Impulse response analysis of cointegrated systems (Q1186063)

A vector autoregressive model with cointegrated variables and white noise disturbances is considered, and the estimation of the impulse responses is studied. Various economic interpretations and relations are explained. Giving the ML-estimates of the asymptotic variance-covariance matrices of estimators of coefficients matrices and of Gaussian noise, the asymptotic distribution of the estimate of the impulse response and of the forecast error variance components are derived. The possibilities to absorb seasonal dummy variables into the model as well as to impose some constraints on parameters are also discussed. The theory is illustrated by a large example analysing German money demand system.