Emprical Relevance of Ambiguity in First Price Auction Models

We study the identification and estimation of first-price auction models with independent private values where bidders are risk averse and there is ambiguity about the valuation distribution. When bidders' preferences are represented by the maxmin expected utility of [Gilboa and Schmeidler, 1989], we provide sufficient conditions for nonparametric identification of the valuation distribution and bidders' attitude toward ambiguity, separately from the risk aversion (CRRA, CARA). We propose a semi-parametric method and apply it to two datasets, one from experimental auctions and the other from USFS timber auctions. We find, for both cases, that bidders are not only risk averse but also ambiguity averse. In addition, we consider the multiplier preferences of [Hansen and Sargent, 2001] and identify the valuation distribution using the same conditions, and show that normalizing, additionally, (any) one quantile of the value, e.g. upper bound of the support, is sufficient to identify the ambiguity parameter separately from the nonparametric utility.