graber book forum part 2: the attack on neo-classical economics

Part 1.


I often teach the graduate course that introduces students to major themes of macro-sociology. I start off with rational choice theory, which, as you can imagine, triggers teeth grinding rage. I then ask students: “If people aren’t following their preferences, then what are they doing?” Answer: silence.

Of course, there are good answers to this question. Most economists, especially behavioral economists, would probably argue for a model that is close to rational choice, but includes biases. Sociologists often take this a step further and argue that people respond to social conventions, follow norms, heuristics, or employ cultural tool kits. The difference between the textbook rational choice model and what many sociologists believe lies not in the maximizing part of the model but in how individuals construct the options and judge alternatives.

I bring up this pedagogical example because contemporary economics is built on a number of simple assumptions that appear obvious and incontrovertible but can actually be successfully critiqued. The surface plausibility of standard economics is hard to argue with by novices, which is why first year graduate students often get stumped by the question I asked.

One important difference between economists and other social scientists lies in their willingness to entertain serious alternatives to the rational choice model. Psychologists are  so used to thinking about different models of decision making that they find the insistence on the rational choice model a bit puzzling.

The problem, however, is that for many routine social science questions, it is hard to articulate a simple alternative model that is easy to understand and can easily be the foundation for normal science. The rational choice model has handful of simple axioms, it’s easy to formalize, and easy to tweak.


So what does this have to do with David Graber’s book on this history of debt? Aside from being a radical criticism of debt, Graeber offers one of the few successful attacks on academic economic theory. He doesn’t attack the rational choice model directly. Rather, he, in my view, attacks one of the core ideas of economics that is tied to the rational choice model.

When you read the nitty gritty of economics, you often see the following jump. You start with a description of the rational choice axioms: people have options, rank them, and act upon them in a consistent way. The jump is this: money is the natural way that you should figure out what people prefer. Money is the natural expression of needs and the money economy is the natural resolution of the economic problem of distributing goods. Without money, you’d need to barter to pursue your own personal goals and that’s very inefficient.

The first chunk of Graeber’s book is a anthropological account of barter. Where does it exist?   Is it actually true that the money economy represents a solution to the problem of barter? Graeber claims that barter is actually exceedingly rare. According to him, barter makes little sense at all. Why pile up on specialized goods and wait for other people to pile up on what you want and then trade? That’s bizarre.

Instead, what happens in most non-monetized cultures is that people engage in generalized exchange. If you need X, Fred will give you X, but you (or someone else in the group) has to help Fred sometime later. Thus, most groups engage in a debt economy, not a direct trade (barter) economy. Of course, there are some exceptions, such as trade between hostile groups or prisoners from Western societies. But overall, Graeber claims that the overwhelming theme in economic ethnography is that barter simply doesn’t exist.

The conclusion? Adam Smith was wrong to say that people have a natural tendency to “truck and barter.” Why? It’s a strange, unintuitive form of economic exchange. Therefore, money is not the natural solution to barter, since barter, for the most part, does not exist.


According to Graeber, the anthropology literature, composed of observations of dozens and dozens of societies, undermines the link between self-interest and modern capitalist institutions. Classical economists, as well as their contemporaries, have made a deep error in assuming that a Western economic practice is the natural functional solution to economic issues that arise in all societies. I myself have even promoted this argument in my undergraduate class on economic sociology.

We’ll discuss the next step in Graber’s argument next week, but for now, I’ll conclude on the implications of Graeber’s attack on the barter-money link. If direct exchange of goods (barter) is not the embodiment of rational action, then what is? The answer, I think, is generlized exchange. A true believer in economics text books would correctly point out that generalized exchange can be described in terms of utility functions. Fair enough, but that’s not the point.

The real deep point is that monetary exchange, credit markets, and a whole host of other modern financial institutions are in no way natural. Furthermore, there’s actually an alternative to price theory, which uses money as it’s main variable (e.g., “clearing price”). The anthropologist’s version economics would start with indirect exchange as the main variable, which has a better claim to universality than prices, and then describe all institutions as recorders and shufflers of debt.

Adverts: From Black Power/Grad Skool Rulz

Written by fabiorojas

December 15, 2011 at 7:08 pm

Posted in books, culture, economics, fabio

22 Responses

Subscribe to comments with RSS.

