Comprehensive data on consumption expenditures have historically not been collected in US longitudinal household surveys. The Panel Study of Income Dynamics (PSID) expanded its expenditure data collection in 1999 and 2005. We examine these new expenditure data, highlighting several unique features of the PSID data. We then compare the PSID expenditure data with those in the Consumer Expenditure Survey (CE). We document that the PSID data cover nearly the entire scope of the CE data, and the mean statistics of total expenditures compare favorably between the two surveys. However, significant differences remain for certain expenditure categories.

Using data recently collected by the Panel Study of Income Dynamics, we find that the intergenerational correlation in expenditures is no larger than that in income, suggesting limited intra-family risk-sharing. On the other hand, even after controlling for the intergenerational correlation in income, the expenditures correlation remains significant. This suggests that other factors such as preferences, access to credit, and non-pecuniary inter vivos transfers potentially played a role in consumption smoothing across generations within a family. We also find that the correlation coefficients estimated using food and imputed total expenditures are smaller than that estimated using the measured total expenditures.