The phenomenon of performativity has recently sparked debates about the status of the economic discourse. This paper aims to discuss the subjectivist idea that if economics ‘performs’ social reality, rather than merely reflects it, then every theory can be considered ‘true.’ My main goal is to point out three limits of performativity. First, not all theories can be performative since some do not produce empirical landmarks for agents. Second, social institutions restrict performativity. Third, I emphasize the necessity that a theory (...) to be self-fulfilling. This article is a prelude to a new kind of performative studies based on an original definition: a theory performs the world if it implies a behavioral regularity which leads to a general coordination between agents. That is to say, if it becomes a convention à la David Lewis. (shrink)