Program
Papers and Abstracts
1. What Does a Badge Add? Platform Certification and Restaurant Sales
Review platforms increasingly certify sellers on top of the reviews they host. We ask what certification adds for restaurants that already have many reviews. We study Dianping’s 2024 Must-Eat List, which the platform compiles from user reviews. Monthly transaction data cover five major cities that host 17 percent of the national list. In a difference-in-differences design with propensity-score-matched controls, first-time listing is associated with a 23.6 percent increase in sales. The gain comes from orders rather than prices. Losing the badge is associated with a 25.8 percent decline in sales. Within one kilometer, never-listed restaurants show no sign of large sales losses. Certification thus appears to retain value for sellers even when reviews are plentiful.
2. Choose for You, or Let You Choose? Recommendation Algorithms and Platform Interface Design
Digital content platforms deliver recommendations through two contrasting interfaces: playing items sequentially one at a time (push, as in TikTok) or presenting a grid of previews and letting users choose (menu, as in Xiaohongshu or YouTube). We study how this design choice governs both consumer search and creators' content provision. When content is fixed, covers allow consumers to screen out poor matches using private taste information, meaning the menu dominates push only when opening a video is sufficiently costly relative to glancing at covers. Because a more precise recommendation algorithm reduces residual match uncertainty, algorithmic improvements act as an information substitute for user screening, favoring push. On the supply side, the interface alters creators' effort allocation between content quality and thumbnail appeal (clickbait). Under a menu, creators divert effort into packaging, inducing an equilibrium signal-jamming rat race that crowds out substantive quality. Higher algorithmic precision curbs this crowd-out by rewarding intrinsic quality through platform curation. Combining consumer search and creator incentives, we show that the platform adopts a menu interface when its algorithm is imprecise, but switches to push as recommendation capabilities mature.
3. Most-Favored-Nation Clauses and the Scope of Entry
This paper examines how selective most-favored-nation (MFN) clauses affect entry and competition in industrial procurement. We develop a model in which an entrant chooses which buyers to serve before competing with an incumbent in prices, and show that an MFN can encourage entry while limiting the entrant's competitive scope. By protecting one buyer, the incumbent commits to a softer pricing response at the uncovered buyer. The entrant benefits from this response, but expanding to the protected buyer removes the benefit. The welfare implications depend on how broadly the entrant would compete without the clause. When entry would otherwise be broad, any voluntarily adopted selective MFN that changes outcomes lowers both expected total surplus and consumer surplus. In contrast, when the entrant would already serve only one buyer, the MFN raises total surplus under our maintained conditions by transmitting the benefits of competition to the otherwise monopolized account. Such agreements can be privately supported when protection goes to a buyer unlikely to attract the entrant on its own. Selective coverage can be optimal even before entry costs are known. The cost of obtaining buyer consent can also make the incumbent prefer to protect a smaller buyer, even though protecting a larger buyer provides a stronger pricing commitment.
4. Procrastination and Competition Failure
We develop a model of price competition with procrastinating consumers in which market discipline is supposed to arise from both the initial selection of providers and the possibility of switching providers. As in other theories, consumers may forego large gains by sticking with their initially chosen offer, so competition at the switching stage is weak. Unlike in other theories, consumers — who falsely expect to switch soon — may also fail to select the best starting offer, so competition at the initial stage is weak as well. This mechanism can translate temporary product differentiation into permanently high prices, greatly enhance the price effect of persistent differentiation, or generate high markups even with perfect substitutes. Reflecting the same mechanism, sign-up deals do not serve their classically hypothesized role of returning ex-post profits to consumers, but instead often exacerbate the failure of price competition. We complement our analysis with a tailored survey of consumers, confirming the logic of procrastination underlying our model. Consumer procrastination thus emerges as a novel source of competition failure that applies where other theories do not, helping to explain high average prices in many markets with switching costs.
5. Tying in Two-sided Markets
This paper offers a strategic explanation for the prevalence of tying in two-sided markets. We extend the standard Hotelling duopoly model to allow each platform to serve two groups of users whose participation decisions are linked by cross-group externalities. When these externalities are sufficiently strong, prices on the single-homing side can become strategic substitutes, reversing the usual pattern of strategic complementarity in the one-sided model. A price cut by one platform then induces its rival to raise its price. By tying, a platform commits to a lower effective price on that side. This commitment reduces the rival’s profit but can increase the tying platform’s own profit. Tying can therefore be profitable even when the platform accommodates its rival, extending the rationale for tying beyond entry deterrence. When both platforms can tie, they may face a prisoner’s dilemma in which each chooses to tie even though both would earn higher profits under separate selling. Tying can also increase total welfare.
6. Consent-or-Pay: Pricing Privacy and Data Sharing
We study the welfare effects of consent-or-pay, a pricing strategy that lets consumers accept tracking for a lower fee or pay more to avoid it. Consent-or-pay raises total welfare relative to tracking-only service when basic access is free. When both basic fees are positive, we identify conditions under which the gains from protecting high-privacy-cost consumers outweigh the pricing effects of privacy-cost dispersion. Under affine inverse hazard rates, consent-or-pay outperforms no-tracking service when its basic fee is positive, but free basic access does not guarantee this ranking. A modest cap on the privacy surcharge can improve welfare when basic access is free but backfire when it is paid; a cap on the total premium price can instead lower both prices and improve welfare. When the firm chooses data intensity, too few buyers may share data even though the firm collects too much from those who do at prevailing prices. When consumers learn their privacy costs after subscribing, a committed refund can induce efficient data sharing and improve welfare.
7. Business Model Competition and the Supply of Privacy-preserving Apps
Privacy-conscious consumers should sustain a market for subscription-funded apps that charge a fee and collect little data, yet paid apps account for only 3-5 % of apps in major app stores. We explain this puzzle with a market failure: ad-funded apps benefit from data pooled across the advertising ecosystem, whereas subscription-funded apps do not. By raising ad revenue per user in every category, these data-driven network effects strengthen ad-funded apps and tilt competition in their favor. We model business-model competition across a continuum of app categories with endogenous entry and an equilibrium data stock determined as a fixed point. Because data raises targeting accuracy, which in turn draws more users to ad-funded apps, an exogenous increase in data is amplified in equilibrium by a data multiplier. Better targeting is pro-competitive on the intensive margin but deters privacy-preserving entry on the extensive margin. With costless entry, better targeting always benefits consumers; with a positive entry cost, consumer surplus is single-peaked in targeting accuracy. When most apps are ad-funded, consumers benefit from limits on data combination and from digital-advertising taxes. We characterize when consent mandates can implement consumers' preferred market structure and show that a digital-advertising tax can implement it whenever laissez-faire targeting is excessive.