Monday, January 7, 2013

Performativity, Realism and Social Construction


[This post is going to be a little abstract but I hope, not too jargon-ridden.  I wrote it mostly to clear my own head.  But hopefully, it's useful to others too.]

A largely problematic concept in contemporary Science and Technology Studies (STS) is the issue of
performativity.  Performativity was explored first by Robert Merton in his discussion of what is sometimes called a "self-fulfilling prophecy."  The classic example is bank-runs.  A line forms outside the bank of people who think that the bank is going bust and who'd like their money back.  These people are seen by others who join the queue, and so on and on until the bank actually goes bankrupt because all its depositors demand their money back.  Whether the bank was really going to go bankrupt before that initial group of people queued up to demand their money is largely beside the point.  A bank run is a self-fulfilling prophecy: a belief that brings its referent into existence. 

Merton used the word "self-fulfilling prophecy" when the prophecy itself was false.  And self-fulfilling prophecies are considered, on the whole, to be bad science, at least according to normative theories of science like
Karl Popper's.  Take, for instance, Popper's notion of falsifiability as being the key idea that distinguishes science from mere pseudo-science.  Psychoanalysis was, according to Popper, a pseudo-science because it was not falsifiable.  Thus, the analyst could diagnose his patient as suffering from a certain psycho-sexual malaise; if the patient denied that he suffered this way, the analyst could attribute that to denial.  Thus there was no way in psychoanalysis to prove the analyst wrong.  Here, the psychoanalyst’s diagnosis is a self-fulfilling prophecy; by denying what the analyst has posited, I am indirectly confirming his hypothesis.  Marxism too, Popper thought, was something similar, despite its pronouncements about being scientific.  Instead of denial, Marxism used the trope of ideology and false consciousness to refute its detractors - thereby rendering itself impervious to falsification. 

Self-fulfilling prophecies, then, are considered to be bad science, by Popperian standards.  Or at least, they are not science.  Are scientific theories self-fulfilling prophecies?  This is an interesting question, philosophically speaking, but not a very interesting one [1].  [These numbers stand for the end-notes at the end of this post.] It is in the social sciences like economics that ideas about performativity (which is understood as a self-fulfilling theory performing itself into existence) start to become really interesting.  An interesting issue explored by economic sociologists is performativity of prices.  Are prices "real"?  Is there something out there in the world that prices correspond to -- stable, unchanging features that are outside human intervention?  Or are prices self-fulfilling prophecies?  If prices are self-fulfilling prophecies, how do we know that "bubbles" -- which are scenarios where prices have shot through the roof and are clearly out of sync with "fundamentals" -- exist?  As the title of Donald Mackenzie's
famous book suggests, financial models are "an engine, not a camera."  That is, financial models are self-fulfilling prophecies; they bring the realities they describe into existence.

In a
long, thoughtful, meditative blog-post, Kieran Healy points out the different senses in which the word performativity is used, and in particular, how one can understand the word in order to make sense of Mackenzie's argument -- without understanding financial models are complete hoaxes.  Because, clearly, one could interpret performativity trivially -- this is that economic models have no correspondence with reality. (Which would mean that economists are fraudsters and economic models are a hoax.)  This is clearly a possibility but only turns out to be the start of the problem.  For, if economic models have no correspondence with reality, whatever that is, then what really constrains these models?  Could it be that economists could just say whatever they like, and their scholarly ideas would just perform the economy into existence?  Just intuitively, this seems untenable.  It is not so easy to change the world.  I may have any number of ideas about how to do things -- but there is no way for me to just "perform" them into existence. Changing the world is hard and difficult.

In what sense then can economic models be said to perform, or enact, economic realities?  Here, Healy draws on
Wittgenstein's ideas about rules, languages games and forms of life.   Briefly, in his analysis of rules, Wittgenstein shows that rules only make sense when considered against a form of life.  That is, rules take for granted a certain social form.  The rule "if the light is green, then start driving the car" can only be understood in a world of cars, traffic lights, roads and junctions.  One could modify the rule to say "if the light is green but there is a car at the intersection, start driving only after that car has left the intersection," or "if the light has just turned green, then decide depending on the traffic on the road if you can just stay at the intersection for a bit to decide which is the right way to go."  And so on.  IN ordinary life, these are routinely ways in which this rule is "followed" and there could be infinite variations on this.  Social life needs to be understood not as a following of rules; rather, rules need to be understood as the reifying of conventions that draw on the largely taken-for-granted background of social life. 

