Measuring top incomes using tax record data: a cautionary tale from Australia

Atkinson et al. (J. Econ. Lit. 49(1):3–71, 2011) survey an important new literature using income-tax-based data to measure the share of income held by top income groups. But changes in tax legislation that expand the tax base to include income sources (e.g. capital gains, dividends, etc.) disproportionately held by these groups will conflate such an expansion with an increase in the share of income they hold. We provide a cautionary tale from Australia of how comprehensive tax reform legislation in 1985 substantially altered Australian top income series, especially those that do not separate taxable realized capital gains from other taxable income. Drawing on the Household, Income and Labour Dynamics in Australia (HILDA) Survey we then estimate the size and distribution (across income groups) of taxable realized capital gains in 2006 and 2009, and compare these results with those using accrued capital gains, finding substantially different distributions. More importantly, we find substantial differences across our measures in how capital gains changed between 2006 and 2009. Our results suggest that yearly taxable realized capital gains, often included in studies of top incomes, might be a poor proxy for the theoretically more appropriate yearly accrued capital gains.