The Unassuming Economist

From a paper by Nora Lustig and Andrea Vigorito:

“Inequality measures based on household surveys may be biased because they typically fail to capture incomes of the wealthy properly. The “missing rich” problem stems from several factors, including sampling errors, item and unit nonresponse, underreporting of income, and data preprocessing techniques like top coding. This paper presents and compares prominent correction approaches to address issues concerning the upper tail of the income distribution in household surveys. Correction approaches are classified based on the data source, distinguishing between those that rely solely on within-survey information and those that combine household survey data with external sources. We categorize the correction methods into three types: replacing, reweighting, and combining reweighting and replacing. We identify twenty-two different approaches that have been applied in practice. We show that both levels and trends can be quite sensitive to the approach and provide broad guidelines on choosing a suitable correction approach.”