Markus H. Hahn

A major literature using tax data measures the share of income captured by the top of the income distribution. We correct existing Australian estimates by removing employers’ social contributions from the denominator and explain the limitations of using public record tax tables to capture the numerator. We show that Australian Tax Office unit record sample data are only able to accurately measure incomes of top income groups below the top 1 per cent. We conclude that greater access to the entire unit tax record population will be necessary to bring Australian research in this area up to that in the United States and United Kingdom.

We use Cross-National Equivalent File (CNEF) data from the United States and Great Britain to investigate the association between adults’ health and the income inequality they experienced as children up to 80 years earlier. Our inequality data track shares of national income held by top income percentiles from the early 20th century. We average those data over the same early-life years and merge them to CNEF data from both countries that measure self-reported health of individuals between 1991 and 2007. Observationally, adult men and women in the United States and Great Britain less often report being in better health if inequality was higher in their first five years of life. Although the trend in inequality is similar in both countries over the past century, the empirical association between health and inequality in the United States differs substantially from the estimated relationship in Great Britain. When we control for demographic characteristics, measures of permanent income, and early-life socio-economic status, the health–inequality association remains robust only in the U.S. sample. For the British sample, the added controls drive the coefficient on inequality toward zero and statistical insignificance.

In this article, a general Blinder–Oaxaca decomposition for nonlinear models is derived, which allows the difference in an outcome variable between two groups to be decomposed into several components. We show how, using nldecompose, this general decomposition can be applied to different models with discrete and limited dependent variables. We further demonstrate how the standard errors of the estimated components can be calculated by using Stata's bootstrap command as a prefix.