The dynamic effects of aggregate supply and demand disturbances: further evidence

For the decomposition of GDP into its trend and cyclical components, Blanchard and Quah (1989) proposed to extract the cycle by imposing a long-run restriction on the moving average polynomial of a bivariate autoregressive process for GDP and a cyclical indicator. We investigate the robustness of this decomposition with respect to the specification of the process and the choice of the cyclical indicator. We find that low order VARMA processes are sufficient for capturing the joint dynamics of the bivariate process. The decomposition seems rather stable, however, there are differences in the impulse response functions associated with permanent shocks. Moreover, the inflation rate, though used in several studies, seems not suitable for the usage as a cyclical indicator.