Incentive-driven Inattention

This paper establishes a link between incentives and limited attention in economic decision-making. We explore how agents' attention reacts to incentives versus the arrival of information and how each factor affects the quality of decisions. We analyze a unique survey dataset where professional forecasters decide when to update a forecast and there is a formal incentive in the form of a contest rewarding forecast accuracy. There is also a major piece of information arriving right after the contest. We empirically establish that the contest is the primary driver of updating decisions and accuracy improvements. Then, we develop and structurally estimate a rational inattention model where agents choose how much attention to allocate to updating. The estimated model fits the data and allows us to perform counterfactuals to quantify the value of the contest and how it affects updates and accuracy, as well as to establish the optimal timing of the contest.