2023 Marketing Symposium
Frontiers of Marketing Science

Information

Time

July 25, 2023, 13:00~

Place

Tama Campus, Chuo University
中央大学多摩キャンパス2号館4階研究所会議室2

Access

https://www.chuo-u.ac.jp/access/tama/

Campus Map
https://www.chuo-u.ac.jp/uploads/2023/06/access_tama_02.pdf?1688357972075

多摩モノレール駅から2号館までは徒歩5～10分程度です．

キャンパス内は複雑ですので，守衛所で経路等をお尋ね下さい．

Host Organization

Institute of Business Research, Chuo University, Tokyo

Organizer

Hiroshi Kumakura (Chuo University)
Masakazu Ishihara (New York University)
Takashi Teramoto (Chuo University)
Kyung Tae Lee (Chuo University)

Program

13:00～13:15

Hiroshi Kumakura (Chuo University)

Title

Introduction: Frontiers of Marketing Science

Slides (Presenters only)

Abstract

In the age of big data and IT, marketing science has greatly advanced and worthily uncovered hidden and complicated human behavior with quantative methods. On the other hand, marketing science faces several key challenges. First, other academic fields claim some topics that could be viewed as marketing topics. That is, other disciplines are flooding into marketing with their own methos/ways, and they are solving problems connected to the real life challenges in marketing by more sophisticated manners. Second challenge is to balance between methodology and practical issues. We say third challenge is rises of social interests in regulation, consumer activism and ideology, and public policy. Then, under above challenges, several qualitative methods and substantive areas/issues that should be studied in the near and distant future are shown.

13:15～13:45

Baek Jung Kim (U of British Columbia), Masakazu Ishihara (New York University) & Vishal Singh (New York University)

Title

The Impact of Peer Usage on User Retention for Crowdfunding Platforms

Abstract

The public school education system in the U.S. displays significant inequities in resources, exacerbated by the budget cuts after the Great Recession. Consequently, schoolteachers have had to find alternative resources to fulfill student needs and have increasingly turned to online crowdfunding. This paper examines U.S. public school teachers' adoption and repeat use of Donorschoose.org (the largest crowdsourcing platform in this domain). Our particular focus is on understanding the presence and nature of peer effects on short-term adoption and long-term repeat-use behaviors (i.e., user retention). Our newly assembled database captures information on approximately 233,000 teachers from 46,000 public schools from 2002 to 2014 (academic year). We supplement this data with various information, including attributes of schools (e.g., poverty and ethnic mix of students) and the government funding in school districts. Our key strategy for identifying causal peer effects relies on a rich set of fixed effects, lagged peers' decisions, and an instrumental variable approach to resolve endogeneity of the installed base peer effects in addition to the homophily, unobserved correlates, and simultaneity (i.e., reflection) problems. Furthermore, we document additional empirical evidence of peer effects using an exogenous shock from teachers' transfer between schools as a "mover design" framework. Confirming the findings in previous literature, our results suggest strong positive peer effects on adoption, implying the role of information played by experienced colleagues. However, peer effects on repeat use are adverse, discouraging teachers from using the platform repeatedly. The paper highlights the common-core nature of this context as teachers compete with each other for limited (primarily local) donors as a potential explanation.

13:45～14:15

Kohei Hayashida (U of California San Diego), Masakazu Ishihara (New York University), Makoto Mizuno (Meiji University) & Kosuke Uetake (Yale University)

Title

Addiction and Alcohol Taxation: Evidence from Japanese Beer Industry

Slides (Presenters only)

Abstract

This paper studies the effects of taxation and regulation on addictive alcohol consumption. Exploiting the changes in tax policies and sales regulation in the Japanese beer market, we first show some descriptive evidence that consumers (i) are addicted to alcohol, (ii) are forward-looking and stockpile, but potentially present-biased, and (iii) substitute across categories in response to policy changes. To quantify the impacts of policy changes, we then estimate a dynamic structural model of alcohol purchase and consumption where consumers can be present-biased. A series of counterfactual simulations show that the current Japanese alcohol tax system is suboptimal in that alternative policies can increase tax revenues while keeping alcohol addiction lower. Finally, we derive the optimal alcohol tax policy, taking both externalities and internalities into account.

14:15～14:45

Makoto Mizuno (Meiji University) & Hiroki Takikawa (U of Tokyo)

Title

Do People Signal Their Socioeconomic Status on Social Media? -An Analysis of “Conspicuous Tweeting” by Machine Learning

Slides (Presenters only)

Abstract

Our study focuses on the signaling of socioeconomic status (SES) signaling on social media platforms, which we refer to as "conspicuous tweets." We collected tweets from about 4,000 Twitter users, identified their respective occupations, and translated them into three SES scores. These scores were developed using preliminary research on approximately 500 different occupations.

Our application of topic modeling (LDA) revealed that the latent topics of the tweets are associated with both SES and the demographics of the users. Furthermore, it appears that Twitter users' interest in specific brands may vary in their SES. These findings suggest that disparities in SES may result in different tweeting behaviors, including conspicuous tweets.

However, machine learning techniques, such as L2 regularized logistic regression, could not accurately predict each user's SES based solely on the content of their tweets. This suggests that if a machine is unable to predict SES from tweet content, it is likely that humans will struggle as well. Therefore, at this point, there isn't compelling evidence to support the existence of conspicuous tweets generally. The paper will also discuss potential reasons for this finding and propose areas for future research to further investigate this phenomenon.

