Who Lives Where? Household Composition and Income Statistics

We examine how alternative household definitions and equivalence scales affect aggregate measured average income, inequality and poverty. Using Icelandic data, we compare the household compositions that respondents report in surveys to those that appear in administrative register data and find that the two do not always match. These differences have a substantial effect on household income, which comes from individual tax-records, and the subsequent aggregate income statistics. Using register instead of survey information on household composition results in smaller households, which in turn produces lower figures for household income, both in raw data and when equivalised. This also produces higher figures for the Gini coefficient and the incidence of poverty within the sample. In a separate comparison, holding the household definition fixed and instead comparing the weighted survey sample to the full population, both inequality and poverty are lower in the population. These results highlight the importance of ensuring the comparable measurement of household composition in income analyses, and underscore the need for surveys to represent the entire population in order to produce reliable and comprehensive income statistics.