Sufficient Statistics for Measuring Forward-Looking Welfare

We provide a method to measure welfare, in money-metric terms, taking into account expectations about the future. Our two key assumptions are that (1) the expenditure function is separable between the present and the future, and (2) there are some households that do not face idiosyncratic undiversifiable risk. Our sufficient statistics methodology allows for incomplete markets, lifecycle motives, non-rational expectations, non-exponential time discounting, and arbitrary functional forms. To apply our formulas, we require estimates of the elasticity of intertemporal substitution, goods and services' prices over time, and repeated cross-sectional information on households' income, balance sheets, and expenditures. We illustrate our method using the PSID from the United States. We find that static measures overstate cost-of-living increases for most households, particularly younger and poorer households. Our estimates can be used to study the welfare consequences of dynamic stochastic shocks that affect households along different margins and time horizons. For example, we find that involuntary job loss is associated with a 20% reduction in money-metric utility for households younger than 60 years old