  1. Great summary! Jim Johnson, political theorist at Rochester, argued pretty persuasively that the easier inroad to critiquing Rat Choice is knowledge, not preferences. Most RC models rely on pretty strong assumptions about knowledge (of outcomes, potential states of the world, etc.), even when they incorporate risk/probabilistic indeterminacy. The problem with critiquing preferences is that they can be defended almost tautologically (much as you implicitly do with your question to your students). But, as Quiggin eloquently notes, the tautological version of preferences is empty – it has almost no predictive power, since preferences are exogenous. So, economists jump to much more restrictive models of monetized self-interest. When critiqued, they oscillate between these two arguments “as sin 1/x as x approches 0” in Quiggin’s mathy metaphor.

    More german to Graeber, Gavin Kennedy of Adam Smith’s Lost Legacy has a multipart critique of the book – and unsurprising defense of Smith – starting here. It might be of some interest to you or other readers. From my reading of Smith and of Graeber (not yet complete), I think Graeber is definitely overstepping his claims in his critique of Smith (if not always of contemporary economics).


    Dan Hirschman

    December 15, 2011 at 7:27 pm

  2. Fabio,

    Glad to see this. I haven’t read the book yet but have heard good things. A couple points to push back on, and I’m curious to see your response

    First, your quick description of RC as “people have options, rank them, and act upon them in a consistent way” is how RC is often taught but it is not perfectly accurate. A more accurate one is “people have options and act on them in a consistent way” is most accurate. RC is ultimately about consistency. This is why it can be used for humans or animals or genes. There’s nothing about cognition or instrumental rationality. Thus, the “people rank them” is really an add-on that is not fundamental to RC but is often added because of its explanatory convenience, and then unfortunately people then erroneously think it epitomizes RC when it is really an add-on. In fact, the “people rank them” is where I often see the hang-ups with people who don’t like RC. After all, people can be consistent (even stochastically so) in following conventions or in the way they fall prey to other behavioral biases. In short, the biggest complaint that I see people have about RC is really not about something fundamental to RC but is rather an appendage to RC.

    Second, I’ve never seen anyone claim that barter is the embodiment of our modern notion of rational action, which, by the way, is not how Adam Smith conceived of rational action. Yes, I agree with the main point that barter exchanges as conceived in modern models is not historically representative, but that is in part because I think these models look at impersonal exchange with strangers whereas (I think, and please correct me if I’m wrong) the generalized exchange historically was not with strangers. This is a big difference that doesn’t itself undermine how money improves upon barter in impersonal exchange settings. It does, of course, make a good case that the impersonal exchange benchmark is a funny one.



    December 15, 2011 at 7:34 pm

  3. Not so fast, “Mike.” Utility functions – which are the whole point of RCT – have implicit rankings. just take the inverse utility function and you can recover ranking. anytime you have utility functions, you always have rankings, even if they are degenerate in some cases (e.g., indifferent between all options).



    December 15, 2011 at 7:38 pm

  4. Dan,

    I think my comment applies to your comment as well. Criticizing RC based on actors’ knowledge is again assuming instrumental rationality, which is not core RC but is an appendage.

    That is why predictive power of RC models must be placed in context. Core RC is a methodological approach and is thus not refutable. Adding certain assumptions to RC, such as assuming a certain preference structure is a way to use the RC methodology, but empirical studies that refute claims made using an RC approach do not undermine RC overall, they just provide evidence contrary to the particular application of RC.



    December 15, 2011 at 7:39 pm

  5. I think I’ve got you here, too, Fabio. Utility functions do not necessarily represent instrumental rankings. Ultimately utility functions only represent the consistency in action. The proper way to understand a utility function is not that a person has a ranking that she maximizes. The proper way is that the individual acts consistently so that it is AS IF the person was maximizing preferences that can be represented by the utility function. Again, no cognition or instrumentality is necessary.

    Think about revealed preference. All you need is choices to be consistent in a way so that it is AS IF the person has preferences that she is maximizing.

    The idea of preferences is a story that economists use because it lends itself to folk psychology explanations of the sort “person i likes X more than Y and so chooses X.” But that is just a story.



