Do Managera exploit their Workers' Overconfidence

The behavioral principal-agent literature shows that biased self-beliefs can affect effort and performance, which may benefit principals. Experimental evidence documents systematic confidence distortions in performance settings, suggesting that principals can benefit from these. However, they can only do so if they correctly understand how these distortions affect performance, raising the question of whether principals recognize and exploit them in practice. To study this, we propose an online experiment in which agents perform a real-effort task under either an overconfidence-relevant environment, where the piece rate depends on relative ranking in an IQ task (top/bottom half), or an overconfidence-irrelevant environment, where piece rates depend on a random number draw. Before agents work, principals decide whether to reveal the piece rate and thus whether agents learn their ranking. If principals anticipate overconfidence in the overconfidence-relevant environment, where agents overestimate the likelihood of receiving the high piece rate, they should be less likely to disclose this information. A within-subject design, in which principals make decisions in both environments and also for themselves before performing the task, allows us to test whether disclosure decisions are strategic responses to agents’ biases.