent factor model that generates a joint spatial mapping of products and brands to provide insights into the underlying competitive market structure. The proposed model combines review text with review rating through a topic model and simultaneously performs market segmentation and brand positioning. We apply our model to car review data and show how the model may be used by managers to identify changes in consumer perception in online platforms and visualize the competitive market structure.
TD10
Meeting Room C3.5
Better Marketing for a Better World
Contributed Session
Chair: Jarrod Vassallo, University of Sydney
1 - From Cash to Taps: Exploring the Pain of Payment in Mobile Transactions
Keiko Yamaguchi, Nagoya University, Nagoya, Japan, Satoshi Takahashi
Payment transparency is the relative salience of payment compared to paying by cash. It explains how spending decisions vary by mode through the pain of payment. Additionally, mobile payments, perceptually distant from cash, have low transparency, leading to less pain of payment. Users spend more when using mobile payment services than other modes, such as cash or credit cards. Prior studies implicitly assumed that the level of transparency is specific to the payment mode. Furthermore, they have explored approaches to control other mediating variables to solve the overuse problem. In contrast, this study proposes an approach that challenges this implicit assumption. Directly controlling the transparency of mobile payments, we examine a new mechanism that supports payment decision-making based on sound financial sense in the modern context. The first phase verifies the potential to control transparency in mobile payments. Our approach supposes that the user interface design could induce various sensations in users to add more salience to mobile payment applications. We conduct multiple online experiments using payment applications designed to provide a realistic sense of payment. These experiments aim to determine whether changing the user interface of payment apps increases the perceived salience (transparency) and whether increasing mobile payment transparency intensifies the pain of payment after the purchase. This study loosens the implicit assumptions about payment transparency. Furthermore, it provides a fresh perspective for proposing methods to control transparency and facilitate better decision-making.
2 - David vs Goliath: Can Disintermediation Level the Playing Field for Microbusiness Owners in the UK Food Delivery Market?
Shuai Qin, Aston University, Birmingham, United Kingdom, Xiaolan Chen, Wanqing Zhang, Pradeep Chintagunta

Today’s marketplaces greatly benefit from the convenience offered by digital platforms, with such platform giants playing a crucial role in mediating buyer-seller interactions. The platform economy is rapidly expanding, with revenues soaring from a projected €3 billion in 2016 to €14 billion in 2020, according to the European Council (2021). However, a power imbalance exists between the two sides of the platform marketplace. Marginalised microbusinesses continue to grapple with numerous challenges, including excessive platform commission charges, limited access to customers, and a disadvantaged position in negotiations with intermediaries (Carter et al., 2004).

Microbusinesses are growing more adept at direct customer acquisition while bypassing digital platforms (Muzellec et al., 2015). Yet, research on disintermediation remains notably scarce. Therefore, we conducted multiple studies to 1) explore what are the disintermediation tactics adopted by microbusinesses (study 1: two-years ethnographic research that involve interviews with customers and restauranteurs that delivered disintermediation); 2) the effects of different disintermediation tactics in marketing practices (study 2: field experiment of implementing those tactics through working with small restaurants to engage their customers); 3) customers' perception of varied tactics and the moderation/mediation effects in shaping their views (through a series of lab experiment with customers).

This comprehensive research unravels what are disintermediary approach adopted by sellers, examine how these approaches are effective in practices, and why customers’ react to these approaches in one way rather than another, which hugely contributes to the disintermediation and builds the marketing knowledge of disintermediation activities.

Keywords: Disintermediation, Digital Platform, Microbusinesses


3 - Energy-Saving Competition among Young Children: A Field Experiment
Shiwen Gao, Wenzhou-Kean University, Zhejiang, China, People's Republic of, Xiuping Li, Leonard Lee, Yan Zhang, Yih Hwai Lee
Energy crises and water shortages continue to threaten many countries in the world. Most of the campaigns run by governments and organizations targeted adults and resulted in more awareness but minimal actual resource conservation. Since previous research has suggested that children can significantly influence their parents’ behavior, in this research, we designed a competition-based intervention program targeting young children from 12 primary schools in Singapore, to study the short- and long-term effects of different incentive combinations (education only with no competition incentives, competition with publicly announced social incentive, competition with publicly announced financial incentive, and competition with privately announced financial incentive) on energy and water conservation at the household level. Our results of the difference-in-differences analysis show that there was no difference in electricity and water consumption between households where children only received education only with no competition incentives and those competed for publicly announced incentives. However, electricity and water consumptions were significantly higher among households where children competed for privately announced financial incentives compared with everyone else in the study. Households, where children competed for publicly announced financial incentives, did not consume more energy or water than their privately announced financial incentive counterparts. Furthermore, the overconsumption of households where children competed for privately announced financial incentives persists even three months after the program ends. The negative effect of the competition with privately announced financial incentives is driven by the negative parents’ attitudes.
4 - Market Expansion and Targeting While Balancing Social and Financial Objectives: Evidence from Microfinance Organizations in Bangladesh
Jarrod Vassallo, University of Sydney, Sydney, Australia, Ahmed Khwaja, Jaideep Prabhu
This paper analyzes the strategic targeting decisions of social enterprises in balancing their social and financial objectives in an emerging market, high risk and low-resource context. We extend existing oligopolistic models of market entry to this hybrid organizational form to examine how social enterprises strategically expand into different market segments to manage the trade-off between these dual objectives. We assemble a novel archival dataset from multiple sources for the entire entry and expansion history of the three major microfinance organizations in Bangladesh. Using a longitudinal sample on 7,807 branches for 464 markets from 1992 to 2012, we estimate the game theoretic model using a control function approach that embeds a generalized residual procedure to account for discrete endogenous competitor actions. Contrary to the prevailing “mission drift” thesis, our results indicate that rather than drifting away from poorer markets, social enterprises engage in dynamic targeting to manage risks to their financial viability. They modulate their targeting of poorer markets and markets at risk of extreme weather events over time (“structured-flexibility”). Thus, targeting decisions are dynamic and evolve over time with the organization’s exposure to risk. Our findings may assist managers, impact investors, and policy makers in scaling up and supporting social enterprises. More broadly, our work contributes to understanding the role of market-based solutions to social challenges such as financial inclusion and income inequality.
TD11
Meeting Room C3.6
Social Media Engagement 2
Contributed Session
Chair: Işıl Büdeyri Turan, Özyeğin University, Istanbul, Turkey
1 - Music-Motion Synchronicity: A Crossmodal Transformer Model of Multi-level Customer Engagement with Social Media Videos
Erya Ouyang, Temple University, Philadelphia, PA, United States, Wen Wang, Xueming Luo

Short-form videos have become the dominant content format on social media platforms, enjoying a surge in popularity. These videos inherently encompass various modalities, incorporating images and audio to capture intricate interactions. Prior research has underscored the importance of cross-modal interactions in management and business, influencing aspects such as product evaluation, brand perception, and work performance. However, a knowledge gap remains regarding how marketers can systematically quantify reinforcement across modalities, leveraging insights as a decision support tool for multi-level social media engagement. This paper introduces the Music-Motion Synchronicity (MM Sync) framework, expanding on the Multimodal Transformer model (MulT), a leading multimodal deep learning algorithm. MulT's cross-modal attention mechanism facilitates the learning of interactions between multi-modal sequences across distinct time steps, dynamically adapting streams from one modality to another. Experimental results demonstrate that the MM Sync model significantly predicts multi-level customer engagement, encompassing both shallow and deep engagement levels. Heterogeneity analyses are conducted across influencer types and music genres, showcasing the diverse informative role of MM Sync in Multi-level Engagement. In conclusion, our findings and methodologies provide managerial implications for influencers, brands, and short-video platforms aiming to amplify the popularity and engagement of social media videos in this burgeoning entertainment market.

2 - Unhealthy and Healthy Choices: Exploring the Impact of Food Visuals on Social Media Engagement in the Grocery Industry
Eugene Cha, CJ CheilJedang, New South Wales, Australia, Fuxuan Jia, Tianle Chen, Chuqiao Liang, Jennny Park, Junbum Kwon

It is well-known that the grocery sector offers a kaleidoscope of food choices, presenting a unique challenge: how to effectively capture consumer attention in a space flooded with diverse food imagery? Our study ventures into this territory, identifying the most effective types of food visuals for social media presentation, then delving into how these can be optimized in different ways. We pay special attention to the intriguing dichotomy between healthy and unhealthy food imagery and their distinct impacts on consumer engagement.

We conducted a content analysis of Instagram posts from four top North American grocery stores. The findings indicate that posts featuring unhealthy food visuals garnered significantly more likes and comments compared to those with healthy visuals. Moreover, we observed that color factors, such as hue and symmetry, moderated the influence of unhealthy food imagery on social media engagement. Meanwhile, the engagement with healthy food visuals seemed less reliant on their visual appeal and was more influenced by the nutritional information of the food elements, particularly the sodium content in the food visuals.

This study contributes theoretically to social media marketing in grocery retail by analyzing the how different food visuals influenced on consumer engagement. Practically, it guides marketers in selecting food types and presenting them on social media.

3 - Age Stereotype of Celebrity Endorsement
Nan Zhang, University of Science and Technology Beijing, Beijing, China, People's Republic of, Xiaodan Zhang, Mei Huang , Fang Wan, Victor Chen
As an emerging interactive marketing channel, celebrities or influencers-generated content plays critical roles on social media platforms. Yet little is known about whether and how post types by celebrities across different age groups drive online followers’ engagement. This research investigates the diverging effects of celebrity endorsement and celebrity age on followers’ engagement on social media. Across four distinct studies, utilizing both publicly available big data from real social media posts (Study 1a and Study 1b) and experimental data (Study 2 and Study 3), this study examines the diverging effects of post type (brand-related vs. non-brand-related post) and celebrity age (younger vs. older) on followers’ engagement, and the underlying mechanism of interest fit and social identity. The results reveal that older celebrities triggered less social media engagement from followers by brand-related posts, while for younger celebrities, there existed an insignificant difference in posting brand or non-brand-related content, as their followers demonstrated more positive emotional responses towards brand-related posts. The effects occurred contingent upon the perceived interest fit and social identity. Specifically, an attenuated sense of interest fit and social identity among older celebrities with brand-related posts correlates to less social media engagement. Our findings contribute to the literature on celebrity endorsements on social media, and enrich the literature on self-branding by investigating a novel factor, the moderating role of a celebrity’s age in determining the effectiveness of their social media presence. This research provides practical guidance for celebrities and brands seeking to optimize their influencer marketing tactics in the digital realm.
4 - The Impact of Extremely Hot and Cold Temperatures on User Engagement
Işıl Büdeyri Turan, Özyeğin University, Istanbul, Turkey, Tolga Akcura, Kıvılcım Döğerlioğlu Demir

User-generated (UG) content on social media generates engagement, and high engagement improves brand performance metrics. Recent studies on UG contents focus on image and text analytics in creating higher engagement. User engagement to social media contents, however, involves a convoluted process which can be affected by individual mood and preferences and environmental cues. Temperature levels, as a part of environmental cues, affect consumers’ mood, product consumption, and decision making. Yet, we don’t know whether and how temperature levels affect user engagement on social media. This study examines how and why temperature influences user engagement using an experiment. Then, an empirical study is conducted with comprehensive real-life data that span five years to validate the experimental findings. The results indicate that extremely high temperatures decrease user engagement through decreasing the happiness level. Gender, weather sensitivity, hot/cold weather preference, environmental comfort and perceived temperature, and social media usage frequency are amongst the control variables. The research is the first study that studies the temperature impact in user engagement context while establishing the underlying causal mechanism. There are implications for researchers, social media managers, campaign planners and brand managers.

TD12
Meeting Room C4.1
Financial Decision Making
Contributed Session
Chair: Savannah Shi, Santa Clara University, Santa Clara, CA, United States
1 - Who Pays the Price for Bad Advice? The Role of Consumer Vulnerability, Learning and Confirmation Bias
Christine Eckert, EBS European Business School, Oestrich-Winkel, Germany, Julie Agnew, Hazel Bateman, Fedor Iskhakov, Susan Thorp
We study how consumers learn about the quality of expert services for credence goods.
For these goods, sellers provide the service, and also act as experts who determine consumers’ needs. Due to the nature of credence goods, consumers often must rely on their interpretation of ambiguous seller signals to judge seller quality. We focus on the context of financial advice and examine how consumers interpret ambiguous signals to infer
a financial adviser’s quality. Using data from an experimental survey, we propose and
estimate a model of learning by advice clients. Survey participants ("clients") make
subjective interpretations of advice on topics of varying difficulty, simultaneously learning
about the quality of the advisers giving the advice. Most participants fail to follow
Bayesian updating rules and use ambiguous signals from difficult topics to confirm their
prior beliefs about the advisers. Impulsive participants are more prone to such confirmation
bias. Participants who combine impulsiveness with low financial skills, and thus an
inability to judge the quality of advice, will pay too much to poor advisers –particularly
to malevolent advisers who present advice strategically. Our model can identify strategies
to help vulnerable consumers in the context of financial advice and beyond.
2 - Identifying the Value of Consumption Features in FinTech lending
Xiaodan Zhang, University of Science and Technology Beijing, Beijing, China, People's Republic of, Maiju Guo, Yanlai Chu, Guoqun Fu

The rapid advancement of digital technology enables FinTech companies to access insights into users' consumption tendencies, yet little is known about whether and how consumption features influence consumers' financial behavior. Using a transaction dataset from a microloan platform with assembling a unique purchase record dataset from a large e-commerce platform, this study constructs a metric for consumption diversity through natural language processing and explores its impact on subsequent loan and repayment behavior. Empirical analysis reveals a significant influence of borrowers' consumption diversity on their future loan application amounts and default probability. Specifically, higher consumption diversity is associated with increased loan application, whereas decreased default probability. This research contributes to the existing literature on FinTech default prediction and consumer financial decision-making by highlighting the potential utilization of prior consumption information not only for credit screening but also for mitigating default rates in microloan lending, thereby offering practical guidance for targeting and credit risk management.

3 - Banks are No Longer the Only Players: Measuring M-Banking Service Quality from Online Reviews
Seoyoung Lee, Dongguk University, Seoul, Korea, Republic of, Sunnyoung Lee, Minkyu Ku

As the demand for mobile payments increases rapidly, FinTech is reshaping the future of the banking industry (Agarwal et al. 2020; Agarwal and Zhang. 2020). In response to the competition from FinTech, traditional banks introduce integrated applications for digital transformation (Dapp et al.2015), while FinTech invests in security systems to enhance consumer trust (Kang, 2018). To compare them in a competitive relationship, we identify key dimensions - security, practicity, design, sociality, and enjoyment (Arcand et al. 2017) - to measure service quality through reviews. This study aims to understand consumer priorities in m-banking services and propose the survival strategies for banks.

The data from Google-Play-Store includes 523,772 reviews for banks and 777,670 for FinTech with 4 or more words from 2019 to 2023. We apply Support Vector Machine (SVM) to classify each review in terms of its emphasis on five extracted dimensions. We further employ dynamic topic modeling to capture time-related aspects of service quality and use correspondence analysis to present a perceptual mapping of m-banking on satisfaction attributes.

Among the topics classified by SVM, practicity related to FinTech has the most positive polarity with the sentiment score of 0.38, closely followed by security related to banks with the second-highest of 0.37. The result verifies the relative importance of service quality dimensions from unstructured data for FinTech or banks, influencing customer satisfaction and determining the competitive positioning of both. This study implies banks to fulfill user support and positive experiences as well as security to survive in the banking sector.

4 - In the Eyes of Investors: The Role of Attention in Investors' Reaction to Earnings Announcements
Savannah Shi, Santa Clara University, Santa Clara, CA, United States, Siqi li, Amanda Badger

Utilizing eye tracking technology to directly measure visual attention, we conduct an experiment to examine whether placing specific and quantitative information that summarizes the earnings news in the headline section of an earnings release affects investors’ reaction to the release, and how attention mediates such an effect. We find that investors make larger valuation changes when the headline section of an earnings release contains key financial information than when it does not. More importantly, only the attention to the headline section, measured three variables including headline dwell time, headline revisits, and recency of the last visit to the headline, mediates the effect of headlining key financial information on investors’ valuation judgement when earnings news is good. In addition, investors’ attention varies when they read different sections of the earnings release; and including key financial information in the headline section exerts a priming effect on the attention paid to other parts of the earnings release, despite that the information is being repeated. Taken together, our study contributes to the literature by providing new insights into the black box of investor attention and information processing when reading earnings releases.

TD13
Meeting Room C4.2
Game Theory and Analytical Models for Fake News
Contributed Session
Chair: Birger Wernerfelt, MIT, Cambridge, MA, United States
1 - Modeling Misinformation Spread for Policy Evaluation: A Parsimonious Framework
Richard Staelin, Duke University, Durham, North Carolina, United States, Yiting Deng
We develop a parsimonious framework for evaluating the efficacy of different approaches for limiting the spread of misinformation. We use this framework and simulation studies to determine the evolution of truthful and fake messages on social media platforms and then investigate the following policy interventions: 1) our suggested approach of having the platform require senders of messages to also state their perceived (possibly incorrect) veracity of the message; 2) provide some accuracy nudge to increase the number of potential readers who can identify fake messages; 3) have the platform flag fake messages; and 4) have the platform demote or down-rank fake messages. We find when a significant number of senders are able to correctly identify the veracity of the message, the market can self-regulate under our suggested approach. If this is not the case, we find the other approaches are effective in reducing the spread of misinformation.
2 - Navigating Rumors: Firm Defense and Regulation
Yifang He, University of Pittsburgh, Pittsburgh, PA, United States, Yue Wu
Media often generate rumors about celebrities or firms to attract consumers, but these rumors may be fake news and hurt their targets’ reputation. The targets of rumors can respond by remaining silent or defending against rumors, although such defenses may also be skeptical for consumers. Moreover, defense is costly and may inadvertently cause broader dissemination of the rumor. To protect consumers, regulators can penalize media or firms that send false information. In this paper, we adopt a game-theoretic model to explore the dynamics of rumors and interactions between dishonest media, firms and regulations. We seek to answer whether strengthening regulation can help reduce the dissemination of false rumors and related social harm. The model demonstrates that increasing the penalty for false signal can increase consumers’ belief that the rumor is truthful, creating a stronger incentive for the firm to actively defend their reputation. We show that a firm’s defense against rumors can hurt its reputation by generating additional discussions. This heightened attention discourages the firm from defending against the rumor, leading to more consumption of false rumors and higher social harm.
3 - The Origin and Function of Arbitrary Signals: Making False Statements, Having Long Hair, and Smoking Virginia Slims
Birger Wernerfelt, MIT, Cambridge, MA, United States

In the 1960’s a lot of men grew their hair long, in the 1970s some women smoked Virginia Slims cigarettes, and today many Americans, while knowing better, voice the view that the 2020 presidential election was stolen. Each of these actions had two functions: They promote a specific political view (a relaxation of social norms, gender equality, and Donald Trump’s populism, respectively.) and they give the actor a ticket to membership in the group of others who send the same signal.[1] At the same time, the signals were costly; many other members of society looked at the senders with disdain, considering them uncivilized, confrontational, and naive, respectively. And yet, since their willingness to bear these costs tells others that they place a high value on objectives, they increase the felt solidarity between signalers thus further increasing the value. Beyond this, costs go down as more signalers join the case, since the outlier behavior becomes less unfamiliar to others (Benabou and Tirole, 2006). The theory differs from more traditional analyses of social movements because the costs of bearing punishments from non-senders and the value of membership in an attractive peer group may outweigh the intrinsic value agents place on the cause.



[1] It is not hard to find examples in which a person’s opinions influence how others treat them: Many Americans would not socialize with individuals who support particular aspects of the opposite political ideology (e. g. about global warming, abortion, or gun rights). Similarly with religion, morality, race relations, etc..

TD14
Meeting Room C4.3
Advertising: Machine Learning Approaches
Contributed Session
Chair: Fatemeh Gheshlaghpour, University of Chicago Booth School of Business, Chicago, IL, United States
1 - An Image Deep Learning Analysis on Movie Posters to Optimize the Localization Effect of Advertising
Jongdae Kim, Chonnam National University, Gwangju, Korea, Republic of, Youseok Lee, Junhee Seok
We propose an image deep learning-based approach to gauge the impact of localization in international advertising. Specifically, this study focuses on scrutinizing the efficacy of movie poster images as a promotional tool in the film industry. Employing convolutional neural networks (CNNs), we extract visual information from movie poster images and quantify the level of localization. Assessing the influence of poster localization on the U.S. film's financial performance in a foreign market (South Korea), we find a notable a U-shaped curve, indicating that the optimal degree of poster localization lies at either extreme: either fully non-localized or highly localized. This research is poised to contribute by proposing an empirical framework for examining the impact of international advertising strategies through the lens of movie poster localization. Furthermore, we expect that the outcomes of our study will provide valuable insights for decision-makers in determining the appropriate level of localization.
2 - Exploring the Interrelations of Music on Advertising and Chart Performance
Yana Asenova, KLU, Hamburg, Germany, Jan Becker

Licensed music has been used by the advertising industry for many years as a tool to catch the attention of the viewers and trigger specific emotions they connect with it. Its positive impact on brand recall and purchasing behavior has therefore been known to industry professionals and academic researchers for a long time. Considering the sometimes-enormous reach of advertising campaigns (e.g., Super Bowl commercials), the positive effect should not be limited to the success of the advertisement only, but also to the licensed song itself. Yet, the interdependencies between advertising and chart success have not been investigated so far.

This study explores the relationship between advertising and music marketing success, using the motion picture industry as an example. For a large sample of movie trailers, we use a vast and comprehensive set of longitudinal data on the characteristics and performance metrics of the movie trailers, the movies, and the music. The multiple data sources allow for the implementation of several machine learning techniques to analyze the dynamic interrelationships between advertising and music marketing success in the empirical study. Preliminary results provide insights that are relevant for both the advertising and music sector and contribute to the existing research on music in advertising.

3 - Strategic Personalization
Fatemeh Gheshlaghpour, University of Chicago Booth School of Business, Chicago, Illinois, United States, Sanjog Misra, Pradeep Chintagunta

Firms often face the challenge of implementing personalized marketing policies across multiple marketing instruments in order to maximize a specific objective function, e.g. short-term profits. In this paper we evaluate different marketing policies that differ in terms of which subset of instruments is personalized across customers. We compare these personalized policies to assess the instrument for which profits are most sensitive to personalization. We use structured DNNs to capture individual heterogeneity and design the optimal marketing policy based on that information. Our results from a large-scale direct-mail field experiment for a financial institution in South Africa that varied the advertising content and interest rate offers to consumers to promote loan applications, show how effective personalized advertising content can permit the firm to offer higher interest rates. Thus, effective advertising content can pay off for the firm not only by increasing the apply probability, but by supporting a higher interest rate. We see that while the full-personalized policy increases the expected profits per unit loan amount by 174.5%, personalizing only advertising content increases the expected profits per unit loan amount by 172.9% in the absence of personalized interest rates, and only offering the optimal personalized interest rates does not affect the expected profits per unit loan amount as much.

TD15
Meeting Room C4.4
Influencer Marketing: Inferences from Quasi Experiments and Big Data
Contributed Session
Chair: Mengjie (Magie) Cheng, Harvard Business School, Boston, MA, United States
1 - Content Management Policy and Influencer Marketing on Social Media Platforms: A Study of Shopping Links in Sponsored Content
Junni Chen, Shanghai University of Finance and Economics, Shanghai, China, People's Republic of, Qi Sun, Siqi Pei, Yong Liu

Firms increasingly turn to influencers on social media platforms to promote brands and products. Yet, we know little about how a platform’s content management policy affects the effectiveness of influencer marketing. Employing a difference-in-differences approach, this paper utilizes a quasi-experiment on a leading social media platform, which abruptly restricted the inclusion of shopping links in sponsored posts, to examine how the platform's content policy change affects influencers' creative behaviors and user engagement. Contrary to the platform's goals, we find that influencers crafted sponsored content that resembled previous sponsored (organic) content to a greater (lesser) extent following the ban on shopping links. The increased similarity to sponsored content led to reduced user engagement with sponsored content after the ban. We further show changes in user behavior, notably users refraining from saving sponsored content for future reference, may have contributed to the decline in engagement. Compared with influencers with more followers, those with fewer followers experience a greater reduction in user engagement following the policy change. Highlighting the importance of considering influencer and user behavior changes when evaluating content management policies, these findings provide important implications for social media platforms, influencers and users, and scholars interested in social media and influencer marketing.

2 - Sponsorship Disclosure and Influencer Marketing
Feier Li, Peking University, Beijing, China, People's Republic of, Qingfu Chen, Hongju Liu

In the burgeoning landscape of influencer marketing on social media, sponsorship disclosure has drawn increasing attention from academics and practitioners. This study delves into the repercussions of a significant policy change concerning sponsorship disclosure on a large social media platform. Originally mandating automatic disclosure for all sponsorships, the platform later allowed influencers the discretion to voluntarily disclose after the policy change. Focusing on influencers in the beauty and lifestyle category, we collect a comprehensive dataset encompassing user-generated contents, engagement metrics, and sponsorship indicators.

A difference-in-differences analysis reveals that clearly disclosed sponsorships, in contrast to non-disclosed ones, yield a significant increase in likes, albeit accompanied by a reduction in the rate of gaining new fans. This audience response triggers a supply-side shift in influencers' content creation behavior. Following the policy change, influencers create more content, including both sponsored and non-sponsored posts.

While the relaxed sponsorship disclosure requirement proves beneficial for influencers and the platform in terms of attracting new fans, increasing content quantity, and boosting platform advertising revenue, it also leads to a change in engagement behavior. This study underscores the complex decisions influencers and platforms must make in strategically managing sponsorship disclosure, considering its diverse implications on audience engagement, content creation, and overall effectiveness of influencer marketing.

3 - The Value of Silence: The Effect of UMG vs. TikTok Licensing Dispute on Music Demand
Mengjie (Magie) Cheng, Harvard Business School, Boston, United States, Elie Ofek, Hema Yoganarasimhan

Our work studies how original content can spur social media trends. While many social media posts are original, a significant portion are content recreations that incorporate elements such as music, actions, objects, or text taken from the original (or genesis) post. The participation of users in recreation behavior over time can form a social media trend. To understand such content diffusion patterns, we analyzed 2,044,432 genesis videos that created original music posted by U.S. creators on TikTok. We tracked subsequent recreations leveraging the same original music within 120 days of the genesis video. Our data suggests that only 43,571 out of the 2 million genesis videos (2.13%) inspired recreations. Yet, genesis videos with recreations achieved, on average, 15 times more views and engagement than those without recreations. We find that the number of content recreations follows an inverted-U pattern as a function of time since the genesis video was posted, and that a 1% increase in recreations leads to a 0.61% increase in views for the genesis videos. In addition, while genesis videos typically garner the most attention, recreation videos can gain a 'second-mover' advantage, especially if they join the trend early enough and their re-creator has a larger follower count. Furthermore, genesis videos inspire more recreations when their first few recreation videos diverge from the genesis content and are from creators with more followers. Our study contributes to the literature on innovation and creativity, influencer marketing, and copycat behavior, providing important managerial implications for creators, brands, and platforms.

TD16
Meeting Room C4.5
Livestreaming 4
Contributed Session
Chair: Ziyi Wang, Renmin University of China, Zhongguancun No.59,Haidian District, Beijing, 100872
1 - Value Co-creation in Live Streaming E-commerce
Dingning Liu, Monash University, Caulfield East VIC, Australia, Shanfei Feng, Junzhao Ma

Presenting author: Dingning Liu, Monash University, dingning.liu@monash.edu

Abstract:

Live streaming e-commerce is a new retail format that blends the different activities of shopping, entertainment and socialising into a single experience. Since its inception in 2016, it has become one of the most important retail formats in China, accounting for 20% of China’s total e-commerce sales now. Despite its commercial significance, there is relatively little academic research on the topic and the reasons behind its success remain poorly understood.

This research aims to close this research gap by investigating the factors contributing to the success of live streaming e-commerce through the lens of the value co-creation theory, which posits that collaboration among multiple stakeholders enhances collective value (Lusch & Vargo, 2006; Vargo & Lusch, 2016; Vargo et al., 2008; Vargo et al., 2020). Field data obtained from a major live streaming platform was used to examine the effect of the three key dimensions of value co-creation (i.e., interaction, experience and relationship) on sales performance. The results show that these three dimensions of value co-creation impact sales performance differently for different product category characteristics. This research highlights the moderating role of product category characteristics and is among the first to leverage panel data over traditional one-period surveys, offering new insights for effective marketing strategies in live streaming e-commerce.

Keywords: E-commerce, Live Streaming, Value Co-creation

2 - Persuasive Storytelling in Live Streaming: A Dual-Route Approach to Increasing E-Commerce Sales
Yutian Shen, Shenzhen Technology University, Guangdong, China, People's Republic of, Pei Yu Chien, Mei-Chih Hu, Ching-Yan Wu

While live streaming e-commerce has generated significant business value, this value is primarily created by a few top-tier live streamers. In fact, many small streamers struggle with low traffic and a low conversion rate. Attractive, persuasive narrative content can help increase sales. Specifically, a live streamer's persuasive narrative can connect with customers, convincing them to make a purchase and thereby drive sales. However, very few studies have been dedicated to exploring this effect. The Elaboration Likelihood Model (ELM) is a dual-route psychological theoretical framework designed to understand how persuasive messages are processed and, consequently, influence an individual's attitude or decision-making. Thus, we propose a model built on the ELM to contribute to the literature on live streaming e-commerce. More specifically, this model explains how persuasive narrative content by live streamers influences customers' purchase decisions by applying two crucial factors within the ELM framework: central cues and peripheral cues. Additionally, while peripheral cues are superficial information, they are likely to influence the relationship between central cues and customers' behaviour. Therefore, we consider peripheral cues as moderators in the model as well. These findings aim to offer strategies for small streamers to improve sales and, consequently, increase overall sales in live streaming e-commerce. More specifically, streamers can refine their content strategy based on these findings to effectively attract customers. Additionally, these findings can be used to train new live streamers on how to better attract customers' attention and increase sales.

3 - Now or Never: Time Pressure in Livestream Shopping
Sophia Yiu, RMIT, Melbourne, Australia, Ashish Kumar, Daniel Rayne

Livestream shopping has rapidly evolved as a dynamic e-commerce platform, significantly influencing customer online purchase behaviour. This study investigates the impact of time pressure strategies on sales and engagement within the livestream shopping environment. A comprehensive analysis was conducted on 1824.5 hours of content from 271 livestream sessions by 12 sellers. These sellers were categorised based on experience and further classified as mega, macro, or micro, considering their follower counts and sales figures. This approach aims to provide an in-depth understanding of time pressure's impact across different seller levels. Utilising advanced AI for video-to-text transcription, the study involved transcribing a month's livestream sessions. The content was then analysed using Natural Language Processing (NLP) and Latent Dirichlet Allocation (LDA) to identify persuasive strategies and pinpoint instances of time pressure application.

The research categorises time pressure into three distinct strategies: Limited Time Scarcity, Limited Quantity Scarcity, and Hard Sell. It also examines the role of seller characteristics in the effective application of these strategies. Findings indicate a statistically positive correlation between time pressure and both sales and customer retention. This aligns with existing literature that highlights scarcity-induced promotions as key drivers of purchase intention. Notably, sellers with higher rankings, extensive experience, and independence are adept at leveraging time pressure strategies to enhance sales and retention. This study underscores the critical role of time pressure in customer decision-making within livestream shopping. It provides insightful implications for practitioners and contributes to the broader understanding of contemporary e-commerce dynamics.

4 - I am Being There: Will the Background Influences Purchase Intention in Live Streaming E-commerce? Insights from Physical Presence Perspective
Ziyi Wang, Renmin University of China, Beijing, China, People's Republic of

Purpose: This paper aims to investigate the effects of different backgrounds in live streaming e-commerce (LSE) on purchase intention from the physical presence aspect.

Design/methodology/approach: This research used a multimethod approach with four studies. After an exploratory qualitative study conducted for the model, three designed experiments were employed to prove casual relations between the hypothesized relations. Then, the data collected from Douyin provides new support for the framework.

Findings: The results suggest that purchase intention is higher with a real-world (vs. green-screen) background. Physical presence mediates the main effects of real-world background on purchase intention. Importantly, the differences between those backgrounds are mitigated when focusing on distal (vs. proximal) sensory imagery and strong zero-sum beliefs.

Practical implications: These findings offer actionable insights for small and tiny marketers on how to effectively select the LSE background, providing practical strategies to enhance marketing effectiveness, which is also helpful in shaping the LSE industry norms.

Originality/value: The previous study is not sufficient to explore the background in LSE, where consumers will perceive various information and stimuli. Rather than focusing on the social presence in live stream, this study examines the physical presence as a sense of being there, which will be evoked with the stimulus of the background in the LSE situation. Drawing on SOR theory and distance theory, the finding enriches the research on environment psychology in the LSE field.

Keywords: Live-streaming e-commerce, Live-streaming background, Physical presence, Purchase Intention, Sensory imagery, Zero-sum beliefs

TD17
Meeting Room C4.6
Health Care Marketing: Provider Decision Making & Insurance
Contributed Session
Chair: Jong Yeob Kim, Nanyang Technological University, Singapore, Singapore
1 - Prescribing Sustainability: How ESG Impacts Physicians' Drug Choices
Minjung Kwon, Syracuse University, Syracuse, New York, United States, Si Cheng
While investor focus on Environmental, Social, and Governance (ESG) initiatives in pharmaceuticals surges, their practical impact on healthcare consumer behavior remains largely unexplored. This study delves into this critical gap by investigating how ESG events influence brand switching among pharmaceutical products. We adopt a unique lens: analyzing the prescribing behavior of healthcare providers. We examine whether adverse ESG headlines about a manufacturer trigger increased switching to substitute drugs, potentially eroding market share and impacting long-term market presence. To explore this complex relationship and obtain comprehensive insights, we leverage multiple rich data sets. RepRisk, MSCI ratings, and RavenPack data provide quantitative and qualitative context into ESG events, while claim-level prescription data offers granular details on healthcare providers' drug choices. This multifaceted approach allows us to shed light on the previously unexplored link between ESG, healthcare providers, and drug choices. Our findings offer valuable insights for both pharmaceutical marketers and investors navigating the evolving landscape of sustainable business practices.
2 - Balancing IQ and EQ: A Field Experiment on Human-AI Collaboration
Huazhong Zhao, City University of Hong Kong, Kowloon, New York, Hong Kong, Ting Hou, Qiang Li, Meng Li
The emergence of generative artificial intelligence (GenAI) is leading to a profound evolution in many industries. Despite the consensus that AI outperforms humans in many aspects, especially in low logic requirement tasks, many scholars believe that AI is unlikely to reach human intelligence level as a whole package in the foreseeable future. Nevertheless, there is still a lack of detailed discussion on the nuanced aspects where AI falls short compared to humans in the current literature. This paper aims to fill this research gap by conducting a field experiment on a healthcare platform focusing on two aspects of comparison between human intelligence and AI: Intelligence Quotient (IQ) and Emotional Quotient (EQ). First, we find that AI excelled in medical writing tasks with low requirements for professional expertise due to the high IQ. However, in tasks requiring higher levels of professional expertise, AI lags behind human physicians. We attribute this difference to the insufficient EQ of AI, evidenced by its inability to match the emotional and logical depth achieved by human professionals. In order to mitigate AI's EQ limitations, we further investigate the potential of human-AI collaboration by providing AI with additional human assistance in two aspects: emotional content or logical structure. We find that emotional content written by a real person does not help AI to generate better medical writing, while logical structure preset by a physician significantly improves readers' perception. This provides profound insights into addressing the current shortcomings of AI and guides the development of large-scale language models.
3 - Impact of Influencers and Microgrant in the Adoption and Valuation of Subscription-based Services
Prakash Satyavageeswaran, Indian Institute of Management Udaipur’, Udaipur, India, Madhur Mohan, Vedha Ponnappan, Raghunath Rao

Out-of-Pocket Expenditure (OOPE) in healthcare is a significant challenge in low- and middle-income countries. Prepayment-based subscription service offers a solution to overcome the debt trap and financial hardships due to OOPE. However, the uncertainty about future usage, provider credibility, and liquidity crunch could present significant challenges in the adoption of such plans. The authors examine the impact of two marketing levers - influencer and credit provision - on the adoption of subscription services and the relative efficacy and valuations of these levers relative to mere information provision (control). Using a randomized controlled field experiment among a tribal population in India, authors find that microgrant significantly increases adoption and valuation relative to the control arm, signaling the importance of credit provision in such liquidity-constrained contexts. While the local influencers greatly enhanced the valuation of the service, they did not boost adoption relative to the control arm. The authors also estimate heterogenous treatment effects using ML based causal forest approach. The research contributes to the theory and practice by extending the conversation on the causal effect of influencers and credit provision on the adoption and valuation of a subscription service where future usage is uncertain.

4 - Health Insurance and the Dynamics of Patient Decision Making
Jong Yeob Kim, Nanyang Technological University, Singapore, Singapore, Masakazu Ishihara, Vishal Singh
This article exploits variations in eligibility age thresholds and coverage amounts in the South Korean dental market to examine healthcare utilization, strategic delays, and the dynamics of treatment choices by elderly patients. We use two primary data sources: a nationally representative health survey and a long panel of data on individual-level prescription and treatment choices from a large university hospital in South Korea. Our empirical approach exploits temporal variation in eligibility thresholds, and out-of-pocket costs for different forms of treatments to compare age cohorts just meeting the insurance criterion vs. not. We find large increases in healthcare utilization post insurance expansion, and strong evidence of strategic delays—a particular form of moral hazard—in treatment choices for patients just below eligibility thresholds. Evidence from reduced-form analysis guides us to develop a dynamic structural “patient life-cycle” model to examine patients’ strategic delays of treatment and adoption of new technology under different dental insurance policies. From a methodological point of view, we demonstrate how the age threshold can be used as an exclusion restriction to estimate discount factors in dynamic models of healthcare choices. Our results show that lowering the eligibility threshold from age 65 to 60 increases treatment timing and improves patient welfare by 6.8% but also leads to a significant increase in government spending, resulting in a net negative overall change in social welfare. Implications of our findings for other contexts, such as the expansion of Medicare in the United States to include dental benefits and lowering age thresholds, are discussed.
TD18
Meeting Room C4.7
Game Theory and Sustainability
Contributed Session
Chair: Jiaxuan Li, Tianjin university, Tianjin
1 - How Should a Platform Promote Sustainable Products? Push versus Pull
Chen Cao, Hong Kong University of Science and Technology, Kowloon, Hong Kong, Mengze Shi
This paper examines the optimal approach for an e-commerce platform to promote sustainable products. Inspired by real-world practices, we propose new corporate social responsibility initiatives that link sustainability with promotions. We analyze two simplified promotion strategies: pull strategy as rewards to consumers and push strategy as commission fee discount to a seller. We consider the platform's promotion strategy as the premise for both the seller's decision and the consumer's purchase decision, and seek to determine the most effective promotion types that maximize the platform's profit in different scenarios. Specifically, we analyze the cases where there is only one seller and where there are two sellers on the platform. Our results show that for a platform with a single seller, both the platform and the seller benefit from cash rewards to consumers. In the case of a platform with two sellers, although both pull and push strategies can make the platform more profitable, it is more effective for the platform to offer cash rewards to consumers. Our findings contribute to the existing literature by combining corporate social responsibility with promotions, and provide insights for e-commerce platforms seeking to promote sustainable consumption.
2 - Sustainable Consumption: A Strategic Analysis
Wilfred Amaldoss, Duke University, Durham, NC, United States, Siddharth Prusty
Consumers’ growing concern for the environment has motivated firms to offer sustainable products in several categories. An exploratory survey shows that many consumers desire sustainable products and are willing to pay more for them, but some consumers dislike sustainable products and want to pay less for them. Using a theoretical model where consumers are heterogeneous in their preference for sustainable products and firms are horizontally differentiated, we investigate the strategic implications of sustainable consumption. First, our analysis shows that product valuation moderates the impact of consumer preference on price, yielding results that run counter to intuition. Second, irrespective of product valuation, we find that competing firms’ profits can decrease with consumers’ desire for sustainability but increase with consumers’ dislike for sustainability. Third, we clarify when and why enforcing minimal sustainability standards can backfire and reduce aggregate consumer welfare. Fourth, we show that competing firms could earn strictly lower profits if they offer a customized product for each consumer segment: a sustainable product for consumers who desire sustainability and a baseline product for those who dislike sustainability. Moreover, symmetric firms could adopt an asymmetric product strategy. Fifth, in contrast to the luxury goods literature, we show that snobbishness hurts the profits of firms offering sustainable luxury goods despite softening price competition. Finally, we find that firms’ profits can increase as consumers’ political orientation shifts to the right.
3 - Audit Dilemmas: Responsible Procurement Amidst Supplier-Auditor Collusion.
Jiaxuan Li, Tianjin University, Tianjin, China, People's Republic of, Xiaopei Zhao, Yunchuan Liu, Jianxiong Zhang
In the implementation of the audit, the government is faced with the choice of allowing the manufacturer to conduct its own internal audit or hiring a third-party audit firm. However, both options have potential problems, such as internal audits can lead to corruption of the manufacturer itself, while external audits can lead to collusion between the manufacturer and third parties. In considering the government's economic and equity objectives, the main issue of this study is to determine when the government should allow internal audits and when it should opt for external audits in order to achieve optimal audit results. In this paper, a game theory model is established to study the influence of government audit strategy selection on supplier product quality information disclosure under collusive risk. We draw some counter-intuitive conclusions: internal audits are not a blanket bad option: while there is a risk of corruption by manufacturers themselves, in some situations it may be more effective for governments to allow manufacturers to conduct internal audits themselves, especially in terms of reducing audit costs and increasing efficiency. External audits are not an absolute guarantee: While external audits are considered to provide independent audit results, our research found that governments still need to be concerned about possible collusion between manufacturers and third parties when selecting external audits.
These findings provide a new perspective on government decision-making when auditing manufacturers' products, and are expected to provide substantial reference for achieving a balance between economic and equity objectives.
TD19
Meeting Room C4.8
Digital Platforms: Pricing, Data and User Experience
Invited Session
Special Sessions
Chair: Yi Zhu, University of Minnesota, Minneapolis, MN, United States
Co-Chair: Lei Zhuang, Fudan University, Shanghai
1 - Buy Now, Pay Later: Integrated Versus Disintegrated Consumer Installment Services
Xiabing Zheng, University of Science and Technology of China, Hefei, China, People's Republic of, Buqing Ma, Bo Zhou, Ying-Ju Chen
Firms are increasingly adopting buy-now-pay-later (BNPL) services to facilitate consumers' purchase decision. This paper investigates the impact of BNPL services on the firms' profits and consumer surplus when the product and the installment plan are offered by one firm (i.e., integrated model) or two independent firms (i.e., disintegrated model). Depending on the service model, firms optimally decide on the retail price and the interest rate, and consumers with heterogeneous incomes decide when and how to make the purchase in a two-period model. First, we find that BNPL enables firms to engage in both price discrimination and reverse price discrimination. In particular, high-income consumers end up paying a higher effective price than lower-income consumers with BNPL who default later, but lower than that paid by lower-income consumers with BNPL who make the full payment. Second, the introduction of BNPL can lead to a lower market coverage and drive out consumers who may, ironically, need the service more, because of the higher price from BNPL. Third, compared to its absence, BNPL can lead to a higher consumer surplus. Surprisingly, the integrated model can yield a lower consumer surplus than the disintegrated model. Lastly, we consider the revenue-sharing contract between the seller and the lender in the disintegrated model. We find that the seller can be worse off from a higher revenue-sharing rate for herself.
2 - Enhancing User Experience: Unveiling the Role of the Quality Score Metrics
Lei Zhuang, Fudan University, Shanghai, China, People's Republic of, Zsolt Katona, Yi Zhu
Social advertising platforms and search engines use the so-called quality score to favor certain advertisers in their advertising auction. Quality scores might be based on a number of factors including ad relevance, click-through rate, and landing page experience. This work examines the role of quality score metrics by the platform in enticing advertisers with heterogeneous investment costs to improve their page quality, and its impact on advertisers' profits and the ad platform's revenue within an environment where page quality is an endogenous decision for advertisers. The results suggest that a quality score function that rewards advertisers in excess of their click-through rates motivates advertisers to invest in their page quality in a non-monotonic way. At the same time, investments that are incentivized by the platform's rewards have dual effects on auction dynamics. On the one hand, advertisers have the potential to reduce the cost per click through investments. However, such investments also intensify auction competition by raising advertisers' expected value of clicks. We find that when advertisers' investment costs are similar, the intensified competition could offset the benefits of quality investments. Therefore, higher rewards do not always translate into higher profits, particularly for advertisers ending up with top positions. In terms of the platform's revenue, despite advertisers' increased bids, a quality score function in excess of click-through rates often comes with a sacrifice of the platform's revenue, with the exception of the case where advertisers bid the most aggressively. Several managerial implications for both advertisers and the platform are discussed.
3 - How do Peer-to-Peer Platforms Affect Durable Asset Prices? Theory and Evidence from a Natural Experiment
Raghunath Rao, University of Texas at Austin, Austin, TX, United States, Andreas Kraft, Garrett Sonnier

Peer-to-peer (P2P) platforms enable the sharing of durable assets, especially in short-term housing and transportation. Since owning an asset can bring owners potential income through the utilization of slack capacity, P2P platforms may enhance overall asset value and thus increase prices. However, platform services may obviate the need to own the asset in the first place, depressing values. Due to these countervailing forces, the effect on its price is ambiguous.

We use the 2016 regulatory-induced exit of Uber and Lyft from Austin, TX, as a natural experiment to study the impact of P2P platforms on vehicle prices. Utilizing data on over 5 million transactions, we estimate the causal impact of P2P platforms on asset prices. We find robust evidence of both positive and negative price effects depending on this cost. First, we find that the platform increases the prices for fuel-efficient vehicles and reduces the prices for relatively fuel-inefficient vehicles. This result implies that environmental gains from the use of fuel-efficient vehicles on the platform are potentially crowded out due to these vehicles becoming more expensive.

Overall, our estimates suggest that platforms led to a modest increase in new vehicle prices across sedans but decreased prices across pickups. Conversely, the prices of newer vintages of used sedans, highly utilized on the platform, experience a significant increase. However, the older vintages show a significant softening of the prices. Disaggregate analysis suggests that consumers have different preferences for vehicles when using these for pure consumption versus income generation.

TD20
Meeting Room C4.9
Private Labels and Branding
Contributed Session
Chair: Kaye Chan
1 - Private Label Supply Strategies of Dual Branders: A Generalizable Empirical Analysis
Kusum Ailawadi, Tuck School of Business at Dartmouth College, Hanover, NH, United States, Mercedes Martos-Partal, Yu Ma, Oscar Gonzalez-Benito

Private label (PL) is a very substantial portion of total CPG sales in many countries. National brand (NB) manufacturers love to hate the competition they face from PL and yet, many of them actually make PL products for retailers in addition to their own NBs. The very limited empirical research on such “dual branding” examines the supply of PL by a NB manufacturer to a specific retailer in a category and the impact on the performance of the supplier’s NBs in the retailer’s stores. However, the decision to get into the PL business is, or at least should be, a strategic one, and there is substantial variation in the broader PL supply strategies of dual branders. Some supply PL only in their NB categories while others also, or only, do so in categories where they don’t offer NBs; some supply only to one retailer while others supply to multiple retailers; some shrink their PL supply business over time while others grow it. The purpose of this research is to identify segments of dual branders based on their PL supply strategy and examine the association with characteristics of their NB business. We take an “empirics first” approach yet aim to provide generalizable insights. We do so using a unique dataset of over 700 dual branders who supply PL to one or more of six major retailers in the Spanish grocery market across 270 CPG categories.

2 - The Role of Consumer-based Brand Equity on the Prolongation of Trademarks
Louisa Pfeifer, University of Hamburg, Hamburg, Germany, Thomas Schreiner, Henrik Sattler

Trademarks are an integral part of a firm’s brand equity. While extant research looked at the outcomes of trademarks, e.g., a firm’s financial valuation, profitability, and productivity, limited knowledge exists regarding the factors influencing a company’s decision to prolong or terminate trademarks. For the first time, the presented research investigates the effects of consumer-based brand equity (CBBE) dimensions on the decision to prolong or terminate trademarks and how trademark characteristics, and the level of regulation in a product category moderate these effects. Utilizing a unique dataset covering 25 countries and 57 product categories from 2001 to 2019, tracking 1,446 trademarks, this research establishes empirical evidence for CBBE’s critical role in trademark prolongations. Specifically, the CBBE dimension relevant stature (reflecting brand knowledge, esteem, and relevance) positively affects trademark prolongation, whereas the CBBE dimension energized differentiation (reflecting brand uniqueness) negatively impacts trademark prolongation. These results translate into important managerial implications: Our results suggest that companies should invest in building brand knowledge, esteem, and relevance, as brands high in relevant stature tend to show an increased likelihood of trademark survival. Moreover, managers of differentiated brands should think beyond plain trademarks, foster brand associations such as slogans, logos, color, font, and shape related to the brand name, and protect them via a brand association trademark since these associations mitigate the negative effect of energized differentiation.

3 - From Store Shelves to Screens: How Platform Private Label Entry Affects Consumer Engagement
Wenyan Tuo, University of Arizona, Tucson, AZ, United States, Xinying Hao, Yong Liu

Platform Private Labels (PPLs) are reshaping consumer engagement and competition in e-commerce. Companies like Amazon launched their Amazon Basics line, leveraging online data to shift away from conventional brick-and-mortar in-store private labels and capitalize on the search-based online market.

This paper investigates the intricate impact of PPL introduction through the lens of consumer engagements. We utilized a Difference-in-Differences analysis with propensity score matching at the product level, leveraging a dataset of nearly 3 million product reviews provided by consumers for 106,511 products across 7 categories. Utilizing machine learning, we clustered keywords into market segments and quantified product descriptions and images with state-of-the-art language and image processing models.

Preliminary results suggest that the introduction of PPLs significantly stimulates consumer engagement in the related product market. This effect potentially stems from the diversification of the product assortment, the encouragement of innovation, or the escalation of competition.

This paper contributes to a comprehensive understanding of private labels in the digital economy and provides pivotal insights into the strategies that platforms can employ to foster consumer engagement through PPLs.

4 - The Impact on Branded Consumption Behaviour Due to Environmental Turbulence
Kaye Chan, University of Technology Sydney, New South Wales, Australia

Environmental turbulence can occur from technology, market, and competitor changes. These changes impact the market and how consumers behave and purchase brands. It has been theorised that environmental turbulence can impact customers’ choices (Rego et al, 2021). While research has predominantly focused on the impact of environmental disruption on brands and retailers (e.g., Lesscher et al., 2020; Kumar et al., 2018; Akturk and Katzenberg, 2022), limited research evaluates the impact on consumer choices and subsequent behaviours. Through a quasi-experiment, the impact of environmental turbulence on physical availability and the consumer's branded purchase behaviour is evaluated. This paper addresses the impact of retail disruption on individual households, accounting for the large heterogeneity in purchase behaviour previously demonstrated. The study will model the individual impact on branded purchases due to the turbulence, including the impact of heavy and light buyers' purchasing behaviour and their resultant branded purchasing. A sub-analysis of individual households that did not purchase from the retailer in the 12 months following closure will also be undertaken to evaluate the potential underlying drivers. This paper contributes to the brand and channel literature by showing the impact of retail disruption on consumers at a household level. This highlights the need for brand owners to consider their consumer communications to respond to changes in retail channel access. Further, it highlights the importance of understanding the impact of such environmental turbulence at the individual level due to the loss of important data at an aggregated level due to consumer heterogeneity.

TD21
Meeting Room C4.10
Consumer Behavior: Response to Algorithms
Contributed Session
Chair: Ivan Ho
1 - Algorithms are Gendered: The Masculine Algorithms and Their Influence on Consumers
Yegyu Han, IE Business School, Madrid, Madrid, Spain, Sang Kyu Park, Chris Janiszewski

Algorithms are rapidly integrated to our daily lives, from curating music and predicting stock prices to driving cars and providing medical advice. To better understand how consumers perceive and interact with algorithms, prior literature has dedicated to understanding when and how consumers form relationships with algorithms and shape attitudes toward them, as well as the various factors that influence algorithm adoption. This research aims to contribute to the stream of literature on how people perceive and interact with algorithms. Drawing on prior research demonstrating that consumers often attribute gender to various inanimate objects and abstract concepts, we suggest that consumers ascribe masculinity to algorithms. We theorize that this association stems from the predominant masculine traits of algorithms and the male-dominated STEM industry in which algorithms are developed. Through a series of studies, we investigate the downstream consequences of the algorithm-masculinity association on choice and consumption experiences, as well as its effect on perceptions of algorithm-created outcomes.

2 - When It Comes to Love, Do You Believe in Fate or Choices? Consumer Reactions to Algorithm (vs. Human) Dating Recommendation Agents
Xi Song, University of Macau, Zhuhai, Macau, Matthew Tingchi Liu, Mingxia Zhu, Ziying Mo

The proliferation of recommendation algorithms has altered the way people consume, as well as how people meet and start relationships. Prior research suggested that consumers are averse to having algorithms execute human-oriented tasks. In matters of love, finding a date is inherently subjective, whereas online dating services present a modern way of developing romantic relationships, with the promise of increased likelihood of finding an ideal mate. It becomes intriguing whether consumers would prefer human matchmakers from the platform over algorithm recommendations. Drawing upon the sense of agency literature, people’s perceived controls and subsequent behaviors are influenced by whether the context is agency-supportive or agency-controlling, with the former emphasizing opportunities for free choice and the latter making people believe that their actions are imposed by others. This study assumed that among dating service users, human matchmakers were perceived as agency-controlling and algorithm recommenders as agency-supportive. We posit that consumers have an implicit theory about the nature of love, such that some consider their fate of love to be preordained and outside of individual influence (algorithm prediction), whereas others believe that their love fate is malleable and can be changed by one’s efforts (human intervention). Thus, we suggest that algorithm dating recommendations (human matchmaker) would make consumers perceive higher (lower) sense of agency, which positively relates to service adoption intention. Such relationship is stronger when people hold a belief of fixed love fate. The research findings would contribute insights and practical implications to online relationship marketing and human-algorithm interaction literature.

3 - Please Give High Priority to My Data:The Effect of Self-data Priority Belief on Perceived Algorithm Recommendation Quality
Yi Li, Macquarie University, Sydney, Australia, Ivan Ho

Although the development and adoption of artificial intelligence, machine learning, and algorithms for offering consumer advice is on the rise, research has shown that consumers demonstrate resistance to algorithm-based advice despite its objectively superior quality. To overcome this resistance to algorithm-based advice, this research documents a lay belief that consumers have about how use of their data by algorithms can improve recommendation quality. Specifically, consumers believe that their data are more important and relevant than other user data for algorithms to generate high-quality personalized recommendations. Based on this lay belief, simply framing algorithms as giving high priority to individual user data can improve perceived recommendation quality and adoption intentions. Several managerially relevant boundary conditions for this lay belief have been identified; the effect holds when consumers 1) perceive their needs to be unique (vs. not unique), 2) have an accuracy consumption goal (vs. exploratory consumption goal), and 3) believe that their data are of high quality (vs. low quality). The findings across six studies using both hypothetical and real behavioral consequences provide a simple yet effective framing strategy to enhance consumer perceptions of algorithm-based advice quality and adoption intentions.

4 - Educating on the Spot: The Use of Chatbot in Driving Consumer Motivation to Support Sustainable Causes
Ivan Ho, Macquarie University, Sydney, Australia
Despite the rapid advancement of AI, chatbots are mostly used to automate repetitive tasks, such as delivering simple customer services, answering frequently-asked-questions, and collecting customer complaints. As a result, the implementation of chatbot is often associated with consumers’ perception of companies’ effort to cut cost. Until recently, chatbots are gaining attention as a new type of educational tool, which can be used to advise and support users, ultimately delivering an interactive learning experience. However, consumer research has devoted very limited attention to the implementation of educational chatbots for marketing purposes. This research suggests that educational chatbots are valuable assets for enriching consumers’ knowledge and understanding of companies’ products and initiatives, and subsequently motivating consumptions, particularly for conservation organizations. The present research defines educational chatbots as text-based conversational AI-based agents capable of providing information regarding the company, its products, and/or initiatives. It is argued that educational chatbots can be used by conservation organizations to engage in interactive exchange of information with consumers . This research argues that the design of the educational chatbots (menu-based vs. knowledge-based) will influence consumers’ motivation and behavior to support the companies’ environmental initiatives, and expects that knowledge-based chatbots will be more effective in encouraging donation, compared to menu-based chatbots. This effect can be explained by a sequential process in which consumers’ perception of expertise influences uncertainty reduction. Findings will also provide practical insights to conservation organizations in educating and empowering customers with knowledge at the point-of-purchase, which motivate them to support the company’s initiatives.
TD22
Meeting Room C4.11
Applied Theory: New Considerations for Products
Invited Session
Special Sessions
Chair: Z. Eddie Ning, Sauder School of Business, University of British Columbia, Vancouver, BC, Canada
1 - Retail Bundling of Competing Manufacturers' Products: Should Manufacturers Worry?
Samir Mamadehussene, University of Texas-Dallas, Richardson, TX, United States, Nanda Kumar, Vithala Rao
While retailers often bundle products of competing manufacturers, the literature in which the intermediary, in a decentralized channel, makes the bundling decision is scant, and no existing work allows the retailer to decide both whether to bundle and what to bundle. We consider a setting in which if the retailer bundles, a decision must be made on whether the bundle comprises products from the same manufacturer (self-bundle) or from different manufacturers (cross-bundle). When the retailer is able to bundle products from different manufacturers, it can direct consumers who are loyal in only one category to buy the product that provides the retailer with the highest margin in the other category. Therefore, the practice of cross-bundling may intensify price competition between manufacturers. We find that when consumers’ brand-preference across categories are positively correlated, cross-bundling serves as a price discrimination device that allows the retailer to target low-valuation consumers (who have no brand-preference) with minimum revenue loss from high-valuation (loyal) consumers. We find conditions under which the possibility of the retailer to cross-bundle enhances manufacturers' profits. Moreover, there is a region of parameters in which the ability of the retailer to cross-bundle provides a win-win-win situation for all parties: manufacturers, retailer, and consumers.
2 - Communicating Attribute Importance under Competition
Jungju Yu, KAIST, Seoul, Korea, Republic of, Jae-Yun Lee, Jiwoong Shin
When consumers shop for unfamiliar products, they may be uncertain about which product attributes are more important than others in determining their utility. This uncertainty can make it difficult for consumers to make comparisons between products and thus discourage consumers from engaging with the product category. We study firms' strategic communication of information about product attribute importance under competition. Consumers are aware of product attributes and attribute levels but are uncertain about which attribute is the more important. Two firms independently receives a noisy signal about the true state and sends a cheap-talk message to a buyer, highlighting one of the two product attributes. After observing the messages from the two firms, the buyer decides whether to further engage with the category by visiting the stores at a positive cost. We show that there exists a truthful equilibrium in which both firms honestly announce the noisy signal they have received. In this equilibrium, the collective message can credibly convey information about the more important attribute, especially when their messages coincide. Having resolved the uncertainty, the buyer is encouraged to visit the stores and buy a preferred product. Interestingly, one of the firms can be better off sending a truthful message even if the message reveals its competitive weakness. Moreover, we show that if only one firm could communicate with consumers, the truthful equilibrium does not exist. Thus, the presence of the competing firm enables credible communication of the more important attribute and makes both firms better off.
3 - Reputation for Privacy
Jesse Yao, Chinese University of Hong Kong, Shanghai, China, People's Republic of
As consumers become increasingly concerned about their privacy, firms can benefit from committing not to sell consumer data. However, the holdup problem prevents them from doing so in a static setting. This paper studies whether the reputation consideration of the firm can serve as a commitment device in a long-run game when consumers have imperfect monitoring technology. A patient enough monopoly can commit because its reputation will be permanently destroyed if consumers observe the data sale. The persistent punishment provides the monopoly a strong incentive not to deviate. In contrast, reputation may fail to serve as a commitment device when there are multiple firms. The penalty for selling data is smaller when consumers cannot know exactly which firm sold the data. Also, other firms can hurt the reputation of a particular firm even if that firm does not sell data. The authors find some sufficient conditions under which the incentive to deviate is so strong that firms lose the ability to commit. Reputation failure in the presence of multiple firms persists with endogenous or asymmetric monitoring.
4 - Signaling through Commitment to Product Equity
Z. Eddie Ning, Sauder School of Business, University of British Columbia, Vancouver, British Columbia, Canada, Kinshuk Jerath
Many companies commit to creating inclusive and equitable products as part of their diversity, equity, and inclusion (DEI) agenda. Such design considerations aim to provide the same level of product experience to minority and majority consumers. Emerging research suggests that DEI initiatives have a measurable impact on business outcomes, but the economic mechanisms that influence consumer demand remain under-explored. We characterize conditions under which profit-seeking firms can use public commitment to product equity to signal their competence. Equitable product design can increase demand even from consumers who do not directly benefit from the added product features. We analyze how the economic power of minority consumers affects strategy, profit, and consumer welfare and show that a policy intervention that mandates equity can backfire. We test our theoretical predictions using lab experiments by examining how consumers’ beliefs are influenced by a firm’s product equity and inclusion statements.
Friday, June 28, 8:30 AM - 10:00 AM
FA01
Meeting Room C2.5
Pricing: Inflation and Perception
Contributed Session
Chair: Youngeun Lee
1 - How Soaring Inflation Impacts Consumers’ Spending Patterns
Ada Choi, University of New South Wales, Sydney, New South Wales, Australia, Harald J. van Heerde, Marnik Dekimpe, Jake An

Inflation is back. After several decades in which consumer prices rose hardly at all in most developed countries, inflation has recently rebounded to levels that are severely affecting the lives of households – directly impacting their income and subsequent budget allocation decisions across a broad set of spending categories. While extant literature has studied how macroeconomic shocks (such as cycle fluctuations), and macro-inflation (as reflected in a rising consumer price index) affect consumer decisions, the impact of micro-inflation remains unexplored. Micro-inflation, defined as the rate of increase in the prices of goods and services within a household’s own consumption basket over a period of time, poses an essential yet unexamined factor in understanding how consumers adapt their individual-level consumption expenditures to cope with escalating inflation. To address this critical gap in knowledge, this study leverages a massive dataset consisting of the full daily banking transaction records of approximately 350,000 Australians, spanning seven years from 2016 to 2023. Adopting an Empirics-First approach, this study will employ a seemingly unrelated regression model to gain insights into the dynamic effect of micro-inflation on individual-level consumer spending decisions across diverse categories, ranging from transport and housing to education. The findings of this study carry significant implications for both marketers and policymakers, allowing them to better understand consumer responses to micro-inflation and develop targeted strategies to support consumers in managing their household expenditures. This research will not only contribute to academic discourse but also provide practical guidance for navigating the complexities of an inflationary environment.

2 - Heterogeneity in Internal Referencing Price Formation: The Role of Ego Involvement
Drs L.I. Son MBA, Leiden University, Leiden, The Netherlands, Netherlands, Prof.dr. J. I. van der Rest

This study explores the role of Ego Involvement in price information in the formation of Internal Reference Prices (IRP). It seeks to develop an Ordered Alternative Questionnaire (OAQ) drawing on Social Judgement Theory, which specifies Ego Involvement as the importance or centrality of a particular issue to a person's life, in this study the importance of price information in consumer buying decision-making. The exploratory design was informed by a constructivist grounded theory approach. Data were collected from 28 in-depth in-home interviews transversing supermarket-shopping occasions. Based on advanced axial categorization coding the study reveals variance by three interdepended vintage points: price information, store-selection and time/effort exerted. Differences were found in these orientations, indicating differences in stances on price information Ego Involvement and thus IRP-formation. Interestingly, although the use of IRPs is not in the public domain, participants felt that there is a norm that should be adhered to. Specifically, those with less extreme stances felt remorse, guilt or even shame for not acting accordingly. The findings contribute to theory by deepening our insight into consumer heterogeneity, in particular the number and type of price cues included and the width of resultant IRPs, and the role of organic cues which pertain to the inner psychophysical processes that affect consumer decision making. Important managerial and marketing implications can also be drawn, as ignoring consumer heterogeneity in IRP formation can undermine pricing and promotional strategies with significant risks for reducing long-term profitability.

Keywords: Internal Reference Price, Heterogeneity, Social Judgement Theory, Ego Involvement, Price Information.

3 - Are Consumers Price Trapped in a Spiral of Concentration in Grocery Markets?
Lisa Asher, University of Sydney, Sydney, Australia, Catherine Sutton-Brady, Satheesh Seenivasan

The degree of market concentration within western economies of grocery retailing has become particularly important, with the cost-of-living crisis and proposed “greedflation” (Dekimpe & Van Heerde, 2023). In Australia and New Zealand, duopolies dominate the landscape. The implications of concentration has the attention of government, as it is impacting consumers, producers, manufacturer competition, private label, and prices. We propose the unlinked accelerators of power are private label and the retailer’s ability to control markets at a category level. Through linking industrial organization theory, marketing empirical first studies, industry evidence, and our own empirical work, we have developed a theory called “retailer spiral concentration theory”. Within its propositions we argue there are three degrees of market concentration occurring that are interconnected as a spiral. 1. Competing retailers, 2. The retailer playing the dual role of customer and competitor with private label, 3. The retailer reducing manufacturers in categories to make room for their private label, increasing their profits and control. This is a spiral of consolidating competition in favour of the retailer with private label, and large national brands who have some degree of power, of which this activity is unregulated. The ongoing concentration of these three areas results in a layered consumer price trap of disproportionately high prices. We propose testable propositions to understand this interconnected concentration, with proposed solutions. One major solution is the need for government intervention within competition law to address the issues of concentration at these lower market levels.

4 - Shrinkflation: Evidence on Product Downsizing and Consumer Response
Youngeun Lee, Boston College, Chestnut Hill, MA, United States

Product downsizing, also referred to as shrinkflation, is a strategy used by firms in response to cost increases: firms indirectly increase unit prices by reducing package sizes while keeping item price constant. This paper documents the extent of product downsizing and estimates the impact on price and quantity sold to measure the consumer response. To provide systematic evidence of product downsizing, I leverage a comprehensive dataset of products across a wide range industries.The sample includes reduced products from across 56 product categories and 295 product sub-categories. The size of reduction is non-trivial, where the median product is reduced by 11 percent of its package size. I show that package size reduction leads to an increase in unit price by 9 percent at the median, whereas item prices remained consistent. I find minimal quantity response to package size reductions, and analyze whether quantity reaction varies across product types and unit pricing regulations.

FA02
Meeting Room C2.6
Game Theory for Competitor Strategy 1
Contributed Session
Chair: Arjang Nikbakht, University of Alberta, 11135 83 avenue North west, Edmonton, AB, T6G 2c6, Canada
1 - How Does Oligopolies Competition affect Exploration vs. Exploitation?
Yaping Li, Tianjin University, Tianjin, China, People's Republic of, Yunchuan Liu, Jianxiong Zhang
We study how does oligopoly competition for consumer attention affect tradeoff of recommendation algorithms between exploration and exploitation.
We design a game-theoretical model where firms recommend content to multihoming consumers who decide how to allocate attention among firms content.
Different from previous research that limit competition to duopoly assumption, cool feature of our model is that we relax competition setting to more than two firms.
We show that firms' incentive to explore (innovation) is non-monotonic with oligopolies number.
2 - Color Substitution, the Lipstick Effect, and Optimal Advertising
Anand Krishnamoorthy, University of Central Florida, Orlando, FL, United States, Norris Bruce, Ashutosh Prasad

This paper examines analytically a scenario where consumer preferences for different colors in their consumed goods are influenced by the state of the economy and it proposes how firms can schedule their advertising to exploit such effects. To determine the optimal advertising expenditures for such scenarios, we propose a differential-game model of the sales-advertising relationship, which features sales and advertising at the brand level, color substitution, and the effect of brand strength. As in many product-adoption models, advertising influences non-adopters to adopt the product with some probability, thereby capturing competitive effects between the brands and colors. Each firm advertises to consumers at the brand level and sets pricing to retailers at the color level. Furthermore, the retailer selects the retail prices, which also affect the sales. The analysis yields explicit solutions for the optimal advertising, value function, and comparative statics.

3 - Why Would Firms Add Excess Information to Advertising Messages?
David Soberman, University of Toronto, Toronto, ON, Canada, Yi Xiang
The idea we explore is designing advertising messages to make information incrementally more difficult to process. People are assumed to have a limited ability to process information. When a firm includes extra elements of information in a message, people are less likely to retain any specific piece of information. This may mean that important things are forgotten. The benefit of adding excess information when consumers have complete information is the creation of segment(s) that are uninformed about that firm. This creates a captive segment for the competitor leading to an increase in price. Because of the complementarity of prices, the focal firm benefits from the competitor's decision to increase price. The cost of adding excess information is a loss of demand for the firm that adds information. The optimal strategies are a function of the strength of the two effects noted above. If consumers only consider a product when fully informed (the base case), there is an incentive to add excess information but the price increase needed to support this strategy is high. When consumers employ deduction to determine the expected location of a firm, the incentive to add excess information can be amplified or reduced depending on whether the firms are differentiated or not. In sum, the study demonstrates that firms have an incentive to move the market away from perfect information.
4 - Equilibrium Refinement for Asymmetrical Stackelberg Leadership Duopolies
Arjang Nikbakht, University of Alberta, Edmonton, AB, Canada, Paul Messinger, Bora Kolfal

Often in Stackelberg duopolies both firms prefer the same role (e.g., leadership). The current paper proposes an equilibrium refinement for role selection in such cases. The refinement shows how applying a performance criterion assures a form of perceptual congruence between the equilibrium outcomes and the assumed player roles: a very dominant (weak) firm cannot assume to be the consistent follower (leader). This paper gives a general tool for identifying and ruling out implausible leadership structures for the purpose of model design. Also, ruling out implausible equilibria makes the equilibrium predictions more managerially relevant. Lastly, applying this approach leads to game predictions that can be tested empirically.

Our approach is to identify all cases where at least one role selection is feasible using the performance criteria. In some games and parameter configurations, only one firm can consistently assume the preferred role (suggesting who should be the leader); in others, both firms can consistently assume the preferred role (suggesting indeterminate leadership); in yet others, no firm can consistently assume the preferred role (suggesting Stackelberg game structures are not applicable). In one interesting subcase, a weaker firm can only exist assuming the preferred role (offering an explanation for entry deterrence). We demonstrate the application of our refinement for four commonly used models: two where both firms prefer leadership (Cournot and vertical channel competition) and two where both firms prefer followership (market base demand and consumer utility).

