No abstract

In a centralized marketplace that was designed to be simple, we identify participants whose choices are dominated. Using administrative data from Hungary, we show that college applicants make obvious mistakes: they forgo the free opportunity to receive a tuition waiver worth thousands of dollars. At least 10 percent of the applicants made such mistakes in 2013. Costly mistakes have externalities: they transfer tuition waivers from high-to low-socioeconomic status students, and increase the number of students attending college. To shed light on the mechanisms underlying mistakes, we exploit a reform that substantially increased the selectivity of admission with financial aid in some fields of study. Increased admission selectivity raises the likelihood of making obvious mistakes, especially among high-socioeconomic status and low-achieving applicants.Our results suggest that mistakes are more common when their expected cost is lower. Still, the average cost of a mistake in 2013 was 114-365 dollars.

Traditionally, strategy-proofness is considered an extreme form of strategic simplicity. When a strategy-proof mechanism is in place, honesty is the best policy: no matter what actions others are taking, the best course of action is to report one's true preferences. For this reason, such mechanisms are often referred to as truthful, and the preferences that are reported to them are interpreted at face value.Truthfulness is thought to reduce the costs of strategizing, promote equity (by not giving advantage to more sophisticated players), provide robustness, eliminate the costs associated with the collection of information on others, and simplify the interpretation of reported preferences. 1 These desirable features led many centralized markets, especially in education and entry-level labor markets, to adopt truthful mechanisms. More specifically, the mechanisms in use are often based on the applicant-proposing version of the Deferred Acceptance (DA) algorithm, which is strategy-proof for applicants.But do truthful mechanisms actually induce truthful reporting? Recent evidence that emerged from the Israeli Psychology Master's Match (IPMM) strongly indicates that the assumption of truthful reporting is false. Additional evidence from the field and from the lab suggests that this phenomenon of preference misrepresentation is pervasive. A recurring finding is that misinterpretation rates are higher in weaker segments of markets. This motivates us to further investigate who are the individuals that misrepresent their preferences under DA and what drives this behavior.Our hope is that our findings will inform market designers and policy makers about the prevalence of misrepresentation and its systematic nature. A better understanding of when and why individuals misrepresent their preferences can guide these practitioners in designing specific and targeted interventions to promote truthful reporting. For example, if members of particular groups are more likely to (err and) misrepresent their preferences, then this behavior may have * Hassidim: Bar-Ilan

Since no stable matching mechanism can induce truth-telling as a dominant strategy for all participants, there is often room in matching markets for strategic misrepresentation (Roth [25]). In this paper we study a natural form of strategic misrepresentation: reporting a truncation of one's true preference list. Roth and Rothblum [28] prove an important but abstract result: in certain symmetric, incomplete information settings, agents on one side of the market ("the women") optimally submit some truncation of their true preference lists. In this paper we put structure on this truncation, both in symmetric and general settings, when agents must submit preference lists to the Men-Proposing Deferred Acceptance Algorithm.We first characterize each woman's truncation payoffs in an incomplete information setting in terms of the distribution of her achievable mates. The optimal degree of truncation can be substantial: we prove that in a uniform setting, the optimal degree of truncation for an individual woman goes to 100% of her list as the market size grows large, when other women are truthful. In this setting, we demonstrate the existence of an equilibrium where all agents use truncation strategies. Compared to truthful reporting, in any equilibrium in truncation strategies, welfare diverges for men and women: women prefer the truncation equilibrium, while men would prefer that participants truthfully report.In a general environment, we show that the less risk averse a player, the greater the degree of her optimal truncation. Finally, when correlation in preferences increases, players should truncate less. While several recent papers have focused on the limits of strategic manipulation, our results serve as a reminder that without the pre-conditions ensuring truthful reporting, even in settings where agents have little information, the potential for manipulation can be significant.

No abstract