Consumer Welfare Under Individual Heterogeneity

We propose a nonparametric method for estimating the distribution of consumer welfare from cross-sectional data with no restrictions on individual preferences. First demonstrating that moments of demand identify the curvature of the expenditure function, we use these moments to approximate money-metric welfare measures. Our approach captures both nonhomotheticity and heterogeneity in preferences in the behavioral responses to price changes. We apply our method to US household scanner data to evaluate the impacts of the price shock between December 2020 and 2021 on the cost-of-living index. We document substantial heterogeneity in welfare losses within and across demographic groups. For most groups, a naive measure of consumer welfare would significantly underestimate the welfare loss. By decomposing the behavioral responses into the components arising from nonhomotheticity and heterogeneity in preferences, we find that both factors are essential for accurate welfare measurement, with heterogeneity contributing more substantially.