orgtheory.net

French conventionalist economics

with 10 comments

Over at my favorite obscure journal, Journal for the Theory of Social Behaviour [sic] there is a recent article by John Latsis on the “new” French convetionalist school of economics (what the author describes as a strange mixture of “Durkheimian” sociology, Austrian influences and German institutionalist historical economics). This “new” movement was also recently reviewed by Biggart with a view of adopting some of their insights into American Economic sociology. While the JTSB article has little to say about American econ soc, it does a good job of laying out the basics of the French conventionalist approach, and in characteristic JTSB style, also delves into the “ontology” of this school of thought (hint: it’s “intersubjective”). It seems that this might be the next big wave of French theory to hit the states (at least in the econ soc parts) so this article might be required reading for some of you.

One funny thing is that the author of the article describes the conventionalist critique of neoclassical economics as if it was new or recent or could be dated to 1989 (when the conventionalist “manifesto” appeared) when in fact it almost reproduces verbatim the American econ soc critique of mainstream economics (history and institutions are important, timeless economic laws don’t exist, institutions must be endogenized, categories of economic actors are socially constructed and historically deconstructed, etc.). There is some important new lingo, and it appears that the idea of “externality” does a lot of theoretical work. For instance, Latsis notes that:

The conventionalists use a number of different techniques in order to develop analyses of the social world that differ in motivation, method and substance from more orthodox economic accounts. Yet one central claim epitomises their contribution to economics: conventions transform social actors’ experiences andinteractions by imposing an externality onto the objects and social relations that they are in interaction with and participate in. How conventions impose externality onto actors experiences is not always clear, particularly because this may vary according to the specific context in which the convention arises. However it is clear that the conventionalists see human co-ordination as carried out within normative frameworks that are not always decided or explicitly agreed to by the actors. Furthermore, these normative frameworks are arbitrary in the sense that they are not the natural outcomes of the optimal operation of a quasi-mechanical system; they could have been otherwise (262-263).

So my guess is that this would count as “heterodox” economics only for the most close-minded neo-classical. It is certainly common sense for any economic sociologist and I would guess for most Austrian economits.

Written by Omar

April 9, 2008 at 8:37 pm

Posted in economics, just theory, omar

10 Responses

Subscribe to comments with RSS.

  1. what the authors describe as a strange mixture of “Durkheimian” sociology, Austrian influences and German institutionalist historical economics

    The French Conventionalists have their origin in America, beginning with David Lewis’s analysis in Convention, which itself borrows, refines and generalizes Tom Schelling’s idea of “focal points” as a solution to co-ordination problems.

    Kieran

    April 9, 2008 at 9:37 pm

  2. Latsis goes over Lewis’ work and claims that the new French convetionalists criticize and mostly reject it, due to its “individualism.” Here’s a snippet:

    In practice the tensions involved in establishing an individualistic theory of convention can be shown by examining an alternative, overtly individualistic, account of convention. The philosopher David Lewis (1969) pioneered the modern analytic treatment of convention. Lewis sees conventions as arising from situations where agents have a mutual coincidence of wants. He analyses convention as the solution to a co-ordination game. Thus, the problem faced by agents is to co-ordinate around one among a number of equally satisfactory possible outcomes, given that any co-ordinated behaviour is better than none at all. It is essential that precedent plays a central role in this form of co-ordination. If there is a clear precedent from past instances of the co-ordination game, and if this precedent is well known to the relevant population of agents, then the agents would have a reason to co-ordinate around one salient equilibrium. The essence of the game theoretic concept of convention is:

    . . . a self-perpetuating system of preferences, expectations, and actions capable of persisting indefinitely. (Lewis, 1969: 42)

    Precedent leads to the formation of expectations about how people will act in the future. This leads to conformity in the actions of individuals which is represented by a behavioural regularity that conforms to one of the possible equilibrium strategies of the co-ordination game. A final essential element of the Lewisian framework is common knowledge. According to Lewis, for a population to sustain a convention it must be common knowledge that a given behavioural regularity is the convention. This along with the mutual ascription of common inductive standards, background information and rationality further ensures stability of behaviour.

