Revealed preference analysis with framing effects

In many settings, behavior is observed to vary based on seemingly arbitrary factors. Such framing effects cast doubt on the welfare conclusions from revealed preference analysis. We relax the assumptions underlying revealed preference analysis to recover ordinal preference information from choices when framing effects are present. Plausible restrictions of varying strength permit either partial- or point-identification of preferences for the decision-makers who choose consistently across frames. Recovering population preferences requires understanding the empirical relationship between decision-makers’ preferences and their sensitivity to the frame. We develop tools for studying this relationship and illustrate them with data on automatic pension plan enrollment.