“Bayesian Identification: A Theory for State-Dependent Utilities” by Dr. Jay Lu
The University of California – Los Angeles
We provide a revealed preference methodology for identifying beliefs that allows for utilities to vary across states. A notion of comparative informativeness is introduced that is weaker than the standard Blackwell ranking. When information is private, beliefs and utilities can be identified using stochastic choices from two treatments where one is strictly more informative than another. When the signal structure is public, stochastic choice from a single treatment can lead to full identification. These results illustrate novel identification methodologies unique to stochastic choice. Applications include identifying discrimination and biases in hiring decisions, medical advice and college admissions.