How should we measure changes in consumer welfare given observed data on prices and expenditures?
This paper proposes a nonparametric approach that holds under arbitrary preferences that may depend
on observable consumer characteristics, e.g., when expenditure shares vary with income. Using total
expenditures under a constant set of prices as our money-metric for real consumption (welfare), we
derive a principled measure of real consumption growth featuring a correction term relative to
conventional measures. We show that the correction can be nonparametrically estimated with an
algorithm leveraging the observed, cross-sectional relationship between household-level price indices
and household characteristics such as income. We demonstrate the accuracy of our algorithm in
simulations. Applying our approach to data from the United States, we find that the magnitude of the
correction can be large due to the combination of fast growth and lower inflation for income-elastic
products. Setting reference prices in 2019, we find that (i) aggregate real consumption per household in
1955 is underestimated by 11.5% by the uncorrected measure, and (ii) the correction reduces the annual
growth rate from 1955 to 2019 by 18 basis points, which is larger than the well-known "expenditure
switching bias" over the same time horizon.
Xavier Jaravel and Danial Lashkari
30 November 2022 Paper Number POIDWP045.pdf
Download PDF - Measuring growth in consumer welfare with income-dependent preferences
This POID Working Paper is published under the Programme on Innovation and Diffusion.