Performative theory, when applied to economics, is too limiting. The economy and social relations require a more nuanced understanding. (mainly indicts Callon but can be used for any performative author) Brisset 16

The idea that economics creates the world rather than describes it has been the subject of several debates (Aspers,2007; Felin & Foss,2009a, 2009b; Hodgson,2010; Miller,2002; Mirowski & Nik-Khah,2007). Indeed, claiming that a theory, even if wrong at first, can shape the world and become right seems too strong for many economists, especially if they consider themselves heterodox economists. This affirmation seems too radical because then any theory could be true if it becomes a social norm. From its conceptual point of view, ANT assumes that people act in the markets in accordance with their expectations of other agents’ behavior. These expectations are based on a stock of information and economic theories associated with this stock. Therefore, agents’ expectations can be similar to the concept of expectations used in economic theory: ‘I believe that others act according to a specific theory and I believe that others believe the same thing as I do. ’Thus, the theory gradually becomes a coordinating device (Callon,2007, p. 322), and there would be no wrong theory per se. Two kinds of criticism follow:

(1) Firstly, Mirowski and Nik-Khah (2007) but also Miller (2002) emphasize the conservative way in which performativity theory is engaged when it claims, in opposition to most critics of rationality assumptions, that homo oeconomicus does exist, because of the embeddedness of the economy in economics. To argue that economic theories shape the world is to stop any critics of dominant theo-ries that effectively influence the world.

(2) If economics matters in the construction of reality, we need to explain how and why. A recent debate brings this question to light (Felin & Foss,2009a, 2009b;Ferraro, Pfeffer, & Sutton,2005, 2009). The key question is: If science has the nature of a convention in the market place, is this convention purely arbitrary? If not, it is necessary to understand why, in a situation of competition between several economic theories, that one is adopted and not the others. Even if we consider this choice as a purely ideological one, ‘an immediate question then is which of many competing ideologies ought to be adopted and why?’ (Felin &Foss,2009b, p. 676). Felin and Foss call for a ‘reality check’ and argue that only true theories impact social reality because agents only choose to keep a theory after its confrontation with reality. This contests the direction causality. With performativity, a theory T is applied, and then influences agents’ expecta-tions. Felin and Foss reverse this: a theory T is a good description of the social reality; this is why it is used by agents.

I think these debates could be endless because the definition of performativity is vague. I also think Felin and Foss’s idea of a ‘reality check’ idea is important. Neverthe-less, their concept of truth is problematic because it ignores the main idea in performa-tivity theory which is its constructivist stance. So I propose rebuilding the theory of performativity using some classical tools of economists. Economists themselves have focused on the impact of theory on economy (Brisset,2012a,2012b). This interest has gone hand-to-hand with the issue of predictions in social sciences. Morgenstern (1928,1935, 1972) first emphasized the problem of interaction between social scientists and agents in concluding that social phenomena cannot logically be foreseen. His reasoning rests on a logical analysis: given a certain prediction or explanation, the economist knows that agents will react to it. As a consequence, he anticipates this reaction and revises his prediction which, he also knows, will be taken into account by agents, etc. This process is known as the Morgenstern Paradox.4 It is the starting point for several works on the predictability of social events when agents react to the predictions of econ-omists: a public economic theory is reflexive. Two important articles, Simon (1954) and Grunberg and Modigliani (1954), focus on the possibility of public prediction regarding a vote were the result. Their main goal is to escape from the infinite regress of Morgen-stern’s reasoning. The core argument is that there exists at least one fixed point in the agents’ expectation function–i.e. the way agents build predictions regarding econo-mists’ prediction–provided the variables predicted have upper (K) and lower bounds(k) and a reaction function that is continuous over the interval k≤P≤K, where P is the public prediction (Grunberg,1986, p. 478). In other words, there exists at least one pre-diction which remains correct even if agents take it into consideration in forming their expectations:‘The fixed point theorem demonstrates, when its assumptions are satisfied, only that there exists at least one point, F, in which the actual vote (or the expected vote in the probabilistic case) equals the predicted’(Simon,1982, p. 610). This prediction is, thus, self-fulfilling.5The Rational Expectations Hypothesis is a direct implication of the Simon and Grunberg and Modigliani result. Muth (1961, pp. 316, 317) gives the following defini-tion:

The hypothesis asserts three things: (1) Information is scarce, and the economic system generally does not waste it. (2) The way expectations are formed depends specifically on the structure of the relevant system describing the economy. (3) A ‘public prediction,’in the sense of Grunberg and Modigliani (1954), will have no substantial effect on the opera-tion of the economic system (unless it is based on inside information).

