Predicting tax rate changes: Insights from the permanent income hypothesis

A large literature demonstrates that changes in aggregate tax rates are basically unpredictable. These previous results typically assume that the alternative to a random walk in tax rates is either serially correlated tax rate changes, or tax rate changes closely tied to lagged government spending. We show that the permanent income hypothesis implies that consumption movements should predict income tax changes if the changes are anticipated, regardless of the reason for the anticipated change. The evidence for a range of countries supports this implication, thus demonstrating that the random walk characterization of aggregate tax rates found previously sterns entirely from the limited set of alternative hypotheses considered.