This article illustrates the causal relationships according to Granger between the seven French monthly industrial production variables uncorrected for seasonal variations. Dickey-Fuller's tests are used to decide the stationarity of all the variables. The series considered in a wide sense as stationary are analyzed and modelled within an autoregressive vectorial (VAR) model constructed step by step. The infinite moving average (MA) representations of this VAR with orthogonalized innovations is made in order to analyze for each component of the time series vector its sensitivity if shocks take place on its multiple causal variables. Finally for each component in the system VAR we constructed the equation of the variance decomposition for the different horizons.
INDJEHAGOPIAN, J.P. and MOURAD, M. (1993). Causality and Shock Analysis in the French Industrial Sector. ESSEC Business School.