I explain puzzles in the school assignment literature using a many-to-one matching model in which participants on one side of the market, the students, are endowed with ego-utilities à la Köszegi (2006). Ego concerns generate a form of information avoidance that results in non-truthful participation in DA matching mechanisms. In particular, students’ best replies may be non-monotonic in school ranks. Furthermore, students may be sensitive to signal garbling, in the sense of Blackwell (1953), even when keeping admission probabilities constant. In terms of policy, the results imply that admissions committees’ reliance on application dimensions that are seemingly weak proxies of academic performance may be beneficial. Other implications relate to the effectiveness of affirmative action policies that directly lower admission requirements: whenever students’ best replies exhibit non-monotonicity in schools’ selectivity, such policies might backfire.

We show experimentally that people undervalue options they find complex. We document this phenomenon for tasks as diverse as belief updating, visual perception, and compound risk. This behavior is incompatible with expected utility, even when accounting for risk aversion and incorrect beliefs; instead, it suggests people dislike the cognitive uncertainty they experience in the face of complexity as a form of ambiguity aversion. The data supports this explanation: our effects vary with both cognitive uncertainty, ambiguity aversion, and their interaction. At a broad level, our results suggest that individual preferences (attitudes toward complexity) matter in cognitive models. At a narrower level, our paper informs the literature on non-Bayesian updating, which overlooks complexity aversion, and the connection between compound lottery and ambiguity aversion, which, we show, holds mostly for subjects who find compound lotteries complex.

I study partisan gerrymandering when district composition affects candidates' policy positions and, consequently, voters' behavior. In the U.S., primary elections determine which candidates compete in general elections, with a district's ideological composition shaping who emerges as the nominee. Thus, redistricting affects not only which party wins but also the ideology of competing candidates. I find that classical gerrymandering strategies can backfire when candidates emerge endogenously, particularly in districts where extreme voters may select non-viable candidates. However, when properly designed to account for both voter affiliation and preference intensity, gerrymandering can be a more powerful instrument than traditional approaches that consider only party affiliation. I show how methods from optimal transport theory can be used to characterize the optimal redistricting plan, which creates districts that maximize ideological distance between competing candidates. Using these findings, I analyze two implications for the U.S. House of Representatives: how gerrymandering contributes to political polarization and its consequences for minority representation.

In this paper, we quantify the distortions induced by career concerns within the Federal Open Market Committee (FOMC). We combine a structural approach with an unanticipated change in the information available to the public about internal committee deliberations. We show that— given the policy preferences of Fed Presidents and Board Governors serving in the FOMC— agents' incentives to appear competent and unbiased outweigh the distortions induced by antipandering and conformity. Relative to a counterfactual with no reputational considerations, career concerns improve the welfare of an unbiased principal. Given our estimates of career concerns, Transparency improves welfare relative to an Opaque regime in which internal deliberations are not made public. In a counterfactual exercise, we show that greater heterogeneity in regional shocks reduces conformity but increases policy errors under Transparency.