I propose a model that captures the evolution of overconfidence as being a result of the constrained utility maximization problem of a decision maker who is rationally inattentive to information, but at the same time biased towards more optimistic subjective beliefs. Empirical studies have shown that individuals with fewer skills have more confidence, but as the skill level increases, overconfidence decreases. The phenomenon is well-known as the Dunning-Kruger effect in the psychology literature. In my model, this effect is explained by the simultaneous choice of objective and subjective information. Beyond the traditional materialistic utility and cost of information, a non-materialistic utility component induced by overly optimistic subjective beliefs is considered, which is constrained by the cost of information distortion. The setup is tractable in a range of economic problems. The first application shows why individuals with fewer initial skills are less motivated to learn than those already having a higher skill level. The second application explains excess business entries and high rates of failures due to overconfident entrepreneurs.
Recent scandals in science have brought attention to the problem of detecting fraud and attributing punishment in the context of research teams. We examine the problem theoretically and consider the socially optimal scheme for assigning culpability. We consider the simplest possible environment with two scientists, only one of whom is capable of committing fraud. Our theoretical analysis shows that a regime of group accountability that incentivises researchers to monitor other members of the group achieves the best social outcomes. Given this regime, the model yields the counter-intuitive prescription that punishing non-culpable members of the team for participating in a fraudulent project is the most promising tool for increasing the fraction of research that is honest.
The prevalence of conspiracy theories is a concern in western countries, yet the phenomenon is rarely addressed in experimental economics. In two preregistered online studies (NStudy 1 = 97, NStudy 2 = 203) we examine the relationship between exposure to conspiracy modes of thinking, self-reported conspiracy mentality, self-reported manipulativeness, and behaviour in an economic game that measures strategic sophistication. Part of our design was based on Balafoutas, Libman, Selamis, and Vollan (2021), who found a positive relationship between exposure to conspiracy modes of thinking and strategic sophistication. Our results did not corroborate their findings in an online setting. Our measures of conspiracy mentality were modestly correlated with strategic sophistication in Study 2, but not in Study 1. Although we expected manipulativeness to be positively associated with both conspiracy mentality and strategic sophistication—thereby linking conspiracy mentality and strategic sophistication indirectly—it was only associated with conspiracy mentality.
This paper investigates the gender differences in two types of overconfidence: overestimation and overplacement. We aggregate findings from 39 experimental studies in economics, published between 2000-2020, all of which employed real-effort tasks, in a preregistered Bayesian hierarchical analysis. A previous meta-study by Bandiera et al. (2022) concluded that both men and women exhibit overconfidence, with no significant gender differences. However, Moore and Healy (2008) define multiple types of overconfidence, which are sometimes inversely related. Following their terminology, we divide our meta-study sample into two categories–overestimation and overplacement–and test whether the original results still hold. In the overall sample, we find evidence of overconfidence among men but not among women; however, the difference in the magnitude of their biases is not statistically significant. In the subsamples, neither men nor women exhibit statistically significant overestimation or overplacement, and no significant gender differences emerge in the magnitude of these biases for either type of overconfidence
This project revisits the model of adverse selection under asymmetric information with the power of the rational inattention framework. I depart from the setup of Akerlof (1970) by revising its extreme information asymmetry assumption. Instead of assuming that the Seller is fully informed and the Buyer is fully uninformed, I consider a setting in which both parties are able to gather information, but at a cost. As a result, both the Seller and the Buyer become partially informed, and the information asymmetry is the consequence of the asymmetry in their incentives and unit information costs. This enhanced framework provides new insights into the implications of incomplete information for market outcomes, efficiency and welfare. When information asymmetries occur endogenously, they do not lead to market collapse, but they do create market inefficiencies. The Buyer is better off and the Seller is worse off compared to the efficient symmetric information benchmark.