Drilled by Overconfidence?

Asymmetric Beliefs, Moral Hazard, and Catastrophic Environmental Risk

Rik Chakraborti and Jason F. Shogren

Abstract: In principal-agent models, the two players can look at the same effort level and still hold asymmetric beliefs about the likelihood of a good and bad state of nature being realized. They can “agree to disagree,” a classic idea that behavioral economists have stressed should be included more frequently in optimal contracting models. This paper examines how agreeing to disagree about the likelihood of an environmental risk generated by a project (e.g., a hydropower dam) can affect the design of the optimal contract given the agent can invest in precautionary care. We focus on the case of the overconfident agent—one who believes either that his precautionary care is extra effective at reducing risk, or that his lack of care poses minimal risk of an environmental catastrophe.

Extending Laffont’s (J. Pub Econ, 1995) classic model, we show that the agent’s overconfidence about the environmental risk matters in two ways. First, if the agent overestimates the effort-contingent likelihood of a good state—the likelihood the dam does not collapse when he exercises full precautionary care—then the regulator can exploit this feature. She writes a contract that generates a positive social surplus by reducing risk at a lower social cost. Second, if the agent overestimates the baseline likelihood of the good state—the chance the dam does not collapse when she exercises no precautionary care—the contract is now less efficient.  Since the agent overestimates the chance things will be good even if no care is taken, his marginal return from costly care is lower. The regulator now has to write a contract to make the effort-contingent desired state even more desirable. This can require a high payoff that extracts all the social surplus, or makes the project infeasible, or both.