    Leading conventionalists present ample evidence to suggest that their accounts are incompatible with this model. Two simple observations highlight this lack of fit. First, radical uncertainty is at the basis of the conventionalist account, yet the Lewisian model employs probabilistic reasoning and expectation formation that would not (by definition) be possible in a situation of such uncertainty. Whereas common knowledge is at the basis of Lewis’s model, common ignorance is at the basis of the conventionalist one. Second, the conventionalists explicitly set out to discuss co-ordination in real-life scenarios, including situations in which some conflict exists. This is a significant departure from Lewis, who explicitly concerns himself with an abstract model of a pure co-ordination game, that is a game which excludes the possibility of conflict.

    Some conventionalists are aware of this inconsistency, as their comments on Lewis and contemporary game theoretic modelling show:

    Convention does not emerge automatically under specific external conditions, in a given situation, such that whenever such situation exists, the convention will be found. A situation may itself be identified (interpreted) in many different ways, and thus it may lead to quite different actions from one moment to another. (Salais & Storper, 1997: 18)

    This last statement highlights a third fundamental difference between the two approaches. In the above quotation, the authors effectively deny the possibility of identifying necessary and sufficient conditions for the production of conventions. This project was, however, one of the major driving forces behind the publication of Lewis’s book and the subsequent development of game-theoretic analyses in this area. The inconsistency between conventionalist pronouncements and leading individualistic theories of convention sheds further doubt on the compatibility of either form of MI with conventionalist analysis (269-270).

    Omar

    April 9, 2008 at 9:46 pm

  3. due to its “individualism.”

    Hmm, I see. Though I find this a little irritating because Lewis’s analysis is an actual, worked-out analysis of how conventions could emerge where people have a desire to see a convention emerge but there’s no central authority to decree the outcome. Saying that Lewis “excludes the possibility of conflict” doesn’t seem like a particularly productive criticism, because including the possibility of conflict or power differences (and thus, the ability to impose some arrangement on people) actually makes the theorist’s job easier, not harder. It gives you a giant extra degree of freedom.

    Lewis, in particular, was interested in what it might mean to say — as people often do — that language is a conventional system of signs, and wanted to see whether there was a way, in principle, for such a system of conventions to emerge from the ground up, given that by definition the conventions of language couldn’t be established by a bunch of speakers sitting down and decreeing (in that language) what the rules were going to be.

    Kieran

    April 9, 2008 at 10:02 pm

  4. Très intéressant…

    Steve Vaisey

    April 10, 2008 at 12:50 am

  5. I had thought of conventions as the informal, everyday interaction equivalent of organizational routines. The Biggart and Beamish Annual Review piece nearly equates the two. According to them conventions are habits, customs, etc. that “organize and coordinate action in predictable ways.” Sounds a lot like routines to me.

    But Latsis never really nails down what he means by convention. It seems to be a very broad definition. The section of the paper about unemployment is more about categories than habits or customs. My impression was that the conventional approach he describes seems like a French substitute for institutional theory, focusing on how new logics and categories are constructed and institutionalized.

    Most orthodox models of the economy take it for granted that there are a
    specific number of agents in employment and a further number (still capable of
    work) who are actively seeking it. Nevertheless, in practice any measurement of
    the composition of these two broad categories involves a complex siphoning process
    that allows the government statisticians to establish firm and isolable categories.
    More often than not the categorisations actually made have the effect of
    changing the nature of social reality for those to whom they are supposed to refer.
    To take a simple case, a person without employment above the age of 65 in most
    European countries is considered to be retired; as such he is not accounted for in
    the unemployment statistics and does not benefit from the state apparatus of
    unemployment provision.

    This simple case hints at a number of deep ethical and scientific problems that
    those who construct statistical categories for policy usage must face. The most
    important problem is both straightforward to identify and nearly impossible to
    answer: what is the role of creators of statistical and economic categories? To
    discover and delineate pre-existing social categories, thus doing justice to the
    reality of existing external social conditions? Or to apply explicitly normative
    principles to the formation of categories which will create and maintain a social
    reality in line with the ethical, scientific and technological conditions of the society
    in question? While traditional economic analysis has assumed such concepts are
    situational features or initial conditions and that we are powerless before the “laws
    of the market”, conventionalists reject this assumption. For them, the emergence,
    maintenance, transformation and sometimes disappearance of social categories is
    an important part of the economic and social sciences.
    (pg. 258).

    brayden

    April 10, 2008 at 5:28 am

  6. Omar and Kieran,

    Excellent discussion about Lewis and individualism in French economics of conventions. Maybe you spotted some reason why this “school” always needs to struggle for its “unity” (they’re producing manifestos every once in a while), which at some point seems to be splitting into regular (or almost regular) economic sociology on the one hand and sophisticated (although also almost regular) economics with cognitive and evolutionary bits on the other hand.

