Recent empirical work on captial gains implies that realizations are highly sensitive to marginal tax rates. Because they are based on cross-section data, however, these estimates cannot distinguish between permanent responses to tax rate changes and the timing of realizations to take advantage of the normal fluctuations in any individual's tax rates over time. The purpose of this paper is to distinguish transitory from permanent tax effects by analyzing panel data for taxpayers. Controlling for permanent and transitory income and other variables, the estimates suggest both transitory and permanent effects, although the permanent tax rate effect is not significant in all cases.

Charitable Giving, Income, and Taxes: An Analysis of Panel Data by Gerald E. Auten, Holger Sieg and Charles T. Clotfelter. Published in volume 92, issue 1, pages 371-382 of American Economic Review, March 2002