A further distinction made by analytic philosophers is the notion of
regulative rules versus constitutive rules.  Regulative rules are rules like the ones I used as examples in the last paragraph: "if the light turns red, stop the car" or "if the light turns green, start driving."  These rules are supposed to regulate activities.  Constitutive rules, on the other hand, do not merely regulate, but constitute activities.  Thus the rules of chess, for e.g., "the bishop only moves diagonally," "the rook moves in straight lines," actually make up the game of chess.  [A rule like "if the player's hand is on the chess-piece, then he has not yet completed the move" can be seen to be a regulative rule.]  What Wittgenstein shows is that the difference between regulative and constitutive rules is a difference in degree, not in kind.  All rules are at heart constitutive rules.  They constitute a form of life, and the form of life constitutes them. 

Healy explains the performativity of economic models in the sense of being transformative rules.  In this case, some rules merely constitute a form of life -- but other rules, game-changing ones, can actually transform the form of life itself radically.  This is like the notion of finding a "trick," say, in a card game. If the trick is trivial, then the game remains the same.  If the trick is truly radical, then others start imitating it, and in time, it transforms the card game completely. 

Economic models (like the
Black-Scholes formula) are like truly game-changing tricks (or rules).  They are performative in the sense that as they are increasingly adopted (primarily because they seem to confer an advantage on those who use them), they change the game itself -- in this case, finance.  And it is that sense that financial models *have* truly been transformative: an engine, not a camera.

Is Healy's argument a form of social construction, or is it not?  I think it is, although arguments about whether a theory is constructionist or realist are often debates over values and postures that theory-makers should adopt.  Because, Wittgenstein, as I see him, was a social constructionist: he argued that all rules ground themselves in social conventions; that, at bottom, most structured activities are a matter of how we agree or disagree with each other.  Our methods for agreeing or disagreeing with each other, of checking up to see if we believe an assertion or don't, are all different in different fields of activity.  The BSM formula, in Mackenzie's case, "worked," in some sense -- and the fact of its working was what allowed it to be a game-changer, performing a different kind of economic reality into existence. 

Orgtheory had a separate post (part 1, part 2) from Ezra Zuckerman who argued against "pure" social construction.  Again, his blog-posts are about the performativity of prices: how are prices constrained by more than just the beliefs of market participants?  Zuckerman suggests that a sociological theory about prices must take insights from both, a realist, and a constructivist perspective.  The constructivist perspective, he argues, has taught us that beliefs play an important role in price determination (why else would people talk about a "bubble"?).  But they make it seem as if prices are only constrained by what he thinks are "subjective" factors.  He argues that there are "objective" factors that also determine/constrain prices -- and it is not enough here to say that prices are conventions, although not easily changeable.  Rather, a theory must specify what factors constrain prices -- and the list of these factors must include both subjective and objective ones. 

However, his definition of "objective" is cultural.  Objective factors are those, he suggests, drawing on
Andrew Abbot's work, that resist and cannot be changed by short-term cultural work.  Thus our beliefs about the price of Manhattan apartments and GE stocks as being worthwhile investments are objective; because short of an earthquake destroying both Manhattan and all GE plants, this is a reality that is not going to change soon.  (One could probably undertake a long-term project, a social movement, to change Manhattan's status, but that might take hundreds of years, if not more.)  Prices, he therefore argues, have lower bounds (say for an apartment in Manhattan) that are not amenable to short-term cultural work.  So there are objective features that determine/constrain prices.    

This also explains, for Zuckerman, why bubbles exist and proliferate despite the existence of skeptics who warn that a bubble is occurring.  However, these skeptics were not able to express themselves in practice -- which is why the prices of homes rose and rose (until they fell, in his view, because they violated the objective constraints).  This explains bubbles: a bubble happens because of a self-fulfilling prophecy, that violates the objective reality (which is not amenable to short-term cultural work).  It also explains the presence of skeptics who doubt the bubble -- but who are unable to do anything about it in substantive terms.  It explains why a bubble happens even when so many people think that there is a bubble.