    Unity is a problem, because economics of conventions is supposed to include such different works as Luc Boltanski’s on moral sociology or André Orléan’s on mimetic rationality (or Alain Desrosières and Laurent Thévenot’s on the construction of statistical categories).

    Another harsh, local ongoing discussion is about whether or not this approach is left-wing or not (in an agitated book published in 2005 in a collection initiated by Pierre Bourdieu, Bruno Amable and Stephano Palombarini said that the conventions school is bad because they are mild Blairites and that they should do some Carl Schmitt instead).

    typewritten

    April 10, 2008 at 7:47 am

  7. It is kind of funny that a lot of the work on the construction of statistical categories goes by the name of “convention” since that surely would be file under some sort of Berger-Luckmann “social construction” file if an American were to do the same study. Maybe “convention” is just a nice way to say “collective representation” among French economists?

    Omar

    April 10, 2008 at 9:29 am

  8. To be sure, nice ways of saying things is definitely a most crucial scholarly concern in French social sciences. The best is when you have a concept that is difficult to translate in English because of its intrinsic nuances. That is an achievement. For instance, a spontaneous hypothesis about “convention” is that it may be meant to secretly resonate with Robespierre and the Jacobins.

    typewritten

    April 10, 2008 at 9:39 am

  9. Convention’s pretty common as a catchphrase for persistent social patterns throughout economic literature, isn’t it? Samuel Bowles and the behavioralist/complexity economics folks make extensive use of it, and I’d gotten the impression they simply borrowed it from earlier (post-classical) game theory lit.

    Anyway, I’ve always had problems with this claim, no matter which school it’s thrown into:

    “timeless economic laws don’t exist”

    On a somewhat trivial and petty level, it’s obvious that “timeless economic laws don’t exist” is in itself a timeless economic law, if true, but more to the point — if there’s no time invariant law to be found, what exactly is everyone looking for? Whenever we invoke some context and say “within this context this law is valid” — isn’t it just a time invariant law with a caveat? And isn’t the natural scientific quest to expand that law to endogenize the context, too?

    Meh. Too much self-fulfilling defeatism in declaring bounds to generality for me. The social sciences are young! Give it time and enthusiasm.

    praxeologicalGoertzelite

    April 10, 2008 at 1:19 pm

  10. Indeed, there is a severe lack of unity among conventionalists. I was not able to access to the article, but a large part of this school holds Lewis as a major reference (and also, as mentioned, Schelling). The ‘89 papers which founded it mostly did, you might know the philosopher Jean-Pierre Dupuy, who teaches at Stanford. He is one of its founder (and, I think, the most interesting), whose influences come from a deep critique of austrianism and a large use of pragmatism.

    Another stream follows a more radical departure of individualism methodology, but is more recent, and probably closer to Orlean’s interpretation of Keynes, though I am not sure. Anyway, these divisions in such a tiny school of thought make me suspicious of the future and the exportability of their methodology, though there are certainly some interesting things to take for a sociologist, as well as for economists (even mainstream). Boltanski, Dupuy, may be Orlean have certainly some success abroad, but more as individuals than as collective founders of an fruitful heterodoxy.

    As for the definition, in my not-so-old days in Nanterre, I was always taught from the lewisian examples : how do people choose to drive on a certain side of the road?, or game theory : how do people coordinate in simple games?, etc. and then extended to the firm (a convention as an organisation scheme, a labour effort setting), the market, etc. It might not be perfectly convincing, but the basic definition didn’t seem so fuzzy.

    (after writing this long comment I just noticed the post dated from April. Late, but, oh well…)

    Markss

    June 12, 2008 at 12:57 pm


Leave a Reply