FA03
Cockle Bay Room 1
Sharing Economy 1
Contributed Session
Chair: Thomas Schreiner
1 - Exploring the Sharing Paradox in Collaborative Consumption: Conceptualizing and Measuring Paradoxical Sharing Behaviors
Kimmy Chan, Hong Kong Baptist University, Hong Kong, Hong Kong, Lawrence Sze Lok Choi, Howard Ling

Despite the growing trend of collaborative consumption in the “sharing economy”, there is a notable research gap in how peer service providers manage the conflicting roles of being both owners and sharers simultaneously. This study builds upon psychological ownership and paradox literature to introduce the concept of “paradoxical sharing behaviors” (PSB), which describes peer service providers’ competing, interrelated behaviors undertaken to meet demands arising from their different roles, simultaneously and over time. Through in-depth interviews and five main studies (n= 2,237 participants, spanning across the UK, Europe, US, Hong Kong, and China in sharing contexts of ride, accommodation, and luxury sharing), we develop a scale for measuring PSB, according to dimensions of control, self-identity, intimacy, and territoriality. Our research affirms the psychometric properties of the scale, including convergent, discriminant, nomological, known-group, and predictive validity. The findings also reveal that service providers with a holistic mindset and strong economic and social motives tend to exhibit PSB. These behaviors can potentially enhance service quality and promote extra-role behaviors in both the short- and long-term. This research enriches the existing body of work on paradoxes by expanding its application to the sharing economy, providing novel insights for researchers and practitioners to understand barriers that prevent service providers from sharing their properties and offering a tool to measure service providers’ ability to handle such a paradox.

2 - CANCELLED What Influences Behavioral Loyalty on Airbnb? Analyzing the Factors Affecting Repurchase of Listings
Xiaoting Xie, University of Science and Technology of China, Hefei, China, People's Republic of, Xiabing Zheng, Xiaobei Shen

Airbnb has been a prominent player in the sharing economy, revolutionizing the way people book accommodations and disrupting the traditional hotel and rental industry. Nowadays, attracting loyal consumers to rebook a specific listing on the platform becomes crucial in maintaining Airbnb’s sustained growth. This study explores the factors (i.e., various features of listings) that may affect repurchase of listings on Airbnb. We collected consumer-facing information about Airbnb listings and hosts directly from Airbnb websites, focusing on seven representative European cities. The ridge regression model was adopted to address potential multicollinearity issues and analyze the factors influencing repurchase behavior of specific listings on Airbnb, as it is particularly suited for situations where there is a high degree of correlation among independent variables in regression analysis. The results show that host score has a significant positive effect on repurchase, while listing score is negatively associated with it. When considering the moderating effect of price, higher listing score are more likely to encourage consumers to repurchase among high-priced listings. In contrast, the findings show a negative correlation between listing score and repurchase among low-priced listings. Additionally, the findings suggest these joint effects of price and listing score only exists when listing score is higher than host score. Our results offer guidance for hosts on Airbnb to develop effective marketing strategies to foster consumer loyalty and encourage repurchase.

3 - Customer Responses to Platform Exploitation Attempts by Service Providers in the Sharing Economy
Prithwiraj Mukherjee, Ahmedabad University, Ahmedabad, India

Platform exploitation, a phenomenon where service providers "cancel" a service request and provide it to the customer off the books for cash, to avoid revenue sharing with the platform, is a common problem in the sharing economy. Such "canceled" service requests not only represent lost revenue for platforms, but could also potentially jeopardize customer safety, trust and satisfaction. Such exploitation is especially rampant, and often the norm, in short-term gigs like taxi rides, one-off home repairs, and similar contexts, in emerging economies.

Initial studies of this phenomenon (e.g. Zhang et al, 2021) focus on the impact of agent characteristics and agent-customer relationship duration on the propensity to exploit the platform. This study, on the other hand, focuses on gigs (e.g. taxi hailing) where long-term relationships are not possible. We focus on the factors that influence customers' compliance to platform exploitation proposals by service provider agents.

In this ongoing study, we propose a series of experiments to study the effects of coercion and justification (of platform exploitation) by service providers, along with the moderating effects of scarcity, on the likelihood of customers to accept platform exploitation proposals. Initial results suggest that unlike buyer-supplier contexts (e.g. Hausman and Johnston, 2010), coercion by the service provider actually reduces the likelihood of customer acceptance (compliance) of platform exploitation. We further observe that a scarcity of providers does not have a moderating effect on the influence of coercion on compliance.

4 - How to Communicate Physically Shared Offerings in the Sharing Economy
Thomas Schreiner, UNSW, Sydney, Sydney, Australia, Franziska Volckner, Harald van Heerde, Valentyna Melnyk, Tobias Hinze, Magdalena Bekk

A key element of the sharing economy is that resources, such as shared accommodation on Airbnb, can be physically shared. Despite the importance of marketing communication for sharing platforms, it is unclear whether and how providers should communicate physically shared offerings differently from non-physically shared offerings. To address this gap, this research takes a psychological ownership perspective and investigates how provider communication and pricing drive demand and how these effects are moderated by whether an offering is physically or non-physically shared. Monthly field data of 130,595 Airbnb offerings across six large U.S. cities and close to five years show that for physically (versus non-physically) shared offerings, provider availability and the amount of provided information are more efficient in increasing demand, while provider self-investment is less efficient. Moreover, physically (versus non-physically) shared offerings are subject to a higher price sensitivity of potential customers. A text analysis of Airbnb consumer reviews, based on first-person possessive determiners and pronouns, shows that physically (versus non-physically) shared offerings convey a lower sense of psychological ownership. This finding explains why giving a sense of psychological ownership through marketing communication is especially efficient for physically shared offerings. The authors discuss implications for providers and platforms that provide access to physically shared or non-physically shared or both types of offerings.

FA04
Cockle Bay Room 2
Generative AI/Content Creation in Marketing
Invited Session
Special Sessions
Chair: Hortense Fong, Columbia Business School, New York, NY, United States
1 - Predicting Predictability: Quantifying Movie Predictability from Video Trailers using Generative AI
Hortense Fong, Columbia Business School, New York, NY, United States, George Gui

When does a video trailer give away too much information? Video trailers are designed to pique the interest of viewers but also preserve the suspense of movies (and shows). A trailer that gives away too much information cannibalizes movie content, which could decrease interest in watching the movie. This paper proposes a strategy that utilizes Generative AI (GenAI) to quantify how predictable a movie is based on its trailer. We demonstrate that this strategy not only gathers valuable insights but also provides an explanation for viewer interest in watching movies.

2 - Using Multimodal LLM to Extract and Discover Features from Ad Images
Shawndra Hill, Microsoft, New York, NY, United States, Jeremy Yang, Qianyun Zhang, Saurabh Verma, Audrey Burgess, Ethan Meng

Advertisers’ interest in understanding what creative elements predict ad performance is as old as the history of advertising. However, extracting features from ads has either been a manual process given their multimodal nature (e.g., text, image, video) or narrowly focused on a small number of predefined features when aided by supervised algorithms. Here we explore the possibilities of using a general purpose Multimodal Large Language Model (LLaVA) to extract any features that can be queried with natural language from image ads. With a random sample of about 2 million ad images on Facebook and Instagram, we extract features corresponding to the content (e.g., cognitive, affective, experiential) and execution (e.g., comparative, endorsement, entertainment, imagery, mnemonic) of the ad images; we also ask the model to provide open-ended descriptions of the ad images. We show four sets of results. First, we validate the accuracy of the model with a small sample of images labeled by human coders. Second, we describe the differences in ad features by advertiser and campaign attributes. Third, we show the correlation between ad features and performance. Fourth, we explore what novel performant features can be discovered from the open-ended descriptions. We also discuss how these results can be used to enhance ad ranking, provide creative guidelines for advertisers, and guide the creation of generative ads.

3 - Product Design Using Generative Adversarial Network: An Application in Artistic Template Design
Hui Li, The University of Hong Kong, Hong Kong, Hong Kong, Jian Ni, Fangzhu Yang

The development of Generative AI enables large-scale automation of product design. However, this automated process may struggle to incorporate diverse consumer preference information from a company's internal dataset. To address this, we propose an efficient model that integrates consumer preferences into the automated product design process on a large scale. In particular, our model can leverage rich consumer preference information in user-generated content (UGC) from external websites, which helps overcome the "cold-start" problem faced by small or start-up companies with limited internal data. We utilize Continuous conditional Generative Adversarial Networks (CcGAN), combining a prediction model with a Deep Convolutional generative adversarial network (DCGAN). We first train a prediction model using consumer choice data and use the predicted preferences to guide the generation of new product designs. We train and evaluate the model in an unmanned photo gallery setting where consumers select templates for self-taken photos. The model incorporates templates and consumer preferences from both the company's internal dataset and external UGC. We compare three methods: (1) DCGAN without consumer preference, (2) CcGAN with consumer preference in the company's internal data, and (3) CcGAN with consumer preference in both internal and external data. Our empirical results demonstrate that incorporating consumer preference, particularly external preference, generates more appealing template designs compared to simple GAN models.

4 - Using Generative AI for More Generalized Consumer Research: Stimuli Sampling through Large Language Models
Liu Liu, University of Colorado Boulder, Boulder, CO, United States, Nicholas Reinholtz

Recently, large language models (LLMs) like GPT4 have gained significant attention due to their mastery of language generation as well as their ability to solve tasks from a wide variety of domains such as coding, psychology, and medicine, showing sparks of artificial general intelligence (AGI). Researchers are now exploring ways to adopt LLMs in market research, for example replacing traditional human surveys with answers from ChatGPT. In this article, we propose a new emergent use of LLMs: generating experiment stimulus in consumer research. Specifically, we examine the potential of the latest LLMs to mitigate issues related to stimulus under-sampling and experimenter bias, thereby improving the generalizability and validity of consumer research findings. We begin by discussing the roles of stimuli in consumer research, the process of stimuli creation in current research methodologies, and identifying key challenges and opportunities. We then illustrate how the latest cohort of LLMs is well suited to overcome these challenges and help improve the efficiency, precision, objectivity, and generalizability of consumer research. We demonstrate the process, feasibility, and performance of the proposed application of LLM in the research process via a few studies. Additionally, we offer some thoughts on practical considerations, including issues related to open science and potential biases introduced by LLMs.

5 - AI in Disguise: AI-generated Ads Outperform Human-made Ads if They Don't Look Like AI
Oded Netzer, Columbia University, New York, NY, United States, Yannick Exner, Jochen Hartmann, Shunyuan Zhang

2023 was the year of generative AI. Its advance sparked the imagination about the role of generative AI in creative industries in general and advertising in particular, and raised the question of whether it can meaningfully disrupt the advertising industry. Answering this question is nearly impossible using individual A/B tests as these only provide a snapshot into specific application contexts, which are subject to the researchers' discretion. We partner with a leading online ad platform, which released a generative AI-powered ad maker in 2023, to obtain a dataset covering more than 2 million daily ad-level observations across various industries. This dataset includes both human-made and AI-generated ads in a quasi-experimental setting.

In our quasi-experimental setting of ads that were created by the same advertiser at the same time, and as part of the same campaign but vary in terms of the usage of AI, we find that the click-through rates (CTRs) are higher if advertisers employ AI to generate advertising images, but not necessarily for the ad caption. Interestingly, this superiority of AI images holds only if the AI-generated images do not look like AI. We find that AI images are more likely to include a human face, but the presence of a human face tends to disguise the ad as being generated by AI. Our findings provide evidence that advertisers can benefit from leveraging mass-market generative AI tools to boost their online ad performance.

FA05
Meeting Room C3.1
Marketing Strategy: Innovation
Contributed Session
Chair: Yongchuan Bao, University of Alabama in Huntsville, , Huntsville, AL, United States
1 - Enhancing Open Innovation Success: Exploring the Role of Marketing Capability on Enhancing Success with Inbound Innovation
Sanjay Sisodiya, University of Idaho, Moscow, ID, United States

Open innovation allows firms to leverage external and internal knowledge to improve innovation outcomes (Chesbrough 2003). External knowledge coming into a firm is often termed inbound, while knowledge (or inputs) from within the firm going external are often referred to as outbound innovation. Previous studies have considered the role of inbound and outbound open innovation, and to the extent to which performance is improved (Schroll and Mild 2011; Michelino, Caputo, Cammarano, and Lamberti 2014). Considerable research has been performed on both inbound and outbound perspectives (Enkel, Gassman, and Chesbrough 2009), yet there is still opportunity to delve deeper into the capabilities that firms must have or develop to be successful with open innovation.

Gassman, Bell, and Hogenkamp (2011) highlight the importance of having capabilities in place to yield improved outcomes. One such opportunity is with the R&D and marketing capabilities of the firm. While R&D capability is critical to advancing innovation within an organization, the marketing capability may be critical to providing a complementary capability needed to enhance open innovation outcomes.

This study further explores the role of inbound open innovation using a combination of established measurement scales in a survey along with secondary data. We test our hypotheses using a combined sample of over 200 responses. After controlling for firm size and evaluating a range of performance measures, we find that there are promising results to support the importance of not only pursuing open innovation, but also ensuring there are established capabilities within the firm.

2 - The Effect of Big Data Comprehensive Pilot Zones on Corporate Innovation Efficiency
Zhixuan Yang, Wuhan University in China, Wuhan, China, People's Republic of, Fue Zeng

The rapid advancement of the Fourth Industrial Revolution has led to the emergence of new technologies such as artificial intelligence and big data, bringing opportunities and challenges to corporate innovation. To follow the trend of big data, a series of policies have emerged. However, the impact of such policies on corporate innovation lacks in-depth exploration. Therefore, we use empirical data from 2010 to 2022 and use the establishment of big data comprehensive pilot zones as a quasi-natural experiment. Based on the difference-in-difference model, we study the impact and the mechanism of the BDCPZ on corporate innovation efficiency. We found that the BDCPZ can significantly improve the innovation efficiency of listed companies, and this conclusion still holds after a series of robustness tests. The mechanisms include increasing supply chain diversity and improving supply chain efficiency. Specifically, the BDCPZ can augment the diversity of both upstream suppliers and downstream customers for enterprises, providing new growth opportunities for multi-stakeholder holistic innovation integration. Moreover, owing to its large volume and high velocity, the BDCPZ can significantly enhance the speed of information sharing and collaborative interactions along the supply chain, thereby positively affecting innovation efficiency. Additionally, the BDCPZ can play a stronger stimulating role when companies face relatively low industry competition pressure and financing constraint pressure, or when they exhibit a moderate strategic proactiveness. Our findings can assist managers in leveraging big data policies to stimulate corporate innovation vitality.

Keywords: Big data comprehensive pilot zones;innovation efficiency;supply chain diversity;supply chain efficiency;difference-in-difference

Authors: Zeng Fue,Yang Zhixuan (speaker)

3 - CEOs with Throughput Career Experience: The Double-edged Sword Effect on Firm Innovation
Wenting Zhu, Southwestern University of Finance and Economics, Chengdu, China, People's Republic of, Feifei Liu

Although the literature has well documented that a firm’s innovation is negatively associated with its CEO’s throughput career experience, few studies examine its positive side for firm innovation. We address this gap by adopting an efficiency logic and examining how and when CEO’s throughput career experience deters and facilitate firm innovation. Based on a sample of listed firms in China from 2008 to 2017, our results show that a firm whose CEO has throughput career experience spends less on R&D but is more efficient in turning innovation inputs to outputs. In addition, for state-owned firms (v.s non state-owned firms), the negative impact of CEOs’ throughput career experience on innovation input becomes less significant while its positive effect on innovation efficiency is significant only for state-owned firms. These findings reveal a double-edge sword effect of CEOs’ throughput career experience on firm innovation and help a firm with a focus on innovation effectively appoint CEOs based on their career experience.

4 - Collaborative Problemistic Search Capability and Digital Innovation: The Moderating Effects of IT Capability and Legal Development
Yongchuan Bao, University of Alabama in Huntsville, Huntsville, AL, United States, Hua Liu, Shaobo Wei

Representing a new mode of collaboration in the digital economy, collaborative problemistic search capability (CPS) refers to a firm capability to collaborate with downstream (customers) and upstream partners (suppliers) on digital infrastructure. Drawing on the organizational search theory and the institutional theory, this study unpacks CPS into CPS with customers (CPS-C) and CPS with suppliers (CPS-S) and examines the effects of CPS-C and CPS-S on digital innovation and their interactions with IT capability and legal development in the context of China. Specifically, we argue that both CPS-C and CPS-S exert a positive effect on digital innovation, and IT capability demonstrates a complementarity effect by enhancing the effect of CPS-C and a substitution effect by reducing the effect of CPS-S, whereas legal development magnifies both the complementarity effect and the substitution effect of IT capability. A survey among multiple managers of a sample of firms in China supports the hypotheses. The research results make important contributions to marketing literature on digital innovation and advance existing research stream on collaboration with customers and suppliers towards innovation.

FA06
Meeting Room C3.2
Modeling Consumer Shopping Behavior
Contributed Session
Chair: Yinxing Li, Tohoku University, Sendai, Japan
1 - CANCELLED Beyond Home Delivery: An Empirical Examination of the Impact of Self-Pickup Stations on Online Shopping
Chenxi Liao, Chinese University of Hong Kong, Hong Kong, Hong Kong, Jingcun Cao, Ying Xie
Click-and-collect services through self-pickup stations have gained popularity as an alternative to traditional home delivery in e-commerce. There are, however, concerns, such as the cannibalization of home delivery, reduced order sizes, and competition with offline shopping. Additionally, understanding the diverse responses of experienced and new customers to self-pickup stations is essential, particularly in the post-COVID-19 era. We analyze a rich individual-level transaction dataset from a leading online shopping platform in Asia. We find that the number of nearby self-pickup stations has a positive association with individual customers' shopping frequency and overall weekly spending on the online platform, despite lower self-pickup free-shipping thresholds. The number of such stations also has a positive spillover effect on home-delivery sales, suggesting the presence of a complementary rather than competitive effect between the two delivery channels. Lastly, the effects of self-pickup stations vary based on whether customers joined before or after COVID-19 and their online shopping experience: greater availability of nearby self-pickup stations leads to increased spending and shopping frequency among experienced customers but a decrease in both among newer customers. This research provides crucial managerial implications for the practitioners. Self-pickup stations can complement the home-delivery channel, suggesting that e-commerce platforms should consider offering both options. New customers with limited experience in online shopping may however be drawn back to the offline shopping format when frequenting self-pickup stations, and platforms must therefore cultivate and solidify the online shopping habits of newly acquired customers.
2 - Effect of First-visit to a Merchant in an E-marketplace
Sadat Reza, Nanyang Technological University, Singapore, Singapore, Xueli Zhang
In this study we examine the impact on consumers’ probability of purchasing from a merchant when they are attracted to visit that merchant first upon arriving at an e-marketplace. Once the consumers enter the platform, the consumers need to decide which merchant to visit. At this point, any merchant can conduct activities, such as banner advertising on the landing page, that can potentially entice the consumers to choose that merchant as the one to visit in the first instance. The question that the merchant needs to examine is under what conditions it is reasonable for it to engage in the above-mentioned activity. Engaging in such an activity is reasonable only if the choice of first-visit merchant increases the probability of purchase from that merchant. Our paper focuses on how to estimate this effect using observational data. Since the consumers self-select their choices of first-visit merchant, the estimand is not generally identifiable from observed propensities to visit the merchants. We discuss the identification challenges and propose to generate the required counterfactual estimates using two approaches: a bounds estimation method and a non-linear difference-in-differences method. We apply the estimation procedures on a data collected from an online cashback platform. Our results indicate that the effects of first-visit to a merchant in the cashback platform vary with merchant characteristics. The results also indicate that the larger merchants stand to gain more from enticing the visitors to the platform to visit their portals first. The results have useful managerial implications.
3 - Receipt Embedding and Shopping Mission Segmentation
Yinxing Li, Tohoku University, Sendai, Miyagi, Japan, Nobuhiko Terui

A battery of studies has applied the Word2Vec model to marketing problem by using large-scale shopping data, such as Prod2Vec, Item2Vec, and Meta-Prod2Vec. They show that the framework of Word2Vec outperforms existing models in the prediction of sales. However, these existing approaches lack the interpretability of the model since the Word2Vec framework cannot evaluate the effect of variables, which may limit its use in the marketing, such as the effective personalization and targeting. This study proposes a machine learning model by extending these works in several directions: (i) embedding receipt that is characterized as multiple purchase in a shopping trip. This derives implications by vector representation of receipt, that is, (ii) identifying shopping mission of customers by clustering receipt vectors, (iii)identifying price and seasonal effects. Our model contains the seasonal and promotion terms in the form of state space prior. We proposed a novel approach in this study by involving the concept of receipt vector into the Item2Vec framework, as well as the prior structure which represents the dynamic preference shift of the receipt by incorporating state space prior combined with the likelihood constituted by product2vec model. Our study highlighted the importance of the marketing environment when forecast the market basket for the future trips. The results of empirical results help managers understand the purchasing patterns and preference shift for a certain customer in different marketing environment.

FA07
Meeting Room C3.3
Marketing Implications of Responsible AI
Invited Session
Special Sessions
Chair: Kalinda Ukanwa, University of Southern California, Los Angeles, CA, United States
1 - Reliably Detecting Increased Intercommunity Support After Hate Acts
Breagin Riley, University of Notre Dame, Notre Dame, IN, United States, A. Chyei Vinluan
Despite a decrease in violence, overall, the US has experienced an increase in violence motivated by bias against race, color, religion, national origin, sexual orientation, gender, gender identity, or disability, or simply, hate acts. Ample evidence demonstrates negative economic consequences for the geographical area where the incidents occur, but more nascent work identifies positive economic consequences for the community of people to which the target individuals belong. For example, when the police violence against Black people in America was salient, people from outside of Black communities increased support for Black owned restaurants (c.f., Sharma, Frake, and Watson, under review). We demonstrate how to responsibly use data to detect positive intercommunity economic behaviors in response to hate acts committed against phenotypically Asian people in the US. We first use secondary data linking anti-Asian hate acts by date and county to corresponding news coverage; longitudinal grocery purchases by household, date, and county; and racial and ethnic demographics by county. We show that intercommunal purchases of Asian signaling groceries increase when anti-Asian hate acts become salient to non-Asians. Then, we discuss how to responsibly use various kinds of data to train AI to detect intercommunal economic support after hate acts.
2 - Why Firms Should Want Algorithmic Accountability
Bill Rand, North Carolina State University, Raleigh, NC, United States, Kalinda Ukanwa, Peter Pal Zubcsek
Because of growing concerns about responsible use of artificial intelligence, European and US regulators recently introduced legislation to protect consumers from algorithmic bias. These policies hold firms accountable for the fairness of their algorithms while relying on consumers to report when unfairness occurs. Our research reveals unintended consequences of these policies resulting from differences between how firms and consumers assess fairness. Current algorithmic fairness standards use firm data and measures of statistical parity to determine if demographic groups are being treated similarly. However, the average consumer does not have access to firm data. Consumers are left to assess fairness by gathering information from their social networks about the firm's service treatment. We model how consumer assessments can trigger the spread of unfairness beliefs, even when the firm's algorithm is fair. We show that a lack of algorithmic accountability may lead consumers to paradoxically believe that a firm with an unfair algorithm is less biased than a firm with a fair algorithm. We also demonstrate how accountability via a third-party watchdog institution could reconcile these different perceptions of fairness and provide both firms and consumers with better insights into algorithmic bias.
3 - CANCELLED You have the Right to Remain Private: The Effect of Online Privacy Legislation on Purchase Intentions
Andre Martin, University of Notre Dame, Notre Dame, IN, United States, Katrijn Gielens, Tarun Kushwaha
Despite the increasing relevance of online privacy laws among consumers and legislators, it is not clear what their impact is on consumers’ decision-making and, in turn, brand outcomes. We add to the nascent privacy policy literature by investigating (1) how online privacy laws affect consumers’ stated purchase intentions of sufficient brands and (2) how online display advertising (ODA) moderates the impact of online privacy laws on purchase intentions. Using a quasi-experimental setting where one US Census region, the West, received the treatment of the first online privacy law (California Consumer Privacy Act - CCPA), we gather regional, daily stated purchase intentions of 1,985 brands from 2012 to 2021. Results from a difference-in-difference model show that although, on average, consumer purchase intentions decrease more in the West after the introduction of CCPA, virtual brands (brands with a primarily online presence) receive a boost in purchase intentions. However, human service brands (service brands that require brand-employee interaction) receive a boost in purchase intentions when they increase their ODA. The opposite is true for virtual brands and goods brands (brands that primarily offer physical goods). We discuss the managerial and policy implications for brand engagement under the umbrella of online privacy laws.
4 - How Word-of-Mouth Affects Consumer Response to Algorithmic Bias
Kalinda Ukanwa, University of Southern California, Los Angeles, CA, United States, Roland Rust
News about biased algorithms has made people realize that decisions made by artificial intelligence systems can be influenced by demographic details like race, ethnicity, and gender. Often, people hear about these decisions—like getting a loan, insurance, or school admission—through friends or family who share similar backgrounds. For instance, if women hear that other women are being denied credit cards by a company, they might assume that gender plays a role in these decisions and decide not to apply to that company, even if they don’t know why the others were denied. This study explores how these shared stories about acceptance or rejection by services can shape people’s choices, especially when it comes to services that screen consumers. We find that this specific type of word-of-mouth, which we call service acceptance word-of-mouth, can guide historically marginalized consumers towards companies that use fair algorithms and non-marginalized consumers towards companies using unfair algorithms. Interestingly, we also find that when people from different backgrounds share and value each other’s experiences, they might all choose the same firm—even if the firm's algorithm is not in everyone’s best interest. We demonstrate that this shared choice can arise from people trying to maximize their own benefits, even without any explicit concern for social justice, fairness, or morality.
FA08
Meeting Room C3.4
Gender, Markets, and Society
Invited Session
Special Sessions
Chair: Xinlong Li, Nanyang Technological University, Singapore, Singapore
1 - Are Female-Led Trials Safer?
Manuel Hermosilla, University of Illinois at Chicago, Baltimore, Maryland, United States
Prior literature suggests that women can exercise more cautious decision making when the interests of others are at stake. Motivated by this idea, we investigate whether the gender of the principal investigator (PI) may affect the safety of patients in clinical trials. We present causal evidence linking female PIs with 4.3% fewer patients affected by serious adverse events compared to those led by male PIs. This safety advantage is concentrated in experimental groups exposed to active comparators rather than placebo or experimental interventions, indicating more cautious trial design. Female PIs are also linked with a higher propensity to terminate trials early due to safety concerns, reflecting more vigilant trial management. These findings suggest that policies that aim to alleviate the marked underrepresentation of women in clinical trial leadership can also contribute to improved patient safety.
2 - What Explains the Growing Gender Education Gap? The Effects of Parental Background, the Labor Market and the Marriage Market on College Attainment
Michael Keane, Johns Hopkins University, Sydney, Australia, Zvi Eckstein, Osnat Lifshitz
In the 1960 cohort, American men and women graduated from college at the same
rate, and this was true for Whites, Blacks and Hispanics. But in more recent cohorts, women graduate at much higher rates than men. To understand the emerging gender education gap, we estimate a model of individual and family decision-making where education, labor supply, marriage and fertility are all endogenous. Assuming preferences that are common across ethnic groups and fixed over cohorts, our model explains differences in all endogenous variables by gender/ethnicity for the ‘60-‘80 cohorts based on three exogenous factors: family background, labor market and marriage market constraints. Changes in parental background are a key factor driving the growing gender education gap: Women with college educated mothers get greater utility from college, and are much more likely to graduate themselves. The marriage market also contributes: Women’s chance of getting marriage offers at older ages has increased, enabling them to defer marriage. The labor market is the largest factor: Improvement in women’s labor market return to college in recent cohorts accounts for 50% of the increase in their graduation rate. But the labor market returns to college are still greater for men. Women go to college more than men because their overall return is greater, after factoring in marriage market returns and their greater utility from college attendance. We predict the recent large increases in women’s graduation rates will cause their children’s graduation rates to increase further. But growth in the aggregate graduation rate will slow substantially.

3 - Goodbye My Friends and Goodbye My Career: Evidence from the Movie Industry
Xinlong Li, Nanyang Technological University, Singapore, Singapore, Andrew Ching
By analyzing a dataset with more than 180,000 actors, directors and producers in the motion picture industry over a span of 20 years, this research provides causal evidence that network connections can significantly influence an actor's career development. Specifically, we focus on quantifying the impact of losing a connection with a producer or director due to their death on an actor's career trajectory. Our identification strategy leverages the insight that the timing of one's death is exogenous to the network evolution, and hence it creates an exogenous shock to their network. Following the passing of a producer or director, our study reveals a decline of 6.76% in acting opportunities for actors who had previously collaborated with them. Our findings highlight that actors with greater experience are less affected, and female actresses are more affected by the loss of connections due to these death shocks. In particular, the detrimental effect to female is more than 2.5 times of that to male.

We also find evidence that for every additional year that passes since the connection loss event, a treated actor experiences a 0.7% more acting opportunities decline. This finding provides empirical support for the potential presence of the Matthew Effect, where the already connected individuals tend to become even more connected and successful, while those who experience connection loss become increasingly disconnected and less successful.
FA09
Meeting Room C3.5
Branding and Logos
Contributed Session
Chair: Jingpeng Li, No.111, North Section of Second Ring Road, Chengdu, Sichuan, China, 610000
1 - An Empirical Analysis of Bundling Strategy
Wei Fan, The Ohio State University, Columbus, OH, United States, Tianqi Xue
Bundling is a popular market strategy where multiple products are sold together as a package to consumers. Retailers use this approach by offering discounted rates on product bundles to encourage sales. Although the theoretical understanding of bundling is rich, there is a lack of empirical insights into its impact on both sellers and consumers, especially for bundling physical goods. We aim to fill this gap by conducting an empirical analysis using a unique dataset from a chain optical store. We examine the following question: (i) How does bundling affects consumers' welfare compared to free bundling? (ii) What is the impact of bundling strategies by different brands on consumers' benefits? (iii) How do various bundling strategies affect sales and costs? The research shows that bundling complementary products provides benefits to consumers by reducing search costs and increasing the utility of each bundled item. The sellers also benefit from this strategy as it helps them to concentrate on the distribution of consumers, which may impact prices and costs. This empirical analysis provides practical insights into the existing theoretical literature on bundling strategies in the retail sector.

2 - Spillover Effects in Search and Advertising for Umbrella-Branded Products
Zhengwei Zhang, The University of Texas at Austin, Austin, TX, United States

This study examines brand spillover effects within the automotive industry, focusing on the role of umbrella branding in shaping consumer perceptions. Leveraging Google Trends data, we investigate how search behaviors for different vehicle models under the same brand are influenced by brand spillovers. Utilizing dynamic factor models (DFM), we estimate parent brand and sub-brand factors for major manufacturers like Toyota and GM between January 2013 and July 2023. Our findings reveal significant brand spillovers, such as positive forward spillovers from the parent brand to the Toyota RAV4 and reciprocal spillovers within GM models. Incorporating advertising data as covariates demonstrates the impact of advertising on online brand searches. By highlighting the dynamics of brand information search, this research contributes to a deeper understanding of brand spillovers at the upper outcomes of the purchase funnel.
Keywords— Google Trends, brand spillovers, advertising spillovers, dynamic factor models

3 - Do Logo Complexity Impact Souvenir Authenticity Perception?When Complex is Authentic
Jingpeng Li, Southwest Jiaotong University, Sichuan, China, People's Republic of

People crave authenticity in tourism. However, empirical research on logo design elements’ effect on souvenirs’ perceived authenticity has been largely ignored in the existing literature. We fill this gap by focusing on logo complexity and examining its impact on souvenir authenticity. Across four studies (N= 936), we demonstrate that complex logos lead to higher perceived authenticity, with the explanatory mechanism behind this being perceived oldness. The consumption type moderates this effect. We find that the effect of logo complexity is mitigated for utilitarian consumption. Our findings contribute to the literature on perceived authenticity and provide practical advice for local souvenir marketers when designing brand logos.

FA10
Meeting Room C3.6
Social Media
Contributed Session
Chair: Xin Jiang
1 - Complement or Substitute? Examining the Relationship between Social Media and TV Channel Consumption of Live Sports Events
Yupin Yang, Simon Fraser University, Burnaby, BC, Canada, Qiang Lu, Rohan Miller

This paper empirically examines the substitute and complement patterns between social media content and TV consumption on live sports events. Particularly, we study the relationship between the social media (Twitter) content and TV ratings of live broadcasting programs on Supercar Championship racing events in Australia. We first apply a topic model method, TopSBM (topic stochastic block model), to analyze the social media content to find the topics in each time period, and then build a regression model to examine the relationships between TV ratings and various social media topics in each 15-minute period. In the results, we find that the topics with strong emotions (i.e., glory/exciting moments, exciting race incidents) are positively related to TV rating in the following period. These findings suggest that for some unique moments, consumers prefer to consume the sports through TV broadcast as it provides richer and stronger emotions to consumers with high quality images and videos. This finding suggests that social media complements TV broadcast for the “excitement” content. However, informational content such as race process description, social media is a strong substitute to TV broadcast as this topic is more informational than emotional. Last, the car driver chatter content on social media increases fans’ interests in the events and likelihood to watch live broadcast on TV channel. Our study provides insights for firms to utilize the synergies between social media and TV experiences to facilitate fans experiences and build a longer and stronger relationship with fans.

2 - CANCELLED Curate to Engage: Analyzing the Effectiveness of Content Curation Strategies on Digital Platforms
Pankhuri Malhotra, University of Oklahoma, Norman, OK, United States, Mina Ameri
Creating constant content to keep audience engagement is a demanding task for content creators on digital platforms. This paper investigates a key strategy used by content creators – content curation – which involves selecting and sharing existing content alongside creating original material. Our study examines the impact of this strategy on digital platforms, particularly focusing on the micro-blogging platform Twitter. Through an extensive analysis of 10 million tweets from 800 creators, we explore how content curation impacts creator effectiveness. Utilizing log-linear models with robust fixed effects, our research uncovers a nuanced, non-linear relationship between content curation and creator effectiveness. This relationship is significantly moderated by the size of the creator's network. Our findings reveal that larger creators experience diminishing returns from frequent reposting, whereas smaller creators benefit from consistent reposting in terms of engagement. Another notable aspect of our study is the role of content diversity. The contrast between the curated content and a creator's original posts significantly impacts how audiences perceive and interact with the curated content. Additionally, we employ a Double Machine Learning model to assess the impact of curation on engagement, controlling for high-dimensional image and text features and a rich set of fixed effects. Through this research, we address a relatively ignored aspect of content creation, specifically content curation, which is widely used by creators to maintain audience engagement.
3 - CANCELLED How Can Influencers' Healthy Lifestyle Video-log Motivate Public Health Behavior? The Effect of Voice-visual Matching
Xin Jiang, China University of Mining and Technology, Xuzhou, Jiangsu Province, China, People's Republic of

Abstract: How social media influencers promote people's healthy lifestyle is attracting increasing attention. The current paper focuses on how influencers’ healthy lifestyle video-log can motivate people to engage in health behavior. Based on 3324 video logs of influencers' daily healthy lifestyle from a short video platform in China, and through machine learning and econometric methods, the structured data, and unstructured data such as voice features, visual features, and comments of videos were acquired and analyzed, and then a psychological experiment was combined to test the mechanism. This study found that when the users browse the video logs of influencer participating in health behaviors, the matching performance of voice and visuals in videos can improve the users' intention to participate in health behaviors. Specifically, in an influencer’s video log, high voice arousal matching high visual variation, and low voice arousal matching with low visual variation can make the users generate higher influencer identification and thus higher health behavior intention. This paper can provide marketing implications for how to effectively motivate people to follow influencers' lifestyles.

Keywords: social media influencer; health behavior; identity mechanism; voice arousal; visual variation

FA11
Meeting Room C4.1
Theory and Practice in Marketing
Invited Session
Special Sessions
Chair: Doug J. Chung, University of Texas at Austin, Austin, TX, United States
1 - How Does Firms' Social ESG Rating Influence Brand Equity
Lingling Zhang, China Europe International Business School, Shanghai, China, People's Republic of, Yajin Wang, Hongyu Shan, Ziqing He

ESG-related issues have gained increasing importance worldwide. Corporates have invested significantly in ESG strategies, and their ESG performance is closely monitored by various third-party rating agencies such as Reprisk, Refinitiv, and Sustainalytics. However, there are important gaps in the current ESG-rating literature and practice. The ESG performance is largely measured at the firm level, instead of the brand level, so that current EGS ratings lack the granularity to guide actionable marketing strategy for brand managers. Furthermore, current ESG rating systems primarily represent data from public news outlets instead of social media, the latter of which is where consumers learn and discuss brands’ ESG performance nowadays. To fill these gaps, we propose the innovative social ESG rating, which measures how consumers react to firms’ ESG performance on social media. We collaborate with the largest data platform in China, which provides real-time coverage to almost all the major social media sites worldwide. Our ESG rating sheds light on the volume and sentiment to the online ESG-topic mentioning for each brand. Our initial results show that a brand’s social ESG rating is positively correlated with its online brand equity but provides unique orthogonal information. This research also answers the question of whether ESG ratings have an impact on consumers’ awareness, relevance, and affinity to a brand. The findings of the project generate important implications for marketers on how to effectively utilize social media and better engage with consumers on ESG issues.

2 - Between the Originality and Popularity: How to Create Successful Digital Content
Minjee Sun, University of Iowa, Iowa City, IA, United States, Kang-Pyo Lee, Matthew Osborne

As digital content consumption becomes more routine, a range of content creators, from novices to professionals, are increasingly engaged in creating digital content. Despite the proliferation of digital content, few pieces of work achieve success and financial rewards. For a content creator to achieve a successful outcome, the content should follow a popular trend to capture the attention of potential customers, and also for algorithms to expose it to customer searches. At the same time, the content must have some originality: in other words, it must have some distinct features that ensure it can maintain customer attention in the long term.

This research explores how content creators balance popularity and originality in their content creation. We utilize natural language processing (NLP) and text mining to identify popular and trending topics, then employ graph theory to construct network graphs for these topics, assessing the novelty and originality of content aligned with specific trends. Essential outcome measures, such as view counts and subscriptions, guide the determination of the optimal blend of popularity and originality.

Our preliminary findings reveal that the emphasis on novelty varies significantly depending on the content's topic. Consumers exhibit preferences for stereotypical content in specific topics, lean toward balanced content in others, and seek highly distinctive content within each topic. Next, we plan to extend our findings to diverse digital content to ensure generalizability. Our research can assist content creators in developing more engaging and effective content, and support digital platform managers in identifying and promoting promising content creators.

3 - Time Dependence and Preference: Implications for Compensation Structure
Doug J. Chung, University of Texas at Austin, Austin, TX, United States, Byungyeon Kim, Byoung Park

This study jointly examines agents’ time dependence—period effects within instantaneous utility—and time preference—the behavior on discounting future utility. The study considers the start- and end-of-period effects for time dependence and exponential and hyperbolic discounting for time preference. It provides formal identification arguments and sufficient conditions for both time constructs, including those regarding the duration of time dependence. The empirical application uses granular individual data and variations in the compensation structure to separately identify the two time constructs. Taking agents’ time assessment into account, counterfactual studies examine changes in their behavior, and, thus, the sales outcome, in response to alternative compensation structures. The results demonstrate a trade-off between long- and short-quota cycles; the nonlinearity in sales performance with respect to the duration of the compensation cycle; and how quota-bonus plans and commissions can be used to motivate different types of agents.

FA12
Meeting Room C4.2
Customer Journey 1
Contributed Session
Chair: Jing Niu, University of Liverpool, Jouy en Josas, France
1 - Product Category Classification Based on the Consistency of Information
Geunho Baek, Keio University, Seoul, Korea, Republic of

In light of the growing prevalence of omnichannel strategies among marketers, academic research has intensified in exploring the concept's development and assessing the effectiveness of each omnichannel action. Numerous studies have highlighted critical factors contributing to the realization of omnichannel, with particular emphasis on the consistency of information across channels owned by a firm. This consistency, which ensures uniform brand and product information across diverse consumer touchpoints, yields benefits such as reduced confusion and preserved brand imagery. However, realizing these benefits hinges on the information sources consumers utilize, with optimal outcomes achieved when consumers rely on sources within the firm's control. Despite the significance of understanding consumer information sources, research in this area remains limited.

To address this gap, this study examines data from over 2000 Japanese consumers across 14 product categories, investigating their usage of information sources. By defining consumer-centric information consistency as consumer behavior characterized by reliance solely on firm-controlled information sources (i.e., firm's official online website, sales agency), the study categorizes product types based on the proportion of high-level consistency with consumers relative to total consumers. Furthermore, the investigation delves into the relationship between consumer-centric information consistency and various consumer attributes and purchasing behaviors, elucidating the underlying reasons and outcomes of consumers' high-level consistency usage. The findings suggest tailored information provision strategies based on product and consumer characteristics, enabling practitioners to enhance their omnichannel strategies.

2 - A Cross-Categorical Customer Journey Segments on Serendipity
Rei Tada, Yokohama National University, Yokohama , Japan, Sotaro Katsumata

Numerous studies pointed out the importance of understanding customer journey when firms develop a marketing strategy. This trend has also seen in marketing practice. Specifically, business magazines and consulting firms publish company rankings on customer experience. The literature on customer experience has recently focused on serendipity in consumer decision-making. Serendipity refers to an individual’s unexpected and fortunate discovery through chance events. Prior studies examined the role of serendipity in specific retail environments such as e-commerce and subscription services. We further need to examine the conditions under which serendipity is likely to occur in consumer decision making from many more different perspectives. It would be significant to identify the relationship between serendipity and customer touchpoints. The purpose of this study is to divide consumer decision process into several journey segments by customer touchpoints and to show the impact of these segments on customer perceived serendipity and satisfaction. We conducted some questionnaire surveys and examined customer touchpoints and decision-making outcomes across more than one product category (e.g., food and books). The results of latent class analysis allowed us to categorize journey segments and revealed the relationship between these segments, serendipity, and customer satisfaction. We also discussed the potential applications of the results to marketing practice.

3 - In Search of Balanced Metrics: Understanding the Role of Consumer Mindset Metrics in Online Interactions
Kate Gunby, University of Sydney, Balgowlah, Australia, Rahul Govind, Vincent Mitchell

The rise of both digital advertising and eCommerce platforms has provided marketers with a direct way to track consumers' online advertising exposure and response through website visits and sales. While these new data have enabled marketers to see an immediate response to advertising, results are commonly interpreted in isolation without understanding the existing predisposition towards a brand. This is important as this predisposition is often the result of brand-building activity. Without understanding this relationship, we risk overemphasising the impact of digital advertising activities in driving a consumer response.

Using an innovative approach to data collection with passive metering technology capturing data at the individual level, this research demonstrates how consumer brand attitudes influence interactions with brands online for 27 brands across three categories. Hurdle model analysis enables us to measure the effect of brand attitudes on the probability of a brand website visit and the frequency of such visits. Results show that 85% of all visits come from consumers with at least one positive brand attitude (a top 2 rating for either liking or familiarity), and those with the most positive mindsets are up to 24 times more likely to visit a brand website. Furthermore, the optimal manner of representing attitudes is not via the traditional measures from a hierarchy of effect models (awareness, consideration and preference) nor, as prior research has suggested, purely using cognitive measure, but rather via separate measures of familiarity and liking.

4 - A Meta-Analysis of Financial and Non-Financial Returns on Engagement Initiatives
Jing Niu, University of Liverpool, Jouy en Josas, France

Engagement initiatives offered by companies, aimed at enhancing firm-customer interactions and interactions among customers themselves, have ongoing momentum in marketing. While not primarily focused on sales, engagement initiatives have been shown to be beneficial for companies in various studies. However, the magnitudes and relative comparisons of both financial and non-financial returns on Engagement Initiatives remain inconclusive. To address this, this meta-analysis synthesizes evidence over 580 effect sizes from more than 70 empirical studies in the last two decades. On average, both financial and non-financial return on engagement initiatives are positive, with non-financial return more pronounced. Furthermore, a multivariate meta-regression reveals that firm characteristics (industry type and firm size), customer characteristics (customer participation intensity and purchase stage), and the operationalization of engagement initiatives play a relevant role in determining the magnitudes of the financial and non-financial return. This study contributes theoretical understanding of engagement initiatives and provides a basis for developing hypotheses on the variability in returns on Engagement Initiatives across different business contexts.

Keywords: Engagement initiative, customer engagement; customer lifetime value, customer relationship management, meta-analysis

FA13
Meeting Room C4.3
Online Display/Banner Advertising
Contributed Session
Chair: Andrew Prendergast, Ex. Google, Nielsen // NetRatings, BBDO, PO Box 198, South Yarra, 3141, Australia
1 - Sequential Scroll and Search Decision
Jae Hyen Chung, The Chinese University of Hong Kong, Sha Tin, Hong Kong

I build upon Weitzman’s sequential search model and incorporate the scroll decision by consumer in the model. In the proposed model, a consumer faces two dynamic programming problems---scrolling down the product list and searching among products that are scrolled through. Besides, the proposed model allows the consumer to update his belief about the distribution of the observable product features along the scroll process.

By allowing consumers to search among only a subset of products, the proposed model relaxes the assumption typically made in the consumer search literature that, prior to the search process, the consumer is aware of the observable features of all products presented. In addition, by incorporating the scroll cost in the model, the proposed model precisely measures the impact of the product position within the list on consumer's scroll and search behavior.

I first demonstrate the parameter recovery by the proposed model and estimation strategy with the numerical experiments. In addition, by combining the estimation strategy with EM algorithm, I demonstrate that model parameters are identified even when the scroll decision and/or the search order are not observed in the data. In the empirical application of the proposed model to the dataset from Expedia, I quantify two opposing impacts of the price discount on the scroll and search behavior.

2 - Leveraging Generative AI in Digital Advertising: What to Generate?
Remi Daviet, University of Wisconsin-Madison, Madison, WI, United States, Yohei Nishimura

Generative artificial intelligence (AI) is a groundbreaking technology that has the potential to disrupt and revolutionize the digital advertising industry. However, a wide range of uncertainties persist regarding the integration of generative AI into traditional advertising processes, including finding effective implementation, training methodologies, and anticipated performance gains. Moreover, the large space of variations that can be generated makes it challenging to identify content that is both performing well and compatible with the brand's standards. This paper addresses these critical concerns by proposing a novel creative design process driven by combining the generative AI with two deep Bayesian prediction models. The first model aims at identifying high-performance content, while the second assesses acceptability with respect to brand standards. To minimize training costs, both prediction models undergo training through repeated testing of informative batches of creatives, in an active-learning fashion. We demonstrate the effectiveness of our approach with a field application to scene setting in ads for an outdoor activity company. By providing a framework guiding the integration of generative AI in digital advertising, this paper seeks to bridge the gap between theoretical potential and effective practical applications.

3 - Font Dynamics:Font Weight and Its Role in Crafting Digital Content Appeal
Yanyan Zhu, University of Science and Technology of China, Hefei, China, People's Republic of, Yongjun Li, Baohuan Zhou
With the rapid rise of the digital media era, consumers are increasingly relying on images and text to make product choices on digital platforms. In recent years, the significance of font design has attracted widespread attention from scholars in the field of marketing. This study aims to delve into how the design features of book cover titles' font on digital reading platforms affect the visual appeal of the product and to identify whether this effect is influenced by other factors. Research shows that assuming similar readability, there is a significant positive U-shaped relationship between cover font weight and click-through rate. Specifically, books with thicker or thinner cover fonts tend to have higher click-through rates, while click-through rates drop significantly when the font weight is at a medium level. Furthermore, we extensively explored changes in this effect under different visual processing scenarios. Findings indicate that positioning the title at the top of the image strengthens the relationship between font thickness and click-through rates, with a more pronounced effect observed in female-oriented books. By delving into the design characteristics of fonts on digital reading platforms, this study provides important insights into the field of digital marketing, specifically web elements, online advertising, and web design. At the same time, it expands the understanding of sensory marketing in digital environments and provides practical guidance for enterprises to choose suitable font design, especially brand identity.
4 - CANCELLED - Predicting the Performance of Digital Advertising
Andrew Prendergast, Ex. Google, Nielsen // NetRatings, BBDO, South Yarra, Australia

A first principles exploration of ethically sound, privacy-preserving simulation, prediction and evaluation of campaign optimization in a digital advertising setting, including publication and description of a number of anonymised paid advertising datasets from search and display campaigns across a multitude of clients.

We analysed the practical application of performance marketing by a digital media buying team in a large advertising agency and explored challenges faced by the business school graduate level campaign analysts in predicting performance of digital advertising transacted in Vickery, silent bid and private deal settings, and explored the utility of truthful and non-truthful bidding WRT risk preferences. Our study focuses on micro-conversion based ROI optimization of direct-response search and display activity, but found that the techniques developed are also applicable to “above the line” branding focused digital campaigns. We then rigorously executed several multi-million dollar search campaigns using the developed techniques and validated the Vickery hypothesis that accurate assessment of placement valuations and truthful bidding maximises long run expected utility and campaign optimization stability.

The techniques presented include a practical approach to placement valuation & bidding which uses a simple Bayesian prior that can be calculated in Excel. We compare it’s predictive performance to more exotic models using a “poor-mans-simulation” ML model evaluation technique and find the results are competitive. The evaluation technique is presented and we demonstrate its apriori simulation of future campaign performance from past ad-server data collected aposteriori. A selection of datasets to aid in replication and improvement of our experimental results are also provided.

FA14
Meeting Room C4.4
Influencer Marketing: AI and Machine Learning Approaches
Contributed Session
Chair: Simone Wies, Goethe University Frankfurt
1 - Thank You for Sponsoring:The Impact of Disclosure on Social Media Marketing
Wenyi Huang, University of Michigan, Ann Arbor, Michigan, United States, Yiwei Li, Justin Huang
The Federal Trade Commission (FTC) 2019 guidelines require that influencers clearly disclose any material connections that they have with sponsors. However, the guidelines leave room for influencers to choose how they would disclose this information. In practice, we observe two broad classes of disclosure strategy. The first is active disclosure, which is characterized by the influencer’s use of personalized language (e.g., “Thank you to brand for sponsoring”) to acknowledge and thank the sponsor. The other is passive disclosure, characterized by the usage of platform tools (e.g., hashtags or badges) to disclose the sponsor relationship. While extant research has primarily investigated consumers’ negative responses to the latter, this research identifies both strategic options and quantifies their impact on both social media and brand engagement. We find that for a small influencer (around 1,000 followers), passive disclosure significantly increases user engagement by 5%, and for a typical large influencer (around 100,000 followers), active and passive disclosures boosts brand engagement by 5.3% and 6.6% respectively. Moderation analyses provide mechanistic evidence that this occurs because of disclosures’ roles in indicating sponsored relationship between influencers and brand. Our findings highlight that both influencers and sponsors can derive benefits from disclosing advertising, with encouraging implications for fostering consumer advertising awareness and policy implementation.
2 - Measuring Influencer Empathy with Multimodal Deep Graph Learning Model
Simone Wies, Goethe University Frankfurt, Frankfurt, Germany, Xueming Luo
Influencer marketing has been established as one of today’s most impactful marketing strategies, and a burgeoning body of literature has started to investigate relevant criteria to select the most effective influencers. We focus on a core criterion of influencer effectiveness, influencer empathy, which is tied to the essence of influencer marketing—the fostering of deep cognitive and emotional connections with their audience. Our research offers important insights into influencer empathy by exploring two fundamental inquiries: (1) Measuring influencer empathy with multimodal deep graph learning methods and (2) documenting the sales effectiveness of influencer empathy. Our model is based on text, audio, and image data and their interactions, which we believe offers the most comprehensive account of empathetic expressions. From a managerial standpoint, our research offers actionable insights for marketers in influencer marketing. The methodology developed to measure empathy provides a practical tool for brands to assess and select influencers. Brands can easily leverage this tool to identify influencers whose empathetic behaviors align with their marketing objectives, ensuring a more effective influencer-brand fit.
FA15
Meeting Room C4.5
Leveraging Machine Learning for Causal Inference in Marketing
Invited Session
Special Sessions
Chair: Pantelis Loupos, University of California, Davis, Davis, CA, United States
1 - Estimating Effects of Long-Term Treatments
Shan Huang, University of Hong Kong, Seattle, WA, United States
Estimating the effects of long-term treatments in A/B testing presents a significant challenge. Such treatments --- including updates to product functions, user interface designs, and recommendation algorithms --- are intended to remain in the system for a long period after their launches. On the other hand, given the constraints of conducting long-term experiments, practitioners often rely on short-term experimental results to make product launch decisions. It remains an open question how to accurately estimate the effects of long-term treatments using short-term experimental data. To address this question, we introduce a longitudinal surrogate framework. We show that, under standard assumptions, the effects of long-term treatments can be decomposed into a series of functions, which depend on the user attributes, the short-term intermediate metrics, and the treatment assignments. We describe the identification assumptions, the estimation strategies, and the inference technique under this framework. Empirically, we show that our approach outperforms existing solutions by leveraging two real-world experiments, each involving millions of users on WeChat, one of the world's largest social networking platforms.
2 - DoubleMLDeep: Estimation of Causal Effects with Multimodal Data
Sven Klaassen, University of Hamburg, Hamburg, Germany, Jan Niclas Teichert-Kluge, Philipp Bach, Victor Chernozhukov, Martin Spindler, Suhas Vijaykumar
This paper explores the use of unstructured, multimodal data, namely text and images, in causal inference and treatment effect estimation. We propose a neural network architecture that is adapted to the double machine learning (DML) framework, specifically the partially linear model. An additional contribution of our paper is a new method to generate a semi-synthetic dataset which can be used to evaluate the performance of causal effect estimation in the presence of text and images as confounders. The proposed methods and architectures are evaluated on the semi-synthetic dataset and compared to standard approaches, highlighting the potential benefit of using text and images directly in causal studies. Our findings have implications for researchers and practitioners in economics, marketing, finance, medicine and data science in general who are interested in estimating causal quantities using non-traditional data.
3 - Unconfoundedness with Network Interference
Pantelis Loupos, University of California, Davis, Davis, CA, United States
We study non-parametric estimation of treatment and spillover effects using observational network data. Peer effects in selection into treatment are challenging to accommodate in a nonparametric setting because they can induce high-dimensional network confounding, where the total network and covariates of all units constitute sources of selection bias. Most of the literature instead assumes confounding can be summarized by a known, low-dimensional function of these objects. We show that graph neural networks are well suited to adjust for high-dimensional network confounding. We establish a network analog of approximate sparsity, demonstrating that models with complex interference in potential outcomes and selection into treatment can have low-dimensional structure. We apply our methodology to Venmo and explore how the use of emojis in Venmo transactions influences user engagement.
FA16
Meeting Room C4.6
Health Care Marketing
Contributed Session
Chair: Nah Lee, SKK GSB, Seoul, Korea, Republic of
1 - Is the Price Really High? Investigating the Influence of Price Unfairness Perception on Patient Engagement
Wenjing Zhang, Harbin Institute of Technology, Harbin, China, People's Republic of, Mengdi Wang, Dong Li
Price is the cost patients pay for access to physicians’ services. In turn, the physicians should make the patients feel that their sacrifice is fair. However, patients often perceive the market-driven pricing practices of online medical consultation(OMC) as unfair, especially upon learning that they have paid more than in-person consultation or have not received the expected diagnosis and suggestions. Indeed, encouraging patients to continuously engage in the OMC service is challenging if managers fail to avoid price differences or mitigate such unfair perceptions. Despite the demonstrated importance of price perceptions for outcomes such as (re)purchase behavior and patient engagement, surprisingly little is known about how real price and perceived price influence patient engagement in the OMC context. We crawled and analyzed a data set with over 140000 reviews based on 7136 physicians in 12 departments from a leading OMC platform(https://wy.guahao.com) in China, followed by the use of text analysis and econometrics to test our hypotheses. The results indicated that the relationship between OMC price and patient engagement is U-shaped for graphic consultation and inverted U-shaped for voice consultation, while price unfairness, physician status, and disease risk affected patient engagement and moderated the above relationships. These findings enrich the OMC literature and provide meaningful insights into the pricing strategy of OMC platforms.

Keywords: Online medical consultation; Price unfairness; Patient engagement
2 - The Impact of Generic Substitutability Laws on Physician Payments and Generic Adoption
Rahul Narnindi, SMU, Singapore, Singapore, Singapore, Ernst Osinga, Sandeep Chandukala

The US has witnessed an estimated $2.2 trillion in savings over the past decade due to the substitution of expensive prescription drugs with their cheaper generic alternatives. In this research, we investigate the effect of various US state laws governing the substitution of branded drugs with cheaper generic substitutes (henceforth referred to as “generic substitutability laws”) on physician payments and generic adoption. These laws, that vary across states, establish a framework for pharmacists to substitute branded drugs with FDA approved generics. By combining two publicly available datasets – 1) physician payments made by pharmaceutical firms from the year 2013-2022 released by CMS, and 2) the annually released list of all approved drugs and their therapeutic equivalents (known as “Orange Book”), we investigate whether generic substitutability laws influence the degree to which pharmaceutical firms adjust physician payments after the release of a generic competitor. We expect pharmaceutical firms to more drastically reduce payments in states where generic substitution is encouraged. We explore several moderating factors to this relationship. In addition, using a publicly available database, Medicare part D prescriber data, we test whether heterogeneity in generic substitutability laws influences the adoption of generics. We aim to highlight physical payments as an important lever for managers to affect the primary demand for their drug. We discuss the implications of our findings for policymakers and manufacturers.

3 - The Impact of Inter-Hospital and Intra-Hospital Telemedicine Models on the Strategies of Coopetitive Hospitals
Yufei Ge, Tianjin University, Tianjin, China, People's Republic of, Yanfei Lan

As information and data technologies continue to advance, an increasing number of patients can access telemedicine services from anywhere with the internet. However, due to the complexity of diseases, telemedicine still needs to be integrated with in-person visits. This integration makes it necessary for information such as patient data to be transferred between telemedicine organizations and offline healthcare facilities. The organization and the hospital can be the same institution, i.e., the intra-hospital telemedicine mode, or can involve cross-hospital telemedicine, i.e., the inter-hospital telemedicine mode. We construct a stylized model to investigate the impact of the introduction of these two telemedicine modes on the competition and cooperation patterns between the two hospitals. The results of the study reveal that the introduction of a telemedicine mode does not always lead to higher profits for the hospital that adopts it. The introduction of an intra-hospital telemedicine mode can alleviate competition between the two hospitals under certain conditions, thus creating a win-win situation, but this condition is more stringent and challenging to realize. In contrast, the introduction of an inter-hospital telemedicine mode can also result in a win-win situation and, under specific conditions, make both hospitals more profitable than the other mode.

4 - Hospitals' Reactions to Patients' Reviews
Nah Lee, SKK GSB, Seoul, Korea, Republic of, Richard Staelin

This paper examines the extent to which the content of online hospital patient reviews affects healthcare provider's subsequent quality of their delivered healthcare experience. The authors construct an eight-year panel of 2,252 hospitals containing information on the hospital's demographics, yearly quality of care, and all of its patient's Google reviews. Then, using prior research on the types of topics discussed in patient reviews, how these topics impact the patient's overall evaluation of the service, and the classification of these topics into clinical and non-clinical experiences, the authors create two star ratings associated with each review, one for the clinical experience, the other for the non-clinical experience. They aggregate the individual-level review ratings up to the hospital-year level. Then, using General Methods of Moments (GMM) and instrumental variable methodology, they estimate the marginal effect of patient review content on subsequent hospital performance. They find significant segments of their sample of hospitals deliver improved future quality of care after experiencing a negative shock of clinically focused reviews and this effect is greatest for hospitals with good financial health. Conversely, they find instances where a negative shock of non-clinical review content results in a decrease in subsequent healthcare quality.

Keywords: hospital reviews, longitudinal analysis, casual analysis, clinical performance, quality of care, instrumental variable methodology

FA17
Meeting Room C4.7
Sustainability Special Session
Invited Session
Special Sessions
Chair: Dionne Nickerson, Indiana University, Bloomington, IN, United States
1 - Assessing B-Corp Certification's Impact on Firm Innovation: Beyond a Transient Fad, Toward a Sustainable Legacy
Yufan Lin, Cal Poly , Pomona, CA, United States, Mike Nguyen, H. Erkan Ozkaya
This research investigates the relationship between B Corporation certification and firm innovation. B Corps, committed to balancing profit with environmental, social, and governance (ESG) concerns, demonstrates notable growth and performance advantages. However, the impact of their certification on innovation remains uncertain. The rigorous certification process could potentially hinder innovation, while the purpose-driven nature of B Corps might enhance it by promoting stakeholder relationships, transparency, and attracting talent. This study compiles a unique dataset comparing B Corp certified firms with their counterfactuals (i.e., top five closest competitors) in terms of their patent publications and filings, sourced from Owler.com and the Google Patent database. Then, it uses a stacked Difference-in-Differences (DiD) methodology, with endogeneity correction and matching methods, to analyze the short and long-term effects of B Corp certification on patent publications and filings. This approach ensures a comprehensive counterfactual analysis and overcomes the data limitations often associated with private firms. Results indicate a significant, enduring impact of B Corp certification on innovation, evident in both short-term (three years) and long-term (five years) analyses. Even firms that did not renew their certification showed sustained innovation benefits. These findings challenge the view of purpose-driven initiatives as detrimental to innovation, highlighting the strategic advantages of social responsibility certifications in fostering persistent corporate innovation beyond mere financial returns.
2 - Can Selling Lead to Female Empower? Experimental Evidence from India
Rajesh Chandy, London Business School, London, United Kingdom, Iris Steenkamp, Om Narasimhan, Gaurav Mehta
In this paper, we introduce a methodological innovation to overcome some of the major challenges that have plagued the field of psychological (empowerment) testing in employment contexts, especially in less developed countries (Ludeke and Larson 2017; Laalaj et al. 2018). We introduce a unique form of data, namely video interviews, to assess the personality and related psychological empowerment dimensions. To the best of our knowledge, we are the first to introduce video interviews to measure psychological change in the context of emerging markets.
Our preliminary results show that women engaged in sales employment are more empowered than women earning equivalent amounts in the placebo employment condition, and women assigned to a no employment control condition. This paper has direct implications for managers eager to reach rural populations to access the large and relatively untapped pool of customers in rural parts of developing countries (Mahajan 2016). Creating rural sales forces is one of the most effective and commonly applied strategies for reducing the transaction costs of commerce in these rural markets (Mahajan 2016). Notably, such setups provide gainful employment to some of the most deprived members of society, e.g., rural women. In addition, sales employment can empower women and their families, multiplying the benefits. Our research therefore suggests that businesses can achieve the twin goals of “doing good” (superior sales) while “doing well” (resilient and empowered women).
3 - Decomposing the Demand for Sustainable New Products
Youngtak Kim, University of Tennessee, Farragut, TN, United States, Sundar Bharadwaj, Kristopher Keller
Sustainable is no longer a strategic differentiator but a necessity with increased consumer demand for sustainable products serving as key drivers of sales growth. As a result, companies now have greater incentives to create sustainable products that meet the needs of consumers and contribute to environmental and social well-being, to increase firm value. However, the corporate sustainability and social responsibility literature in marketing is limited to firm-level initiatives with fewer examinations at the product level. Consequently, the sources of growth or the role of the marketing mix in influencing growth for sustainable products is not well understood. Therefore, this study aims to investigate how the introduction of sustainable new products drives growth using a demand decomposition approach. Using retail scanner and detailed data on all relevant product characteristics (e.g., various sustainability aspects), the authors decompose the demand for sustainable new products in the CPG industry to determine if they serve as sources of primary (i.e., market expansion) and/or secondary demand growth (i.e., substitution, cannibalization). In addition, the study examines the effects of distribution, promotion, price, and branding in moderating the sources of growth and thus provides a comprehensive view of the growth of sustainable products. By accounting for these factors, this study offers theoretical and practical insights when considering the launch of sustainable new products.
FA18
Meeting Room C4.8
E-Commerce Platforms
Contributed Session
Chair: Sudeep Kumar
1 - The Value of Sharing Data through Data Analytics on Digital Platforms: Empower Small Businesses on Alibaba's Taobao Marketplace
Aishen Li, Tsinghua University, Beijing, China, People's Republic of, Yubo Chen, Xuebin Cui, Banggang Wu, Liu Yang

Advanced data analytics, employing AI and machine learning techniques, enable firms to extract value from data while mitigating privacy concerns. Large digital platforms, renowned for their exceptional data collection and analytical prowess, are ideal conduits for data sharing with sellers through sophisticated data analytics. In collaboration with the world's leading online marketplace, we leverage the initiation of free access to advanced analytics as a quasi-experiment to examine the effects of such data sharing practices on sales performance. Employing a synthetic difference-in-differences approach, our findings reveal: (1) the adoption of advanced analytics resulted in a 30% increase in sales revenue for firms, as an addition to basic descriptive analytics; (2) market-level analytics generate greater value than firm-level analytics, particularly for smaller sellers with limited data, and demonstrating increasing returns over time; (3) the performance disparity narrows as smaller firms experience more substantial growth compared to larger firms; (4) the observed impacts stem not from changes in pricing or advertising budgets but through product innovation; adopted sellers tend to release more new products and are more inclined to expand into new categories with less competition. Our findings contribute to the global debate on data regulation, proposing a novel role for digital platforms in efficiently utilizing non-rival customer data to empower small businesses.

2 - Fresh Price for the Fresh: Investigating Pricing and Demand Response on Grocery E-commerce Platform
Tao Han, The University of Hong Kong, Hong Kong, Hong Kong, Jingcun Cao, Jia Liu

Online grocery retailing is experiencing significant growth, driven by the increasing demand for fresh products with instant delivery. However, compared to durable products, the e-commerce platform faces unique challenges in understanding the dynamic nature of the market. On one hand, the demand for fresh products exhibits high-frequency fluctuations throughout a day due to local consumers’ dietary rhythms. On the other hand, fresh products incur higher costs for storage and preservation, and they can perish quickly, leading to substantial waste. This study investigates consumers’ heterogeneous price elasticity at different time especially when purchasing fresh products. Specifically, we focus on a leading fresh product supermarket, which provides a wide range of fresh products and instant delivery service. We employ double machine learning with instrument variables to address potential issues of high-dimensional confounders and price endogeneity. By identifying the price elasticity at different time windows, we seek to understand how local consumer’ price sensitivity on fresh food evolves throughout a day. We also explore heterogeneity across different categories, such as fruit, vegetables, meat and packaged food. Leveraging the identification results, we demonstrate the effectiveness of an inventory-based dynamic pricing strategy through simulation to maximizing daily profit, by striking a balance between revenue generation and the costs associated with waste.

3 - Experience Infusion: How to Improve Customer Experience with Incidental Activities
Ashish Sinha, University of Qeensland, Brisbane, Australia, Mathew Chylinski, David Lie

The conceptualization of consumers as cognitive misers has profoundly influenced the design of online shopping platforms, leading to streamlined processes aimed at reducing the effort and complexity involved in purchasing decisions. This efficiency-driven paradigm is exemplified by industry giants like Amazon, which optimize their platforms to expedite the purchase funnel. However, the emergence of social eCommerce challenges this model by infusing the online shopping experience with engaging, autotelic activities such as games and social interactions. Despite the growth of social eCommerce, the integration of autotelic activities within instrumental shopping tasks remains underexplored in marketing research. This paper models the dynamic processes underpinning such business models utilizing data from three time series experiments coupled with a Bayesian Dynamic Linear Model, analysed using Hamiltonian Monte Carlo methods, to assess how infusion of autotelic activities affects consumer experience and decision-making accuracy over time. There are significant theoretical implications of our model, as the current models of consumer behavior have largely been developed for instrumental activities. Equally, our findings indicate that social eCommerce platforms can significantly enhance the shopping experience and decision accuracy by embedding casual games and social recommendations into the product evaluation process on a periodic basis. To our knowledge, this is the first academic inquiry into the efficacy of blending autotelic activities within the instrumental consumer journey of social eCommerce. The simplicity and effectiveness of this approach suggest that it may be applicable across a range of industries, providing far-reaching implications for both established digital marketplaces and emerging social eCommerce ventures.

4 - The Impact of Label Adoptions on Sales: Evidence from Amazon's Launch of "Small Business" Badge
Sudeep Kumar, University of Minnesota, Minneapolis, MN, United States, Qi Xie, Linli Xu

Platforms adopt labels (e.g., top-rated, Amazon’s choice) that signal good quality of products, to increase their sales. But recently, many platforms have also adopted labels that do not intend to signal quality but rather highlight the socio-economic status of business. The idea is to promote such businesses by appealing to the social-economic justice sense of consumers. However, such labels can also activate the negative bias of the customers, potentially leading to loss in sales. Thus, it is important to understand how adoption of such labels affects sales and when it’s beneficial for brands to adopt such labels. We aim to answer these questions using weekly data collected from Amazon during the time when “small business” label was introduced. We find that having the label has an overall detrimental effect on sales across a variety of categories. An analysis of heterogeneity of treatment effects implies that consumers may infer low quality for products with such labels, leading to decrease in sales.

FA19
Meeting Room C4.9
Consumer Behavior: Social Influences and WOM 1
Contributed Session
Chair: Yunseob Choi, Yonsei University, Seodaemun-gu, Korea, Republic of
1 - Capturing Positivity: The Influence of Visual Content on Star Ratings in Online Reviews
Yiyi Li, University of Texas at Arlington, Arlington, TX, United States

Many studies in the realm of social media and online reviews have focused on the impact of visual content on audience responses to posts. However, there remains a gap in understanding how the inclusion of visual content with a post affects the review writer’s evaluation of a product. This research seeks to investigate the relationship between the inclusion of visual content and the star ratings assigned by reviewers and to provide insights into the underlying reasons for the observed phenomenon where reviewers who include photos in their reviews tend to assign higher star ratings. Leveraging a large dataset collected from Yelp.com, we demonstrate that restaurant reviews featuring photos tend to receive higher star ratings compared to those without visual content. Through cutting-edge text and image processing techniques, coupled with supplementary lab experiments, we further show that the positive association between photos and star ratings is not only due to heightened reviewer engagement or a desire to assist others. Rather it can also be linked to confirmation bias, a phenomenon where individuals use photos to visually confirm positive aspects of their experiences while downplaying negative elements. These outcomes carry significant implications for marketers, suggesting potential strategies to encourage users to incorporate photos in their reviews. Moreover, this research contributes to a broader understanding of online communication, social influence, and the psychological dynamics inherent in user-generated content.

