Juessen, FalkoLinnemann, Ludger2012-03-072012-03-072012-03-07http://hdl.handle.net/2003/2937910.17877/DE290R-3291We use a panel VAR to assess the dynamic effects of government spending on unemployment rates in OECD countries. We first present Monte Carlo evidence that the Hahn and Kuersteiner (2002) estimator produces almost unbiased estimates of impulse responses in an annual macro panel VAR. In the application, we find that positive shocks to government spending identified either through a Cholesky decomposition or by sign restrictions tend to lower the unemployment rate in the short run, though signifi cance depends on identification assumptions.enDiscussion Paper / SFB 823 ; 08/2012fiscal policy effectspanel vector autoregressionssign restrictionssimulationunemployment310330620working paper