We introduce a flexible framework to study probabilistic choice that accommodates heterogeneous types and bounded rationality. We provide a novel progressive structure for the heterogeneous types to capture heterogeneity due to varying levels of a behavioral trait. Given an order of alternatives, our progressive structure sorts the types by the extent to which they align with this order. Unlike the random-utility model, our model uniquely identifies the heterogeneity, allowing policy makers to perform an improved welfare analysis. As a showcase, we provide characterization of a well-studied type of bounded rationality: “less-is-more.” In addition, we provide conditions for unique identification of the underlying order for the less-is-more structure.

This paper demonstrates theoretically and experimentally that in first-price auctions overbidding with respect to the risk neutral Nash equilibrium might be driven from anticipated loser regret (felt when bidders lose at an affordable price). Different information structures are created to elicit regret: bidders know they will learn the winning bid if they lose (loser regret condition); or the second-highest bid if they win (winner regret condition); or they will receive no feedback regarding the other bids. Bidders in loser regret condition anticipated regret and significantly overbid. However, bidders in the winner regret condition did not anticipate regret.