2 - The Salience of the Core-Periphery Structure and Social Influence in the Diffusion of New Experience Goods in Online User Communities
Soobin Lee, Yonsei University, Seodaemun-gu, SEOUL, Korea, Republic of, Jae Young Lee, Tae-Hyung Pyo

An Online User Community (OUC), a virtual society where people gather to communicate over their shared interests, has evolved into a key player in determining the success of diffusing new experience goods. In OUCs, there are core members with dense connections to other members and peripheral members with loose connections to their neighbors. This research sheds light on how the salience of the core-periphery structure in OUCs (i.e., the extent to which an OUC has a salient distinction between core and periphery) affects the importance of social influence in the diffusion of new experience goods. For empirical analyses, we analyzed in-group diffusion of new songs in OUCs hosted in Last.fm, one of the world’s most influential online communities over music. According to our empirical findings, social influence majorly drives the diffusion of new songs in OUCs with a less salient core-periphery structure. Interestingly, such a relationship between social influence and the core-periphery structure becomes stronger when the focal song has greater brand equity and greater compatibility with the focal OUC’s shared interest. Our findings have several implications for researchers and decision-makers in practice.

3 - Don't Leave Me on Read! How Avoidance-Based vs. Busyness-Based Attribution of Read Receipts Influences Relationship Outcomes
Yige Wan, Stanford University, Stanford, CA, United States, David Fang

In 2012 instant messaging (IM) services such as Facebook Messenger and Whatsapp released the read receipt feature, allowing users to see when someone has read their message. Through a series of 4 online studies (N = 1421) this paper investigated whether individuals would provide a “desire to avoid” attribution or “too busy to respond” attribution when they send a message that was read without reply (RWR) or when they received a message that they (RWR) and how interpersonal relationships (e.g., trust, closeness, and reliability) become affected by RWR messages. Studies 1 and 2 found that both senders and receivers were more likely to attribute RWR messages to busyness (vs. desire to avoid). Study 3 posited that perceptions of interpersonal factors such as trust, reliability and closeness all dropped for both senders and receivers of RWR messages and that the drop was moderated by an individual’s desire to avoid attribution level. Study 4 further examined this effect in three relationship scenarios of varying closeness levels (parents, friends, and strangers) and found that strangers who sent a message that was RWR tended to underestimate the other party’s desire to avoid them whilst experiencing greater decreases in trust, closeness, and reliability.

4 - The Effect of Incongruency between Attitudes and Behaviors on Social Influence
Yunseob Choi, Yonsei University, Seodaemun-gu, Korea, Republic of, Youjung Jun, Tae-Hyung Pyo, Jae Young Lee

This research examines the effect of incongruency between attitudes and behaviors on social contagion in music consumption context. In this paper, we used a unique data set obtained from Last.fm, one of the most popular music social network websites, including individual users’ music play history, Loved track list (i.e., a list of user’s favorite music), network information, and social tags (i.e., user-generated keywords related with songs). After quantifying coordinates for consumers’ attitudes and behaviors and calculating distance between them, we analyzed how the effect of peers' influence on consumers' adoption of new songs varies by incongruency between attitudes and behaviors. The results show that consumers are more likely to adopt new songs played by their peers as their incongruency between attitudes and behaviors increases. Furthermore, we have found that this effect is moderated by embeddedness ratio – the proportion of followings shared with peers in the focal user's entire network. Consumers with high (vs. low) incongruency between attitudes and behaviors are less (vs. more) likely to be influenced by peers who are sharing lots of common followings with them (i.e., high embeddedness ratio). Based on these results, we discuss theoretical and managerial contributions for marketing researchers and practitioners.

FA20
Meeting Room C4.10
Virtual / Augmented Reality
Contributed Session
Chair: MinChung (MC) Kim, Ulsan National Institute of Science and Technology (UNIST), Ulsan, 44919, Korea, Republic of
1 - Adapting Sense of Presence to Product Uncertainties in Crafting Augmented Reality Enhanced Experiences
Bing Han, Shanghai University of International Business and Economics, Shanghai, China, People's Republic of, Hua Fan

With the advent of metaverse, a groundbreaking development in tourism and hospitality practices is the exhibition of products via augmented reality (AR) mobile apps. However, tourism practitioners face a great challenge of optimizing customers’ sense of presence. Grounded in situated cognition theory, this study examines the different effectiveness of sense of presence (i.e., social vs. spatial presence) created by AR apps in crafting AR-enhanced experience under product quality and fit uncertainties. An online survey with 1970 respondents reveals that, social presence is superior than spatial presence in triggering informative customer experience, while inferior than spatial presence in triggering entertaining customer experience. A field experiment on 1584 users of a hotel chain’s mobile app validates these findings and further reveals that, when product quality uncertainty is high, the superiority of social presence (vs. spatial presence) in driving informative experience will be amplified. But when product fit quality is high, the superiority of spatial presence (vs. social presence) in driving entertaining experience will be mitigated.

2 - Customer Payment Choices in Virtual Reality Retail: Implications for New Sensory Environment in the Metaverse
Yoonsun Jeong, University of Texas at El Paso, El Paso, TX, United States, Nandini Nim, Jessica Felix, Leah Smith
This research explores the nature of customers' payment method choices in virtual reality (VR) retail settings. Blockchain-based payments and cryptocurrency have gained momentum as customer payment methods in the metaverse. In the past decade, digital wallets such as Apple Pay have become popular across multiple retail channels and have become a preferred method. Payments present a customer touchpoint that can create a customer experience problem without proper ubiquity and interoperability across various VR and non-VR retail channels. Thus, leading banks, financial institutions, technology firms, and retailers need to optimize payment technologies and methods for omnichannel retail. We study customer payment preferences among – credit cards, digital wallets, platform currency, and cryptocurrency. Using a VR grocery store and online grocery shopping simulation, we study the impact of an immersive sensory environment. Over two experiments and a qualitative study, we find that customers are more likely to choose digital wallets (Study 1) and platform currency (Study 2) over cryptocurrency across both VR and non-VR retail settings. We build a case for retailers and other stakeholders to build blockchain-based digital wallet capabilities to provide customers with familiarity in the transactions and payments phase of the customer journey in VR retail. This research offers future research questions in the field of VR and metaverse retail payments.





3 - "Human-Technology Symbiosis": the Impact of Technological Embodiment on Consumer Experience in Virtual Reality
Yao He, Southwestern University of Finance and Economics, Chengdu, China, People's Republic of, Xiaorong Fu

The emergence of embodied devices has disrupted the human-technology relationship in virtual reality, whereby the machine becomes a sensory extension of the human body, shaping consumer experiences of purchasing online. Drawing on embodied cognition theory, this study delves into the effects of technological embodiment on consumer experiences and decision-making patterns, as well as the partially mediated role of cross-modal imagery in virtual experiences. Specifically, a VR experience with high technological embodiment can, by generating cross-modal imagery, impact consumer consciousness, leading them to devote greater attention to the sensory information of the product when making subsequent consumption decisions. This research contributes to understanding the impact of multi-sensory VR experiences on consumer behavior, highlighting the potential for forward-thinking marketers and researchers to leverage the latest technologies to create more immersive online experiences for future consumers.

Keywords: technological embodiment, virtual reality experience, cross-modal imagery, decision-making patterns

4 - Effects of Avatar-self Similarity in Brand Metaverse
MinChung (MC) Kim, Ulsan National Institute of Science and Technology (UNIST), Ulsan, Korea, Republic of, Jaehyun Lee, Yeolib Kim
A ‘brand’ metaverse is a virtual space where customers experience the brand via digital avatars. With the advancement of augmented and virtual reality technologies, a brand metaverse is an important medium for communicating the brand with customers. This study focuses on the resemblance between a customer’s self and his/her avatar (i.e., self-avatar resemblance) in the brand metaverse and examines its influence on brand attitude and purchase intention. Although prior literature studies self-avatar resemblance in virtual gaming platforms, few studies examine how it influences customers’ exploration in the brand metaverse. We fill this research gap by uncovering that engagement in the brand metaverse platform mediates the relationship between self-avatar resemblance and brand attitude. Based on the interactive nature of metaverse, we also hypothesize that copresence―the number of avatars concurrently exploring the brand metaverse―strengthens the mediation. By conducting an experiment using a fashion brand’s virtual world in a popular metaverse platform, we measure self-avatar resemblance and test the hypotheses. We find the results largely consistent with the hypotheses, which provide important implications to practitioners considering ‘metaverse marketing.’
FA21
Meeting Room C4.11
Online Reviews: Insights from Large Language Models
Contributed Session
Chair: Chunhua Wu, University of British Columbia, Vancouver, BC, Canada
1 - Using LLMs for Sentiment Analysis of Online Reviews
Peter Danaher, Monash University, Caulfield East, Australia, Stephan Ludwig, Xiaohao Yang, Ehsan Abedin, Lan Du, Yu-Ting Lin

Online customer reviews are commonly summarized by the average star rating (as an indicator for overall satisfaction). While ratings come in a straightforward structured rating scale format, the details of why customers hold this overall sentiment, is hidden in the unstructured review texts. A number of simple and speedy, lexicon-based methods are available for summarizing such review content (e.g., LIWC and VADER). However, as these are created to be applicable across communication contexts, their classification accuracy for any review content aspects or sentiments are poor.

In this study, we adapt a pre-trained LLAMA2-7b-chat model to extract and quantify customer sentiment towards marketing relevant aspects in customer reviews. LLAMA2, provided by Meta AI, is an auto-regressive language model that introduces architectural refinements to existing transformer models. To train the LLAMA2 model, we employed human coders to code for mentions of marketing relevant aspects (i.e., price, product, place, promotion and service) and the corresponding sentiment (positive, negative, or neutral) in a subset of 2,000 Amazon product (across 23 product categories) and 2,000 Yelp service reviews (across 22 service categories). We find that our LLAMA2 method correlates very well with human-coded sentiment evaluations. Indeed, it is better than LIWC, EV 2.0, VADER, a transformer-based model and even GPT4.

2 - UGC or A.I.- GC: Can We Really Detect Product Reviews Generated by A.I. vs. by Consumers?
Lan Luo, University of Southern California, Los Angeles, CA, United States, Liye Ma
“ChatGPT has revolutionized the way people engage with and create product reviews... Its ability to understand and generate human-like text enables it to generate high-quality and engaging product reviews, empowering consumers to make well-informed purchasing decisions.” That was ChatGPT’s own assessment of its impact on product reviews. The proliferation of Large-language models (LLMs) has spurred considerable interest in understanding how they affect digital word-of-mouth (WOM). With that come numerous models/tools to assess whether a particular text is written by a human or an LLM, and such tools have been used in considerable extant research. Nevertheless, to our knowledge, there has been very little research on examining the effectiveness of such LLM detectors on product reviews. In this study, we analyze the performance of such LLM detectors in detecting whether a product review is written by A.I. or consumers. Using a large volume of real-world review data, we find that prevailing LLM detectors either fail to make informative predictions, or make informative predictions with high noise. Even among the informative detectors, there appears to be a consistent bias of labeling longer reviews and higher-rated reviews as being written by A.I. Accounting for the limitations of the LLM detectors, we discuss the implications of our “artificial” ability to detect whether a product review is generated by A.I. or actual consumers.
3 - Coping Through Precise Labeling of Emotions: A Deep Learning Approach to Studying Emotional Granularity in Consumer Reviews
Ali Tamaddoni, Deakin Business School, Burwood, Australia, Ali Faraji-Rad, Atefeh Jebeli

Research in psychology has shown that people’s ability to describe their negative emotions granularly is correlated with successful coping. We introduce the concept of emotional granularity to the consumer behavior literature and develop a deep-learning-based method using the BERT Large Language Model to measure how granularly consumers describe their emotions in their language use. Our method unobtrusively measures the construct at the situation-specific level, using textual data generated by consumers. We study how granularity in describing negative emotions, as a predictor of coping success, predicts online reviewers’ rating of service providers after negative experiences. We demonstrate several novel effects using two dataset of over 12 million reviews: 1) especially when the overall experience with the service provider is negative, higher emotional granularity in describing negative emotions in reviews is associated with more positive ratings of the business; 2) as reviewers write more reviews, they describe their negative emotions more granularly, which predicts higher ratings for negative service experiences from reviewers with a more extensive history of writing reviews; 3) when the service experience is negative, the temporal distance between the negative consumption experience and posting the review is associated with more positive rating of the business, and this effect is mediated by higher granularity in describing negative emotions as a predictor of coping success. Our results have implications for understanding the role of emotional granularity in consumer decision making, understanding the predictors of online review ratings, and profiling consumers based on psychological traits inferred from the online content they generate.

4 - Shifting Standards or Changing Preferences? Unraveling Review Polarization via LLMs
Chunhua Wu, University of British Columbia, Vancouver, BC, Canada, Limin Fang
This paper studies the dynamics of online review polarization – the trend of increasingly extreme reviews over time – and its economic impacts. Analyzing Yelp reviews from 2005 to 2022 through a predictive approach with Large Language Models (LLMs), we distinguish between shifts in rating standards (scale effects) and changes in consumer preferences (content effects). Our results indicate a dominant scale effect in the surge of 5-star reviews, while the increase in 1-star reviews is mainly attributed to content changes. We find minimal influence of societal trends such as political polarization on review polarization. To evaluate the economic impact, we match Yelp reviews to Texas restaurant sales data. Our analysis shows that polarized reviews significantly affect sales. Notably, consumers’ restaurant choices are more significantly influenced by observed ratings than revealed content, highlighting the significant impact of the scale effect with review polarization.
Friday, June 28, 10:20 AM - 11:50 AM
FB01
Meeting Room C2.5
Pricing: Advanced Methods
Contributed Session
Chair: Ushio Dazai, Fukuoka University, Jonan Ward, Nanakuma, Fukuoka, 8140180, Japan
1 - Neither a Picasso nor a Da Vinci: A Multi-modal Model for Pricing of Novice Artwork
Sharmistha Sikdar, Dartmouth College, Hanover, NH, United States, Ishita Chakraborty, Nika Dogonadze

Art pricing is a fundamentally difficult problem especially when the reputation of the artist is unknown. Emerging artists who sell artwork for consumption and aesthetic purposes often find it difficult to determine the right price for their artwork. Online advisory sites determine prices based on artwork dimensions and other heuristics, e.g., price per square inch and artist’s labor rate per hour. However, these advisory sites do not consider several other cues (e.g., quality of images uploaded, painting description, artist’s past sales successes and reviews, etc.) that might affect a buyer’s willingness to buy. Our empirical context focuses on novice artwork sold on Etsy. We use scraped data from Etsy.com on prices, images, and other characteristics of sold artwork and artist shops to build a scalable, multi-modal neural network (NN) model to predict customer willingness to pay (WTP). Our model uses multiple modalities of information, i.e., structured inputs like artwork dimensions, and unstructured inputs like textual data from painting descriptions and visual images of the actual artwork. We find that while the full model with all modalities performs the best, structured inputs contribute most to the overall predictive performance, followed by textual inputs with visual inputs being the least important. We performed a WTP experiment and found that our model predicts the customer WTP better than the online advisory sites. Further, based on our full model, we develop an online application that can be used in real-time by novice artists to determine prices for their artwork.

2 - Impact of Peak-End Effects and Consumer Heterogeneity on Reference Price Formation
Hiroshi Kato, Hakuhodo DY Holdings Inc., Tokyo, Japan, Ryo Kato

The purpose of this study is to investigate the heterogeneity of peak-end effects on prices formed by purchasing behavior. It is hypothesized that peak-end effects on price influence consumers' reference price formation and that the strength of these effects varies with consumer heterogeneity.

This hypothesis is tested for two product categories: green tea and beer. Consumer heterogeneity was expressed using demographic information (gender, age, etc.), behavioral economic measures such as the BIS/BAS scale, the Information Processing Style Scale (IPSJ), the Self-Control Scale, and the time discount rate.

The peak-end effect on price was captured as a weighted average of the highest and most recent prices observed for each consumer. Brand choice behavior was represented as a multinomial logit model, and peak-end effects were represented within brand choice behavior by considering reference price information as a model variable.

The results showed that the addition of reference price information in the multinomial logit model, which accounted for the peak-end effect, improved the data fit and predictive performance of the purchase model. Furthermore, the data fit and predictive performance of each consumer's brand choice behavior were improved by accounting for consumer heterogeneity. This indicates that the peak-end effect is significantly related to reference price formation, and that consumer heterogeneity significantly affects reference price formation through the peak-end effect. The relationship between behavioral economic measures and the peak-end effect, expressed as consumer heterogeneity in this study, is a novel finding in behavioral economics and contributes to a more sophisticated understanding of brand choice behavior.

3 - Gender Inequality And Household Purchase Decisions: The Case of Automobiles in China
Han Zhong, University of Toronto, Toronto, ON, Canada, Zemin (Zachary) Zhong, Nan Chen

We study how gender inequality influences major household purchases in the context of the Chinese auto market. Using administrative data on car registration, we document a significant gender ownership gap that is correlated with local gender inequality. Consumers in more gender-equal regions are also more likely to buy cars of female-preferred colors. We use the introduction of compulsory education in China, which reduced the gender education gap, as a natural experiment to test the causal effect of gender inequality on car purchases. We find that reducing gender education inequality increases both female ownership and female-preferred colors in car purchases.

4 - Price Sensitivity and its Regularity in Subscription Service
Ushio Dazai, Fukuoka University, Fukuoka, Japan

In recent years, many subscription services, such as Netflix or Amazon Prime, have raised their prices. While concerns regarding customer churn abound, understanding the nature and degree of customer response to these price hikes remains elusive.

This study explores price sensitivity in subscription services for digital goods. First, the author examines the distribution of price sensitivity using the price sensitivity measurement method for users and shows that the distribution closely resembles that of the convenience goods.

Following the confirmation of price sensitivity, the author explores the relationship between customer satisfaction and price sensitivity, a crucial aspect when contemplating price increases. However, contrary to existing research suggesting that higher customer satisfaction increases price sensitivity or willingness to pay (WTP), the author did not find such a relationship. The study introduces the concept of weak payment awareness in subscriptions.

Furthermore, the author demonstrates the observation of a law of ratio-transformed price sensitivity. Drawing from Weber's law, which asserts that people perceive physical stimuli in terms of ratios, the study presents empirical results indicating that ratio-transformed price sensitivity remains consistent across different price ranges of services. This includes no statistically significant differences in the ratio of amounts perceived as "cheap" or "expensive" across various subscription services. The paper also explains the lack of perception of small price changes, outlining the presence of a low price sensitivity region.

The findings on price sensitivity in subscriptions presented in this paper will serve as a reference when revising prices.

FB02
Meeting Room C2.6
Game Theory for Competitor Strategy 2
Contributed Session
Chair: Parneet Pahwa, University of Texas- Dallas, Richardson, TX, United States
1 - How Collusion Ends: The Case of Advertising Agencies
Sharon Horsky, Hebrew University at Jerusalem, Jerusalem, Israel, Robert Zeithammer

This paper examines the longstanding practice of compensating advertising agencies with a fixed commission on media billings, and empirically shows what eventually led to its demise. Using a combination of archival research and a unique longitudinal dataset we assembled, we first document a collusive arrangement between agencies and publishers, and then we show how the collusion unraveled. Our data shows that the collusive arrangement was gradually weakening due to increased transparency and price pressure. While various “sliding scale” and “discount” contracts emerged over time these were still expressed as a percentage of media commission. The core institution of compensating agencies based on commission on media billings only disappeared after the entry of a new type of agency player which led to widespread unbundling of the advertising tasks. Our analysis has implications for the analysis of cartels, as well as for understanding the compensation methods currently evolving in the digital advertising market.

2 - The Impact of Environmental Regulations on Business Operations
Xuecheng Fan, Sichuan University, Chengdu, China, People's Republic of, Yichen Zhao

Regulatory policies often have a certain orientation, which may bring unexpected situations while guiding the regulated object. For example, encouraging the development of certain industries through policy may lead to overcapacity. Conversely, suppressing the development of certain industries through policy may cause many enterprises to withdraw from the market, which increases the market share of incumbent enterprises in a disguised way. In this way, from the perspective of some regulated persons, regulation is not necessarily a bad thing. As an integral part of the modern environmental governance system, environmental regulations have the potential to alter the dynamics of competition and cooperation, resulting in a phenomenon known as the “relatively better” effect. In light of this phenomenon, we have developed a static game model based on incomplete information to analyze its implications. Through the exploration of Nash equilibrium, we have discovered that the "relatively better" effect allows environmentally regulated clean enterprises to transfer a portion of their costs to polluting enterprises. Consequently, clean enterprises experience lower operating costs compared to their polluting counterparts under environmental regulation conditions. This conclusion is supported by empirical analysis conducted using Chinese enterprise data.

3 - The Conditions for Competitive Offering and Segmentation to Improve Consumer Wellbeing: An Analytical Approach.
Zecong Ma, St. Mary's University, San Antonio, Texas, United States

In this research, a mathematical framework encompassing critical marketing constructs is proposed. Drawing upon existing marketing literature, mathematical definitions have been formulated for consumer-centric constructs including need, satisfaction, wellbeing, and value, along with business-oriented constructs such as offering, cost, and market. Utilizing these definitions, coupled with a set of foundational axioms, two key theorems are derived and validated. The first theorem elucidates the conditions under which competitive offerings improve customer wellbeing, while the second theorem delineates the conditions enabling segmentation strategies to improve customer wellbeing. This analytical methodology presents an innovative pathway for the advancement of marketing theories.

4 - Stand on Social Issues: Should Brands Take a Position?
Parneet Pahwa, University of Texas at Dallas, Dallas, TX, United States

Many brands take a position on social causes. A study by Deloitte found 75% of respondents saying that taking a stand demonstrates that the company cares about more than just profits and with 70% saying that it would help attract customers and partners. But this positive attitude is not without cynicism. May customers believe that brands that take a stand do so for PR and marketing purposes and still others believe that when a brand speaks out it is just "jumping on the bandwagon".

We develop a game-theoretic model in which two firms compete in a market to serve customers who differ in their position as it relates to a social issue. Firms' core values are common knowledge and may be for, against or neutral to the social issue. Firms endogenously decide whether or not to take a public position on the social issue, while recognizing the consequences of taking a position which is (is not) consistent with its core values and the position of its core customers. We identify market conditions where in equilibrium firms adopt symmetric strategies and take a neutral position or are in favor/against the social issue. We also find that under certain market conditions in equilibrium firms adopt asymmetric strategies with one firm taking a position in favor while the other takes a position against the social issue. Finally, we also explore whether it ever optimal for firms with identical core values to take different positions on a social issue.

FB03
Cockle Bay Room 1
Sharing Economy 2
Contributed Session
Chair: Luqian Sun, University of Minnesota, 1400 S 2ND STREET, Minneapolis, MN, 55454, United States
1 - AI Divide Versus Inclusion: Evidence from Algorithmic Task Assignment in a Food Delivery Platform
Tae Jung Yoon, KAIST College of Business, Seoul, Korea, Republic of, Yeonseo Kim, Jiyong Eom

With the growing popularity of artificial intelligence (AI) in many business sectors, gig economy platforms are at the forefront of introducing various algorithmic approaches in their workplace. Specifically, these platforms have widely adopted algorithmic task assignment systems to efficiently match customer service requests with platform workers, enhancing operational efficiency and workforce management. However, there are growing concerns about whether this algorithmic management approach also benefits all platform stakeholders, and whether all of them equally benefit from the algorithm. In a unique setting of a food delivery platform, we study both the impact of algorithmic task assignment on delivery workers’ productivity, including disparities among them, and the eventual quality of customer service. Our results show that the adoption of task assignment AI improves worker productivity on average by nearly 3%, with the improvement primarily concentrated among medium-skilled workers. We find evidence that task assignment AI enables delivery workers to deliver more orders simultaneously (i.e., stack more orders) by assigning them appropriate orders based on their current situation, increasing daily orders and earnings in low- and medium-skilled workers. These results indicate that algorithmic task assignment can partially alleviate existing productivity disparities among workers. These AI-driven changes in the workplace shorten customer waiting time when a single order is delivered, but this is not the case for stacked order delivery which involves workers’ opportunistic behavior. Our study demonstrates that algorithmic task assignment may not benefit all platform stakeholders, but it can at least be a Pareto-improving strategy for them.

2 - Investigating the Sales Impact of (In)formal Entrepreneurship in Home-sharing Platforms: The Contingent Roles of Institutional Environments
Zhenxin Xiao, Xi'an Jiaotong University, Xi'an, China, People's Republic of, Siqiu Lian, Xiaoxiao Liu, Xiang Gong
With the rapid advancement of the sharing economy, numerous opportunities have emerged for individuals to establish their own businesses. While existing literature has classified entrepreneurship into three categories: informal, semi-formal, and formal, there is a lack of research that compares the financial returns of these types of entrepreneurship. Drawing on institutional theory, this paper aims to quantify the sales performance of different types of entrepreneurship in the home-sharing economy and explore how they are influenced by external institutional environments. Our study focuses on Airbnb, the leading global home-sharing platform, offering an appropriate context to address our research questions. We categorize hosts with official registration licenses from government agencies as formal entrepreneurs, while those displaying self-declared natural person merchant information as permits are deemed semi-formal entrepreneurs. Hosts without any permit information fall under informal entrepreneurs. We utilize a unique dataset derived from Airbnb listings and area statistical yearbook data, and our difference-in-differences analysis covers approximately 411,577 properties from 112,258 hosts between December 2021 and July 2022. Our findings indicate that hosts utilizing semi-formal entrepreneurship experience the highest sales performance, while hosts employing formal entrepreneurship achieve the lowest sales performance. Furthermore, our analysis demonstrates that the impact of (in)formal entrepreneurship on sales performance is contingent upon regional legal development. This paper not only enriches the home-sharing economy literature by incorporating aspects of entrepreneurship and institutional theory research but also underscores the nuanced dynamics among entrepreneurship types and their performance within the home-sharing economy.
3 - Shared Journeys, Diminished Gratuities: An Empirical Analysis of Tipping Behaviors in Shared Ride Services
Luqian Sun, University of Minnesota, Minneapolis, MN, United States, Linli Xu

In gig economy, platforms like Uber, Lyft and food delivery services often provide users with the option to tip drivers through their apps. The convenience of digital transactions has made tipping more accessible, and it has become a means of acknowledging and rewarding the efforts of individuals who provide services in the online space. The norms and expectations for online tipping can vary across platforms, but the practice has undeniably become a significant part of the digital economy. We aim to understand the tipping behavior on digital platforms. Specifically, we examine how and why customers tip differently when they use UberX Share, a service that offers cost-effective and environmental benefits that allows riders heading in the same direction to share a ride. Using trip-level data from New York City, riders using the UberX Share services are less likely to and, when they tip, they tip less (in absolute amount as well as tip percentage). We further provide evidence to explain such behaviors.

FB04
Cockle Bay Room 2
Panel on AI-Driven Marketing in a Digital World
Invited Session
Special Sessions
Chair: Pankhuri Malhotra, University of Oklahoma, Norman, OK, United States
1 - Panelist - Koen Pauwels
Koen Pauwels, Northeastern University, Newton, United States
2 - Panelist
Lan Luo, University of Maryland-College Park, College Park, MD, United States
3 - Panelist
Ryan Dew, Wharton, Philadelphia, United States
4 - Panelist
Sharmistha Sikdar, Dartmouth College, Hanover, NH, United States
5 - AI-Driven Marketing in a Digital World
Pankhuri Malhotra, University of Oklahoma, Norman, OK, United States

This special session is dedicated to bringing scholars and practitioners with a keen interest in the intersection of Artificial Intelligence (AI) and Marketing Research. It will feature a series of talks and panel discussions focused on the dynamic changes in marketing strategies and tactics prompted by rapid advancements in AI.


We're also excited to introduce the new Artificial Intelligence Special Interest Group (AISIG), a forum designed for engaging discussions, sharing cutting-edge research, and establishing best practice guidelines in this AI-transformed marketing landscape.


Details of the SIG - https://www.ama.org/ai-sig/. Key topics include AI-driven applications in generating customer insights, optimizing marketing strategies, enhancing customer service and advertising, and the ethical dimensions of AI in marketing, such as privacy and responsible use.

6 - Panelist
Unnati Narang, University of Illinois at Urbana Champaign, Champaign, IL, United States
7 - Panelist
Davide Proserpio, University of Southern California, Los Angeles, CA, United States
FB05
Meeting Room C3.1
Marketing Strategy: Alliances, Mergers and Acquisitions
Contributed Session
Chair: Mitsukuni Nishida, Johns Hopkins Carey Business School, 100 International Drive, Baltimore, MD, 21202, United States
1 - The Construction of State-owned Enterprises Under the Rule of Law and the Efficiency of Mergers and Acquisitions-Empirical Evidence Based on the General Legal Advisor System
Ju Mao, Renmin University of China, Renmin, China, People's Republic of

The system of general legal advisors for SOEs is an important governance structure design for the construction of rule of law state-owned enterprises (SOEs), but there is little literature on its effectiveness. This article is based on the scenario of mergers and acquisitions (M&A), and takes the efficiency of M&A as the starting point to examine the implementation effect of the general legal advisor system for SOEs. Research has found that: (1) Companies implementing the system of general legal advisors for SOEs have higher efficiency in M&A, manifested in better operational and market performance, and this effect depends on the professional ability and power of the general legal advisor. This indicates that the general legal advisor for SOEs has played a positive role in the decision-making process of M&A, which helps to improve the efficiency of M&A. (2) The impact of the general legal advisor of SOEs on the efficiency of M&A is heterogeneous, and this positive effect is mainly reflected in local SOEs and commercial SOEs. (3) The path of this impact is that the general legal advisor of SOEs can help reduce ineffective M&A in advance, enhance the ability to integrate M&A after the fact, and reduce the risk of goodwill impairment and bankruptcy. This article reveals the impact of the construction of rule of law SOEs with the general legal advisor system as the core on merger and acquisition activities, providing intuitive evidence for the implementation effect of the general legal advisor system in SOEs.

2 - Consumer Switching Behaviour After Mergers: Evidence from the T-Mobile/Sprint Merger
Lachlan Deer, Tilburg University, Tilburg,, Netherlands, Shrabastee Banerjee, Jura Liaukonyte, Daniel McCarthy

This paper investigates consumers' decisions to switch brands in the aftermath of a horizontal merger. Despite a large literature that provides theoretical- and counterfactual-predictions about how consumers respond to mergers there is limited empirical evidence on how consumers actually respond to mergers by changing their product choices. We leverage monthly consumer choice data over 2018-2022 from a large representative sample of US consumers to measure the effect of the merger of T-Mobile and Sprint on consumer brand choices and switching between providers. The merger took place in April 2020 with the intention of improving the competitiveness and signal coverage to compete with AT&T and Verizon, particularly in rural areas. Our preliminary findings indicate that on average the merged T-Mobile/Sprint entity attracts new consumers away from Verizon and AT&T, with substantial heterogeneity in the magnitude of the effects between urban and rural consumers. The analysis connects heterogeneity in consumer’s switching decisions to underlying geographical differences in improved product quality offered by the merged T-Mobile/Sprint offering viz-a-viz to its competitors. The analysis further documents the impact of localized store closures as a result of the merger on a consumer’s decision to change providers.

3 - When Mergers Fail: Evaluating a Role of Intangibles behind a Slowdown after an Acquisition
Mitsukuni NIshidi, Johns Hopkins Carey Business School, Baltimore, MD, United States
Although the profitability of firms drives the merger motives, nearly half of all mergers fail. This paper analyzes whether a past merger made merging firms more or less profitable, and through what mechanisms did the merger affect profitability. Using an acquisition in the Japanese convenience-store industry as a case study, I quantify the effects of the merger on revenues, costs, and intangible factors in conjunction with geography. The model relies on limited and publicly available information on sales and store counts of the convenience-store chains in 47 geographical markets, before and after an actual merger event. The descriptive analysis documents that the merging firms experienced a slowdown in sales performance, and the magnitude of the slowdown differs across geographic markets. To examine the effect of the merger on firm profitability, I estimate a dynamic game of outlet expansion with latent state variables using sequential Monte Carlo methods. Intangible factors are incorporated through unobserved state variables that are serially correlated and heterogeneous across chains. The estimates reveal these intangible factors declined following the merger, and that this decay is associated with the distance from the parent company's headquarters.
FB06
Meeting Room C3.2
Retailing: Store Choice/Loyalty/Shopping Behaviors
Contributed Session
Chair: Sonika Singh, University of Technology-Sydney, Sydney, Australia
1 - The Relationship between Context of Creating a Shopping List and Planned Purchase Execution
Takashi Teramoto, Chuo University, Hachioji, Japan, Satoshi Nakano, Naoki Akamatsu, Shin Sato
We examine the relationship between the descriptions on shopping lists used for offline-shopping and the execution of planned purchases. Shopping lists are addressed in the following arguments: the relationship between bringing shopping lists and unplanned purchases, and the relationship between content of shopping lists and purchases. However, the following contexts of creating a shopping list have not been argued: 1) planning granularity of listed items and 2) time difference between list and purchase. Our study fills these gaps. We use single-source data of mobile-app shopping list and purchase histories of 788 shoppers which consist of 4012 observations. The dependent variable is whether the list items are planned purchased, and the independent variables are three levels of granularity of item planning, the time difference between the list and the purchase, and the interaction of these variables. A random-effects binomial logit model is applied to the analysis. The results show that the more specific the granularity of the plan and the smaller the time difference, the more likely the planned purchase is executed. In addition, the specific-category level, which means a medium level of planning granularity, has a higher probability of planned purchase for a larger time difference than the other levels. The theoretical contributions of our study are 1) incorporating the context of shopping list creation and 2) applying single-source data of mobile shopping lists and ID-POS data to research in planned/unplanned purchase behavior, which has mainly applied data based on shopper intercept survey.
2 - Do Household Life Transitions Change Store Loyalty?
Satheesh Seenivasan, Monash University, Caulfield, Australia, Fatima Madani

In households, multiple life events transpire, such as marriage, divorce, childbirth etc., which transition them across different stages in their life cycles. These transitions engender economic and psychological changes within households and could have substantial implications for their shopping behavior. This study investigates the impact of Household Life Cycle (HLC) transitions on store loyalty. Building on the theoretical foundation of sense of belonging, we hypothesize that experiencing HLC transitions that decrease the sense of belonging will decrease household store loyalty and vice versa. Using Nielsen panel data and employing difference-in-differences models with propensity score matching, we show that household’s loyalty trait increases after childbirth, but decreases after divorce and empty nesting. We further study the moderating effects of retailer and household characteristics on the effects of HLC transitions on store loyalty. Our findings contribute to HLC and life events literature and have important implications for retailers and store managers.

3 - Uncovering Consumer Heterogeneity in Big Data: A Hybrid Marketing Science - Deep Learning Approach
Seung-Wook Kim, Bentley University, Iowa City, IA, United States, Gary Russell

Traditional choice models for understanding market segments are computationally expensive, hampering calibration in Big Data. A potential solution lies in machine learning; however, its lack of marketing theory foundation and interpretability hinders managerial use. Given these challenges, this paper describes an approach to the calibration of a choice model with unobserved consumer heterogeneity in a Big Data context. Our proposed Autoencoder-Latent Class Model (ALCM) commences with utilizing autoencoder – a type of feedforward deep neural network – to create parsimonious latent representations of a large volume of consumers based on their different shopping patterns. We then use the latent representation with stratified sampling to produce a high-quality sample of the original data. The choice model is then fit to this high-quality sample. Simulation demonstrates that the autoencoder-generated latent representation consistently enhances consumer clustering accuracy, reducing sampling errors by up to 50%. Application of the ALCM to real-world data from the carbonated soft drink industry achieves 82% accurate consumer segmentation prediction with a remarkable 95% reduction in computational time compared to traditional models. This innovative modeling approach offers promising avenues for large retailers and researchers grappling with conventional marketing science models in the realm of Big Data.

4 - In-store Shopping - Does Fluidity of Time and Money Matter?
Sonika Singh, University of Technology-Sydney, Sydney, Australia, Svetlana Bogomolova, Harmen Oppewal

In this study we examine consumers’ fluidity of time and money during grocery shopping – that is the difference between the actual and planned spending of time and money in a supermarket. Marketers are increasingly allocating larger proportions of their marketing budgets on in-store promotions to attract consumer attention and generate sales. While sales data reveal how consumers eventually allocated their monetary budgets across categories and brands, yet insights into how these sales were generated are typically limited. Understanding how monetary and time budgets relate to activities undertaken by a shopper in an individual shopping trip is crucial for generating more insights into how consumers act on their intentions to arrive at their purchase decisions. While most studies consider time and money budgets as important, these variables are mostly used as control variables. The notion of fluidity of consumers’ time and money has received little attention in the marketing literature. Using eye tracking data from supermarket intercept survey before and after the shopping trip, we empirically investigate the drivers of consumers’ fluidity of time and money. Our results indicate that in-store promotions, shopper attitudes and demographic characteristics influence the fluidity of money. Planned shoppers exhibit fluidity in their spending, and the actual time spent and money spent increases if the store department is on the shopping list.

FB07
Meeting Room C3.3
Causal Inference and Public Policy
Invited Session
Special Sessions
Chair: Unnati Narang, University of Illinois at Urbana Champaign, Champaign, IL, United States
Co-Chair: Ali Goli, University of Washington, 2901 NE BLAKELEY ST. APT 131, Seattle, WA, 98105, United States
1 - Information Overload vs Signal Reaffirmation: Empirical Evidence from Goodreads.com
Shrabastee Banerjee, Tilburg University, Tilburg, Netherlands, Roshini Sudhaharan
The proliferation of written content on online platforms has created an enormous volume of user-generated content (UGC), including reviews, ratings, and Questions and Answers (Q&A), with the potential for interactions among these pieces of information remaining a less-explored area of study. Online platforms utilize these UGCs as observable information signals to alleviate product uncertainty. This article studies the dynamics between these signals by examining the effect of Q&A on subsequent ratings. Leveraging a natural experiment of staggered Q&A implementation at Goodreads and comparing reviews and ratings at Amazon, where there was no Q&A, we find that Q&A leads to a decrease in subsequent ratings by 0.1 stars on average, mainly in the case of fiction books. Q&A has no significant effect on the ratings of non-fiction books. The differential impact can be attributed to fiction books' relatively more subjective nature than non-fiction.
Moreover, we find that more subjective questions lead to negative ratings. Thus, the extent of the decrease in ratings due to Q&As is contingent upon the degree of concordance between information signals. Our findings illuminate the boundary conditions of information provision, especially in contexts where products possess subjective attributes, inviting conflicting information signals, resulting in heightened product uncertainty and diverse consumer preferences. These insights offer valuable guidance for e-commerce platforms in optimizing the integration of diverse information signals to enhance consumer decision-making processes.
2 - The Impact of Fintech in Emerging Market Retail
Shreya Kankanhalli, Cornell University, Ithaca, NY, United States, Steve Anderson, Sridhar Narayanan, Leonardo Iacovone

Across emerging markets, cash is the ubiquitous means of transaction for small-scale retailers and the customers they serve. Policymakers and marketers are interested in understanding how to bridge this gap and stimulate the growth of electronic payments in emerging markets. They broadly view usage of electronic payments as a tool for small-scale retailers to scale-up, improve their performance and raise their transparency. However, no prior study has rigorously analyzed the impact of using electronic payments on the performance of small-scale retailers, nor the mechanisms by which performance gains can occur. Our study does so via a randomized control trial in Guadalajara, Mexico. It promotes the adoption of an electronic payment technology for randomly selected retailers to study its eventual impact on retailer performance. 845 small-scale retailers are randomized into four groups:

  • Control: No intervention.
  • Technology: These retailers are given a smartphone that is enabled with technology to receive electronic payments.
  • Technology + Marketing: These retailers also receive assistance with marketing the benefits of e-payment technology to customers.
  • Technology + Marketing + Commissions: These retailers also receive a four-month waiver for the commission normally paid on each e-payment transaction.

We collect data, 24 months post-intervention on outcomes such as sales, costs, and profits to answer how e-payment adoption impacts retail performance. Moreover, we show mechanisms by which these performance gains occur, namely how e-payments enable i) access to new customer segments, ii) accumulation of working capital for productive investments, and iii) adoption of other digital marketing technologies.

3 - Managing Online Responses for Optimizing Future Reviews and Demand
Ishita Chakraborty, University of Wisconsin Madison, Madison, WI, United States, Hulya Karaman, Shrabastee Banerjee
Online reviews have now become more interactive with businesses responding to customer complaints/praises on public review platforms. Management responses (MR) can impact both future reviewers as well as demand. While the impact on future reviews has been explored, the impact on demand has not been studied. Also, the important question of how to respond has remained largely unexplored. In this paper, we leverage a conversational dataset comprising of online reviews, responses and subsequent sales over 4 years from a major international hotel group’s proprietary CRM system to study the impact of both these decisions (whether to respond and how ) on subsequent ratings and demand. We employ NLP techniques, such as topic modeling, context-specific dictionaries, and transformer-based models (RoBERTa), to categorize response text into strategies beyond simple metrics like response length. This comprehensive list of strategies covers responses to both positive and negative reviews, including gratitude, loyalty mentions, revisit requests, regret, denials, excuses, and explanations. Overall, we find that responding has a positive impact on both ratings and demand. The impact of response strategy is nuanced; more importantly, the type of response can have differential impact on future reviews and sales based on whether it is a response to a positive or negative review. Thus, optimizing for better ratings does not necessarily optimize for higher demand. For instance, while engaging with negative reviews using explanations and regret can make future reviews more negative as it can encourage more complaints, yet it enhances demand as prospective buyers can see managers acknowledging shortcomings.
FB08
Meeting Room C3.4
Marketing, Digitization, and Society
Invited Session
Special Sessions
Chair: Ziwei Cong, Georgetown University, Fairfax, VA, United States
Co-Chair: Yixing Chen, University of Notre Dame, Notre Dame, IN, United States
1 - The Value of External Data Capabilities in Search Markets: Evidence from a Field Experiment
Ananya Sen, Carnegie Mellon University, Pittsburgh, United States, Yixing Chen, Xiaoxia Lei
This paper measures the relative performance implication of expert-based and algorithm-based recommendations in the context of business-to-business (B2B) e-commerce. Using data from a B2B e-commerce platform that connects buyers and suppliers in construction industry, we document an efficiency-equity dilemma inherent in algorithm-based recommendations. On the one hand, on average, the algorithm outperforms human experts in terms of suppliers’ response rates and bid-hit ratio. On the other hand, relative to human experts, the algorithm tends to magnify market inequities by directing a larger volume of orders to well-established suppliers than to their less-established counterparts. Our findings highlight the importance of human-algorithm integration in the realm of B2B digital commerce.
2 - When the Giant Falls: The Unintended Consequences of Bud Light's LGBTQ+ Collaboration
Chen Jing, Boston University, Boston, MA, United States, Dokyun Lee, Shuba Srinivasan

Brands increasingly face risks when engaging in sociopolitical issues. Although their initial intention is to benefit the firm and subsequent responses aim to mitigate any risks or harm, there is limited empirical evidence about the unintended consequences of such risk events on a brand's products, its industry, and its competitors. This study investigates such a scenario that unfolded in April 2023 when Bud Light, a renowned beverage brand, launched a hasty promotional campaign in partnership with LGBTQ+ influencer Dylan Mulvaney. The campaign, however, quickly faced significant backlash from conservative groups, leading to a substantial decrease in the market value of Anheuser-Busch, Bud Light's parent company. Using social media, brand health tracking data, and stock market, our research quantifies the overall impact of Bud Light's sociopolitical risk event and the company's subsequent responses on firm and competitor’s performance. We aim to explore the heterogeneity in public reactions, focusing on identifying customer segments most likely to react to both the risk event and specific firm responses. Additionally, the study examines the ripple effects of such events on competing brands and industries, investigating changes in market structure. We offer a comprehensive understanding of the dynamics involved when brands engage in sociopolitical issues, along with practical managerial implications for the competing category.

3 - Unlocking the Power of Joining a Virtual Team: The Heterogeneous Impacts of Team Signaling in Telehealth
Siliang Tong, Nanyang Technological University, Singapore, Singapore, Shiyi Wang, John Dong, Jianjun Wang

The recent proliferation of Telehealth platforms has engendered immense business potential in improving patients’ welfare with accessible medical resources and facilitates doctors to expand services to geo-distant locations. In particular, the development of communication and information technology empowers the formation of virtual teams in which doctors could integrate medical resources and specialty knowledge across institutions and departments for more comprehensive medical services. This study aims to empirically examine the business value of joining a virtual team on individual doctor's patient demands. Drawing upon signaling theory, we develop a set of hypotheses to provide a theoretical explanation by which virtual team information could affect different types of doctors (e.g., job titles, tenure, popularity, and region) and what types of virtual team (e.g., specialty vs institution based) benefit individual doctors most. We leverage a quasi-natural experiment setting, in which an interface revamp on a leading Telehealth platform in China results in the team consultation tab showing up on the personal page of each doctor in the team. Applying Synthetic Difference-in-Differences (SDID), Generalized Synthetic Control (GSC), and Double Machine Learning (DML) to strengthen the causal identification, we document several novel findings. Firstly, our results suggest that doctors could receive 27.0% more consultations by joining the virtual team. Secondly, we find that the impact of joining virtual teams is more significant for disadvantaged doctors. Thirdly, we find joining the institution-based virtual teams could bring more benefits to doctors compared with joining the specialty-based teams.

4 - Brain Speaks Louder than Words
Qi Zhao, Penn State University, State College, PA, United States, Min Ding

It is common for marketing researchers and practitioners to collect attitudinal opinions in person on attitudes toward new products, technologies, and relevant controversial topics (such as ethics, sustainability, and privacy). However, it is well known that such responses have biases for various reasons. Leveraging digital neuroscientific techniques, this research devises a novel approach to infer respondents’ truthful answers directly from cognitive processing that is clean from responding biases. The proposed approach draws upon the cognitive dissonance theory and sequentially presents synthetic stimuli to artificially induce more systematic processing when the presenting stimulus is incongruent with the respondent’s truthful opinion. We regard exposures to synthetic stimuli as focal events and adopt the event study method to detect the underlying cognitive processing through the proposed sequential stimuli presentation. For validation, we implement a lab study to collect and infer respondents’ preferences on Chinese dishes, a common product category in which our college student respondents have experience. We find significant neural evidence supporting more systematic processing upon incongruent stimuli exposure, while the presence of a heuristic cue attenuates this significance. Compared with benchmark methods, the proposed approach achieves significantly higher accuracy in predicting respondents’ real preferences (e.g., 63% vs. 40%). Contrary to existing methods, the proposed approach entails no training data to make individual-level inferences and no direct reporting from participants, making itself appealing to studies requiring a continuous participation experience.

FB09
Meeting Room C3.5
Brand Management Using AI and Text Analysis
Contributed Session
Chair: Masahiko Yamanaka, KSP-SP Company, Ltd., NO.R Hamamatsucyo 4F, Minato-ku, 1050013, Japan
1 - GPT-4V: A Faster, Cheaper, More Accurate Non-brand Name Suggestiveness Coder?
Larissa Bali, Ehrenberg Bass Institute for Marketing Science, Adelaide, Australia, Zachary Anesbury, Peilin Phua
We assessed the ability of Artificial Intelligence’s new visual processing capabilities to replace human coders on a time consuming and expensive task. Specifically, we used GPT-4V to classify over 500 Distinctive Assets (i.e., non-brand name branding elements including logos and taglines) from twelve categories and three countries as either suggestive or non-suggestive. Brand suggestiveness refers to a branding element that conveys relevant attribute or benefit information relevant to a particular product or category context (Keller et al. (1998)). Our results demonstrate the high potential of this early-stage model. It produced promising intercoder reliability at 75% (from comparing GPT-4V’s output to a benchmark from three human coders). Surprisingly just 50% of Distinctive Assets are suggestive, going against the recommendations in academic literature (Gunasti et al. 2020; Lam et al. 2013; Sen 1999) and marketing practitioner preference (Romaniuk 2022). Our brand suggestiveness study demonstrates that GPT-4V might be a tool for all sorts of academic research in marketing. Already, GPT-4V provides marketing academics and practitioners with a time and cost-efficient way to produce scientific knowledge.
2 - Shelf Face Synergies Among Items within the Same Brand - Leveraging POS Data for Co-selling Sales Patterns Utilizing Concurrent Selling Ppatterns of Items by POS Data
Masahiko Yamanaka, KSP-SP Company, Ltd., Minato-ku, Japan, Kazuharu Kure

Through the analysis of co-selling patterns in POS data, we have identified a positive effect among three items of the same brand. In the Japanese frozen food market in 2023, we discovered a synergy effect of shelf faces between a flagship product and two new products. The term " shelf face synergy" refers to the situation where selling additional products together with a flagship product results in higher sales than selling the flagship product alone.
Our developed methodology not only identifies positive effects like the shelf face synergy but also uncovers its negative effects as well as the independent relationships among items. Our approach involves two steps. First, using monthly POS data, we identify potential relationships between flagship products and additional items. For this purpose, store-specific data is aggregated and compared based on sales patterns. The second step involves selecting stores where flagship products and additional items are displayed together. We use machine learning and non-linear regression to analyze the relationship between the daily sales amount (PI) of flagship products and additional items. Explainable AI (XAI) is employed to facilitate interpretation.
We believe that the relationships among targeted items derived from co-selling patterns in POS data can be widely utilized for optimizing product lines and strengthening brand management.

FB10
Meeting Room C3.6
Social Media Marketing
Contributed Session
Chair: Ayesha Hossain, Sydney, Australia
1 - Individual Conformity Behavior: The Metaphorical Power of Video Brightness
Yixun Lin, School of Economics and Management, Beihang University, Beijing, China, People's Republic of, Chenxi Li, Siwen Ding, Xinyue Xia

Live comments, as a unique form of viewer-generated commentary, present an opportunity to study the conformity effect, which could potentially rival word-of-mouth in shaping promotional outcomes for videos. Prior studies indicate that human factors (e.g., individual difference), content characteristics (e.g., comment valance) and surrounding contexts (e.g., sensory of warm) influence individual conformity behaviors. However, there is a notable gap in the literature regarding the impact of surrounding context of brightens in videos on live comments. Displayed concurrently with video content, live comments are integrated into the visual environment created by the video. Although previous research has recognized brightness as a critical visual cue with metaphorical significance, its influence on individual conformity behaviors in response to live comments remains unclear. Employing metaphor theory, this research empirically addresses the gap through mixed methods. Based on a secondary data analysis on one of the largest video platforms, our findings demonstrate that increased brightness intensifies conformity behavior. Furthermore, the above relationship is moderated by the degree of exposure and the comment type. In addition, the results of a follow-up laboratory experiment indicate that the influence of brightness on conformity behaviors is explained by positive emotions. This research offers video producers valuable insights into leveraging visual elements like brightness to enhance conformity behavior, providing marketers with strategies to craft more effective, high-conformity content. Additionally, the principles uncovered extend beyond video promotion, serving as a versatile framework for enriching video production across multiple contexts.

2 - How Firms Can Steer Social Media Conversations Using Dialogic Listening
Harald van Heerde, University of New South Wales, Sydney, Australia, Mohammad Saljoughian, Kelly Hewett, William Rand

Many firms struggle with how to craft their messages in conversations with customers on social media. The problem is compounded by the fact that these conversations take place in different, simultaneous threads, each of which potentially requires a different approach. This paper studies how firms can adapt their responses in individual social media conversations such that the sentiment in each conversation becomes more favorable. In doing so, the paper uses a new conceptualization to conceptualize five components of dialogic listening: 1) empathetic understanding, 2) unconditional positive regard, 3) spirit of mutual understanding, 4) presentness, and 5) genuineness. Based on an analysis of close to a million tweets capturing over 200k threads involving four major US banks across 10 years, this paper provides concrete managerial guidance about how to respond to have the best possible effect on the resulting user sentiment. This advice is provided for conversations that are both critical of the firm or positive about the firm. The paper also illustrates that lifting user sentiment in social media conversations leads to an increase in stock price, an important measure of firm performance.

3 - Informative or Emotional? How MGC Impacts Consumer Engagement Across Social Platforms?
Xiaoran Xie, Shanghai Jiao Tong University, Shanghai, China, People's Republic of, Peng Liu, Suguo Du

Abstract

Most brands release MGC (Marketer-Generated Content) on social platforms in a "one-size-fits-all" approach, publishing the same content on all social platforms. However, the same MGC may lead to different levels of user engagement (e.g., likes, shares, comments) across different social platforms. In this study, we propose two social platform characteristics: emotional characteristics (e.g., platform intimacy, platform loyalty) and content characteristics (e.g., content presentation form, credibility). We develop an empirical model to study how social platform characteristics interact with MGC, and how this interaction leads to different consumer engagement. We use real customer response data from popular Chinese social media platforms (Study 1) and experimental data (Study 2) to help marketers to understand the optimized MGC marketing strategy across social platforms to lead the best consumer engagement. We differentiate the impacts of emotional characteristics and content characteristics, and their interplay in driving user online engagement. Merchants should adopt different MGC delivery strategies according to campaign advertising intent (increasing sales vs. increasing brand awareness).

Keywords: social media platform, advertising campaign intent, user engagement
4 - The Impact of Social Media on Music Demand: Evidence from a Quasi-Natural Experiment
Nils Wlomert, Vienna University of Economics & Business, Vienna, Austria, Daniel Winkler, Christian Hotz-Behofsits, Jura Liaukonyte
The digital age has significantly changed how music is consumed, promoted, and monetized. Social media platforms like TikTok are playing a pivotal role in this transformation. This shift has sparked a debate within the music industry: While some stakeholders see social media platforms like TikTok as opportunities to boost songs to viral status, others raise concerns about potential cannibalization effects, fearing that such exposure might reduce revenue from streaming services like Spotify. In this paper, we evaluate the effect of a song's presence - or absence - on social media on its demand on music streaming services using a quasi-natural experiment: Universal Music Group's (UMG) - one of "The Big 3" record labels - decision to remove its entire content library from TikTok in February 2024. We use representative samples covering close to 50% of the US and 94% of the German streaming markets, employing a difference-in-differences approach to compare the streaming consumption of songs that were removed from TikTok with those that were not. We find that UMG's removal of music from TikTok led to a 2-3% increase in streams on audio platforms for affected songs, indicating substitution effects. However, this average treatment effect masks significant heterogeneity: older songs and songs with less promotional support elsewhere saw a decrease in streaming consumption, suggesting that TikTok helps consumers discover or rediscover content that is not top of mind for consumers.
FB11
Meeting Room C4.1
Machine Learning for Personalization and Marketing Measurement
Invited Session
Special Sessions
Chair: Yuyan Wang, Stanford University, Stanford, CA, United States
1 - Personalized Algorithms and the Virtue of Learning Things the Hard Way
Omid Rafieian, Cornell University, New York, NY, United States, Si ZUO
Personalized recommendation systems are now an integral part of the digital ecosystem. However, the increased dependence of users on these personalized algorithms has heightened concerns among consumer protection advocates and regulators. Past studies have documented various threats personalization algorithms pose to different aspects of consumer welfare, through violating consumer privacy, unfair allocation of resources, or creating filter bubbles that can lead to increased political polarization. In this work, we bring a consumer learning perspective to this problem and examine whether personalized recommendation systems hinder consumers' ability to learn their own preferences. We develop a utility framework where consumers learn their preference parameters in the presence of a recommendation system. We introduce a notion of regret, which is defined as the regret when consumers make decisions on their own. We theoretically show that the presence of the recommendation system acts as a barrier to consumer learning. We then empirically investigate this phenomenon using the MovieLens data. We show that consumers who rely more on personalized recommendations make worse decisions in the absence of recommendation systems. Finally, we discuss a variety of consumer protection policies that help improve consumer learning and document the welfare implications of each.
2 - Interpretable Recommender System and Decision Support Using Geometric Deep Learning
Yan Leng, University of Texas at Austin, Austin, TX, United States, Xiao Liu, Rodrigo Ruiz
We introduce a novel and interpretable algorithm for recommender systems (RSs), capable of integrating user and business network data effectively. Incorporating network data brings two challenges. First, network connections are heterogeneous, but network data typically reveal only the existence of these connections. Hence, identifying important connections and eliminating less significant, or “noisy,” links is crucial. Second, while complex deep learning (DL) models can merge network data with traditional data sources, their black-box nature tends to impede user trust and acceptance. Hence, ensuring model interpretability without compromising accuracy presents a significant concern. Addressing these issues, we propose the Multi-Graph Graph Attention Network (MG-GAT) model, a geometric deep learning approach with two primary components. The first component is the neighbor importance graph, a feature that uses an attention mechanism to weed out noisy network connections. The second component is feature relevance, which offers insights into the prediction-making process behind the recommendations, thereby enhancing their interpretability. When benchmarked against leading black-box deep learning models, the MG-GAT model exhibits competitive performance in recommendation tasks, as evidenced by results from two extensive Yelp datasets and four conventional recommendation datasets. A controlled experiment further corroborates that the explanations the RSs provide based on MG-GAT bolster user acceptance and trust in the recommendations. Furthermore, we demonstrate the effectiveness of MG-GAT as an interpretable decision support tool, particularly useful for user search, business targeting, and platform data acquisition.
3 - Unified Marketing Measurement: How to Fuse Experimental Data with Marketing Mix Data?
Nicolas Padilla, London Business School, London, United Kingdom, Ryan Dew

Digital marketers are struggling to measure campaign effectiveness due to the loss of customer-level tracking, rendering multi-touch attribution models obsolete. Moreover, constantly running experiments may be a costly alternative if effectiveness changes over time. As a consequence, firms have turned to using classic measurement tools like media mix models, which have always been built on potentially endogenous aggregate measures of campaign spend and performance.

We propose a unified marketing measurement (UMM) framework that allows us to measure time-varying marketing effectiveness. Methodologically, we use a modern Bayesian nonparametrics framework that fuses the (available) experiments with aggregate media-mix data and leverages the exogenous variation in experiments to de-bias a media mix model. Using Gaussian Processes, our model regularizes ad effectiveness over time smoothly, allowing the experiments to separate marketing effectiveness from the correlation between sales and ad spending close to the experiment.

Our UMM framework also provides uncertainty quantification on ad effectiveness, which can be leveraged to determine if further experiments are needed. Using a series of simulations, we show the conditions for properly inferring ad effectiveness over time. We further show that endogeneity bias in observational data induces higher posterior uncertainty on the effectiveness and structural correlation estimates, which does not decrease with more observational data. This means we can use posterior uncertainty quantification to diagnose when additional experiments are needed.

We showcase our model using data from a large retailer that includes sales and marketing spend at the aggregate level, and previous A/B tests conducted by the retailer.

4 - Data Augmentation Strategies for E-Commerce Personalization
Alex Miller, USC Marshall School of Business, Los Angeles, CA, United States
In this project, we explore ways to leverage modern AI/ML-based methods for improving e-commerce personalization. We study the value of Large Language Model (LLM) embeddings and generative neural network methods for enhancing traditional techniques in personalized marketing, specifically investigating their potential for augmenting tabular data in customer targeting models. The research studies the effect of these methods on the performance and learning rates of targeting models in the online retail context. Using offline, off-policy evaluation with historical A/B testing data, we conduct a systematic study to identify optimal augmentation techniques that offer improved accuracy while maintaining cost and computational efficiency.
5 - Optimizing Consumer Experience by Leveraging Behavior Insights: An Intent-Based Recommendation Framework
Yuyan Wang, Stanford University, Stanford, CA, United States, Cheenar Banerjee, Samer Chucri, Minmin Chen

Recommender systems overly focusing on short-term engagement can inadvertently hurt long-term consumer experience. In response to this challenge, advanced sequence modeling techniques and reinforcement learning have emerged as promising directions. However, these approaches still rely on modeling and optimizing single-step \emph{item}-level interactions as the short-term focused systems do: They predict the next item to be consumed based on a sequence of past interactions using black-box machine learning (ML) models.

In this work, we show the benefits of going beyond item-level black-box predictions and incorporating a higher level and more abstract understanding of consumers. We propose to align the recommender systems better with the consumer decision-making process through incorporating consumer behavior insights into the design of the ML solutions. Consumer behaviors are fundamentally driven by their underlying intents. Consequently, understanding and modeling these consumer intents are crucial for the ML recommender systems to optimize their long-term experience. To this end, we develop an intent-aware recommendation framework that predicts consumer intent in real-time and tailors recommendations accordingly.

We conducted extensive analyses in simulated environments and live experiments, and found that when the consumer’s novelty-seeking intent is incorporated into our intent-based recommendation framework, it is able to provide better contextualized recommendations based on their underlying intent. In particular, when consumers have a higher propensity of novelty-seeking, the framework recommends more high-quality novel content that the consumer is likely to engage with. Our work proposes and validates a principled framework to incorporate consumer behavior insights in the design of machine learning systems.

FB12
Meeting Room C4.2
Customer Journey 2
Contributed Session
Chair: Juliana Sanchez Ramirez
1 - Consideration: An Integrative Rethink for a Rapidly Transforming Marketplace
Pallavi Jha, Georgia State University, Atlanta, GA, United States, R Venkatesh, Rabikar Chatterjee

Product consideration and consideration set formation have been central to our understanding of consumer behavior since at least Howard and Sheth (1969). Whereas quantitative and behavioral articles pre-2000 have provided the bedrock for our analysis of consideration and its consequences, the emergence of online commerce, digital products, wired consumers and big data methods over the past two decades have dramatically impacted the dynamics of consideration set formation as well as its analysis. Even more profound changes tied to the platform revolution, IoT, decision making tools, and generative AI will arguably redefine product consideration and its implications for managers in fundamental ways. Against this backdrop, our study addresses two research questions:

  • Based on an integrative review and synthesis of the extant literature, what foundational elements of consideration and their shifts can be delineated, in terms of their theoretical, methodological and managerial implications?
  • Extrapolating from our review and synthesis, how are the upcoming marketplace transformations likely to further redefine the meaning and significance of consideration as well as its antecedents and consequences?

Our study is founded on a comprehensive review of 40 years of academic research in marketing and allied disciplines, in-depth interviews with leading business executives on how they foresee consideration-driven opportunities and challenges, and sophisticated analyses of the textual dataset we have assembled. These reveal that product consideration as a topic of study is very dynamic, the field is ripe for fresh research, and the most interesting research problems are arguably ahead of us.

2 - Modeling Customer Journey with AI
Zipei Lu, University of Maryland, College Park, MD, United States, P. K. Kannan
When observing a sequence of customer interactions, it is important for a firm to understand how these interactions relate to key objectives -- qualified customer leads, customer conversion events, or churn. Inspired by recent development in Large Language Models (LLM), this research proposes a transformer-based framework that models the sequence of customer interactions in a way that resembles modeling a sequence of words in a sentence. Our proposed model accounts for the varying effects of an interaction when it occurs at different times and the complex interactions between the various touchpoints on other touchpoints and events. We present an application in the multi-channel marketing context where the firm observes sequences of customer-firm encounters across multiple touchpoints. We show that our model successfully predicts the next touchpoint and the evolvement of conversion rate over time at the individual level, as well as the contribution of each touchpoint to the conversion event. By unraveling the patterns within rich customer interaction data, our model provides a useful tool to help managers better understand and manage the customer journey.
3 - Customization and the Customer Journey: Lasting Effects on Retailer Loyalty
Suh Yeon Kim, Georgetown University, Washington, DC, United States, Rebecca Hamilton, Tongil TI Kim, Michael Lewis

Retailers frequently offer customers the opportunity to customize products in addition to buying ready-made products. How does the experience of customizing a product—contributing to the design of the product by choosing features and options—affect the customer’s loyalty to the retailer and future purchases? Using customer transaction data from a retailer, we measure the immediate and lasting effects of customizing products on retailer loyalty. Next, we conduct a series of longitudinal experimental studies to disentangle the effects of the customer’s inherent preference for customizing from their experience with customization. We find that retailers can shape preferences for customizing products across stages of the customer’s journey, offering opportunities to increase both satisfaction and loyalty to the retailer.

4 - Enhancing Journey Insights through the Integration of Usage Pattern Modeling
Juliana Sanchez Ramirez, IESEG School of Management, Lille, France, Kristof Coussement, Arno De-Caigny, Dries Benoit, Lauren Waardenburg
With the evolution of big data and customer analytics, businesses are storing a diverse set of data sources resulting from multiple touchpoints, including marketing interactions, support, training, transaction history, and usage. Integrating this information into a holistic view of the customer journey provides insights into consumer decisions, allowing analysis of short and long-term impacts of marketing interactions and guiding program designs for the entire customer lifecycle.

In this project involving real-life use case, we first explore the integration of different data sources as part of the customer journey (CJ). Second, we investigate the integration of usage data through the CJ to identify different usage patterns. Finally, we evaluate the impact of these usage patterns on business-relevant metrics such as customer lifetime value (CLV) and churn rate.

This research contributes to the literature on the exploration of usage data in a network-based setting. It aims to capture relevant aspects such as the temporal structure of usage and the integration of heterogeneous inputs. Our research has the potential to provide a transferable methodology capable of capturing and incorporating usage patterns in different applications of business analytics models, enhancing their adaptability and effectiveness across diverse contexts.
FB13
Meeting Room C4.3
Advertising: Measuring Effectiveness
Contributed Session
Chair: Shuge Li, Unit 602 21 Marquet St, 2138
1 - Native Advertising Effectiveness Under in Newsfeed Apps
Mengzhou Zhuang, The University of Hong Kong, Hong Kong, Hong Kong, Er Fang

Although newsfeed advertising is increasingly popular among marketers, it has received limited attention in the marketing literature. We are interested in three related research questions: (1) What are the distinct states of consumer newsfeed browsing behaviors? (2) How do consumers in each behavioral state react to newsfeed ads? (3) How do newsfeed ad features affect reactions to ads in each of the behavior states? This research reveals the effects of two newsfeed ad features—relevance (the consistency between the post and consumers’ interests) and organicity (the extent the post looks like organic posts)—on two advertising outcomes: the click-through rate and dwell time. We use a hidden Markov model to model the consumer’s sequential clicking and reading decisions while accounting for the endogeneity of newsfeed selection. We apply our framework to unique clickstream data with more than 395,000 newsfeed posts from a leading newsfeed app. We identify three consumer states of browsing behaviors: Title skimming, Variety seeking, and Focus reading, and we show that the effects of ad relevance and organicity on clicking and reading decisions vary with the behavior states. Finally, our post-hoc simulation shows that feeding ads that are optimal for each consumer’s behavior states can increase the click-through rate by roughly 26% to 170% and the dwell time by roughly 7% to 50%.

翻译

搜索

复制

2 - Throw a Sprat to Catch a Whale? An Empirical Study on Automobile Product Placement in TV Shows
Lingyu Lin, Nanyang Technological University, Singapore, Singapore, Yupeng Chen, Junhong Chu, Yanlai Chu, Siliang Tong

Product placement, where products are featured in video productions (e.g., shows and movies), allows firms to promote their products subtly and has been widely adopted in practice. Existing empirical studies have mainly explored the positive impact of product placement on sales of consumable products, such as cigarettes. However, this may not the case in the category of high-involvement product, which is durable, expensive and may carry a higher risk to the purchase decision. Therefore, despite of substantial progress made by the research regarding the causal impact of product placement in real-world settings, an important question persists: Can product placement impact consumers’ purchasing behaviors in high-involvement product categories like automobiles? Our research aims to address this gap.

We examine the impact of product placement by analyzing the sales change of car models exclusively placed in all Chinese TV shows from 2017 to 2019. We apply an advanced method for the causal identification, the synthetic difference-in-differences (SDID) method of Arkhangelsky et al. (2021), which combines the attractive feature of the widely used difference-in-differences and synthetic control methods. Our findings reveal that automobile product placement in TV shows has a significant positive effect on its sales. Additionally, we observe a spillover effect: the placement of a particular car model positively impacts the sales of other models within the same brand, especially those with similar prices. To the best of our knowledge, this study is the first to investigate the causal impact of product placement on the sales of a high-involvement product (i.e., automobiles).

3 - Strategic Multi-Media and Cross-Channel Synergies During Economic Fluctuations
Shameek Sinha, University of Auckland Business School, University of Auckland, New Zealand
We investigate the role of consumer-oriented media choice and channel selection on purchase intentions in retail markets across categories, how this influence changes from a stable economic situation to a recessionary period and what are its repercussions on marketers’ strategic media communication and channel selling. We build an analytical framework drawing on the behavioral economics literature where consumers' media choice and channel selection decisions are based on a mental accounting model which considers the tradeoff between alternative sources of information available from media communication and selling channels. This allows consumers to optimally decide on the possible synergies or lack thereof in making choices. This decision making process is continuously updated as the consumers encounter fluctuating macroeconomic environment, not only because of their own behavioral adaptations, but also based on the changing resource allocations by retail marketers. We use survey data from the US for the 2008-2011 period to investigate the individual and combined effects between conventional and contemporary media and regular, online and discount channels across essentials and durables. We find that combined media and channel synergies are positive during stability; however during recessions, influence of individual media or channel is stronger than the combined effect, which sometimes has negative synergy. Though synergy between media consumption and channel shopping patterns can be leveraged during stability, during recessions, marketing strategy should be tweaked to focus on individual media or channel. Marketers can choose between conventional and contemporary media and direct their targeting efforts across the preferred regular, discount and online channels.
4 - The Effect of Femvertising on Consumption: A Case Study of Always Campaign
Shuge Li, UNSW Sydney, Sydney, Australia, SunAh Kim, Maggie Chuoyan Dong

Consumers express increasing interest in interacting with brands that are socially responsible. In response to this trend, many brands have adopted ‘femvertising’ - a prevalent advertising strategy that showcases a commitment to women’s empowerment and addressing gender stereotypes. While prior lab studies have documented the positive impact of femvertising, there is a lack of empirical evidence regarding its effect on actual market consumption. In this paper, we quantify the impact of a brand’s adoption of femvertising. In June 2014, the brand Always released a video that challenged stereotypes associated with the phrase “like a girl”, sparking a wider discussion on gender equality and women empowerment. Using market purchase data, we examine the effect of femvertising campaign on market demand. We further explore heterogeneity in brand choice to understand which households are most responsive to femvertising.

