Journal Article
Buy-it-now or Take-a-chance: Price Discrimination through Randomized Auctions (with Elisa Celis, Markus Mobius and Hamid Nazerzadeh). Management Science. Forthcoming.Abstract

Increasingly detailed consumer information makes sophisticated price discrimination possible. At fine levels of aggregation, demand may not obey standard regularity conditions. We propose a new randomized sales mechanism for such environments. Bidders can "buy-it-now" at a posted price, or "take-a-chance" in an auction where the top d > 1 bidders are equally likely to win. The randomized allocation incentivizes high valuation bidders to buy-it-now. We analyze equilibrium behavior, and apply our analysis to advertiser bidding data from Microsoft Advertising Exchange. In counterfactual simulations, our mechanism increases revenue by 4.4% and consumer surplus by 14.5% compared to an optimal second-price auction.

Student Portfolios and the College Admissions Problem (with Hector Chade and Lones Smith). Forthcoming in The Review of Economic Studies . 2013.Abstract

We develop a decentralized Bayesian model of college admissions with two ranked colleges, heterogeneous students and two realistic match frictions: students find it costly to apply to college, and college evaluations of their applications are uncertain. Students thus face a portfolio choice problem in their application decision, while colleges choose admissions standards that act like market-clearing prices. Enrollment at each college is affected by the standards at the other college through student portfolio reallocation. In equilibrium, student-college sorting may fail: weaker students sometimes apply more aggressively, and the weaker college might impose higher standards. Applying our framework, we analyze affirmative action, showing how it induces minority applicants to construct their application portfolios as if they were majority students of higher caliber.

PDF onlineappendix_final.pdf dataappendix.pdf
Moral Hazard, Incentive Contracts and Risk: Evidence from Procurement (with Patrick Bajari) . Forthcoming in the The Review of Economic Studies. 2013.Abstract
Deadlines and late penalties are widely used to incentivize effort. Tighter dead- lines and higher penalties induce higher effort, but increase the agent's risk. We model how these contract terms affect the work rate and time-to-completion in a procurement setting, characterizing the efficient contract design. Using new micro-level data on Minnesota highway construction contracts that includes day-by-day information on work plans, hours worked and delays, we find evidence of ex-post moral hazard: contractors adjust their effort level during the course of the contract in response to unanticipated productivity shocks, in a way that is consistent with our theoretical predictions. We next build an econometric model that endogenizes the completion time as a function of the contract terms and the productivity shocks, and simulate how commuter welfare and contractor costs vary across different terms and shocks. Accounting for the traffic delays caused by construction, switching to a more efficient contract design would in- crease welfare by 22.5% of the contract value while increasing the standard deviation of contractor costs | a measure of risk | by less than 1% of the contract value.
timeincentives_final.pdf supplementary-appendix_final.pdf
Asymmetric Information, Adverse Selection and Online Disclosure: The Case of eBay Motors. American Economic Review. 2011;101 (4).Abstract
Since George A. Akerlof (1970), economists have understood the adverse selection problem that information asymmetries can create in used goods markets. The remarkable growth in online used goods auctions thus poses a puzzle. Part of the solution is that sellers voluntarily disclose their private information on auction webpage. This defines a precise contract to deliver the car shown for the closing price which helps protect the buyer from adverse selection. I test this theory using data from eBay Motors, finding that online disclosures are important price determinants; and that disclosure costs impact both the level of disclosure and prices.
Supplementary Appendix.pdf Final Working Paper.pdf PDF
Procurement with Time Incentives: Theory and Evidence (with Patrick Bajari). Quarterly Journal of Economics. 2011.Abstract

In public procurement, social welfare often depends on how quickly the good is delivered. A leading example is highway construction, where slow completion inflicts a negative externality on commuters. In response, highway departments award some contracts using scoring auctions, which give contractors explicit incentives for accelerated delivery. We characterize efficient design of these mechanisms. We then gather an extensive data set of highway projects awarded by the California Department of Transportation between 2003 and 2008. By comparing otherwise similar contracts, we show that where the scoring design was used, contracts were completed 30–40% faster and the welfare gains to commuters exceeded the increase in procurement costs. Using a structural model that endogenizes participation and bidding, we estimate that the counterfactual welfare gain from switching all contracts from the standard design to the efficient A+B design is nearly 22% of the total contract value ($1.14 billion).

On-line Appendix Working Paper Version Data Appendix PDF
Working Paper
Just starting out: Learning and price competition in a new market (with Ulrich Doraszekski and Ariel Pakes). 2013.Abstract

Deregulation of the frequency response market in the UK allowed electricity firms to compete on price in an otherwise stable environment. We provide an analysis of the evolution of the deregulated market from the date it started. Initial activity was volatile, with some firms exploring different prices, while others made few price changes. This was followed by a period in which prices fell and the variance in the cross sec- tional distribution of bids declined markedly. By the end of our study price changes had became relatively rare and small, consistent with convergence to a static Nash equilibrium. We examine how well models of learning do in predicting play during the period prior to convergence but after the initial volatility. Models where perceptions of competitors’ play depend on past play suggest that firms’ weight recent play dispro- portionately. We also find evidence of statistical learning about the underlying demand parameters conditional on competitors’ play. A model that combines these two features fits quite well: it is able to explain 37% of the share-weighted variation in prices, even though none of the model parameters are chosen to fit the pricing behavior.

Who benefits from improved search in platform markets? (with Albert Wang). 2013.Abstract
Online platforms invest large sums in their search technology. Motivated by this, we investigate how lowering search costs affects the welfare of market participants, in a model where buyers with horizontally differentiated tastes search and compete for goods in an auction. We identify a "matching effect", whereby lower search costs endogenously shift market participation in favor of some goods and against others. We prove that there is a unique equilibrium, and demonstrate that the decentralized market achieves the social planner's solution. Decreasing search costs thus improves joint welfare; and yet surprisingly joint seller revenue may fall.
A Demand System for a Dynamic Auction Market with Directed Search (with Matthew Backus). 2012.Abstract
We develop a demand system for a dynamic auction market with directed search. In each period, heterogeneous goods are exogenously supplied and sold by second price auction, and incumbent bidders choose which good to bid on and how much to bid. Bidder valuations are multidimensional, private and perfectly persistent, and the population of bidders evolves according to an exogenous entry and endogenous exit process. We prove that the state of the market (which includes active bidders' types and information sets) evolves as a geometrically ergodic Markov process. We characterize best responses as solutions to a partially observed Markov decision problem and provide conditions under which the econometrician can identify equilibrium strategies from time series data. We provide additional conditions under which this allows nonparametric identification of preferences. When the market is large so that each bidder's actions are informationally small, we show bidderwise identification: the valuations of bidders whose individual time series includes a bid on every product are identified. Two-stage nonparametric and semiparametric estimation procedures are proposed, and shown to work well in Monte Carlo and counterfactual simulations.