This paper tests Barro's tax-smoothing hypothesis and Mankiw's revenue- smo
othing hypothesis, using quarterly data for four countries-Canada, France,
the UK, and the US. The tests are conducted using recent advances in the th
eory of integrated regressors, the single-equation approach (with the time
series properties of the data imposed in estimation and hypothesis testing)
as well as the multi-equation VAR approach, which treats all variables as
part of a joint process. There is considerable evidence for tax-smoothing (
and inflation- smoothing) but not for revenue-smoothing.