Individual and household debt: Does imputation choice matter?

Type Working Paper - University of Pretoria Working Papers
Title Individual and household debt: Does imputation choice matter?
Issue 202141
Publication (Day/Month/Year) 2021
In the case of sensitive requests - such as those made of survey respondents to reveal their earnings, their individual assets, debts or even net worth - complete answers are rarely forthcoming. Thus, there are numerous non-responses. We apply a bevy of imputation methods in an attempt to reduce the proportion of missing data on individual and household debt that is present in the National Income Dynamics Study panel data. Our application of Multiple Imputation by Chained Equation (MICE) yields additional observations on these variables than are available in the NIDS imputation data. Although our imputations do alter the distribution of the debt data across the first four waves, especially for individual level debt data, the effect of that alteration, once aggregated to the level of the household, is negligible.

Related studies