We study the effects of preselection on an expert's incentive to give truthful advice. In a decision environment in which certain decisions yield more precise estimates about the expert's expertise, a mediocre expert's advice is biased. We show that this bias can be undone by the introduction of a preselection stage, where the decision maker himself sometimes studies the case, and thereby alters the expert's perception of the problem. We identify a parameter range in which the decision maker's choice is inefficient if it is not possible to commit to a certain preselection level.
D82 - Asymmetric and Private Information ; D83 - Search, Learning, Information and Knowledge ; D72 - Economic Models of Political Processes: Rent-Seeking, Elections, Legistures, and Voting Behavior