performativity of markets and endogeneity
I believe someone on this blog banned the word “performativity” some time ago — but, I’m going to lift the ban for a second and give a quick report on the performativity session at the Academy of Management in Montreal last week.
The session was well attended and good fun. It involved Fabrizio Ferraro and Daniel Beunza, Yuval Millo, Nicolai Foss and me, and Bruce Kogut. As I’m sure that the Socializing Finance folks will also give a report, I’ll just give a quick summary of our presentation.
First, we (this is co-work with Nicolai) discussed some good news (slide #1) about the performativity argument vis-a-vis markets:
- It attempts to open the black box (construction) of “markets.”
- Performativity focuses on actors and their subjectivity, along with the specific tools, devices and socio-political machinery of market construction.
- Often provides rich histories of economic phenomena.
Then the bad news about the performativity of markets argument:
- Simply: it’s wrong (this was slide #2).
More seriously, one way to perhaps summarize the issue with the performativity argument is that it suffers from a serious endogeneity problem. Namely, the performativity approach selects a particular model (theory, device, etc) and traces it’s “effect” (diffusion, use) through history, post hoc, without looking at the set of possible models (or devices) that might have been chosen or created. In fact, performativity assumes that models, a priori, are “arbitrary” and thus ignores (even rejects) the underlying “reasons” for why the particular model (or device etc) has an effect or perhaps is better than feasible alternatives. So, let me elaborate:
- The performativity argument mistakes uncertainty for performativity. In other words, what seemingly looks like performativity is actually the “best efforts” and guesses of actors to understand states of the world or create devices to explain it. Given this uncertainty it is unfair to simply label this initial work by actors as a performance, though this is precisely what the performativity argument does: it takes imperfect models, developed by informationally-constrained actors, and traces the diffusion, uses and aberrations of these models (the data of performativistas).
- Put differently, performativity samples on the dependent variable by only focusing on the successful models rather than the set of possible models (or looking at how, given heterogeneity, a particular model emerges).
- Furthermore, performativity not only dispenses with notions of truth, but also comparative notions such as “better.” According to performativity, there is no way to judge a priori, the truth of models is directly linked with the diffusion or social acceptance of these model, and their self-fulfilling nature.
- So, performativity, then, essentially confounds social acceptance and diffusion with performance itself without recognizing the rationality and informational constraints of human agents. That’s a problem.
- Performativity rejects any notions of being (granted, subjectively) able to assess the environment. To get more concrete — Fabrizio for example talked about how socially responsible investment vehicles (tools, markets, etc) were created, and cited performativity as the explanation. But clearly the construction of this market is not impervious to an actual, real market opportunity. Or? Can any market be performed?
- The whole counter-performativity notion directly refutes performativity itself. The performativity argument is self-refuting in several ways. First, performativity has to logically allow, though doesn’t, for some kind of a priori reality that is being modified by the performance. Second, if performance is the only act in town, then there is no way for counter-performance to emerge. For example, the notion of a bubble (well, specifically the “pop” of a bubble) is logically not explainable from a performativity perspective.
- I have often wondered how the performativity program reflexively sees itself. So, for example, if, as performativistas say, economists are engaged in an arbitrary performance, what is the performativity program itself doing? It seems to implicitly take on a rather privileged stance vis-a-vis itself, though this stance (again) is self-refuting given the assumptions that performativity makes about the exercise that others are engaged in.
- Yuval kept talking about “testable reality” and “prediction” vis-a-vis performativity. His point, I think, was that social acceptance is the test, but note again that agents are deliberating about the potential uses of a model or device under uncertainty, with feasible alternatives and informational constraints, and that is quite different from performativity. One can’t simply point at Ptolemy’s geocentric model and tell a performativity story.
- A final concern is that the performativity of markets/finance argument builds directly on the Edinburgh strong programme of research (and science and technology studies), which fundamentally dispenses with the very notions of science and knowledge (a la Barnes and Bloor: prediction and explanation, reason and rationality are thrown out the door).
OK, that’s my quick laundry list.
Overall, despite my pessimism about performativity argument, I do think that we need additional theorizing and empirical unpacking of the notion of “markets” — something that the performativity program is attempting to do. There are clearly important processes of social construction going on in markets, and understanding the actors, their subjective assessments and models, influence processes and mechanisms, tools, devices etc seems central for a proper understanding of markets. So, I’m glad the performativity folks are also addressing these types of issues and are willing to engage in friendly interactions and lively debates.
