Research

Publications

 

    Working Papers


        Abstract. This paper introduces a theory of signal perception to study how people update their beliefs. By allowing perceived signals to deviate from actual signals, we identify the probability that people miss or misread signals, giving indices of conservatism and confirmatory bias. In an experiment, we elicited perceived signals from choices and obtained a structural estimation of the indices. The subjects were conservative and acted as if they missed 65% of the signals they received. Also they exhibited confirmatory bias by misreading 17% of the signals contradicting their prior beliefs.


                 

Abstract. This paper investigates how ambiguity attitudes affect people's willingness to enter a gender incongruent field dominated by the opposite gender. In an online experiment, subjects performed both a male-typical task and a female-typical task. They had to choose one of the two tasks to compete with other subjects and win a prize if they were among the top 50% performers. We elicited subjects' ambiguity attitudes concerning their ranking in the two tasks. We found that 49% of men, as compared with 33% of women, chose to compete in the gender incongruent task. Both women and men were more ambiguity averse towards their ranking on the gender incongruent task. Controlling for actual performance, this extra ambiguity aversion made people less willing to compete in  gender incongruent tasks. This suggests that ambiguity is an important factor contributing  to the underrepresentation of one gender in professions populated by the opposite gender (e.g., women in STEM fields). Our findings opened up an additional channel, ambiguity attitudes, where policymakers may tap in to reduce the ongoing gender inequality.


                 

Abstract. This paper proposes a unified framework for optimization over two or more components, e.g., time and risk. Using a century-old theorem on macro-micro aggregation we show that many existing debates in the literature, including incentive compatibility of the random incentive system, hedging in the Anscombe-Aumann framework for ambiguity, equity in Harsanyi’s veil of ignorance, and multiattribute risk aversion, all concern the same bifurcation question “row-first or column-first aggregation?”. For decisions with a single component, behavioral models typically relax (strong) separability assumptions while preserving monotonicity. This paper shows that when two or more components are involved, relaxing separability while maintaining monotonicity is no longer possible. Relaxing separability then comes at the cost of giving up monotonicity for at least one of the components. The question of which monotonocity to give up, is equivalent to the (“bifurcation”) question of which component to aggregate over first. Our analysis identifies the common piece underlying many existing puzzles, and sheds new lights on many ongoing debates. We further provide diagnoses and techniques for overcoming undesirable restrictions. A mathematical online appendix shows how our framework can be used theoretically to generalize many preference axiomatizations.


                

Abstract. This paper introduces source theory, a descriptive theory of decision under ambiguity that models ambiguity attitudes through probability weighting functions that depend on the source of uncertainty. It gives a theoretical foundation for the empirically popular source method. Source theory tractably explains the main empirical findings on decision under ambiguity [HB1] without resorting to expected utility for risk or multistage optimization. We characterize ambiguity attitudes through Arrow-Pratt-like transformations of the source-dependent probability weighting functions. A mathematical novelty is that these transformations occur “within” rather than “outside” the functions.



Work in Progress