Vincent P. Crawford: Meaningful Theorems: Nonparametric Analysis of Reference-dependent Preferences

Vincent P. Crawford (University of Oxford and UC San Diego) presenting his talk titled "Meaningful Theorems: Nonparametric Analysis of Reference-dependent Preferences" (abstract below) at the Conference on Mechanism and Institution Design held at Corvinus University of Budapest, Hungary.

The session is dedicated to the Professor's 75th birthday and his fundamental contributions to economic theory, game theory, and the Society for the Promotion of Mechanism and Institution Design.

Abstract: joint paper with Laura Blow
This paper derives nonparametric conditions for the existence of reference-dependent preferences that rationalize a price-taking consumer’s demand behavior. Unless reference points are modelable and sensitivity is constant, reference-dependent models of consumer demand are flexible enough to fit virtually any data. Assuming modelable reference points and constant sensitivity, we characterize continuous reference-dependent preferences, relaxing Kőszegi and Rabin’s (2006; “KR”) strong functional-structure assumptions. We use our characterization to re-analyze Farber’s (2005, 2008) data on cabdrivers’ labor supply. Relaxing KR’s assumptions greatly increases a nonparametric measure of predictive success. For many drivers, a relaxed reference-dependent model has greater success than its neoclassical counterpart.