Abstract
Barro's tax smoothing hypothesis (TSH) implies that the government runs a 'budget deficit' whenever it anticipates the growth rate of national income to increase or the growth rate of its expenditure to decline. We test this implication of the hypothesis by examining the implied cross-equation restrictions on a vector autoregression (VAR) model using U.S. data for the period ranging from 1929 to 1988. Our formal tests reject the hypothesis for the full sample period, but cannot reject it for the post-1947 period. Further investigations show that the statistical rejection should be attributed to sharp differences in the statistical properties of the pre-1947 and the post-1947 data rather than the failure of the hypothesis itself. c 1993.