Le problème du réalisme des hypothèses en économie politique

This is a version, slightly corrected in 2010 with regard to form, of a working paper produced in 1968. Its subject is the problem of the realism of assumptions in economics. It offers an interpretation of Milton Friedman's famous essay of 1953 in which, contrary to most discussions, Friedman's solution to the problem does make sense. Under that interpretation, Friedman does not assert that one should test the consequences of a theory but not its assumptions, that one can predict but not explain, or that individuals behave as if they were rational and firms as if they maximized profits. Such assertions do not make sense and ascribing them to Friedman makes criticism of his position much too easy. A correct formulation of that position is that some phenomena belonging to some specifiable classes can be analysed, explained and predicted as if some propositions, for instance about rationality or maximization, were true, even though, when supposed to concern the real world, they are false. The paper includes a discussion of some philosophical questions, such as that of instrumentalism, raised by that solution.