    December 15, 2011 at 7:45 pm

  6. @Mike –
    Fair enough. Another Jim Johnson argument is that RC should be understood as a research program, not a single theory. You have evolutionary game theory on one side, and neoclassical economics on another, with strategic game theory in between (and lots of other variations). The assumptions about knowledge, intentionality, content of preferences, etc. vary tremendously. So, “RC” has no predictive power because it’s an approach not a theory. Individual RC theories may have predictive power, but often at the cost of unrealistic, assumed preferences, that may not be stable across time and context (but are often described as such). The slipperiness is when RC proponents justify particular RC theories by defending the larger RC framework.

    For reference, I really recommend Johnson 1996 How Not To Criticize Rational Choice, which is a review/critique of Green and Shapiro’s Pathologies of Rational Choice.


    Dan Hirschman

    December 15, 2011 at 7:45 pm

  7. To be clear, I think the subtlety I’m bringing up is lost on most economists. Some of them mistakenly think that being rational is equal to being selfish.



    December 15, 2011 at 7:46 pm

  8. Be careful, Mike. “Instrumental” has a slightly different meaning in sociology than it does in other fields. In soc, instrumental means “toward an external goal” rather than expressive, which may not be goal directed. E.g., the person may pray for instrumental reasons (they want to impress others in church) or expressive reasons (they are pious).

    You are saying that RCT does not assume that people explicitly rank options, which I agree; I was sloppy for having said so in the post. However, option rankings are implicit in RCT, even if they are not arrived at through cognition.



    December 15, 2011 at 8:20 pm

  9. PS. Mike, unrefutable is a very dangerous word around here…



    December 15, 2011 at 8:21 pm

  10. @Mike:

    I got hung up on the ranking thing and overlooked your main issue – barter as the embodiment of rational action. Graeber, I think, is on target. I’ve been in many classes where this line was pushed. I have said it myself in lecture thinking it was the truth. I have heard many people cite Smith’s passage in WoN as evidence that barter is normal and that money is an evolved improvement over barter.

    It might be the case that specialists in money and credit within economics do not believe this. Or economists who study non-monetized economies. One weakness of Debt, which I’ll likely discuss later, is no discussion of specialists, just Adam Smith and a few others. Maybe money and credit specialists have a different take or economic historians.

    But this is not a trivial matter. The standard story that is often given for various institutions is that they are functional responses to problems (e.g., barter leads to money because money solves barter problems). Graeber is pushing a different angle – power struggles lead to settlements, which are regulated through economic institutions. The reason I started off with RCT is that it is an example of something presented as irrefutable or obvious when it is not. Also, a lot of textbooks will start off with stories, which turn out to be patently false, to motivate various theories.

    In an odd way, there’s a kinship between Graeber and Coase. Each attacks a “just so” story in economics that underwrites massive theory. For Coase, it’s the light house, and for Graeber, it’s barter.



    December 15, 2011 at 8:55 pm

  11. Good to make me clarify, Fabio.

    I meant instrumental as goal-oriented. Goal orientation is not fundamental to RC. It is commonly added to action consistency in RC theories, but it is an add-on that need not be made. Having a utility functions does not imply goal-orientation. Consistency implies utility function.

    And by unrefutable I meant the common sense notion of falsifiable (which of course is problematic philosophically). I definitely didn’t mean that it was immune to criticism. I am critical of many applications of RC, but am less critical of the RC methodological approach. The criticisms I’ve read of RC often mistakenly attack RC itself rather than attacking the application.



    December 15, 2011 at 9:05 pm

  12. But be careful extrapolating from small samples. I too have seen money presented as a solution to barter, but this story doesn’t even come up in many econ courses, and so my problem is in using this one example to make a more sweeping statement about RC. The RC approach and economics can stand quite well whether or not money is a solution to barter.

    But the distinction between impersonal and personal exchange is crucial. The models of money I know focus on the former, and this is where I think economists are getting it wrong. Yes, money will solve barter problems in the former, but the book is saying that that’s not the interesting question because exchange didn’t arise in impersonal settings.



    December 15, 2011 at 9:25 pm

  13. Mike, the whole point of debt is to argue that historically, barter *does* not exist. So it can’t be the case that the reason money was invented was to eliminate barter. And this is an argument that is not limited to a small sample of people.

    Seriously, ask most folks who are intellectuals: why was money invented? The modal answer will be a version of the barter story. Perhaps, economists have a different answer, but I have not heard it. Graeber even cites a number of well read econ textbooks (and Adam Smith himself) showing that lots of people believe the functionalist explanation of money.



