TESTING THE POSITIVE THEORY OF GOVERNMENT FINANCE

Researchers characterizing optimal tax policies for dynamic economies have reasoned that tax rates should follow a random walk. We conduct a frequency-domain examination of U.S. tax rates and reject the hypothesis that the first difference in the series is white noise. The source of the rejection is pronounced activity of tax changes at an eight-year cycle. Regression analysis confirms the finding that there is a cyclical component to tax changes corresponding to changes in political party administration. The results suggest that the positive theory of government finance needs to be refined to incorporate features of political equilibrium. © 1990.