This paper develops a three-step empirical methodology to test the rational electoralcycle hypothesis. The first step consists of testing for the existence of electoral cycles in fiscal policy. The second step conducts three tests for how such cycles should depend on election outcomes as suggested by recent political agency models. The third step is to regress electoral success on fiscal policy. This three-step approach is applied to a panel of Swedish local governments with more than 2000 observations from elections. The findings are as follows: (i) spending is raised and taxes are cut in the election year, (ii) in the election year, spending is higher for a government that will be re-elected as compared <p> to those that will not be re-appointed, (iii) in the post-election year, spending is higher and taxes are lower for re-elected governments than for newly elected ones, (iv) reelected governments spend less and tax more in the post-election year as compared to the election year, (v) conditional on taxes, spending is positively related to electoral success. These set of findings are consistent with Rogoff’s equilibrium budget cycle model where a government signals its competence through cycles in fiscal policy.
