5/2019 Jaanika Meriküll, Tairi Rõõm. Are survey data underestimating the inequality of wealth?

Working Papers of Eesti Pank 5/2019

This paper uses administrative data from registers and survey data from interviews to analyse unit and item non-response in a wealth survey. It draws on the Estonian Household Finance and Consumption Survey dataset, where the survey data on income and wealth are complemented by information on the same variables from administrative sources for all the people sampled. The results show that the non-response contributes to the underestimation of wealth inequality in the survey data, as the Gini coefficient is underestimated by 6 percentage points and also the top wealth shares are substantially underestimated. The downward bias is originating from item non-response and not from unit non-response. Imputation can address the problems caused by item non-response across most of the net wealth distribution, but does not eliminate the downward bias at the top of the wealth distribution.

JEL Codes: D31 (Personal Income, Wealth, and Their Distributions); E21 (Consumption, Saving, Wealth)

Keywords: wealth distribution, unit non-response, item non-response, participation bias, wealth survey, income survey, Household Finance and Consumption Survey, Estonia

The views expressed are those of the authors and do not necessarily represent the official views of Eesti Pank or the Eurosystem.