The quantitative effects of tax foresight: Not all states are equal

Abstract

This paper explores the effect of federal tax news on state economic activity. We estimate a factor-augmented vector autoregression (FAVAR) model, which allows us to consider the possibility that unobserved factors – such as credit and fiscal conditions – might be relevant for modeling the dynamic response of aggregate and state-level economic activity. We identify tax foresight as a shock to the implicit tax rate, measured by the yield spread between the one-year tax-exempt municipal bond and the one-year taxable Treasury bond. Our results suggest that an increase in the implicit tax rate raises national output over much of the anticipation period. In addition, anticipated tax increases give rise to expansions in state personal income and employment. We find that the variation in the responsiveness of economic activity across states is mostly explained by differences in industrial composition and income distribution, as well as by some demographic characteristics such as median income and education. Finally, using a proxy for exogenous changes in federal tax revenues, we investigate the dynamics of state-level personal income and employment. Our results point to considerable heterogeneity in the response across U.S. states. Moreover, they reveal that the long-run multiplier for an anticipated increase in tax revenues is about a tenth of the short-run multiplier for an unanticipated increase in taxes.