Holger Herz
holger.herz@unifr.ch
+41 26 300 8377
https://orcid.org/0000-0003-3495-1640
-
Ordentliche_r Professor_in,
Departement für Volkswirtschaftslehre
PER 21 bu. G428
Bd de Pérolles 90
1700 Fribourg
Forschung und Publikationen
-
Publikationen
15 Publikationen
Control aversion in hierarchies
De Chiara, Alessandro and Engl, Florian and Herz, Holger and Manna, Ester, The Journal of Law, Economics, and Organization (2025) | Artikel -
Forschungsprojekte
Incentive Contracts with Nonstandard Beliefs and Errors in Statistical Reasoning
Status: LaufendBeginn 01.09.2022 Ende 31.08.2026 Finanzierung SNF Projektblatt öffnen Evidence from psychology and behavioral economics suggests that individuals suffer from various errors in statistical reasoning that systematically bias their beliefs and perceptions. The economic literature on incentive provision, however, is still mostly based on models with rational agents and standard preferences. This is surprising given that beliefs and perceptions are crucial determinants of behavior, and may thus have important implications in this domain: (1) Individuals may systematically base decisions on biased perceptions and beliefs and (2) adopt their behavior if they anticipate biased perceptions and beliefs of others. Some researchers have started to incorporate nonstandard preferences into incentive theory, but nonstandard beliefs and errors in statistical reasoning have received almost no attention. I present four projects that aim to demonstrate their relevance for the field. The first project focuses on biases in the risk perception of explicit performance contracts. The informativeness principle (Holmström, 1979) states that if an agent is risk averse, the optimal incentive contract should include any measure (with optimal weighting) that is correlated with exogenous factors affecting the agent’s performance measure, because doing so reduces the agent’s risk exposure. However, research on errors in statistical reasoning suggests that individuals may fail to appropriately account for correlation, potentially counteracting the fundamental logic of the principle. In fact, it is possible that agents who neglect correlation may perceive such contracts as more risky and may thus be averse to such contracts, reverting standard predictions. The other three projects focus on subjective performance evaluations. Many applications in organi- zational and personnel economics focus on situations in which a principal infers agent performance (or ability) from a noisy, non-verifiable performance measure (for example for promotions or the distribution of bonuses). This literature has taken rational inference for granted. I propose to study how errors in statistical reasoning and biased belief formation can systematically affect a principal’s inference in such settings. The second project again focuses on correlation neglect. I consider the case in which agents compete for promotions and a principal wants to promote the most able or hard-working agent. She thus needs to filter out stochastic factors that affect each agent’s performance measure (such as project specific shocks). I propose that correlation neglect can induce an evaluation bias in such settings. The third project studies the impact of motivated beliefs, which is a desire to hold specific beliefs about oneself. Here, I focus on team production tasks and want to study whether motivated beliefs lead to self-attribution bias of team leaders, which in turn affects performance evaluations of team members. Finally, in the fourth project, I want to study the impact of information projection error. If principals have superior information, they may be unable to put themselves into the shoes of less-informed agents, which can downward bias their assessment of an agent’s ability. In all three projects, I propose to not only study the principal’s evaluation bias but also its anticipation by agents and the potentially resulting incentive distortions. The overall aim of this project is to incorporate insights from behavioral economics research on beliefs and errors in statistical reasoning in commonly used principal-agent frameworks. To test the resulting behavioral predictions, I intend to run controlled experiments. I believe the results of the project could have profound impact on principal-agent theory as well as important implications for the design of incentive systems. Behavioral Foundations of Power and Control
Status: Abgeschlossen