Industrial Organization, Quantitative Marketing
Regulating Privacy Online: An Economic Evaluation of the GDPR (American Economic Journal: Economic Policy, 2024)
Joint with Garrett Johnson & Scott ShriverAbstract: Modern websites rely on personal data to design better content and to market themselves. The European Union’s General Data Protection Regulation (GDPR) was intended to make access to such personal data much more difficult, with the goal of protecting user privacy. We examine the GDPR's impact on website pageviews and revenue for 1,084 diverse online firms using data from Adobe's website analytics platform. Using a panel differences approach, we find a reduction of approximately 12% in both EU user website pageviews and website e-commerce revenue recorded by the platform after the GDPR's enforcement deadline. To understand the policy's impact, we must separate the GDPR's effects on real outcomes from its impact on data recording. We derive informative bounds on both components by first examining site-usage patterns of selected users that we continue to observe post-GDPR. We bound changes to data recording, attributed to users opting-out of data collection, to be between 4% and 15%. This implies that at least 7% of our GDPR estimates (0.8% of total) arise from the consent effect on data recording. We then find larger effects on traffic from email and display ad marketing channels—specific targets of the GDPR. We conclude that at least 8% of our GDPR estimates (0.4% of total) arise from the GDPR's real effect on marketing alone. However, we do not find evidence that consent interfaces dissuade users from browsing sites.
Privacy & Market Concentration: Intended & Unintended Consequences of the GDPR (Management Science, 2023 - Featured Article)
Joint with Garrett Johnson & Scott ShriverAbstract: We show that websites' vendor use falls after the European Union's General Data Protection Regulation (GDPR), but that market concentration also increases among technology vendors that provide support services to websites. We collect panel data on the web technology vendors selected by more than 27,000 top websites internationally. The week after the GDPR's enforcement, website use of web technology vendors falls by 15% for EU residents. Websites are more likely to drop smaller vendors, which increases the relative concentration of the vendor market by 17%. Increased concentration predominantly arises among vendors that use personal data such as cookies, and from the increased relative shares of Facebook and Google-owned vendors, but not from website consent requests. Though the aggregate changes in vendor use and vendor concentration dissipate by the end of 2018, we find that the GDPR impact persists in the advertising vendor category most scrutinized by regulators. Our findings shed light on potential explanations for the sudden drop and subsequent rebound in vendor usage.
Designing Monitoring Programs, 2023 (Draft available upon request)
Abstract: I study the design of voluntary consumer monitoring programs at a large U.S. insurer using data from almost 8 million unique drivers. Monitoring technology tracks driver behaviors and provides performance incentives for improved driving, in the form of discounts on future premiums. The profitability and welfare of these programs depend on who participates (driver selection) and how much drivers respond to performance incentives by reducing risk (moral hazard). I am able to separate these two mechanisms by leveraging unique variation in both performance and participation discounts. Using regression-discontinuity designs, I document that drivers respond to performance incentives by reducing accident risk by approximately 0.6%, on average – or 10% of the mean accident risk in my sample. These results inform a structural model that accounts for selection and moral hazard. The model suggests modest welfare gains due to limited participation. I then use the model to examine counterfactual incentive designs that highlight a key trade-off for program design: increasing driver participation comes at the cost of stronger performance incentives. My results highlight how monitoring programs can be used by firms to screen risky consumers and incentivize risk reduction.
Generative AI in Equilibrium: Evidence from a Creative Goods Marketplace, 2025 (Revision Requested, Quarterly Journal of Economics)
Joint with H. Tai LamAbstract: Artists and policy makers are concerned that Generative Artificial Intelligence (GenAI) may lead to disappearance of non-GenAI content. In this paper we study the implications of GenAI for the production and consumption of creative goods; such as images, music, and writing. We first introduce a simple model of technology adoption and production that highlights how GenAI may influence market equilibrium. Then, using a difference-indifferences design, we causally estimate the impact of GenAI on production, firm entry, a measure of product quality, variety, and sales. We find that GenAI is a substitute for non-GenAI content, increases competition in markets, and crowds-out the production of non-GenAI content. Overall this leads to an increase in the quality and variety of produced and sold goods, and increased sales. Thus, our results imply that unregulated GenAI poses a substantial threat to non-GenAI production but is likely beneficial for most consumers. We leverage heterogeneity across markets to examine the role of market structures and legal differentiation and labeling in mitigating or enhancing GenAI adoption and influencing market equilibria. Evidence suggests policy that regulates content labeling and enforces clear disclosure can mitigate concerns that poor quality GenAI products may lead to market collapse.
Does Transparency Matter in Opaque Product Markets? Insights from Privacy, 2025