The tax-smoothing hypothesis implies that: 1) the optimal tax rate follows a pure random walk; 2) a budget surplus equal to the expected present value of changes in government expenditures. In this paper random walk tests of tax rate are performed for the period 1970-2002 and use a vector autoregression model to the period 1970-1997 in the case of the state of Rio Grande do Sul public finances. The tests reject the tax-smoothing hypothesis for the case in study for both periods. That is, the evidences show that the state government has not behaved as tax-smoother.
tax-smoothing hypothesis; vector autoregression; Rio Grande do Sul