A branch of the French social sciences has emerged in the 1980s, known as ‘the economics of conventions’. The term convention is introduced to subvert the utilitarianism of neo-classical economic theory. However, there is no consensual and stabilized definition of this notion. According to the investigation undertaken in this paper in order to clarify basic concepts, this term should be used for naming a research perspective. Researchers involved in ‘the economics of convention’ face two issues : the methodological issue raised by the interactions between institutions and action; the political issue raised by the assessment and the improvement of institutional measures. This article aims to show how the pragmatic philosophy of John Dewey allows for dealing with these issues.
Lewis sets out to give meaning to the notion of language convention based on the coordination of rational agents.To do so, he makes use of coordination games which require concepts such as the common knowledge of rationality that turn out to be disputable. Consequently, the link he proposes between convention and rationality appears to be at the same time too strong and too weak. On the one hand, it is hard to demonstrate that conformity to a convention as defined by Lewis may arise from a purely rational behaviour.On the second hand, the concept of convention he uses rules out coordination games in which individual rationality leads to suboptimal choice.
This article deals with the idea of the performativity of economics. We use the notion of convention in order to emphasize a necessary condition for performativity of the scientific conventions. We show that performativity could be seen as the translation of a scientific convention into the social world. We demonstrate that such a translation needs the scientific concept to take a peculiar form: an empirical one. We study the example of the performativity of economic rationality, which is now a central concept of a new kind of public policy: “nudge” economics.