Ellsberg Revisited: An Experimental Study


  • Yoram Halevy

    1. Dept. of Economics, University of British Columbia, 997-1873 East Mall, Vancouver, BC V6T 1Z1, Canada; yhalevy@interchange.ubc.ca; http://www.econ.ubc.ca/halevy
    Search for more papers by this author
    • A co-editor and three anonymous referees provided thoughtful comments and suggestions that improved this work substantially. I appreciate helpful discussions with Paul Beaudry, Gorkem Celik, Larry Epstein, Haluk Ergin, Itzhak Gilboa, David Green, Edi Karni, Dan Levin, Sujoy Mukerji, Emre Ozdenoren, Mike Peters, Zvi Safra, and Uzi Segal. Special thanks to Peter Wakker. I also thank workshop participants in Michigan, University of British Columbia (UBC), Hebrew University, Tel Aviv, Haifa, RUD 2003 in Milan, The 2004 Canadian Experimental and Behavioral Economics Workshop in Calgary, The 2005 Annual International Meeting of the ESA in Montréal, and the 2005 Econometric Society World Congress. Rodney Chan, Fernando Har, and Haifang Huang provided superb research assistance. The CASSEL staff, and especially Walter Yuan, provided outstanding support and assistance. Financial support from UBC's Hampton Fund is gratefully acknowledged.


An extension to Ellsberg's experiment demonstrates that attitudes to ambiguity and compound objective lotteries are tightly associated. The sample is decomposed into three main groups: subjective expected utility subjects, who reduce compound objective lotteries and are ambiguity neutral, and two groups that exhibit different forms of association between preferences over compound lotteries and ambiguity, corresponding to alternative theoretical models that account for ambiguity averse or seeking behavior.