Economic Theory Seminar Series: A Policy for Standards of Evidence

Carlos Oyarzun | University of Queensland

A decision maker can commit to an ex-post inefficient standard of evidence to reduce the incentives of an agent, whose type is bad, to undermine the informativeness of evidence. We fully characterize when the optimal standards of evidence of the decision maker are harsh and soft, i.e., require more and less support, respectively, from the evidence to choose the alternative preferred by the agent, than what would be statistically optimal in absence of concerns about the agent’s incentives. The optimal commitment for the decision maker is not necessarily to make the standard more demanding: standards are typically soft for low prior that the type is bad, and harsh for high priors. Indeed, under mild conditions, in equilibrium with standard commitment, the agent is more likely to persuade the decision maker than in absence of commitment, for a range of prior beliefs. We also characterize when optimal standards are confirmative and conservative (c.f., Li (2001)), i.e., wh! en optimal standards are adjusted to favor or disfavor, respectively, the decision that would be made if only prior information was available, without looking at the evidence. We provide sufficient conditions for uniform confirmativism to be optimal in equilibrium. When both (good and bad) types can make effort to affect the evidence, then, under certain conditions, the decision maker is uniformly conservative in equilibrium.