Revealed Preferences in a Heterogeneous Population

This paper explores the empirical content of the weak axiom of revealed preference (WARP) for repeated cross-sectional data or for panel data where individuals experience preference shocks. Specifically, in a heterogeneous population, think of the fraction of consumers violating WARP as the parameter of interest. This parameter depends on the joint distribution of choices over different budget sets. Repeated cross-sections do not reveal this distribution but only its marginals. Thus, the parameter is not point identified but can be bounded. We frame this as a copula problem and use copula techniques to analyze it. The bounds, as well as some nonparametric refinements of them, correspond to intuitive behavioral assumptions in the two goods case. With three or more goods, these intuitions break down, and plausible assumptions can have counterintuitive implications. Inference on the bounds is an application of partial identification through moment inequalities. We implement our analysis with the British Family Expenditure Survey (FES) data. Upper bounds are fre- quently positive but lower bounds not significantly so, hence FES data are consistent with WARP in a heterogeneous population.