Keyword: femvertising, quasi-experiment, gender stereotyping

FB14
Meeting Room C4.4
Effective Influencer Marketing I: Seeding Campaigns
Invited Session
Special Sessions
Chair: Maximilian Beichert, HEC Paris, Paris, France
Co-Chair: Andreas Lanz, University of Basel, Jouy-en-Josas, France
1 - CANCELLED Spreading Seeds: Using Community Detection to Optimize Viral Marketing Campaigns
Ralf Van der Lans, Hong Kong University of Science & Technology-HKUST, Kowloon, Hong Kong, Dingli Wu, Xi Chen

Companies increasingly adopt viral marketing on social media where consumers share advertising messages to their friends. Because most firms have limited budgets to initiate viral marketing campaigns, the seeding decision – selecting a small set of influential people – is critical for its success. The goal of the seeding strategy is to select those people that maximize the reach of the campaign. However, this optimization problem is computationally challenging, as optimal solutions cannot be computed in polynomial times. As a consequence, previous research has used heuristics to seed customers based on individual characteristics or the position of individuals in the social network or a combination of both. A limitation of these approaches is that they consider the selection of a seed independent of other the selection of other seeds. This may be inefficient, especially if seeds share similar network connections. To address the potential overlap of friends of seeds, we propose to spread seeds in the social network, such that each seed is assigned to a different part in the network. To do so, we propose using community detection algorithms to divide the social network into smaller subnetwork. To validate our methodology, we use agent-based simulations and a field experiment to benchmark our seeding strategy against seeding approaches proposed in previous literature. We find that our proposed community seeding approach not only outperforms all benchmark approaches but that our strategy also targets individuals with fewer friends, which significantly reduces the costs of the seeding campaign.

2 - The Role of Crowd Engagement in Seeding Campaigns
Florian Dost, Brandenburg University of Technology / Alliance Manchester Business School, Cottbus, Germany, Ulrike Phieler

Recent studies suggest that nano-influencers, with their smaller but engaged follower bases, perform better than macro-influencers in influencer marketing campaigns. Seeding campaigns take this concept further by distributing products to thousands of ordinary individuals, encouraging them to spread the word within their personal networks.

Traditional influencer marketing tools are impractical for managing large crowds of seeds. Instead, seeding campaigns focus on identifying a suitable crowd of volunteers, connecting them to the brand, and managing their crowd engagement through a dedicated platform.

While maximizing crowd engagement is an explicit goal of campaign managers, we propose that over-engaging seeds can actually hinder their outreach efforts, reducing campaign reach and sales impact.

Analyzing a dataset of 151 seeding campaigns, representing 700,000 seeds, we uncover an intricate relationship between crowd engagement and campaign performance. Initially, increasing crowd engagement leads to broader reach, but beyond a certain point, it starts to diminish outreach, particularly in the offline communication that dominates personal communication of the seeds.

We also demonstrate that a more engaged crowd affects follower/friend engagement and transmissions, further emphasizing the need to strike a balance between crowd engagement and outreach for optimal campaign performance.

To provide practical guidance, we examine how various campaign design decisions influence crowd engagement and the relationship to outreach. Our findings suggest that sending extra product samples, inviting more homogeneous crowds, or running larger campaigns can shift the optimal engagement levels, offering campaign managers valuable optimization strategies.

3 - Creator Incentive: The Dual Aspects of Brand Collaboration
Bindan Zhang, Kellogg School of Management, Northwestern University, Evanston, IL, United States

The generation of platform content is essential for platform growth and competition. Typically, platforms incentivize participation and content creation through monetary rewards. Despite the explosive growth of social media platforms, there is little research on the effectiveness of creator incentives. I collect a novel panel dataset of creators with detailed video information from one of the largest video-sharing platforms in China. I examine the impact of a special type of incentive, namely the launch of a brand-creator matching platform that facilitates brand-creator collaborations on creator behavior and audience engagement. To account for self-selection, I combine a difference-in-differences approach with propensity score matching. I find that the launch of the brand-creator matching platform leads to an increase of 25% in creator productivity, but this increase does not lead to an increase in better audience responses measured by audience engagements. Furthermore, the increase in productivity is asymmetric across content categories, with a positive association with the total number of videos with brand collaborations in each category. By taking random samples of videos with and without branded content from the same creator and controlling for both observed and unobserved characteristics, I also find that inserting branded content leads to a decrease in audience engagement. Furthermore, well-established creators with more followers suffer more from the negative impact of incorporating brand-name content. This paper contributes to the literature on platform incentives and influencer marketing and provides insight into maintaining a long-term healthy ecosystem for creators, audiences, brands, and social media platforms.

FB15
Meeting Room C4.5
Marketing Application of Unstructured Data
Invited Session
Special Sessions
Chair: Qiang Lu, University of Sydney, Belfield, Australia
Co-Chair: Jian Ni, Virginia Tech, Blacksburg, VA, United States
1 - Skin Color Congruence in Access-Based Consumption: Evidence from Fashion Rentals
Xiaohua Zeng, Peking University HSBC Business School, Peking, China, People's Republic of, Chuang Tang, Liyuan Wei
The demographics in the U.S. have become more racially diverse and more multiracial, posing new questions for marketing scholars and practitioners about interracial marketplace interactions, especially when consumers must share products on access-based platforms. Drawing on the racial congruence literature, this research uses a large dataset of renter selfies posted alongside their reviews on a fashion rental website to examine empirically conspicuous phenotypical characteristics such as skin color affect subsequent consumers’ rental decisions. To obtain renter skin color, we used the facial recognition technology on Baidu AI open platform (https://ai.baidu.com/tech/face). Baidu’s algorithm first detects human faces in a photo and measures the skin color based on visible facial skin. Baidu facial recognition technology uses deep convolutional neural networksand itis trained on 1.2 million images of celebrities collected from websites. Baidu’s approach can extract information from image data with over 99% accuracy rate of labelled face in the wild (Chaudhuri 2020; Liu et al. 2015). This technology has been widely used in industry to streamline airport boarding and to provide new experience at fast-food chains. Baidu face recognition program measures the skin color on a 6-point scale, with 1 being the lightest and 6 the darkest. We find strong and robust skin color-based congruence effects where renters are more likely to follow previous renters of similar skin colors. We further test underlying mechanisms and find that product contagion, social standards of fashion,and information provision explain the congruence effect. Our findings provide guidance to online fashion rental platforms on socially responsible design.
2 - Dynamic Fatigue Management in Customer Service: Unveiling Vocal Biomarkers for Enhanced Customer Satisfaction
Ye Zheng, University of Sydney, Sydney, Australia, Qiang Lu, Jiang Qian, Jake An
Voice interaction remains a pivotal mode of communication with customers, significantly contributing to the proliferation of call centres. Within these dynamic environments, representatives encounter a spectrum of customer interactions, leading to varying degrees of fatigue throughout their shifts.

Grounded in the Conservation of Resources Theory (Hobfoll, S.E., 2001), this paper posits that fatigue engenders a perceived depletion of personal energy resources among representatives. Consequently, this perception catalyses a strategic conservation of remaining resources, manifesting as diminished effort in work tasks to mitigate further resource depletion. This conservation behaviour is hypothesized to result in a decline in work efficiency, quality, and, critically, customer satisfaction. Accordingly, the paper suggests that tasks of higher complexity should be strategically assigned to less fatigued representatives to uphold service quality.

Employing innovative machine learning techniques, this study utilizes vocal biomarkers to dynamically monitor the fatigue levels of call representatives throughout their shifts. The primary objective of this research is to elucidate the temporal patterns of agent fatigue and its subsequent ramifications on customer satisfaction.

By analysing call centre recording data from an Australian financial investment firm, this investigation illuminates novel strategies for task allocation based on fatigue levels, aiming to avert workload saturation and, most importantly, enhance customer satisfaction. This paper contributes to the nascent body of literature on dynamic fatigue management and its impact on service delivery, offering valuable insights for optimizing operational efficiency and customer engagement in call centre settings.
3 - Match of the Titans: The Sales Implications of Seller-Customer Communications in Online Marketplaces
Jordan Truong, Melbourne University, Melbourne, Australia, Stephan Ludwig, Liliana Bove, Dennis Herhausen, Dhruv Grewel
In online marketplaces, product descriptions provided by sellers and reviews offered by customers both influence sales. Sellers can control their own product descriptions and images, which they might seek to align with the content of customers’ reviews. With longitudinal and secondary data, including 1,926 seller descriptions and 199,236 customer reviews over four years, the authors leverage automated text and image analyses to determine that the impact of embedded value (features/attributes) and value-in-use (usage outcomes) foci in seller descriptions depends on how closely they match the respective foci in customer reviews. According to polynomial regressions, a greater match on different product value foci exerts positive effects on sales, which diminish over time. The number of reviews and product price also affect this relationship. For products with fewer reviews or relatively low prices, closely matching a review focus on embedded value increases sales. For products with more reviews or relatively high prices, matching the review focus on value-in-use exerts a greater impact. These findings show sellers how to adapt their product descriptions and imagery strategically, to align effectively with customer reviews, the number of reviews, and product prices in online marketplaces. Further, we develop LexoCraft (https://lexocraft.my.canva.site/)—a webapp designed to help online sellers adapt their descriptions effectively with customer reviews. The app generates a 3-D visualization illustrating sellers’ current performance in comparison to the optimal outcome, and offers tailored recommendations to enhance how they convey the value of their products, which in turn maximize their sales potential.
4 - Does Your Money Follow Your Ideology? An Empirical Analysis of Online Crowdfunding
Tao He, Virginia Tech, Blacksburg, VA, United States, Jian Ni, Paul Herr
Crowdfunding platforms emerge as a crucial alternative financing option for entrepreneurship and even some political fundraising. We study how the political leanings of investors may affect their project choices and try to quantify the extent of political preferences among investors and assess their impact on project financing. We explore whether investors' political beliefs align with their funding patterns, aiming to reveal patterns and degrees of influence. Built on the NLP and causal inference literature, we construct a (pseudo-)experimental setting of crowdfunding projects using the projects descriptions, and then investigate the effects in a matching framework and explore the possible heterogenous effects using the random forest approach. Our research illuminates how political biases shape investment behaviors on crowdfunding platforms, offering valuable insights for investors, project initiators, and policymakers, and enriching our understanding of crowdfunding market dynamics.
FB16
Meeting Room C4.6
AI in Healthcare
Invited Session
Special Sessions
Chair: Meng Zhu, Virginia Tech, Baltimore, MD, United States
1 - Artificial Intelligence for Mental versus Physical Health
Haiyang Yang, Johns Hopkins University, Baltimore, MD, United States
Findings of a series of studies suggest that, for a mental health issue versus a physical health issue of similar significance, people can be more averse to using artificial intelligence for treating the former. This research explores the underlying mechanism and discusses the implications.
2 - Algorithmic Bias and Physician Liability
Shubhranshu Singh, Johns Hopkins University, Baltimore, MD, United States
AI algorithms are increasingly used to help support clinical decision making. Their potential bias, as reflected in disparities in accuracy across patient populations, has received growing attention. The U.S. Department of Health and Human Services (HHS) recently proposed an anti-discrimination liability rule, which states that providers may be liable for erroneous medical decisions made in reliance on biased algorithms. In this paper, we model and analyze how such anti-discrimination liability rules affect both upstream AI development decisions (by an AI developer) and downstream AI deployment decisions (by a physician). We derive several results with important policy implications.
3 - Health Inequity Implications of Digital Health Interventions (DHI) Adoption
Sakshi Singh, Marriotts Ridge High School, Marriotsville, Maryland, United States
This paper examines the tradeoffs and incentives that influence the provision and acceptance of digital health interventions. We develop a theoretical framework to study how various considerations of policymakers, service providers, and consumers interact to determine the resulting health inequity.
FB17
Meeting Room C4.7
Antecedents of Firm ESG (Environmental. Social, and Governance) Performance
Contributed Session
Chair: Tao Luo, Sichuan University
1 - Does Sibling Management affect ESG Performance of Family Firms? Evidence from China
Qingmei Tan, Tianjin University, Tianjin, China, People's Republic of, Yani Sun, Peixuan Geng, Can Qu

Based on the perspective of socioemotional wealth (SEW) and agency theory, this study explores the effect of sibling management on ESG performance of family firms. This study also investigates the moderating roles of top management team (TMT) faultlines and pyramidal layers. Using the sample of Chinese listed family firms from 2010 to 2020, this study finds that sibling involvement in management significantly promotes ESG performance of family firms. Both increased TMT faultlines and pyramidal layers weaken the positive effect of sibling management on ESG performance of family firms. These results are robust considering endogeneity and a series of robustness tests. Heterogeneous analysis indicates that the positive effect of sibling management on ESG performance is more pronounced in family firms more deeply influenced by Confucian culture and Clan culture, and in family firms located in central and western regions of China. Our results demonstrate the significance of kinship heterogeneity of family managers, TMT faultlines and pyramidal layers on family firms’ ESG strategies.

2 - AI in the Hands of Good or Bad: Firm’s ESG Performances and Inferred Corporate Social (Ir)Responsibility Leads to Divergent Consumer Reaction to Firm AI Adoption
TaeWoo Kim, University of Technology Sydney, Sydney, Australia, Suyun Mah, Michelle Kim, HyeJin Lee, SunAh Kim

Emphasis on AI has increasingly become a common trend in various industries. Our research examined how consumers respond to firms’ tweets announcing increased focus on AI in their businesses. We analysed 3,214 corporate tweets from S&P 500 companies and the valence of consumer tweets about the firms before and after the announcements. In this analysis, we adopted a compartmentalized approach regarding corporate social (ir)responsibility and sustainability performance by dividing them into three independent sections: “Environment,” “Social,” and “Governance” performance. The most salient finding was observed in the Governance domain characterized by business ethics (e.g., corruption, lack of privacy protection): low performance in the Governance domain led to strong negative reactions among the consumers regarding the corporate’s AI adoption. Corporate performance in the Social and Environment domains produced mixed results. Follow-up experiments revealed that Governance performance translates into the perception of good or bad corporate intentions, leading to different levels of support for the corporate AI adoption. These findings show the public sentiment regarding AI that technology is neutral, and it is important who uses the technology for what purposes. Implications on the relationship between corporate social (ir)responsibility and technology utilization are discussed.

3 - Does Social Responsibility Reform Curb Corporate Greenwashing: Evidence from a Quasi‑Natural Experiment in China
Tao Luo, Sichuan University, Chengdu, China, People's Republic of

Greenwashing has significantly impeded the advancement of environmental, social, and governance (ESG) practices and undermined the rights of stakeholders. To better fulfill the social responsibility of corporations, the Chinese government spearheaded reforms among state-owned enterprises in 2016. Using data of A-share listed companies from 2011 to 2021, this study uses a difference-in-differences (DID) model to investigate the effects of the social responsibility reform on corporate greenwashing. We find that social responsibility reform significantly restrains corporate greenwashing, and this effect is heterogeneous by firm characteristics. Further analyses indicate that the improvement in the propensity of environmental performance disclosure by corporations and the moderation of regulatory inquiries by the CSRC are important channels to limit corporate greenwashing. Overall, our findings provide new insights into reducing corporate greenwash behavior and offer microscopic evidence for the objective evaluation of the effects of China’s social responsibility reform.

FB18
Meeting Room C4.8
Platforms: Content, Consumer Search, and Market Power
Invited Session
Special Sessions
Chair: Olivia Natan, UC Berkeley, Berkeley, CA, United States
Co-Chair: H. Tai Lam, UCLA Anderson, Los Angeles, CA, United States
1 - Consumer Inferences from Search Rankings
Olivia Natan, University of California, Berkeley, CA, United States, Jessica Fong, Ranmit Pantle

Consumers' responses to information presentation order, especially in ranked lists (e.g. search results), paginated displays, or physical spaces (e.g. store shelves), are well-documented. However, the source of such ranking effects is under-explored. Consumers may draw inferences from the order of items. In online settings, where ranking algorithms commonly reflect aggregate tastes (e.g. popularity based rankings) and individual heterogeneity (e.g. personalized rankings), consumers may infer that higher-ranked alternatives are higher quality or better match their tastes. However, these position effects need not exclusively reflect equilibrium beliefs about how rankings are formed. Instead, consumers may have differential costs to search or attend to items based on their ordering or location. We develop an experimental paradigm, combined with a new structural model of search behavior, to understand the source of ranking effects in consumer search contexts. We propose a model of search that separately accounts for search costs and consumers’ beliefs about product quality. We use data generated from our experiment to estimate this search model. We mitigate consumers’ inferences about product quality by informing some experimental subjects that the products are randomly ordered, which produces baseline cost estimates. We find that consumers learn slowly about changes to rankings. The source of position/rank effects is critical for understanding the welfare implications of alternative information design strategies and regulations. Evaluating the impacts of platform steering requires understanding whether consumers draw the correct inferences about top ranked products or simply search them due to frictions in the search process.

2 - Ad-funded Attention Markets and Antitrust: Youtube Content Market
H. Tai Lam, UCLA Anderson, Los Angeles, CA, United States
Many valuable digital products (e.g., Youtube, Google Maps, LLMs) are provided at no charge. But they are not "free" nor costless to supply, rather the "price" consumers pay is with attention to advertising, which funds their production. The importance of ads for enabling attention product markets are not well understood and elasticities important for welfare lack study. Despite this, the Department of Justice is suing Google to improve competition in the $224b upstream advertising market, with unclear effects on downstream consumers. Lower ad prices may reduce production and increase ad load, but ad quality improvements may benefit consumers. Using data on Youtube's $29b market, I apply RDDs and IVs to derive causal estimates of attention (price) elasticities of demand and supply to inform the welfare consequences of proposed antitrust action. Content creators can choose between platform ads (ad rolls), and disintermediated ads (sponsorships), with an additional ad unit reducing viewership by 22% and 7%, respectively. I use a simple structural model of video consumption and content creation to analyse the effects of potential reduction in ad revenue and improved ad quality to understand consumer welfare under different antitrust outcomes.
3 - The Equilibrium Impact of Generative Artificial Intelligence on Content Markets
Samuel Goldberg, Stanford Graduate School of Business, San Francisco, CA, United States, H. Tai Lam

Generative artificial intelligence (GAI) promises to greatly change the economics of content creation by reducing both the costs, and barriers to entry, of producing content. Still, it remains unclear how buyers perceive AI generated content, and which types of creators will choose to substitute towards using GAI. For platforms, answers to these questions will influence how different types of content should be priced, labeled, and suggest that the equilibrium level of content quality on the platform may shift. This project aims to better understand the impact of GAI on marketplaces for creative content and inform both the economics of how AI and workers collaborate, as well as platform policy. A first order question is do consumers substitute between AI generated content and traditional content and how does this vary across markets? If the two types of content are strong substitutes, and differences in WTP are small, we might expect the majority of content creators to adopt GAI. On the other hand, if WTP differs substantially there will likely be heterogeneity in which creators adopt. This raises a second set of questions: which creators adopt GAI? GAI likely reduces the costs of producing content. Do new creators enter, having adopted the new technology, or do existing high cost creators adopt? How does this vary with market competitiveness? Using data on hundreds of thousands of images on the Adobe Stock Images marketplace, we examine the effects of a late 2022 policy change allowing GAI images to be sold in all image markets.


4 - Does Information Mitigate Behavioral Gaps Due to Market Inexperience
Uyen Tran, Arizona State University, Phoenix, Arizona, United States
Standard search models assume consumers know the price distribution, so providing them with price information does not change their behavior. We evaluate this assumption using detailed individual-level purchasing and search data on 2,449 consumers in the UNC-Chapel Hill textbook market. We treat a random subset with information about the price of their assigned textbook at various retailers. The information treatment: leads to more online search and purchases; reduces the gap in online search behavior between consumers across experience levels; and increases take-up of the textbook at the extensive margin. Inexperienced consumers do not update their beliefs about price dispersion.
FB19
Meeting Room C4.9
Consumer Behavior: Social Influences and WOM 2
Contributed Session
Chair: Ying Yue, Communication University of China
1 - Consumers' Tendency to Reciprocate Online Help Is Contingent on Helpers' Name Disclosure
Yunqing Chen, Xiamen University, Xiamen , China, People's Republic of, Leilei Gao

In offline helping interactions, help-recipients typically feel motivated to reciprocate the kindness they have received, whether directly to the helper or indirectly, by helping others. This tendency facilitates future helping behavior and has clear potentials to benefit organizations and websites that seek to encourage helping or collaboration among users. Yet given that reciprocity is socially motivated, it is unclear whether it arises in online settings, where many users offer help anonymously. The present research shows that, if a help-provider discloses their name in online communications, the recipient is more likely to repay the favor by helping others. The effect occurs because name-disclosure makes the helper a real individual in the mind of the recipient, which in turn boosts the recipient’s motivation to reciprocate, thereby encouraging them to engage in helping behaviors. Seven preregistered studies and an analysis of field data robustly support the relationship between helper name-disclosure and the recipient’s helping behavior, as well as for the proposed underlying mechanism. The research contributes to a deeper understanding of the dynamics of online reciprocity and user engagement in digital communities.

2 - How Social Information Exposure Affects user Click Intention in Social Media: The Role of Emotional Value and Social Value
Fenghua Wang, School of Business and Management, Key Laboratory of Brain-Machine Intelligence for Information Behavior (Ministry of Education and Shanghai), Shanghai International Studies University, Shanghai, China, People's Republic of, Mohan Wang, Mengting Wang, Xiangbo Kong

ABSTRACT With the booming development of social media, the scale of social users is becoming even larger. Many platforms have realized the important role of social information in influencing potential users' decisions and have exposed social information on their pages to attract potential users. In this paper, we investigate the impact of social information exposure on user decision-making, the role of emotional value and social value play in-between, as well as the moderating effects of content type. We conducted three behavioral experiments to test our hypothesis. First, exposure to peer browsing information significantly influences user click intention. User click intention is significantly higher when exposed (vs. not exposed) to peer browsing information. Further, user click intention is significantly higher when peer identity is exposed (vs. not exposed). Second, Social value and emotional value play a positive and significant mediating role in the relationship between peer browsing information and user click intention. Additionally, content type plays a moderating role in the influence of peer browsing information on click intention through the emotional value. Finally, theoretical and practical implications are also discussed.

KEYWORDS: information exposure; peer influence; user click intention; social media

Authors: MohanWang,Mengting Wang,Xiangbo Kong, Fenghua Wang*(Corresponding and presenting author)

3 - Exploring the Dynamics of Shopping Companions on Consumer Behavior in Retail Environments
Atanu Adhikari, Indian Institute of Management Kozhikode, Kozhikode, India, Dev Prasad Murthy

Purchasing goods is a common social activity frequently done with friends or family. Previous research emphasizes the significant role shopping partners play in shaping various aspects of the shopping experience. This research explores the social and psychological mechanisms underlying the effects of shopping companions on mall patrons. Specifically, it investigates normative influence on arousal, attentiveness, apprehension, and intention to purchase during shopping.

We examine how factors such as gender and social identity (friend or family member) influence these crucial facets of consumer behavior, drawing on psychosocial theories. While most prior research contrasts the attitudes and actions of solitary shoppers with those of shoppers with companions, our investigation attempts to comprehend the complex implications of various features of shopping companions. Our theory posits that having a shopping companion activates normative values, subsequently influencing consumer behavior.

Our study addresses three essential areas, contributing substantially to the marketing literature. First, we investigate how shopping partners influence arousal, attentiveness, and anxiety when shopping in a store. This information is crucial for retailers as it illuminates differences in the emotional states and attentiveness of accompanied customers. Second, we examine the affective influence that gender combinations have on consumer behavior within shopping groups, allowing retailers to tailor gender-specific sales strategies. Third, we explore how the companion's social relationship with the shopper affects arousal, attentiveness, and apprehension associated with shopping. Our results highlight the selective nature of consumer responses, illustrating how the presence of various genders or social relationships can more effectively sensitize consumers to products.

4 - The Effect of Childhood Socioecomomic Status on Indulgent Consumption
Ying Yue, Communication University of China, Beijing, China, People's Republic of, Chenghu Zhang, Guifeng Meng

Can childhood socioeconomic status (SES) influence the desire for indulgence in adulthood? The present research, spanning four experiments across three domains of indulgence-related decisions and utilizing different methods of measuring childhood SES, sheds light on how individuals from different childhood SES backgrounds choose when faced with a conflict decision between self-control and indulgence. Experiment 1 finds that individuals with high childhood SES are more likely to choose indulgence compared to those with low childhood SES. This effect is replicated in experiment 2 and find to be independent of an individual's current level of SES, which is mediated by differences in pleasure pursuit. In other words, individuals who grew up wealthy are generally more likely to pursue novel experiences and potential rewards from decision-making, thus increasing their choice of indulgences. Experiment 3 finds that this effect is diminished when individuals feel lower deservingness. Experiment 4 rules out "sense of control" as a psychological mechanism for the effect. Overall, this research demonstrates how, why, and when childhood SES influences the desire for indulgence, suggesting that the effects of childhood SES may be etched into our adult psychology, continuing to influence adult consumer decision-making regardless of one's socioeconomic situation later in life. These findings contribute to the growing literature on how consumer behavior is formed and influenced.

Key words: childhood socioeconomic status; indulgence; pleasure pursuit; deservingness

FB20
Meeting Room C4.10
Visual Content: Insights from Eyetracking and Field Experiments
Contributed Session
Chair: Moshe Unger, Tel Aviv University
1 - Postcolonial Acculturation Culture in South Korea: An Analysis of My Neighbor Charles, An Immigrant Reality TV Show
Jiwen Zhang, Dongguk University, Seoul, Korea, Republic of, Yulee Choi, Seungwoo Chun

This study examines how Korea's colonial history and its contemporary role in the global economy influence the acculturation culture in Korea. This dynamic cultural environment, conceptualized as postcolonial acculturation culture, is instrumental in shaping the hybrid identities of both the locals and immigrants, as well as influencing their interactions. This study primarily focuses on the differing portrayals and societal perceptions of immigrants from the Global North and South. This is achieved by analyzing the 44 episodes of the immigrant-focused reality TV show, My Neighbor Charles. Additionally, the study performs text analyses, including topic modeling and sentiment analysis, on 30,809 comments from the show's YouTube pages. This approach reveals distinct aspects of postcolonial acculturation culture in Korea. The findings of both analyses show that Global North immigrants are depicted as educated and successful, while those from the Global South are shown overcoming struggles in Korea. The program emphasizes the importance of English and other colonial languages as social and economic capital, especially for Global North immigrants. Interaction dynamics also differ, with Global North guests often portrayed in reciprocal, culturally appreciative interactions compared to the more transactional interactions of Global South guests. The findings suggest that Koreans generally show favoritism towards immigrants from the Global North, seeking validation for their country's development. Conversely, there are ambivalent attitudes towards those from the Global South, coupled with an expectation to see their efforts toward achieving the Korean Dream. These insights make significant contributions to the fields of acculturation and postcolonial consumer research.

2 - The Impact of AGI Summaries on Video Content Consumption: Insights from A Randomized Field Experiment
Rui Qin, Hong Kong Polytechnic University, Hong Kong, Hong Kong, Yue (Katherine) Feng, Dai Yao

Artificial general intelligence (AGI) has been increasingly adopted in content generation in marketing applications. A unique feature of AGI is its ability to generate condensed and synthetic summaries of content, which is originally of greater lengths and complexity. To understand the effects of AGI summaries on the consumption of original content, we conduct a randomized field experiment on a major video content platform in China. We supplement AGI summaries to the videos in the treatment group at various stages after they are posted, and track various viewer engagement activities of these videos and comparable control videos. Our results reveal that the inclusion of AGI summaries overall enhances video consumption activities such as likes and shares, while we did not observe a significant impact on view counts. Additionally, the effects of AGI summaries are found to be nuanced and plausibly more pronounced on videos that are more difficult to consume, including those that are utilitarian (hence are slightly more complex), posted earlier (hence are less familiar to viewers), and heavily commented (hence are more likely controversial). These findings provide valuable insights into the informative role played by AGI summaries and contribute to the growing body of knowledge surrounding the utilization of AGI in content consumption, which offer important implications for marketers, content creators, and platform providers.

Keywords: AGI summaries, video content consumption, user-generated content, customer engagement, field experiment

3 - Text Message Screen Positions in Video Advertising
Camila Lopez, UNSW Business School, Kensington, Australia, Junbum Kwon, Fang Liu, Harmen Oppewal, Eric Trinh

Video ads are becoming increasingly popular in marketing. Video ads typically display text information in order to effectively communicate their messages. It is therefore critical for marketers to understand how the positions of text messages in video ads affect viewers’ attention. By embedding experimentally designed text messages in real video ads and using eye-tracking, we find that a text message receives a longer fixation duration and a greater share of total attention when it is located just above a focal object. Second, a text message with a hashtag or having more characters draws more attention when it is placed under or to the right of a focal object, respectively. Our findings thus suggest the most effective location for a text message in a video ad. The study also demonstrates a new experimental approach to studying video ads.

Keywords: visual attention, video ads, eye-tracking.
4 - Predicting Consumer Choice from Raw Eye-tracking Data using the RETINA Deep Learning Architechture
Moshe Unger, Tel Aviv University, Tel Aviv, Israel, Michel Wedel, Alexander Tuzhilin
We propose the use of a deep learning architecture, called RETINA, to predict multi-alternative, multi-attribute consumer choice from eye movement data. RETINA directly uses the complete time series of raw eye-tracking data from both eyes as input to state-of-the art Transformer and Metric Learning Deep Learning methods. Using the raw data input eliminates the information loss that may result from first calculating fixations, deriving metrics from the fixations data and analysing those metrics, as has been often done in eye movement research, and allows us to apply Deep Learning to eye tracking data sets of the size commonly encountered in academic and applied research. Using a data set with 112 respondents who made choices among four laptops, we show that the proposed architecture outperforms other state-of-the-art machine learning methods (standard BERT, LSTM, AutoML, logistic regression) calibrated on raw data or fixation data. The analysis of partial time and partial data segments reveals the ability of RETINA to predict choice outcomes well before participants reach a decision. Specifically, we find that using a mere 5 seconds of data, the RETINA architecture achieves a predictive validation accuracy of over 0.7. We provide an assessment of which features of the eye movement data contribute to RETINA's prediction accuracy. We make recommendations on how the proposed deep learning architecture can be used as a basis for future academic research, in particular its application to eye movements collected from front-facing video cameras.
FB21
Meeting Room C4.11
Large Language Models: Applications 1
Contributed Session
Chair: Tianyu Gu, University of Utah, Salt Lake City, UT, United States
1 - Enhancing Human Persuasion With Large Language Models
Minkyu Shin, City University of Hong Kong, Hong Kong, Hong Kong, Jin Kim
Although large language models (LLMs) are reshaping various aspects of human life, our current understanding of their impacts remains somewhat constrained. Here we investigate the impact of LLMs on human communication, in the context of consumer complaints in the financial industry. Employing an AI detection tool on more than 780K complaints gathered by the Consumer Financial Protection Bureau (CFPB), we find evidence of LLM usage in the writing of complaints—shortly after the release of ChatGPT. Our analyses reveal that LLM usage is positively correlated with the likelihood of obtaining desirable outcomes (i.e., offer of relief from financial firms) and suggest that this positive correlation may be partly due to the linguistic features improved by LLMs. We test this conjecture with a preregistered experiment, which reveals results consistent with those from observational studies: Consumer complaints written with ChatGPT for improved linguistic qualities were more likely to receive hypothetical relief offers than the original consumer complaints, demonstrating the LLM’s ability to enhance message persuasiveness in human communication. Being some of the earliest empirical evidence on LLM usage for enhancing persuasion, our results highlight the transformative potential of LLMs in human communication.
2 - Deciphering Customer Interactions with Voice Assistants using LLMs
Ziting Liao, University of Maryland, College Park, MD, United States, Wendy Moe, Liye Ma

In this paper, we examine user interactions with an AI assistant, with the goal of inferring purchasing intent. With the proliferation of ambient computing, firms have more opportunities than ever before to connect with their customers, and users are interacting with their AI assistants to do more than just shop. Here, we text analyze user-initiated interactions and identify features of utterances that predict purchasing intent. Specifically, we build a bipartite network of nouns and verbs and measure the distance of specific words to “golden” purchasing words, such as “purchase,” “buy,” or “order.” We then predict purchasing intent as a function of this distance along with other linguistic features. One challenge of this research is in measuring purchasing intent. We do this by using large language models, specifically Chat-GPT3.5, to annotate our data with a measure of purchasing intent. We validate this method by comparing the results of our analysis with Google Ads CPC, with the assumption that higher CPCs correlate with higher purchasing conversion probabilities. We find that the words used in an utterance can be mapped onto a network graph and effectively predict purchasing intent. Note that the analysis is limited to a single utterance and so no customer tracking across interactions is necessary. Additionally, we validate the ability of large language models to predict purchasing intent.

3 - Artificial Intelligence-Generated Content (AIGC) in Digital Marketing: The Interaction between Content Characteristics and Brand Characteristics
Yang Yang, Beijing Technology and Business University, Beijing, China, People's Republic of, Yuduo Li, Haitao Xiong

As large language models are rapidly gaining popularity, artificial intelligence-generated content (AIGC) receives wide attention both from the academic and the industry. However, extant research did not fully explore AIGC in the domain of marketing. More specifically, AIGC can respond to marketers' requirements automatically and efficiently in digital marketing, while consumers' reactions to AIGC of different brands are not always positive. In Study 1, with 47,151 Weibo marketing real cases from 11,610 bloggers, this paper creates corresponding AIGC cases with ChatGLM. On this basis, we compare the features of human-generated content and AIGC in digital marketing and conduct linguistic analysis (including vocabulary features, topic analysis, and sentiment analysis) to mine content characteristics (i.e. syntactic simplicity and language concreteness) of AIGC with deep learning methods. In Study 2, this paper conducts lab experiments to study the interaction between content characteristics and brand characteristics (i.e. brand uniqueness and brand innovativeness) and test the boundary conditions for AIGC's effect on consumers' purchase intention. The findings can provide effective detection features for practitioners to distinguish AIGC and human-generated content. Besides, the results would figure out when the AIGC has a favourable effect.

4 - Decoding the Language of AIs: A Study on GPT's Craft of Product Descriptions
Tianyu Gu, University of Utah, Salt Lake City, UT, United States, Hongye Sun
This paper investigates the capabilities of Large Language Models (LLMs) like GPT in optimizing product descriptions for both utilitarian and hedonic categories. Across six comprehensive studies, we assess GPT's alignment with human standards in effective product description writing. Our findings reveal GPT's adeptness in enhancing product descriptions, but also unveil nuanced performance variations when following specific versus general optimization directives. Notably, while GPT aligns with human standards in enhancing concreteness under broad directives, it encounters challenges under targeted instructions for hedonic products. This distinction underscores the complex dynamics in AI-driven content optimization, emphasizing the differing impacts of generic and specific guidance on AI tasks. Consistency in assessments by GPT, its rival Bard, and human evaluators further suggests that AI evaluations on product descriptions are not only reliable but also efficient, potentially reducing the need for extensive human reviews. Moreover, we point out limitations in traditional lexicon-based metrics, advocating for more sophisticated tools to capture content nuances. This study enriches the discourse on AI in digital marketing and textual analysis, spotlighting the complex interplay between AI capabilities and human insight in content creation and optimization.
Friday, June 28, 1:00 PM - 2:30 PM
FC01
Meeting Room C2.5
Price Discrimination and Willingness to Pay
Contributed Session
Chair: Jared Wong
1 - Optimal Price Points in Differential / Discriminatory Pricing: Theoretical Model and Empirical Evidence
Katerina Kormusheva, Australian National University, Canberra, OH, Australia

Marketing relevance: This paper seeks to expand current knowledge in the area of discriminatory and differential pricing, a main area of Marketing research. In designing a pricing strategy, one of the key parts of the marketing mix, a customized targeted pricing approach, also called ‘differential pricing’ or ‘discriminatory pricing’ in research, is an approach which has long been applied in the contexts of perishable inventory and service-based industries (hotels, airlines, telecommunications, and others). A key variable to manipulate is the number of different price points for the same product or service, when applying differential pricing strategy. By developing a framework and a model for determining how many price points to implement in differential pricing, our research addresses a gap in existing research. We focus on choosing the levels of differentiation, derive a function form of the model framework proposed, and lastly, test it empirically with data from a large-scale marketing pricing experiment of services in telecommunications.

Keywords: pricing, differential pricing, discriminatory pricing, price points

Contribution: We define a framework for choosing the levels of differentiation in discriminatory pricing, based on the intersection of two main characteristics – demand flexibility and cost of differentiation. Using data from a large-scale marketing pricing experiment in telecommunications, we demonstrate that by applying the proposed differentiation framework for differential pricing, the firm would be able to double the number of customers and bring a 69% improvement in monthly revenue for the firm.

Author and presenting author: Dr. Katerina Kormusheva

2 - Brand-Loyalty-Driven Competitive Price Promotions
Dan Horsky, Rochester University, Rochester, NY, United States

Brand-Loyaly-Driven Competitive Price Promotions

Game theoretical results suggest the use of price promotions by competing profit maximizing manufacturers of frequently purchased consumer goods may stem from the existence of the consumer phenomenon of brand loyalty. We thus investigate empirically whether managers of brands in categories in which consumers do not exhibit loyalty do not offer price promotions, whereas for brands in categories where loyalty exists, the frequency of offering price promotions is positively related to the magnitude of consumer loyalty.

We examine 183 brands from 20 product categories that are part of the IRI household panel data set. We use an MCMC algorithm to jointly estimate, models of consumers demand and managers price promotions supply, of all 20 categories, thus addressing the issue of "non-random" marketing-mix variables. We find that in all product categories, consumers exhibit brand loyalty and brands offer price promotions. Moreover, controlling for category, store, and brand factors, managers of brands with higher consumer-loyalty effects indeed offered price promotions more frequently.

3 - Alternate Elicitations of (Maximum) Willingness to Pay
Jared Wong, Yale University, New Haven, CT, United States, Jin Kim, Andrew Meyer, Nicholas Surdel, Shane Frederick

This research investigates the potential differences between maximum willingness to pay (WTP) and standard WTP elicitation methods. Prior research suggests that if consumers interpret reservation prices as a range, maximum WTP solicitations may lead to considerably higher valuations than standard WTP inquiries. However, in a series of six studies involving 1,704 participants, we find minimal disparities between these procedural approaches. Our findings suggest that the distinction between maximum WTP and WTP solicitations may not have substantial implications for measuring consumer preferences. Understanding the relationship between different procedural methods in WTP elicitation can inform future research and enhance the accuracy of consumer valuation assessments.

FC02
Meeting Room C2.6
Game Theory for Platforms
Contributed Session
Chair: Michelle Lu
1 - The Impact of Competitive Intelligence Services on Online Marketplaces
Arcan Nalca, Smith School of Business / Queen's University, Kingston, ON, Canada, Taner Bilgic, Mehmet Gumus, Mohammad Nikoofal

Recent innovations have driven a steady increase in online marketplace transactions. To remain competitive, numerous marketplace platforms and independent data providers offer Competitive Intelligence Services (CIS), enabling sellers to explore not only their market potential but also that of their competitors. In this paper, we employ a game-theoretical approach to competitive learning to analyze the impact of CIS on participants with varying market shares in an online marketplace. In the presence of noisy demand signals, the online platform benefits from offering CIS free of charge to both sellers. This is because high demand noise makes demand exploration difficult for each seller in the first period. Consequently, price competition under poor knowledge of the price-demand relationship in the second period leads to a lower payoff for each seller as well as the platform. However, as demand uncertainty decreases, the platform prefers inducing CIS exclusively for the seller with the larger market share. This scenario leads to signal-jamming behavior between the sellers, which results in a win-win-win situation for both sellers and the platform provider. Finally, we consider various model extensions and discuss the managerial implications for the design and regulation of competitive intelligence services in online marketplaces.

2 - Consumers' Subscription Mix: Menu Pricing in Two-Sided Media Platform Competition
Yunhyoung Kim, Marketing, University of Minnesota Twin Cities, Minneapolis, Minnesota, United States
This article studies the impact of subscription mix, "consumers' multi-homing behaviors based on video service platforms' menu pricing strategies", in two-sided media platform competition. Recently, there are huge portions of consumers who subscribe to more than one video streaming service. Platforms like Netflix and YouTube have introduced varied subscription models—including premium, ad-supported, and free access—to not only increase their subscriber base but also to appeal to a broader range of advertisers. The study questions the efficacy of such strategies, considering the unpredictability of consumer subscription combinations and the potential decline in advertising revenue if platforms fail to target unique consumer segments effectively. Given these dynamics, two different media platforms are considered: an advertising-focused media platform and a subscription-focused media platform. An analytical model, grounded in two-sided platform competition literature, is employed to explore these phenomena.

Key findings reveal that a consumer's mix of subscriptions encourages platforms to adopt asymmetric menu pricing strategies, allowing them to mitigate intense competition on both the consumer and advertiser fronts. These strategies are found to be largely dependent on the marginal utility derived from multi-homing. There is an identified optimal level of marginal utility increase from multi-homing that enables platforms to diversify subscription plans and maintain a competitive edge, resulting in mutual benefits for all involved up to a certain threshold. However, if the perceived value of multi-homing drops or exceeds this threshold, the advantageous effects diminish, presenting a critical balance that platforms must navigate.
3 - A Model of In-Consumption Social Interactions
Michelle Lu, China Europe International Business School, Beijing, China, People's Republic of, Ji Shen

In digital content consumption, social interactions become vital components that drive consumers’ overall experience and willingness-to-pay for digital consumption. Readers of digital media obtain utility not only from informational gists they draw but also from comments shared by other readers. Learning experience on online education sites is largely shaped by student interactions in course communities. Viewers of livestreaming content enjoy interactions with other viewers. Hence, managing social interactions is a looming challenge for digital content providers – who can participate and how will social interactions change the paywall strategy? To answer these questions, we construct a continuous time model in which a consumer’s willingness-to-pay for the product is dynamically shaped both by idiosyncratic shocks and endogenous interactions with other consumers. We coin a term "social interaction elasticity," which measures how sensitive consumers' social utility respond to social interactions. We find that in the presence of negative social influence, content providers' profitability hinges on social interaction elasticity. Our model framework and results can also be applied to the context of physical goods and provide managerial insights to firms that consider building interactive online customer communities.

FC03
Cockle Bay Room 1
AI for Marketing
Invited Session
Special Sessions
Chair: Song Lin, Hong Kong University of Science and Technology, Hong Kong, Hong Kong
1 - Competitive Model Selection in Algorithmic Targeting
Tony Ke, Chinese University of Hong Kong, Shatin, Hong Kong, Ganesh Iyer

We study how market competition influences the algorithmic design choices of firms in the context of targeting. Firms face a general bias-variance trade-off when choosing the design of a supervised learning algorithm in terms of model complexity or the number of predictors to accommodate. Each firm has a data analyst who uses the chosen algorithm to estimate demand for multiple consumer segments, based on which, it devises a targeting policy to maximize estimated profits. We show that competition induces firms to strategically choose simpler algorithms which involve more bias but lower variance. Therefore, more complex/flexible algorithms may have higher value for firms with greater monopoly power.

2 - CANCELLED The Illusion of Interpretation: Post Hoc Explanations Aren't a Silver Bullet for Business Research
Tong Wang, Yale University, New Haven, CT, United States, Ronilo Ragodos, Feng Lu, Yu Jeffrey Hu
In the realm of business research, the need to demystify black-box machine learning models is paramount. Post hoc explanation methods, such as LIME and SHAP, have gained widespread popularity for this purpose. Common practice involves constructing a black-box model and subsequently employing these post hoc tools to interpret the model's output. However, this approach harbors fundamental flaws. This paper delves into the potential misleading effects of post hoc explainers when used to interpret phenomena underlying a model's training data. Our study focuses on what we term the "Prediction Insight Pipeline" in business research, where we identify and illustrate critical issues. We reveal that post hoc explanations not only often deviate from the actual truths in the data but also display inconsistency among themselves, further muddling the understanding of these models. To address these challenges, we introduce and evaluate straightforward and intuitive mitigation strategies. These strategies form the basis of recommended steps for utilizing post hoc explanations, the effectiveness of which is validated through both real and simulated datasets.

3 - Discovering Panoramic Market Structure via Large-scale Consumer Movement Data
Xiao Liu, New York University, New York, NY, United States, Eunkyung An, Natasha Foutz, Baohong Sun
Mobile trajectory data capture consumers’ granular store patronage and embed rich information about individuals’ brand preferences. Leveraging this emerging source of big data,we discover novel panoramic market structures of brick-and-mortar retailers across industries via a heterogeneous network representation learning model. Empirical analyses on 36 million records of location data from two major U.S. metropolitan areas demonstrate the effectiveness of the proposed model in deriving pan-industry consumer segments, discovering targeting strategies, identifying competition radius as well as inter-category strategic alliances.
4 - Towards Intelligent Shopping Assistant: An Early Exploration
Song Lin, Hong Kong University of Science and Technology, Hong Kong, Hong Kong, Zijun Shi, Xinyi Sun
We present an exploratory study investigating the potential of Generative AI, specifically Generative Pre-trained Transformer (GPT), to enhance consumer decision making in the context of E-commerce. We build an online shopping platform where users can browse products, view product descriptions and consumer reviews, placing products in a shopping cart, and make purchase decisions. Additionally, we design a chatbot assistant, fueled by GPT-3.5 from OpenAI, capable of aiding consumers in searching for products and summarizing product descriptions and consumer reviews. To evaluate the impacts of AI assistance on shopping behaviors, we conduct incentive-aligned experiments on the shopping platform, by randomly assigning website traffic into five design conditions: (1) a control condition without any AI function, (2) a review keywords filtering function, (3) GPT-generated review summaries without a chatbot, (4) a GPT-powered chatbot assistant, and (5) a GPT-powered chatbot assistant with keywords filter. Our analysis explores how different designs influence various aspects of consumer shopping behaviors, including time spent, search activity, purchase decision, price paid, perceived information overload, and the overall shopping satisfaction. We find some evidence that a GPT-powered shopping assistant can encourage more search, increase purchase likelihood, and reduce consumers’ perception of information overload.
FC04
Cockle Bay Room 2
Meet the Editors 1
Invited Session
Chair: Valentyna Melnyk, UNSW, Sydney
1 - JCR
Bernd Schmitt, Columbia University, New York, NY, United States
Journal of Consumer Research
2 - Marketing Science
Olivier Toubia, Columbia Business School, New York, New York, United States
Marketing Science Journal
3 - Journal of Retailing
Katrijn Gielens, University of North Carolina-Chapel Hill, Chapel Hill, North Carolina, United States

Journal of Retailing

4 - JACR
Vicki Morwitz, Columbia University, NEW YORK, New York, United States
JACR
5 - Journal of International Marketing
Aysegul Ozsomer, Koç University CASE, Istanbul, Turkey, Kelly Hewett

Journal of International Marketing

6 - Journal of Choice Modelling
Harmen Oppewal, Monash University, Caulfield East, Victoria, Australia

Journal of Choice Modelling

7 - Management Science
Wilfred Amaldoss, Duke University, Durham, NC, United States
Management Science
FC05
Meeting Room C3.1
Marketing Finance Interface
Contributed Session
Chair: Ujwal Kayande, Melbourne Business School
1 - Consumer and Investor Reactions to Media-exposed Corporate Tax Avoidance
Dominik Papies, University of Tuebingen, Tuebingen, Germany, David Scheuermann, Samuel Staebler
Corporate tax avoidance activities of multinational corporations have attracted a lot of media attention in recent years. Marketing, however, has not studied this development, although it is unlikely that this type of public attention will go unnoticed by consumers and investors. Hence, it remains unclear how consumers and investors react to this type of public attention. To address this void, we develop a conceptual framework that delineates the similarities and differences between corporate tax avoidance and related phenomena such as corporate social irresponsibility. We then study nearly 500 of the largest publicly traded US firms over a 10-year period. We analyze their tax payments and utilize several modern Machine Learning approaches to understand the content and sentiment of the media coverage of their tax behavior. In several empirical analyses, we explore how these two constructs affect the brand strength and the firms’ stock market performance. We find that, although such revelations hurt brand strength and increase share trading immediately after the revelation of a questionable tax practice, the evidence for an effect on firm value is inconclusive. This suggests that, although consumers and investors exhibit heightened awareness and reactions to reports on aggressive tax strategies, investors do not uniformly hold negative views on such practices and their associated media exposure. Our findings challenge the notion that market forces alone can effectively deter questionable tax management, and policymakers should consider alternative regulatory measures to curb excessive corporate tax avoidance.
2 - Market Your Share! Marketing Capabilities, Investor Relations and Firm Performance
Andre Tomano, Maastricht University, Maastricht, Netherlands, Niels Holtrop, Joost Pennings, Thomas Post

We examine the influence of a firm’s marketing capability (MC) on the quality of its Investor Relations (IR) team. Further, we investigate the individual and collective impact of these constructs on the firm’s financial performance, postulating IR quality as a mediating factor.
Past research has shown that a firm’s MC has a positive impact on its financial performance (e.g. Ang et al., 2022; Cao et al., 2023). MC can therefore serve as a credible signal to investors about the future prospect of a company. Yet, the underlying mechanism(s) linking MC and firm performance remain unclear.
In this research, we investigate an essential link between firm-specific information and analysts and investors, that is, a firm's IR team. Empirical evidence has established that active, high quality, or just the mere activation of IR is related to better financial market metrics (Agarwal et al., 2016; Karolyi et al., 2020; Neukirchen et al., 2023).
Utilizing secondary data for a sample of around 3,000 US firms from 2016 to 2021, the preliminary results show a positive effect of MC (estimated via Stochastic Frontier Analysis) and IR quality (measured by an annual survey conducted by Institutional Investors) on raw stock returns and Tobin’s q. The analysis also reveals a significant positive effect of MC on IR quality and that IR quality serves as a mediator between MC and Tobin’s q.
Hence, the ability of these firms to better ‘market’ their shares to analysts and investors, leads to increased financial performance over firms that lack this ability.

3 - In the Eyes of the Stakeholders: Performance Implications of Incoherent Corporate Reputations
Ujwal Kayande, Melbourne Business School, Carlton, VIC, Australia, Adnan Yusuf, Rajdeep Grewal, Peter Roberts

Because a firm’s reputation is multidimensional (e.g., product quality, social responsibility), it can vary across dimensions. We term such a variance across dimensions as “reputation incoherence” and contribute to academic research on corporate reputation to help corporate and marketing managers understand how stakeholders perceive and assess firm reputations in the marketplace and how those perceptions can translate into financial outcomes at the firm level. Reputation incoherence could have varying effects on firms’ financial performance level and performance variability, depending on the extent to which stakeholders attend to the distinct dimensions. If each class of stakeholders focuses on only one or two distinct dimensions, this reputation incoherence likely will create greater performance variability than if all stakeholders attended to all dimensions. In that latter case, reputation incoherence might lead the stakeholders to discount all reputation information, with negative implications for its performance level. With data from 453 publicly listed firms, pertaining to 2017–2022, the current research confirms these hypotheses and reveals that a firm’s visibility moderates these effects. Because both lower performance levels and higher performance variability produce diminished firm value, managers should develop marketing strategies that ensure coherent reputations; to do otherwise means putting their firm’s market value at risk.

Key words: corporate reputation, firm performance, firm visibility, market-based assets, performance variability, reputation incoherence.

FC06
Meeting Room C3.2
Online Retailing: Tackling Abandoned Shopping Carts
Contributed Session
Chair: Karthik Sridhar, Baruch College - The City University of New York, 1 Bernard Baruch Way, New York, NY, 10010, United States
1 - Role of Incentives in Abandoned Shopping Cart Recovery Funnel
Yoonju Han, Ewha Womans University, Seoul, Korea, Republic of, Sandeep Chandukala, Srinivas Reddy, Yong Chin Tan

A pervasive challenge in online retail is cart abandonment (i.e., customers adding items to their carts, but leaving without purchasing). Retailers usually provide incentives to recapture lost sales from abandoned carts. In this research, we examine the role of incentives in moving customers along the cart recovery process (i.e., revisiting the retailer, converting to purchase, returning products post-purchase). In addition, the moderating influence of incentives on the effects of various customer (i.e., coupon-proneness and average size of previous transactions) and cart characteristics (i.e., number of cart items, proportion of private-label products, and device usage during cart abandonment) in the cart recovery process are explored. We jointly model the decisions along the purchase funnel using a Hierarchical Bayesian model. Collaborating with a leading online fashion retailer, we conducted a field experiment that involved manipulating the presence of incentives in recovery interventions. Our findings reveal that incentives encourage faster revisits to the retailer, facilitate conversion to purchase while also increasing product returns. Furthermore, coupon-prone customers are more likely to return products and this effect is enhanced when they are provided with incentives. However, incentives decrease the likelihood of returning products when provided to customers with larger average transaction value. Finally, incentives increase the likelihood to return products when the abandoned cart has a larger number of items and larger proportion of private labels. Findings from our study would provide guidance for retailers to ascertain if recovery incentives are appropriate to achieve their business objectives, i.e., increase revisits, conversions, or reduce product returns.

2 - When Do Shopping Cart Interventions Increase Purchases?
Arun Gopalakrishnan, Rice University, Houston, TX, United States, Young-Hoon Park

In the realm of electronic and mobile commerce, retailers frequently offer online shopping carts, allowing customers to temporarily store items. However, a significant proportion of items in these carts are not purchased, prompting firms to target cart-abandoning customers with strategies from reminder messages to coupon offers. Existing literature on the effectiveness of these shopping cart interventions is sparse. This research addresses this gap through a comprehensive field experiment that investigates the effects of mobile reminders and coupons on the conversion funnel with an online retailer. Specifically, this study examines two critical factors that may influence the performance of these interventions: (1) customers’ historical cart utilization patterns and (2) the timing of the intervention, assessed over a wide temporal range of 1 to 30 days.

We find that mobile coupons significantly increase purchase rates, with the lift ranging from 56% to 155%. This lift in purchases largely results from coupon redemptions of items already in the cart, rather than through advertising or spillover effects. Interestingly, a mere reminder without a coupon does not significantly change customer behavior compared to a no-communication control group. Additionally, we find that customers’ prior cart usage patterns significantly influence the effectiveness of coupons, offering important insights for targeting customers who abandon carts. Finally, we find that the increase in purchases is highest when interventions are timed between 4 and 9 days after items are left unpurchased in the cart, suggesting that timing is a crucial moderator for the performance of such interventions.

3 - Tackling Abandoned Cart Recovery as a Classification Problem
Karthik Sridhar, Baruch College - The City University of New York, New York, New York, United States, Ram Bezawada, Ashish Kumar

Online retailers face significant revenue losses due to cart abandonment during the checkout process. The prevalence and increase in the frequency of such customer actions lead to a decrease in customer loyalty and firm engagement in the long run. This study utilizes actual purchase data to offer actionable strategies for reducing cart abandonment and facilitating recovery in the presence of abandoned cart reminder instruments. The study leverages a rich dataset spanning multiple years, with information on cart abandonment and subsequent partial or full recovery of abandoned cart items with or without exposure to reminder instruments. The dataset encapsulates novel explanatory variables, such as the presence of popular items, payment mechanisms, cart value, and customer recency and frequency actions, that can be used to understand the antecedents of cart abandonment and recovery processes. We investigate cart abandonment and subsequent recovery behaviors as a classification problem and employ classification algorithms in machine learning to delineate cart recoveries as full, major and minor recoveries. Although very managerially relevant, these different types of recoveries in relation to abandoned carts have not been investigated by prior research. By jointly modeling cart abandonment, reminder email response, and the types of cart recovery, the research provides insights while controlling for self-selection bias. This study's implications are relevant for online retail firms, offering strategies to curtail cart abandonment and expanding the understanding of different recovery scenarios. Online retailers can benefit from tailored approaches to address specific customer behaviors, ultimately improving their revenue and customer relationships.

FC07
Meeting Room C3.3
Public Policy Applications of Causal Inference
Invited Session
Special Sessions
Chair: Unnati Narang, University of Illinois at Urbana Champaign, Champaign, IL, United States
Co-Chair: Ali Goli, University of Washington, 2901 NE BLAKELEY ST. APT 131, Seattle, WA, 98105, United States
1 - Stochastic Product Failures: Identifying the Impact of False Positive Tests on Product Usage
Dinara Akchurina, University of Toronto, Toronto, ON, Canada, Avi Goldfarb

In many product categories, such as healthcare, the underlying product quality is challenging to assess even upon consumption. This raises the question of whether occasional product failures impact the regular consumption and usage of products in this category. Our empirical setting revolves around a large-scale COVID-19 rapid screening program run across various enterprises and sectors throughout Canada, in which participants were encouraged to regularly screen for COVID using rapid antigen tests. Using matching estimators, we show that receiving a false-positive result has long-term consequences for product usage: users who receive a false positive result are more likely to drop out of the program weeks later after receiving the false-positive result. The effect is especially salient for users whose first experience with a product resulted in a product failure. Our results underscore the importance of occasional product failures in shaping consumer beliefs and subsequent product usage.

2 - From Pills to Puffs: How Prescription Drug, OTC, and PSA Ads Shape Smoking Mitigation Efforts
Ali Goli, University of Washington, Seattle, WA, United States, Erfan Loghmani
This study delves into the interplay between different forms of advertising and the demand for cigarettes and their substitutes. We investigate how public service announcements (PSAs) by government entities, e-cigarette advertising, prescription drug advertising (e.g., Chantix), and advertising by nicotine replacement therapies (NRTs) like patches and gums influence consumer behavior. Our findings reveal a significant reduction in cigarette demand due to advertising by competing products and PSAs, with a notable shift in demand towards substitutes. Interestingly, prescription drug advertising exhibits a spillover effect, influencing the uptake of over-the-counter options, particularly e-cigarettes. This spillover varies systematically with the extent of insurance coverage across different geographies. Our research underscores the emergent role of e-cigarettes as a smoking cessation alternative and highlights the influence of insurance coverage in shaping consumer preferences towards different smoking cessation products.
3 - Market Structure and Political Influence in the Auto Retail Industry
Sarah Moshary, University of California, Berkeley, Berkeley, CA, United States, Cailin Slattery
We study the relationship between market structure and lobbying in the automobile retail industry. The size distribution of an industry can affect lobbying—lobbying is a public good, and larger firms will internalize more of the benefits of lobbying activity. We create a new data set of lobbying at the state level and exploit variation in the misalignment of local political and product markets to identify the public goods effect of market structure on lobbying. A first finding is that the majority of lobbying in the industry occurs via trade associations. These are industry groups that coordinate firms and could potentially solve the collective action problem in lobbying. To the contrary, we find that mergers of large auto retailers increase lobbying and lead to favorable outcomes for the industry.
FC08
Meeting Room C3.4
Diversity, Equity, Inclusion (DEI)
Contributed Session
Chair: Qinghua Li, Northwestern University, Evanston, IL, United States
1 - Cinematic Reflections: Race Matching in Filmmaking and Audience Appeal
Jia Li, Wake Forest University, Winston Salem, NC, United States, Koleman Strumpf

This study investigates the relationship between the racial identity of primary figures in film production, such as main actors/actresses and directors, and the film’s reception within racially congruent communities. Utilizing a mixed-methods approach, we analyze box office data and community reception patterns across diverse racial groups. Our dataset encompasses films from various genres and periods, with special emphasis on those where the director or lead actors/actresses belong to minority racial groups. Our corresponding model explores whether films directed by or starring individuals of a particular race tend to resonate more strongly within communities of the same race. This study contributes to the understanding of racial representation in media and its impact on community preferences. It offers insights into how film as an art form can both reflect and shape cultural identities, highlighting the importance of diversity in the film industry.

2 - Beauty in a Spectrum of Sizes: Sales Impact of Advertising Models' Body Shape
Jingpeng Hong, Harvard Business School, Boston, MA, United States, Zijun (June) Shi, Shunyuan Zhang

In recent years, the emergence of more inclusive body sizes in models has marked a transformative shift in the fashion industry, championing diversity and inclusivity. Though brands are increasingly integrating models of more inclusive body shapes into their showcases, the consumer response remains ambiguous. This paper investigates the influence of using more inclusive body size models on sales by using a unique three-month longitudinal dataset from a large online marketplace in Asia. The dataset contains 168,841 apparel products with transaction and clickstream data. Images on product pages allow us to measure the models' body sizes by extracting their body mass index (BMI), along with a rich set of other characteristics. We find that a 1% increase in the advertising model’s BMI correlates with a 1% rise in sales. In particular, the sales lift is attributable to a concurrent lift in both the likelihood of clicking on the product page and the likelihood of purchasing upon the click. Furthermore, we observe larger effects of increasing model’s BMI on sales for slim-fit items (1.33%) than for relaxed-fit items (0.65%). Next, to verify the causal link and mechanisms suggested by the field data, we conduct a controlled experiment in which we use generative AI techniques to manipulate the body shapes of advertising models, thereby providing additional causal support to our findings. This research highlights that a tangible commitment to inclusivity can contribute to commercial success and hence provides significant implications for brands that wish to enhance inclusivity in designing or advertising their offerings.

3 - The Beauty Bias Effect in Academic Careers within Business
Qinghua Li, Northwestern University, Evanston, IL, United States, Dawei Wang, Brian Uzzi

Despite discussions surrounding various biases, the influence of a person's physical appearance on decision-making is sometimes overlooked. This study delves into the role of facial beauty in shaping the career success of scientific scholars.

While scholarly assessment ideally relies on one’s public scientific output, the presence of the “halo effect” suggests that physical attractiveness can also impact career success. This cognitive bias leads people to assume physically attractive individuals possess other positive qualities such as intelligence, competence, and likability. Outside academia, empirical studies demonstrate that physical attractiveness can enhance career success, including job interviews, sales, and political roles.

To explore the relationship between facial beauty and career success among scientific scholars, we conducted multiple studies using diverse samples of scholars. Initially, we extracted over 20,000 scholars from Google Scholar. We then expanded the scope of our inquiry to the earlier stages of the scholars’ careers, with a specific focus on assistant professors in top business schools. We used three independent machine learning models to assess their facial beauty based on their profile pictures. Variation in the quality of photos was also reduced using machine learning.

Our findings consistently linked scholars' facial beauty to higher university rankings, including junior faculty members in business schools, irrespective of their academic performance. This underscores the importance of promoting diversity and equity in scientific fields, ensuring equal opportunities for all scholars regardless of appearance or background.

FC09
Meeting Room C3.5
Channels
Contributed Session
Chair: Tomokazu Kubo, Chuo University, Tokyo, Japan
1 - How Channel Integration Benefits Manufactures' Online Market Entry Agility? The Mediating Role of Dealers' Dependency Behavior
Hua Fan, Shanghai International Studies University, Shanghai, China, People's Republic of, Bing Han

Marketing Relevance: The effectiveness of channel integration performed by manufacturers in facilitating their online market entry is still an under-researched topic. This study fills the research gap by investigating and comparing the effects of channel integration strategies and components implemented by manufacturers.

Methodology/Approach: A questionnaire was designed and distributed to purchasing managers in downstream dealers. The survey yielded 2051 valid responses, and data analysis employed hierarchical and polynomial regressions, along with response surface analyses.

Findings: First, dealers’ dependence plays a crucial role in triggering manufacturers’ online market entry. Second, manufacturers’ channel consistency and transparency are more relevant to dealers’ active dependence, while information and interaction integration lead to dealers’ passive dependence. Third, channel ambidexterity is superior to integration ambidexterity in ensuring dealers’ dependence.

Contribution: Significant contributions to omnichannel integration research are made by introducing four distinct types of channel integration strategies. Enhancements to the interfirm dependence literature are achieved through a comparative analysis of the effectiveness of channel integration strategies in shaping dealers' dependency behavior.

Implications: Manufacturers should carefully evaluate the effectiveness of these strategies, avoiding a blind pursuit of all forms of channel integration ambidexterity. Strategic leverage of channel feature ambidexterity is recommended as a cornerstone for fostering dealers’ dependence and achieving success in online market entry.

Key words: interfirm dependence, market entry agility, channel consistency, channel transparency, information integration, interaction integration

2 - Regional Network Configurations and Franchised Outlet Failure: A Network Theory Perspective
Li Ji, Hong Kong Metropolitan University, Hong Kong, Hong Kong, Xu Zheng, Haibin Yang

Franchised outlets are not isolated actors; instead, they are embedded within a regional network consisting of the focal outlet and their uniform counterparts. While extant research has acknowledged the interlinks between component stores, few studies have examined how an actor’s performance is affected by the network configurations. Taking a network theory perspective, we view a focal franchised outlet to be embedded within a regional franchise network, consisting of both multi-unit franchisees (MUFs) and single-unit franchisees (SUFs). In addition, each network configuration differs in terms of the overall network connectedness, which refers to the ratio of all actual to possible ties in the network, and network heterogeneity, which refers to the imbalance in the outlet number of each franchisee. We conceptualize and empirically test how the overall network configurational characteristics, including network connectedness and network heterogeneity, individually and jointly, affect individual outlet failure. We conduct the complementary log-log regression on the survival duration of 17,959 outlets across 26 franchisors operating in United States from 2002 to 2021. The result shows that network connectedness has a U-shaped impact on outlet failure, and network heterogeneity increases an individual outlet failure. Furthermore, network heterogeneity steepens the U-shaped impact of network connectedness on outlet failure. Our research provides actionable insights for franchisors to strategically configure their franchise networks. In particular, franchisors should watch out the back firing effect of MUFs by avoiding an over-connected network and can set an imbalanced franchisee configuration to mitigate the positive effect of high network connectedness on outlet failure.

3 - Channel Value Prediction - Clinic Channel with Skincare Products
Kuo Hao Huang, National Chung Hsing University, Taichung, Taiwan, Shihyu Chou, Ming Chih Tsai

The concept of channel value has evolved into an indispensable key element within the realm of marketing strategy. Despite its significance, there exists a notable scarcity of in-depth research on channel value. Prior studies have largely overlooked theoretical discussions and the development of predictive models, leaving a knowledge gap within the academic community. This research endeavors to address this gap by conducting a more comprehensive exploration of the importance of channel value and the development of predictive models associated with it. Taking inspiration from Kumar's (2018) concept of customer future profit, we aim to elucidate customer value from a more holistic perspective. Utilizing transaction data from a Taiwanese skincare products company, specifically focusing on its clinic channel customers, we employ a two-stage clustering technique to identify distinct types of clinic channel value. Following the identification of channel value type and the labeling of channel clinics, we establish connections between the type of channel value and relevant variables, as outlined by Kumar (2018), along with some fundamental clinic characteristics. This is achieved by training a neural network model. Through this study, our objective is to furnish the skincare products company with profound insights into channel management. By refining the understanding of channel value, we aim to contribute to the overall enhancement of revenue performance within the enterprise. This research endeavors to bridge the existing knowledge gap and serve as a valuable resource for academia and practitioners.

Keywords: Marketing channels, Channel value, Predictive models, Customer future profit, Neural network model, Clinic channel

4 - Does Marketing Channel Structure Affect Channel Performance? The Moderating Role of Competitive Strategy and Market Orientation
Tomokazu Kubo, Chuo University, Tokyo, Japan, Sho Yuki
Manufacturers use distributors or vertically integrate them to deliver their products to end customers. Vertical integration of distributors by manufacturers is expected to affect marketing channel performance, but the relationship is expected to be contingent on competitive strategy and market orientation. The competitive strategy hypothesis is that vertical separation will enhance channel performance for cost leadership, while vertical integration will enhance the one for differentiation strategies. Furthermore, as a market orientation hypothesis, vertical separation is expected to enhance channel outcomes for reactive market orientation, while vertical integration is expected to enhance channel outcomes for proactive market orientation. In the empirical analysis, endogeneity is problematic in that manufacturers with higher channel outcomes choose vertical integration. In this study, we use propensity score matching to extract vertically integrated/vertically separated pairs with equivalent level covariates. We empirically analyze our hypotheses using data collected from Japanese BtoB manufacturers.
FC10
Meeting Room C3.6
The Spread of Political Information on Social Media
Invited Session
Special Sessions
Chair: Justin Huang, University of Michigan Ross School of Business, Ann Arbor, MI, United States
Co-Chair: Pinar yildirim, University of Pennsylvania, Philadelphia, PA, United States
1 - Reducing Misinformation Sharing at Scale Using Digital Accuracy Prompt Ads
Nils Wernerfelt, Northwestern University, Evanston, IL, United States, Hause Lin, Haritz Garro, Jesse Conan Shore, Adam Hughes, Daniel Deisenroth, Nathaniel Barr, Adam Berinsky, Dean Eckles, Gordon Pennycook, David Rand
Interventions to reduce misinformation sharing have been a major focus in recent years. Developing “content-neutral” interventions that do not require specific fact-checks or warnings related to individual false claims is particularly important in developing scalable solutions. Here, we provide the first evaluations of a content-neutral intervention to reduce misinformation sharing conducted at scale in the field. Specifically, across two on-platform randomized controlled trials, one on Meta’s Facebook (N=33,043,471) and the other on Twitter (N=75,763), we find that simple messages reminding people to think about accuracy—delivered to large numbers of users using digital advertisements—reduce misinformation sharing, with effect sizes on par with what is typically observed in digital advertising experiments. On Facebook, in the hour after receiving an accuracy prompt ad, we found a 2.6% reduction in the probability of being a misinformation sharer among users who had shared misinformation the week prior to the experiment. On Twitter, over more than a week of receiving 3 accuracy prompt ads per day, we similarly found a 3.7% to 6.3% decrease in the probability of sharing low-quality content among active users who shared misinformation pre-treatment. These findings suggest that content-neutral interventions that prompt users to consider accuracy have the potential to complement existing content-specific interventions in reducing the spread of misinformation online.
2 - Language Identity and Foreign Propaganda: Evidence from the Russia-Ukraine War
Andrey Simonov, University of Chicago, Chicago, IL, United States, Sergey Sanovich
We measure the ability of the Russian government to exercise propaganda in Ukraine, leveraging the popularity of news on the Telegram app in the country. Historically, most Ukrainian Telegram news channels have been in the Russian language, due to the bilingual Ukrainian population living in the south and east of the country. Since February 2022, these proportions have reversed to 70% of news in Ukrainian, due to a strong preference for the Ukrainian identity. Using detailed Telegram API data on news production and readership, we show that this language transition has decreased the engagement of news readers with the content, presumably since it is harder for traditionally Russian-speaking Ukrainians to switch their news consumption to the Ukrainian language. We then measure the substitution patterns of these readers, checking if some of this substitution happened to Russian propaganda news groups focused on Ukraine and measuring the unintended consequence of helping Russian propaganda to reach more Ukrainians.
3 - Can We Talk about Race and Racism on Social Media? Evidence from a Feed Experiment
Dante Donati, Columbia Business School, New York, NY, United States, Lena Song, Thomas Lilly

Racial attitudes remain divided in the US despite a national reckoning. Does this mean that we cannot talk about race and racism? This paper studies what shapes the comment section of racial justice content on social media, and its subsequent effects on beliefs, attitudes, and behavior. The project consists of two phases. Using an online field (feed) experiment on Facebook, Phase 1 studies how the quantity and quality of engagement (conversation/comments) with social media posts supporting racial justice vary across users with ex-ante different ideologies. To do so, we exogenously vary the characteristics of the content and the composition of users who see and engage with the same post on the Facebook feed (i.e., users see the posts along with a diverse group, or a homogenous group of like-minded users). The results will inform companies and organizations on how to promote healthy conversations with different users on divisive social issues and avoid backlash. Phase 2 investigates how exposure to different narratives expressed in the comment section of a post affects individuals’ subsequent beliefs, attitudes, and behaviors. For this, we categorize comments collected in Phase 1 based on their prevailing narrative (endorsing or rebutting the message of the post) and randomize their exposure across individuals on Facebook. The findings will contribute to our understanding of how social media platforms can be leveraged for social change, both on racial justice and other social issues.

