This paper presents summary statistics on the occupations of taxpayers in the top percentile of the national income distribution and fractiles thereof, as well as the patterns of real income growth between 1979 and 2005 for top earners in each occupation, based on information reported on U.S. individual income tax returns. The data demonstrate that executives, managers, supervisors, and financial professionals account for about 60 percent of the top 0.1 percent of income earners in recent years, and can account for 70 percent of the increase in the share of national income going to the top 0.1 percent of the income distribution between 1979 and 2005. During 1979-2005 there was substantial heterogeneity in growth rates of income for top earners across occupations, and significant divergence in incomes within occupations among people in the top 1 percent. We consider the implications for various competing explanations for the substantial changes in income inequality that have occurred in the U.S. in recent times. We then use panel data on U.S. tax returns spanning the years 1987 through 2005, to estimate the elasticity of gross income with respect to net-of-tax share (that is, one minus the marginal tax rate). Information on occupation allows us to control for other influences on income in a flexible way using interactions among occupation, position in the income distribution, stock prices, housing prices, and the business cycle. We also allow for income shifting across years in response to anticipated tax changes, for the long-run effect of a tax reform to differ from the short-run effects, for heterogeneous mean-reversion across incomes, and for heterogeneous elasticities across income classes. In a specification that does all this, we estimate a significant elasticity of 0.7 among taxpayers in the top 0.1 percent of the income distribution. Outside of the top 0.1 percent of the income distribution, we find no conclusive evidence of a positive elasticity of income with respect to net-of-tax shares. We find that the estimate for the top 0.1 percent is not robust to controlling for a spline in lagged income that is very flexible at the upper reaches of the income distribution, suggesting that the method used to allow for income dynamics is very important. Allowing for income shifting across years in response to anticipated tax changes has important consequences for the estimates.
