Information Projection: Model and Applications
31 Pages Posted: 25 May 2009 Last revised: 10 Aug 2012
Date Written: May 1, 2008
People greatly exaggerate the extent to which their information is shared with others. I present a general model of such information projection, and apply it to a variety of settings. When assessing an expert's competence using ex-post information, jurors overweigh how much they learn from failed and underweigh how much they learn from successful predictions, and they underestimate the competence of experts on average. In turn, even risk-neutral experts are too reluctant to base predictions on ex-ante information that complements, and too eager to base predictions on ex-ante information that substitutes, for what jurors independently learn ex-post. Greater scrutiny decreases the reputation of physicians and the productivity of the medicine practiced at the same time. An expert's fear of information projection reduces her incentives to exercise care, and negligence schemes that tie her compensation too closely to effort often backfire. Optimal monitoring is coarser and career incentives are typically weaker than under fully Bayesian assumptions. Communication protocols that nudge experts to talk, but restrict the use of messages that complement the speaker's expertise, reduce favoritism and strictly improve welfare.
Keywords: hindsight bias, favoritism, medical malpractice, communication
JEL Classification: D8, C9, D2
Suggested Citation: Suggested Citation