Research (illustrated)
Where Does Advertising Content Lead You? We Created a Bookstore to Find Out
(with Anna Tuchman) [data and codes, online supplement]
Forthcoming at Marketing Science
Funded by: $50,000 Amazon Research Grant
We study how advertising content influences consumers' decisions. To this end, we create a simulated online bookstore that imitates a real online shopping experience. We then conduct a pre-registered and incentive-compatible experiment in which we randomly expose store visitors to display ads, randomizing both advertising exposures and content. We find that ad content plays a major role in shaping advertising effects. Ads that reveal the book's attributes generate substantial effects on search and choice, over and above the impact of ads that do not reveal these attributes. Ads that call attention to a book's low price reveal that the book is less expensive than consumers thought, universally increasing demand for the advertised book. By contrast, ads that highlight the book's genre induce some consumers to search and buy the book but lead others to reject this option without search. These polarized responses may increase or decrease the total number of searches and purchases of the advertised book depending on the share of consumers who favor the revealed genre. Overall, our findings suggest that advertisers should carefully choose which product attributes to reveal in their ad copies.
The Promotional Effects of Live Streams by Twitch Influencers
(with Yufeng Huang) [latest draft] Conditionally Accepted at Marketing Science.
We study the effectiveness of influencer marketing in the video game industry. To this end, we collect novel high-frequency data from Twitch.tv, a major video game streaming platform, by monitoring live streams every 10 minutes for eight months. Using plausibly exogenous variation in the broadcast hours of Twitch influencers, we show that sponsored Twitch streams generate a small and short-lived increase in the number of players and game sales. We also show that sponsored streams encourage other influencers to broadcast the game organically, amplifying the initial effect of the sponsorship. Additionally, live streams serve as informative advertisements that benefit lesser-known games released by small publishers and games with appealing attributes. Combining these results, we provide back-of-the-envelope calculations showing that although sponsored live streams benefit some games, they are not worth the investment in most cases. This result questions the conventional wisdom that influencer marketing generates higher returns on investment than traditional advertising.
Welfare Effects of Personalized Rankings
(with Rob Donnelly and Ayush Kanodia)
Many online retailers offer personalized recommendations to help consumers make their choices. While standard recommendation algorithms are designed to guide consumers to the most relevant items, retailers may deviate from them and instead steer consumers toward profitable options. We ask whether such strategic behavior arises in practice and to what extent it reduces consumers' benefits from personalized recommendations. Using data from a large-scale randomized experiment in which an online retailer introduced personalized rankings, we show that personalization makes consumers search more and generates more purchases relative to uniform bestseller-based rankings. We then estimate a model of search and rankings and use it to reverse-engineer the retailer's objectives as well as to estimate how personalized rankings affect consumer welfare. Our results reveal that, although the current algorithm does put positive weight on profitability, personalized rankings still increase consumer surplus. Using this case study, we argue that online retailers may generally have incentives to adopt consumer-centric personalization algorithms as a way to retain consumers and maximize long-term growth.
Measuring Benefits from New Products in Markets with Information Frictions
(Job Market Paper)
I study how much consumers benefit from new products in markets with information frictions. I analyze new products in the U.S. hard drive market, which is characterized by ample product innovation. Using unique click-stream data, I measure the magnitude of two frictions, category consideration and costly search, and show that both play a crucial role in shaping consumer demand. To estimate consumer surplus from new products, I develop a search model that captures both frictions and propose a novel Bayesian estimation method to recover its parameters. I then show that ignoring information frictions leads researchers to underestimate the consumer surplus from new hard drives because it appears that consumers do not value the combinations of attributes these hard drives offer. Partly eliminating frictions, through marketing efforts or market-wide transparency initiatives, can help consumers to more fully internalize the benefits of new product launches.
Estimation of Preference Heterogeneity in Markets with Costly Search
(with Stephan Seiler, Xiaojing Dong, and Liwen Hou)
We study the estimation of preference heterogeneity in markets where consumers engage in costly search to learn product characteristics. Costly search amplifies the way consumer preferences translate into purchase probabilities, generating a seemingly large degree of preference heterogeneity. We develop a search model that allows for flexible heterogeneity in preferences and estimate its parameters using a unique panel dataset on the search and purchase behavior of consumers. The estimation results reveal that when search costs are ignored, the model overestimates standard deviations of product intercepts by 66%. We show that this bias leads to incorrect inference about price elasticities and seller markups and has important consequences for targeted marketing.
Make Every Second Count: Time Allocation in Online Shopping
(with Rafael Greminger and Yufeng Huang) [latest draft]
We study how the opportunity cost of time influences online search and purchase decisions of consumers. To this end, we construct a comprehensive e-commerce dataset that describes how consumers allocate their shopping time across retailers, categories, and products. We show that during cold weather, when outdoor activities become less appealing and the opportunity cost of time decreases, consumers spend more time shopping online. They deepen their search by examining more products per category and broaden it by visiting more product categories, which leads to increased purchases. To capture the role of time constraints, we develop and estimate a novel model of time allocation and search. We estimate the average opportunity cost of time to be $33 per hour, which aligns with the average hourly wage. We also find that search costs increase with age and income, and that low- and medium-income consumers use their available time to shop and buy more, thereby partially offsetting their limited budgets.