    December 15, 2011 at 9:28 pm

  14. Fabio, I think we may be talking past each other. I don’t disagree with the book’s main point that there is no historical barter and hence money didn’t emerge to solve it. That’s why I would say the money models are likely misspecified. Extending this point about barter to make a larger point about RC or economics as an entire discipline today is where I take issue.

    BTW, you can generate laboratory settings where debt is prohibited and where money emerges to promote exchange. So, theoretically, money can emerge in a way that seems to match the functionalist explanation. However, the subjects are, of course, exposed to modern market life with money and hence likely pre-disposed to finding something to function as money, and this weakens the argument that the functionalist explanation is correct. And I take the point of the book as making an important point that these settings are missing the crucial historical role of debt.



    December 15, 2011 at 11:37 pm

  15. However, the subjects are, of course, exposed to modern market life with money and hence likely pre-disposed to finding something to function as money, and this weakens the argument that the functionalist explanation is correct.

    I guess when you’re seeking to demonstrate that something can emerge spontaneously as a solution to a problem, it may help to populate your experiment with subjects who already know what the solution to the problem is.



    December 16, 2011 at 2:05 am

  16. I haven’t read this book, but I wonder the following from your post. Why does the book not consider debt of the type you discuss here under the term generalized exchange, barter over time? It is, after all, an exchange (I give you this, you give me that). The only difference is that it occurs in two time periods, the present and some unspecified future date. It seems to offer a solution to one problem of barter (the person with whom I must trade has nothing I currently wish to trade for) that creates a second problem (the person to whom I extend credit today may refuse to repay me in the future).

    It is possible, then, that The invention of money solves this problem by transferring the risk of default from an individual where it is hard to evaluate, to whomever is issuing the currency, typically a sovereign or an institutionalized entity.

    Maybe the book covers this?



    December 16, 2011 at 12:26 pm

  17. I have no brief for neo-classical economics; my concern is with Adam Smith’s legacy, usually placed within Classical economics. I am working through ( a critique of Dr Graeber’s well-written and interesting book on “Debt”, so I will not rehearse my critique so far.

    Paragraph 1 above raises a non-issue: “Why pile up on specialized goods and wait for other people to pile up on what you want and then trade? That’s bizarre.” No, it is the image that is “bizarre”. Barter is inconvenient because specific surpluses may not coincide with other’s people’s surpluses. It is from an accidental surplus that may arise in the course of daily life that a person at least theoretically may seek to exchange them for something more useable by her. But if nothing is available, “no exchange”, says Smith, can be made” (WN I.iii.2: p 37). He reports that that “things were frequently … valued according to the number of cattle and women which had been given in exchange for them” (p 38). The difference on this supposed transaction between Smith and Graeber is that the latter uses evidence to suggest that actual barter exchanges did not in fact take place. However, Dr Greaber says the male-dominated theory was that the value of human life was immeasurable, yet life-long debt servitude, in terms of cattle and sex slavery, were a common consequence.

    Dr Graeber’s point that the evidence does not show a mere transition from barter to money must be well taken (and I do). Smith did not have access (nor did anybody else in the 18th century) to the post-1850s literature and the research of thousands of anthropologists that Dr Graeber has at the click of his laptop. But Smith was right that the “inconveniences” of “non-coincidental demands”, whatever their cause, were resolved eventually by the invention of money from c.3,000 BC (BCE)..

    Smith’s asserted that the “propensity to truck, barter and exchange” originated “probably with the acquisition of reason (not rationality!) and speech”. (WN I.ii.1: p 25), References to the Oxford English Dictionary settle his meaning. Multiple references to Dr Graeber’s fascinating descriptions of the forms of exchange in place in ancient societies mostly involved the consequences of the institutional tyrannical disposition to control the exchange of women and children. I found these descriptions somewhat distressing in terms of humanity. Dr Graeber named these social systems as the age of “Human [!] Economies”

    Dr Graeber conflates modern theories of rational utility maximization (at best merely mathematical equations) and all that follows from them, as being representative of modern economics, which he is most welcome and justified to critique, but for some reason he believes, wrongly, that they were the views of Adam Smith, which they most certainly were not. None of these abstract mathematical structures developed since 1870 are representative of Adam Smith’s moral philosophy and political economy (1759 and 1776), hence I strongly disagree with Dr Graeber’s assertion that “generalized exchange can be described in terms of utility functions”. The central point in all of Adam Smith’s philosophy was that “exchange” was, and is, a “universal”, common across the human species since it formation, and “the necessary consequences of the faculties of reason and speech”.