Teppo, let’s get things straight:
First, you keep referring to performativity as if it were a stand-alone research programme or maybe some kind of meta-actor. It’s neither. It is one of the analytical tools that the Social Studies of Finance (SSF) uses to analyse the behaviour of markets and it cannot be understood outside context. The same way you cannot refer to the embeddedness approach by reducing to only one finding in say, in one of Brian Uzzi’s paper, then when you talk about performativity of markets you need to refer to the SSF body of work (or, at least some of it). Otherwise, you’d be debating with straw men you created.
Second point, which is related to the first. You say that the ‘performativity approach’ “selects a particular model and traces it’s “effect” (diffusion, use) through history, post hoc, without looking at the set of possible models (or devices) that might have been chosen or created.’” Again, I don’t know which empirical research you are referring to (and SSF is avidly empirical), but my studies in the historical sociology of financial markets refer to different ‘candidate models’ and shows how one was adopted and had a performative on the market (check the AOS paper, for example). Given that, it is obvious that my work does not assume ‘that models, a priori, are “arbitrary”’.
In short, I would be happy to debate and discuss my work and other SSF and SSF-related work, but, Teppo, you got to read it first and refer to specific points.
Other points
You say that ‘what seemingly looks like performativity is actually the “best efforts” and guesses of actors to understand states of the world or create devices to explain it’. Market actors always face and deal with uncertainty, so how come we don’t see performative effects in the market all the time?
‘Performativity has to logically allow, though doesn’t, for some kind of a priori reality that is being modified by the performance.’ If this is true, then where did the options’ exchange come from? I hope you’re not trying to claim that the Black-Scholes model created it. The exchange was an arena for performativity and it existed before this process took place and was still there when the process ‘reversed’ (counter-performativity).
About the paragraph about my talk: I honestly do not understand what you’re saying there. Please explain.
Yuval
August 18, 2010 at 12:25 am
I think Douglas Adams’ bowl of petunias had something to say about this…
But one thing that I want to add: Michel Callon coined the phrase “the performativity of economics”. Callon is one of the founders of ANT, which was itself a critique (albeit a loving one) of the strong program. And I think it’s useful to keep performativity’s genesis in ANT in mind, because it focuses our attention on how actors (“calculative agencies”) are made, not simply on how theories are made true or false. That options traders can trade implied volatility at all is an example of performativity – implied volatility simply wasn’t something one could see and thus trade before the models that included it were created. (Yuval or anyone else – correct me if I’m wrong about that last bit!)
Dan Hirschman
August 18, 2010 at 2:14 am
Yuval:
I’m specifically pushing on the matter of philosophical and logical consistency within the performativity program of research. So, for something to be meaningfully, truly “performed” it could a priori be ‘right’ or ‘wrong,’ right? (In fact, it often counts on the wrongness, to show an effect – but setting this aside.) The point then is that performativity doesn’t care about the a priori, it’s “arbitrary” (Callon, 1998), wrong or right (I won’t quibble with how reality is assumed here). And, that’s essentially where things go wrong — surely actors deliberate and purposefully create (again, the best they can — or we could make a power-interest argument, whatever) models and devices that, yes indeed, could be wrong, but also these are selected and created from possibilities and based on deliberation (which, again, Barnes and Bloor, 1982, explicitly don’t allow: “there are no context-free or supercultural norms of rationality”). And importantly, models and devices are revised based on feedback.
So, we can talk about performativity (and again, the historical and empirical examples are interesting), but in my mind the label does not add anything to an explanation theoretically. I think we can construct models of markets without needing to resort to performance. (That said, I do think that conceptions of ‘markets’ often remain overly abstract, structural and taken as a given — rather than ‘constructed.’)
teppofelin
August 18, 2010 at 6:07 pm
Daniel: I only now saw this story (just barely ran into it) about you in Wired Magazine — http://www.wired.co.uk/wired-magazine/archive/2010/04/start/daniel-beunza-market-anthropology?page=all
Nice job!
teppofelin
August 20, 2010 at 6:40 am
Teppo, I think you misunderstand Callon. In citing Callon’s (1998) use of the word “arbitrary”, did you read the next two paragraphs in his paper? He states: “This 2500-year old conception is not convincing” … “as the 1987 episode shows, it cannot for all of that be considered an (arbitrary) convention. The content of the formula matter.” Here Callon strongly affirms the existence of the outside reality you speak of.
I’ve read much of this literature closely and I can’t make sense of what you mean by “the performativity argument.” Are you suggesting that performativity scholars argue that the content of theories is irrelevant? That theories become true simply by means of self-fulfilling prophecy? None of the literature I’ve read makes that strong claim.