4 - Politically-biased Moderation Drives Echo Chamber Formation: An Analysis of Content Removals on Reddit
Jangwon Choi, Santa Clara University Leavey School of Business, Santa Clara, CA, United States, Justin Huang, Yuqin Wan
Echo chambers - online spaces where individuals are met with reinforcing viewpoints and insulation from opposing viewpoints - are of increasing interest to social media platforms and policymakers amidst rising political polarization in the US. Selective participation and algorithmic recommendations are commonly cited as drivers of echo chamber formation. In this work, we identify and examine a novel mechanism for the formation of echo chambers on social media sites: the politically-biased content removal decisions of human moderators. Applying a combination of natural language processing and network analysis to characterize the political leanings of users and their textual content on a dataset of 1.2 million users' removed content on Reddit, we document various patterns of adverse content removal (e.g., left-leaning content gets removed by right-leaning moderators, and vice-versa) spanning different types of content (e.g., highly toxic comments vs. others) and moderation environment (e.g., subreddits with high vs. low moderator political diversity). Then, using a propensity score matching approach, we further show that content removals have an indirect chilling effect on censored users' subsequent political speech. We conduct counterfactual simulations to assess the impact that politically biased content removal has on a measure of echo chamber intensity in political segregation. These findings are of broad interest to users, platforms and regulators who weigh the costs and benefits of freedom of expression with the logistics of moderating massive online spaces.
FC11
Meeting Room C4.1
Models for Targeting and Personalization 1
Contributed Session
Chair: Xinying Hao, University of Arizona, Tucson, AZ, United States
1 - Understanding Consumer Heterogeneity: A Comparative Study of Hierarchical Bayesian Models and Machine Learning Approaches
Yuki Kotakeyama, Hakuhodo DY Holdings, Tokyo, Japan, Hiroshi Kato, Masataka Ushiku, Shonosuke Sugasawa
Understanding how consumer heterogeneity affects marketing actions is essential for advancing marketing strategies, such as realizing micromarketing. The hierarchical Bayesian model framework is commonly used to express consumer heterogeneity in market response models such as brand choice models. In this framework, the response parameters of the utility function of the brand choice model are assumed to be different for each consumer and are estimated. On the other hand, in the machine learning domain, there are approaches such as Bayesian Causal Forest (BCF) that can account for consumer heterogeneity. Machine learning approaches are flexible and can estimate complex interactions and nonlinear effects, even when handling high-dimensional variables. In this study, we examined two methods to express consumer heterogeneity: a hierarchical Bayesian model and a machine learning-based approach to evaluate the impact of brand contact on brand awareness. We modeled the relationship between brand contact and brand awareness using data on brand contact and awareness for each consumer and considered consumer demographic information and well-being as consumer heterogeneity. The results suggest that the hierarchical Bayesian model is superior in accuracy and interpretability when the actual structure of the measure effects is clear and simple, while the machine learning-based approach is superior in exploratory effect estimation in complex and unknown data structures. These results suggest that using a machine learning-based approach to get an overview of the data in the initial stages of effect testing and then applying a hierarchical Bayesian model is considered an effective process for understanding consumer heterogeneity.
2 - Harnessing Causal Inference for Marketing Insights: Estimating and Applying Heterogeneous Treatment Effects
Ting-Tse Chen, University of Sydney, Belfield, Australia, Qiang Lu

Causal inference is a vital tool in data analysis, particularly for estimating treatment effects from observational data in scenarios where Randomized Controlled Trials (RCTs) are impractical. This methodology is especially crucial in marketing, as it facilitates an understanding of how various treatments influence different subgroups, thus informing strategies like targeted advertising and personalised pricing. This paper advocates for the incorporation of machine learning based causal inference methods in marketing research, emphasising the estimation of heterogeneous treatment effects (HTE). We present a systemic literature review of both traditional and contemporary methods in machine learning for estimating HTE, with a focus on static and time-series data. The discussion encompasses methods for static data, including the meta-learners family for controlling pre-existing variables and tree-based techniques for algorithm-driven variable selection. We further explore methods that utilise instrumental variables to address unconfoundedness assumptions. In the context of time-series data, we examine heterogeneity-robust Difference-in-Differences (DID) estimators, which are applicable to both time-invariant and time-varying treatment scenarios. The paper also reviews relevant industry practices and their integration into academic research, pinpointing scenarios where these methods prove empirically significant. We conclude by identifying and discussing promising directions for future research in this field. The primary aim is to deepen researchers' comprehension of HTE methodologies, providing guidance on selecting suitable methods based on the specific features of data and treatments. This is intended to facilitate the effective application of causal inference in marketing research.

3 - Fair Active Learning For Targeting
Zhuoyu Shi, Erasmus University Rotterdam Rotterdam School of Management, Rotterdam, Netherlands, Aurelie Lemmens

­As causal machine learning increasingly guides targeted marketing campaigns, concerns about algorithmic biases have grown tremendously. Many recent scandals have shown that algorithms can reproduce and amplify historical disparities present in the data on which they are trained. To address this issue, we propose a new methodology based on active learning for designing fair experiments, named Fair Active Learning (FAL). The contribution of this paper is twofold: First, unlike other approaches, our method tackles the problem of unfairness at the data acquisition stage, enabling its use in combination with any models that predict heterogeneous treatment effects. Second, FAL focuses on ensuring equal predictive performance across different groups, offering an alternative perspective to the state of the art.

We validate our approach with extensive Monte Carlo simulations, demonstrating that FAL enhances the fairness of targeting policies optimized based on fair experiments. The results reveal that the accuracy-fairness tradeoff varies with experiment size, treatment effect distribution, and their relation to protected attributes or unobserved factors. Under certain conditions, FAL improves outcomes for the protected group at the cost of worse outcomes for the unprotected group. In other cases, both groups experience improved outcomes. Finally, we confirm the effectiveness of FAL through an actual field experiment. This research represents a promising step toward mitigating algorithmic biases in targeted marketing and promoting fair and equitable practices in causal machine learning applications.

4 - Emojis as New Targeting Language: A Multi-modal Emoji Mining Approach
Xinying Hao, University of Arizona, Tucson, AZ, United States, Vijay Mahajan
Emoji targeting has emerged as a practice in marketing, but practitioners have been puzzled by its mixed outcomes. In this paper, we develop a multi-modal emoji mining approach and validate its incremental effect on emoji usage in targeted messages across different levels of product involvement. The marketing literature has not yet developed an emoji-mining approach, so we aim to fill this technical knowledge gap by combining visual and textual information to predict relevant emojis. Specifically, we use state-of-the-art computer vision and natural language processing tools to extract visual and textual features and design a two-stream deep network that fuses information from both modalities to make a joint prediction. By training and testing this network on a large Twitter dataset, our results demonstrate that our method can successfully recommend the relevant emojis, and it outperforms a series of baselines. We further validate our approach by investigating the incremental effects of using emojis on targeting outcomes (e.g., open rate) via lab and field experiments.
FC12
Meeting Room C4.2
Multichannel Shopping Behavior
Contributed Session
Chair: Yuping Liu-Thompkins, Old Dominion University, Norfolk, VA, United States
1 - Multichannel Shopping Routine on Consumers' Retail Choice - From Consumer Questionnaires and Purchase History Data in Single-source Form
Junji Miyamoto, Keio University , Tokyo, Japan

This study aims to research consumers’ multichannel use of retail in terms of questionnaires and purchase records. Previous research has been conducted mainly in terms of online and offline perspectives. There are also studies that capture multichannel characteristics by classifying consumers by psychographic attributes, such as innovativeness and motivation (e.g., Neslin et al., 2006).

This study conducted empirical analyses regarding multichannel usage of actual behavior, considering various retail channels. Two data sets were collected from the same people and analyzed as a single source. First, the frequency of channel choice was asked in the questionnaire survey to measure channel use as a feeling. Second, purchase history data was collected with their receipts and actual behavioral data.

This study examines the consistency between these two data and distinguishes whether consumers’ channel choice behavior is routine or ad hoc. For this aim, the purchase data is divided into two periods, and the data is identified according to whether it is a weekday or a weekend. Additionally, multichannel shopping routine is considered with variables such as consumers’ hedonic shopping motivation, market mavens, and impulse purchases (e.g., Ailawadi, Neslin & Gedenk, 2001; Arnold & Reynolds, 2003; SteenKamp & Gielem, 2003), which were asked in the questionnaire.

In terms of academic significance, this adds an aspect of actual behavior to previous channel choice research, not just psychographics in questionnaires. In practical terms, it is helpful for retailers’ strategies by capturing the routine and ad hoc of consumers’ retail choices.

2 - The Effect of the Gap Between the Ratio of Consumers' use of Channels and the Ratio of Firms' use of Channels on Consumer Purchasing Behavior
Shunnosuke Fukuchi, Graduate school of Business and Commerce, Keio University, Tokyo, Japan
This research aims to explore the relationship between consumer behavior and the different channels they use to make purchases, as well as the channels that companies use to sell their products. The study looks at the impact that the growth of e-commerce (EC) has had on consumer behavior, with many consumers now shopping both online and in physical stores. Meanwhile, companies are adopting new strategies to increase their sales through EC. However, there has been no research to examine the relationship between the actual purchasing behavior of consumers and the optimal ratio of EC use by companies according to the degree of channels use by consumers. Therefore, this study aims to clarify the relationship between the gap between the ratio of consumers' use of stores and EC and the ratio of firms' use of stores and EC, and consumers' actual purchasing behavior in the cosmetics and daily necessities categories. The study also aims to identify the characteristics of consumers who use physical stores more, those who use both channels equally, and those who use EC more compared to the ratio of use channels by firms. The sample for the study is collected from Japanese consumers. This research will contribute to multichannel research, especially in revealing firms’ optimal use of EC that is positively related to consumers’ actual purchasing behavior.
3 - Channel Choice and Customer Value
Shirsho Biswas, University of Washington, Seattle, WA, United States, Hema Yoganarasimhan, Haonan Zhang

Customers often switch to being omni-channel (purchasing offline + online) from offline-only due to several reasons. We investigate whether customers who made this switch (a) voluntarily or (b) due to Covid-19, differ in terms of their post-switching behavior. We find that the consumers who switched to omnichannel shopping from offline-only do not differ systematically from those who made the same switch organically in terms of demographics, pre-switching purchases, and post-switching purchase value, order frequency, and variety of brands and categories purchased. However, they do differ in terms of the share of offline purchases post-switching. For the retailer, since offline purchases are usually more profitable than online purchases, this means that covid-switchers while generating similar revenues as organic switchers post-switch, may be more profitable. Similar to previous literature, we find that both sets of omni-channel customers increase their total monthly purchase value, order frequency, and variety of brands and categories purchased, compared to offline-only customers. Our results suggest that nudging consumers to become omnichannel can increase the profitability of these consumers even beyond that of customers who voluntarily become omnichannel.

4 - Retailer Loyalty "Calculus" Top-Down vs. Bottom-Up Processes
Yuping Liu-Thompkins, Old Dominion University, Norfolk, VA, United States
Many retailers interact with consumers in multiple channels such as physical stores, websites, and mobile channels. Traditionally, loyalty toward a retailer has been construed as a unitary construct encompassing the entire retailer, regardless of the interaction channel. The current research argues that, besides overall loyalty toward a retailer (i.e., traditionally defined retailer loyalty), consumers can also hold different levels of loyalty toward the same retailer in different channels (i.e., channel-specific loyalty). Analyzing transaction history and survey data from a large multichannel retailer, this study proposes three distinct processes in which channel-specific and overall retailer loyalty are formulated: (1) a top-down approach, where a consumer’s overall loyalty toward the retailer determines how loyal they are in each channel; (2) a bottom-up approach, where a consumer combines their loyalty levels in different channels to form the overall loyalty toward the retailer; and (3) a hybrid approach, where a consumer’s loyalty in one primary channel drives their overall loyalty toward the retailer, which in turn drives their loyalty in other channels. The empirical results showed systematic differences in cross-channel synergy vs. cannibalization depending on a consumer’s loyalty “calculus” approach. By identifying these hidden consumer segments with different trajectories in their loyalty across channels, this research can deepen our understanding of customer loyalty in a multichannel setting and allow retailers to personalize their multichannel efforts across different consumer segments.
FC13
Meeting Room C4.3
Advertising Optimization
Contributed Session
Chair: Uğurcan Dündar, WU Vienna, Heumühlgasse 14/19, Vienna, 1040, Austria
1 - LOLA: LLM-Assisted Online Learning Algorithm for Content Experiments
Zikun Ye, University of Washington, Seattle, WA, United States, Hema Yoganarasimhan, Yufeng Zheng

This paper introduces the LOLA, a novel framework that integrates Large Language Models (LLMs) with adaptive experimentation to optimize content delivery. Leveraging a large-scale dataset from Upworthy, which includes 17,681 headline A/B tests. We first investigate three broad pure-LLM approaches: prompt-based methods, embedding-based classification models, and fine-tuned open-source LLMs. We then introduce LOLA, which combines the best pure-LLM approach with the Upper Confidence Bound algorithm to adaptively allocate traffic and maximize clicks. Our numerical experiments on Upworthy data show that LOLA performs the best among several benchmarks.

2 - Robust Pricing and Advertising with Unknown Demand
Thomas Weber, EPFL, Lausanne, Switzerland

The optimal pricing and advertising of innovative goods must proceed without precise knowledge of the demand curve and its dependence on advertising expenditure. This paper provides a data-driven robust method with a relative performance guarantee by maximizing a worst-case performance ratio. Under weak assumptions about the structure of consumer demand, the performance index—which amounts to a worst-case performance ratio—exhibits an “envelope property,” in the sense that it is fully determined by its behavior at the boundary of the parameter space. This allows for an efficient computation of an optimal robust price and advertising expenditure, generalizing and robustifying the seminal approach by Dorfman and Steiner (1954). With constant marginal cost, the method applies to any nonlinear demand curve with bounded slope. With convex cost, the method applies, for example, to any constant-elasticity demand curve of unknown elasticity and scaling factor. For the general case of convex cost and nonlinear demand, the optimal pricing and advertising can also be obtained, just without the numerical shortcut of the boundary representation of the performance index. A comparison with the standard worst-case and minimax regret criteria reveals substantial improvements in both absolute and relative performance, at only a small cost relative to the maximized expected profit.

3 - Advertising for Blood
Olivier Rubel, University of California Davis, Davis, California, United States, Prasad Naik

Over 100 million blood donations are collected annually worldwide. To insure safe blood supplies, blood collection agencies need to act conservatively. What does being conservative mean in this setting? Should a decision maker increase or decrease advertising for blood when being conservative? We answer these and related questions by incorporating conservatism in marketing decision-making. Specifically, we introduce the H-Infinity approach that provides a rich framework to incorporate conservatism in dynamic advertising decisions. In this approach, Nature generates disturbances strategically as an equilibrium of the game rather than drawing them randomly from a probability distribution. Accordingly, a conservative decision-maker takes into account the lack of knowledge of the distribution of disturbances. As conservatism increases, the decision-maker accommodates for a wider range of perturbations to the blood collection rate. When conservatism is ignored the usual random disturbances emerge as a special case. In other words, the usual assumption of random disturbances results in the least conservative control policy. We analytically characterize the optimal robust advertising strategies and provide an empirical validation of the proposed approach.

4 - Finding the Right Pace: A New Pacing Heuristic for Real-Time Bidding
Uğurcan Dündar, WU Vienna, Vienna, Austria, Nadia Abou Nabout
In real-time bidding, advertisers use different heuristics to allocate their budget over a certain time span (e.g., hours). The most common heuristic is even pacing and aims at an equal spend across time intervals of equal length; it is the standard option in most campaign management systems. Yet, many industry experts claim that waterlevel pacing is a better option: While even pacing tries to spend the same budget even when there are few ad impressions in the market, waterlevel pacing spends the budget according to fluctuations in traffic and exhausts the budget faster when traffic increases. Yet, advertisers use waterlevel pacing rarely as campaign management systems do not offer it as a standard option. The literature suggests profit-maximizing pacing, which maximizes profit over all time intervals and is superior to even pacing or waterlevel pacing in theory. Yet, advertisers might avoid profit maximizing pacing because it is hard to implement, slow and requires accurate prediction of various input metrics for each time interval. In this paper, we aim to identify a pacing heuristic that generates higher profits while being easy to use by practitioners and fast to run. Specifically, we test three new heuristics that extend the idea of waterlevel pacing by accounting for clickthrough and conversion rates. To study their profitability, we first run a counterfactual analysis with different campaign datasets and benchmark the generated profits to profit-maximizing pacing. Then, we run a simulation study, in which we more systematically examine the profitability of those metrics.
FC14
Meeting Room C4.4
Effective Influencer Marketing III: Exploring Influencer Integrity
Invited Session
Special Sessions
Chair: Maximilian Beichert, HEC Paris, Paris, France
Co-Chair: Andreas Lanz, University of Basel, Jouy-en-Josas, France
1 - How Much Influencer Marketing is Undisclosed? Evidence from Twitter
Yanting He, Imperial College London, London, United Kingdom, Daniel Ershov, Stephan Seiler
We quantify the prevalence of undisclosed influencer posts on Twitter across a large set of brands based on a unique data set of over 100 million posts. We develop a novel method to detect undisclosed influencer posts and find that 96% of influencer posts are not disclosed as such. Despite stronger enforcement of disclosure regulations, the share of undisclosed posts decreases only slightly over time. Compared to disclosed posts, undisclosed posts tend to be associated with younger brands with a large Twitter following and are posted from smaller accounts that generate higher engagement per follower.
2 - The Dark Side of Influencer Marketing: How Brand Partners's Corporate Social Irresponsibility (CSI) Scandals Harm Future Brands' Consumer-Influencer Engagement
Christian Hughes, University of Notre Dame, Notre Dame, IN, United States, Yang (Jenny) Guo

The rapid growth of influencer marketing has sparked many studies trying to maximize the benefits of influencer marketing, with little research examining its dark side – the risks posed to influencers and brands. We study how and why a brand partner’s corporate social irresponsibility (CSI) scandal may affect consumer engagement with influencers’ subsequent posts and potential response strategies for brands and influencers. Corporate social irresponsibility (CSI) behaviors do not harm core products or services but violate moral standards (Xu, Bolton, and Winterich 2021), such as animal testing.

Based on the mere association effect, consumer judgment depends not only on its own behaviors but also on its associated cues (Dimofte and Yalch 2011). We propose that although influencers may not be responsible for unethical behavior, they may become entangled with a brand partner’s CSI scandals, leading to a mere CSI association for the influencer and subsequent brand partner. We argue that since morality positively affects inferred intrinsic motivation (Chernev and Blair 2021), a mere CSI association will reduce influencers’ perceived intrinsic motivation and therefore harm consumer-influencer engagement.

An analysis of sponsored influencer posts confirms this negative effect. Experiments indicate that this effect occurs because associating with a CSI brand reduces influencers’ perceived intrinsic motivation, and that the effect weakens when brands frame their CSI behavior as intrinsically (vs. extrinsically) motivated or when the subsequent influencer post highlights the brand partner’s CSR feature. Overall, our findings uncover the role of mere CSI associations due to brand CSI scandals on subsequent consumer-influencer engagement.

3 - Walk the Talk: The Interplay Between Corporate Leadership and Social Media Representation
Amy Pei, Northeastern University, Boston, MA, United States, Keran Zhao, Pankhuri Malhotra
The relationship between a brand's social media strategy and corporate leadership is an emerging area of marketing research. As influencers increasingly shape a brand's narrative on social media, it is important to understand how their representation aligns with the company's leadership. This study investigates the relationship between the racial diversity of a company's corporate leadership and that of the influencers hired for its social media campaigns. Analyzing data from various sources, including social media campaigns, board diversity metrics, and company performance indicators, the authors find that racial diversity in both corporate leadership and social media influencers is associated with higher consumer engagement. However, a discrepancy in these two dimensions of racial diversity can diminish this positive association. Consumer engagement declines when social media campaigns showcase more racial diversity than corporate leadership. Surprisingly, the opposite discrepancy strengthens the positive association. The key findings highlight the need for genuine diversity in corporate leadership that aligns with social media representation. To maximize impact, brands should prioritize board diversity and collaborate with influencers who authentically mirror this diversity.
4 - The Impact of AI Influencers on Human Influencers
Serim Hwang, SKK GSB, Sungkyunkwan University, Jongno-gu, Korea, Republic of, Xiao Liu, Kannan Srinivasan

Brands are turning to AI influencers for cost-effective marketing, avoiding human-associated risks like ethical issues, public conflicts, and scandals. However, little is known about whether AI influencers complement or displace human influencers. Using Instagram data from 2015 to 2020, we find that human influencers previously sponsored by brands using AI influencers face displacement, while those not previously sponsored experience a complementary effect. More importantly, subset of influencers by demographics and behavior characteristics are more susceptible to AI displacement, a finding contrary to existing literature. We propose mechanisms explaining the treatment effects.

FC15
Meeting Room C4.5
Bayesian Analysis of Textual Data
Invited Session
Special Sessions
Chair: H. Alice Li, The Ohio State University, Columbus, OH, United States
Co-Chair: Greg Allenby
1 - Accounting for Formative and Reflective Topics in Product Review Data for Better Consumer Insights
Greg Allenby, Ohio State University, Columbus, OH, United States, Joachim Buschken, Thomas Otter

Observations of product and service reviews suggest that textual product reviews may contain statements that talk about the overall experience (“We had a great time”) or, similarly, whether to recommend a particular product. We argue that such statements encapsulate an overall assessment and hence are not independently informative about, but rather reflect overall ratings. We propose a model that allows for the distinction between topics that contribute to and topics that merely reflect an overall evaluation and apply it to a data set consisting of luxury hotel reviews. Compared to a standard supervised LDA, we find our model to better fit the data and to improve customer insights by resulting in more semantically coherent topics which point at specific attributes with positive and negative relationships to customers’ evaluation of their experience.

2 - Co-Authoring the Consumer Journey - An Integrated Hidden Markov and Topic Model
Liye Ma, University of Maryland, College Park, MD, United States, H. Alice Li

Understanding the consumer purchase journey is crucial yet challenging due to prolonged purchase cycles, unobserved decision-making stages, and shared initiations of touchpoints by both consumers and firms. This paper adopts a novel perspective, treating touchpoints as a collaborative effort between consumers and firms, while the generation of both the firm-initiated touchpoints and consumer-initiated touchpoints are connected to consumers’ latent purchase states. We employ an integrated hidden-Markov model (HMM) and multiple topic models to handle the extensive touchpoints on consumers’ purchase journeys. The HMM helps connect data at various granularity levels, revealing consumers’ latent purchase states, capturing the sequences and dynamics of their touchpoints, and demonstrating resilient results even when the firm loses partial data access. Meanwhile, the topic models capture different types of touchpoints from various sources, reducing data dimensions while retaining meaningful insights. Leveraging data from a multinational software firm, the model effectively summarizes and categorizes touchpoints, uncovering three latent purchase states that offer insights into transition dynamics on consumers’ purchase journeys. This model’s novel perspective of the co-authorship between the firm and consumers and its adaptability position it as an effective tool for extracting actionable insights and enhancing decision-making in analyzing purchase journeys.

3 - Screening Consumer Complaints for Recall Management: A Topic Model for Decision Automation
Wen Shi, Central South University, Changsha, China, People's Republic of, Yujie Qu, Jia Liu

While consumer complaints are recognized as the primary catalyst for product recalls in numerous sectors with high recall rates (such as automobiles, food, beverages, and pharmaceuticals), both firms and regulatory bodies face challenges due to limited human and technological resources when it comes to screening these complaints for trend analysis. Addressing this gap, we introduce a semi-parametric topic model named the hierarchically dual Pitman-Yor process (HDPYP). The HDPYP is designed to automatically process and analyze vast volumes of consumer complaints alongside their associated recall statements. The HDPYP not only extracts defect-related topics but also predicts the significance of each consumer complaint and forecasts the topic distribution of subsequent recall statements. We apply the HDPYP using consumer complaint datasets and vehicle recall data from the U.S. automobile sector. Our findings demonstrate the value of the HDPYP to aid firms and regulators in crucial decision-making processes, such as pinpointing pivotal consumer complaints warranting further examination (or those deemed “summary-worthy” in subsequent recall statements), identifying product defects, and forecasting recall occurrences in advance. Furthermore, by integrating the outputs of the HDPYP with Large Language Models (LLMs), regulators can efficiently and effectively review and authenticate the recall statements submitted by firms.

4 - Using Text Analysis in Serial Mediation Analysis
H. Alice Li, The Ohio State University, Columbus, OH, United States, Judy (Zijing) Zhang, Greg Allenby
The proposed model extracts mediating variables using textual and fixed-point data in a serial mediation analysis. The model is built on the Seeded Latent Dirichlet Allocation (LDA) model and the Grade of Membership (GoM) model and integrates the treatment and outcome variables to estimate the serial mediation effects in one step. Treatment variables affect the topic compositions of the text data, and the topics are mapped to the membership probabilities related to some extreme profiles. The membership probabilities then predict outcome variables. The proposed model integrates text and fixed-point data, extracts mediators that reflect the rationale of decision-making in a more flexible way, and identifies the relationships between mediators.
FC16
Meeting Room C4.6
Marketing of Health Services and Products
Invited Session
Special Sessions
Chair: Jian Ni, Virginia Tech, Blacksburg, VA, United States
Co-Chair: Jiancai Liao, Aalto University School of Business
1 - Improving Healthcare Accessibility and Equity: The Role of Online Health Platforms
Qiyuan Wang, Hong Kong Polytechnic University, Hung Hom, Hong Kong, Charles Weinberg, Chunhua Wu
Online health has become increasingly popular among patients and doctors. In this paper, we investigate whether and how online health platforms improve healthcare accessibility and reduce inequity. We develop a structural demand and supply model for an online health market and incorporate important features of the empirical context, such as patients' price sensitivity and location preference and doctors' economic and prosocial incentives. Our results suggest that, compared with offline healthcare, online health enables patients to have better access to nonlocal healthcare resources. The benefits are greater for underserved patients (i.e., those living in areas with relatively few doctors and hospitals) than for well-served patients, thus mitigating healthcare inequity. Counterfactual analysis reveals the potential for the platform to further enhance these benefits through its pricing design. Compared with doctors deciding their own prices, setting a uniform low price would discourage high-quality doctors from providing services which consequently limits the potential patients' gains. Instead, a two-tier pricing strategy that sets a high price for high-quality doctors and a low one for the rest balances doctors' incentives and patients' affordability, further improving accessibility and equity for patients while having a minimal effect on doctors' surplus.
2 - The Roles of Packaging and Distribution Strategies in Price Competition: The Case of Bottled Green Tea Market
Masakazu Ishihara, New York University, New York, NY, United States, Hiroshi Kumakura, Masayoshi Moteki
In the consumer-packaged goods industry, packaging strategy plays an important role in product differentiation and price competition. Packaging strategy includes a package design, but also container type and size, product information (e.g., health claims) written on packages, variety of packages to offer, etc. Firms choose to distribute certain package varieties of a product through only certain distribution channels. This paper investigates packaging strategy and associated distribution strategy using aggregate-level data from the bottled Green Tea market in Japan. Bottled green tea products are relatively homogenous in terms of taste and benefits, and thus getting packaging strategy right is crucial. They are offered in a variety of package sizes and container types. We collect weekly sales and price data, as well as product-level information (e.g., packaging, health claims), for all SKUs between 2013 and 2022. Three important features of the data are: (1) frequent redesign of packages, (2) information about the stores in which each SKU was available, and (3) new government regulation in 2015 regarding the use of health claims on packages. We propose a new empirical framework in which the demand-side model extends the standard random-coefficients logit model by allowing for latent store group clustering and the supply-side model captures firms’ decisions on distribution and pricing decisions. Using the structural estimates, we conduct counterfactual experiments and examine the degree to which packaging strategy influences price competition and firm profits. Our results provide useful insights for how firms can effectively manage packaging and distribution strategies in relatively homogeneous goods markets.
3 - The Paradox of Abundance: How Enhanced Drug Availability Affects Consumer Compliance
Jiancai Liao, Aalto University School of Business, Helsinki, Finland, Jian Ni, Chuhan Liu
Drug shortages commands substantial global concern and governmental intervention. However, the ramifications of drug availability on patient behavior, particularly adherence to prescribed regimens, remain poorly understood. Our paper studies the impact of heightened drug accessibility on consumer compliance with medical prescriptions. We leverage a policy initiative rolled out in certain targeted municipalities, markedly augmenting the availability of certain pharmaceuticals. Employing a causal inference framework, we uncover a 43% larger expansion in the compliance gap within the cities participating in the policy compared to their non-participating counterparts, indicating a paradoxical decrease in medication adherence following an increase in drug accessibility. Furthermore, our analysis delineates the variable effects of such policies across markets with divergent concentration levels, observing that consumer adherence in more saturated markets exhibits diminished sensitivity to alterations in drug availability. Moreover, through the application of a generalized causal forest methodology, our findings elucidate that while the policy-induced expansion in drug availability substantially dampens consumer compliance, this effect is disproportionately amplified for pharmaceuticals that undergo significant price decreases as a consequence of the policy relative to those with more modest price reductions. These insights contribute to a nuanced understanding of the behavioral implications of drug availability enhancements and underscore the necessity for tailored policy frameworks to mitigate unintended consequences on medication compliance.
FC17
Meeting Room C4.7
Consequences of Firm ESG (Environmental. Social, and Governance) Performance 1
Contributed Session
Chair: Byoungho Jin, North Carolina State University
1 - Escaping the Green Dilemma: Marketing Interventions on Sustainability Adoption
Zoe Ziqi Zhong, London School of Economics and Political Science, London, United Kingdom, Xiaolin Li
This research tackles the "Green Dilemma," focusing on consumer resistance to sustainable product adoption. It examines factors like perceived quality, functionality, pricing, and social influences, using behavioural modeling, topic modelling, sentiment analysis with experimental data to assess attitudes towards corporate green strategies and Corporate Social Responsibility (CSR). The study addresses a gap in literature by exploring eco-positioning's dual impact on consumer behaviour. It contrasts the "Sustainability Liability" hypothesis, suggesting eco-positioning devalues services and reinforces inertia, with the idea that observable CSR enhances consumer utility and reduces inertia. A novel utility function, underpinned by four experimental studies, assesses consumer inertia and eco-positioning's net effect. Key findings include eco-positioning's quantifiable impact in counteracting consumer inertia, especially in long-term service users, challenging the status quo bias. A significant discovery is that a 3.12% reduction in emissions equates to a $1 incentive, aiding strategic CSR investments. The study also highlights privacy concerns in sustainable service adoption, finding eco-positioning's influence in reducing inertia is more potent and less affected by privacy issues than monetary incentives. This insight suggests new research directions on the interplay between privacy concerns, eco-positioning, and consumer inertia, offering vital implications for businesses and policymakers in sustainability and marketing strategies.
2 - How B2B Firms Utilize ESG Content to Engage Social Media Users
Yue Cao, The Hong Kong Polytechnic University, Hong Kong, Hong Kong, Fang Flora Gu, Danny T Wang

Abstract

This study investigates the effectiveness of B2B social media marketing, with a specific focus on the influence of ESG content on social media engagement. Using a deep learning model, we analyzed 23,791 posts shared by 130 B2B companies on their official social media platforms (i.e., WeChat Accounts) between January 1, 2022 and March 2, 2023. Our findings reveal that companies emphasizing environmental and social domains of ESG in their posts experience higher levels of social media engagement. However, a focus on governance-related content appears to reduce engagement. Additionally, we identified the moderating roles of competition intensity and firm size, both of which positively moderate the relationship between ESG content and social media engagement. This study makes a valuable contribution to the existing literature by offering a fresh perspective on B2B social media marketing, specifically highlighting the role of ESG content in affecting social media engagement. Furthermore, by leveraging a deep learning model, we gain a more precise understanding about the differential impacts of E-, S-, and G-related content, thereby enriching the growing body of literature that explores the intersection of ESG and marketing practices. This understanding empowers businesses to develop targeted social media marketing strategies, fostering better interaction and engagement with stakeholders.

Keywords: ESG, B2B social media marketing, social media engagement, deep learning

3 - How Can We Enhance the Effectiveness of Sustainability Claim? Interaction With Brand Types
Byoungho Jin, North Carolina State University, Raleigh, NC, United States, Yoo Won Min, Jiwoon Kim

When advertising sustainability products, companies often employ either self-benefit appeals, focusing on consumer’s comfort, or other-benefit appeals, highlighting societal and environmental advantages. Between the two, which benefit appeal will be more effective? Can self-benefit or other-benefit appeal be applied across all brands? Is self-benefit appeal more suitable for brands emphasizing sustainability (sustainable brands) or fast-fashion brands?

The effectiveness of self- or other-benefit appeals is inconsistent in literature; this posits that is effectiveness is dependent on brand types – sustainable vs. fast-fashion brands – and proposes to test interaction (benefit appeal x brand types) to explain clearly. Sustainable brands prioritize altruism like environmental protection, while fast-fashion brands highlight consumer demands for affordability and trends. Therefore, this study views that sustainability claim can be maximized when sustainable brands use other-benefit appeals and fast-fashion brands employ self-benefit appeals. This study also posits that the sustainability claim depends on message credibility, the perceived truthfulness of communication, and thus employs it as a mediator.

Based on the above, this paper proposes two propositions: 1) Benefit appeals and brand type will interactively affect the purchase intention of sustainable fashion items; Sustainable brands (vs. fast-fashion brands) with other-benefit (vs. self-benefit) will increase the purchase intention, and 2) Message credibility will mediate the relationship between benefit appeals and brand types on purchase intention. This framework recommends that brands tailor their marketing strategies based on their sustainability focus. It enables sustainable and fast-fashion brands to position their advertising uniquely, thereby boosting sustainable fashion consumption.

FC18
Meeting Room C4.8
Platform Analytics
Invited Session
Special Sessions
Chair: Zijun (June) Shi, HKUST, Clear Water Bay
Co-Chair: Shuo Zhang, Shanghai Jiao Tong University, Shanghai
1 - Gender Neutral Marketing
Jiaqi Chen, Shanghai Jiao Tong University, Shanghai, China, People's Republic of, Tong Guo, Zijun (June) Shi, Shuo Zhang
In this paper, we examine the effects of adopting gender-neutral language in product
titles on sales performance. Focusing on a major e-commerce platform in China, we
identify products that altered their titles to be more inclusive of diverse gender groups
across various categories. By comparing sales data of these products with synthetic
controls created from a broad selection of similar items, we demonstrate that gender-
neutral marketing notably enhances sales, particularly for products traditionally aimed
at female consumers, such as cosmetics. Our findings reveal that this increase in
sales occurs without significantly impacting the retention of existing customers, while
simultaneously attracting a considerable number of new users. We further discuss
the broader implications of gender-neutral marketing and investigate the mechanisms
underlying these observed trends.
2 - Measuring Amount of Product Information in Images: Application in Scalable Image Selection
Yuting Zhu, National University of Singapore, Singapore, Singapore, Xinyu Cao
Images contain valuable product information that matters for consumers’ online shopping. How to design and select product images for each product is an important and challenging task for online retailers. In this paper, we propose a method to measure the amount of product information a set of images contains. Using crowd-sourced data, we show that our proposed metric can well capture people’s perceived amount of information a set of images contain. Using field data from a major fashion outlet company in China, we show that our metric is a strong predictor of online sales. Furthermore, we demonstrate how our proposed metric can guide the selection of product images for an online store.
3 - Buy Now, Pay Later Service in the Platform
Buqing Ma, University of Science and Technology of China, Hefei, China, People's Republic of, Bo Zhou
Many firms are increasingly introducing buy-now-pay-later (BNPL) services to assist sellers and facilitate consumer purchasing decisions. This paper investigates how BNPL services affect seller competition, profits, and consumers on two distinct types of platforms: Pure and Hybrid. In the Pure platform model, the marketplace do not sell directly to consumers (only third-party sellers do). By contrast, the Hybrid platform acts as a first-party seller while providing a marketplace for third-party sellers to reach consumers. Consumers who differ in their disposal incomes can choose whether to utilize BNPL services for their purchases. Our findings reveal several impacts of BNPL services. First, the introduction of BNPL can reduce sellers’ price competition, regardless of the platform's type. Second, in a Pure marketplace, BNPL services widen the price discrepancy between competing sellers, whereas in a Hybrid marketplace, this discrepancy can narrow. Third, compared to its absence, BNPL can lead to higher consumer surplus. Additionally, our results indicate that consumers have weakly higher surplus in the pure marketplace than that in the hybrid marketplace.
4 - The Spillover Effects of Copycat Apps and App Platform Governance
Nan Li, Tongji University, Shanghai, China, People's Republic of, Jingcun Cao, Avery Haviv
From taking inspiration to outright copying, new apps are frequently developed using the ideas of successful incumbents. Mobile app platforms, such as the Apple App Store and Google Play Store, have policies intended to screen these “copycat” apps. We study the effect these “copycat” apps have on the demand and behavior of incumbents by leveraging data from multiple Android app markets in China. We first identify which apps copied others using a supervised machine learning approach. Our identification strategy leverages the fact that we observe the same app competing against different competitors on different platforms. Our findings show that copycats tend to be more prevalent in smaller app markets and non-gaming categories. On average, copycats significantly reduce the demand for incumbent apps. This effect is most prominent in larger, highly rated apps. Furthermore, we find that the entry of copycats affects the innovation of incumbents. Incumbents who are exposed to copycats make larger changes to their descriptions and file size, while simultaneously seeing a reduction in their user ratings, and become less likely to use the platforms for subsequent releases. Our results suggest that the filtering of copycat apps confers significant benefits to incumbent app developers but reduces overall downloads of the app platform. This finding may inform recent discussions on the value that app platforms provide to developers.
FC19
Meeting Room C4.9
Social Influence in Marketing
Invited Session
Special Sessions
Chair: Andrew Ching, Johns Hopkins University, Carey Business School, Baltimore, MD, United States
Co-Chair: Shervin Tehrani, University of Texas-Dallas
1 - More, Faster, and Better? Effects of Rewards on Incentivizing the Creation of User-Generated Content
Elisabeth Honka, UCLA Anderson School of Management, Los Angeles, CA, United States, Mahsa Paridar, Mina Ameri
We quantify the effects of rewards from peers and from the platform on the quantity, timeliness, and quality of user-generated content (UGC). Users can receive both monetary and non-monetary rewards. Understanding the effectiveness of different types of rewards provides platforms with guidance on how to design reward systems to encourage desired user content generation. Our unique data come from an online board game platform and consist of information on three types of UGC: initiating forum threads, replying to peers’ questions, and writing game reviews. Our results show that both monetary and non-monetary peer rewards, i.e., tips and likes, lead to more, longer, politer, more complex, more informative, but less timely content. In contrast, compensation and badges, i.e., the monetary and non-monetary reward from the platform, have the opposite effects. We discuss how platforms can incentivize the production of specific types of UGC and how rewards impact the UGC posting behavior of different types of users, e.g., top contributors or original content creators.
2 - Demand Estimation and Policy Design under Network Effects in the Electric Vehicles Market
Jiarui Liu, New York University, New York, NY, United States
I quantify the heterogeneous network effects and price elasticities in consumers' Electric Vehicles (EV) purchases. I consider both direct network effects from social influences and indirect effects from charging stations. Given the estimated effects, I design targeted pricing policies for the firms. To examine the equilibrium effects of counterfactual policies under social influences, I model consumers' decisions jointly as equilibrium outcomes. Multiple equilibria might arise, posing challenges to estimation and identification. I show that if the average social influence effect is within my derived bounds, then a unique equilibrium is guaranteed even under counterfactual policies. I investigate whether the data patterns suggest a unique equilibrium; if so, estimation requires searching for parameters within the derived bounds. Another challenge comes from the endogenous charging stations and prices. I construct instruments for the endogenous variables and show the identification of each effect. Using zip code level vehicles and charging stations data in Texas over six years, I find positive heterogeneous social influence effects; moreover, ignoring social influence effects underestimates the mean price elasticities by 11%. The socio-economically disadvantaged group is less affected by social influences and more sensitive to prices. I design a profit-maximizing targeted pricing policy that charges this group $15k less than the others on average. Under the recommended policy, the average firm's annual sales increase by 35.7% and annual profits increase by 2.7%. The recommended policy improves not only firms' profits but also distributional equity among consumers.
3 - Consumer Interactions and Peer Effects in Socially-Connected Digital Products
Yulia Nevskaya, Queen’s University, Kingston, ON, Canada, Yijun Chen
Social interactions are integral to socially-connected digital products like online games, social media platforms and other virtual spaces. These social interactions are formed and typically measured during product use, making them not independent of outcomes such as user engagement and churn. This endogeneity makes it challenging to measure causal peer effects, which are crucial for decision-making in customer relationship management (CRM). This study tackles the endogeneity of social connections in the marketing strategies of socially-connected businesses. Using data from a popular online game, we develop and estimate a joint model of network formation and peer effects on churn. Our key findings are: (i) Social connections are largely driven by unobserved gamer characteristics. Ignoring the endogeneity leads to underestimating peer effects. (ii) Networks with a wider range of gamer characteristics have less balanced social connections. (iii) The preference for a more or less balanced network depends on the firm’s CRM objectives. A balanced network is favorable for mitigating the negative impact from popular gamer’s churn, while a less balanced network is more favorable for leveraging the positive influence of popular gamers. These findings suggest that companies seeking to utilize peer effects in marketing strategies should consider network composition and endogenous network formation in their product design and community management strategies.
4 - Choosing Music by Your Ears or Eyes? Estimating Consumers' Demand Under Social Influence
Shervin Tehrani, University of Texas-Dallas, Dallas, TX, United States, Ata Jameei Osgouei, Andrew Ching
This paper develops a model of dynamic decision-making, which incorporates both self-learning and learning from others. Unlike previous studies, our unique approach delves into the under-explored area of the interaction between self-learning and observational learning. Leveraging the MusicLab dataset, which captures individual music search and selection behaviors on an online platform, we shed light on the complexities of using popularity disclosure as a noisy signal for product quality. Neglecting this interaction in analyzing clicking and conversation rates may result in biased estimations of consumer preferences. In markets with diverse entertainment options and varying consumer tastes, our approach models how individuals actively update their beliefs using both popularity and self-experience feedback.

In addition to allowing individuals to naively follow others (i.e., irrational herding), we also model individuals consciously try to discern their own preferences from crowd preferences when making their decisions (rational herding). We use our estimated model to investigate how platforms should adapt and share product popularity information to mitigate harmful herding behavior and enhance consumers' use of such information in different scenarios. In our counterfactual experiments, we explore various disclosure policies, including (i) not disclosing any popularity information, (ii) fully disclosing all popularity information, (iii) only disclosing unbiased popularity information, and (iv) employing personalized disclosure through a machine learning approach called "content-based filtering." Our results indicate that personalized disclosure significantly reduces negative consequences due to naive herding, offering a beneficial balance for both platforms and consumers in utilizing social influence such as product popularity.
FC20
Meeting Room C4.10
Visual Content: Insights from Machine Learning and AI 1
Contributed Session
Chair: Peng Wang, Bayes Business School, University City of London
1 - Visual Variation and Consumer Responses to Video Ad
Masayoshi Moteki, Waseda University, Tokyo, Japan
The aim of this paper is to examine the link between visual variation and consumer reaction to video ad. Visual variation is the sum of the differences between frames taken at equal intervals from the video (i.e., the sum of the amount of change in the video). Therefore, the use of visual variation can provide suggestions on how to create video from the aspect of visual stimuli (e.g., the number of scenes and the speed of camera work). Quantitative marketing, computer science, and other fields have conducted research in recent years to predict consumers' responses to video ad. Most of these approaches use Deep Learning or Machine Learning. It is difficult to interpret the obtained features in these approaches. The author's approach overcomes this problem by extracting features that are easy to interpret. There are several features to be considered, among which this study deals with visual variation. By using visual variation, we can incorporate the aspect of visual stimuli into advertisement production. Regression analysis was conducted using visual variation as the explanatory variable and consumers' responses to the advertisements as the objective variable. The results showed that visual variation and its squared term were significant. This result suggests that there is an inverse U-shaped relationship between visual variation and consumers' responses to advertisements, and that visual variation should be neither too low nor too high. This suggests that "just right" visual stimuli are required to stimulate demand.
2 - All Smiles but No Deal: Smile Analytics in Marketing Videos
Ling Peng, Lingnan University, Hong Kong, Hong Kong, Geng Cui, Yuho Chung, Yongfu He

Despite the dynamic nature of facial expressions, extant studies have mainly used static images in experiments, an approach that lacks external validity and accuracy in reflecting observer perceptions. Emotion feigning, a common practice in marketing to align with display norms, effectively engages and influences customers. Unlike genuine expressions, acted ones are produced and perceived asymmetrically, allowing the symmetry of expressions to offer valuable insights into the spontaneity, intensity, and valence of emotions. Yet, this aspect remains underexplored.


Building upon emotional labor research and smile physiognomy, we develop a comprehensive smile analytics framework. It considers both morphological and dynamic attributes of smiles, employing intensity, authenticity, and timing as primary dimensions for evaluating smile properties and their influence on consumer responses. Intensity reflects a smile's valence and arousal, whereas timing captures the spread of smiling expressions during a presentation. Asymmetry, indicating muscular involvement variance between the face's two sides, serves as a morphological indicator of authenticity. Duration, the smile's length, operates as a dynamic marker. Interactions between these dimensions allow differentiation of smile types and effects, enabling detailed analysis.


Our empirical study uses Udemy pitch videos, revealing that smiling faces can negatively impact sales. Face asymmetry and display timing moderate the smiling-sales effect. These findings highlight the nuanced and intricate effects of smiling and the underlying mechanism. The methods used for the automatic extraction and measurement of facial expressions and asymmetry have the potential to advance smile analytics and improve e-marketing research and practice.

3 - All Work No "Likes" The Effects of Visual Narrative Structures on the Performance of Marketing Videos
Geng Cui, Guanddong University, Guangzhou, China, People's Republic of, Geng Cui, Yuho Chung, Ling Peng
Short videos now dominate social media and online marketing. The performance of videos, however, shows a long-tail distribution, as most videos go unfinished in viewing and record only a handful views or likes, largely due to poor editing and production. Unlike full-length movies, marketing videos are shorter with a smaller temporal window. Viewers’ limited attention span and visual processing capacity highlight the critical role of visual narratives. While videos vary in text/speech and music content, they mainly rely on visual narratives – sequences of visual images or events – to tell stories and generally have certain distinctive patterns and structural properties. To analyze video performance at scale, we draw from the visual narratives and visual cognition literature, adopt the state-of-the-art methods to extract the three key metrics of visual narrative structures, i.e., pacing, volatility, and sequencing (time to peak), and examine their effects on the viewing outcomes. Analyses of 48,831 videos from Kickstarter show that pacing has an inverted-U effect on viewing outcomes, while volatility enhances persuasion. Interestingly, pacing has a negative correlation with volatility, suggesting that fast pacing via too many cuts or motions on a constant basis in fact jeopardizes the appeal of videos and viewing outcomes. Contrary to the conventional wisdom, peaking early is superior to climax in the middle or towards the end. These findings provide novel insight into the role of visual narrative structures in videos and furnish meaningful implications for visual computing, video analytics and marketing.
4 - A Study on the Impact of the Short Video Features towards Customer Purchasing Intention
Peng Wang, Bayes Business School, London, United Kingdom
The massive popularity of short video platforms introduces a great business opportunity for the advertising industry. Despite considerable interest from the advertising industry, there are not many research studies on how short-form videos for advertisement impact consumer decisions. This study aims to investigate the critical factors of short video advertisements that impact customer engagement behaviour. Compared to mainstream studies in this field, this study focuses on the unstructured data of short-form videos to quantify the key features. Through the social media platform TikTok data, we construct and evaluate an algorithm to predict product sales and customer engagement. This study has several contributions. First, it contributes to the social media advertisement literature since it explores how video features affect customer behaviour by digitizing those features. Besides, this study enhances marketing research from a methodological perspective by utilizing big data, natural language processing, and advanced deep learning algorithms. This study aims to offer practical insights and managerial implications for sellers and marketers by developing a marketing tool that assists companies in analyzing and predicting customer purchasing behaviour in response to short video advertising.
FC21
Meeting Room C4.11
Large Language Models: Applications 2
Contributed Session
Chair: Liying Qiu, Carnegie Mellon University, Pittsburgh, PA, United States
1 - Can LLM Adoption Promote Greater Equity? Evidence from Online Consumer Activism Petitions
Yanqing Gui, Cornell University, Ithaca, NY, United States, Piyush Anand, Vrinda Kadiyali

While online consumer activism is on the rise, it is unclear if various demographic segments are equally effective in these movements. We examine consumer activism on Change.org, a leading petition-based activism platform. Our study, involving 16,378 petitions from the pre-LLMs era (2009-2022), reveals significant racial and gender disparities in petition performance, as measured by signature counts. To address these gaps, we explore the potential of Large Language Models (LLMs) in revising petition texts. Our approach is to prompt LLM to rewrite petitions in a more persuasive manner. This is a more modest and platform-acceptable approach than unconstrained changes in petitions. We train a Transformer-based model to predict signatures for these counterfactual petitions. Our findings suggest that the platform-side integration of LLM can reduce the racial performance gap by 38%. However, petitioner-side spontaneous adoption of LLMs shows negligible impact on racial equity (1% gap widening), due to unequal awareness of LLMs across demographics. This underscores the importance of platform-side intervention. We also find that the proposed platform-side intervention will marginally change the gender gap (3% widening), where females currently outperform males. We develop a decorrelated Transformer model to attribute LLM’s causal effect to unintended content changes, a potentially problematic feature of LLMs. This model ensures that LLM's improvements do not disproportionately alter information content in petitions from disadvantaged groups. Our methods for quantifying LLMs' causal effect and attributing it to specific changes are generalizable to other research contexts involving LLMs.

2 - Does Big-Data Correlational Analysis Predict Causal Effects of Language on Decisions?
Akshina Banerjee, University Of Michigan, Ann Arbor, MI, United States, Oleg Urminsky

With the proliferation of large observational datasets, correlational “Big Data” analyses are increasingly used in marketing research. In this paper, we rigorously evaluate their reliability, particularly in understanding language's impact on consumer behaviors and evaluate the extent to which the theoretically possible bias from correlational analyses actually occurs in practice. We compare correlational and experimental findings from large-scale datasets of online news platforms. We code headlines for 50 language constructs and 15 content topics, linking these to click-through rates, thus providing insights into how language influences consumer behavior digitally.

In Study 1 on Upworthy.com, using 1,741 experiments and 7,763 headlines, a positive correlation (r=.67, p<.001) was found between causal and correlational analyses. But, this masks considerable differences: 32% of constructs diverged in direction between methods. In the causal model, 42% of constructs were significant, compared to only 4% in the correlational model (all p’s<.05). Only 8% of constructs were significant and aligned in both analyses, with 58% non-significant in both. Study 2 used 68,037 experiments with 205,541 headlines from various news outlets. In it, the correlation between causal and correlational estimated effects was weak (r=.19, n.s.), with disagreements in direction for 38% of the constructs. Additionally, only 14% of the constructs were significant and in the same direction in both analyses, and 24% were non-significant in both analyses.

This research is, thus, essential for marketers, cautioning against over-reliance on Big Data and advocating for a mix of correlational and experimental methods to accurately understand consumer behavior.

3 - Enhancing Customer Service Automation: Evaluating Large Language Models in Addressing Unrecognized Credit Transactions
Marcel Goic, University of Chile / Instituto de Sistemas Complejos de Ingeniería, Santiago, Chile, Daniel Schwartz, Felipe Hernández

Large Language Models (LLM) have revolutionized business operations in several industries. Among the advantages of LLMs are the breadth of knowledge contained in the models that allow them to be extrapolated to new domains with little training and the ability to understand nuances of unstructured language that enable them to provide answers to complex environments. These strengths suggest that LLM can be a powerful tool for automating customer service. However, given the generality of their capabilities, LLMs have been criticized for hallucinating and providing answers that are not properly justified. In the context of customer service, this is a major risk as it can provide incorrect information affecting the quality of service. In this project, we investigate the performance of different training strategies on effectiveness, hallucination, efficiency, and resolvability in the context of banking customers seeking to clarify unrecognized credit transactions, where having control of the responses is particularly sensible for the firms.

4 - How Much Should We Trust LLM Results for Marketing Research?
Liying Qiu, Carnegie Mellon University, Pittsburgh, PA, United States, Param Singh, Kannan Srinivasan
This study evaluates how well Large Language Models (LLMs) such as GPT-4 can replicate consumer decisions in market research, particularly in scenarios where these decisions are influenced by consumer risk preferences. The LLM's decisions were evaluated by comparing them to actual consumer choices when faced with a range of real home insurance plans, each featuring different premium-deductible options. Findings reveal that LLMs tend to avoid extreme options, suggesting a basic decision-making process rather than a deep understanding of consumer risk preferences. Further tests using simpler scenarios to counteract the tendency to avoid extremes revealed that LLMs apply risk preferences in their decisions. However, these risk preferences significantly deviate from those documented in existing literature. For example, LLM outcomes imply that a user might opt to pay $200 extra in premium to decrease their deductible by $500. This contrasts sharply with real consumer behavior, where there's a tendency to pay an additional $100 to reduce the deductible by $500. These disparities could only be justified by unusually high risk aversion coefficients or extreme loss aversion parameters that are far more than what is typically observed in literature. This indicates that current LLMs may not reliably predict consumer preferences in market research and need more refinement for effective use in understanding consumer behavior.
Friday, June 28, 2:50 PM - 4:00 PM
FD01
Meeting Room C2.5
Recommendation Systems
Contributed Session
Chair: Shuxiao Wang, The University of International Business and Economics, Beijing, N/A, China, People's Republic of
1 - Human vs. Machine: Rethinking AI Marketing Strategy in E-commerce
Satadruta Mookherjee, Grenoble Ecole de Management, Grenoble, France, Hanieh Sardashti

As online retailers continue to increasingly rely on automation, marketing researchers must study the impact on consumer behavior and address the potential societal impact. Therefore, in this research, we study consumer interactions with artificial intelligence (hereafter AI), such as algorithmic shopping aids (e.g., product recommendation systems) to shed light on the underlying mechanisms, such as autonomy and decision under uncertainty.

The agency theory (Singh and Sirdeshmukh, 2000) suggests greater dependence of the principal (consumer) on an agent in the presence of uncertainty. This suggests that consumers may depend on an AI-based marketing tool when there is high uncertainty. However, in other situations, the agency of AI-based tools may backfire resulting in loss of agency of the consumer and negatively impacting their shopping experience.

Through a series of randomized between-subjects lab experiments, we examined the mediating role of perceived autonomy in the relationship between algorithmic interaction and shopping aids. Our findings show that consumers dislike being overwhelmed with pages crowded with multiple recommendation systems featuring too many products. However, heightened perceived purchase risk led consumers to rely more on AI recommendations.

The findings challenge the benefit-maximizing assumption that consumers' only motive is to maximize consumption outcomes through AI. The research contributes by addressing the strategic gap in marketing literature to identify factors that influence the shopping experience when consumers interact with AI-enabled shopping aids that can help managers design better AI marketing strategies.

2 - Accelerating Learning-by-Doing through IT-Assisted Recommender Systems: The Case of Taxi Drivers
Shuxiao Wang, The University of International Business and Economics, Beijing, China, People's Republic of, Junhong Chu

We investigate how taxi drivers make location choices through learning by doing using a unique data set of detailed trip records and trajectories. We assume that drivers choose where to search for passengers by solving a dynamic programming problem and update their belief about demand conditions in a Dirichlet learning fashion. Our findings suggest that taxi drivers improve their productivity mainly by learning from experience and making more informed decisions about where to search for passengers. Our counterfactual analyses reveal that new drivers experienced a significant daily earnings loss due to their inaccurate prior belief about the demand conditions. To mitigate this, we propose recommendation systems that guide new drivers to locations with the best earning opportunities and find this can accelerate their learning and reduce their earnings loss.

FD02
Meeting Room C2.6
Game Theory for Channels
Contributed Session
Chair: Cheng Hsun Hsieh, Ming Chuan University, Taipei - 250 Zhong Shan N. Rd., Sec. 5, Taipei 111, Taiwan, Taipei City
1 - Strategic Demand Forecasting: Inefficiencies and Remedies
Yue Li, Shandong University, Jinan, China, People's Republic of
This paper builds a game theoretical model of a supply chain that faces uncertainty in the demand for a perishable good. A supplier chooses a wholesale price, and a retailer can strategically conduct costly data analytics to forecast demand before choosing its inventory level. We identify two distortions in the decentralized supply chain. First, due to the producer's markup on the wholesale price, the retailer faces an inefficiently high cost of overstocking. Second, since the producer takes only a fraction of the total channel profit, the retailer faces an inefficiently low cost of understocking. The first (second) distortion results in an inefficiently low equilibrium inventory and an inefficiently high (low) forecast accuracy. In aggregate, the retailer stocks an inefficiently low inventory level and may acquire too much or too little information compared with the channel optimum. These inefficiencies can be resolved with existing supply chain practices such as (1) a joint implementation of buy-back and data sharing, (2) a store-within-a-store business model, or (3) vertical integration. Our results also generate insights on the pricing of retail data and the regulation of vertical mergers.
2 - How to Incentivize Showroom Service? From Performance-Driven to Process-Driven
Hongchen Duan, Huazhong University of Science & Technology, Wuhan, China, People's Republic of, Jun Yang, Yi Xu, Bo Zhou

Abstract: Due to the significant effects on reducing returns and attracting consumers, showroom is becoming a popular strategy for online retailers to provide physically experience service and delivery product information. As pure service providers, the absence of order fulfillment and sales function reinforces the agency relationship between online retailers and offline showrooms, requiring retailers to design compensation and incentive programs. We develop a salop model imbedded with a principal-agent model to study the incentive contracts design between online retailers and showrooms. We first discuss the selection of service modes for showroom in a centralized system. A showroom with premium- (basic-) service such as flagship (community) store will be build when return cost is high (low). Second, we design the equilibrium contracts with performance-driven linear incentives. Interestingly, retailers in the decentralized system with incentive contracts are better off than those in the centralized system, when return cost is high. This is because the competition mitigation effect dominates the agency cost effect. We thus further develop a process-driven contract menu with step-compensation, which draws on data and algorithm to infer showroom service efforts. This new contract is always more incentive efficient than the centralized system. Our work provides insights into how omni-channel retailers achieve incentives and synergies across channels, and provides a groundbreaking discussion of the role of data-driven in process monitoring and organization incentives.

Keywords: Showroom, Incentive Mechanism, Contract Design, Principal-agent, Data-driven

3 - Free Delivery or Low Commission Rate: The Online-Food-Delivery Platform’s Optimal Delivery Fee And Commission Rate
Lu Hsiao, National Chung Hsing University, Taichung, Taiwan, Hui Xiong, Cheng Hsun Hsieh

Food delivery platforms have spread across several countries and played an important role in recent years. On these platforms, the food sellers serve their consumers and pay a proportion of the revenue from each sale to the platforms as a commission. Moreover, the consumers have to pay the delivery fees to the platforms for the delivery-to-door services. These platforms face decisions regarding their commission rates and delivery fees. In this paper, we investigate the roles of the commission rate and the delivery fee in the channel coordination of the online food delivery industry. We build a game theoretical model to study the interaction between the commission rate and the delivery fee decisions in the online food delivery industry. Moreover, we implement numerical experiments to illustrate the strategy choices.

We find that the platform should charge consumers no fee for the delivery-to-door service in the monopolistic scenario. In contrast, the consumers may have to pay for the delivery services when there is competition between sellers. Thus, the platform should offer a low commission rate to eliminate sellers’ operation costs and charge a high delivery fee without exacerbating the price competition when products are quite substitutable. In addition, the within-group positive externality increases the competition between sellers and the incentive for the platform to charge no commission and collect high delivery fees. Therefore, the sorting function used to find the best-selling products and the sales volume information on the platform can increase the competition and influence the platform and the sellers’ profits.

4 - Channel Choice in the Presence of Heterogeneous Consumer Channel Preferences
Cheng Hsun Hsieh, Ming Chuan University, Taipei, Taiwan, Lu Hsiao
Compared to the traditional resale channel, where the manufacturer sells the product through the retailer, agency selling alleviates the channel's double marginalization. In contrast, the reselling channel offers the manufacturer greater pricing control. Thus, determining which channel yields a higher profit for the manufacturer is crucial. To illustrate the manufacturer's channel choice, we present an analytical model. Specifically, we build a model consisting of a manufacturer, a platform, and a retailer, which leads to three channel choices: Reselling, marketplace, and dual-channel (combination of reselling and marketplace). First, our results indicate that when the commission rate is exogenous, the reselling channel can be optimal for the manufacturer. However, when the commission rate is endogenous, the marketplace becomes a more attractive option. Second, competition plays a different role in the performance of the manufacturer and the retailer, depending on whether the commission rate is exogenous or endogenous. Lastly, we show that moderate consumer channel preference can result in lower total channel profit. Our study provides insights into the channel structure selection problem, highlighting the importance of balancing double marginalization costs and pricing control for manufacturers.
FD03
Cockle Bay Room 1
Generative AI in Marketing: Opportunities, Risks, and Practical Applications
Invited Session
Special Sessions
Chair: Alex Burnap, Yale University, New Haven, CT, United States
1 - Generative Interpretable Visual Design: Using Disentanglement for Visual Conjoint Analysis
Alex Burnap, Yale University, New Haven, CT, United States, Ankit Sisodia
This paper develops a method to automatically discover and quantify human-interpretable visual characteristics directly from image data. Our method is generative, and creates new "ideal point" visual designs for targeted consumer segments. The approach developed here builds on "disentanglement" methods in deep learning using variational autoencoders, which aim to discover underlying statistically independent and interpretable characteristics constituting an object's visual representation. While the deep learning literature shows that supervision with "ground truth" characteristics is required to obtain unique disentangled representations, these are typically unknown in real world applications and are in fact exactly the characteristics we aim to discover. Our method instead uses readily available structured product characteristics as supervisory signals to enable disentanglement, and discovers and quantifies human interpretable and statistically independent characteristics without domain knowledge on the product category. The approach is used with a dataset on watches to automatically discover interpretable visual product characteristics, obtain consumer preferences over these characteristics, and generate new ``ideal point" visual product designs targeted to specific consumer segments.
2 - The Power of Generative Marketing: Can Generative AI Create Superhuman Visual Marketing Content?
Yannick Exner, Technical University of Munich, Munich, Germany, Jochen Hartmann, Samuel Domdey
Generative AI’s capacity to create photorealistic images has the potential to augment human creativity and disrupt the economics of visual marketing content production. This research systematically compares the performance of AI-generated to human-made marketing images across important marketing dimensions. First, we prompt seven state-of-the-art generative text-to-image models (DALL-E 3, Midjourney v6, Firefly 2, Imagen 2, Imagine, Realistic Vision, and Stable Diffusion XL Turbo) to create 10,320 synthetic marketing images, using 2,400 real-world, human-made images as input. 254,400 human evaluations of these images show that AI-generated marketing imagery can surpass human-made images in quality, realism, and aesthetics. Second, we give identical creative briefings to commissioned human freelancers and the AI models, showing that the best synthetic images also excel in ad creativity, ad attitudes, and prompt following. Third, a randomized field experiment with more than 173,000 impressions demonstrates that AI-generated online ads can compete with professional human-made stock photography, achieving an up to 50% higher click-through rate than a human-made image. Collectively, our findings suggest that the paradigm shift brought about by generative AI can help advertisers produce marketing content not only faster and orders of magnitude cheaper but also at superhuman effectiveness levels with important implications for firms, consumers, and policymakers. To facilitate future research on AI-generated marketing imagery, we release “GeneratedImageNet” that contains all of our synthetic images and their human ratings.
3 - Find the Perfect Fit: Leveraging Generative AI to Optimize Visual Brand Presence in Synthetic Influencer Imagery
Julia Rosada, University of Hamburg, Hamburg, Germany, Maximilian Witte, Jochen Hartmann, Mark Heitmann

In the dynamic digital marketing landscape, influencer marketing has become a highly sought-after strategy, particularly on visual platforms like Instagram. However, the process of content creation inherently creates tension, particularly when determining the degree of prominence allocated to the influencer compared to the brand in these visuals. To investigate this tension and explore the variation in the placement of the sponsored products, we collect more than 1,400 brand-related influencer images, originating from Instagram, covering 447 brands. Based on 14,200 user ratings, we indeed find substantial variation in the perceived brand and influencer presence, suggesting that finding the right brand balance in sponsored visual content is a nontrivial task. To remedy this issue, we propose a solution that leverages recent advances in generative artificial intelligence. Utilizing a refined latent text-to-image diffusion model, we generate photo-realistic influencer images that showcase an optimized brand presence while preserving the influencer's distinct visibility. Our conducted study indicates that the content created achieves significantly higher performance ratings in terms of conveying the brand’s identity and individual presence compared to original collaborations. The integration of generative artificial intelligence in influencer marketing holds the promise of evolving into a scalable, time-efficient, and cost-effective approach to content creation. We conclude by discussing the impact of generative AI on brand-influencer collaborations, where trade-offs between different objectives need to be minimized.

FD04
Cockle Bay Room 2
Meet the Editors 2
Invited Session
Chair: Valentyna Melnyk, UNSW, Sydney
1 - JMR
Kapil Tuli, Singapore Management University, Singapore, Singapore, Singapore
Journal of Marketing Research
2 - JM
Pradeep Chintagunta, University of Chicago, Chicago, Illinois, United States
Journal of Marketing
3 - IJRM
Koen Pauwels, Northeastern University, Newton, Massachusetts, United States
International Journalof Research in Marketing
4 - JSR
Markus Groth, UNSW, SC, Australia
Journal of Service Research
5 - JAMS
Rajendra Srivastava, Indian School of Business, Hyderabad , India
Journal of the Academy of Marketing Science
FD05
Meeting Room C3.1
Moving Beyond Return on Marketing: New Perspectives on the Marketing-Finance Interface
Invited Session
Special Sessions
Chair: Arvid Hoffmann, University of Adelaide, Adelaide, Australia
1 - Does Advertising Create or Just Inflate Firm Value?
Hauke Wetzel, University of New South Wales, Sydney, Australia, Harald van Heerde, Ljubomir Pupovac
Advertising has an indirect effect on firm value via sales growth and profit (customer route) and a direct positive effect (investor route), yet this direct effect is not well understood. While advertising could help investors to price a firm objectively by signaling fundamentals, it could also lead investors to misprice (overvalue) the firm because the persuasive effect advertising has on customers could “spill over” to investors. To understand whether the effect is fundamental or due to mispricing, the authors apply a new framework from finance that decomposes market-to-book value into a fundamental firm value and an overvaluation component. Employing rigorous econometric models to a panel data set for 1506 firms with more than 20,000 observations spanning more than 20 years, the authors show that advertising only has a positive significant effect on fundamental firm value if firms have a CMO. In line with the signaling argument, CMOs can channel advertising into fundamental firm value. The findings also demonstrate that the effect on firm overvaluation is significant and positive when firms are more diversified and when cash flows are less predictable. This is consistent with the notion that advertising spillover effects on investors are greater if investors have more opportunities to encounter a firm as a customer (i.e., when firms are more diversified) or when investors have to rely on their experiences as a customer to a greater degree (i.e., when firms have more unpredictable cash flows). The findings have important implications for marketing researchers, investors, firms, and policy makers.
2 - Shareholder Litigation Risk and Product Recalls
Arvid Hoffmann, University of Adelaide, Adelaide, Australia

Research examining the antecedents instead of consequences of recalls is relatively sparse and has not considered whether firms’ likelihood to recall products is influenced by legal changes that could induce managerial opportunism. Drawing from agency theory and a business ethics perspective, the authors develop a conceptual framework proposing that the recall decisions of publicly-listed firms reflect the outcome of a trade-off between managers’ private incentives to try and avoid a recall and their risk of being sued by shareholders and held personally liable for damaging the firm when doing so. To examine this notion empirically, the authors exploit the staggered adoption of universal demand laws across U.S. states as a quasi-natural experiment affecting managers’ exposure to shareholder litigation risk. Using a difference-in-differences analysis, they find that when it becomes more difficult for shareholders to sue managers for alleged wrongdoings, firms subsequently become less likely to recall products. This effect is weaker in the presence of organizational mechanisms constraining managers’ self-interest-seeking behavior, such as a corporate culture focused on customer interests or the exercise of normative control through monitoring by institutional investors. The authors do not find support for a potential alternative explanation of operational improvement and therefore higher product quality driving their findings.

