Anytime individuals receive feedback about how they perform, organizations also choose what to reveal about their peers' performance. I show that this choice changes how much effort individuals exert in anticipation of the feedback, even absent material consequences. In particular, comparing optimal effort across three standard feedback designs isolates two motives: a competitive motive to outperform peers, and an information motive to shape what individuals infer about their own relative ability. I implement this framework in a laboratory experiment where I elicit each participant's effort choice under the three designs to identify the motives at the individual level. I find that both motives are present on average and their magnitude varies across participants. In turn, this heterogeneity reorders how individuals' performances compare. Feedback design is thus a cheap motivator for organizations but one that also changes the identity of the best performers.
Presented at: 2026 - PhD Workshop (Naples School of Economics); BID/BEE Seminar (Toulouse School of Economics); ESA (Universitat Pompeu Fabra); EEA (University College Dublin); ASFEE (Lille); Enter Jamboree (Universidad Carlos III de Madrid); FUR (University of Alicante); Job Market Workshop (Toulouse School of Economics); Workshop on Social Economy for Young Economists (University of Bologna); BEERS Seminar (GATE-LAB); 2023-2025 - BID/BEE Workshop (Toulouse School of Economics); PhD Workshop (Toulouse School of Economics).
Whenever individuals evaluate an outcome against a reference point formed in the past, they must remember that reference point. We show that this memory is imperfect and shapes how the outcome is evaluated. We survey high school students three times: before their final exams, and six and eighteen months after they receive their grade. In the first survey, we elicit the average grade they are aiming for and establish that this goal acts as a reference point. In the other two, we ask them to recall the goal they stated and collect official transcript data. We find that the remembered goal causally shapes satisfaction. In line with our theoretical model, recalling a less ambitious goal raises satisfaction, which creates an incentive to manipulate memory in that direction. However, students do not seem to act on this incentive: recall shows none of the signs of motivated memory that the theory derives, and is instead pulled toward the grade received.
(Third survey is taking place on December 2026.)
Individuals often invoke excuses to justify selfish behavior. We study how the existence of real excuses shapes behavior when agents care about their reputation. Two types of agents, honests and liars, repeatedly choose between a prosocial and a selfish action. In each period, an unobserved excuse may prevent an agent from acting prosocially. Otherwise, honests act prosocially, whereas liars weigh the gain from selfishness against its reputational cost. Without excuses, a single selfish action reveals a liar, so liars with strong enough reputational concerns always act prosocially and their type remains hidden. With excuses, liars act selfishly more often since the reputational cost is lower. Because they do so more frequently than honests, both types are eventually revealed. Excuses thus make each action less informative but behavior over time more informative. This creates a trade-off for a designer who controls the likelihood of excuses and values both prosocial behavior and learning agents' types.