No abstract

Joseph Schumpeter, as a spoilsport (and probably under Paul Samuelson's influence), looked pessimistically upon the internal coherence and methodological progress achieved by the ordinalist revolution: If [the ordinalists] use nothing that is not observable in principle, they do use "potential" observations which so far nobody has been able to make in fact: from a purely practical standpoint we are not much better off when drawing purely imaginary indifference curves than we are when speaking of purely imaginary utility functions. (Schumpeter 1954, 1067) Schumpeter actually raised a methodological issue more than a practical one. Indeed, once Vilfredo Pareto had faded from the scene, the status 1 of indifference curves within demand theory was rarely discussed for its own sake by the main protagonists of the ordinalist revolution.

Ce document est un preprint d'un article publié. Il peut contenir des changements mineurs par rapport à la version publiée. Référence pour citation :

Probabilistic models of choice (between sure prospects) have become a standard modeling practice since the 1980s, notably through the widespread use of the Logit Multinomial Model pioneered by McFadden (1974). However, the idea to model consumer's behavior as a probabilistic behavior, hence accounting for some kind of behavioral instability in the pure theory * Correspondence may be addressed to. Acknowledgments. I am most grateful to Philippe Mongin, Jean Baccelli, Ivan Moscati, and colleagues who have attended various presentations of this article for their constructive and insightful comments. I am sole responsible for any remaining error.

No abstract