FD06
Meeting Room C3.2
Loyalty Programs
Contributed Session
Chair: Andrew Petersen, Penn State University, 457 Business Building, State College, PA, 16802, United States
1 - A Rising Tide that Lifts All Boats: The Effects of Collective Recognition Programs on E-Commerce Sellers and Regional Economic Development
Wei Miao, University College London, London, United Kingdom, Liqiang Huang, Chris Tang, Xianchu Xu
In this paper, we examine the causal effects of a collective promotion program, Taobao Town Certification, on sellers’ business performance and inclusive economic development.
Using the synthetic difference-in-differences approach, we find that sellers in TTC-certified towns experienced substantial growth with a 32.1% increase in annual sales and a 20.8% increase in the annual number of orders. The cohort-specific treatment effects reveal that the benefits of TTC are more pronounced for later cohorts. We also examine the heterogeneity of treatment effects and find both demand-side mechanisms (through enhanced seller visibility) and supply-side mechanisms (through increased selling efforts). We find that sellers in less economically developed regions with a higher level of collectivism receive more substantial benefits; female and younger sellers derive more benefits from TTC. Overall, our results imply that TTC can reduce regional economic disparities and support inclusive economic development. Finally, we examine the spillover effects from TTC-certified towns to uncertified neighboring towns and discover positive externalities. Our findings have important managerial implications to e-commerce platforms for supporting and sustaining inclusive economic development in rural areas.
2 - Effort Begets Business: On the Effectiveness of Appreciation Letters in Inducing Repeat Purchase
Yan Zhang, National University of Singapore, Singapore, Singapore, Zheshuai Yang

Consumers tend to switch frequently between online retailers. To boost repeat purchases, retailers employ a variety of strategies. One commonly used approach involves enclosing an appreciation letter in the shipping package. This research investigates the effectiveness of this practice. A two-wave laboratory experiment found that handwritten letters significantly increased the likelihood of repeat purchases compared to printed letters or no letters, as they signalled a higher level of effort from the seller. Interestingly, in another experiment, we further demonstrated the role of perceived seller effort by showing that letters with less-refined calligraphy exerted a stronger positive influence on repeat purchases than aesthetically pleasing ones, because consumers inferred that the former conveyed sellers’ personal effort, while the latter were crafted by external hires and therefore signals a professional persuasion tactic. The results of a randomized field experiment that tracked the actual purchases of 2000 consumers over six months mirrored the laboratory findings. In this field experiment, we found that compared with the control condition, letters with bad calligraphy could increase repeat purchase count by 0.375 and letters with ordinary calligraphy by 0.164, whereas letters with good-looking calligraphy were not effective. In addition, printed letters and letters produced by copy machines were ineffective. We also demonstrated that compared to the control condition, the handwritten letter with bad-looking calligraphy significantly increased the profit for the seller. Taken together, this research uses both lab experiments and field experiments to show that handwritten letters, particularly those signalling the seller’s personal effort, effectively boost repeat purchases.

3 - Investigating Multi-level Effort-reward Program Designs with Pinch Points
Andrew Petersen, Penn State University, State College, PA, United States, Yashar Bashirzadeh, Robert Mai, Ali Nadalizadeh
Multi-level effort-reward programs are prevalent in many physical and digital marketplaces. Examples include contexts such as tiered loyalty programs, gamified services, and online gaming. In a multi-level effort-reward program, customers invest a series of efforts (e.g., spending time, spending money, and/or inviting friends) to achieve higher levels/tiers to earn monetary and/or non-monetary (e.g., badges) rewards. In this research, we study the impact of pinch point (vs. smooth) effort-reward program designs on customer engagement using the context of freemium games. Pinch point multi-level effort-reward program designs consist of a series of gentle effort-reward levels followed by a much steeper effort-reward level. Smooth multi-level effort-reward program designs consist of a series of steadily increasing effort-reward levels. In this research, we conduct a series of lab and field studies to understand the impact of pinch point (vs. smooth) multi-level effort-reward programs on customer engagement. We find in a lab study of consumers and managers that smooth designs are generally preferred. However, unlike conventional wisdom suggests, our field data shows that users in the pinch point (vs. smooth) multi-level effort-reward program do not churn any faster over time. Further, users in the pinch point (vs. smooth) multi-level effort-reward program are more likely to in-app purchase, especially at the steeper levels due to the momentum accumulated in the gentle levels. This multi-study research also provides insights into the underlying process and provides evidence for the generalizability of the proposed inertia-based framework for the commonly used multi-level effort-reward programs across physical and digital marketplaces.
FD07
Meeting Room C3.3
Public Policy
Contributed Session
Chair: Yuxin Su
1 - The Geographical Spillover Effect of a House Purchase Deregulation Policy: Evidence from A Policy Experiment in China
Zhe Lin, Xi'an Jiaotong-Liverpool University, Suzhou, China, People's Republic of, Fernando Moreira, Sheng Zhao

We use a policy that removes restrictions to property purchases in a sub-district of Suzhou City, China as a natural experiment to investigate the potential spillover effect of a house deregulation policy on the house prices and the transaction quantity in an area not directly impacted by the policy but geographically near the region targeted by the policy. Apart from the exogeneity of the policy adoption in the context of our study, the geographical proximity between the regions considered and the unanticipated nature of the policy help to further reduce the influence of omitted factors in our Difference-in-Differences analyses. We find that the listing prices and transaction prices in the area surrounding the deregulated district significantly decrease as a result of the policy while the number of houses sold significantly increases. These findings can be explained by the response of sellers in the neighboring area: with the cancellation of purchase restriction in the policy-targeted area, residents who were not eligible to buy houses in Suzhou City before become eligible to buy properties in the deregulation-targeted district. Such change reduces the future demand for the properties in the neighboring area, which motivates sellers to reduce the listing prices of their properties and accept lower transaction prices based on their updated demand belief. Our conclusions and the pertinence of the method used are supported by robustness tests while additional tests shed light on the mechanisms driving the spillover effect identified.

2 - The Pass-through of Retail Crime
Johannes Kasinger, Tilburg University, Tilburg, Netherlands, Carl Hase
This paper shows that local market prices in the Washington State cannabis industry increase in response to (organized) retail crime, uncovering an often overlooked aspect of retail crime's social costs. Our identification strategy exploits the quasi-experimental variation from the timing of store-specific crime incidents to establish causal links between crime and local market outcomes. Utilizing detailed scanner data and crime records between 2018 and 2021, we find that victimized stores increase prices by around 1.5 percentage points on average post-robbery, relative to stores in distant non-victimized markets. Competitors in affected markets increase the prices by around 1.2 percent with a two-month delay. In contrast, local demand and wholesale costs in victimized markets are unaffected, suggesting that price increases result from increased security spending. Price adaptions of nearby stores seem not to be strategic but rather the result of spatial ``knowledge'' spillovers. Our heterogeneity analyses show the largest effects in independent stores, for less price-sensitive goods, and in more competitive markets, shedding light on the underlying mechanisms and the influence of market structure in the pass-through of retail crime costs.
3 - Unveiling the Dynamics of Collective Recycling: Insights from a Large-Scale Educational Recycling Project in Shenzhen
Yutong Wang, Nanjing Audit University, Nanjing, China, People's Republic of, Yuxin Su, Wenyu Dou, Song Gao, Yulin Gong, Zheyu Gu, Peixuan Yang, Yitong Zhu

The findings reveal that areas with extensive promotion of the recycling program achieve significantly higher recycling scores compared to less promoted areas, indicating the effectiveness of promotional efforts. Analyzing scores over 47 months, the dynamic panel regression results show a positive correlation between individuals’ scores in consecutive months and find that students exhibit moral licensing when they receive rankings in the previous month, leading to decreased recycling scores. Conversely, teachers' recycling behavior is negatively influenced by school-level ranking recognition, implying a negative impact of goal achievement on their recycling behavior.

Examining various types of recycled goods, the study identifies that milk carton recycling demonstrates greater stability compared to plastic and paper recycling. Received rankings leading to an increase in recycling tendencies for paper recycling, but a decrease in the other goods. Furthermore, the study uncovers additional determinants influencing individual recycling behaviors, including the education level and gender of the school president, as well as school's student-teacher ratios.

Our findings offer valuable insights for policymakers in designing recycling initiatives. Tailoring interventions based on rankings, participants' identities, considering transaction costs associated with different types of goods, and addressing school-level characteristics can contribute to more successful and targeted recycling campaigns, thereby fostering sustainable behaviors within communities.

Keywords: Collective recycling behaviors, Environmental sustainability, ESG

FD08
Meeting Room C3.4
Marketing & Education
Contributed Session
Chair: Rebecca Chae
1 - Who Follow the Policy? Decoding Varied Responses to the Double-Reduction Policy in China’s Publishing Industry
Yuqing Wang, Renmin University of China, Beijing, China, People's Republic of, Ju Mao, Yanlai Chu, Tana Wang

Abstract: Using data from a major technology company covering 80% of sales in the Chinese publishing industry, this study employs a staggered Difference-in-Differences approach to analyze the effects of the Double Reduction policy, aimed at reducing students’ homework and off-campus training burden. Results reveal significant decrease in the sales of teaching reference books. Proximity to Beijing moderates the policy’s impact, with the effect weakening as the distance increases. Moreover, provinces with high competition level in the National College Entrance Examinations experience non-significant changes in sales of teaching reference books. These findings offer practical implications for managers and policy makers.

Key Words: Double Reduction policy; China’s publishing industry; Staggered DiD

2 - The Impact of Diversity in Tech-Assisted Online Learning
Chen Lin, Fudan University, Shanghai, China, People's Republic of, Yuxin Chen, Jeongwen Chiang

Distance learning have created greater inequality for socioeconomically disadvantaged groups (Parolin and Lee 2021). This research finds that such effect is more salient and detrimental when diversity in virtual classroom composition is overlooked. It is well documented that diversity matters in the traditional classroom, as racial, ethnic, gender and other multiple relationships between teachers and students will affect student performance, especially for disadvantaged groups (Dee 2005), one potential mechanism of is through observable social cues and categorization (Mortensen and Hinds 2001). However, existing studies in the online setting seem to show mixed evidences and concerns for endogeneity (Knippenberg et al. 2004). In distance learning or virtual group setting, teachers and students may not know each other's real identity or background. We use a large dataset from a leading online EduTech company, which provides video records of nearly half a million randomly-formed live classes delivered to 250,432 students by 4991 teachers in year 2021. We find that class diversity itself (especially geographic dispersion) hinders learning outcome and creativity in virtual learning, even though teachers’ and students’ backgrounds are unobserved to each other. However, teachers’ support and diversity experience can help mitigate the negative outcome, especially for the socioeconomically disadvantaged students. Managerial and policy implications are discussed for educators on designing effective and inclusive online classes.

3 - The Impact of Social Comparison in Online Learning: Evidence from a Large-Scale Field Experiment
Rebecca Chae, Santa Clara University, Santa Clara, CA, United States, Xuebin Cui, Xiaojing Dong, Banggang Wu
Online learning has seen tremendous growth, yet encouraging students to stay engaged and finish the course remains challenging. Through a large-scale field experiment on an online learning platform in China, this paper examines whether social comparison information can improve remote learners’ engagement. It exploits data of 7,691 users (76,910 observations) who register for a national accounting certificate exam and take online courses in preparation for the exam. Results suggest that in contrast to users without social comparison, users who engage in downward comparisons with others making less study progress increase their engagement. However, those who engage in upward comparisons show reduced motivation to engage with online learning. This impact of social comparison is more pronounced when users receive information about their classmates rather than information about all users. While the overall impact of social comparisons, especially those with classmates, is strong on engagement, it diminishes near the exam date. Furthermore, there exists a strong spillover effect of this intervention: encouraging users to engage in social comparisons in one course influences their engagement not only in that course but also in another course users take to prepare for the certificate exam. However, the pattern of the spillover effect appears to vary with changes in the course context. These findings offer practical implications to marketers and consumers on how to enhance motivation in online learning.
FD09
Meeting Room C3.5
Qualitative Consumer Research
Contributed Session
Chair: Reyaz Jeffrey, University of Newcastle
1 - Tailoring the Future: Development of Customer Centric Mult Experience Framework for Enhanced Personalization and Customization
Sahil Dua, FORE School of Management , New Delhi, India, Shilpi Jain, Varsha Khattri
Modern businesses are transforming user interactions through a multiexperience paradigm, incorporating diverse devices and sensory modalities to facilitate seamless transitions and address psychological needs. The surge in experiential goods, propelled by information and communication technology advancements, presents challenges for UX designers and marketers.
Measuring multidevice experiences proves challenging due to device diversity, user behaviors, and varied data sources. Traditional metrics struggle to capture the intricate nuances of the multidevice user journey. To address this, our research employs a multi-method exploratory approach to develop a comprehensive framework. This framework, grounded in Experience theory, Embodied Cognition theory, and Schema theory, seeks to identify unique aspects of multiexperience, providing businesses with a tool to optimize key dimensions and enhance the overall multiexperience.
Utilizing a qualitative methodology, we collected data from diverse sources, including extensive literature analysis, content analysis using NLP on 150,000 Google Play store reviews, and 15 semi-structured interviews with UX designers. Our findings contribute to the existing UX framework with the introduction of 5Cs (consistency, continuity, complementarity, convenience, and customisation) as new constructs signifying customisation and personalisation dimensions.
This study is pivotal for both theory and practice. The framework improves conceptual clarity, serving as a foundation for academic discussions on user interactions and technological advancements. It categorizes unique dimensions of multiexperience, providing a nuanced perspective. Given the dynamic nature of multiexperience, marketers can use the framework to adapt their strategies in real time. This agility is crucial in responding to evolving consumer expectations and technological advancements.
2 - Communicating Sustainability Considerations to Customers where there is Organizational Complexity
Mark Uncles, University of New South Wales, Sydney, Australia, Melissa Barden, Adrian Payne

Abstract. We investigate the strategic challenge of communicating sustainability considerations when the marketing organization is multi-leveled and multi-divisional. In such complex organizations, tensions are common and often paradoxical (e.g., scientific accuracy versus consumer understanding and engagement, or transparency with customers to build trust versus drawing attention to unresolved concerns). We present empirical studies and discuss the theoretical and methodological challenges of this type of research. Specifically, findings are based on: (i) extended-interviews with respondents from multiple marketing organizations where environmental sustainability is of major concern, and (ii) in-depth case analysis of a single, complex marketing organization where environmental sustainability is in the spotlight at an industry level.

Contributions. We: (a) examine the tensions and paradoxes that exist at multiple levels in complex organizations (including in and across marketing departments, at different organizational levels, and at the industry level); (b) unpack the organizational logics that influence marketing decision-making (such as decisions about whether and how to feature environmental sustainability in customer value propositions or CVPs); and (c) consider how marketers use communication choices to (attempt to) manage paradoxical sustainability tensions (notably through trade-offs, spatial separation, temporal separation, integration, and accommodation).

Keywords. Communicating sustainability. Customer value propositions. Paradox theory. Tension management. Organizational logics.

3 - Understanding Australian Muslims' Home Loan Consumption Decision-Making; A Qualitative Study
Reyaz Jeffrey, University of Newcastle , Newcastle , Australia, Philip Rosenberger III, Guilherme Pires

This qualitative study explores an under-researched area: the drivers of Muslim consumers' uptake of home loans in Australia. Drawing upon the theory of planned behaviour (TPB) and literature, antecedents were investigated to determine attitude, subjective norms, and perceived behavioural control influencing intention and behaviour. Data was gathered via semi-structured, in-depth interviews with 17 Muslim community leaders on their perceptions of the identified antecedents influencing and driving Muslim consumer intention and behaviour in choosing between conventional and Islamic home loans.

Findings suggest that, despite knowledge of and concern for Islamic tenets about the consumption of banking and financial products, including home loans, most Muslims viewed conventional home loans (CHL) as being permitted, essential or innocuous. Reasons for such views include interpretation of Islamic tenets, limited awareness of available Islamic home loan (IHL) offerings, convenience factors related to accessing CHL offerings, service-quality levels, availability of technology-based services such as Internet banking and the influence of social groups and communities. As such, it is posited that religiosity had overarching implications across all antecedents, although higher religiosity did not necessarily translate to a higher uptake of IHL.

This research has theoretical and managerial implications. Theoretically, it identifies factors influencing decision-making related to home loans by Australian Muslims, contributing a theoretical framework to investigate Muslims' consumption of financial products. Managerially, this research helps marketers understand attitudes, subjective norms, and behavioural factors related to Muslim consumption of home loans, which can facilitate the development of financial products and marketing strategies that better appeal to Muslims.

FD10
Meeting Room C3.6
Search Engine Marketing
Contributed Session
Chair: Subramanian Balachander, University of California, Riverside, Riverside, CA, 92521, United States
1 - Image Distinctiveness
Qingli Zeng, Singapore Management University, Singapore, Singapore, Sandeep Chandukala, Ernst Osinga

This study introduces and focuses on the concept of image distinctiveness. Image distinctiveness is defined as the degree to which an image differs from the set of images in a given context based on the scope of assessment. We operationalize image distinctiveness based on the concept of informativeness which, in turn, relies on measures of similarity. We empirically demonstrate the effect of image distinctiveness on consumer search. Our data, from a leading Southeast Asian real estate platform, includes over 8,000 search sessions from more than 2,000 users. We find that image distinctiveness significantly impacts user behavior and this effect varies with the user's channel of origin and their search history. Specifically, image distinctiveness is positively associated with click-through rates for new users and users from channels such as display and social media ads, while returning users and users from direct visits or search engines tend to avoid distinctive images. These effects are economically significant: enhancing image distinctiveness from the 25th percentile to 75th percentile increases click-through rates for new users from social media advertisement by 46% (from 5.30% to 7.74%), but decreases click-through rates for returning users coming from search engine by 30% (from 12.84% to 8.98%). This research offers a novel perspective on the role of images in consumer search by considering images in the context of other images instead of in isolation.

2 - Search Advertising, Price, and Quality: The Role of Price and Quality in Advertisers' Search Advertising Decisions
Yichen Wang, Singapore Management University, Singapore, Singapore, Qi Yu, Ernst Osinga, Sandeep Chandukala

This study focuses on how product quality influences firms’ search advertising strategies and price sensitivity. We conduct several field experiments with an online property trading platform in South-East Asia. The platform provides three distinct advertising tools that allow agents to adjust the positioning of their property listings within relevant search results. Agents face the critical task of selecting which listings to advertise and determining the most appropriate advertising instruments to use. In the experiment, we categorize properties into high and low-quality groups based on various quality metrics. We then systematically vary the prices of search advertising tools within each quality groups. We find that the property agents are responsive to price changes of search advertising tools. Moreover, their sensitivity to advertising prices are heavily influenced by property quality. In addition, we collect clickstream data from property seekers and study property seekers’ response to search advertising. Our study thus sheds light on advertisers’ complex interplay between quality assessment and advertising decision-making. It also offers valuable insights for platforms on how to recommend effective search advertising strategies, which can help retain experienced advertisers and boost search advertising revenue.

3 - Multilevel Channel Competition and Sponsored Search Advertising
Subramanian Balachander, University of California, Riverside, Riverside, CA, United States

Manufacturer and retailer decisions on undertaking sponsored search advertising are complicated by numerous considerations including intra-channel competition for advertising to consumers. In this paper, we use an analytical model to explore how competition at the manufacturer and retailer levels interact with consumer decision journey considerations to determine the sponsored search advertising decisions of manufacturers and retailers.

FD11
Meeting Room C4.1
Models for Targeting and Personalization 2
Contributed Session
Chair: David Huang, Harvard Business School, Boston, MA, United States
1 - Can Transactional Metrics have Mental Representations? Evidence in Purchase Frequency Patterns
Rajesh Sinha, Indian Institute of Management-Indore, Indore, India
Purchase frequency (PF) is a metric that counts the number of transactions from a customer which occurred during a unit time. PF is an important descriptor of customer characteristics and aids customer classification. Although PF often is considered as metric; this research shows PF patterns as customers’ response behaviour. This research performs quantile regression on a panel data of eatable basket and explains PF responses as outcomes of the interplay among three transactional constraints (i.e., mental budget constraint, shopping cost constraint, and holding constraint). The observed PF response patterns are found to be different at the different relative levels of PF itself, which the research explains through the lens of transactional constraints interplay. In this research, the transactional constraint based explanation of PF patterns underpins on four well-known transactional metrics (i.e., amount spent per transaction, total amount spent, units purchased per transaction, total units purchased). The underpinning transactional metrics based plausible explanation of the observed PF patterns evidences the internal mental representations of external transactional metrics. This research contributes to the understanding of PF as a response behaviour and can be applied for designing appropriate targeting plans for PF enhancements.
2 - Synergizing Experiments: Designing Personalized Marketing Interventions through Incrementality Representation Learning
Ta-Wei Huang, Harvard Business School, Boston, MA, United States, Eva Ascarza, Ayelet Israeli

In the pursuit of personalization, companies aim to tailor marketing offerings to match individual customer preferences. Traditional methods typically consist of two stages: first, a set of predefined interventions are tested and evaluated across various customer segments; second then, the most effective intervention for each segment is assigned. Although these methods provide some benefits of data-driven personalization, they often fall short in achieving complete customization, constrained by the capacity to test only a few interventions within a set number of segments. Furthermore, these methods do not provide insights for designing new interventions or for targeting different segments, thereby restricting the scope and potential of data-driven personalization strategies.

This research introduces a novel approach enabling companies to design personalized interventions more efficiently by capitalizing on historical experimental data. We develop a flexible causal machine learning framework, termed incrementality representation learning, which estimates the conditional average treatment effect (CATE) based on intervention characteristics and customer covariates and extracts low-dimensional representations of these features to capture the heterogeneity in treatment effects. This approach allows companies to leverage past experiments to address crucial questions, such as identifying the most effective type of intervention for any specific customer segment or determining which segments are most likely to respond to new, untested interventions. Leveraging 549 marketing campaigns conducted through an online reward platform, we show that our framework significantly improves targeting efficiency and offers substantial insights for designing new marketing interventions.

FD12
Meeting Room C4.2
MultiChannel: Driving Sales
Contributed Session
Chair: Yiwei Wang, Zhejiang University
1 - Enhancing Offline Retail Sales: The Differential Impact of Instant Apps on Pure Physical and Hybrid Stores
Sha Zhang, University of Chinese Academy of Sciences, Beijing, China, People's Republic of, Zecheng Fan

To tackle the issue of transaction abandonment due to mandatory app installations, instant apps (also known as mini-programs in China) have emerged as a solution. These apps are a novel mobile application format, allowing users to access a condensed version of an app without installation, thereby transforming consumer interactions with digital services and blurring the lines between online and offline commerce. Despite their popularity, evidenced by over 500 million daily active users in China and 3 billion uses on Google Play, the empirical impact of instant apps, particularly on a firm’s offline channels, remains unexplored. This study categorizes brick-and-mortar stores into pure physical and hybrid stores, the latter integrating with third-party delivery platforms like Doordash. We obtained a unique dataset from a large retailer, encompassing 175 stores and 149,712 transactions from 01/01/2021 to 10/31/2021. Employing the PSM-SDID model, we find that adding instant apps generally boosts offline sales, though the mechanisms differ between pure physical and hybrid stores. In pure physical stores, an increase in offline sales was due to larger order sizes after instant app usage. Conversely, in hybrid stores, the addition of instant apps led to a rise in the frequency of offline orders. Shopping utility theory provides a framework to hypothesize and interpret these findings. Furthermore, our research confirms that store size and delivery range positively moderate the relationship between instant app introduction on offline sales. This study underscores the strategic importance for both pure and hybrid stores to consider adding instant apps into their channel portfolios.

2 - The Impact of Bundle Variety on Consumer Purchasing
Jeeyeon Kim, La Trobe University, Kaohsiung City, Taiwan
The bundling strategy is commonly used in digital retail platforms and offers advantages in encouraging them to purchase a variety of products. This study examines how digital retailers can effectively manage multi-channel and single-channel customers using different types of bundling strategies. It analyses the purchasing data of 45,561 customers in the cosmetics sector across online and mobile channels. Specifically, it investigates how customers who purchase variety bundles and non-variety bundles diversify their purchases and the speed at which they make these purchases. The results provide substantial findings. Firstly, variety bundles are more effective than non-variety bundles at expanding the range of purchases beyond bundled products. Second, multi-channel customers purchase a wider variety of products compared to single-channel customers, but they do not acquire products more quickly or in larger quantities. Third, the purchase of non-variety bundles by multi-channel customers diminishes its effectiveness. These results provide new insights into the behavioral outcomes of digital channel purchases based on bundling formats and offer valuable recommendations for enhancing the efficiency of multi-channel customer management for retailers.
3 - The Value of Curated Boxes: Evidence from an Omnichannel Fashion Retailer
Yiwei Wang, Zhejiang University, Hangzhou, China, People's Republic of, Lauren Lu

Curated boxes offer customers a carefully chosen assortment of products, allowing them to purchase desired items and return the rest. While this channel strategy enhances customer shopping experience, its impact on retailer profitability remains largely unexplored. Employing field experiment data from an online omnichannel fashion retailer, we investigate how briefly using curated boxes impacts customers’ purchase behavior in the online sales channel and their renting behavior in the subscription rental channel. We observe that, during the treatment period, curated boxes boost omnichannel total sales due to (1) direct purchases from curated boxes and (2) demand spillovers onto other channels. In the online sales channel, treated customers’ purchase quantity increases slightly, whereas their purchase variety, i.e., the number of categories purchased, expands significantly during the treatment period. In the subscription rental channel, the treated customers’ renting quantity and variety increase, implying reduced profitability for the retailer. During the posttreatment period, demand spillovers amplify in the online sales channel but decline in the subscription rental channel. We provide suggestive evidence to support mechanisms behind these consumption patterns. Depending on their pretreatment rental and online purchase frequencies, demand spillovers induced by curated boxes vary across customers, based on which we propose a data-driven strategy to improve the value of curated boxes by targeting specific customer segments.

FD13
Meeting Room C4.3
Game Theory for Online Reviews and Consumer Search
Contributed Session
Chair: Dina Mayzlin, University of Southern California
1 - Searching for Rewards
Xu Zhu, The Chinese University of Hong Kong, Hong Kong, Hong Kong, Tony Ke, Jiwoong Shin
Loyalty programs are a commonplace in many markets. It allows enrolled members to receive rewards based on their past purchases and redeem these rewards for future purchase. Based on a repeated ordered search framework, we propose a new theory for loyalty programs in a frictional market. Consumers search for product information sequentially over many firms, each of which offers two independent products in two periods. Consumers make a purchase from one firm in each period. We find that firms use both price discounts and rewards to attract consumers, where price discounts deter consumers from continuing to search other firms in the current shopping trip, while rewards enable the firm to gain prominence in consumers’ next shopping trip. Consequently, as the search cost gets higher, firms offer lower price discounts but higher rewards. Overall, loyalty programs always decrease industry profit and raise/reduce consumer surplus when the search cost is high/low.
2 - Fake It Until You Make It? For How Long? The Dynamic Impact of Online Fake Reviews on Competition and Reputation
Zhiyao Zhang, Shanghai Jiao Tong University, Shanghai, China, People's Republic of, Peng Liu, Suguo Du
User-generated content on online platforms, such as Dianping and Yelp, can be utilized by consumers to make decisions. Therefore, firms are incentivized to recruit fake reviews to manipulate the overall ratings to enhance their market share. In this paper, we examine how firms use fake reviews to gain consumers and boost their sales. Since consumers treat reviews as a quality signal of the product, we split the quality signal from all reviews into several aspects, including the real quality signal, the signal from the positive fake reviews and the signal from the negative fake reviews. We propose a theoretical model to investigate firms’ online review manipulation strategies, which focuses on consumer attraction and consumer retention induced by fake reviews. Furthermore, we construct a dynamic version of the game theory model that captures the variations of quality signals caused by fake reviews, enabling us to dive deeper into the fake review strategies. The empirical evidence of this study is derived from the real firm data crawled from Dianping. We leverage the current sybil detection method implemented by a machine learning model to distinguish fake reviews. The results of the study can explain the review manipulation strategies between firms on online platforms.
3 - Strategic Presentation of Consumer Reviews
Dina Mayzlin, University of Southern California, Manhattan Beach, California, United States, Leif Brandes, David Godes
The literature on online reviews has documented extensively the effect of different review presentation formats on consumer behavior. However, far less is known about the reasons why such formats differ across platforms. This paper addresses this question and investigates the relationship between a platform's strategy, in the form of its business model, and its incentives to provide review-based insights in different formats. Specifically, we analyze a platform in order to derive its optimal solicitation, processing, and presentation of consumer reviews. We envision the platform in this regard as an intermediary receiving information from previous consumers and transforming it into actionable data for prospective buyers. The platform has control in how it gathers, transforms and displays data. For example, the platform can control the precision of the multi-item scales when it gathers feedback and can decide whether to include multi-media submissions, direct Q&A, and machine learning-enabled summarization. While some of these interventions clearly benefit the consumers, we observe a great deal of variation in the extent to which platforms invest in them, including for example the amount and richness of available information provided to prospective buyers. In order to propose an explanation for this variation, we allow for a range of different platform business models, from an advertising model to third-party retailers to direct sellers. We then investigate the extent to which these models give rise to differing incentives for both gathering and displaying consumer feedback data.
FD14
Meeting Room C4.4
Effective Influencer Marketing II: Content Dynamics
Invited Session
Special Sessions
Chair: Maximilian Beichert, HEC Paris, Paris, France
Co-Chair: Andreas Lanz, University of Basel, Jouy-en-Josas, France
1 - Does Tipping Affect User-Generated Content Creation? Evidence from a Natural Experiment
Shijie Lu, University of Notre Dame, Granger, IN, United States, Ruibin Geng, Sha Yang, Xi Chen

Allowing consumers to make voluntary payments or tips to contributors has become an increasingly popular practice on user-generated content (UGC) platforms to incentivize content generation. In this research, we study whether and how the introduction of tipping affects the quantity and novelty of UGC. We address this question by using the launch of the tipping function on Sina Weibo, the largest microblogging site in China, as a natural experiment. We define users on Sina Weibo as the treatment group and users from a similar blogging site without the tipping function as a control group. Utilizing coarsened exact matching and the difference-in-differences model, we find a dual effect of tipping on UGC creation: it increases content novelty, measured by the degree of differentiation from both the creator’s past work and peers’ concurrent work, but reduces content quantity. Further, the treatment effect on content quantity and novelty is more pronounced for content creators who are more extrinsically motivated, more established on the platform, and more efficient in acquiring social recognition. These findings are generally consistent with the notion that tipping affects content novelty by aligning consumers’ appreciation for new information with content creators’ economic incentives to satisfy consumer demand.

2 - Influencer Branding Over Lifecycle
P. K. Kannan, University of Maryland, Bethesda, MD, United States, Lingling Zhang, Guo Bai
Influencer marketing has gained increasing importance worldwide. While brands’ strategy has received much attention, limited research exists on the strategy and implications of influencers. On social media platforms, influencers should be viewed as a business managing their own personal brand. On one hand, influencers post organic content to grow and engage with followers; on the other hand, they need to monetize the fandom to remain sustainable. Therefore, it is important for influencers to manage the congruence between their personal brand and the collaborating business brands. In this research, we empirically examine the influencers’ branding strategy using a large-scale dataset. Our data include more than 2 million records from over 40,000 influencers on the largest social commerce platform in China. We adopt the Bidirectional Encoder Representations from Transformers (BERT) to analyze the semantic similarity among posts. Our initial findings reveal that influencers can maintain a brand image by sticking to a consistent theme in their organic posts; the similarity between organic and ad posts have impacts on follower engagement of the ads; and influencers can develop a co-branding strategy by choosing which ads to sponsor. Our research has implications on how to optimally match influencers and brands as well as on how influencers should do branding throughout their lifecycle.
3 - Recommendation Algorithms on User-Generated Content in Social Networks: the Quantity-Quality Dilemma
Ziwei Cong, Georgetown University, Washington, DC, United States, Jia Liu

This paper investigates the role of recommendation algorithms in shaping content creation in online social networks. Our study leverages a quasi-experiment conducted by Zhihu, the largest knowledge-sharing platform in China. Zhihu originally used a content-based filtering algorithm, which recommends content to users based on the topics to which each user has subscribed. After more than a year, Zhihu moved to a social filtering algorithm, which recommends content with which users’ social connections are engaged. We find that the intervention reduced the volume of overall answer contribution on the platform by 23%, reduced the number of answers per question by 59%, and increased the answer response time by 188%. However, the intervention increased users’ perceived quality of answers by 4.9%, as measured by the ratio of upvotes to total votes of answers. We explore possible explanations for this quantity-quality tradeoff. We find that the scale of the user social networks (measured as a user’s follower size), is typically smaller than the topic subscriber networks (measured as a topic’s subscriber size), resulting in a narrower content reach under the social filtering. However, the social filtering achieves a better match between users and content, primarily due to the high homophily of the social network on Zhihu; that is, users connected on the network often share similar interests and expertise. These findings highlight that the social filtering algorithm can be a double-edged sword that leads to a narrower but potentially more precise content reach, depending on the characteristics of a platform’s online social network.

FD15
Meeting Room C4.5
IV-free Causal Inference in Marketing
Invited Session
Special Sessions
Chair: Rouven Haschka, University of Cologne, Cologne, Germany
1 - A New Instrument-free Method to Correct for Endogeneity in Time Series With Mechanistic Causes and Random Confounders
Florian Dost, Brandenburg University of Technology / Alliance Manchester Business School, Cottbus, Germany

This research introduces a novel approach to correcting endogeneity in observational time series data. All correction methods rely on untestable assumptions, such as the existence of instruments or specific distributional assumptions. This new method offers assumptions different and complementary to existing methods, particularly those that are instrument-free. The approach relies on Takens’ Embedding Theorem for attractor representations of deterministic dynamic systems and derives an unconfounded predictor empirically from past data of the potentially endogenous variable. The testable and untestable assumptions of this approach are different from existing approaches and practical: The predictor must contain a mechanistic, and hence predictable, element from an exogenous process (its own, or other antecedents such as instruments), while potential confounders should be random.

Simulations conducted in this research demonstrate the capabilities and limitations of the approach. While the approach may fail if the predictor is too random (which can be detected), or worse, if a confounder is mechanistic (which cannot be detected), an extension with added seasonal fixed effects has the potential to address this issue. The proposed approach is robust to various distributional changes that could bias other methods, such as the Gaussian copula or latent instrumental variable approach, and is also robust to weak or confounded instruments that can bias traditional instrumental variable corrections. In conclusion, the new approach presents an addition to the existing toolkit of methods for correcting endogeneity in observational time series data, and has the potential to provide accurate results in cases where other methods fail.

2 - Overcoming Endogeneity and Sparse Data Bias in Consumer-level Random-Coefficient Discrete Choice Models Using Copulas
Fan Yang, NEOMA Business School, REIMS, Marne, France, Xixi Hu, Yi Qian, Hui Xie

The most popular method for estimating discrete choice demand models for data with unmeasured product characteristics is the two-stage product-market controls approach (hereinafter PM approach), initially developed by Berry (1994) and Berry, Levinsohn, and Pakes (1995); with subsequent extension to consumer-level data such as Berry et al. 2004, Goolsbee and Petrin 2004, Chintagunta, Dube and Goh 2005. The PM approach calculates the product-market fixed effects in the first stage and use the instrumental variables to correct for endogeneity of unmeasured variables in the second stage.

We develop an alternative instrument-free control function approach. By simply adding generated regressors to control for the effects of unmeasured product characteristics, our approach is both easily implemented and applicable to other models that the PM cannot work (e.g., Bajari et al. 2007; Fox 2008; Hendel and Nevo 2006). Moreover, the PM approach requires estimating the product-market fixed effects, thus can suffer from curse of dimensionality. The large product-market dimensionality can cause high computational cost (Berry et al. 2004, Dube et al. 2012, Goolsbee and Petrin 2004, Petrin and Train 2010, Reynaerts et al. 2012), and the data sparsity, limited number of purchase occasions of certain brands in certain market, can also lead to significant finite-sample bias in the product-market fixed effect estimates, making the PM approach sensitive to close- to-zero market shares (Berry et al. 2004, Petrin and Train 2010). Our control function approach avoids estimating these high-dimensional product-market fixed effects and thus is free from these computational and sparse data bias issues.

3 - Set Identification of Price Elasticities of Demand - A Narrative Sign Restriction Approach
Rouven Haschka, University of Cologne, Cologne, Germany, Helmut Herwartz
Providing insights into both consumer behaviour and market dynamics, information about the price elasticity of demand enables companies to optimize pricing strategies and improve their overall competitiveness. As prices and quantities are determined jointly, the empirical evaluation of price elasticities of demand requires a sound understanding of structural (i.e., truly causal) relations. We suggest using structural vector autoregressive models endowed with consensual sign restrictions to causally identify price elasticities, and analyse the demand for beer at a large scale for 1,953 stores in the US. For many stores (i.e., at least 40%), recent endogeneity-robust single-equation methods such as IV-based identification, Gaussian copula endogeneity correction, SORE, etc., yield positive price elasticity estimates that oppose behavioural fundamentals and, hence, are hardly informative for the design of pricing strategies. Instead, weak sign restrictions coupled with intuitive narrative restrictions imposed on the functioning of sales promotions turn out to effectively sharpen the identified sets of price elasticities, resulting in elasticity estimates that are sufficiently precise to improve the design of retail pricing strategies.
FD16
Meeting Room C4.6
Product and Packaging Design
Contributed Session
Chair: Jiae Kim, The Ohio State University, 4780 Holmes St, Dublin, OH, 43016, United States
1 - Using Deep Learning and Eye Tracking to Optimize Package Design
Jingling Yu, HKUST, HongKong, Hong Kong, Ralf van der Lans, Bertram Shi

Package design is an important marketing mix element that plays a critical role when consumers visually search for products. Because package design is a highly subjective and complex process, it remains difficult for firms to quantitatively assess the effectiveness of designs, and consequently, consumers often struggle to find their desired product in cluttered environments, especially under time pressure. Our research aims to develop a methodology that assists managers in enhancing package design to improve product findability while maintaining aesthetic appeal. Our contributions are threefold: 1) We develop a dynamic model that describes the eye movements of shoppers while searching for a product on a shopping website. As model input, we use automated machine learning to extract product features. This approach allows us to understand how packaging design affects visual attention processes during search. 2) We illustrate how the parameter estimates of our eye-tracking model in combination with generative machine learning tools can be used to improve package design, and 3) We validate our findings in a follow-up experiment, which illustrates how firms can create appealing and attention-grabbing packaging supported by a data-driven approach.

2 - Aesthetic and Technological Dimensions in New Product Development: An Exploration from the Dual Perspectives of Firms and Consumers
Brian Yiu, Hong Kong

Background:

Despite the extensive body of research on new product development (NPD) as a performance driver, particularly within the technological domain, there is a paucity of empirical studies investigating the pivotal role of aesthetical product development in driving firm performance. This study addresses the research gap by exploring the roles of both aesthetical and technological product development from the perspectives of firms and consumers simultaneously. The conceptual framework highlights both aesthetical and technological drivers could be equally important meditating factors to affect business performance.

Methods:

The firm-level study utilizes a secondary dataset of 600 Chinese manufacturers listed at Shanghai and Shenzhen Stock Exchanges between 2013 to 2022 from CSMAR. The OLS regression model is deployed, incorporating 5-year lagged terms to ascertain explanatory and interaction effects amongst factors leading to firm performance. The consumer-level studies of two focal product arms respectively in headphones and sport-shoes are conducted. The first study is in-depth interviews to identify the key attributes. The second study, conjoint experiment with 201 respondents, quantifies the influences from aesthetical versus technological attributes. The third study is the real focal product trial experiment to investigate consumers’ psychology toward willingness-to-buy.

Results:

The preliminary results by the conjoint analysis show aesthetical and technological attributes have played equivalent roles in buying decision.

Conclusions:

This study could likely shed light on the role of aesthetical as well as technological product development in mediating the effect of NPD on business performance.

3 - Consumer Heterogeneity in Archetypal Analysis: The Hierarchical Dirichlet Process Prior and Its Application to Product Line Design
Jiae Kim, The Ohio State University, Dublin, OH, United States, Greg Allenby, YiChun Liu, Peter Kurz

Consumer preferences are inherently diverse, and understanding their heterogeneity is crucial in marketing research. The diversity of demand and the wide variety of product features make the design of product lines challenging. Evaluation of marketing opportunities when there are many usage contexts and product features requires integration of information on what and when features are demanded, when they are demanded and by whom. We propose a Bayesian nonparametric approach to an archetypal analysis that combines data on the context of consumption, alternative products used, and feature preferences. We incorporate the Hierarchical Dirichlet Process (HDP) Prior to model consumer heterogeneity in mixed membership models and choice models that include fixed-point rating-scale responses and conjoint responses. Our presentation illuminates how the HDP prior reveals latent consumer archetypal characteristics, and how this provides a richer understanding of the underlying factors influencing consumers' product preferences and choices. Our modeling effort better responds to consumers' needs by providing them with greater welfare as measured by maximum attainable utility. This leads to useful product line design and managerial decisions.

FD17
Meeting Room C4.7
Consequences of Firm ESG (Environmental. Social, and Governance) Performance 2
Contributed Session
Chair: Menghan Wang
1 - Sustainable Marketing: A Comprehensive Analysis through Bibliometric and Topic Modelling
Kirtika Deo, University of Technology Sydney, Sydney, Australia

Sustainable marketing is becoming one of the fastest-growing areas for research; however, a detailed review of the topic is limited. This study employs bibliometric analysis to provide a rigorous review and assessment of sustainable marketing trends using publications from the past two decades. Moreover, topic modelling was carried out to strengthen the bibliometrics findings by using Latent Dirichlet Allocation (LDA) to review papers in large quantities for transparent, reliable, faster and reproducible findings. The bibliometric analysis has confirmed that social and environmental dimensions of sustainable marketing in the current literature are widely researched compared to the economic dimensions. The temporal variations based on modelled topics highlight the latest trends in the literature related to sustainable marketing are broadly focused on green products, sustainable development and corporate social responsibility. The study only includes peer-reviewed literature, excluding industry and government reports published elsewhere but presents a novel methodology which can be replicated easily for additional insights for marketing research.

2 - Sustainability Orientation: A Meta-analysis of its Antecedents, Consequences and Situational Contingencies
Bipul Kumar, Indian Institute of Management Indore, Indore, India

Literature conceptualizes Sustainability Orientation (SO) as a strategic resource with dynamic capabilities that help firms attain superior competitive advantage and better performance (e.g., Claudy et al., 2016; Roxas et al., 2017; Khizar et al., 2022). Researchers explored SO from the perspective of marketing, stakeholder, and entrepreneurial orientations (Kohli & Jaworski, 1990; Lumpkin & Dess, 1996; Berman et al., 1999).

Exploring the conceptualization of SO, this study develops a framework identifying antecedents and consequences of SO, including some situational contingencies. We test this framework meta-analytically as meta-analysis estimates the true population effect size with the help of an integrative view of inconsistencies in earlier studies and allows investigation of the effect of contextual factors using moderation analysis (Wilson & Lipsey, 2001; Kumar et al., 2022). Following an extensive literature search, this study uses effect sizes from 42 independent studies for meta-analytically testing the relationships on antecedent and consequence sides of SO.

The findings after meta-analysis provide an integrated view of SO and suggest marketing orientation, top management support, and stakeholder integration as some of the antecedents of SO and marketing performance measures such as brand equity, market share, and a few financial performance indicators like return on investment as consequences of it including a few moderators like competitive intensity. This study contributes to the literature on the interface of sustainability orientation and marketing by providing an integrated framework and also provides managerial and policy insights, such as SO leading to superior brand equity.
3 - ESG Incidents and Firms' Response on Social Media
Menghan Wang, University of Hong Kong, Hong Kong, Hong Kong, Chu (Ivy) Dang, Roni Michaely
This paper examines whether, when and how firms respond to negative ESG incidents on social media. Based on the analysis of Twitter posts from S&P 500 firms and Top 500 brands, we find that firms did not tweet more, but their tweets received more user engagements in the days following the incidents. The effect is more pronounced when the incidents are more severe and widespread, among firms which are more exposed to risk such as climate change, and during the period when ESG attracts greater attention. The effect is also stronger for firms in consumer-oriented industries, when firms are active on Twitter and attract higher user engagement prior to the ESG incidents. We also observe a spillover effect among peer firms within the same industry. Furthermore, textual analysis of tweet contents reveals diverse response strategies employed by firms. Our study provides insights into public relations and brand management strategies.
FD18
Meeting Room C4.8
Platforms/Two-Sided Markets: Insights from Machine Learning and Quasi-Experiments
Contributed Session
Chair: Bhoomija Ranjan, Monash Univesity, Caulfield North, Australia
1 - Does Polarizing Content Pay Off?
Shunyao Yan, Santa Clara University, Santa Clara, CA, United States, Klaus Miller

This paper examines consumer demand for politically polarizing news content using individual-level panel data on a major European news website. We first develop textual measures of political polarization in a multi-party-political system from parliament debates, political parties’ media releases, and election programs. We then characterize how consumers click on and subscribe to these polarizing news articles. We identify two sets of instrumental variables to resolve the potential endogeneity of news coverage and news consumption. The first set of variables exploit the news website's sudden paywall introduction, which is exogenous to and limits consumers' news consumption. The second set of variables capture "suspenseful" and "surprising" news events whose results are uncertain in nature, such as national election. Using these two sets of instrumental variables, we assess how polarizing news articles impact consumers' news consumption as measured by the page impressions generated on the website and subscriptions. We find robust evidence that polarizing news articles negatively affect consumer subscriptions. However, articles polarizing to the political left positively impact the page impressions of consumers more likely to hold right-leaning political beliefs, which is inconsistent with confirmation bias. Overall, our results highlight the different impacts of polarizing content on the tension between news media subscription and advertising revenue.

Keywords: Content Monetization, Digital Media, Media Bias, Confirmation Bias

2 - Now or Book: Service Order Time Dilemma
YItong Chen, Tianjin University, Tianjin, China, People's Republic of, Zhongbin Wang, Yunchuan Liu

In recent years, third-party on-demand food delivery (OFD) platforms have gained popularity among traditional restaurants seeking to expand their customer base and time-constrained individuals searching for an alternative to dining in-person. However, the inherent delays associated with OFD can negatively impact the overall dining experience. To address this issue, the platform has introduced a reservation order service. This study examines the effects of rider pick-up times set by OFD on consumers, platforms, and merchants. We establish a service system where the OFD's selection of the rider's meal pick-up time is influenced by factors such as product failure rate and system congestion level. The findings indicate that while implementing a booking service benefits OFD themselves, it may not necessarily be advantageous for consumers and restaurants alike. When customer orders are expected to be delivered within less congested periods, it is advisable for the platform to allow riders more flexibility in picking up meals later while restaurants should consider offering reservation order services. Conversely, during peak delivery times when customer orders are anticipated to face higher congestion levels, it is recommended for platforms to prioritize early meal pick-ups by riders while restaurants may opt against providing booking order services. Finally, we propose an innovative service mechanism that grants restaurants flexibility in deciding whether or not to offer reservation order services in order to maximize operational profit.

3 - Service Provider Performance in the GIG Economy: Role of Commission Structures
Bhoomija Ranjan, Monash Univesity, Caulfield North, Australia, Benedict Dellaert, Yongdong Liu
In the sharing economy, the platform’s key role lies in creating value by facilitating matches between service providers (supply) and prospective consumers (demand). Platform profits depend on the simultaneous growth of both demand and supply sides of the market. To maintain this, multiple platforms have recently introduced subscription services for customers and service providers (e.g. Uber One, DashPass). Subscription services incentivize demand/supply actors to provide regular (e.g. monthly) payment to the platform in exchange for better service (e.g. lower delivery fee for customers) or superior job matches for vendors.In this project, we explore the role of commission structures in determining downstream service provider performance. We exploit a quasi-experimental variation in service provider commission structures using a unique dataset from an Indian sharing economy platform that introduced a subscription model for one of its offered product categories. The subscription model changed the commission structure from a fixed rate (per-job) to a flat monthly fee in 2018. This changes the incentives faced by the vendor and consequently affects their performance. Our results indicate that subscription causally reduces the demand uncertainty faced by service providers. In equilibrium, average demand to each vendor also increases. We posit two explanations for this phenomenon. First, service providers work longer (more jobs per day) and more intensively (more days per week) to recover the cost of subscription. Second, vendor competition increases leading to churn. Our results provide meaningful insights for sharing economy platforms in designing optimal incentive structures to maintain demand and supply forces in equilibrium.
FD19
Meeting Room C4.9
Multichannel: Spatial Aspects
Contributed Session
Chair: Chao Ma, Xiamen University, Fujian, N/A, China, People's Republic of
1 - Product Availability in Online Grocery Retailing: Can Online Retailing Reforest Food Deserts?
Rafael Becerril Arreola, USC Darla Moore School of Business, Columbia, SC, United States, Yilian Du
Online retailing holds the promise of equalizing consumer access to healthy foods across geographies and socioeconomic groups. We assess the extent to which this promise is honored by quantifying the availability of grocery products offered through online retailing across geographies and product categories. We collect and analyze data from a leading multi-channel retailer in the United States, which, like many others, offers last-mile delivery services. We find that the average probability of a particular product to be available through last-mile delivery services is only about 49%. The degree of availability from warehouse fulfillment (i.e., shipping) is mostly determined by product-related factors, whereas the degree of availability from store fulfillment (i.e., home delivery) is mostly determined by geographic factors. Geographic variation of consumer preferences seems to play only a minor role. Instead, infrastructure seems to be a dominant restriction in the case of home delivery whereas product attributes that hinder long-distance transportation are dominant constraints for shipping. The retailer offers fewer products in designated food deserts, but the difference is only about 3.6%, partly because of the broad reach of shipping services.
2 - Profiling BOPIS Orientation From Multiplatform Behavioral Data
Syagnik Banerjee, University of Michigan, Grand Blanc, MI, United States, Fareena Sultan, Moeen Butt
One of the trends in global retailing is buy-online-pick-up-in-store (BOPIS), which has accelerated post-pandemic. Approximately 70% of US retail consumers have utilized buy-online-pick-up-in-store (BOPIS) more than once (Swanson, July 14, 2022). When the transaction is displaced to a remote virtual platform rather than in-person, the retail store’s ability to influence at the point of contact, moves from a synchronous face-to- face experience, to that of a remote entity, via a digital platform. Several scholars have assessed the customer’s propensity to engage in BOPIS via the Technology Acceptance Model (TAM), measuring performance expectancy, trust, and compatibility with BOPIS shopping, hedonic motivation, and social influence through survey methodologies (Han, Jang, & Shin, 2020; Han, Kim, & Jang, 2019). However, in this research, our key research question examines how we can predict remote to in-person transaction from secondary behavioral data. We draw insights about the impact on BOPIS from a database of more than 300,000 stores in Ohio from 2019, in the Pre-Covid era. The database contains data from a variety of retail outlets, such as restaurants and other eateries, grocery stores, general merchandise stores, and clothing stores. We utilize store-level data from store visitors’ foot traffic, dwell times in stores, and online/physical cross-visit patterns, to predict the percent of remote to in-person transactions. The findings can help store managers better plan engagement with their clientele, based on the mix of behavioral, geographic, and demographic variables.
3 - Spatial Competition with Online Platforms: An Empirical Analysis of the Wealth Management Product Market
Chao Ma, Xiamen University, Xiamen, China, People's Republic of, Shuoxun Zhang
We study wealth management products (WMPs) issued by banks and find higher prices online than offline. The explanation is: banks must charge lower prices at physical branches to compensate customers for traveling costs. The online-offline price gaps decrease with the share of online-banking users, the number of incumbent banks, and a bank’s branch density and clustering. WMP markets provide an ideal scenario to identify the effect of spatial competition on the online-offline spread because online and offline channels for issuing WMPs (selling financial products) do not differ in delivery, search, or waiting costs or the convenience of inspecting product quality.

Keywords: Digital economy, Digital finance, Spatial competition, Online prices, Banking

FD20
Meeting Room C4.10
Visual Content: Insights from Machine Learning and AI 2
Contributed Session
Chair: Chaehyeon Lee, 50, Yonsei-ro, Seodaemun-gu, 03722
1 - Consumer Reactions to AI-Generated Visual Content
Chaehyeon Lee, Yonsei University, Seoul, Korea, Republic of, Sue Ryung Chang

Amid the rapid development and widespread accessibility of generative AI tools, the emergence of AI-generated images is gaining remarkable traction. While previous research has explored AI-generated texts in marketing, the study on AI-generated images remains understudied. In this study, we examine whether consumers like AI-generated images and how image characteristics and creator characteristics influence the consumer reactions to AI-generated images. We especially focus on two important image characteristics in evaluating AI-generated images (i.e., naturalness and complexity) and two creator characteristics (i.e., professional and influencer status).

For empirical analysis, we collect digital images, whether generated by human or AI, along with related information about the creators from a global showcase platform for artists. We employ automatic image analyses to operationalize image characteristics. We find that people tend to like images less when they recognize the images are generated using AI tools. Regarding image characteristics, however, the naturalness of image weakens the negative effect of using AI in generating images, but complexity enhances it. As for creator characteristics, the influencer status of the creator mitigates the negative reactions to AI-generated images, while art professional status of the creator strengthens them. Our research contributes to the literature of AI usage in marketing by providing valuable insights into consumer reactions to AI-generated image and effective company strategies for utilizing AI-generated image without eliciting repulsive emotions from consumers about the use of AI.

2 - Structural Evolution of Ad Design: Computational Analysis of 40 Years of Super Bowl Ads
Nina Wang, Massachusetts Institute of Technology, Cambridge, MA, United States, Madhav Kumar, John Hauser
Over the past half-century, advertising strategies have evolved in response to shifts in consumer behavior, business priorities, and cultural values. Our paper introduces a scalable computational approach to analyze key design elements in video advertisements. We deconstruct ads into constituent visual, audio, and textual modalities, then apply deep learning techniques to assess five dimensions of ad design: 1) complexity, 2) informativeness, 3) narrator persona, 4) emotional tone, and 5) energy. When applied to 40 years of Super Bowl commercials, our framework reveals trajectories in ad design choices. For instance, ad complexity has increased over time, while informational content and formal language have declined. We discuss how these evolving strategies reflect changing brand-consumer relations in the modern advertising landscape. Overall, our work provides data-driven insights into advertising design principles and their role in brand messaging.
3 - How to Read an Image Like Text
Amrita Dey, University of Denver, Denver, Colorado, United States, Tianyu Gu, Yu Zhu, Stephen Carson

This article explores the dynamic realm of unstructured data analysis, with a particular focus on images in the field of marketing, underscoring its profound impact on managerial strategies. While substantial research has enriched our understanding of unstructured textual data, the potential for image analysis remains relatively untapped. The authors propose three crucial metrics for image analysis, borrowed from well-established text analysis constructs: readability (ease of perception), concreteness, and uniqueness. These metrics encapsulate well-established visual theories like Gestalt Psychology and Visual Complexity Theory, bridging the gap between textual and visual data evaluation. To validate these metrics, the authors employ real-world use-case crowdfunding dataset. This paper equips marketers with a useful tool that facilitates the interpretation of images using simplified textual constructs, thereby enhancing visual efficacy in diverse marketing communication contexts.

FD21
Meeting Room C4.11
Opportunities and Challenges of LLM Applications
Invited Session
Special Sessions
Chair: Chengfeng Mao, MIT, 143 Albany St, Cambridge, MA, 02139, United States
1 - The Challenge of Using LLM to Simulate Human Behavior: A Causal Inference Perspective
Olivier Toubia, Columbia Business School, New York, NY, United States, George Gui
Large Language Models (LLMs) have demonstrated impressive potential to simulate human behavior. Using a causal inference framework, we empirically and theoretically analyze the challenges of conducting LLM-simulated experiments, and explore potential solutions. In the context of demand estimation, we show that variations in the treatment included in the prompt (e.g., price of focal product) can cause variations in unspecified confounding factors (e.g., price of competitors, historical prices, outside temperature), introducing endogeneity and yielding implausibly flat demand curves. We propose a theoretical framework suggesting this endogeneity issue generalizes to other contexts and won't be fully resolved by merely improving the training data. Unlike real experiments where researchers assign pre-existing units across conditions, LLMs simulate units based on the entire prompt, which includes the description of the treatment. Therefore, due to associations in the training data, the characteristics of individuals and environments simulated by the LLM can be affected by the treatment assignment. We explore two potential solutions. The first specifies all contextual variables that affect both treatment and outcome, which we demonstrate to be challenging for a general-purpose LLM. The second explicitly specifies the source of treatment variation in the prompt given to the LLM (e.g., by informing the LLM that the store is running an experiment). While this approach only allows the estimation of a conditional average treatment effect that depends on the specific experimental design, it provides valuable directional results for exploratory analysis.
2 - Custom-tuned LLMs vs. Professional Analysts for Identifying Customer Needs
Chengfeng Mao, MIT, Cambridge, Massachusetts, United States, Artem Timoshenko, John Hauser
Identifying customer needs (CNs) is critical to new product success. For over 30 years, the gold standard has been trained analysts carefully reading interview transcripts or user-generated content. Recently, researchers and firms have developed machine learning methods to facilitate the customer need analysis, by focusing human effort on the most informative content. The process is still expensive and time-consuming, but large-language models (LLMs) show promise. Using professionally-identified CNs in 10 categories, we custom-tuned an LLM to formulate CNs from online reviews. We applied the custom-tuned model to three new categories. We recruited experienced professional analysts to evaluate new-category CNs formulated by either the custom-tuned LLM, an open-source LLM, or by other professional analysts (from prior client-sponsored studies). The source of the CNs was randomized and blind to the analysts. The custom-tuned model formulated CNs comparable to those formulated by professional CN studies and sufficiently specific for product development. The custom-tuned model performed similar to human analysts, and substantially better than the open-source model.
Saturday, June 29, 8:30 AM - 10:00 AM
SA01
Cockle Bay Room 1
Emerging Research on DEI in Corporate and Brand Strategy
Invited Session
Special Sessions
Chair: Dionne Nickerson, Indiana University, Bloomington, IN, United States
1 - Brand Communication of Diversity, Equity, and Inclusion on Social Media and Consumer Response
Seoyoung Kim, Fordham University, New York, NY, United States, Sundar Bharadwaj
This study explores the factors influencing consumer responses to brands' inclusive marketing communication on social media, particularly focusing on diversity, equity, and inclusion (DEI). Modern consumers consider a brand's commitment to DEI when making purchasing decisions, leading brands to integrate DEI into their social media communication. However, not all brands benefit from inclusive marketing, as consumers often perceive it as performative or insincere. Existing research lacks clear guidance on effective DEI communication for brands.
Drawing on sociology and marketing communication research, this empirical study analyzes 3,952 DEI-oriented Instagram posts (2010–2021) using natural language processing and causal inference. The study identifies key communicational factors influencing consumer responses, including comprehensibility, factuality, sincerity, and legitimacy. Results indicate that effective DEI communication should include sufficient elaboration, verifiable information, and consistent implementation to generate favorable consumer responses.
The study challenges the common emphasis on timing and specific DEI topics, suggesting that how a brand communicates may be as critical as when or what they communicate. Our findings are corroborated by real-world examples (e.g., Pepsi's 2017 campaign) which underscore the potential for brands to suffer reputational damage when their DEI communication lacks authenticity.
In conclusion, this research provides practical guidance for crafting effective DEI marketing communications. Brands should prioritize how they communicate DEI issues rather than solely focusing on timing and specific DEI topics. These findings contribute to a more nuanced understanding of consumer responses to DEI communication and offer valuable insights for brands navigating the complex landscape of inclusive marketing on social media.
2 - Race (un)conscious: Understanding Racial Ideologies in Brand Political Activism Statements
Rowena Crabbe, University of Texas Austin, Austin, TX, United States, William Scarborough

This paper examines if and how company statements contain varying racial ideologies (i.e., the cultural frameworks used in discourse). We examine the content of companies’ statements following George Floyd’s murder. We propose a two-dimensional framework for identifying racial ideologies in companies’ statements. The first dimension, race consciousness, describes the degree to which individuals acknowledge or avoid directly discussing race and racial inequality. The second dimension, attribution level, describes whether inequality is attributed to individual characteristics or structural access to opportunity. We construct a unique lexicon to measure the two dimensions of racial ideologies. We then use a bag of words technique to apply the lexicon to companies’ statements. We find that companies’ statements contain distinct configurations of present-day racial ideologies across dimensions of race consciousness and attribution level. This article contributes to the marketing literature by introducing racial ideologies to the marketing literature and demonstrate their presence in company communications. We create a typology of racial ideologies along two dimensions. Second, we create a lexicon to understand the presence of racial ideologies in company communication. These findings advance understandings of BPA and race in the marketplace, with implications for managers and scholars.

3 - When Silence Is Not Golden: The Openly LGBTQ+ Corporate Board Members and Their Impact on Firm Performance
Zhiling Bei, University of Missouri, Columbia, MO, United States
About 5.6 percent of US adults self-identify as lesbian, gay, bisexual, and transgender (LGBTQ+) in 2021. However, openly LGBTQ+ people occupy 0.5 percent of the board seats in Fortune 500 companies in 2021. This visible 10-fold difference (5.6 vs. 0.5) indicates that LGBTQ+ people are underrepresented and/or are fear to disclose their sexual identity in Fortune 500 boardrooms. This paper examines whether and how openly LGBTQ+ corporate board members (OLBMs) affect firm performance. Using a longitudinal analysis of Fortune 500 firms from 2012 to 2019, the authors find that an increase of OLBMs in the Fortune 500 boardrooms is positively associated with (1) a higher degree of corporate sexual equality, measuring how a firm treats its LGBTQ+ employees, consumers, and investors, and (2) higher long-term financial performance (i.e., Tobin’s Q). They identify that corporate sexual equality partially mediates the relationship between OLBM and long-term financial performance. This gives much-needed insights into why the relationship between OLBM and long-term financial performance exists. We also find social acceptance of LGBTQ+ and LGBTQ+ self-identification moderate the relationship between OLBM and long-term financial performance. Our findings have important implications for scholars, business leaders, and investors.
SA02
Cockle Bay Room 2
Influencer Marketing: Insights from Text/Image/Video Analysis
Contributed Session
Chair: Xiangnan Zhang
1 - The Impact of Face Presence in User-generated Videos on Consumer Engagement
Ying Hu, Renmin University of China, Beijing, China, People's Republic of, Ying Ding, Xia Wang

Influencers are increasingly showing their faces in user-generated videos (UGVs). However, a comprehensive understanding of the effects, mechanisms, and contingencies of this approach is still lacking. Findings from a large-scale empirical study of Bilibili videos reveal that the presence of faces in UGVs significantly increased consumer engagement. However, this effect is found to be attenuated when influencers exhibit commercial intent or when the videos are aesthetically more appealing. We further demonstrate the underlying mechanism through a pre-registered experiment and an eye-tracking study, which suggests that face presence in UGVs can better capture consumers’ attention, thus building intimate relationships with consumers and consequently enhancing consumer engagement with videos. Our findings make meaningful contributions to the extant literature on UGVs and consumer engagement and provide detailed managerial guidelines on how to strategically manage the presence of faces in UGVs.

2 - Navigating Influencer Success: Unraveling the Impact of Cognitive and Affective Appeal on Social Media
Mengchen Zheng, Boston University, Boston, MA, United States, Wenyan Tuo, Shuba Srinivasan
Integrating empirical modeling with controlled experiments, our research delves into the intricacies of personal brand development on social media. While micro-influencers are valued for their relatability and authenticity, they may grapple with perceived competence. We propose integrating cognitive appeal and delivering informative content can elevate their popularity. Conversely, mega-influencers, known for their knowledge and authority, may face challenges in establishing emotional connections. We propose utilizing verbal and visual appeal to evoke emotions can foster a stronger bond with the audience.
Our dataset comprises 15,018 videos from 90 YouTube channels, stratified sampled by follower numbers and categories. Channel-specific information, including join dates and gender, was collected, alongside daily view and subscriber growth data from 2018 to 2021. Utilizing the advanced analytical capabilities of generative AI, we construct a robust set of video-specific features, which encompass visual elements from the cover pages and videos, textual analysis of titles and subtitles, as well as linguistic and acoustics patterns within the audio tracks. This multifaceted approach allows for a nuanced understanding of content engagement on YouTube.
With a mixed effect model and controlled experiments, we uncover a robust interaction: cognitive appeal becomes more influential as subscriber numbers decrease, while affective appeal is more effective as subscriber numbers increase. Our longitudinal analysis contributes to a comprehensive understanding of the enduring effects of cognitive and affective appeal.
This research offers actionable insights for cultivating a personal brand. Influencers can optimize their appeal strategies in order to build and maintain a strong presence on social media.
3 - Hatespeech Management in Livestreaming: A Multimodal Deep Learning Approach
Xiangnan Zhang, Temple University, Philadelphia, PA, United States, Xueming Luo, Wen Wang

The prevalence of toxic speech in live streaming nurtures the urgent need for influencers to choose appropriate coping strategies. Based on the psychological coping theory and influencer marketing literature, we utilize an advanced video analytics approach and modern Large Language Model techniques to develop a multimodal framework to identify influencers’ toxic speech coping strategies from livestreaming videos. The four different toxic coping strategies we defined are: Disengagement Strategy, where influencers don’t engage in the toxic speech; Humor Strategy, where influencers try to diffuse the situation in a humorous way; Aggressive Strategy, where influencers choose to attack back; and finally, Calm Strategy, where influencers express themselves frankly without invalidating the attackers. Our research examines the effect of influencers’ adoption of coping strategies on viewers’ engagement, diffusion of toxic speech comments, and influencers’ revenue. We adopted a Panel VAR model approach on data at the minute level from 1,155 live streams to quantify the effects of different coping strategies. We find that the implementation of the Calm strategy has been observed to enhance streamer revenue. The adoption of the Aggressive strategy generally yields only detrimental outcomes. The strategy of using Humor can be seen as a double-edged sword; while it has the potential to boost revenue for influencers and enhance viewer engagement, it may result in the diffusion of toxic comments. Finally, the Disengagement Strategy has mixed and overall limited effects on both viewers’ engagement and revenue. We also explored the moderation effects of influencers’ characteristics on the effectiveness of coping strategies.

SA03
Meeting Room C3.2
Online Retailing: Payment and Credit
Contributed Session
Chair: Chen-yu Lin, No. 100 Wenhua Road, Xitun District, 40724
1 - The Value of Co-branded Credit Cards in Online Retailing: Empirical Evidence from HKTVmall
Haoyu Liu, University of Technology Sydney, Ultimo, Australia, Huijun Chen, Kristiaan Helsen
Co-branded credit cards are sponsored by two parties, usually a retailer and a financial institution, the holders of which may enjoy exclusive discounts or rewards. Being an important part of issuers' business, co-branded credit cards' value in the context of online retailing is unclear: How do co-branded credit cards affect consumer behavior in online retailing? In this paper, we investigate this problem by leveraging a quasi-experiment involving the issue of a co-branded credit card by Citibank, a leading global bank, and HKTVmall, the largest online retailer in Hong Kong. We perform difference-in-differences (DiD) analysis with matching co-branded credit card users and non-users. We find that the co-branded credit card increases both consumers' ordering frequency and ordering size. Most of the increments come from the online-retailer-side benefit (HKTVmall's extra 5% discounts every Thursday) rather than the financial-institution-side benefit (Citibank's 5-times point rewards every Monday). The increments are subject to consumer heterogeneity. The findings provide co-branded credit card design implications for online retailers and financial institutions, and the robustness of the findings is established by considering alternative model specifications.
2 - An Immediate Pain or A Long Sorrow? The Effects of Credit Repayments on Customer Purchase Behavior
Lu(Lucia) Meng, Renmin University of China, Beijing, China, People's Republic of, Zelin Zhang, Peter Popkowski-Leszczyc, Qiang(Kris) Zhou, Zhixiang Huang

Major retailers like Amazon, Taobao, and JD are increasingly incorporating store credit services into their platforms. Similar to credit cards, such services offer two credit repayment options: a) Lump-sum repayment, allowing customers to settle the entire credit within a one-month grace period with no interest penalty. b) Periodic repayment, allowing customers to separate credit repayments across multiple periods with interest charges. While periodic repayment can generate revenue, it will also remind customers of their debt associated with the retailers, which could decrease their subsequent purchases. Therefore, it is imperative to understand the impact of credit repayment.

Our paper examines the impact of Periodic vs. Lump-sum credit repayments on customers’ subsequent purchases, using a unique transaction dataset from JD.com, a leading E-commerce platform in China. We employ a single-hurdle difference-in-differences approach combined with propensity score matching, which enables the simultaneous estimation of a two-stage purchase process. In the first stage we estimate the latent purchase likelihood and in the second stage the purchase amount.

The results show that periodic repayments lead to a decrease in customers’ purchase likelihood, however, after overcoming this hurdle, they tend to spend more. We find that the negative effect on purchase likelihood stems from a mental hurdle effect and the positive effect on purchase amount arises from a liquidity relaxation effect. Our findings provide valuable insights for retailers in designing credit repayment policies that enhance the profitability of their stores, and that align with customer behaviors for improved financial outcomes.

3 - The Mechanism of Trust Transfer on Consumer Acceptance of Retailer-based m-payment Service in Taiwan
Chen-yu Lin, Feng-Chia University, Taichung City, Taiwan, Nguyen-Phan-Thao Nguyen, Li-Wei Wu

This study aims to understand the trust mechanism in determining Taiwanese consumers’ intention to accept a retail-based m-payment service. Despite m-payment becoming one of the most popular payment services, our understanding of the mechanism of trust transfer in retail-based m-payment is still limited.

This research develops a synergistic model to highlight the critical role of trust in consumers’ adopting a retail-based m-payment service. The trust transfer theory was adopted and extended with two a trust triggers (i.e., store reputation and perceived service quality). The sample of 300 respondents was obtained through an online survey distributed to PX Mart’s customers in Taiwan. A research model and hypothesis were tested with structural equation modeling (SEM) using Smart PLS software.

The results reconfirmed the existence of direct mechanism of trust transfer (i.e., trust in retailer to trust in a retail-based m-payment). Furthermore, the empirical results revealed two indirect trust transfer mechanisms, namely a trust-perceived usefulness route and a trust-perceived risk route. Consumers’ trust in retailer could increase their trust in a retail-based m-payment through enhancing perceived usefulness and reducing perceived risk of a retail-based m-payment. The results also showed the positive influence of retailer reputation and retailer service quality in consumers’ trust in the retailer.

Our findings complement trust transfer theory by demonstrating that three mechanisms of trust transfer routes from a retailer to its m-payment service. This finding encourages more retailers to offer their m-payment services by enhancing high quality in-store services and accumulated more store reputation.
SA04
Meeting Room C3.3
Chatbots
Contributed Session
Chair: Abdulaziz Alharbi, NUspace, The University of Newcastle, Newcastle NSW 2300
1 - Functional and Hedonic Consumption of Personified AI Agents on Online Customer Experience and Purchase Intention
Yuanxin Wang, Minzu University of China, Hong Kong, China, People's Republic of, Jiaqi Li

Currently, AI agents play significant roles within the e-commerce industry, as it effectively improves the overall consumer experience and bridge the emotional disconnect between humans and machines by incorporating anthropomorphic attributes. The present research aims to examine the persuasive influence of anthropomorphic features on consumers’ inclination to make purchases in both functional and hedonic product contexts.

