Income measurement and household consumption elasticities: evidence from dual income reports in the CFPS

Using 2014–2022 CFPS data, we study whether the income report used for identification changes estimated household consumption–income elasticities. In a selected urban working-class spouse-report sample, instrumenting self-reported wage-income growth with household-reported total-income growth raises the elasticity from 0.157 to 0.564 while holding the sample, outcome, controls, and specification fixed. This within-sample sensitivity is informative because it shows that income measure may change both the magnitude of a micro elasticity and the income variation underlying the estimate, even without changing the empirical design. The OLS – IV difference may reflect reporting error, income-component reweighting, or both. A persistent-income reweighting interpretation requires consumption to respond at least as strongly to persistent as to transitory income changes and the household report to place less weight on transitory variation. The response ordering is unverified; smoother household-report dynamics support, but do not establish, the loading condition. Residual non-labour income and the unresolved shared-financial-respondent channel may bias IV upward. We interpret 0.564 as an upper benchmark for common-report income variation in this restricted sample, not as a population or persistent-income elasticity.