Contract Theory, Economics of Information, Strategic Experimentation, Innovation, Education.
Control, Cost, and Confidence: Explaining Perseverance in the Face of Failure (with Inga Deimen), Games and Economic Behaviour 2022, vol.134, pp. 52 - 74. (Accepted version)
We study effort provision and the development of the belief that effort matters over time: a student is uncertain whether she has control over success through her effort or whether success is determined by her innate ability, which she also does not know. In each period, what she can learn about her control and her ability depends on the level of effort she exerts. The student's optimal effort policy in this two-dimensional bandit problem takes the form of a linear belief cutoff rule and typically features repeated switching of the effort level. Moreover, we define perseverance and procrastination as indices for the student's behavior over time and analyze how they are affected by control, cost, and confidence. Finally, we relate our results to findings in educational psychology and discuss policies to foster perseverance and to lower procrastination.
Optimal Delegated Search with Asymmetric Information: Screening without Transfers (with Elnaz Bajoori)
We study delegated search when a principal relies on an informed agent, effort is observable, but outcome-based transfers are infeasible. Motivated by procurement, we show that all incentive-compatible mechanisms reduce to simple threshold rules. We characterize the optimal pooling rule, where thresholds rise with failed searches as the principal updates beliefs, and the optimal separating menu, which elicits the agent’s private information through minimum quote requirements. Separating mechanisms always dominate pooling and provide a novel rationale for widely used procurement rules mandating multiple quotations.
More effort or better technologies? On the effect of relative performance feedback (with Gwen-Jiro Clochard and Guillaume Hollard)
Providing rankings - in addition to individual performance only - has been proved a powerful tool to enhance performance. However, the exact channel through which ranking feedback affects performance is still a matter of debate. A common practice is to add an ad hoc term to the utility function when rankings are made available. We here propose an alternative approach where additional information provided by rankings increases performance by allowing individuals to identify better technologies. Our model captures several features not encompassed in previous models. In particular, it accounts for (1) the persistence of the feedback effect in multiple rounds tournaments, (2) performance improvements all along the performance distribution, including the worse performers. We bring our model to data using a field experiment consisting of a tournament based on a series of four math tests taken by high school girls.
Feedback and Learning in Tournaments
We study the optimal feedback policy in a dynamic tournament with strategic experimentation. In many situations performance feedback is needed for agents to evaluate the quality of their work and adjust their strategy accordingly. Feedback makes it possible to learn about the productivity of a technology or the potential of a project and to discard one in favour of a new one after negative feedback. However, in a tournament setting, such as when agents compete for a promotion, precise feedback does not implement the output-maximising technology choice. The leader of the tournament will hold on to a subpar technology if she has a sufficient lead while the follower will discard an above-average technology to have a chance to catch up with the leader. A different inefficiency emerges with private feedback, where the competitor's performance is not revealed. Agents discard moderately above-average technologies in order to compete with an opponent who has an increased expected performance due to his strategic switching behaviour. We show that the efficient switching decision can be induced with partial feedback, where the principal restricts herself to giving recommendations to switch or keep the technology when it is efficient to do so. In a setting where agents both choose technology and costly effort partial feedback continues to be the superior policy. Among the considered policies, partial feedback induces the highest level of effort.
The Kindness of Strangers (with Alessandro Iaria)
We study the phenomenon of kindness towards strangers. Seemingly altruistic behaviour can follow from purely selfish motives if agents face risk. In a repeated dictator game setting, a charitable equilibrium can be sustained if dictators have a positive probability to change roles, even with anonymous transactions. This also holds if behaviour cannot be monitored perfectly. The main driving factor for charitable behaviour is the desire to sustain the social norm of kindness, from which the charitable agent herself might benefit in the future. This can be interpreted as an informal insurance arrangement in the absence of enforceable contracts. We continue to examine how cooperation is complicated by inequality, in terms of heterogeneous risk exposure. We study how more persistent differences in risk make cooperation increasingly hard to achieve. We show how heterogeneity can lead to a fragmentation of society, where cooperation is only possible within subgroups, leading to losses in welfare. The model allows for interesting interpretations of social divisions in societies of varying heterogeneity.