Causality and interdependence in econometric analysis and in economic theory

This is the Inaugural Address for the first established Chair of Econometrics in Italy at the Università Cattolica del Sacro Cuore of Milan (A.A. 1964–65). The author first focuses his attention on the contraposition between causal (recursive) and interdependent models in econometrics. The tools adopted to analyse the former category of models are much simpler than those necessary for the latter category, while the bulk of current econometric analysis is carried out by means of fully interdependent models. The author claims that this preference has its roots at a more fundamental level: the Neoclassical theoretical framework, which has been the dominant approach in economics. Its logic is best represented by the Walrasian general equilibrium model. As is well known, this model is explicitly characterized by the full interdependence of all the economic variables of the system. Yet, there are other approaches, like the Classical and the Keynesian ones, where some important relations are best represented by means of causal, rather than interdependent, relations. As an example, the author recalls the Ricardian theory of income distribution, where its recursive structure is crucial in showing the essential determinants of the distributive variables and, in particular, the residual nature of profits with respect to wages; or the Sraffa price system, where the rate of profit is determined before any other unknown and – as in Ricardo – independently of rents. The author concludes by outlining a distinction between two sets of economic relations: a first group of relations called “natural”, that are independent of the institutional setting of the economic system and are thus common to all production systems; and another group of relations, specific to the institutional context, which can differ from system to system, and can sometimes be fruitfully studied with the help other social sciences.