Fishing for Fools

How big is the e¤ect of a few fools on market outcomes? We argue that in auctions, even a small share of overbidding behavioral agents have a large e¤ect, because the auction format “…shes”for the highest-bidding behavioral buyers. Through this …shing mechanism, behavioral agents disproportionately increase auction pro…ts. They also generate a large welfare loss by crowding out the demand of rational agents, who obtain the good with delay or not at all. The welfare e¤ect of a few fools can be further ampli…ed when the market mechanism is endogenous: even with a small share of behavioral agents, sellers may prefer ine¢ cient auctions over e¢ cient …xed-price markets. Evidence from eBay supports the existence of overbidding and con…rms its ampli…ed e¤ect on allocation and pro…ts. Sellers who adjust the details of auctions to exploit buyer inattention earn higher revenue. Our predictions about sellers’market choice match stylized facts on the use of auctions versus …xed-price markets. E-mails: [email protected], [email protected]. We would like to thank Christopher Adams, Stefano DellaVigna, Ed Glaeser, Dan Kahneman, Matthew Rabin and seminar participants for helpful comments. Johnny Dai, Eva Chen, Kimberly Fong, and David Liu provided excellent research assistance. How large is the e¤ect of behaviorally biased agents on market outcomes? The literature in behavioral economics and psychology has documented a number of systematic deviations from the standard economic model.1 While the evidence that such deviations exist is hard to dispute, less is known about the extent to which behavioral biases a¤ect economically relevant decisions and market outcomes (Camerer, 2006; Ellison, 2006). In markets, several forces operate that can reduce the impact of agents with behavioral biases. High-stake incentives and repeated transactions discipline deviations (Stigler, 1958; Becker, 1957). Opportunities to learn and sort reduce the impact of biases relative to their measurement in laboratory experiments (Levitt and List, 2007). Moreover, the impact of any remaining behavioral agents is likely to be small because market outcomes are determined primarily by the large mass of rational agents. These arguments suggest that behavioral biases may not be important for allocations and welfare in practice. In this paper, we make two contributions to this discussion. First, we argue that there exist market mechanisms that amplify, rather than reduce, the e¤ect of behavioral biases. Speci…cally, we show that in auction markets, even a small share of overbidding behavioral agents can have a large e¤ect on allocation, pro…ts and welfare because auctions select highest-bidding behavioral buyers as winners. Second, sellers will optimally choose, when available, market mechanisms that amplify the impact of behavioral types, to maximize pro…ts by exploiting these agents. For example, even if only a small share of agents tend to overbid, sellers may endogenously shift to allocate goods through ine¢ cient auctions rather than e¢ cient …xed-price markets. This endogenous choice of market mechanism further ampli…es the welfare e¤ect of behavioral agents. As a result of these two mechanisms, behavioral biases can have substantial in‡uence on market outcomes and welfare in some environments. These observations also suggest a new explanation for the widespread use of auctions for allocating goods in practice. The economics literature links the success of the auction format to pro…t maximization and e¢ ciency under incomplete information.2 Auctions identify the bidder who values the good the most, and is thus willing to pay the highest price. Bidders may, however, be willing to pay a high price for reasons other than a high valuation. Such concerns about participants bidding “too much”are as old as auctions: In ancient Rome, legal scholars debated whether auctions were void if the winner was infected by calor licitantis, i.e., “bidder’s heat.”3 As we document in Section 1, laboratory studies and …eld data provide evidence that auction participants sometimes bid too much, due to psychological mechanisms such as bidding fever, joy of winning, loss aversion, and inattention towards, or limited memory of, alternative purchase opportunities. For example, we show that a large fraction of eBay auctions end at a price above the …xed price at which the See the surveys by Rabin (1998) and DellaVigna (2008). See, e.g., Milgrom (1987). Corpus Iuris Civilis, D. 39,4,9 pr. See Malmendier (2002).