I can understand the confusion — I’ve often wondered why the literature has adopted the word “performativity” from Austinian speech-act theory. Speech-act theory places a strong emphasis on the importance of language itself in creating social order, while performativity scholars generally give language itself no privileged position in their ANT-inspired case studies. As Callon states in the paper you cited, “it is not the formula itself that can cause that world … to exist. Other forces are involved, other interests.” ANT and speech-act theory have deep, unresolved incompatibilities that ought to be resolved in future research.
But the performativity literature is not, as far as I’ve seen, advancing a general hypothesis about the nature of social science. It’s giving counterexamples to simplistic, conventional thinking about the nature of social science, i.e. Friedman (1953) — whom MacKenzie criticizes specifically in his book.
In other words, the literature I’ve read claims “there exist odd cases where a theory is adopted for reasons other than its predictive ability; moreover adoption of a theory can actually increase its predictive accuracy (at least temporarily for Black-Scholes).” It does NOT claim that “social science theories become true when everyone adopts/believes them.”
As counterexamples, these case studies should foster further philosophical reflection the nature of social science. They should not be rejected out of hand as “wrong” because they’re incompatible with how we ordinarily understand social science, unless you can demonstrate some kind of error in the empirical methodology being employed. Again, as Callon indicates, the content of the formula is NOT arbitrary, and it’s important that we nail down what, precisely, matters when selecting among competing theories if it’s not simply predictive accuracy.
Taylor Spears
August 20, 2010 at 9:06 am
Taylor: Thanks for responding.
So, the arbitrariness (or whatever word one uses, that just happens to capture the ‘problem’ of an a priori reality) seems quite central to the argument (despite several uses of it by Callon) —- yes, I’ve read what you cite and it continues as follows, illustrating the point: ….”Imagine, as MacKenzie proposes a different formula, for instance one with a calculation error or a statistical incoherence. Would it have had the same impact? This question is obviously difficult to answer directly since no one has ever tried the experiment. [And then goes on to discuss Black-Scholes as an example.]” Fine. But if a formula is adopted (with all its implications for decision-making, devices, actor etc), though ‘wrong,’ so what? One doesn’t have to talk about performativity to capture errors like this (Callon then goes on to cite counter-performativity vis-a-vis Black-Scholes and 1987). One can look at the set of possible models and devices that agents can come up with, accounting for uncertainty and the fact that subjective agents, rational as they are (within limits, ‘best efforts’), and then adjust their expectations, devices and behavior based on feedback from the environment.
To say that the latter somehow is inherently congruent with performativity does not seem to work. Performativity’s underlying assumptions about human rationality and reality are nicely captured in Barnes & Bloor (1992), Relativism, Rationalism, and the Sociology of Knowledge (book edited by Martin Hollis and Steven Lukes). There you’ll see just how radical a difference there is between a scientific conception (e.g., see Hollis’s chapter on the ‘social destruction of reality’) and the one argued by the strong programme.
My conception of the process of construction, then, is more conducive with truth seeking (no quotation marks needed) in a social environment, where naturally there are dynamics that lead to construction, but where we can meaningfully talk about reality, agent efforts to garner information/makes sense of the environment, influence and learning, heterogeneity and so forth.
There’s a very deep epistemological divide here (each camp with their own journals in philosophy: for example, compare the journals Episteme and Social Epistemology). Some good sources to read include: Goldman, A. Knowledge in a Social World. OUP or Boghossian’s Fear of Knowledge, OUP. Full citation details for these (and other) books can be found here.
teppofelin
August 20, 2010 at 9:54 pm
[...] More seriously – I don’t buy Dan’s arguments here. As with most stage theories (not only Marx, but also Kuhn), the mechanisms of institutional reproduction and change in his account are sorely underspecified. ‘Contradictions accumulate’ isn’t a much more helpful empirical claim than ‘shit happens.’ To really understand what is happening, you need a proper theory of the underlying conditions for ideational retention and reproduction. Why do some ideas decay into self-parody, while others do not? After all – not all ideas decay (or at least: not all ideas decay at the same rate). Some economic ideas have continued for centuries (the limited liability corporation), while others have disappeared completely, while others yet have disappeared and reappeared. We don’t know why – but if we want to make the kinds of claim that Dan is making, we need to know why, or at the least, have some rough idea. Otherwise, what we have is at best a sometimes-observed empirical regularity melded to a smidgen of intuition, which is not enough (in my book at least) to dismiss a counter-claim (that one particular idea may have a longer shelf life than previous versions) out of hand. The only large scale effort to come up with a proper theory that I am aware of is the sociological literature on performativity, but this is distinctly more useful in explaining how ideas succeed than how they become ossified, and lacks any account of the mechanisms producing variation. [...]
Cowen and Drezner Join the Sixth International (Repentant Libertarians Cadre) | Your Poetry Resource
August 27, 2010 at 6:53 am