On income tax functions: an application of robust, regression-based diagnostics to models of conditional means

This paper implements a variety of robust, regression-based diagnostics to nonlinear models of effective federal individual income tax. Estimates of effective marginal tax rates obtained from these tax models are frequently used in the empirical growth and real business cycle literature to quantify the macroeconomic effects of distortionary taxation. Based on a battery of conditional mean tests, the paper concludes that none of the tax models used in the empirical growth and real business cycle literature provides a good basis for obtaining reliable estimates of the effective marginal tax rates. Therefore, conclusions based on such tax estimates must be treated with caution.