Sign restrictions and statistical identification under volatility breaks -- Simulation based evidence and an empirical application to monetary policy analysis

Apart from a priori assumptions on instantaneous or long run effects of structural shocks, sign restrictions have become a prominent means for structural vector autoregressive (SVAR) analysis. Moreover, second order heterogeneity of systems of times series can be fruitfully exploited for identification purposes in SVARs. We show by means of a Monte Carlo study that taking statistical information into account offers a more accurate quantification of the true structural relations. In contrast, resorting only to commonly used sign restrictions bears a higher risk of failing to recover these structural relations. As an empirical illustration we employ the statistical and the sign restriction approach in a stylized model of US monetary policy. By combining identifying information from both approaches we strive for improved insights into the effects of monetary policy on output. Our results point to a decline in real GDP after a monetary tightening at an intermediate horizon.