15:00～15:30

Baek Jung Kim (U of British Columbia), Jake An (U of Technology Sydney) & Sung Kwan Lee (Chinese University of Hong Kong)

Title

Understanding Irrational Gambling Behaviors under Financial Constraints

Abstract

This paper explores how gamblers become addicted so that they gamble irrationally, even with financial constraints. Using unique individual-level data from an investment app in Australia that includes users' gambling information (online sports betting) and granular user-level financial status, the paper documents that 1) gamblers gradually gamble more (i.e., evidence of becoming serious gamblers) and 2) this tendency is persistent even with financial constraints that are measured by overdraft fee. Particularly, gamblers reduce the amount of gambling when their balance is insufficient, but they recover their original gambling behaviors quickly and then maintain it in subsequent periods.  We also found similar patterns in lottery purchase behaviors. Merging with survey data, we suggest that gamblers' "risk-tolerant" attitude may explain these behaviors. While risk-averse gamblers do not show these behaviors, risk-tolerant ones do gamble more, even with financial constraints.

15:30～16:00

Masakazu Ishihhara (New York University) & Hiroshi Kumakura (Chuo University)

Title

What Distorts What We See in Online Reviews? Assessing the Sources of Reporting Bias in Review Ratings and Comments

Slides (Presenters only)

Abstract

The reliability of online user reviews as an information source hinges on the degree to which online reviews represent unbiased opinions of users. However, due to the voluntary reporting nature, online reviews could be biased both quantitatively (e.g., rating distribution) and qualitatively (e.g., distribution of topics mentioned in comments). While the literature recognizes this reporting bias, empirical evidence regarding what factors distort what aspect of online reviews by how much is limited. This paper uses a unique data set that combines consumer satisfaction survey data with online review data for the Japanese hotel industry, and studies the sources of reporting bias in review ratings and comments. Specifically, we investigate factors (usage experiences as well as consumer, product, and transaction characteristics) that influence the relationship between users' satisfaction and their propensity to post a review. To understand the bias in the qualitative aspect (e.g., text), we apply the Latent Dirichlet Allocation to extract qualitative information from users' comments. We propose a simple Bayesian statistical model that estimates the review reporting probability and quantifies the bias due to each level of each factor. Our results reveal a variety of factors that introduce different degrees of biases. In particular, we find that consumers' prior expectation, frequency of hotel stay, and the price paid relative to a hotel's median price tend to distort both rating and topic distributions. Our analysis provides insights on how consumers and firms should interpret opinions in online reviews so as not to be misguided by the biased reviews.

16:15～16:45

Srinagesh Gavirneni (Cornell University), Miguel I. Gómez (Cornell University), Koichi Yonezawa (U of Bristol) & Xinrong Zhu (Imperial College London)

Title

Leveraging User-Generated Content for Product Improvement: Bridging the Gap between Customer Inputs and Managerial Decisions

Paper (Presenters only)

Abstract

User-generated content (UGC) is a valuable yet underutilized source of information for evaluating product performance. In this paper, we propose an empirical approach to establish systematic relationships between UGC and product or production attributes, enabling customer-centric managerial decision-making. First, we demonstrate how food manufacturers can leverage natural language processing (NLP) tools to analyze UGC. By analyzing approximately 12,000 customer reviews from a major food manufacturer, we identify key customer perceptions related with topics like smell, taste, and packaging. We then examine the relationship between these perceptions and product or production attributes. Our findings identify certain product attributes or types of manufacturing facilities that are associated with distinct perceived issues observed in the UGC data. These insights offer valuable guidance for marketing and management decision-making for consumer packaged goods (CPG) manufacturers. Our approach provides a cost-effective and timely solution for enhancing product management and production quality control, compared to traditional survey methods.

16:45～17:15

Makoto Abe (U of Tokyo)

Title

Preference Reversal: Analytical Framework For Construal Level Theory That Incorporates Discounting

 Slides (Presenters only)

Abstract

Preference reversal is one of the topics which has attracted attention in the choice modeling research. Time discounting is often used to explain preference reversal in behavioral decision theory. However, the discounting theory fails to explain some types of preference reversal. Furthermore, preference reversals are limited to those along time axis (temporal distance). To extend our knowledge of discounting, this research provides an analytical framework that combines time discounting theory in behavioral decision theory and construal level theory (CLT) developed in social psychology. We posit three propositions for discounting: magnitude effect (the higher the construal level, the smaller the discount rate), sign effect (the discount rate is smaller for a loss than for a gain), and generalization of distance (discounting applies to psychological distance such as social distance). These propositions are validated by two studies. In study 1, we conducted a series of three experiments of lottery choice task on two samples of respondents—namely, students and web panel. In study 2, we estimated discount rates of higher-level and lower-level construals by employing multiple intertemporal choice task.

While many choices involve tradeoff among attributes, their consequence of the change in psychological distance is not that apparent. By identifying whether these attributes evoke high or low construal and whether the aspects are related to gain or loss, our approach greatly facilitates the analysis of how evaluation and preference are affected by the psychological distance, and thus that of preference reversal behavior.

17:30～

Small Social Gathering

1号館4階1409教室

懇親会にもご参加お願いします（参加費として1,000円申し受けます）