Muth views the self-fulfilling condition as necessary for a theory to be correct: a theory with Rational Expectations Hypothesis is self-fulfilling, and, thus, the true one. Sargent(1993) points out two components of rational expectations: on the one hand, individual rationality reduced to maximization and, on the other hand, the mutual consistency of perceptions of the environment. The environment, in a strategic context, includes others’ behaviors and anticipations in forecasting the level of a certain variable x. Therefore, an agent a must form beliefs about b’s beliefs which includes a’s beliefs.164N. Brisset Expectations have to be common knowledge and self-fulfilling. With Rational Expecta-tions Theory, these conditions are fulfilled by assuming that all agents have the same representation of the economic system as economist. First, this representation is com-mon knowledge throughout the population (including economists). Second, this repre-sentation is self-fulfilling: given information X, agents make X real, that is, X is a fixed point f(X*)=X*. This is the meaning of Muth’s affirmation that ‘those expectations of firms (or, more generally, the subjective probability distribution of outcomes) tend to be distributed, for the same information set, about the prediction of the theory (or the “ob-jective” probability distributions of outcomes)’(Muth,1961, p. 316). Third, according to Rational Expectations Theory, there is a symmetry between agents and economists(Sent,1998, 2001, 2002), and, therefore, no longer the problem of self-undermining theory as in Morgenstern’s argument when we consider that agents know the objective value of the expected variables.6Nevertheless, one could say that these conditions are not sufficiently restrictive: it is possible to find several sets of expectations which are self-fulfilling and consistent. In other words, that there is no objective probability à la Muth. Indeed this is the central claim of the Sunspots Equilibrium Theory (Azariadis,1981; Cass & Shell,1983).According to this theory, there can be several public self-fulfilling theories which could be adopted by agents in their different cognitive frames and thus imply different future realities. In other words, several self-fulfilling theories could exist even if they initially seem groundless, i.e. they do not affect the fundamentals of economy (i.e. preferences, endowments, or production possibilities).We finally reach the issue at stake in the different critiques of performativity: rela-tivity. This is an important issue needed to explore the relevance of the concept of per-formativity more deeply and avoid the risk underlined by MacKenzie (2009, p. 57),which is the risk of broadening the performativity thesis indiscriminately to all kinds of effects of economics on economy: performativity becomes a loose concept because everything has become performative. As Dequech says, the ability of economics to shape the economy has been both neglected by economists and overestimated by Cal-lon (Dequech,2008). Are there then some limits and conditions to performativity? This question leads us to MacKenzie’s distinction between Barniesian performativity and counterperformativity(2006,p.31or2007, p. 55):Barnesian performativity7:‘practical use of an aspect of economics makes economic pro-cesses more like their depiction by economics’ Counter performativity: ‘practical use of an aspect of economics makes economic processes less like their depiction by economics’ This is an important distinction since it potentially gives us a basis for explaining why some theories perform and not others. Nevertheless, MacKenzie does not provide a clear definition of counterpeformativity beyond what simply seems to be self-destruc-tion. Of course, this is an important idea (as I will later argue). Nevertheless, it gives no clue as to why and how agents use a theory (or a device) in their daily life. Indeed, what is ‘practical use’? And, above all, a practical use of what? Is an entire theory cap-able of being used or integrated in a device (agencement)? In addition, as we will see in Section 5, if MacKenzie uses the idea of counterperformativity as an empirical con-cept (he notes that a theory can fail to perform the social world), he does not theorize this notion as a constitutive element of an entire performative motion. As argued by Butler (2010,p.152) there is a difference between claiming (a) that breakdown and disruption of performative operations can happen, and (b) that the risk of breakdown and disruption are constitutive to any and all performative operations. The first is empirical, but the second is structural. In other words, the problematic element of the Callonian approach to performativity is the break with Austin’s project of defining performativity according to felicity condi-tions. This is why ANT is not effective in answering economists’ questions about the limits of performativity. The inability of the Collonian literature to grasp the idea of ‘performativity failures’ is mainly due to its willingness to remove society from the social landscape: Society is imagined as a context or as a frame. Society is out there and you imagine how to put this strange beast, the market, in this frame. It’s another version of the infrastructure and superstructure metaphor. You have realities that are called markets or economic activi-ties or society and what you try to imagine is the respective positions of these realities. Developing an argument made in the anthropology of science and technology, you must not imagine society as a context for different types of activities including economic activi-ties; you have to imagine the process through which collective relations are constructed, including relations that can be called economic relations. (Callon, Barry, & Slater,2002,p.291)Bruno Latour is clear on this point. First, he denies the monopoly of the social macrostructure in the understanding of individuals’ behaviors, which is one of the most stimulating contributions of his work: For the social sciences to regain their initial energy, it’s crucial not to conflate all the agen-cies overtaking the action into some kind of agency–‘society’, ‘culture’, ‘structure’, ‘fields’, ‘individuals’, or whatever name they are given–that would itself be social. Action should remain a surprise, a mediation, an event. (Latour,2005, p. 45) Second, he goes further by clearly arguing that there is simply a priori no common world(Latour,2011). This is not a minor detail since this point of view leads the per-formativity literature to ignore the fact that to perform the world, a theory has to become a social device, that is to say, it has to be socially accepted in a world made of social macrostructures that can put pressure on it (Brisset,2011, 2014a). Yet for per-formativity to gain explicative capacities, I argue that it would be important to reintro-duce the social world in the analysis in order indicate the social conditions for performativity.