Mechanism Design under Strategy Advice and Sub-Optimal Play: A School Choice Experiment (with Andrew Dustan, Martin van der Linden, & Myrna Wooders)
This paper presents a laboratory experiment evaluating the effects of strategy advice in the two most popular school choice mechanisms. Our analysis is two-fold. First, we measure the effect of strategy advice on player's strategic decisions. In the strategy-proof Deferred Acceptance (DA) mechanism, strategy advice prompts more players to choose the dominant strategy. Likewise, in the Immediate Acceptance (IA) mechanism, strategy advice to implement one of two heuristic strategies that are widely recommended in the field induces players to choose one of those strategies. The second part of our analysis considers mechanisms performance. Strategy advice steers players away from sub-optimal strategies in DA, but has the opposite effect in IA. We exploit this variance in sub-optimal behavior to evaluate mechanism performance under sub-optimal strategies. Standard efficiency and stability tests and that DA consistently outperforms IA. We, then, consider the welfare of the individual players who choose sub-optimal strategies. To do so, we develop a new typology of DA strategies that allows us to partially order sub-optimal strategies. This analysis confirms that DA outperforms IA under sub-optimal strategic behavior, even when players choose strategies "far" from the dominant strategy.
Second-Order Beliefs and Gender in the Lab (with Andrew Dustan & Greg Leo)
One plausibly important, but relatively ignored, mechanism for explaining behavior in a variety of economic settings are second-order beliefs— beliefs about other people's beliefs. Second-order beliefs about gender-specific distributions, in particular, may be important to understanding the mechanisms underlying gender differences in outcomes. In this paper, we develop a methodology to elicit second-order beliefs about the differences between two populations, then apply it to elicit beliefs about the differences between men and women in a lab experiment. We elicit both first- and second-order beliefs about the performance of men and women on a timed math task and an abstract bargaining task. We find that men and women both believe that men score higher on the math task. Participants' second-order beliefs are distorted— both men and women believe that men's beliefs favor men more than they actually do and believe that women's beliefs favor men less than they actually do. First-order beliefs about bargaining also favor men choosing higher minimum acceptable offers in the Ultimatum game, but second-order beliefs are distorted in the opposite direction. Our methodology can be easily adapted to measure second-order beliefs about the differences between any two populations in any measurable personal characteristic.
Worker Beliefs and the Job Application Decision: A Large-Scale Lab-in-the-Field Experiment