I find Zuckerman's analysis largely persuasive [2]. But here's the rub: he would argue that it is a realist (or at least, semi-realist) account while I think it is a through-and-through constructionist account.  Why is this?  Does this difference between what I see as social construction and what he sees as a form of realism tell us something about how these terms are used - and where the substantive difference between them is?  I think so -- more on this below.

First, I think the really real realists would disagree with Zuckerman over his definition of objective factors as those that are not amenable to short-term cultural work.  This suggests that objective factors are amenable to long-term cultural work (even if the possibility that this cultural work will be successful can never be predicted), something that true realists will disagree.  The truly objective never changes; it is timeless.

But I think there is a second, and perhaps, an even more important point where even semi-realists like Zuckerman would differ from social constructionists like me.  That point, I think, is about the variability of social worlds.  Let me explain. 

Let's take the lower-bound on the price of a Manhattan apartment as an example (Zuckerman's objective factor).  Let's say I am able to wind the clock back two thousand years and start again.  Or to put it differently, assume the world is a video cassette, which I rewind back 2000 years, and then play again.  Will we reach the same place we are in today?  In particular, in this alternate world, will a Manhattan apartment still have that lower bound?  Will Manhattan still be Manhattan? 

I suspect that the answer to this question will separate the realists from the constructivists [3].  A committed realist would argue that this alternate world will not be that significantly different from the world today; because the world is full of objective, unchanging, and timeless factors.  As a constructivist, I would argue that there is a large chance that this alternate world will be significantly different because the world is shaped by human actions and social institutions, just as much as it is shaped by objective realities.  And the social world is, as Barry Barnes has argued, largely a self-fulfilling prophecy, as actors act based on how they are expected to act, thereby reinforcing those very expectations.  If the world is shaped by actions and the expectations of actions (i.e. beliefs about those actions) then an alternate world that is produced by my video-cassette time-machine would be, quite possibly, a very different one.

But a semi-realist might say: that even if Manhattan is not Manhattan, there would still be lower bounds on prices.  The real lower bound would be Chicago real-estate prices.  There would still be lower-bounds on some prices, and there would still be objective factors (in Abbott's sense) that constrain prices.  I agree.  The difference though is that as a social constructionist, I can offer no predictions on how this world will look like.  The so-called objective factors in this alternate world will be different -- but the beliefs of market participants (the subjective factors) will still influence the prices.  Once we see that the objective factors are not predictable in this alternate world, the subjective factors (people's beliefs, the way they act on those beliefs and so on) start to seem more fundamental.

To put in yet another way, the difference between constructivists and realists is over the issue of prediction, and in particular over the issue of long-term prediction.  Short-term predictions are possible for both the realist and the constructivist.  But long-term predictions, say, about housing prices or computer prices 50 years from now, will be more difficult for constructivists to make than realists.  It is difficult only because even objective factors that determine prices can be changed by long-term cultural work; and this cultural work is impossible to predict.  The more confident you are about prediction, you shift to the realism side of the spectrum.  The less confident about prediction you are, will make you more of a constructivist.

And this explains, finally, some of the arguments that have been happening in the Orgtheory comment threads.  Would you like your regulator to be a realist or a constructivist?  Realists argue that even the existence of regulators is premised on realism; for if there were no objective factors constraining social facts (like prices) then how would one even begin to regulate something in the first place?  I would disagree.  I think it depends on the time-frame that the regulator is supposed to regulate.  A regulator who is thinking about the future 50 years from now is simply deceiving himself or herself.  For a regulator who is thinking 5 or 10 years down the line: it simply doesn't matter whether he is a realist or a constructivist.

--------
Endnotes:
 
1.  Clearly, in one key sense, they are not.  Because one is able to do things with scientific theories: predict the behavior of projectiles with reasonable accuracy, build rockets, make aeroplanes fly, manufacture cars and plastics and so on.  But in another, perhaps minor, sense they are.  A theory is a piece of representation that can "fit" many different worlds and situations.  I put fit in quotes because the process of "fitting" takes a lot of work and labor (until it gets routinized when that labor simply becomes taken-for-granted).  Galileo's key insight is said to be that he saw the motion of a body on an inclined plane as similar to the motion of an oscillating pendulum.  When students learn mechanics in high school and early college, they are taught to see (to "apply") the problem-sets as examples of the canonical practice problems that they study (Kuhn 1969).   In this way, one might say, is a self-fulfilling prophecy: more and more things can be seen as examples of the theory; the theory essentially fulfills itself. 

