Expectation Formation in Step-Level Public Good Games

This paper focuses on the process of individual expectation formation. Specifically, the question is addressed whether individuals think strategically when they form their beliefs about other players' behavior. Most belief learning models assume that people abstract from strategic considerations. Using an incentive compatible mechanism, experimental data are obtained on subjects' expectations in a step-level public good game and in a game against nature. The evidence is consistent with predictions derived from a general representative of belief learning models: the naive Bayesian model. By modeling the updating process used by experimental subjects, we are also able to study the structure of the errors they make and the rate of updating. We find no evidence of primacy or recency in our data but do observe significant differences across subjects in the error structure.