In this article we describe the dynamic relationships between seven indexes relevant to sectors of French industrial production, using a statistic model. It is assumed that the data-generating process can be accurately described by a vector autoregression (VAR) and the approach suggested by Caines, Keng and Sethi in relation to FPE criteria is used to specify the VAR structure. Granger's causality relationships between the variables deduced from the VAR estimation are studied. Lastly, we suggest a selection procedure for the ordering of the variables of an economic system so that the Choleski method is not applied blindly to compute the variance decompositions and impulse responses. This procedure is applied to the economic system being studied.
INDJEHAGOPIAN, J.P. and MOURAD, M. (1993). Var Modelling of Macroeconomic Time Series : Causality and Shock Analysis. ESSEC Business School.