Correcting for the Missing Rich: Imputing Missing Incomes into Household Data Proportional to Wealth Distribution

We compare alternative imputation methods to correct for missing incomes of the rich in inequality estimates for countries where official tax statistics are not available or reliable. First, we demonstrate that correcting household survey data by imputing missing incomes proportional to wealth distribution is both (i) economically intuitive and macroeconomically consistent, and (ii) yields results close to benchmark studies for Turkey for 2019. We follow Alvaredo et al. (2019) and use the average of the US, China, and France’s wealth distribution (from WID) for countries where wealth distribution data is missing. Second, we validate our method with summary statistics data for Brazil and Chile from LIS. We compare our estimates for corrected Gini for Brazil and Chile with De Rosa et al. (2024), which employs official tax statistics. Our estimates approximate De Rosa et al. (2024) findings both in magnitude of correction and in trends over time. Third, we estimate corrected Gini coefficients for developing countries where no reliable tax data is available and show that official Gini estimates under-estimate inequality by more than 10 percentage points in most places and times. To conclude, we propose a method to correct inequality estimates for missing incomes of the rich that requires only publicly available summary data from LIS (for income distribution), WID (for wealth distribution), and World Bank WDI for GDP and its expenditure components, and no specialized software.