Session 7: Empirical Market Design

We develop a method to recover primitives from data generated by artificial intelligence (AI) agents in strategic environments such as online marketplaces and auctions. Building on how leading online learning AIs are designed, we assume agents minimize their regret. Under asymptotic no regret, we show that time-average play converges to the set of Bayes coarse correlated equilibrium (BCCE) predictions. Our econometric procedure is based on BCCE restrictions and convergence rates of regret-minimizing AIs. We apply the method to pricing data in a digital marketplace for used smartphones. We estimate sellers’ cost distributions and find lower markups than in centralized platforms.

Although social housing is prevalent in many developed countries, there is no consensus over how to design allocation rules. Measuring the impact of a change in rules requires predicting how applicants will respond. In a context where rents are fixed, application data scarce, and allocation rules lack transparency, disentangling an applicant’s preferences from their probability to receive an offer is challenging. This paper develops a dynamic framework which makes use of a novel, comprehensive dataset of French social housing applications to separately identify preferences of applicants to social housing from their expectations over future offers and the allocation rules. This allows to compare the welfare impact of changes in the allocation rules. Results indicate that the current system favors households with French nationality, and disadvantages precarious households like single mothers compared to the rest of the population. Counterfactual analysis suggests better targeting low-income households would significantly improve welfare. Mechanisms based on applicant waiting time like f irst-come-first-serve are shown to be welfare-reducing.

Cap-based regulations are widely used to address distributional disparities in matching markets, but their efficiency relative to alternative instruments such as subsidies remains poorly understood. This paper develops a framework for evaluating policy interventions by incorporating regional constraints into a transferable utility matching model. We show that a policymaker with aggregate-level match data can implement a taxation policy that maximizes social welfare and outperforms any cap-based policy. Using newly collected data from the Japan Residency Matching Program, we estimate participant preferences and simulate counterfactual match outcomes under both cap-based and subsidy-based policies. The results reveal that the status quo cap-based regulation generates substantial efficiency losses, whereas small, targeted subsidies can achieve similar distributional goals with significantly higher social welfare.

In many markets, advance interim contracts include an explicit right to renege, granting one party the option to switch to more efficient matches that emerge later in the search process. This paper studies the formation and welfare implications of such interim contracts, leveraging novel data from a brokerage firm in the trucking industry. The broker allocates advance shipment contracts to carriers through a dynamic auction mechanism and penalizes cancellations through a reputational mechanism. I develop a theoretical model linking the carriers bidding problem to the firms cancellation penalties through a dynamic job-search problem and structurally estimate the model from rich data on bids and cancellations. In counterfactual simulations, I show that the firm is incentivized to lower cancellation penalties as the option value of the right to renege is priced into carrier bids. The results rationalize the large degree of contractual flexibility observed in the trucking industry as an efficient market outcome rather than one constrained by limited enforcement.

Open source software (OSS) is a form of public knowledge widely provided and relied on by the private sector. To study the effects of growing government involvement in this critical public good, I build a new empirical model where high-tech firms choose software inputs and developer labor in competitive equilibrium. For estimation, I create a new dataset of OSS and in-house investment for the global web development industry, where software choices are directly observable. I simulate counterfactuals to assess the global impact of China tightening its recent internet restrictions on cross-border OSS collaboration or increasing its financial support for domestic OSS. I find that stricter restrictions do little to boost domestic OSS investment. Instead, lost spillovers raise web development costs in China by $2 per dollar of disincentive and $7 globally. Heightened subsidies prove more effective at increasing domestic investment and cut global costs by $11 per dollar of subsidy—tripling if the US responds in kind.

College admission platforms aim at achieving a balance between avoiding congestion and allowing for ex-post flexibility in students’ matches. The latter is crucial as the existence of off-platform options implies that some students will drop out of the platform in favor of their outside option, freeing up seats in on-platform programs. Sequential assignment procedures introduce such flexibility, by creating a dynamic trade-off for students: they can choose to delay their enrollment decision to receive a better offer later, at the cost of waiting before knowing their final admission outcome. We quantify this trade-off in a setting in which waiting costs can be heterogeneous. We use rich administrative data on rank-ordered lists and waiting decisions from the French college admission system to estimate a dynamic model of application and acceptance decisions. We find that waiting costs are large, especially for students of low socio-economic status. Yet, by improving students’ assignment outcomes relatively to a standard single-round system, the multi-round mechanism decreases the share of students leaving the higher education system without a degree by 5.4% and leads to large welfare gains for all groups.

We propose a sufficient statistic for designing AI information-disclosure and selective automation policies. The approach allows for endogenous and biased beliefs, and effort crowd-out, without using a structural model of human decision-making. We deploy and validate our approach in a fact-checking experiment. Humans under-respond to AI predictions and reduce effort when presented with confident AI predictions. Overconfidence in own-signal rather than under-confidence in AI drives AI under-response. The optimal policy automates decisions where the AI is confident and delegates the other decisions while fully disclosing the AI prediction. Although automation is valuable, the benefit of assisting humans with AI is negligible.

