Sign up for meetings on the sheet below:
docs.google.com/spreadsheets/d/13ovx0deRg8mA3h5XVO95SLGQcZQMTAcbZiTKaGFTKvs/edit#gid=0
If signing up less than two days before the talk, please also email facultyadmin@economics.ox.ac.uk

Abstract:
Experts play a central role in many economic, political, and social interactions, and the ability of experts to sway decisions in their favor is widely documented. A key ingredient of this ability is the difficulty non-expert decision makers have in de-biasing an expert’s advice. For instance, even if you know your mechanic is recommending excessive repairs, how do you know what repair your car really needs? Canonical models of expertise do not capture this property. In canonical representations, expertise consists of a single piece of information and, as a result, advice can be perfectly debiased and experts are unable to sway decisions in their favor. We develop an alternative model of expertise, one in which expert knowledge consists of a continuum of correlated states and in which advice can be de-biased only imperfectly, even when advice is hard information. We show that this richer conception of expertise provides the expert with leverage to sway decisions in her favor. Yet to do so, the expert must communicate in a new form, conveying ostensibly redundant beyond a simple recommendation. We show that this power to sway outcomes benefits the expert for any degree of complexity beyond that assumed in canonical models, although peaking for decisions of moderate complexity.