Measuring Political Budget Cycles: A Bayesian Semiparametric Assessment

In this paper, we estimate the effects of political variables, including the government's popularity (approval rate) and term in office (tenure), on fiscal variables without any assumption of functional form, using a Bayesian semiparametric regression technique and US data. Our main findings are twofold. First, the effects of term in office are not statistically significant; that is, we find no evidence of either an unconditional or popularity-conditional political budget cycle. Second, the effects of the presidential approval rate on government consumption are U-shaped. This implies that the incumbent has an incentive to reduce government spending when the race for office is neck-and-neck, while the budget can be loosened when the incumbent is either certain to win or certain to lose the next election. The functional form of these effects is invariant throughout tenure. The estimation results indicate that incumbent politicians have an incentive to reduce federal government consumption to signal their competence to voters who are fiscally conservative, and that the political incentive to manipulate fiscal policy arises not only immediately before elections but also throughout the term of office.