In this paper, we argue that the New York City congestion pricing scheme that was launched on January 5, 2025 has a major shortcoming: it has a much more severe impact on the drivers of personal vehicles than on the passengers of taxis and ride-hailing vehicles or on the clients of delivery services. In addition to being inequitable, this shortcoming also makes the congestion pricing scheme relatively ineffective at solving the traffic congestion problem inside the Central Business District, due to the fact that the drivers of personal vehicles constitute a minority of traffic there. We provide empirical evidence from the launch of the current plan, and propose a simple modification to the scheme that addresses this short-coming.

Matching mechanisms that elicit strength-of-preference can exhibit efficiency gains over those that do not. To quantify these gains, we propose a measure of approximate ex-ante Pareto efficiency. We use this notion to quantify the efficiency improvement of the raffles mechanism (which we define) over the deferred acceptance mechanism (DA). We complement our theoretical analyses with experimental results. Using human subjects, we find that the raffles mechanism yields higher average payoffs as predicted by the theory, despite the fact that subjects play only approximately-optimal strategies.

Product testing plays an important role in the functioning of markets for innovative new products, where uncertainty exists regarding the product’s safety, quality, or other attributes. Despite the importance of testing, in practice different products face a wide range of regulatory regimes and private testing incentives—even similar (or identical) products face disparate regulations and private incentives across geography and time. In this paper, we develop a dynamic model of innovation, testing, and competition between firms to examine the interplay between private incentives to test and regula-tory requirements. We calibrate the model using data from the medical device sector and then conduct several counterfactual comparative static exercises. In our calibrated model, we find that: Even in the absence of regulatory requirements, firms have sub-stantial private incentives to conduct their own product tests; this is especially the case for higher quality products. Greater competition tends to widen the gap between social and private testing incentives. Minimum testing regulations both deter entry by low quality products and increase testing by high quality products, with the latter effect the main determinant of optimal testing policy.

I show that most reductions in 401(k) plan fees over the past decade come from updating plan menus to incorporate newly introduced funds with lower fees. Because employer sponsors face transaction costs when selecting and switching plan providers, providers can delay menu updates, preventing participants from accessing these lower-fee funds. To quantify the impact of transaction costs, I develop a dynamic structural model of employers’ provider choice and providers’ fee competition. I estimate that transaction costs contribute 11 bps to plan fees on average or $1 billion in total. However, mitigating transaction costs has limited effects once forward-looking providers revise their fee strategies in response. By contrast, consolidating plans of small employers can generate substantial fee savings.

This paper analyzes the consumer-welfare effects of intermediaries in a pension and annuity market with adverse selection. Intermediaries provide advice, helping individuals improve decisions when understanding products is complex and costly, but may introduce distortions due to agency problems. In an insurance market, intermediary effects on choices can impact adverse selection and, through it, prices. I document the importance of intermediation and its connection to adverse selection in the Chilean pension market, where products are complex and intermediaries have a financial incentive to steer consumers toward annuities. To quantify the effects of potential intermediary regulations, I develop and estimate a dynamic demand model that includes life-cycle decisions, product, and intermediation choices. I find intermediaries have the potential to improve welfare: retirees would give up around 250 USD a year to eliminate frictions in product choices. Despite intermediaries steering a majority of their customers into annuities, a ban on intermediation is approximately consumer-welfare neutral. The variety of annuities allows intermediaries to recommend close substitutes to the outside option, limiting the harm from misaligned incentives. Decision costs without intermediaries and annuity price increases due to adverse selection erode any gains from a ban. In light of policy concerns regarding the role of intermediaries, my results highlight the potential value provided by advisors – even with biased incentives – when choices are complex and stakes are high.

Commission rates for housing transactions are twice as high in the United States than in other countries. Policymakers have raised concerns that the practice of sellers offering buyers’ brokers commissions can lead to high commissions and harm consumers. This paper empirically examines the equilibrium impacts of a proposed policy called “decoupling,” which would require buyers and sellers to each pay their respective brokers. I develop a structural model integrating buyers, sellers, and brokers to characterize the equilibrium house prices, com-missions, and to assess the welfare impacts of the policy. I estimate the model with rich observed heterogeneity and credible sources of identifying variation using shifters of house prices and commissions. I find that decoupling reduces commissions paid by 53%, as sellers no longer have to offer high commissions to attract buyers, and brokers compete for price-sensitive buyers. Sellers and buyers experience a surplus gain of 4% of the total transaction value from hav-ing higher net proceeds than the status quo. I find notable surplus gains for buyers across income groups as sellers pass through part of their commission savings to house prices.

In the last decade, large financial institutions in the United States have purchased hundreds of thousands of homes and converted them to rentals. This paper studies the welfare consequences of institutional ownership of single-family housing. We build an equilibrium model of the housing market with two sectors: rental and homeownership. The model captures two key forces from institutional purchases of homes: changes in rental concentration and reallocation of housing stock across sectors. To estimate the model, we construct a novel dataset of individual homes in metropolitan Atlanta, identifying institutional owners of each house and collecting house-level daily prices, rents, vacancies, web page views, and customer contacts from Zillow. Overall, we find that institutional acquisitions decrease rents and increase rental transactions, leading to large welfare gains for renters. This net benefit reflects two opposing forces: while higher concentration raises rents, higher rental supply lowers rents enough to more than offset the effect of concentration, pushing rents down overall. These renter gains come at the expense of homebuyers, whose welfare falls. On the supply side, institutional acquisitions benefit house sellers but harm the average landlord.