I study the forces that shape economic behavior—from individual beliefs to workplace inequality.

Abstract

We study in the lab how choosing one of two products affects beliefs about their qualities. In a between-subject design, we randomize both ownership mode—choice versus assignment—and whether a product appears favorable by pairing it with an objectively better or worse alternative. This allows us to estimate the causal effect of choice on the perceived quality gap while accounting for contrast effects. We find that choosing magnifies the belief gap between the owned and unowned alternative, driven almost entirely by newly formed pessimistic beliefs about rejected options. Prompting participants to focus on product qualities before choosing attenuates the effect, suggesting an important role for attention. The results help explain reduced switching and markups among near-identical products, and support active-choice policies over opt-out defaults.

Abstract

Overconfidence can arise—or be exacerbated—because of a decision problem itself. In organizational settings, delegation requires decision makers to choose whether an outcome depends on their own performance or another person’s. We test whether this self-reliance dilemma induces motivated reasoning about relative performance. Participants perform a real-effort task and later decide whether to be paid based on their own or another participant’s performance. We vary whether they learn about this decision before or after reporting beliefs. Learning about the dilemma first leads participants to report a smaller performance advantage for the counterpart and to believe incorrectly more often that they outperformed them. These belief distortions arise even though delegation itself is unaffected.

Abstract

After purchasing a product, people receive new information about both chosen and rejected alternatives. We study whether choosing changes how this later information is processed. Participants learn about the fundamental quality of financial investments through repeated price changes; we compare people who chose investments themselves with people who received them exogenously. Learning is stickier after choice: participants react less to both good and bad news about owned and non-owned investments. The effect is not explained by greater attention to chosen investments. A structural model shows that learning after allocation is close to the Bayesian benchmark, while learning after choice is too sticky.

Abstract

I investigate whether people distort beliefs about third parties—such as the ability of scientists to offset one’s environmental impact—to excuse self-interested behavior. In a laboratory experiment, participants choose how much money to take. Whether this money is taken from passive participants depends on a third party’s success in solving a riddle. Treatments differ only in whether success or failure makes the choice harmful to others. Beliefs about third-party success are 13 percentage points higher when failure makes the choice harmful. With monetary incentives for correct beliefs, the gap falls to 6 percentage points and is statistically insignificant. The results suggest that people use beliefs about third-party success as excuses.

Project summary

This methodological experiment tests whether repeated redistribution decisions within a spectator task create carry-over effects: whether exposure to merit- or luck-based inequality in an earlier decision influences redistribution choices in a later one.

Project summary

This project studies how people’s internal representations of uncertain states shape which information they acquire, particularly when some interpretations are more psychologically attractive than others.

Project summary

Choices are informative about moral character, but people interpret them within narratives about others’ conduct. We model agents who care about self-image and can choose beliefs about others within the limits of plausibility. In a hypothesis-testing equilibrium, agents can hold self-serving beliefs when those beliefs remain rationalizable and sufficiently plausible. Limited information can therefore let people excuse selfish behavior or acclaim moral behavior, while full information restores accurate beliefs. The model also shows that greater visibility can reduce aggregate morality when saints are praised, but improve it when sinners are punished.

Project summary

We study sequential public voting when group members care about appearing similar to others. In a three-member binary decision, one member may vote against her private preference to manage her social image. When population types are close to evenly split, this member is more likely to vote against her type when voting second than first. The ranking can reverse when one type is overwhelmingly common. These image concerns can make majority rule fail to select the option preferred by a majority and can make simultaneous voting worse than sequential voting.

Project summary

Using Hungarian linked employer–employee panel data and an event-study design combined with matching, we examine how managers’ illness episodes affect their employee pool. Employee separation rises by 8% after a manager’s health shock, driven mainly by dismissals rather than voluntary departures. Because the total number of employees does not change significantly, the pattern points to restructuring rather than downsizing. The findings reveal a consequential channel through which managers’ personal experiences can reshape workplace outcomes.

Project note

This registered school-based randomized controlled trial investigates whether a role-model intervention can affect students’ aspirations, motivation, and education choices. The draft is available on request.