skip to content

Managing Bidder Learning in Retail Auctions

When firms exploit behavioral biases it is natural to think that, eventually, consumers will learn to avoid their mistakes, limiting their exploitation. Profit maximizing firms, however, have an incentive to undermine such learning. We study the consumer learning dynamics and the firm’s response in a multi-unit descending price auction with a simultaneous fixed price offer. In our panel of 8 million bids by 280.000 bidders, consumers often bid more than the fixed price. Depending on competing bids, an overbid can lead to paying more than the fixed price (overpaying). We argue overpaying increases the saliency of the consumers’ mistake by making it payoff relevant, which is likely to affect consumer learning. Indeed, bidders who overpay subsequently overbid less often and are more likely to leave the market compared to bidders who similarly overbid but did not overpay. We show the resulting loss in future profits makes overpaying undesirable, and document a structural break in our data at which the firm eliminates such overpayments — and the resulting consumer learning — through changes in how it runs its auctions. Methodologically, we discuss identification of our treatment effects using causal graphs and show how these treatment effects identify a three-type structural model of bidder behavior with learning dynamics.