Naturally, the skeptic might ask: well, the theory actually works.  In some sense or the other, one can verify or check up whether it works.  This is true.  But with one qualification, which is of course, how the word "works" is understood.  The sense in which the theory is a self-fulfilling prophecy, however, is that there might be some alternate theory that "works better."  Or "works," but in a different sense.  And a dominant theory, by working reasonably enough, might be letting us overlook other possible theories, because we only care about a theory "working," in one particular sense.  As long as there is only one sense of "working" that we are preoccupied with, then that one dominant theory is a self-fulfilling prophecy.

All that said, the performativity of scientific theories is not really an interesting issue.  [Because the way a theory is taken to "work" is a pretty much unproblematically taken for granted notion in science.  Unless, in Kuhnian terms, there's a scientific revolution brewing, in which case, this becomes a fiercely debated issue.]

2.
  There's also an element of cheating in Zuckerman's model.  It is easy to see that there is a lower bound for Manhattan real estate prices.  Could one think of a lower bound for, say, Phoenix real estate?

3.  For another take on this, see Chapter 3 of Ian Hacking's "TheSocial Construction of What," titled "What about NaturalScience?" where Hacking offers three factors that separate realists from constructivists: the contingency of scientific theories (or in our case, social facts like prices), the nominalism of the analyst, and explanations of the staying power of theories and social facts. 

2 comments:

Graham Peterson said...
This comment has been removed by the author.
Graham Peterson said...

A nice subtle post. Thanks for the Popper reference -- I'd thought that before and didn't know I had been anticipated.

Abhijit Banerjee is a leading economist, and wrote on herding (interdependent beliefs) in 1992: "A Simple Model of Herd Behavior" QJE.

Similar models describe the formation of social norms and conventions. So sure, you can model a price as a "mere" belief. But if we argue that prices are "mere," then everything is mere and we ought to all go do Yoga and read Vendata -- "everything is nothing and the universe is an illusion, brother."

Economics does not create the economy -- it fails to describe how technological change happens, and technological change is central (and exogenous, strangely) to nearly every macroeconomic model of change.

Black Sholes, when first published and used, ended up with huge residuals when compared with actual prices in financial markets. Those gaps closed as people started playing to the model. Economic training makes graduate students more selfish in Ultimatum games. Sure. So too, many a sociology student storms out of his job at a convenience store screaming at his boss about being exploited.

Markets, trade, and capitalism are not an 18th century invention of political economists, and the modern economy is not at the whim of introductory economics students who can't work a supply and demand diagram to save their life a month past their first course. Economists do not claim prices are "real" in an anti-Capital V.1. Given people's preferences, which sociologists are free to explore the origins of, they make welfare enhancing choices to the best of their knowledge, usually. The question of informational deficiency and welfare then, fought between economics and sociology, is whether or not people so totally lack information about their decisions that they're convinced of society-wide lies that degrade humanity, or whether they occasionally lack enough information to not buy bad assets or marry the wrong woman.

Popper was dead on in his comparison of false consciousness to Freudian psychoanalysis. He who believes the world is made of lies will always be right, and never convinced otherwise, because there is no way to reason against him. She who argues against him in this view just a liar, and her evidence merely the lies the rest of the world is convinced of.

Words like "performativity" and "false consciosness" are elegant ways to make extremely severe claims about the ethical underpinnings of the professional choices and political beliefs of swaths of people. It's not surprising they inspire such hateful wars of ideas.

That people are not always or even usually aware of the origins of their own beliefs, or even consciously participating in the accretion of them, does not necessarily imply that those beliefs are welfare reducing. Assuming as much denies the ethical, normative, historical, and performative roots of structural and critical theories themselves -- it denies that the whole point was to go looking for a revolution before understanding well how society worked in the first place.