Abstracts
Abstract
The best known and probably the most frequently used of the models of this class is undoubtedly the Input-Output model of the Québec economy built, continuously updated and operated by the Bureau de la Statistique du Québec. However, the methodology, adapted and sometimes extended has found a number of other, in particular micro-economic applications.
Evidently, the basic inspiration of the methodology in question is to be found in the ideas put forward by Professor Leontief. Certain researches done in France, especially in the late 1950's, but also since then, have exerted considerable influence.
Although these models trace their origins to activity analysis in the sense that they start from the principle that to understand a complex system it is preferable to study in detail its inner structures and workings rather than the evolution over time of the great aggregates characterising the overall behaviour.
By abandoning the postulates of proportionality and of one-to-one correspondence between "products" and "industries", the models discussed here openly give up any pretence to mathematical elegance including the existence of "general solutions" of the kind of those associated with the Leontief inverses. They just become in effect simulation models and at the same time much more convenient and flexible frameworks for the collection, organization and the handling of data, data which are much closer to basic data than the highly processed data incorporated in the traditional Input-Output models. They are also much more easier to update and to incorporate "non-statistical" data.
Although more powerful, in many respects, than the traditional models, they share with them at least two basic weaknesses which, significantly, are not unrelated to each other. They are incapable of handling in a really comprehensive and systematic manner the confrontation of supply and of demand influences and they give no more than a most cursory treatment to the whole range of financial phenomena and a fortiori to the influence of these phenomena on the "real" ones. It is clear that the future work on this class of models will have to put heavy emphasis on trying to reduce these two weaknesses.