Décomposition d'agrégats au moyen d'indicateurs. Une méthode économétrique

In this paper, we develop a method allowing to split up an aggregate according to its homogeneous components, using statistical indicators. This problem is formulated as one of optimal linear forecast in a generalized linear regression model structured in order to explain the time path of an heterogeneous aggregate. This formulation leads to a consistent breakdown of such an aggregate enabling the measurement of the statistical estimation error of the decomposition.