Generalized Two-Step Maximum Likelihood Estimation of Structural Vector Autoregressive Models partially identified with Short-Run Restrictions

This paper presents a generalized two-step maximum likelihood estimation method for partially identified vector autoregressive models. We suggest a likelihood ratio test for over-identification in a sub-system and derive the asymptotics for impulse responses and forecast-error variance decomposition for partially identified models. As an application, we consider an open economy model to investigate the effects of monetary policy on exchange rates and term structures. We find that exchange rates tend to overshoot and term structures have hump-shaped responses to monetary policy shocks