Using a mixed-methods approach, we conduct an online experiment and in-depth interviews. Participants are recruited from online platforms, and around 500 questionnaires are distributed. Subsequently, we employ AMOS software to construct a structural equation model, aiming to investigate the relationship between variables such as the degree of anthropomorphism, consumers' purchase intention, information service quality and artificial empathy. Additionally, In-depth interviews with 8-10 AI agent users provide a comprehensive understanding of the persuasive impact.

The results suggest that the influence of anthropomorphic AI agents on purchase intention is not insignificant in functional scenarios when compared to non-anthropomorphic service. This lack of impact can be attributed to the mediating role of the Information Quality Model (IQM). However, in hedonic scenarios, anthropomorphic AI agents have significant positive effects on purchase intention, which is facilitated by the presence of artificial empathy. Therefore, establishing emotional resonance with consumers is crucial for the effectiveness of anthropomorphic features. Furthermore, the effects of these features are moderated by Maslow's hierarchy of needs in both scenarios.

These insights offer enterprises a precise framework for customizing AI agents to various product categories.

Keywords: AI agents, Anthropomorphism, Product types, Purchase intention, Purchase scenarios

2 - A Conceptual Model of AI-Enabled Chatbot Customer Experiences in Religious Tourism
Abdulaziz Alharbi, The University of Newcastle, Callaghan, Australia, Ameet Pandit

AI-Enabled Conversational Agents (AICAs) – e.g. chatbots – are redefining the way religious tourists interact with and experience spiritual journeys. This technological evolution at the intersection of faith, culture and digital innovation offers personalised-travel-experience opportunities. Despite AICAs’ potential to revolutionise religious-tourism experience, a knowledge gap exists in understanding the full spectrum of AICA effects across the entire religious-tourism journey, from initial interest to post-travel reflection. This deficiency limits understanding of how AICAs can enhance not just moments of direct interaction but the holistic journey, influencing tourists’ cognitive, emotional and behavioural responses in religious tourism.

Addressing this gap, we integrate the Stimulus-Organism-Response (S-O-R) framework with the Customer Journey Model to delve into the dynamics of religious-tourism experiences. This approach allows for a conceptualisation of how interactions with AICAs (S) affect tourists' cognitive and emotional states (O), e.g. perceived chatbot intelligence, ease of use, frustration or enjoyment. These states then influence tourists' behavioural responses (R), e.g. spiritual well-being and the intention to continue using AICAs.

Adopting a mixed-methods approach, the planned research will combine qualitative interviews with quantitative surveys to investigate tourists’ perceptions and behaviours regarding AICAs in religious tourism. This research aims not only to advance theoretical understanding but also to provide practical guidelines for integrating AI technologies in a way that enriches the religious tourist experience, making a compelling case for the importance of this research in navigating the digital transformation of religious tourism

SA05
Meeting Room C3.4
Qualitative Research: Avatars and Virtual Influencers
Contributed Session
Chair: Paul Messinger, University of Alberta, Edmonton, AB, Canada
1 - Exploring Consumer Avatars: Identity and Consumption Behavior in Virtual World
Aditya Billore, Indian Institute of Management, Indore, Indore, Madhya Pradesh, India, Ashish Sadh, Neha Khatri

In today’s hyper-connected ecosystem, the digital space presents marketers with opportunities to build brands using the power of co-creation and communities (Swaminathan et al., 2020). This ecosystem enables the users to socialize, collaborate, have fun, and even trade with other users beyond the limitations of time and space. The users in virtual space use Avatars to present their self-identities, and these avatars may project their actual self or the ideal self. The avatars become one’s interface for interacting with other users in the digital social space, projecting their desired identity (Freeman & Maloney, 2021).

The existing literature explores various aspects of consumer avatars like anthropomorphism (Suh et al., 2011; Dubosc et al., 2021), culture (Aljaroodi et al., 2023), self-identity (Kalyvaki et al., 2023), etc. This study aims to identify the factors consumers consider while designing/customizing and selecting their Avatars and how they see these avatars projecting their identity. Also, to understand the consumption behavior that the consumers are likely to exhibit based on their virtual identity. We use an iterative approach, moving between theory and qualitative data to develop a comprehensive conceptual framework using qualitative data and extant literature but emphasizing data-based findings analytically. The framework developed would identify theoretical mechanisms explaining attitude formation and consumer choices in the contexts of Avatars and Metaverse. The insights that this study would bring will help marketers in customizing the choice of communication vehicle in the virtual world, also providing better opportunities for consumers to express themselves on these platforms.

2 - Navigating Digital Altruism: The Impact of Strategic Gender Representation in Roblox's Philanthropy Market
Lintong Li, University of Sydney, Belfield, Australia, Qiang Lu, Mathew Chylinski

In the digital era, strategic communication in philanthropy is increasingly vital, especially on platforms like Roblox. This paper presents a mixed-method study, blending netnographic observation over three months with a Choice-Based Conjoint (CBC) analysis from 183 American participants, to explore donor preferences and the impact of gender stereotypes in digital fundraising.

Our netnographic analysis of Roblox's "PLS DONATE" feature uncovers a significant usage of emotional, rational, identity, and reward appeals, indicating a trend towards personalized donor engagement. The CBC analysis reveals surprising gender biases, particularly an inverse relationship between masculine traits and preference for female avatars, challenging traditional notions of gender in digital philanthropy.

The study also identifies a generational split in donation behaviors, with younger donors favoring identity-based appeals, contrasting with older donors' traditional preferences. This highlights the need for flexible communication strategies catering to diverse age groups.

Our findings contribute to the understanding of digital philanthropy, showcasing Roblox as a potential model for virtual fundraising. The study underscores the importance of developing inclusive and equitable strategic communication practices in digital marketing and platform development, tailored to the preferences of digital natives and future generations.

3 - A Framework of the Extended Self in the Metaverse: Visual Self-representation in Avatar-mediated Environments
Paul Messinger, University of Alberta, Edmonton, Alberta, Canada, Xin Ge, Kristen Smirnov, Ozan Ozdemir, Feyzan Karabulut

We provide an overview of the relationship between the appearance of avatars in virtual environments and the people they represent. We also consider how other people respond to such self-representations. Our synthesis of selected empirical and conceptual literature constitutes a framework of the extended self in the metaverse.

We argue that the selection of avatar appearance reflects a balance between the motives of self-enhancement and self-verification. Across multiple avatars for different roles, there is evidence of a hierarchy of physical traits whereby people retain core appearance traits that do not change across roles, but traits more peripheral to a person’s core identity vary to fit the occasion and the mood of the individual. There is also evidence of a feedback effect whereby in-world behavioral traits are affected by the relative attractiveness of one's avatar compared with the individual’s real appearance (known as the Proteus Effect).

Concerning how other people view an individual’s avatar, there is evidence overall that the more realistic the avatar, the more trustworthy people perceive the avatar, which, in turn, influences consumer behaviors and outcomes. An important qualification is that the perceived trustworthiness of an avatar is also related to the congruity between (a) form and behavioral realism and (b) form and domain realism. So an avatar with high form realism but low behavioral realism may be ineffective, for example, at persuasion. But a cartoon-like avatar can be as (or more) persuasive for promoting a digital product, as an avatar with high form realism.

SA06
Meeting Room C3.6
Crowdfunding
Contributed Session
Chair: Hyung Sup Bhan, Tulane University, New Orleans, LA, United States
1 - Platform Recommendation in Crowdfunding
Habin Jung, INSEAD, Fontainebleau, France, Maria Ana Vitorino

Online platforms often provide product recommendations to consumers, serving to alleviate potential information asymmetry between sellers and buyers, particularly concerning product quality. We examine the effectiveness of platform recommendations as a tool to reduce this information asymmetry, using data from Kickstarter, a large crowdfunding platform that connects project creators with financial contributors who choose projects to fund from the available options. The data details the funding acquired by all 55,710 projects on the platform over the course of 83 months. We leverage the transition at Kickstarter from a recommendation system allowing project creators to post fake recommendations to one that prevents it. This transition allows us to evaluate Kickstarter's ability to identify high-potential projects. Furthermore, we explore whether the platform's recommendations contribute to improved outcomes for project creators, particularly in terms of the amount of funding they receive.

2 - Donations or Rewards? Insights into Personal Pronoun Use in Crowdfunding
Sheng Liu, Lingnan University, Hong Kong, Hong Kong, Hong Kong, Yiwei Li, Lei Su, Darren Dahl

Language is the currency of human communication, we use words to interact with others. Particularly, personal pronouns are among the most used words according to the Oxford English Corpus. The extant literature suggests that even subtle variations in language use can significantly affect consumers’ perceptions/behaviors. However, although substantial research shows that personal pronoun usage reflects the writer's mindset or traits, it is unclear how pronoun usage impacts the reader.

This project attempts to bridge this gap and investigates how pronoun usage (third-person versus first-person) affects funding behavior in the increasingly important domain of crowdfunding (donation-based versus reward-based). Specifically, we predict that the use of third- (first-) person pronouns increases funders’ intention to contribute to donation- (reward-) based crowdfunding. For donation-based crowdfunding, funders contribute primarily out of altruism and are more likely to trust a fundraiser who writes objectively using third-person pronouns. For reward-based crowdfunding, funders expect rewards and the use of first-person pronouns signals the fundraiser’s expertise in product development.

We will combine empirical analyses with experiments. Empirical analyses based on population-scale data verfity the proposed effects. Field and lab experiments strengthen the causal inference of the posited relationships. The proposed research is among the first that (1) demonstrates that a subtle variation in language use can influence funding behavior, (2) offers an integrated perspective of donation- and reward-based crowdfunding, and (3) causally reveals a psychological process that has significant funding implications for crowdfunding. The managerial implications are also substantial as currently 88% of donation-based crowdfunding overwhelmingly uses first-person pronouns.

3 - From Pitch to Pledge: The Role of Linguistic Legitimacy Cues on Crowdfunding Performance
Nasim Ul Haque, University of Melbourne, Melbourn, Australia, Liliana Bove, Stephan Ludwig, Dennis Herhausen, Dhruv Grewal

As crowdfunding gains popularity among marketers and entrepreneurs, the competition for investment has intensified, resulting in most projects failing to attract sufficient investment. To demonstrate their desirability, legitimacy theory suggests entrepreneurs aiming for crowdfunding success should establish the legitimacy of their projects. Employing a multi-study approach, analyzing 112,804 projects from Kickstarter (investing in creative projects) and 12,417 projects from Kiva (lending to marginalized entrepreneurs), this research highlights the importance of the focus and order of the textual cues of pragmatic (benefits to investors) and normative (adherence to social norms) legitimacy in project descriptions. Descriptions work best when they moderately focus on pragmatic and normative legitimacy, with the introduction of normative legitimacy later in the description helping to sustain investor interest. The optimum ordering of pragmatic legitimacy varies based on the crowdfunding platform's nature. Additionally, the study reveals that distinctive projects on Kickstarter experience significantly higher gains when focusing on legitimacy cues compared to their more similar counterparts, while the effectiveness of legitimacy cues maintains a consistent impact across all Kiva projects. These findings hold implications for researchers studying crowdfunding dynamics and entrepreneurs seeking fundraising success.

4 - Platform Openness and Crowdfunding Creator Economy: Different Roles of New and Experienced Users
Hyung Sup Bhan, Tulane University, New Orleans, LA, United States, Seoungwoo Lee, Hyoryung Nam, Joon Ro

Crowdfunding platforms often choose to increase platform openness by relaxing entry restrictions to ensure a sustainable pool of creators. We study the dynamics of crowdfunding platform under different policies of platform openness by leveraging a unique event: Kickstarter’s relaxation of project application requirements in June 2014. Analyzing a dataset encompassing 291,613 projects spanning five years, we uncover the distinct roles of new and experienced users in shaping the platform’s growth, diversity, and revenues under different policies of platform openness. In the open platform characterized by high information asymmetry, retaining experienced users generates stronger cross-side network effects, greater project diversity, and higher revenues than acquiring new users. Conversely, in the closed platform, acquiring new users yields stronger cross-side network effects, greater project diversity, and higher revenues. We provide insights on how crowdfunding platforms can manage user acquisition and retention strategies under different policies of platform openness.

SA07
Meeting Room C4.1
Marketing Mix Modeling
Contributed Session
Chair: Justin Kim, Mphasize, Publicis Media
1 - Geospatial Marketing Mix Modeling: Harnessing Spatial Autocorrelation for Enhanced Prediction and Insights
Ryoma Yasunaga, Hakuhodo DY Holdings, Tokyo, Japan, Shonosuke Sugasawa, Hiroshi Kato
This study introduces the application of spatial modeling, specifically the Conditional Autoregressive (CAR) model, in marketing mix modeling (MMM). Traditional MMM often overlooks the influence of geospatial information despite its natural relevance to consumer purchasing behavior. A vital aspect of the study was the acquisition and utilization of detailed regional and weekly total sales data, a valuable resource for understanding consumer behavior. Combined with information on product prices, total TV commercial exposure, and store displays, this data provided a comprehensive foundation for our spatially enhanced MMM. With this enriched dataset, we used the Geo-level Bayesian Hierarchical Media Mix Model (GBHMMM) as a baseline and integrated CAR to spatialize the model. Our method, utilizing spatial modeling, captures the spatial autocorrelation and spatial heterogeneity in regression coefficients among regions, a factor the baseline overlooked. The results show that our method improves the predictive performance during the test period and the information criterion, compared to the baseline. From a practical standpoint, this study not only constructs a model with improved performance but also illuminates the intricate relationship between consumer purchasing behaviors and geospatial information. This study extends beyond traditional MMM by offering actionable insights for region-specific marketing strategies, mainly by analyzing spatial heterogeneity in consumer behavior.
The findings highlight how consumer interactions with marketing efforts vary across regions, pointing to new opportunities for targeted marketing campaigns and strategic planning. We will present further details of the experiments during the presentation.
2 - Self-selection, Sunk Cost Fallacy and the Effectiveness of Paid versus Free Coupon Campaign
Qingliang Wang, Xi'an Jiaotong University, Xi'an, China, People's Republic of, Jingguo Liu

It is a general practice for marketers to use paid or free coupons to attract consumers to participate in promotional campaigns. However, it is unclear whether paid or free coupon is more effective. We aim to address this issue by investigating the impact of paid versus free mobile coupons on consumer participation in promotional campaigns. Based on the literature on sunk-cost fallacy and zero-price effect, we propose hypotheses to examine "whether consumers have to pay for the coupon" on consumers' acquisition and redemption of coupons. We collect detailed data on coupon promotion campaigns and individual-level consumer transaction data from a refueling service platform in China. This data is unique in that we can observe both the acquisition and redemption of coupons for each consumer. We empirically identify the self-selection effect (i.e., high-incentive consumers are more likely to purchase) and sunk-cost fallacy (consumers who make a payment are more likely to redeem the coupon) in consumers' coupon acquisition and redemption process. We find that the coupon price could be an effective design tool to improve the effectiveness of promotional campaigns. Our results provide evidence of the sunk-cost fallacy in consumers' coupon redemption behavior. The results have important practical implications in terms of how to design coupon-based promotional campaigns effectively.

3 - Machine Learning Enabled Marketing Mix Modeling
Justin Kim, Mphasize, Publicis Media, New York, NY, United States, Amar Suljic

Machine Learning Enabled Marketing Mix Modeling

The recent emergence of machine learning techniques (ML) has gained a great deal of traction in both marketing practitioners and marketing science community. ML based algorithms are quite often utilized in the fields of marketing analytics. Marketing mix modeling (MMM) is not an exception.

Central to the practical implementation of marketing mix modeling (MMM) are the issues related to model specification and estimation methods.

This research compares machine learning techniques of statistical learning theory (Hastie et. al, 2009) with frequentist and Bayesian approaches (Leeflang, P.S.H. et.al 2000 & Rossi et.al 2005), that have generally been practiced in the MMM industry, in the following aspects. (1) variable selection and transformation algorithms in the pre-model estimation stage including pre-processing the data, (2) model formulation incorporating synergistic effects among sales driver and inter-dependency of cross media channels , and estimation process and cross-validation of model output, and (3) applications of model findings, particularly in the areas of optimization of budget allocation across media and marketing channels and building causal based sales forecasting capability.

Extensive empirical works are conducted using the data across different verticals, such as CPG, Telecom, and Pharma. In order to evaluate the models and in-going hypotheses, a few criteria are employed; (1) statistical integrity, such as model stability and predictive accuracy, (2) applicability and conformity of model findings in the real-world business decisions, (3) Efficiency and flexibility in terms of model building, usage, and recalibration process.

SA08
Meeting Room C4.2
NFTs 1
Contributed Session
Chair: Ashish Galande, IIM Udaipur, 313001
1 - Beyond the Tangible: Unpacking the Impact of NFTs on the Traditional Art Market
Jiang Qian, The University of Sydney, Sydney, Australia, Vanessa Tianying Song, Qiang Lu

The record-breaking sale of Beeple's NFT artwork, "The First 5000 Days," for $69.3 million in March 2021, has catapulted NFTs into the spotlight, making them an increasingly compelling subject that has piqued the interest of numerous investors. The traditional art market plays an important role in investment portfolio in addition to the market of stocks, gold, precious metals and real estate. The wide exposure of NFTs on different media may reshape how customers perceive the value of artworks, and according to Art Basel& UBS, by 2021, 74% of high-net-worth buyers had purchased NFT artworks. Despite their popularity as an investment class, however, the impact of NFTs on the market of traditional tangible artworks remains unclear. To delve into this question, we ask: ‘How does the growing popularity of NFT artworks influence the value of traditional paintings?’

To address this, our study employs computer vision and machine learning techniques to analyze the visual elements of NFT artworks and traditional paintings sold at auction. We aim to identify key features and similarities between the two. To assess their impact on the traditional art market, we utilize a Difference-in-Differences approach combined with Causal Forest methods to estimate and identify heterogeneous treatment effects. The insights gained from this study have significant implications for marketers and art companies.

2 - The Value of Probabilistic Selling: Evidence from NFT Mystery Box
Peiwen Xie, Nanyang Technological University, Singapore, Singapore, Sadat Reza, Eunsoo Kim

The popularity of mystery consumption and the widely-applied uncertainty marketing across industries have attracted much academic attention. Despite ongoing research in probabilistic selling (i.e., introducing uncertainty in product offerings or assignments) over a decade, there is still no empirical validation of the benefits of such a selling strategy to marketers to date. To fill this gap, we examine the value of probabilistic selling and the underlying mechanism in the context of the non-fungible token (NFT) market, in which creators use this selling strategy by launching mystery boxes. We use three estimators (i.e., two-way fixed effect DiD, staggered DiD, and synthetic DiD) to tackle a series of identification challenges in the staggered adoption pattern. We find that probabilistic selling increases adopters’ monthly revenue by alleviating mismatches between supply and demand (i.e., enhancing the matching efficiency of the market and spreading out demand across the component products). In addition, we show that alternative mechanisms, such as the change in the pricing mechanism and novelty effect alone, cannot explain the results. The findings contribute to probabilistic selling literature by validating the value of this selling strategy in long-standing theoretical predictions, and provide practical implications for its implementation.

keywords: probabilistic selling; probabilistic goods; opaque goods; mystery box

3 - The Market for Counterfeit Non-fungible Tokens
Jiani Liu, The University of Hong Kong, Hong Kong, Hong Kong, Hui Li, Xi Li, Shane Wang
Experiencing an exponential surge in popularity since 2021, non-fungible tokens (NFTs) have become widely regarded as an effective means of confirming the authenticity of digital assets thanks to the inherent transparency and immutability offered by blockchain technology. However, substantial media attention and growing empirical evidence confirm that this industry is facing a significant counterfeiting crisis. In this paper, we seek to understand the popularity of counterfeit NFTs, often referred to as “copymints,” by investigating their value determinants and the trading behaviors of investors. We collected a unique dataset on transactions of authentic NFTs and copymints from one of the largest blockchains in existence, Ethereum. Aggregate token-level analyses suggest that the value of copymints is positively associated with the price of authentic NFTs, especially for high-quality or investable copymints. We also find that investors who purchase both authentic NFTs and copymints play a pivotal role in shaping the copymint market. Micro investor-level analysis shows that these investors treat authentic NFTs and copymints as complementary products, which explains the positive price co-movement between authentic and counterfeit NFTs at the aggregate level. The underlying insight from our discoveries suggests that, within the NFT market, profit is often the primary motivator for investors who prioritize a token’s quality or investability over its authenticity when considering copymint purchases. We discuss the implications of our results for both NFT creators and marketplaces.
4 - Is An AI-generated Image Worth a Thousand Words? Using Automated Image Analysis to Understand the Value of AI (vs. Human) Generated NFTs.
Ashish Galande, Indian Institute of Management Udaipur, Sethji Ki Kundal, India, Frank Mathmann, Sara Thaichon
Artificial intelligence (AI) is disrupting creative industries, enabling everyday users, with limited skill or training, to create stunning visual art. Clients increasingly choose AI-created art that is instantly available at low or no cost over expensive services that take weeks to generate. A key question to inform public policy as well as marketing strategy is what drives consumer value for Human vs. AI-generated visual artwork. This research uses automated image analysis to distinguish Human from AI-generated visual artwork to predict sales prices on OpenSea.ai, a major marketplace for non-fungible tokens (NFTs). Recent research has considered the effects of generative AI through the lens of algorithm aversion which suggests that consumers overwhelmingly prefer humans over artificially intelligent algorithms. Building on emerging research that illustrates the importance of consumer self-regulation for AI acceptance, we consider consumers’ regulatory mode orientations as a key factor determining the value of AI-generated imagery. We use 6.1 million transactions of NFTs on Opensea.ai and establish regulatory mode motivations extracted from language in the listings as an important factor determining the sales price. We currently illustrate that NFTs which contain AI-generated imagery are valued less than human-generated imagery, extending algorithm aversion effects to AI-generated imagery. For marketing strategy and allocating public policy attention, this suggests that AI-generated imagery may not be an effective substitute in use-cases where considerations regarding consumer value outweigh speed and cost benefits of AI-generated imagery.
SA09
Meeting Room C4.3
Consumer Behavior: Influencer Marketing
Contributed Session
Chair: Tongyao Lin, Newcastle University, Newcastle upon Tyne, United Kingdom
1 - What Will my Influencer Say? The Impact of Credible Social Media Influencers
Lancy Mac, University of Macau, Macau, China, People's Republic of, YiXin Zhang, Jimmy Chi Ming Lee

There is considerable research evidence on the positive impact of social media influencers on consumer purchase decisions. These influencers are able to sway their followers using their unique resources and characteristics and therefore enjoy celebrity status. While a lot of studies have been done on the impact of influencer credibility, little is known about the mechanism through which such characteristics can affect consumer behavior. This study seeks to uncover the impact of influencer credibility on followers’ information-seeking behavior which in turn contributes to purchase intention.

Specifically, we gauge the impact of trustworthiness and expertise of influencers on their followers’ tendency to seek opinions from them, thus enhancing purchase intention. The emphasis is on the intervening mechanism of information seeking from the influencers which is based on social identity theory and the uses and gratifications theory. When the influencer is credible, this will enhance the follower’s propensity to actively seek confirmation from their influencer, therefore, greatly impact their purchase decision. Results of our survey showed that both trustworthiness and expertise are positively related to information seeking from influencers, which in turn boosts purchase intention.

The results of this study contribute to the existing literature by advancing the understanding of the mechanism through which influencer credibility can facilitate positive consumer behavior, research on which is still scarce. Influencers should focus on leveraging their trustworthiness and expertise to elicit the engagement of followers to always seek their opinion/information before making purchase decisions.

Keywords: social media influencers, information seeking behavior, source credibility

2 - Beyond SENS(E)ation: Exploring Sensory Language Effectiveness in Virtual Influencers' Product Endorsements
Nadine Eckel, University of Münster, Münster, Germany, Christina Okoutsidou, Dipayan Biswas

Although managers increasingly shift their resources from human to virtual influencers, the factors driving the success of their product endorsements still need to be explored. Here, sensory language is known to elicit positive consumer behavior. Yet, what if the influencer cannot sense? This paper investigates human-likeness and product depiction that may impact the effectiveness of sensory cues in virtual influencers’ posts. The authors develop a conceptual framework grounded in mental imagery theory and employ a multimethod approach, analyzing both social media and lab experiment data to test their hypotheses. Contrary to prior belief, the usage of sensory language by virtual influencers exerts a reverse (negative) effect on engagement and purchase intent. Based on our findings, marketers can guide high human-like virtual influencers to use low sensory language to mitigate the uncanny valley effect. Therefore, this paper complements past research on the effectiveness of social media content in (virtual) influencer marketing.

3 - Competitiveness of Virtual Influencers with Human Counterparts on Social Media Platforms: Can They Influence Brands' Marketing Strategies Ingrained in Consumer Shopping Decisions?
Abaid Ullah Zafar, Wenzhou-Kean University, Wenshou, China, People's Republic of, Yangyang Fu, Jianwen Lv, Chen Rui

Rapid proliferation in contemporary technologies has led to the emergence of another type of influencer on social media platforms, called virtual influencers. They are artificial computer-generated influencers rather than humans but can perform similar functions on social media, gaining popularity among netizens and other concerned stakeholders. Exploring their role to fill the gap and offer valuable insights for effective marketing strategies is critical. Accordingly, a research model is suggested in the context of virtual influencers to unveil the systematic role of social media browsing, openness to change, and parasocial relationships in shaping consumers' purchase intention along with boundary factors following the uses and gratifications theory and stimulus-organism-response framework. Data were directly collected from 207 followers of virtual influencers and analyzed by structural equation modeling. Results unveil that hedonic and utilitarian social media browsing positively shapes followers' openness to change, which affects their parasocial relationship. Such relationships with virtual influencers significantly encourage followers to buy the conversed products. Findings of interaction effect indicate that the perceived humanness of virtual influencers substantially enhances the role of openness to change to build a solid parasocial relationship. Unexpectedly, this research could not establish a significant moderating relationship of product-endorser fit and the parasocial relationship towards the followers' purchase interaction. This study provides novel insights for brands to exploit the digital uprising of virtual influencers effectively. It indicates the need to change digital marketing strategies to reinforce consumer engagement strategies arising from the virtual influencers framework, which can help achieve marketing goals worldwide more effectively.

4 - Beyond Virtual Agents: Employing Virtual YouTubers to Enhance Advertising Effectiveness in Influencer Marketing
Tongyao Lin, Newcastle University, Tyne, United Kingdom, Saurabh Bhattacharya, Qionglei Yu

Employing virtual influencers (VIs) as a marketing tool is a new trend in influencer marketing (Appel et al., 2020). Although VIs without self-awareness are easy to control, studies have found that VIs lack credibility with the customers because they are virtual agents, which can negatively impact advertising effectiveness (Lim & Lee, 2023; Ozdemir et al., 2023).

To understand how to mitigate the negative impact, this study differs from the existing research that focuses on virtual agents as influencers by exploring another type of VI: Virtual YouTubers (VTubers) (Lou et al., 2022). VTubers are unique content creators using 2D/3D avatars to interact with viewers. Compared to virtual agents, VTubers have humans to provide their experience, and viewers are aware of the persons behind the avatars (Lu et al., 2021). This awareness may reduce the negative impact caused by virtual agents and have better advertising effectiveness (Luo et al., 2019). However, no studies examine whether VTubers demonstrate better advertising effectiveness than virtual agents.

This study, grounded in social presence and congruity theories (Klein et al., 2019; Osgood & Tannenbaum, 1955; Short et al., 1976), contributes to diversifying the influencer types in the influencer marketing literature by investigating VTubers and comparing the impact of VIs and VTubers on advertising effectiveness. This study aims to provide companies with an enhanced market strategy by employing appropriate influencers for favorable advertising effectiveness. Also, it provides insights into increasing customers' perceived credibility to improve advertising effectiveness for marketers.

SA10
Meeting Room C4.4
Sales Promotion 1
Contributed Session
Chair: Trang Bui, Tilburg University, Warandelaan 2, 5037 AB , Tilburg, Netherlands
1 - Mall-Wide Conditional Promotions
Cheolho Song, Washington University in St. Louis, St. Louis, MO, United States, Jingcun Cao, Tat Chan, Haihao Guo
We study the impacts of mall-wide conditional promotions (MWCPs), during which retail stores in a shopping mall simultaneously offer conditional promotions that allow consumers to redeem rewards after reaching a spending threshold, all planned and coordinated by the mall manager. Using data from one of the largest shopping malls in China, we show that the promotional event in summer 2019 increased the sales of participating stores by 1,355%, with the cost of rewards being only 4.7% of the total revenue. The sales increase can be decomposed into a 199% expansion in extensive margin and a 386% expansion in intensive margin, with the interaction of the two margin expansions accounting for the remaining 770%. Both luxury and nonluxury stores in the mall benefited from the promotional event, and there is a positive spillover to nonparticipating stores. We show data patterns consistent with our conceptual framework of how MWCPs achieve their effect. Further analyses rule out alternative explanations that the sales increase is driven by short- or long-term intertemporal substitutions, suggesting that there is a significant demand expansion. Our findings provide valuable insights for mall managers to design promotions and evaluate their performance.
2 - The Dual Effects of Free-trial Promotion on Customer Acquisition and Retention: A Randomized Field Experiment
Min Zhang, Tianjin University, Tianjin, China, People's Republic of, Yin Bai, Yuxuan Ai, Yuzhuo Li, Yiwei Li

Companies often face the decision on whether they should implement a free-trial promotion. Offering free products can potentially boost customer purchase, as extrinsic motivations can stimulate customers’ incentive behaviour. However, implementing such a promotion might also have some unintended consequences, hindering customers’ retention due to the lack of intrinsic motivation, which stands in a superior stage in long-time behaviour. This research scrutinizes the impact of a free-trial promotion on both customer acquisition and retention metrics, and endeavours to identify an optimal strategy to maximize its benefits. Identifying the causal effect of the free-trail promotion is challenging because companies usually make their decisions based on private information and may have simultaneously implemented other initiatives that could confound the effect. As a result, we conducted a randomized field experiment in collaboration with an insurance company, finding that free-trail promotion can benefit customer acquisition while hindering retention, and offering highly related free-products may make up. Companies should take cognizance of the dual effects of free-trial promotions and provide proper types of free products.

3 - Do Online Free Product Promotions for Consumer Packaged Goods Encourage/Discourage Offline Purchases?
Kaichi Saito, Meiji-Gakuin University, Tokyo, Japan, Satoshi Nakano, Jeff Inman, Akira Shimizu, Takashi Teramoto

CPG manufacturers often attempt to communicate with consumers through company-owned websites. On the other hand, consumers still tend to make their CPG purchases primarily at physical stores. In this research, we focus on online free product promotion campaigns run by CPG manufacturers when introducing new brands into the market and examine the relationship between online entries into the campaigns and offline purchases. To investigate the relationship between online promotion and offline purchases, we use unique single-source panel data combining offline purchases, webpage visits, and TV advertising exposure for two new alcohol brands launched almost simultaneously by the same manufacturer and conducting free product promotion. According to a preliminary descriptive analysis, consumers do not often access pages conveying brand attributes and brand image but rather more frequently access pages announcing promotion campaigns and accepting consumers’ entries into the campaigns. We fitted the single-source data to a bivariate probit model with random effects comprising purchase and campaign entry equations. The results show that (1) entry experience into free product campaigns promotes trial purchases, (2) this positive effect of campaign entry experience on purchases disappears after trial purchases (i.e., campaign entry experience does not promote repeat purchases), and (3) the first campaign entry promotes trial purchases, but the second or subsequent entries do not have any additional effect. This study contributes to the customer journey literature by demonstrating that participating in free product campaigns encourages consumers to make an initial purchase but is not a strong driver to start the loyalty loop.

4 - Can Price Promotions Stimulate Sustainable Purchase?
Trang Bui, Tilburg University, Tilburg, Netherlands, Barbara Deleersnyder, Arjen van Lin

While the retail sector is increasingly leveraging price promotions to drive sustainable purchases, the efficacy of this strategy in promoting sustainable purchases is not straightforward. Against this background, retailers increasingly introduce sustainable private labels (PLs) as an affordable alternative to sustainable national brands (NBs). These PLs emerge as a significant contributor to sustainable grocery purchases.

This study investigates (1) the extent to which immediate and long-term (own) price promotion elasticities differ between sustainable and conventional brands; (2) how promotion effectiveness differs across sustainable NBs and PLs; and (3) how shoppers switch between sustainable and conventional NBs and PLs when they are promoted.

Based on German household panel data supplemented with product sustainability information, we build a two-stage modeling approach to unveil variations in promotion response across sustainable and conventional NBs and PLs. Using an error correction model, we estimate brands’ own and cross promotion elasticities and calculate their ability to influence other brands with promotions (i.e., clout) and their susceptibility to the promotions by other brands (i.e., vulnerability). The findings show that price promotions can indeed boost sales for sustainable NBs and PLs both in the short and long-run. The observed patterns of competition between brands offer valuable insights to retailers and manufacturers, guiding the organization of promotion support. Our research also holds relevance for policymakers developing strategies to foster more sustainable grocery purchases.

Keywords: Price promotions, Sustainable, Retailing, Private labels, Error-correction model

SA11
Meeting Room C4.5
Retailing: Category/Brand Choice 1
Contributed Session
Chair: Pei Yu Chien, National Tsing Hua University
1 - Cross-category Effects of Price Discount: Evidence from Gasoline Retailing
Hung Ho, The University of Chicago Booth School of Business, Chicago, Illinois, United States
We investigate the impact of a price discount on one product on the purchasing behavior of other product categories in the context of gasoline retailing. Analyzing data from over 20,000 gas stations throughout the U.S., our study reveals a positive correlation between cash transactions for gas and subsequent in-store spending. By examining individual data from consumers who visit stations offering cash discounts, we find that encouraging consumers to pay for gas with cash results in an 8.7 percentage point increase in their likelihood of making in-store purchases. We establish the causal effect of using cash for gas on in-store purchases using an instrumental variable approach. We further show that this effect is primarily driven by consumers' need to physically enter the store when choosing cash payments. Additionally, we develop a demand model that incorporates shopping costs, offering insights into the economic benefits of employing loss-leader pricing strategies for retailers.
2 - Product Returns as Reminder Advertising: An Empirical Investigation
Kissan Joseph, University of Kansas, Lawrence, KS, United States, Lauren Min, Boram Lim

Product returns are a significant element of today’s retail environment. To date, a robust finding documented in the literature is the positive impact of return incidence on the level of future sales. Not surprisingly, academic researchers have called for research that unpacks the consumer behavior aspects of this practice. Accordingly, in this research, we propose that a product return event is an exogeneous stimulus that is akin to a reminder ad that highlights the return feature offered by the retailer. Thus, similar to the impact of a reminder ad, sales in the next period are boosted. Empirically, in our analysis of consumer purchases at a South Korean grocery retailer encompassing 1200 matched consumers, we find, on average, a 33.9% increase in basket size among consumers “treated” with return incidence. Moreover, this increase decays with time and is more pronounced for consumers with relatively smaller basket sizes and higher variety seeking propensities in the pre-period. These heightened effects are consistent with reminder advertising being more effective among consumers with greater opportunity to increase their basket size and those benefiting to a greater extent from the safety net of a returns policy. We conclude by outlining specific managerial implications associated with the design of return policies.

3 - Logos and Experiences: A Study of New Brands
Fang Liu, University of Western Australia, Perth, Australia

A brand logo is a representation of a brand which usually contains a wordmark, a graphic symbol, or both. Brand logos are often consumers’ first contacts with brands, and previous branding and advertising research has found that a brand logo significantly influences how consumers evaluate a new brand. Little attention, however, has been paid to the relationship between brand logos and consumer experiences. Employing a quasi-experimental design and using large consumer panels from multiple countries, this paper examines the effectiveness of brand logos for new brands based on different symbolic design principles. This paper further investigates how brand logos interact with cultural values to influence consumer brand adoption. Results show that brand logos significantly influence consumers’ imaginative experiences, which result in differences in purchase intentions. Another core findings is that the extent to which brand logos influence consumers’ imaginative experiences appear be contingent upon cultural values, such as the level of uncertainty avoidance. Overall, the findings of this study suggest that businesses should adopt a more experience-based approach for product packaging and labelling, especially when marketing new products and services in international markets.

Keywords: brand logo, imaginative experience, quasi-experimental, labelling, packaging, new brands, international markets, cultural differences

4 - Unravelling Service Market Disruption: The Role of Complementarity Effects in a Retail Context
Pei Yu Chien, National Tsing Hua University, Hsinchu City, Taiwan, Jack Cadeaux

The introduction of new service assortments by a new service entrant can disrupt incumbents in a retail context, which is called service market disruption. Numerous studies have explored disruptive innovation, many of which are conceptual. Furthermore, while some empirical studies investigate the antecedents of disruptive innovation, they generally concentrate on disruptive product innovation. This study examines how the complementary of a given service category with other service categories sold by a retailer can accelerate the market-wide disruption of an incumbent service provider. Depending on the two types of inter-category complementarity links that are created by market basket lift and confidence measures (i.e., non-directional link and directional link), we identify three different category-level centrality measures within two inter-category complementarity networks: non-directional centrality, in-degree centrality, out-degree centrality. Results reveal that the non-directional centrality of a focal service with other service categories has a significant and positive effect on disruption. In addition, this study finds the out-degree centrality of a focal service category with other service categories can better predict the disruption than the in-degree centrality of a focal service category with other service categories.

SA12
Meeting Room C4.6
International Marketing Research
Contributed Session
Chair: Nayoung Jung
1 - Do Political Ties Help or Hurt Cross-Border M&As? An Empirical Examination of Home-Based Managerial Political Ties in Cross-Border M&A Completion by Emerging Market Firms
Chenxi Zhou, Xiamen University, Xiamen, China, People's Republic of, Kent Hui

Although emerging market firms (EMFs) have been aggressively pursuing cross-border mergers and acquisitions (CBMAs), many of their announced deals have collapsed, making the determinants of deal completion a critical research topic. This study develops a dual-legitimacy perspective that considers both the home- and host-market legitimacy of EMF acquirers to examine how their home-based managerial political ties (HBMPTs) influence the likelihood of completing a CBMA deal. Based on around 20 years of data on CBMA deals announced by Chinese firms, we find that the HBMPTs of EMF acquirers can improve the likelihood of CBMA completion. However, they are particularly beneficial when an EMF expands from a home region with a low level of factor market development to a host country with a low institutional distance. Conversely, when an EMF expands from a home region with a high level of factor market development to a host country with a high institutional distance, HBMPTs become marginally detrimental to CBMA completion. Such findings reveal a paradoxical role of HBMPTs in CBMA completion. Regarding potential tools to mitigate the adverse effects of HBMPTs, we find that managerial overseas connections are more effective than the firm’s corporate social responsibility in its home country. These empirical results shed new light on how EMFs’ political capital influences their global market entry.

2 - Chic or Happy? Cross-cultural Preferences in Fashion through Image Mining
Yijun Chen, University of New South Wales, Sydney, Australia, Suhyoung Ahn, Byoungho Jin, Jiwoon Kim, Junbum Kwon, Yoo Won Min

While emotions are universal, their expression varies across cultures. In Western contexts, expressions of happiness and joy are often preferred, whereas in Eastern cultures, more low-arousal emotions are valued. However, this preference can depend on the context. The fashion industry typically sees chic expressions as more stylish or appealing compared to happy or joyful expressions. Global fashion brands indeed often tailor their marketing strategies to reflect the cultural nuances of their target markets. Do global fashion brands modify the images in their posts depending on the country? Is there a cultural variation in the preference for happy expressions in these fashion brand posts? Additionally, could the color characteristics of these images influence this effect? To get the answers, this study proposes a research framework that explains cultural differences in the preference for the expression of emotions in fashion marketing. In collectivist cultures, expressions of happiness are likely to be seen as contributing to social cohesion and positive group dynamics. Thus, we propose that in the context of global fashion brand posts, collectivist cultures may demonstrate a stronger preference for expressions of happiness compared to individualistic cultures. Additionally, this relationship appears to be influenced by the impact of image color characteristic. For instance, in Eastern cultures, the use of warm hues in fashion brand posts can enhance the appeal of happy expressions. These warm colors, associated with warmth and social harmony, complement and intensify the positive emotional impact of happiness, aligning with cultural preferences for communal cohesiveness.

3 - How Does Watching Global Country Video Influence Consumers' Digital Shopping?
Nayoung Jung, Yonsei University, Seoul, Korea, Republic of, Sue Ryung Chang

As the significance of video-sharing platforms continues to rise globally, marketers are enthusiastic about harnessing these platforms to influence consumer shopping decisions. In particular, the advent of YouTube renders it easier for consumers to access videos related to various countries and cultures. However, empirical studies are still insufficient in exploring the influence of watching videos regarding global countries on consumers’ digital shopping behavior. By examining a dataset about the video-watching history of consumers and their usage patterns on shopping websites/apps, the authors find that watching videos related to global countries can increase the digital shopping duration of consumers. The authors also discover that high familiarity with certain countries can strengthen this relationship, while negative sentiment towards those countries moderates the relationship. The authors further uncover that consumers' general shopping patterns can moderate the effect of watching global country videos on digital shopping duration. This study extends the existing research on video content in marketing, focusing on elucidating factors influencing digital shopping behaviors. It also offers guidance for companies aiming to enhance consumer shopping duration through the development of effective marketing communication strategies via digital media.

Keywords: digital shopping, country-related video, global marketing, country characteristics, consumer shopping pattern

SA13
Meeting Room C4.7
Sustainability: Modeling and Predicting Sustainable Behavior
Contributed Session
Chair: Jochen Eckert, Wiesbaden Business School, RheinMain University of Applied Sciences, Bleichstraße 44, 65183
1 - CANCELLED Eco-Movement Echoes: Tracing Sustainability Shifts with GPS Data
Liyang Zhou, Bocconi University, Milano, Italy, Sara Valentini, Qiaoni Shi
With household emissions contributing a significant two-thirds of global greenhouse gas emissions (United Nation, 2020), understanding the drivers behind consumer shifts towards sustainable behaviors is crucial for reducing individual carbon footprints. This paper analyzes the role of external shocks like eco-movements in enhancing consumer awareness and adoption of sustainable practices. Specifically, we investigate the influence of Greta Thunberg's 2019 environmental activism, including her transatlantic voyage, her Climate Action Summit speeches, and the New York City climate strike. Utilizing device-level mobile GPS location data and analyzing Places of Interest (POI) information, we employ a unique approach to track consumer participation and exposure to these events, and to explore how and where consumers engage in sustainable activities, such as transportation habits, interactions with sustainable brands, and energy usage patterns. We then scrutinize these individual-level sustainable behaviors pre- and post-external shocks to examine the impact of eco-movements, with a focus on distinguishing between short-term and long-term effects. We also investigate heterogeneous effects, differentiating between consumers who were actively involved in these events and those who were passively exposed. Our findings reveal the potential of eco-movement events as a strategic tool to foster sustainable consumerism, offering valuable insights for both policymakers and marketers in promoting environmental sustainability.
2 - From Gamification to Greenification: How Cognitive Absorption and Perception of Corporate Prosociality Make People Go Green?
Qin Zhang, Pacific Lutheran University, Parkland, WA, United States, Junping Ma

Human-centered gamification literature has focused on cognitive beliefs and hedonic motivation to elucidate the formation of individuals’ eco-friendly behavioral intentions, yet little is known about the impact of cognitive absorption—a proximal antecedent to cognitive beliefs—and consumers’ perceptions of corporate prosociality (CPCP). We propose that cognitive absorption and CPCP are critical psychological factors in causing individuals to volitionally engage in eco-friendly behaviors. We also investigate, via a mixed-methods approach, how individuals’ value perceptions of a gamified program impact their subsequent cognitive absorption experience. In contrast with traditional studies, we adopt the notion of polygamous intention, which encompasses two related endogenous variables: the program participation intention and the eco-friendly behavioral intention. Results from an experiment of 106 actual customers and follow-up surveys that involved 378 active subscribers of Alibaba’s Ant Forest—a global pro-green gamified lifestyle platform—indicate that gamification mechanisms and value perceptions positively affect individuals’ cognitive absorption which, in turn, positively influences CPCP. While CPCP and program participation intention directly affect individuals’ eco-friendly behavioral intention, cognitive absorption also influences their program participation intention. By integrating a cognitive belief antecedent and CPCP in a single model, we provide a more complete understanding of individual behaviors relating to eco-friendly gamification, thus contributing to both behavioral information technology research and practice.

3 - Small Change, Big Impact? Displaying Technology-Based Just-in-Time Information to Encourage Fuel-Efficient Route Choices
Jochen Eckert, Wiesbaden Business School, RheinMain University of Applied Sciences, Wiesbaden, Germany, Franziska Schechner, Tatjana Steusloff
Reducing greenhouse gas emissions from road transport is essential to limit global warming. Passenger cars are responsible for approximately 12% of total carbon dioxide (CO2) emissions in the EU alone. Public policy often relies on costly and time-consuming mechanisms to reduce CO2 emissions, such as investments in cycling infrastructure or subsidies for zero-emission vehicles. We explore alternatives that are more cost-effective, faster to implement, and easier to scale up by leveraging technology-based just-in-time information on the environmental impacts of choice alternatives. We thus addressing one of the main barriers to sustainable consumption identified in research, namely uncertainty about the environmental impacts of different decision options and the difficulty of distinguishing good from bad options. Our research focuses on fuel-efficient route choice, a critical but often overlooked factor influencing CO2 emissions. Using discrete choice experiments that mimic the real-world user experience (UX) of Google Maps' eco-route implementation, we show how different presentations of just-in-time information about the environmental impact of route alternatives can significantly alter consumers' choices, leading to stark differences in eco-route adoption and consumers' willingness to trade travel time to save emissions. We apply different choice architecture interventions to fill gaps in route choice research and quantify their individual and combined effects. Our data from 4,639 German car users shows that compared to Google's current UX, the best-performing presentation of just-in-time information can save an additional >300,000 metric tons of CO2 emissions per year globally, the equivalent of taking approximately 70,000 fuel-based cars off the road for a year.
SA14
Meeting Room C4.8
Dynamic Model of Learning, Pricing, and Competitive Strategies in the Digital Economy
Invited Session
Special Sessions
Chair: Chuan He, University of Colorado-Boulder, Boulder, CO, United States
1 - Gender Differences in Bargaining and Selling
Pranav Jindal, Indian School of Business, Chapel Hill, NC, United States, Minkyung Kim, Peter Newberry
We study whether there exist any differences in salespeoples’ wages by gender in a setting
where salespeople sell products and negotiate prices. Based on two years of transactions data from a large durable goods retailer in the U.S., we find that women salespeople earn 23% more than the average salesperson. Differences in demographics and products and order characteristics across gender explain some but not all of the observed gender wage gap. We find that women are better at selling products which have increasing marginal returns to negotiating, and are also better at meeting quota, which again results in a disproportionate increase in wages. Women are also better than their counterparts at negotiating higher prices which is consistent with previous research where gender wage gap is attenuated when women negotiate for someone else or in environments with lower ambiguity (Bowles et al 2005, Bowles and McGinn 2008). Finally, we explore the importance of learning in driving the gender wage gap and find that women perform worse than men at the time of joining but they learn faster than men. We explore the nature of learning and find that women learn to sell more over time but not how to negotiate better, which women do better even at the time of joining.
2 - Streaming Alliances
Abhinav Uppal, Indian School of Business, Hyderabad, India, Nanda Kumar, Manish Gangwar
The OTT video streaming industry has witnessed significant growth in recent years, marked
by the influx of new players. The arrival of new streaming services has heightened competition within the market. At the same time, we are observing an interesting phenomenon where competing services are forming new alliances that facilitate consumer multi-homing. Amazon, for example, offers add-on subscriptions to its allied services, such as HBO, enhancing the combined viewing experience for its customers through seamless integration. In this paper, we build a game-theoretic model to examine how an alliance facilitating multi-homing between two streaming services impacts the price competition in the market. We find that the alliance’s impact on price competition depends on the level of content differentiation in the market — competition intensifies when differentiation is high but relaxes when differentiation is low. Interestingly, the alliance may increase the profitability of a third non-partnering service when the content is sufficiently undifferentiated. We also investigate the partnership decision of a focal streaming service to shed light on who it should partner with and on what terms. Our research offers insights into the current landscape of the OTT streaming market and provides implications for both managers and policymakers.
3 - Auditing Bandits for Online Personalization
Bob Rombach, Rotterdam School of Management, Erasmus University, Rotterdam, Netherlands, Gui Liberali, Yang Li
Reinforcement learning algorithms, such as those based on multiarmed bandits (MAB), have been successfully employed in various marketing resource allocation scenarios including banner advertising as well as other personalization practices. Although these on-the-fly learning algorithms arehighly efficient and accurate, there are concerns regarding the potential propagation of discriminatory societal or systemic biases towards customers. Such adverse effects have prompted calls for regulations like the General Data Protection Regulation (GDPR) by the European Union that protects sensitive customer characteristics. However, we challenge the prevailing beliefthat not using individual sensitive attributes would benefit consumers, and present empirical evidence that such a data restriction in the MAB context could hinder efforts to combat discrimination and inadvertently lead to unjust treatment of minority consumers. Moreover, we propose a real-time auditing framework for MAB thatnot only maximizes the total expected rewards but also safeguards the welfare of all consumers under explicitly selected ethical principles. To achieve this, we employ a differential privacy strategy to separate the access of protected user variables fromthe MAB-based personalization engine. In the process of auditing, we evaluate different finite resource allocation rules and study the intricate relationship between learning and fairness. This innovative MAB approach canbe readily applied to online digital services seeking to allocate finite marketing resources while controlling discriminatory personalization outcomes. The empirical findings also help “myth bust” the notion that data protection laws should simply prohibit the processing of sensitive user data, and instead encourage a better and fairer approach to utilize such data for MAB applications.
4 - Market Differentiation and Behavior-based Pricing
Ruichao Pan, Sciences Po, Paris, France, Wilfred Amaldoss, Chuan He, Shaowei Ke
We examine behavior-based pricing (BBP) in a setting that incorporates both vertical and horizontal differentiation to explore how these two types of product differentiation influence firms’ pricing strategies and the profitability of BBP. Two key parameters influencing the analysis are identified: the cost-adjusted quality difference between firms, which measures their competitiveness in the vertical dimension, and the transportation cost to serve, which gauges the level of horizontal differentiation. When firms possess competitive advantages , it becomes easier for them to retain old customers despite offering higher retention prices. However, due to customers being forward-looking and dynamically optimizing, they anticipate the increased second-period retention prices and strategically choose the firm that is at a relative competitive disadvantage in the first period. This benefits the less advantaged firm. Consequently, when horizontal differentiation is sufficiently strong compared to vertical differentiation, the high-quality firm can profit from BBP. Conversely, when horizontal differentiation is sufficiently weak compared to vertical differentiation, the low-quality firm can profit from BBP. This effect is counterbalanced by the force of relative discounting.
When firms and customers are allowed to discount the future differently, we find that if customers are significantly less patient than firms, no firm can profit from BBP. Specifically, the high-quality firm offers better prices to its past customers when horizontal differentiation is sufficiently weak relative to vertical differentiation, whereas the low-quality firm offers better prices to its past customers when horizontal differentiation is sufficiently strong relative to vertical differentiation.
5 - On-Demand Services and Product Line Design
Bing Jing, Cheung Kong Graduate School of Business, Beijing, China, People's Republic of
On-demand services (e.g., DoorDash and Instacart) have emerged and prospered in several industries such as restaurants and supermarkets. For consumers, using an on-demand service carries the following trade-off: It lowers their transaction costs by sparing a trip to the store, but also decreases their consumption utility due to reduced service level and lack of personal inspection. In a monopoly of vertical differentiation, I study how on-demand services affect the firm’s product line structure. I show that on-demand services favor a longer product line. The reason is as follows: Transaction costs hinder the firm’s ability to extract rent from the low-end market, and on-demand services lower the transaction costs and facilitate second-degree price discrimination. Although on-demand services seemingly add an alternative purchase channel for consumers, they (weakly) increase the firm’s profits and (weakly) decrease consumer surplus. I further examine how on-demand services impact price competition in a duopoly. Here on-demand services may intensify, relax, and does not affect price competition depending on market conditions.
SA15
Meeting Room C4.9
Customer Satisfaction
Contributed Session
Chair: John Dawes, Ehrenberg-Bass Institute, UniSA, 29 HAZELWOOD AVENUE HAZELWOOD PARK, ADELAIDE, 5066, Australia
1 - I'm Sorry: Consumers' Responses to Service Remedies
Julie Edell, Duke University, Durham, NC, United States

This study investigates the effects of the ways service providers resolve a service failure and the impact of an apology and various levels of compensation after a service failure. Using scenario methodology, respondents recorded their emotional responses to the service failure (a long flight delay) and the remedy offered by the airline. Participants were assigned to one of six conditions, varying the amount of compensation received and whether or not the service provider offered them an apology.

The results show that issuing an apology for the service issue significantly effected the consumers’ emotional responses, such that they feel less regret and more satisfaction with their decision to use this provider. Consumers were less angry and happier with the provider when they apologize and even show marginally less disappointment and more delight with the bad outcome. Surprisingly compensation amounts offered ($10, $100 or $350) made no difference in levels of regret, satisfaction, anger or the decision to use this airline again.

However, an apology coupled with ten dollars was as effective in reducing regret, anger and disappointment as receiving $350 but with no apology for the flight delay. Similarly, an apology coupled with ten dollars was as effective in increasing satisfaction, happiness, willingness to use again and recommend the airline as was receiving $350, but without an apology. Mediation analysis showed the emotional responses mediated the willingness to use the provider again and recommend it to others.

2 - A Study of Sustainability Tourism in Japan's Inbound Market
Tatsuru Nishio, Yamaguchi University, Yamaguchi, Japan, Akane Kakishima

The aim of this research is to analyze how destination sustainability initiatives and efforts to address disparities in the inbound market affect tourist satisfaction. In the post-COVID era, new competition arises for destinations, emphasizing the need to devise marketing strategies for sustainable tourism. Destinations must address community understanding (responsible tourism) and disparities among visiting tourists (accessible tourism) when formulating strategies.

Data were collected through a survey targeting individuals who traveled to Japan from the top 12 tourism-sending countries (N=1448), including South Korea, China, Taiwan, Hong Kong, Thailand, Singapore, Malaysia, Indonesia, the United States, Australia, the United Kingdom, and France. The survey included questions about basic attributes such as tourists' country of residence, age, and gender, as well as Likert scale items on "① Satisfaction when visiting natural tourist destinations," "② Destination's response to tourism initiatives," and "③ Response to inequality."

The analysis was conducted using regression analysis and structural equation modeling. In the first stage of regression analysis, "② Responsive Tourism Response" and "③ Response to Inequality (8 items)" were each regressed on the dependent variable "① Satisfaction." In the second stage, after excluding items with strong correlations in "③ Response to Inequality (VIF>3)," the relationship between "① Satisfaction," "② Responsible Tourism Response," and "③ Response to Inequality (2 items for the elderly and wheelchair accessibility)" was analyzed using structural equation modeling.

The study obtained implications for marketing strategies and identified future research challenges from the overall model, regional models, and country-specific models for the top 12 countries.

3 - Does High Customer Satisfaction Generate Positive Non-customer Impressions of the Firm?
John Dawes, Ehrenberg-Bass Institute, UniSA, ADELAIDE, South Australia, Australia
Customer satisfaction is said to pay off in increased sales from existing customers, as well as being the basis of favorable word of mouth about the firm. A large body of work indicates a positive link between satisfaction and willingness or likelihood to recommend a firm. However, the problem of response bias exists for much of this work (i.e. 'I said I was satisfied, of course I'm likely to recommend'). This study overcomes that issue by examining aggregated satisfaction scores and likely-to-recommend scores from customers of a wide range of firms using YouGov survey panel in the UK and USA; and analysing them against the impressions of those firms among non-customers via the YouGov panel. In other words: have satisfied customers spread a favorable impression of the firm to non-customers? The analysis will be undertaken first, cross-sectionally: across a sample of firms, do the ones with higher satisfaction scores also have more positive impressions among non-customers? The analysis will also examine within-firms, over time: if a firm's customer satisfaction score increases or decreases markedly, will we see its positive impression among non-customers change in the same direction? Preliminary analysis for Banks and Insurance companies shows that while satisfaction and positive recommendation metrics are positively correlated, there appears to be little association with non-customer impressions. The analysis calls into question the claim that satisfaction pays off in positive word of mouth directed to non-customers.
SA16
Meeting Room C4.10
Consumer Behavior: Product Information
Contributed Session
Chair: Rafay Siddiqui
1 - Spatial Sensitivity: Unveiling Its Effects on Consumer Choice and Perception
Yehjun Lee, Virginia Tech, Blacksburg, VA, United States

Spatial cognition plays a pivotal role in human perception. For example, if an object has moved 10 meters away, individuals’ perceptions may differ regarding how far the object has moved. Similarly, if a building is 100 meters away, there will be variation regarding perceptions of how far away the building is. There is a notable gap in understanding individual sensitivity to spatial changes and how it influences consumption behavior. This paper addresses this gap by exploring two critical questions: How can individual differences in sensitivity to spatial changes be effectively measured, and what impact do these differences have on perception and behavior?

Through a series of experimental studies, we show that individuals with higher sensitivity to spatial changes exhibit a preference for immediate, smaller rewards, possibly due to an amplified perception of the distance to delayed rewards. These individuals also tend to perceive products as larger, which could influence their willingness to pay, suggesting significant implications for consumer decision-making.

Our findings contribute to both theoretical and practical aspects of marketing. We challenge some assumptions of Construal Level Theory and suggest potential integration with cognitive load theories, offering a fresh perspective on spatial perception's role in consumer judgments. This research advances our understanding of how spatial perception can predict and influence key consumer traits like impatience and self-control. Such understanding is crucial in developing strategies to mitigate impulsive buying tendencies and enhance self-regulatory practices.

2 - Primacy and Recency in Consumer Reference Point Formation: A Consideration Time Model
Francisco Cisternas Vera, Chinese University of Hong Kong, Ma Liu Shui, Hong Kong, Andrew Meyer

Prospect Theory and other reference-dependent models of choice have been incorporated throughout psychology and economics. However, they fail to generate clear predictions in many marketing contexts, in part because consumer reference point formation is not yet well understood. Scanner panel purchase data suggest that consumers’ reference points are primarily influenced by the most recently offered products. We find evidence that this reverses at shorter time scales – such as within an individual shopping session.

We report data from three large-sample experiments in which a total of 12,072 participants each make dozens of independent binary product-purchase decisions. In all experiments, products are randomly assigned to trials, creating exogenous variation in the attribute levels of recently encountered products. As expected, we find large effects of this variation: encountering strong products on previous trials reduces the likelihood of purchasing any product on the current trial. More surprisingly, we find that primacy effects dominate recency effects – the first product encountered has a larger impact than the most recent product, even after consumers have considered dozens of products.

We show that this primacy effect can be explained by longer consideration times for initially encountered products, and we propose a model of reference point formation in which the influence of each reference value is proportional to the time spent considering it. We estimate this model and validate our causal inference with instrumental variables and placebo tests. We demonstrate that weighting reference values by consideration time substantially enhances the predictive validity of consumer reference point estimates.

3 - Attribute Granularity and Consumer Choice
Xin Ge, University of Northern British Columbia, Prince George, BC, Canada

Does the granularity — i.e., the level of detail — at which product attribute information is presented have a systematic impact on consumer choice among alternatives? In a series of six experiments, this research shows that a more fine-grained characterization of alternatives on an attribute dimension (e.g., in terms of two distinct aspects of performance instead of overall performance) promotes choice of the alternative with the most favorable level of that attribute, and that it does so by influencing consumers’ judgments of attribute importance. This effect is attenuated as it becomes easier for consumers to consolidate the fine-grained aspects of an attribute dimension. Moreover, the impact of attribute granularity on choice diminishes (and even reverses) as the anticipated delay between choice and consumption increases. This research demonstrates that the level of detail provided about an attribute dimension has a systematic influence on consumer choice among alternatives, and it sheds light on the mental mechanism that underlies this effect.

4 - The Effect of Product Information Cues on Intertemporal Choice
Rafay Siddiqui, Santa Clara University, Santa Clara, CA, United States, Feifei Huang, Qianqian Liu
In an intertemporal choice, consumers face a choice between smaller-sooner rewards and larger-later rewards. For example, a consumer could either choose to purchase the current version of a smartphone right now (smaller-sooner reward), or an upgraded version of the smartphone that will be released three months from today (larger-later reward). Past work has investigated factors that influence such choices, such as the framing of the rewards, and the framing of the wait time for the larger-later reward. In this work, we investigate how common cues present in product information may affect intertemporal choice. Specifically, we look at how sustainability-related information may increase preference for the larger-later reward. We find that the presence versus absence of a sustainability-related cue in product information shrinks perception of the length of the wait time for the larger-later reward. Consequently, sustainability-related cues increase preference towards the larger-later reward. We theorize that the relationship between sustainability and a focus on the future leads to the larger-later reward feeling closer in time when a sustainability-related cue is present. These findings have significant theoretical and practical implications. We contribute to the intertemporal choice literature by identifying a novel antecedent of wait time perception and patience in intertemporal choice. Likewise, our findings enable product managers to learn how different types of product information may influence consumers’ wait time perception and consequently their preference to delay their receipt or purchase of a product.
SA17
Meeting Room C4.11
Mobile Apps
Contributed Session
Chair: Xuejing Ma, East China Normal University, United States
1 - The Impact of COVID-19 Lockdown on Digital Lifestyle in China
Shi Jia, University of Hong Kong, Hong Kong, Hong Kong, Jianmin Jia, Yun Yuan

A basic question in digital marketing is the extent to which different aspects of digital lifestyle and consumption depend on physical interactions and mobility. We disentangle the relationship between digital consumption and physical activities by combining the mobile app data with city-level foot traffic and instore credit card spending data. In particular, we use anonymized individual-level mobile phone data of 80,533 randomly selected active users of a national carrier in Chengdu, Sichuan. The dataset includes, at the individual-user level, daily data usage for all mobile apps, mini-programs, and web browser destinations (https), and spans December 1, 2019 to April 30, 2020, and July 2 to August 14, 2020, during which there were 3 deescalating levels of lockdowns.

We leverage the mobility restrictions in regression discontinuity in time with multiple cutoffs analysis to quantify the impact of the mobility restrictions on mobile app usage, and also test for age and gender inequalities. Furthermore, we investigate if consumers’ digital consumption patterns (e.g., social media vs. messenger vs. education apps, etc.) are driven (and constrained) by their physical mobility and offline behavior. We used three different mediators; intra-city mobility, outflow mobility (Baidu), and transaction frequency of offline Union Pay payments (at the aggregate-level). Overall, the mediation models demonstrate that physical mobility and activities are fundamental forces driving digital behaviors and lifestyles (for which there are significant demographic heterogeneities), and that restrictions on human mobility were largely responsible for the reduction of digital activities during the pandemic in China.

2 - The Impact of Mobile Apps Improvements on Consumer Purchasing Attitudes
Keita Arakawa, Keio University, Tokyo, Japan

As e-commerce develops, companies are beginning to use their own mobile apps to implement omnichannel strategies. They are developing their own mobile apps to engage customers and create new communication channels with them. Restaurant chains, in particular, are offering membership registration functions, coupon distribution, store locator, and ordering services through mobile apps through their apps. In light of these omnichannel strategies, many studies have focused on mobile apps. However, most of those studies mainly deal with the addition of mobile apps, and few studies focus on enhancing and improving mobile apps. In addition, many studies have focused on company performance and stock returns, and few studies have examined the impact on the actual consumer image and purchasing reality. To address this gap, this study focuses on a global fast food chain that relaunched its mobile app in Japan in February 2021, implementing a member registration function and improving the UI/UX. Using two-time points of data on the image and purchase frequency of approximately 3,000 Japanese consumers before and after the company relaunched its mobile app, this study investigated the impact of the mobile app improvements on consumers' image and purchase frequency toward that fast food chain. The results showed that mobile app improvements significantly increased purchase frequency, although they did not affect consumer image. This study extends research on mobile apps and omnichannel and contributes to practice by showing that investments in mobile apps increase the frequency of purchases by consumers.

3 - Research on the Purchasing Behavior of the Leading-Edge Group Using Receipt Data
Shinichiro Hata, Smart Analytics, Tokyo, Japan, Akira Shimizu, Tadashi Shinoda, Yuko Sakai, Kazumi Sasaki

In this study, we conducted an awareness survey of smartphone receipt application users, matched it with purchase history data, and analyzed the relationship between awareness and behavior from various angles. The users of receipt applications are characterized by the fact that many of them are young people due to the nature of the data collected via smartphones, that they can obtain purchase history at places such as fast food restaurants and convenience stores that cannot be captured by regular card data, and that the receipts contain information on the stores, so the range of a person's activities can be determined. The receipt includes information on the store, so the scope of the person's activities can be determined. In this study, we used these data characteristics to identify the leading-edge consumers through a questionnaire survey and analyzed the stores they use and the products they purchase. The analysis revealed that the leading-edge group purchased more healthy products than the other groups, even within the same product category. They are often used in upscale stores. At the same time, the non-leading-edge shoppers were more likely to support bargain products and discounted products. Furthermore, even within the same region of Japan, the leading-edge group had different preferences depending on the region in which they lived, with a preference for fashionable stores in Tokyo and price-oriented stores in Osaka. The combination of receipt and awareness data reveals what conventional purchase history data cannot reveal.

4 - The Spillover in Mega Apps
Xuejing Ma, East China Normal University, Shanghai, China, People's Republic of, Yuanchen Yang
Mega apps, such as Meta, WeChat, and Alipay, with their integration of functions, have gained significant prominence in the digital landscape. However, despite their widespread usage, little is known about the intricate interplay among different functions within these apps. This study addresses this question by investigating the impact of Yu’ebao adoption on digital payment usage within the Alipay ecosystem. Utilizing a unique and large-scale dataset, our findings reveal a significant positive spillover effect, with Yu’ebao adoption leading to increased digital payment usage, particularly in offline channels. The spillover effect is sustained over an extended period, and becomes even more pronounced in the long term. We also find that the spillover effect primarily occurs through the substitution of traditional payment methods rather than a shift from other digital payment channels. Our findings underscore the significance of spillover effects within mega apps and carry managerial implications pertaining to the promotion of digital payment and the strategic utilization of spillover effects.
Saturday, June 29, 10:30 AM - 12:00 PM
SB01
Cockle Bay Room 1
Reducing Discrimination and Inequities through Interventions
Invited Session
Special Sessions
Chair: Dionne Nickerson, Indiana University, Bloomington, IN, United States
1 - The Divergent Effects of Cognitive and Emotional Empathy on Marketing Performance: Evidence from Microfranchising in a Base-of-the-Pyramid Market Context
Sourindra Banerjee, University of Leeds, Leeds, United Kingdom, Jarrod Vassallo, Jaideep Prabhu, Toby Norman
Microfranchising is a ubiquitous form of “Inclusive Marketing” used to penetrate base-of-the-pyramid markets that is nevertheless understudied, even within marketing. This approach is unique because, besides generating a surplus for the host organization, it aims to empower the poor and disenfranchised to become entrepreneurs (i.e., microfranchisees) and to provide beneficial new products to previously excluded customers. Yet goal asymmetry between microfranchisees and host organizations can lead to sub-optimal outcomes for organizations, microfranchisees and customers. Given this challenge, we ask: ‘which individuals should host organizations select as microfranchisees to improve marketing performance and reduce therisk of perverse outcomes?’ Accordingly, we propose selection based on two types of empathy with divergent effects on performance: cognitive empathy (seeing the world from a customer’s perspective) and emotional empathy (feeling what a customer feels).
We test our empathy divergence thesis using a unique dataset that we collected from a microfranchising operation within the legal service industry run by BRAC, one of the world’s largest social businesses that operates in the base-of-the-pyramid markets. Combining multi-method (archival and survey) and multi-respondent (microfranchisees, host organization managers, and customers) data covering 49-months, our dataset provides a detailed view on microfranchisee pricing decisions, sales volume, service details,and customer demographics.
Our findings show that while greater cognitive empathy of microfranchisees drives up price, sales volume and customer satisfaction, greater emotional empathy drives down price and customer satisfaction while increasing sales volume. More generally, we show emotional empathy plays a complementary role to cognitive empathy in its impact on sales volume, and
2 - The Usage and Impact of Differentiation: Evidence from an Online EdTech Platform
Zekun Liu, New York University, New York, NY, United States, Tulin Erdem, Masakazu Ishihara
We explore the impact of a digital differentiation tool on student learning outcomes across socioeconomic segments, considering teachers' usage and implementation. Using a structural model with a hidden Markov model, we assess the tool's effectiveness in improving student performance and addressing educational disparities. Our findings indicate potential improvement in student outcomes, but limited teacher usage hinders actual effectiveness, regardless of socioeconomic background. Teachers from different segments show varied preferences, with low-poverty school teachers prioritizing medium-achieving students more. To enhance student learning outcomes and address educational disparities, we explore possible interventions in our counterfactual analyses. Cost reduction boosts tool usage and benefits students but disproportionately favors low-poverty schools, widening gaps. Targeted professional training in high-poverty schools could mitigate this, bridging the usage divide. Our results emphasize considering both product design and usage divide to maximize differentiation tools' impact and promote equitable education.
3 - Examining Consumer and Investor Responses to Corporate Social Responsibility Initiatives and Firm Diversity Targets
Dionne Nickerson, Emory University, Atlanta, GA, United States, Karen Wallach, Beth Fossen
Despite increased stakeholder expectations of firms’ diversity efforts, progress in this domain remains challenging for many firms, and limited research has explored the effectiveness of different diversity efforts across multiple stakeholders. Inthis research, we investigate the impact of two crucial firm diversity efforts – corporate social responsibility (CSR) initiatives and diversity targets – onthe responses of two key stakeholder groups – investors and consumers. We specifically examine diversity efforts in the form of diversity targets, which firms utilize to signal their commitment to diversity. Four critical questions guide our research: (1) Do consumers and investors respond differently to CSR and diversity initiatives? (2) How do consumers and investors differ in their reactions? (3) What is the impact of diversity targets on responses to a firm’s CSR initiatives? (4) Are views on diversity initiatives changing over time?
Using an event study methodology, we examine over 700 CSR initiatives by 41 firms from 2015 to 2020. We assess consumer reactions through online word-of-mouth (WOM) by analyzing almost 1MM Twitter responses and measure investor reactions using cumulative abnormal returns (CAR). Our results show that CSR initiatives by firms that have set diversity targets spur about 40% less positive WOM for these firms. However, the reduction in positive consumer WOM reverses when the CSR initiative pertains to diversity. Specifically, when firms set diversity targets, their diversity CSR initiatives generate a more than 200% increase in positive WOM for the firm. This pattern of results is consistent across all five years of our data window.
SB02
Cockle Bay Room 2
Influencer Marketing: Network, Quality and Content Effects
Contributed Session
Chair: Flora Gu, Hong Kong Polytechnic University
1 - Peer Effects in Sponsored Content Creation on Social Media Platforms: Mechanisms and Economic Implications
Mengyang Sun, Shanghai University of Finance and Economics, Shanghai, China, People's Republic of, Qi Sun, Diandian Xiang

We quantify the impact of peer competition effects in influencers’ sponsored content creation on social media platforms. Using detailed data on a leading social media platform in China, we document that influencers’ sponsored content creation decisions are significantly influenced by peer influencers, but such peer effec only affects the quantity of the sponsored content but not the quality of sponsored content. We further identify that such peer effects are primarily driven by the “defensive motive” of top influencers against smaller influencers and the “competition motive” between smaller influencers, but not driven by the “social learning motive” that smaller influencers follow top influencers or the “competition motive” between top influencers. Finally, we show that such peer effects drive influencers to purchase more in-feed advertising service from the platform, and lower their asking price of sponsored content, but do not lead to increased creative effort of the sponsored content.

