Bounds Approaches to Empirical Market Design

National Bureau Of Economic Research Inc, Cambridge MA

Investigators

Abstract

Abstract: This project develops new methods and presents new data to study two important forms of price discovery: auctions and bargaining. Both auctions and bargaining have received attention in the theoretical economic literature but many empirical questions remain, and both forms of transaction are widely used methods of price discovery and trade which affect governments, consumers, and suppliers worldwide, both in developed and less-developed countries. Providing tools for robust analysis of bargaining and auction data will better allow researchers and policymakers to make informed decisions about how to design markets. For example, these tools can help identify markets in which a fixed, nonnegotiable price would be more efficient in practice than allowing for negotiated prices, and identifying cases in which the opposite is true. These tools can also be used for studying how buyers and sellers are affected by different designs of auctions or bargaining settings, measuring the impact of collusion on market participants, or many other questions about the design of marketplaces. The specific real-world settings which this project uses to illustrate these tools are bargaining and auctions transactions for online consumer-to-consumer sales and bargaining and auctions for used cars. At a technical level the project brings together econometric tools for partial identification and bounds estimation and the questions of empirical market design. Bounds approaches provide a framework for analyzing questions about efficiency and optimality of market mechanisms under weak structural assumptions. The project will provide new nonparametric partial and point identification results and estimators of important model features. Estimation will build on recent developments in the moment inequality literature. The first part of the project will demonstrate how data on auction bids and reserve prices can be used to either point or partially identify all features of the auction (depending on whether the reserve price is private knowledge of the seller or publicly observed) while allowing for an unknown number of bidders and unobserved heterogeneity. These results will then be applied to evaluate consumer surplus under different legal regimes regarding used smartphone sales. The second part of the project is concerned with data on alternating-offer price negotiations and will derive bounds on both buyer and seller valuations under minimal assumptions and provide estimators for bounds of functions of these distributions. These bounds will be used to analyze how close the current bargaining mechanism is to a fully efficient bargaining mechanism and document features of products or bargaining parties which contribute to efficiency. Finally, in a combined auction and bargaining setting the proposed research will show that the object of interest in the auction (distribution of buyer values for a given car type) is point identified and the object of interest in bargaining (distribution of seller values for a given car type) is partially identified.