Why the Restricted Bayes/Hurwicz Criterion is Savagian: A Game-Theoretic Perspective of Ambiguity

16 Pages Posted: 19 Mar 2012 Last revised: 12 May 2012

See all articles by Ning Neil Yu

Ning Neil Yu

Nanjing Audit University - Institute for Social and Economic Research

Date Written: March 29, 2012

Abstract

Ellsberg (1961) introduced the Ellsberg urns experiments as evidence against Savage's theory of subjective expected utility. The restrictive Bayes/Hurwicz criterion (Ellsberg, 2001) is proposed to explain the phenomena of subjects shunning the ambiguous urns. Modeling the interaction between the experimenter and the subject as a game, we find sufficient and necessary conditions for the subject to modify beliefs with the reduced-form criterion. We postulate that diverse ambiguity attitudes result from different perceptions about the payoff structure. As a horse race gambler should adjust his belief after the gambling company publishes new odds, the subject can form different beliefs facing different choice problems. The revealed belief discrepancies can be a result of these modifications, and thus does not violate the Savage axioms about the preferences at the moment of each decision. The ambiguity game interpretation is supported by various neuroscientific and behavioral studies.

Keywords: Ambiguity Attitude, Restricted Bayes/Hurwicz Criterion, Restricted Hurwicz Criterion, Ambiguity Aversion/Seeking, Ellsberg Paradox, Ellsberg Urns Experiments, Baboon Infanticide Game

JEL Classification: C72, D81, D83

Suggested Citation

Yu, Ning Neil, Why the Restricted Bayes/Hurwicz Criterion is Savagian: A Game-Theoretic Perspective of Ambiguity (March 29, 2012). Available at SSRN: https://ssrn.com/abstract=2025502 or http://dx.doi.org/10.2139/ssrn.2025502

Ning Neil Yu (Contact Author)

Nanjing Audit University - Institute for Social and Economic Research ( email )

Stanford, CA 94305
United States

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
96
Abstract Views
1,498
rank
348,394
PlumX Metrics