Abstract: The cyclicality of discretionary fiscal policy has been the subject of a large number of papers. However, different studies have come to very different conclusions, even across similar sets of countries and time periods. Conflicting results may stem in part from differences in the included covariates, which can vary substantially from paper to paper. I use Bayesian model averaging (BMA) to estimate the cyclicality of federal discretionary policy in the United States, averaging results from a large number of possible models according to how well each model explains the data. BMA incorporates uncertainty about the true model of fiscal policy conduct and has the additional advantage of clarifying which variables matter to policymakers. The models I consider are motivated primarily by the existing literature, although to my knowledge I am the first to include employment-based measures as cyclical variables.
My results suggest that discretionary policy is countercyclical. However, this finding is driven largely by strong countercyclical responses to recessions. During expansions, in contrast, policy is highly persistent and mostly unresponsive to cyclical variables. I also find that policymakers are much more likely to respond to changes in the unemployment rate than any of the other cyclical variables I consider. Notably, the level of debt does not appear to factor into policy decisions. Finally, policymakers may respond to cyclical conditions during expansions, but with greater delay than during recessions.