Information and Risk Preferences: The Case of Insurance Choice

Despite standard economic models that assume stability of preferences, empirical work on insurance choice has documented a surprising lack of stability of risk preferences across domains. We hypothesize that informational differences can partly explain the observed empirical regularities, and test our hypothesis by conducting lab and field experiments with over 4,000 participants. We exogenously vary the information subjects receive about the underlying risk they face and elicit willingness to pay for insurance. We have three major findings: 1) information plays a significant role willingness to pay for insurance is higher in settings with more complex or ambiguous information; 2) both risk aversion and aversion to complex or ambiguous information are decreasing in the underlying risk probability; and 3) risk aversion and aversion to complex or ambiguous information are negatively correlated, with the correlation coefficient being remarkably invariant to underlying risk and sociodemographic characteristics. Using demand simulation techniques, we illustrate how informational effects can bias the empirical estimation of risk preferences and have a large impact on the allocative efficiency of insurance markets. JEL classification: D12, D14, D81, G22, J33