2 - Influencer or Being Influenced? The Role of Multi-Channel Networks on Influencer Content
Yulin Hao, University of Rochester, Rochester, NY, United States, Xiaojie Li

Social media influencers are increasingly affiliating with multi-channel networks (MCNs). MCNs merge the roles of talent and advertising agencies and, more importantly, are rumored to be directly involved in content creation. This paper provides the first empirical examination of the effects of MCN affiliation on influencer content. We construct a unique dataset that allows us to track influencers' changes in their MCN affiliations. Using a difference-in-differences strategy, we compare influencers who switch their MCN affiliation status with observably similar non-switchers. Our findings reveal that MCN affiliation enhances content engagement and leads to more homogeneous and focused content, steering influencers towards topics with higher advertising prices. After being affiliated with MCNs, the influencers also have more sponsorships and charge a higher advertising price, which are predominantly driven by changes in content resulting from the affiliation. Next, we discuss the mechanisms behind these effects, highlighting the role of production and business resources provided by MCNs and the peer learning and collaboration within MCNs. Our findings suggest that platforms and influencers can benefit from improved engagement and sponsorships resulting from MCN affiliation. Although advertisers face higher advertising prices charged by MCN-affiliated influencers, these prices are justified by more engaging and focused content, which may also beneficial to the advertisers.

3 - How Do Social Media Influencers Affect Product Quality Perception? Evidence from Gaming Industry
Qi Yu, Singapore Management University, Singapore, Singapore, Zhe Lin
Social influencers are a major source of information from which consumers could collect product quality signals before purchasing, especially in industries like computer hardware, video games, and films. This paper studies how consumer’s product quality perceptions are affected by social influencers, who are under agreements with firms to strategically disclose product information. We construct a unique panel data on video games that contain the daily sales and followers of the 1500 top games on Steam, a leading gaming platform, and over 300,000 YouTube videos from top gaming influencers reviewing these games. We find that reviews produced through partnerships on average use more positive language than other reviews. However, these types of reviews are less effective in generating consumer interest and sales. In particular, these reviews bring in fewer game followers on Steam. Conditional on game quality, we don’t find the number and valence of these reviews to affect the sales volume of a game. Both findings suggest that consumers are sophisticated enough to correctly infer the quality of games even from reviews with missing information. We discuss the implications of our findings for product information disclosure.
4 - Does Content Novelty Generate Engagement in Influencer Marketing?
Flora Gu, Hong Kong Polytechnic University, Hong Kong, Hong Kong, Yiwei Li, Fine Leung, Xinyuan Chen

Influencer marketing, a rapidly expanding strategy, shows promise in boosting firm engagement, thus attracting increasing portions of company budgets. This study scrutinizes a prevalent assumption among both practitioners and laypersons: that influencer marketing captivates consumers through its generation of content more novel than firm-controlled methods. This belief stems from the roots of influencer marketing, where influencers organically shared their passions, interests, and consumption experiences voluntarily. However, with influencers now often paid as brand endorsers, the question arises: do they continue to produce novel content that engages consumers?

Our analysis, utilizing a dataset of Weibo posts, uncovers a paradoxical effect of influencer marketing investment on content novelty. Our findings indicate that heightened influencer marketing investment diminishes the influencer’s intrinsic desire to create value for followers, simultaneously amplifying the extrinsic motivation for monetary gain through brand-related goals. Interestingly, this negative correlation diminishes when firms engage influencers who exhibit higher follower interactivity, indicating stronger intrinsic motivation. Additionally, our findings depict an inverted U-shaped relationship between content novelty and consumer engagement. While distinctiveness helps content stand out in a crowded social media landscape, excessive deviation or extreme creativity may alienate wider audiences by catering to niche tastes.

SB03
Meeting Room C3.2
Consumer Behavior: Payment Options
Contributed Session
Chair: Belinda Barton, Bond University
1 - A Meta-Analysis on the Effect of Payment Methods on Consumers' Spending Outcomes.
Lachlan Schomburgk, University of Adelaide, Adelaide, Australia, Alex Belli, Arvid Hoffmann

Over 40 years of research links cashless payment methods to increased consumer spending. Referred to as the “cashless effect,” this phenomenon has recently come under scrutiny as consumers are increasingly familiar with cashless payment methods which could weaken the cashless effect, while other research challenges the robustness of the effect and questions which conditions could strengthen or weaken it. The current study aims to contribute to reaching a consensus in this ongoing debate through a large-scale meta-analysis leveraging a meta-analytical framework that synthesizes the insights from the extant literature. Across 391 effect sizes from 70 papers, we find evidence of a small, but significant, cashless effect. Further, we document that cashless payment method features, consumption situations, and contextual factors influence the cashless effect. Specifically, the cashless effect is stronger for cashless payment methods with higher physical transparency and in conspicuous consumption situations, while it is weaker in pro-social consumption situations. The results also reveal that the cashless effect is stronger in periods of economic growth, while it has generally become weaker over time. Our findings offer new insights for academics, consumers, and practitioners such as retailers, charities, and policymakers interested in the effects of payment methods on consumer spending behavior.

2 - The Dynamics of Payment and Pleasure: Monetary and Social Interactions in Shared Consumption Experiences
Belinda Barton, Bond University, Robina, Australia, Aimee Smith, Natalina Zlatevska

In the wake of a post-pandemic loneliness epidemic, consumer behaviour has shifted toward prioritising shared experiences in spite of increasing financial constraints. While prior studies have separately examined the effects of monetary transactions in gifting and financial management within monogamous relationships, the nuanced role and effect of payment dynamics in these shared experiences remains unclear.

Across a series of experimental studies, we investigate how different payment arrangements (split arrangements vs. consumer vs. other consumer pay the full amount) affect consumer happiness within shared experiences. We find that when consumers are with close social ties, they derive more happiness from paying themselves in full (vs. split or other consumers) for the shared experience. Conversely, in the absence of social connectedness (i.e., with a weak social ties) consumers are happier when they are able to conserve monetary resources (i.e., split or other consumer pays). These patterns are consistent regardless of the value of the experience admission, within ecologically valid parameters, for close social ties, but when the value of the admission increases consumers are even happier when conserving financial resources among weak social ties. Findings suggest that there is a complex relationship between payment dynamics and happiness in shared consumption experiences that is dependent on the social ties and contributes to a deeper understanding of consumer well-being in experiential contexts.

SB04
Meeting Room C3.3
Online Reviews: Biases
Contributed Session
Chair: Douglas Bowman, Emory University, Goizueta Business School, Atlanta, GA, 30322, United States
1 - Bias in Online Reviews: A Perspective from Social Distance
Shida Ni, Harbin Institute of Technology, Harbin, China, People's Republic of, Yukuan Xu, Zili Zhang
A series of studies has demonstrated that consumer online reviews are not invariably accurate and objective. Particularly, during the authoring of online reviews, numerous factors may contribute to biases. In this study, based on a large-scale dataset of restaurant reviews from an online reputation platform, we investigate a novel source of review bias, namely social distance. We first constructed a measure of social distance between consumers and restaurants based on consumers' past average restaurant expenditure (PARE) and the current restaurant's per capita spending (RPCS). Then, through empirical research, we examined the role of social distance from two opposing directions: when a consumer's PARE is higher than the restaurant's RPCS, they are more likely to post negative reviews, and vice versa for positive reviews. Furthermore, the expertise of consumers (reflected in their membership level status) significantly mitigates the rating skew caused by social distance; and when considering gender factors, it was discovered that female consumers are significantly less affected than male consumers. Our research enriches the application of social distance theory in the field of online reviews and provides important references for the algorithms used by online reputation platforms to aggregate and display business ratings.
2 - Truth or Dare: The Effects of Reviews with Videos and Disclosure of Incentives
Shiyu Liu, Lingnan University, Hong Kong, Hong Kong, Yuho Chung, Geng Cui, Ling Peng

The widespread use of sponsored content on social media is blurring the boundaries between paid advertisements and genuine consumer-generated word-of-mouth. Enforcing disclosure requirements for sponsored video content produced by influencers is particularly challenging because such videos often lack any indication of sponsorship or commercial nature. Previous studies on disclosure for sponsored content primarily focus on written text. Videos are multimodal, comprising text, speech, and moving images, and disclosures can appear in any of these modes or in combinations of these forms. Drawing on the literature on disclosure and multi-modal analysis of videos, this study examines the impact of sponsorship disclosure in marketing videos on consumer perceptions and ad performance, comparing visual and vocal disclosures. An empirical study using data collected from a video sharing platform suggests that multimodal disclosure has a more significant impact than unimodal disclosure, and visual disclosure is more effective than vocal disclosure.

3 - Review Solicitation: Intended and Unintended Revenue Consequences
Jinsoul Seo, Simon Business School, University of Rochester, Rochester, NY, United States, Minkyung Kim, Hana Choi
This paper examines how review solicitations affect individual review behaviors and ultimately e-commerce revenues, using a natural experiment and rich individual-level panel data from a leading South Korean e-commerce platform. We first document that review solicitations encourage more reviews, but decrease the number of photos and 5-star ratings, leading to uncertain revenue outcomes. Notably, solicitations more strongly influence new buyers and reduce 5-star ratings for expensive products. We further explore the underlying mechanisms via user and product characteristics, and find that a review solicitation increases the following week sales by 19,282 KRW ($14.6). Our results offer insights on how and when to send out review solicitations for e-commerce platforms.
4 - Suspicious Online Product Reviews
Douglas Bowman, Emory University, Atlanta, GA, United States, Eunhee (Emily) Ko

Concerns over the authenticity of reviews hinder their usefulness. The authors argue that perceived brand strength, brand advertising effort, price, and relative sales each act as signals that help positive reviews of a brand seem more reliable. They investigate this by studying Amazon reviews of branded products from 16 product categories that have the resources and potential desire to advertise. They examine consumer perceptions of review authenticity as perceived by machine learning algorithms trained on human subjects, as well as by direct perceptions of human subjects during validation. The results indicate that having a strong brand and investment in brand advertising, along with price and sales rank, are valuable to managers since they form a certain protection from suspiciousness.

SB05
Meeting Room C3.4
Qualitative Research: Supplier Perspectives
Contributed Session
Chair: Mr Sanaullah, University of Tasmania
1 - Enhancing Supply Chain Resilience through Supply Chain Finance
Zheng Tao, Renmin University of China, Beijing, China, People's Republic of, Hua Song

The growing interest in enhancing supply chain resilience has spurred much research. However, the pivotal role of supply chain finance as a catalyst for resilience is yet to be fully understood. This paper endeavors to bridge this knowledge gap by proposing an evolutionary framework that positions supply chain resilience as an emergent property of strategic supply chain financing, anchored by adept paradox management. Our examination of Xiaomi Finance as a case study reveals the multifaceted nature of supply chain resilience and underscores the need for a spectrum of financial strategies, each tailored to distinct scenarios and levels within the supply chain. Cultivating resilience through supply chain finance is contingent upon a robust paradox management capability, which empowers companies to navigate and reconcile conflicting but interdependent goals through a deliberate equilibrium of separation and synthesis. This paper further elucidates that such capability is not monolithic but a stratified construct, meticulously designed to address the diverse and nuanced resilience requirements across various dimensions of the supply chain.

2 - Innovation via Making, Sharing, and Listening: Overcoming Barriers to Market Learning in Peer-to-Peer Platforms
Gregory Fisher, Miami University, Oxford, Ohio, United States, Aric Rindfleisch

Many peer-to-peer sharing platforms (e.g., YouTube) allow individual contributors to create innovative digital offerings that can be shared with other users in exchange for social media recognition. Unfortunately, the ability of contributors to engage in traditional market learning strategies to craft appealing offerings is constrained by the mediating nature of the platforms, which typically limits the ability of contributors to communicate directly with users to understand their preferences. Our research examines how contributors can overcome this challenge. We focus on the 3D printing user community as our empirical context and conduct a multi-method inquiry. Our first study employs depth interviews with 15 contributors who create and share 3D printable product designs on the platform Thingiverse.com. Our second study combines a survey of 189 contributors to the MyMiniFactory.com platform, along with archival data on the performance of their digital product offerings. Our collective results suggest that contributors learn how to create successful 3D printable designs via a three-step process (i.e., Learning to Make, Learning to Share, Learning to Listen), and that this learning helps foster a communal orientation that enhances the success of their offerings in the form of views, likes, and downloads. In sum, our research identifies a seldom-recognized limitation of peer-to-peer digital sharing platforms and also illustrates how individual contributors can overcome this barrier via an emergent learning process to garner social media recognition for their innovations.

Keywords: user innovation, 3D printing, digital platforms, learning, product design, communal orientation, peer-to-peer sharing

3 - SME's Pathway to Resilience & Market Shaping - Exploring the Role of SME Resources, Resilience Pathways, and Market Shaping
Mr Sanaullah, University of Tasmania, Hobart, Australia, Rajesh Rajaguru, Gauri Laud

This study explores the resilience of small and medium enterprises (SMEs) during crises and their ability to shape the market. The study aims to investigate the processes and pathways that SMEs undergo to build resilience and shape the market.

The results of the study suggest that in the face of the daunting challenges posed by the COVID-19 crisis, SMEs exhibited exceptional resilience through strategic deployment of the integration pathway towards building resilience which eventually triggered market shaping. This transformative approach encompassed operational, external, and personal integration, acting as a powerful shield against the crisis. Within the realm of operational integration, businesses demonstrated remarkable financial acumen, with 35% fostering crucial connections with their network, and 45% engaged in efforts to encourage employees to embrace changes. External integration emerged as a linchpin in crisis resistance in which 70% of SMEs forged collaborative partnerships within their supply-network, 30% strategically integrated new customers, and 20% successfully secured suppliers’ support.

Resilient SMEs were found shaping market through two outcomes: changing the ‘customer and use’, and changing the ‘supply side network’. Changing customers and use were witnessed through SMEs' efforts of creating new customers, new markets, and altering the ways customers accessed their products using technology. Changing supply side networks were demonstrated through supply chain disintegration - bypassing traditional middlemen from the supply chain and giving access to new buyers or the users of product. Hence, the integration pathways helped SMEs to build resilience which triggered market-shaping outcomes.The study contributes to market-shaping theory.

Keywords: market-shaping, SME-resilience, Crisis-management

SB06
Meeting Room C3.6
Games and Gamification
Contributed Session
Chair: May Wang, BNU-HKBU UIC
1 - Self-stated Preferences, Peer Effects, and In-game Purchases
Tianqi Xue, The Chinese University of Hong Kong, Hong Kong, Hong Kong, Chenxi Liao, Ying Xie, Jingbo Wang
In-game virtual goods purchases account for 95% of all user spending in the fast-growing mobile game industry. Despite the rich literature on drivers of physical goods purchases, questions remain if and how those drivers may impact in-game virtual goods purchases. This is because compared to physical goods, virtual goods possess distinct properties in terms of their functionalities and consumption experience. Moreover, many in-game virtual communities foster unique environments that facilitate users’ voluntary preference revelation and promote social interactions among users. In this project, we aim to fill this gap by conducting an empirical analysis using a unique in-game purchase dataset from a social mobile game. We examine the following questions: (i) How do users’ self-stated preferences affect their in-game purchase decisions? (ii) How do peer effects influence in-game purchases, and more specifically, do peer effects vary depending on the type and the strength of the social tie? (iii) What is the interplay between self-stated preferences and peer effects on in-game purchasing? Preliminary analysis reveals that self-stated preferences play a significant role in driving in-game purchases and indicates an intriguing interaction effect between self-stated preferences and peer effects.
2 - The Power of Free: Consumer Choice in Freemium Market
Peng Liu, Santa Clara University, Santa Clara, CA, United States, Cheng Chou, Hai Che
Freemium business model has become popular in recent years. We develop a model of a consumer’s simultaneous choice of the usage of a freemium product and the purchase of its premium add-ons. We estimate the proposed model using the player-level panel data from a
freemium massively multiplayer online game. Our estimates quantify the economic value of free users in a freemium market. We design a counterfactual reward program, in which players will be rewarded with free items for playing the game. The analysis shows that the counterfactual reward program increases the revenue three times more than price promotion.
3 - Analyzing the Affective Embeddings and Mediated Self-expression in Genshin Impact Game: Examining the Impacts of Aesthetic Assemblages and Gambling Features Design
May Wang, BNU-HKBU UIC, Zhuhai, China, People's Republic of, Gaobiao Cao, Yiyan Li

This study explores the impacts of aesthetic assemblages and the gambling features design on self-expression, and the affective embeddings within the immersive gaming experience of playing Genshin Impact. Drawing on social presence theory and flow theory, we aim to gain insights into the mechanisms that contribute to player engagement, game addiction, as well as user experiences of enjoyment, and self-expression within the game's virtual environment. We examine how the presence of other players, both in real-time interactions and through asynchronous elements such as leaderboards and social media integration, enhances the sense of social presence and flow of immersion. Additionally, we explore how these social interactions and player experiences, e.g. enjoyment, and self-expression lead to more engagement with and addiction to Genshin Impact. Using real-world reviews from game player community website, web analytics approached was applied to identify correlation among the key concepts. We also design one laboratory experiment and one field experiment to test the underlying mechanism based on the two theories. We find that gambling features and social presence significantly influence player engagement and addiction, especially for those who with salient sensation-seeking personality. These findings provide novel insights into the ethical concerns of game feature design.

SB07
Meeting Room C4.1
Empirical Models for Competition
Contributed Session
Chair: Richard Grice, INSEAD, Fontainebleau, France
1 - Digital Empowerment: How Online Channels Redefine Bargaining Power for Traditional Manufacturers
Siyu Meng, Peking University, Beijing, China, People's Republic of, Ruobing Ling, Junhong Chu, Yi Xiang

The rising of e-commerce has opened new avenues for traditional manufacturers to reach a broader customer base. Beyond merely expanding their offline shops, traditional manufactures can strategically build their online channels to enhance overall firm performance. Given the significant role of online channels in complementing traditional ones, extensive research has been conducted on the influence of online channels on their offline counterparts. In this paper, we examine whether and how the introduction of online channel by the manufacture impacts the bargaining dynamics with offline intermediary in the traditional channel. By integrating online-offline interactions within a Nash bargaining framework, we model changes in offline sales and the outside options available to manufacturer in the event of negotiation breakdowns with offline intermediaries post-online channel introduction. Our analysis reveals that online channel sales can significantly influence manufacturers' bargaining outcomes, largely dependent on whether the online and offline channels are substitutes or complements. We then test the theoretical framework using data from a women’s apparel manufacture, who has only sold its products in offline department stores before the introduction of the online channel. Employing the leads-and-lags Difference-in-Difference (DID) model, our findings indicate that the introduction of online channel can reduce the commission rates manufacturers are required to pay to offline department stores. In conclusion, the introduction of online channel will increase the bargaining power of manufactures and thus achieve a lower commission rate in the offline channel.

2 - Incumbent's Deterrence Strategies and Potential Entrant's Time-to-Entry: Evidence from the U.S. Airline Industry
Sina Aghaie, Wichita State University, Wichita, KS, United States, Omid Kamran Disfani, Milad Darani, Mike Saljoughian

Previous research illustrates that incumbent firms commonly respond to competitors’ entry threats by cutting prices and/or expanding capacities. While antecedents of these deterrence strategies have been extensively investigated, the consequences have yet to be empirically examined. Drawing upon signaling theory, the authors investigate whether incumbents’ price-cuts and capacity expansions influence Potential Entrants (PEs) entry timing and examine the moderating role of incumbents’ Resources and Capabilities (R&Cs). To test the hypotheses, a parametric hazard model is applied to a rich, multi-market dataset of entry threats between 1997 and 2016 in the U.S. airline industry. The findings suggest that while capacity expansions delay PEs entry, price-cuts expedite it. Furthermore, PEs are more likely to be deterred by price-cuts when incumbents possess high R&Cs. This research advances market entry and competitive strategy literatures, highlights the crucial role of resources in implementing deterrence strategies, and provides insights to managers to better protect their markets.

3 - Planes, Trains, and Co-Opetition: Evidence from China
Shaojun Qin, Temple University, Philadelphia, PA, United States, Maria Ana Vitorino, George John
This paper analyzes the impact of the high-speed train (HST) roll-out on air carriers' route network decisions in China. Using unique hand-collected data on the timing of the HST roll-out and airline networks, we estimate a dynamic oligopoly model of air-route entry and exit that accounts for the competition between airlines and the (positive and negative) spillovers from the HST.
We use the model to assess the impact of the HST on airlines' decisions by simulating a scenario without HST. We find that, despite significant positive spillovers from the HST on airlines, derived from intermodal connectivity, the HST reduced airlines' route presence by about 14\%, and airline profits by 23\%. We find considerable heterogeneity across route-types and regions. Airlines readjusted their networks substituting towards longer routes and more peripheral regions in China thus improving connectivity among regions. We also use the model to explore the benefits of improving intermodal transportation.
4 - Learning Competitors' Identities from the Timing of Pricing Decisions: An Application to Retail Gasoline
Richard Grice, INSEAD, Fontainebleau, France

It is becoming clear that it is not just consumers who limit their consideration when faced with broad product assortments. Managers also restrict their attention to narrow subsets of potential competitors. How then can we identify who it is that managers actually treat as their competitors? I propose an approach exploiting information on the identities of competitors encoded in the timing of price responses. The advantage of this approach is that, to the extent that managers do restrict their attention when setting prices, it identifies who managers treat as competitors, not who consumers consider substitutes.

Because economic theory provides little guidance for modelling the timing of strategic decisions, I found the method in the decision problem of a pricing manager within the ABBE continuous-time retail competition model (Arcidiacono et al. 2016. Rev. Econ. Stud.). I derive from this structural foundation a ‘reduced-form’ expression characterising the hazard-rate of a manager's price-change decisions as a function of their competitors' prices. Estimating this expression with l1–norm regularisation exploits the consistent model selection properties of the LASSO to identify true competitors from potential competitors. Further, it makes implementation a straight-forward application of widely-used machine learning packages, such that the method can be implemented by practitioners as readily as researchers.

I demonstrate the method with an application to high-resolution price data from gasoline stations in Australia. The application reveals that stations from Sydney's suburbs to its surrounding regions all compete and influence each other within a sparsely-connected network structure, rather than distinct geographic markets.

SB08
Meeting Room C4.2
NFTs 2
Contributed Session
Chair: Yujing Chen, University of Surrey, GU1 4AJ
1 - Consumer Behavior Toward Luxury NFTs: A Perspective of Signaling Theory
Yingjiao Xu, North Carolina State University, Raleigh, NC, United States, Hanna Lee, Wenna Han, Xiaohan Lin
Luxury fashion brands are increasingly recognizing the potential of NFTs in enhancing brand engagement by offering exclusive digital collectibles. While the NFT market provides a significant opportunity for an untapped revenue source, a unique challenge for the luxury industry is to successfully facilitate consumers’ perception of the luxury value from digital collectibles. Guided by the Signaling Theory, this study investigates the effects of signaling cues (product prototypicality, visual clarity, and brand prominence) in consumer responses toward luxury NFTs in terms of willingness to buy and willingness to pay price premiums. A 2x2x2 mixed-design experiment was conducted online with 448 US participants. A three-way ANCOVA was employed for data analysis, with consumers’ brand familiarity controlled. Results suggest that while all three signaling cues were significant in enhancing consumers’ willingness to buy luxury NFTs, only visual clarity was significant for willingness to pay price premiums. Additionally, a significant three-way interaction was found, where the positive interaction effect of product prototypicality and visual clarity was stronger when the brand prominence was high. That is, luxury NFTs with a high level of product prototypicality and visual clarity, combined with strong brand prominence, generated the highest level of willingness to buy and willingness to pay price premiums from consumers. From the perspective of quality and value signaling, the results of this study provide practical insights and implications to luxury fashion brands in their endeavors in tapping into the emerging NFT market, which could lead to new revenues and stronger positions in the digital luxury market.
2 - Private Enjoyment or Profit Pursuit? Understanding Skin-Tone and Gender Price Disparities in CryptoPunks NFTs
Sungeun Han, Arizona State University, Scottsdale, AZ, United States, Zhan Shi, Tongxin Zhou

The rise of the Non-Fungible Tokens (NFTs) and avatar collections have gained increasing attention due to their roles in facilitating unique and personalized identity expression in digital spaces. However, notable price differences among various avatar skin tones and gender groups have raised concerns about diversity, equity, and inclusion (DEI) within this market. This study delves into the mechanisms behind price disparities in the market, focusing on how market participants’ motivations vary based on skin tones and gender. Grounded in auction theory, we develop a structural model to decompose participants’ willingness to pay into two components: common value and private value, where the former is linked with financial benefit motivation and the latter is associated with personal satisfaction. We estimate our model using data collected from Cryptopunks, a pioneer digital avatar collection platform. Our findings show that the weight on the private value component is the highest for the Lighter (i.e., Albino, Light) Male group, followed by the Lighter Female group, Darker (i.e., Mid, Dark) Male group, and Darker Female group. These results indicate that people who buy the Lighter skin tone group are more driven by personal satisfaction whereas the Darker skin tone group is driven by financial benefit and a similar stream is associated with the male group and the female group, respectively. These findings potentially highlight an existing bias in how different avatars are valued, emphasizing the need for greater inclusivity and equitable representation in the NFT space.

3 - Decoding Digital Assets: The Impact of NFTs and FTs on Customer Engagement through Ownership in Decentralised Community
Yujing Chen, University of Surrey, Surrey, United Kingdom

Building on signaling theory, this research explores the impact of digital assets, specifically Non-Fungible Tokens (NFTs) and Fungible Tokens (FTs), on trust and customer engagement in decentralised communities, focusing on the mediating role of perceived ownership. Through experimental studies conducted in the UK, the study examines significant differences in trust and perceived ownership between communities utilising NFTs and FTs. Findings reveal that NFTs enhance trust and customer engagement more effectively than FTs, primarily due to their unique characteristics that foster a higher sense of ownership. The study also highlights a unique scenario where offering both NFTs and FTs paradoxically leads to reduced sense of trust and ownership, potentially due to choice overload. This research contributes to understanding the strategic use of digital assets in marketing within decentralised platforms and highlights the importance of managing digital asset portfolios to optimise consumer trust and engagement.

SB09
Meeting Room C4.3
Consumer Behavior: Online Behaviors
Contributed Session
Chair: Jingyi Zhang, Lingnan University, Lingnan, N/A, Hong Kong
1 - Visual Influences on Quantity Perception on Online Grocery Retailing
Wanyi Zheng, Hong Kong Baptist University, Hong Kong, Hong Kong, Michael He Jia, Echo Wen Wan

Online retailing platforms are adopting various practices to present quantities of grocery items on the webpages where consumers make quantity-based purchase decisions. The current research identifies a solution based on visual aids—supplementing the numerical information with a corresponding number of visual icons—to promote product packs containing more individual units. Thirteen preregistered, incentive-compatible studies (N = 6,388) demonstrate that our proposed solution of visual aids of quantity presentation shifts consumers’ preferences from smaller-quantity packs toward larger-quantity packs, compared to the numerical presentation of quantity. This effect is robust across product categories and visual icons and occurs because visual aids of quantity presentation increase the perceived quantitative difference between two packs, consequently enhancing the relative attractiveness of larger-quantity packs over smaller-quantity packs. Furthermore, the visual aid effect is attenuated when visual icons are less discriminable from each other or require more effort to process or when people avoid bulk buying. This research extends the literature by demonstrating that visual aids can even influence perceptions of easy-to-assess small quantities and generates novel managerial insights into the optimization of quantity presentation practice in online grocery shopping.

Keywords: visual aids, quantity presentation, perceived quantitative difference, visual processing, preference

2 - The Bright Side of Low Consumer Trust in Online Marketplaces: A Motivated Belief Perspective
Yu (Anna) Lin, The Chinese University of Hong Kong, Hong Kong, Hong Kong, Xianchi Dai, Wenjie Tang

Platforms invest substantial resources and effort to improve consumers’ trust. Whereas past studies have demonstrated various positive impacts of trust building on platforms, little is known about the potential positive impacts of low trust.

In this research, we propose a positive impact of low trust for e-commerce platforms. Specifically, consumers with relatively low trust in a platform tend to choose products offered at higher prices, which in turn may potentially increase the platform’s revenue. Moreover, we theorize that this effect is driven by a motivated belief about the positive price-quality relationship. Specifically, consumers with lower trust are motivated to develop a stronger belief about the positive relationship between price and quality; guided by this belief, they then deliberately choose higher-priced offerings of a product to avoid the pitfalls of inferior products.

Study 1 examined the effect of consumers’ trust in the platform on their price choice using a unique dataset obtained from a real e-commerce platform. Studies 2 and 3 further examined this effect with a controlled choice set, where we directly measured participants’ trust in a real-world platform or manipulated their trust in a hypothetical platform. Finally, Studies 4 and 5 examined the proposed mechanism by measuring or manipulating consumers’ motivated belief about price-quality relationship.

Revealing a positive impact of low trust, our findings provide implications regarding platforms’ optimal investment to battle inferior products and build trust. Additionally, our findings offer platforms new chances to increase profits through segmentation and targeting strategies based on consumers’ trust in them.

3 - Impact of Avatar Realism on the Inferences of Manipulative Intent
Yuetong Guo, Queen Mary University of London, London, United Kingdom, Jie Sheng, Vignesh Yoganathan

This study explores how the realism of avatars affects consumers’ social perceptions and the effectiveness of persuasion episodes when avatars serve as sales agents in live commerce.

Two types of avatar realism – visual realism (human-like vs cartoon-like appearance) and verbal realism (human vs synthetic voice) – are discussed in this paper. Drawing on the Social Cognition Theory and the Robotic Social Attributes Scale (RoSAS), this research proposes that avatar realism affects consumers’ attachment of social attributes to avatars. Specifically, visual and verbal realism of avatars increase perceived warmth and competence and reduce perceived discomfort; meanwhile, the inconsistency in the two forms of realism creates discomfort and barriers for consumers to recognise avatars as being warm and competent. Further, building on the Persuasion Knowledge Model and Consumers' Processing of Persuasive Advertisements framework, this study examines how the social cognition of avatars affects consumers’ inference of manipulative intent when consumers use heuristics and experiential processing strategies. A series of experiments are conducted to test the main effects of realism in avatars’ visual and verbal design on the inference of manipulative intent, the factorial effects between visual and verbal realism, and the mediating effects of social perceptions.

This research contributes to scholarly discussions on avatar anthropomorphism by examining the impact of avatar realism on consumers via a lens of social cognition and persuasion knowledge. Results from the study are expected to support marketers in designing and exploiting avatars in the sales process.

Keywords: Avatar design; Social cognition; Manipulative intent; Live commerce
4 - Catching the Unlikely Gambler: How and Why Gacha Games Appeal to High Conscientious Consumers
Jingyi Zhang, Lingnan University, Kowloon, Hong Kong, John Yi, Dongjin He, C. Clark Cao

Gacha games, often mobile or browser titles featuring Japanese anime-style characters and designs, are characterized by its eponymous gacha mechanism, wherein players obtain random virtual items using in-game currency purchasable with real-world money. While scoring significant market success, gacha games are frequently likened to gambling by critics due to their chance-based purchase system. Despite this gambling-like feature, here we identify a paradoxical attraction of gacha games for individuals high in conscientiousness, a trait typically linked to self-discipline and gambling aversion. Through five behavioral experiments involving actual gacha game players, we consistently found that higher conscientiousness correlates with increased gacha game engagement, as captured by the acquisition of in-game items, spending, and intentions for future gacha pulls. We further discovered that this is because gacha pulls are typically viewed as avenues for achievement rather than a form gambling, hereby appealing to conscientious consumers. Our findings also rule out the alternative explanation that conscientious players are drawn to the order and structure embodied in in-game items. Taken together, this research reveals an unexpected personality trait predicting gacha game engagement, challenging conventional views of conscientiousness as a deterrent to indulgence and overspending. In so doing, we enhance the field’s understanding of consumer perception and behavior in this unique game genre. Additionally, this study helps identify populations vulnerable to gacha games, offering insights for policymaking.

SB10
Meeting Room C4.4
Sales Promotion 2
Contributed Session
Chair: Nils Wloemert, Vienna University of Economics & Business, Vienna, Austria
1 - Price Promotional Models for EDLP Retailers
David Lie, Monash Business School Monash University, Caulfield East, Australia, Ashish Sinha, Rahul Govind, Sonika Singh

The research aims to develop and estimate price promotion modelling for Every Day Low Pricing (EDLP) retailer. The main challenge of estimating a model for EDLP retailers is the relatively stable and constant pricing, which may hinder model convergence. We propose two-stage method: (1) demand-group-market-share approach to form cross-sectional price variation, and (2) estimating price elasticity model for primary and secondary demand for EDLP chain. We applied the above method to the soup dataset from Walmart, one of the largest US-based EDLP retailer. Initial results identify three demand groups, and price elasticities for these demand groups are estimated. After we have successfully estimate price elasticity, we can then estimate the change in both category and SKU sales. This allows us to optimize the SKU and category sales by changing price. In the Walmart application, early results show that the optimization exercise manages to significantly increase the company’s revenue.

2 - Flying High, Landing Low? Effect of Tensile Price Promotions on Store Traffic and Sales
Han Chen, Monash University, Clayton, Australia, Joydeep Srivastava

This research examines the effectiveness of three types of tensile discounts in advertising storewide promotions: a maximum discount (i.e., up to Y% off), a range of discounts (i.e., X% to Y% off), and a minimum discount (i.e., starting at X% off) on consumers’ initial perceptions and expectations, store visits, and purchase. Using multiple methods, including empirical analysis of data from a large-scale storewide promotion campaign, controlled experiments, a quasi-field experiment, and numerical simulations, the findings suggest that while a maximum tensile discount is most effective in shaping consumers’ initial perceptions and expectations, and thus store visits, it does not necessarily boost purchases in some conditions. Our findings are consistent with the reasoning that because a maximum tensile discount leads to the most favorable discount expectations, the likelihood and the extent of consumers experiencing negative expectation disconfirmation (i.e., realized discount lower than the expected discount) is highest with a maximum tensile discount, thus discouraging purchases and anticipated satisfaction. Together, our results highlight the double-edged nature of using a maximum discount to advertise storewide promotions. There is thus a need to balance between raising expectations of discount and the likelihood of consumers experiencing negative expectation disconfirmation when deciding the optimal tensile discount to maximize store sales.

3 - A Study of Heterogeneous Causal Effect of Price Promotion with a Causal Machine Learning Approach
Jakir Talukder, University of South Austra, Adelaide, Australia, lin liu, Giang Trinh, Jiuyong Li, Anh Vo

The capability to identify buyer groups on which price promotion has greater impacts for targeted marketing, is of utmost importance for businesses. However, accurately estimating the impact of price promotion is challenging due to heterogeneous consumer and brand characteristics. Traditional predictive techniques fall short of predicting changes in customer responses due to the change of promotion status, especially with observational data. We argue that the most effective predictive method for targeting customers using observational data is causal prediction using Conditional Average Treatment Effect (CATE) estimators. In this research, we utilise recently developed CATE estimators of machine learning community, particularly the causal tree-based models, to predict the heterogeneous causal effect of price promotion on customers’ purchases. The findings demonstrate causal tree-based models, in comparison to traditional predictive models and uplift models, can provide more accurate estimates of CATE. Moreover, the research finds that different customer groups respond differently to the price promotion of different brands. By using the predicted CATE by a causal tree, firms can effectively target the buyers to ensure the effectiveness of promotions. The research contributes to the theory of price promotion heterogeneity and enhance the practical application of causal machine learning models in this context.

4 - Privacy Regulations and Advertising in Offline Markets - Evidence from Randomized Field Experiments
Nils Wloemert, Vienna University of Economics & Business, Vienna, Austria, Alexandra Becker, Dominik Papies, Christian Hotz-Behofsits
Growing awareness of the value of privacy has led to the implementation of more stringent privacy regulations and firm policies aiming to protect consumers through greater transparency and more control over their data. In the context of multi-vendor loyalty programs (MVLPs), such regulations can endow consumers with the choice of whether to allow the firm to use their personal data for advertising and other purposes, i.e., consumers have to actively opt-in or -out to receiving personalized targeted advertising. We investigate the consequences of such an opt-out decision for consumers’ long-term shopping behavior in brick-and-mortar retailing. To this end, we utilize two large-scale field experiments in which a random sample of customers of a sizeable MVLP in a European country did not receive targeted promotions for a period of ten months, either from none of the MVLP companies, or from only one specific partner. During this period, customers still received generic point redemption rewards. With these experiments we are able to identify both the general effect of discontinuing personalized promotions on consumers’ long-term behavior and possible spillover and cannibalization effects within the MVLP. We find that not offering personalized promotions generally has a more negative effect on overall expenditures of customers in drug stores than in supermarkets with a wider product range and a higher focus on fast perishable goods. Furthermore, discounters benefit from a complete discontinuation of personalized promotions across all MVLP members. Ongoing research using quasi-experimental data from the introduction of a tiered MVLP membership will complement these findings.
SB11
Meeting Room C4.5
Retailing: Category/Brand Choice 2
Contributed Session
Chair: Baohuan Zhou, University of Science and Technology of China
1 - Subcategory Showdown: Mass or Target Marketing?
Alicia Barker, Ehrenberg-Bass Institute, University of South Australia, Adelaide, Australia, Steven Dunn, Kirsten Victory, Magda Nenycz-Thiel

Subcategory growth is of increasing interest to companies as a means to grow revenue sales. However, minimal research has examined how subcategories compete for buyers. This is important to understand given it provides direction for brands taking a larger role in marketing their subcategory. Without it, companies are left questioning if they should implement mass marketing (penetration) or targeting (loyalty) strategies for subcategory growth.

We address this issue by examining the extent to which purchasing of subcategories follows a Double Jeopardy (DJ) pattern. DJ is commonly observed for brand buying; and summarises how smaller brands have far fewer, slightly less loyal, buyers than larger brands. DJ empirically supports a mass marketing brand strategy.

Using consumer survey data, we benchmark penetration and loyalty for 129 subcategories (e.g., sparkling wine), spanning ten macro-categories (e.g., alcohol) and seven countries. Our results show penetration is the primary difference between higher and lower selling subcategories. Loyalty hardly varies, and does so predictably in-line with penetration.

We also find the NBD-Dirichlet model of marketing generalises to subcategories. Niche/change-of-pace loyalty deviations are also both uncommon (~35% of subcategories), and usually explainable by marked functional differentiation (e.g., coffee pods vs. energy drinks), subscription markets, or subcategories with very small shares.

These results suggest companies seeking to grow their brands through subcategory growth should use resources on mass marketing strategies. Expecting growth by encouraging repeat-loyalty among existing buyers is typically out of step with market structure.

2 - Reexamining Double Jeopardy in the Unexplored Restaurant Category
Yeohong Yoon, Emory University, Atlanta, GA, United States
This research thoroughly explores the double jeopardy (DJ) phenomenon within restaurant service categories, an area previously unexplored in the literature. Leveraging unique credit card panel data, the study uncovers a consistent and notable manifestation of DJ across various restaurant categories, including burgers, chicken, pizza, sandwiches, seafood, and steak. However, categories consumed more frequently in daily life tend to exhibit more significant deviations from predicted model. On the other hand, the results indicate that large brands exhibit superior performance (better penetration rates and purchases per buyer) in sparsely populated regions compared to densely populated urban areas possibly owing to the unavailability of other competitors in these areas. Nevertheless, the study intriguingly demonstrates that the diminished market dominance of large brands in urban areas does not necessarily lead to increased market performance for their mid-sized competitors. The author concludes by emphasizing the significance of physical availability, alongside accessibility and affordability, in shaping the double jeopardy phenomenon within the restaurant service category.
3 - How to Spot a Winning New Consumer Product at Launch
Kirsten Victory, Ehrenberg-Bass Institute, Adelaide, Australia, Magda Nenycz-Thiel, Arry Tanusondjaja, John Dawes, Jenni Romaniuk

Thousands of new products are introduced, but success is not guaranteed. Many new products do not survive beyond their first year. Line extensions (LEs) are common as they leverage existing parent brand assets and buyers. Despite the advantages, failed LEs might damage the parent. This makes it vital to answer: Is it possible to identify (likely) failures early on?

This study investigates the launch-year performance of ‘successful’ and ‘failed’ LEs. Kilts Centre for Marketing sales data is used to identify 7,196 successful LEs (i.e., survived after three years) and 5,294 failed LEs (i.e., reached first-year but failed soon after). Market share has a relationship with survival length, but the role of trial and loyalty are unknown.

We show most failures ‘fail’ to acquire category buyers. Only 10% of the failures have a penetration above the category norm at the end of the launch-year, compared to the half of successful LEs. There is little variation in repeat buyer rate for successful and failed LEs.

This research challenges the importance of loyalty because gaining trial has a much larger role in survival. The marketing discipline has to play a larger role in advancing new product knowledge because the empirical observations about how brands compete apply to LEs.

This research providers marketers an approach to identify failures early on, and either remedy the situation or make a decision to discontinue the LE. This research encourages marketers to stop solely relying on the parent and instead support the LEs with activities to increase trial.

4 - The Impact of Collaboration Network Characteristics on High-level Scholars' Research Performance in the Field of Quantum Information Technology
Baohuan Zhou, University of Science and Technology of China, Hefei, China, People's Republic of, Liang Liang, Yundong Xie, Yanyan Zhu

As a critical component of modern physics, quantum information technology represents a field with discernible thresholds. Evaluating the high-level scholars' research performance in this domain has emerged as a pivotal research focus. Unlike previous studies that examined the impact of collaboration networks on scholars' research performance from a single network perspective, this article explores how the characteristics of scholars' egocentric networks and team networks, within a two-tiered collaboration networks framework, influence the quantity and quality of research performance among high-level scholars. Our research methodology encompasses bibliometrics, social network analysis, and the machine learning infomap algorithm. Initially, we crawled 202,614 papers between 2000 and 2020 in the Web of Science Core Collection SCIE database as primary examples. Incorporating name disambiguation, 6,587 high-level scholars were gathered for collaborative network analysis. In the empirical analysis, we employed both linear regression models, and negative binomial regression models with fixed effects. To ensure the reliability of our results, we conducted a series of robustness tests. In terms of scholars' egocentric networks, occupying an intermediary or structural hole location positively affects scholars' research performance. From the team networks perspective, there is an inverted U-shaped relationship between team size and the quality of high-level scholars' research performance, with the tightness of the teams in which high-level scholars are embedded contributing to the production of higher-quality research outputs. This new exploration of influencing factors relevant to scholars' collaboration networks and their research performance in this empirical analysis, consolidating and broadening the findings of previous studies.

SB13
Meeting Room C4.7
Sustainability: Insights from Conjoint/Choice Models
Contributed Session
Chair: Jungeun Lim, University of Wisconsin-Madison
1 - Sustainable New Product Introduction and Brand Performance
The Hoa Mai, Monash University, Melbourne, Australia, Satheesh Seenivasan, David Sugianto Lie

Introducing sustainable new products enables firms to cater to the ever-increasing consumer demand for sustainable products, while also addressing the increasing market expectation for firms to be environmentally responsible. However, high level of investment required for product innovation and high risk of failure act as barriers to investing in sustainable new product development. Prior literature has focused mainly on determining drivers and barriers to sustainable product consumption while ignoring the effect of sustainable new products on brand performance. This research is conducted to address this gap. Building on signaling theory, we address three important research questions: (1) How do sustainable new products affect their own brand and rival brands' performance? (2) How do sustainable new products affect their own brand’s brand equity? (3) Under what circumstances the effects of sustainable new products are strengthened or weakened? Using a market share attraction model, we estimate the effects of sustainable new product introductions by 14 leading detergent brands in 10 largest US retail chains over a five-year period (2015-2019) on the market shares of own and competitive brands. We also analyze the roles of brand, category, and consumer characteristics in moderating the effects of sustainable new products. Our results help answer the question of whether sustainable new products could signal a brand’s environmental commitment, and thereby increase its brand equity. Managerially, our research answers the question of "why and under what circumstances sustainable new products offers the most benefit for a brand”.

Key words: Sustainability, Brand Equity, Signaling, Competition, Choice Modeling, Econometrics

2 - Consumers’s Collective Influence On Pension Investment: ESG Needs From Pension Participants and Beneficiaries
Yadi Yang, Erasmus University Rotterdam, Rotterdam, Netherlands, Thomas Martens, Onno Steenbeek

To address important Environmental, Social, and Governance (ESG) issues that matters to consumer in pension filed (e.g., pension participants and beneficiaries), the traditional way of maximizing the financial returns should be replaced by striving for long term social gains. However, due to the traditional way of pension fund management, most if not all pension participants and beneficiaries have no say in how their pension would be invested. Dutch pension system is experiencing reforms in recent years and thus it is a good opportunities for pension funds manager and policymakers to explore the needs and the possibilities for ESG investment.

This paper studies how consumers' attitudes and needs for ESG conscious investment are and how will their attitudes and needs, together with risk preference, influence the decision-making process of pension fund managers.

We use quantitative methods to explore the attitudes and needs of ESG investment of pension participants and beneficiaries. Choice-based conjoint analysis has been applied to quantify the trade-off between ESG investment and pecuniary return of ESG investment. The study sample consists of 433 Dutch pension fund participants.

We found that the majority want their pension to be invested in a ESG-conscious way, as long as it does not cost too much. For a hypothetical pension income €2,200, respondents would prefer ESG investment if it does not cost more than €232.56/month. Among the three factors, environment issues gain most focus comparing to social or governance issues. Moreover, risk preference is not correlated with participants’ ESG preference.

3 - Consumer Preferences and Circular Economy: The Case of Reusable Packages
Hossein Eslami, Lebanese American University, Beirut, Lebanon

Circular economy (CE) paradigm has gained much attention recently. Implementing a CE in the Fast Moving Consumer Goods (FMCG) market requires the buy-in from consumers who are the key players on the demand side buying environmentally friendly products and as well as in the reverse logistics returning used products to the manufacturers. Our first objective is to understand consumers’ preferences in the context of reusable packages. We specifically focus on the case of beverages such as soft drinks. To this end, we use an online survey together with a Conjoint choice-based experiment to measure consumers’ preference and willingness to pay for the drinks sold in reusable packages. Moreover, we investigate the heterogeneity in preferences and contrast the self-reported and actual behavior of consumers. The second objective is to measure the effectiveness of different incentive schemes for post-consumer package return. According to our results, while the price of product is a major factor affecting the choice of consumers, they prefer reusable packages over the disposable ones. Based on this, we could analyze the delicate trade-offs that manufacturers are facing in introducing reusable packages. The insights from our study would have implications for the manufacturers and policy makers in moving to the direction of circular economy in the FMCG market.

4 - Exploring the Sustainability Intention-Action Gap in Consumer Packaged Goods via the Utility Maximization Framework
Jungeun Lim, University of Wisconsin-Madison, Madison, WI, United States, Qing Liu, Neeraj Arora

While consumers have shown increasing interests in sustainability and higher willingness to pay for sustainable products, their intentions for sustainability may not lead to the purchase behavior of sustainable products due to perceptions of lower quality associated with sustainable products. This “intention-action gap” is a frustrating problem faced by companies engaging in sustainable initiatives. In this research, we develop a theory-based understanding of the underlying mechanism of such intention-action gap. Drawing on the utility maximization theory of consumer choice and the schema congruity theory, we identify consumers’ perceived incongruence of sustainable products (i.e., deviation from the product category norm) as the core barrier that contributes to the intention-action gap, and test two different ways to reduce the gap by resolving the perceived incongruence. Results from three studies, including hierarchical Bayesian estimates of consumer preferences based on data from conjoint choice experiments, support our hypotheses that resolving the perceived incongruence increases consumer utility of sustainable products and consequently reduces the intention-action gap. We also find eco-labeling to be more effective than equal-quality messaging in increasing consumer utility and willingness to pay for sustainable products. Such effects are more prominent for unknown brands than known brands. Our findings have important managerial implications, providing firms with appropriate guidelines on branding and pricing to accelerate consumer adoption of sustainable products.

SB14
Meeting Room C4.8
Crisis Management
Contributed Session
Chair: Xinyu Chen, Jiangxi University of Finance and Economics
1 - Negative News Spillover and Rivals' Strategic Reactions
Guiyang Xiong, Syracuse University, Syracuse, New York, United States, Xiaobo Lin, Shuai Yang
Conventional wisdom suggests that a company’s negative news is likely to benefit its competitors. In contrast, this research reveals significant heterogeneity in spillover effects across rival firms. By analyzing a large-scale dataset compiled from various sources, the authors (a) compare the spillover effects of various types of negative news, (b) document the heterogenous spillover effects across rival firms and identify the antecedents to such heterogeneity, and (c) investigate rival firms’ strategic reactions in response to the negative news. Results indicate that upon the release of negative news about a firm, the abnormal stock return of an industry rival depends on its strengths in R&D, marketing, and brand equity. In turn, the spillover effects in stock returns can motivate rival firms to make changes in their future product portfolios, brand portfolios, and recruitment emphases. Moreover, there are both firm-level and industry-level moderators that mitigate or amplify such effects. The findings provide novel implications for firms, investors, and policy makers.
2 - The Dynamics of Corporate Misconduct and Online Employer Review Ratings
Andreas Hamann, University of Mannheim, Mannheim, Germany, Lars Gemmer, Andreas Bayerl

Disclosed corporate misconduct can cause severe damage to a company’s reputation among its shareholders and customers. However, little research has investigated how misconduct disclosures relate to the firm’s employees and their portrayal of the employer brand on online employer rating platforms. On the one hand, disclosed corporate misconduct could change how employees perceive their employer and subsequently share their opinions online on employer rating platforms. While a priori unclear, the outcome of this review behavior is crucial for companies as they need a strong employer brand to stand out in the war for talent and attract skilled workers. On the other hand, employees’ review behavior may simultaneously reveal underlying internal working conditions and predict future misconduct. Therefore, this study investigates how disclosed corporate misconduct and the volume, valence, and variance of online employer review ratings relate over time. The results from two panel regressions suggest that misconduct negatively affects review valence but positively influences review volume. Both review measures can, in turn, also predict misconduct, while the review variance does not. Our findings contribute to the research on employer branding, employees’ online reviewing behavior, and the antecedents and consequences of corporate misconduct. They can help managers anticipate and respond to misconduct to protect their employer brand.

Keywords: corporate misconduct, employer reviews, employer branding

3 - Data Breach: Devastating Disaster or New Opportunity?
Jihwan Moon, UNSW, Sydney, New South Wales, Australia, Jake An

Data breaches are ubiquitous and can cause detrimental impacts on businesses. In 2023, the global average cost of a data breach reached $4.45 million – an all–time high. Nonetheless, we observe that some breached firms’ market values do not decrease and sometimes even increase; moreover, after data breaches, firms often raise price. We study firms’ post-breach responses with pricing and data security investments in vertically differentiated markets. Our analysis shows that when high-quality firms suffer a data breach, they might aggressively invest in data security to make consumers less concerned about future breaches, allowing high-quality firms to attract consumers from low-quality firms and increase price. The increased sales help high-quality firms achieve greater profits than a pre-breach level even without passing on investment costs to consumers. These desirable outcomes are feasible when low-quality firms, although not breached yet, are also highly vulnerable to data breaches. However, when low-quality firms are breached, they can neither increase price nor obtain greater demand by investing in security improvement because of high-quality firms’ competitive reactions. Using longitudinal data in the Australian telecommunication market, we provide empirical analyses, consistent with our predictions on post-breach consumer switching and sales for breached and non-breached firms.

4 - The Impact of the Matching Effect Between Data Breach Types and Recovery Strategies on Consumer Forgiveness
Xinyu Chen, Jiangxi University of Finance and Economics, Nanchang City, China, People's Republic of

Some studies focus on how to remedy data breaches, however, little is known about which recovery strategies firms should use in response to different types of data breaches. We identify three different data breach types, namely external breaches, accidental internal breaches and malicious internal breaches. We rely upon signaling theory to develop a theoretical framework connecting data breach types with consumer forgiveness through consumer perception of vulnerability and firm benevolence. Using a database and three experiments, our results show that consumer attitudes towards firms are varied when they face different sources of data breaches. Firms should rely more on apology when external breaches occurred, while firms should prioritize the compensation for consumers’ losses when accidental internal breaches occurred. When malicious internal breaches occurred, we suggest that firms combine apology with compensation. Moreover, the interaction effect between data breach types and firms’ recovery strategies is mediated by consumer perception of vulnerability and firm benevolence. We also demonstrate that the sooner firms apologize the better. As for compensation, a reasonable approach is to wait for a while, but not too long. Our research provides specific suggestions on how and when firms should implement recovery strategies after data breaches.

SB15
Meeting Room C4.9
Customer Service
Contributed Session
Chair: Weizhen Zhou, School of Management, Huazhong University of Science and Technology
1 - Performance above Replacement: Assessing the Impact Differentials across Listing Agents in Real Estate Transactions
Rex Du, University of Texas at Austin, Austin, TX, United States, Sachin Sridhar
In the U.S., more than 5 million homes are sold each year, with a median sale price over $400K and median time on the market over 40 days. Nearly 90% of these transactions are facilitated by listing agents, who perform a multitude of tasks on behalf of the seller. In return, most sellers compensate their listing agent with a commission of 2.5% to 3% of the sale price. The question is: How much better (or worse) would a seller have fared in sale price and time on the market if they had replaced their agent with an “average” one? Answering such a “performance above replacement” counterfactual requires assessing the impact differentials across agents, which presents a causal inference challenge that has heretofore been unaddressed. Leveraging detailed transaction records of 1.2 million single family homes sold through 50,000 listing agents over 17 years in one metropolitan area, we illustrate how this measurement challenge may be tackled. We find a substantial gap in agent performance between the 75th and 25th percentiles: about 5% difference in sale price and 30% difference in time on the market. Such a large dispersion in listing agents’ impact on transaction outcomes highlights not only the amount at stake when choosing an agent, but also the need for identifying markers of high-performance agents, which we investigate by relating observable correlates of agent expertise and effort to transaction outcomes, resulting in a set of empirical findings that should help consumers make more educated decisions when choosing a listing agent.
2 - Deciphering the Impact of AI Chatbots on Customer Satisfaction: A Systematic Literature Review
Xiaolan Xia, Griffith University, Queensland, Australia, Wei Shao, Jiraporn Surachartkumtonk, Haroon Iqbal Maseeh

Within the purview of marketing, artificial intelligence (AI) is pivotal in facilitating customer services, and the automation of digital marketing initiatives (Haleem et al., 2022). The expanding integration of these technologies heralds a transition towards a more technologically oriented nexus between enterprises and consumers (Rathore, 2019). This scholarly inquiry systematically synthesizes and critiques the functions and the characteristics of AI chatbots and examines their impact on consumer satisfaction. Following the rigorous systematic literature review approach (e.g., Chopra et al., 2023; Swain et al., 2023), this study delineates the significance of AI chatbots in fostering consumer satisfaction. Employing the Stereotype Content Model, the investigation appraises the chatbots' technical and social competencies in shaping consumer perceptions of amiability and ability. The study underscores the proliferation of AI chatbots across diverse industries, highlighting their pivotal role in augmenting customer service. It uncovers extant gaps in research and propounds prospective explorations into assorted cultural settings and the trajectory of technological advancements. The review endeavours to furnish strategic guidance for both the academic sphere and industry stakeholders, advocating the advancement and promoting AI chatbots' development and practical use to improve consumer relations and satisfaction.

3 - Algorithm-Augmented and Multitasking Intensity: The Forces of Retrieval and Processing Fluency
Weizhen Zhou, School of Management, Huazhong University of Science and Technology, Wuhan, China, People's Republic of, Yuanyuan Zhou, Jun Yang

Multitasking in instant service emphasizes that agents switch between two or more consumers to increase productivity. However, relevant studies have pointed out that switch cost increases the in-service wait of consumers, which violates consumers’ expectations for a timely response. Given firms are exploiting artificial intelligence (AI) assistants to provide acceptable response strategies for agents and it is still unclear how multitasking intensity influences algorithm-augmented performance. Based on the theory of information fluency, we argue that, in multitasking, although multitasking intensity can improve response speed by increasing the positive relationship between algorithmic cues and retrieval fluency in task switching, it can also reduce performance by increasing the processing disfluency in integrating algorithmic recommendations. Because of the limitation of cognitive ability, there is a complex and non-linear relationship between information load and processing fluency, we infer that the algorithm-augmented performance is heterogeneous across multitasking intensity in an inverted U shape. We test this hypothesis by analyzing the proprietary data of online services of a domestic consumer finance company. The results confirm that algorithm-augmented performance will first rise with increasing multitasking intensity, then fall. These findings highlight that the exploitation of AI assistants in multitasking is more conducive to improving the efficiency of instant services than single tasks. But even though higher intensity multitasking increases the role of algorithmic cues in retrieval fluency, it also triggers processing disfluency and inhibits response speed when integrating algorithmic suggestions.

Keywords: Multitasking, Human–AI collaboration, instant messaging, waiting time


4 - A Study of the Effectiveness of Service Recovery in a Multi-actor Context
Lifei Bai, Southwestern University of Finance and Economics, Chengdu, China, People's Republic of, Tianshu Chu, Xiaorong Fu, Ziyan Huang
The involvement of multiple actors (consumers, service providers, and platforms) complicates the transaction and causes service recovery to face many challenges. The main challenge lies in divergent perceptions of failure causality and responsibility assignment among transactional actors. This leads to the disconfirmation of consumer expectations in service recovery, a significant factor in recovery satisfaction. Based on expectation disconfirmation theory and attribution theory, this study conceptualizes two types of cognitive disconfirmation that create expectation disconfirmation in service recovery—causal cognitive disconfirmation and responsibility cognitive disconfirmation—and investigates how they differ in terms of how recovery satisfaction is affected. The study utilizes the secondary complaint data (text and structured) from an appliance brand on an e-commerce platform. Using a machine-learning BERT model, the study captures the two types of cognitive disconfirmation and explores their interaction with recovery strategies on satisfaction. The results show that two types of cognitive disconfirmation interfere with the efficacy of recovery strategies: in economic recovery strategies, responsibility expectation cognitive reduces satisfaction, while in social recovery strategies, causal cognitive disconfirmation negatively affects satisfaction. Furthermore, the results of the two experiments confirmed these findings. This study is the first to propose a conceptualization of the two types of consumer cognitive disconfirmation in the multi-actor context and explore the mechanisms that influence the effectiveness of recovery strategies. It also provides theoretical support for the development of service recovery research.
SB16
Meeting Room C4.10
Consumer Behavior: Understanding Product Preferences
Contributed Session
Chair: Hyun Young Park, CEIBS (China Europe International Business School)
1 - The Impacts of Consumer Animosity on Attitudes: More Insights From A Perspective of Product Typicality
Ting-Hsiang Tseng, Feng Chia University, Taichung City, Taiwan

The literature on consumer animosity shows inconsistent conclusions regarding its effects on consumers’ attitudes and product judgments. By adopting the perspective of product typicality, this research aims to clarify consumers’ attitudes (both explicit and implicit) toward typical/atypical products from a hostile country. The research conducted two 2 (typical vs. atypical) X 2 (high animosity vs. low animosity) between-subject experiments in Taiwan. China, having both war and economic conflicts, and Korea, having only economic conflicts, were selected as the possible hostile countries of Taiwanese participants in the experiments. 512 and 345 college students of Taiwanese nationality participated in the two experiments, respectively. The results indicate that adverse moderating effects of consumer animosity on explicit attitudes prevail across product categories. Such negative influences can also appear in implicit product attitudes for the target country's typical products but not atypical ones. The study also revealed the positive impact of ethnic product typicality on consumers’ explicit and implicit attitudes toward products from economically hostile countries. However, such positive influences become invalid in explicit attitudes if the target country has war conflicts with the consumers’ home country. This research contributes to the field by exploring the multifaceted adverse effects of consumer animosity in explicit/implicit attitudes across product categories with different levels of ethnic product typicality and revealing the difference in countries with different types of conflicts.

Keywords: Consumer animosity, dual attitudes, ethnic product typicality

Author (Presenting Author): Ting-Hsiang Tseng (Professor, Feng Chia University, Taiwan)

2 - Opulent and Responsible?: Exploring the Shift to Sustainable New Product Development in Luxury Fashion
Ismail Erzurumlu, Neoma Business School, Paris, France, Melike Talay

In this study, we draw upon the luxury brand management literature and the cue diagnosticity theory in consumer behavior to understand the relationship between sustainable new product development and attitudes toward luxury fashion brands. First, we hypothesize whether a sustainably produced new product positively affects the consumer's attitude toward a luxury fashion brand. Second, we explore the theoretical reasoning for the relationship between sustainable new product development and attitude toward a brand by introducing the theoretical construct of the perceived fit between luxury and sustainable innovation. Third, we introduce a moderated mediation hypothesis where we examine to what extent the environmental consciousness of a luxury consumer provides a contingency factor for the mediation hypothesis. We use randomized experimental design research and collect data using student subjects and Prolific participants. Findings reveal that consumers demonstrate a more positive brand attitude toward a new luxury fashion product developed through sustainable practices. Furthermore, this relationship is mediated by the fit between sustainable innovation and luxury brand essence. Also, it shows that consumers' environmental consciousness positively moderates the mediated relationship between sustainable new products and brand attitudes in luxury fashion.

3 - Video Game Character Design: The Role of Diversity, Equity, and Inclusion
Xiaolan Chen, Bayes Business School, London, United Kingdom, Wanqing Zhang, Daniela Cristian

Video games have emerged as the leading form of entertainment and leisure across the globe, surpassing both the movie and the music industries. Despite game characters being a revenue generator for the video game industry and having the power to shape the players’ perception and behaviour, character diversity remains a major challenge. Numerous games lack representation of 'non-white' and 'non-Anglo' characters, not only casts a shadow on the potential for reinforcing gender and ethnic stereotypes, but also shapes prevailing standards of beauty.This research focuses on how DEI related factors (diversity, equity, inclusion) in the context of video games affect player’s preferences, purchase intentions, and engagement. Starting from game characters, the main source video games income, this work examines the effects of character appearance, personality, and nationality. Combining 30 qualitative interviews and a series of experiments (collective N=1940 participants), the results indicate that players generally show a preference for characters with idealised body types and personality, skin tones akin to their own, and origins from countries they have positive associations with. The current findings also indicate that player characteristics, such as international experience, and past video game experiences play a role towards players’ attitudes to diversity. For instance, whereas new players are conservative, experienced players adopt a more inclusive mindset regarding the characters’ skin colour and ethnicity. The current work provides managerial implications for better player segmentation and also suggests that exposing players to a variety of characters fosters greater engagement.

Keywords: Video Games, Character Design, DEI

4 - How Consumers with Hedonic (vs. Utilitarian) Purchase Motive Use Item-Price (vs. Price-Item) Presentation Order as a Mechanism to Justify Their Hedonic Purchase
Hyun Young Park, CEIBS (China Europe International Business School), Shanghai, China, People's Republic of, Jungsil Choi

The present research investigates the moderating role of hedonic and utilitarian purchase motives for the presentation order effect. Past research finds that presenting item first and price later (e.g., 70 items for $29) increases consumers’ purchase intention than presenting the information in the opposite order (e.g., $29 for 70 items). Yet, the effect was observed mostly in hedonic consumption contexts. To address this gap, this research examines whether the effect is applicable for hedonic purchases but not for utilitarian purchases, and why.

In seven experiments, we find that the “item-price” (vs. “price-item”) order increases hedonic purchases, but not utilitarian purchases. Because consumers often feel guilty about hedonic purchases, they engage in motivated information processing when item (benefit) information is presented first and price (cost) information is presented later to perceive greater value from their hedonic purchase, which helps reduce guilt anticipated from hedonic consumption. Two serial mediation analyses show that perceiving greater value reduces guilt. In contrast, presentation order does not affect utilitarian purchases that do not elicit guilt. When a price discount is offered, the order effect is reversed because actual savings from price discount justify hedonic purchases better than perceived savings resulting from motivated information processing.

This research contributes to extant literature by introducing a novel moderator for the presentation order effect and revealing the underlying mechanism. It provides practical guidelines regarding how to promote hedonic products: present item information before price information, unless a price discount is offered, in which case the price should be presented first.

SB17
Meeting Room C4.11
Mobile and Location-Based Marketing
Contributed Session
Chair: Siqi Pei, Shanghai University of Finance and Economics
1 - Enhancing Electronic Coupon Effectiveness by Unlocking the Power of Geolocation Data
Chia-Ying Chen, University of Sydney, Sydney, Australia, Qiang Lu, Jia Li
Electronic coupons (e-coupons) have seen widespread usage in recent years, particularly on mobile devices, enabling firms to leverage geolocation data in conjunction with coupon utilization. This study aims to comprehensively understand the effectiveness of e-coupons concerning user characteristics and proximity to the focal firm through harnessing geolocation data. By utilizing e-coupon usage data and consumer reviews from restaurants located in a major shopping center in China, we initially categorize users based on their frequently visited locations. Subsequently, we identify their primary visitation purposes (e.g., home, work, school) using a combination of supervised and unsupervised machine learning techniques applied to geolocation data. These findings offer valuable insights into the radius of e-coupon effectiveness concerning physical distance and various types of locations. For instance, does proximity to a consumer's home hold the same impact as proximity to their workplace? This research can assist businesses in improving their targeted marketing strategies, better attracting and catering to the diverse needs of various regions and user demographics, ultimately leading to a higher retention rate.
2 - "Please Attain a Free Coupon": Customer Retention Strategy for Hedonic Product in Mobile Commerce
Junbum Kwon, University of New South Wales, Sydney, Australia, Ka Wing Chan, Xincheng Ma, William Gu, Dongwon Lee, Huan Liu, Felix Septianto

With the rise of mobile e-commerce, it is easier and more feasible for m-commerce platforms to implement pop-up coupons to engage and retain existing customers. This study exploits field experiment data on more than 108,265 existing customers who are randomized to receive static, GIF or pop-up coupons. Results suggest that pop-up coupons have significantly higher attainment rate than static coupons and GIF coupons are as effective as static coupons. In addition, the significant positive effect of pop-up (vs. static) coupons on attainment rate is greater for hedonic product than for utilitarian product. Additional analyses find that these results are robust. Exploration of the mechanisms reveals that pop-up coupon had a positive effect because such coupon was perceived as a gift and the moderating role of product on coupon was driven due to the emotion of guilt. These findings offer useful implications for m-commerce platforms and e-retailers in retaining customers.

3 - Predicting Consumer Visits with Co-Location Networks: A Deep Heterogeneous Network Representation Learning Approach
Hanyu Zhang, Emory university, ATLANTA, Georgia, United States, Kunpeng Zhang, David Schweidel

Mobile location data has been used to deliver targeted advertising and predict future behavior. Yet, marketing research leveraging such data largely focuses on the behavior of consumers as independent entities, neglecting their role as part of “social” networks. In this research, we utilize mobile location data to construct a social network based on observed co-locations between two mobile devices, as well as the points of interest (POIs) at which the devices are observed. We empirically demonstrate the value of integrating these co-location relations into a dual-perspective predictive model at the brand-location level. This model is designed not only to forecast new places that users may visit but also to predict potential new users for POIs. We first formulate a heterogeneous information network that captures the structural and semantic relations among different types of entities (e.g., users and POIs). We then implement a deep network representation learning method to investigate relations among users and POIs. We show that incorporating the co-location relations and a heterogeneous information network not only enhances its predictive performance—exceeding benchmark models by over 60%—but also highlights the substantial benefit of this approach for less popular POIs. Our findings demonstrate the potential of mobile location data in constructing such heterogeneous networks, providing valuable support for marketing efforts, particularly for businesses associated with these less popular POIs.

4 - Mobile Payments and Urban Mobility: Assessing the Influence of Mobile Payments on Metro Ridership and Environmental Enhancement
Siqi Pei, Shanghai University of Finance and Economics, Shanghai, China, People's Republic of, Wenyue Xi, Hongshen Sun

How do mobile payments reshape consumer behavior and urban living beyond their acknowledged role in streamlining transactions? The rise of mobile payments and e-commerce expansion have transformed the marketing landscape. While their impact on business growth and consumer purchasing habits is well-recognized, the potential influence of mobile payments on public services largely remains an untapped area in marketing research. This study bridges this research gap by investigating the effects of mobile payment adoption within the metro systems, and exploring its implications on consumer behavior, urban lifestyle, and environmental quality. We gathered daily metro ridership data from 2017 to 2020 for 20 prominent Chinese cities. By capitalizing on diverse adoption timelines across these cities, we utilized a lead-and-lag Difference-in-Differences method to evaluate the impact of integrating mobile payments into metro systems. Our findings indicate that the incorporation of mobile payments significantly boosts daily metro ridership, highlighting its effectiveness as a marketing tool in promoting public transportation. Furthermore, the increased utilization of public transportation considerably contributes to the enhancement of urban environmental quality, emphasizing the importance of mobile payments in fostering sustainable consumer behavior patterns. Our research offers key insights into how payment innovations impact not just consumer behavior but also urban mobility, public services, and environmental well-being. These insights can guide marketers and urban planners in leveraging mobile payment technology to enhance public service delivery and promote sustainable urban lifestyles.