80 likes | 200 Vues
He Who Pays The Piper Must Know The Tune. Robin Hanson George Mason University. The Puzzle of Abstract Experts. Examples: History professor lecture on ancient Rome Newspaper article on event in distant land Doctor says how to avoid rare condition
E N D
He Who Pays The Piper Must Know The Tune Robin Hanson George Mason University
The Puzzle of Abstract Experts • Examples: • History professor lecture on ancient Rome • Newspaper article on event in distant land • Doctor says how to avoid rare condition • We almost never know if they are right when we reward or punish them! • So why would they work to be right?
Previous Research • Heidhues & Lagerlöf 2003 (GEB) • 2 politicians, pick binary policy, get binary clue • 1 voter wants politician so policy = binary truth • Brandenburger & Polak 1996 (RJE) • N managers, pick binary policy, get binary clue • Trader sets stock to prob. policy = binary truth • In both, experts neglect clue for client prior • “neglect” allows partial reveal in mixed str. eq.
Generalizing Those Results • In H&L B&P, if add choice to buy info at tiny cost • They never buy! - Only use info if indifferent on acts • I show that in general, experts never get more info than clients will have when rewarding experts, if • Experts not directly care about client topics or honesty • Info efforts are hidden and costly • Experts can coordinate • On focal info, includes client info, other dim. info is less/more • See previous expert acts, or two player zero sum game
Heidhues & Lagerlöf 2003 (Use some if mix.)
Brandenburger & Polak 1996 (Use some if mix.)
Summary • Previous papers hinted: experts neglect info • Generalize: coordinating experts ignore info • Sharpens puzzle of abstract experts -- why would profs, press, docs work to be right? • Do they directly care? Can we observe effort? • Do we leverage rare cases we can check on? • Why do we let experts coordinate so much? • Do we not really care if experts know more?