Registers or surveys – Does the type of microdata matter for empirical analyses of consumption behavior?

Abstract

In recent years it has come into focus whether longitudinal microdata on consumption derived from administrative registers can constitute an attractive supplement to survey data. This paper explores the consistency between register-imputed and survey-based consumption figures at the household level for Denmark over the period 2002–15. Moreover, it presents estimated consumer demand functions based on the two types of microdata for the same households. The paper finds no significant differences between the marginal propensities to consume out of income estimated on the basis of the two data sources. Furthermore, it demonstrates a close match between total private consumption in the national-accounts statistics and the register-based consumption microdata aggregated over all households.

Monetary Policy and Inequality

“…There are two reasons we do not adopt this measure. First, a recent paper shows that the imputation procedure may be associated with significant measurement error at the tails (Abildgren et al (2018)). This is particularly problematic for studies with a distributional perspective like ours.…”