    My exposition of the role of exchange in the human species, perhaps unique to it, could be expressed, to coin a phrase, as: “Exchange: the first 200,000 Years”.


    Gavin Kennedy

    December 16, 2011 at 2:16 pm

  18. Let me try to rephrase what Mike is saying here, because I had the same reaction when I read Graeber. Fabio makes a two-part argument:

    (1) Graeber convincingly shows that history has never seen a real barter economy (exceptions being when developed nations with established monetary systems collapse). In doing so, he shows that the story told in many economics texts about the origin of money is a myth.

    (2) This undermines some foundation of neoclassical economics.

    I think everyone here, including Mike, agrees with (1). The problem is that it’s not clear exactly how (2) follows from that. Graeber makes a sort of hand-waving argument suggesting that he’s somehow undermined the “founding myth” of neoclassical economics. But it’s not at all clear what part of the neoclassical model depends on this myth. (And I say this as a harsh critic myself of neoclassical theory.)

    I don’t think Graeber ever properly makes the link between (1) and (2), so it’s perhaps unfair to ask Fabio to do so. But, Fabio, you seem to have been convinced by Graeber’s argument, so could you say more about what particular foundation has been undermined?



    December 16, 2011 at 2:53 pm

  19. Thomas,

    Graeber is very explicit about this and it’s an important part of his argument. His whole point is that double coincidence of wants is only a problem if we assume spot market transactions among people who will never see each other again. (And indeed, he agrees that barter does happen under those circumstances). More broadly, one of the central arguments of the book is that market exchange is only made possible by disembedding people and objects from their social contexts.

    In contrast, in your typical primordial gemeinschaft type of place there’s a lot of repeated interaction and so (just as you suggested) delayed reciprocity is the typical form of exchange.



    December 16, 2011 at 5:11 pm

  20. As a preface, let me say that I’ve only read excerpts from the book, and so may not be properly addressing the full argument, but I think I can provide some evidence on what specialists in money and credit within economics have suggested regarding the role of money, in order to evaluate the claim that the lack of historically observed barter economies refute a central or even important claim in neoclassical economics.

    First, if you talk to (neoclassical) economists who are not specialists in money and credit, although they may give some sort of Smithian explanation regarding the role of money, the models that they write down and the way they think about the economy will primarily be in terms of non-monetary exchange but will in no way be based on barter. The Arrow-Debreu general equilibrium model, arguably the heart of what is referred to as neoclassical economics, contains no money but is rather based (perhaps implicitly) on debt claims. The existence of a “price” in this model, rather than a nod to the role of money as possible unit of account, is given by ratios of exchange, in which all goods (including “goods tomorrow”) are fungible due to the common knowledge of these ratios. To be clear, although there exists a “price”, it is manifestly not in nominal terms, but only a ratio of, say, spears to sheep and sheep to cows at which all exchanges are implicitly made. There is no “barter problem” in such a model, and intertemporal exchange is based on something very much like what is described as debt claims.

    If you talk to specialists in money and credit in economics, you might or might not be surprised to hear that extremely little attention is given to the historical emergence of money. This is not to say that there is not quite a bit of thought about the role that money plays, but that such explanations somewhat rarely take the step from explaining a function for money to asserting that there existed historical time periods where people decided to use money because of such a function. One can discuss the question of why people now value money without getting into historical details. Economists are very much concerned with the determinants of the valuation of money (why hold dollars when there exist other financial claims which are as trustworthy and have higher real yields), and this is the basis of arguments based on the exchange value of money. The argument is that (at least some) exchanges are facilitated by the use of money, and for this reason people today value money. Whether this was the case historically is irrelevant if one wants to go on to use this theory to try to understand, for example, the effect of monetary policy on prices or interest rates.

    There is a small literature which goes further and attempts to explain the reasons for which money facilitates exchange, and of this only a small portion corresponds to the traditional “double coincidence of wants” spot barter exchange story (the main reference here is Kiyotaki and Wright), relying on the informational friction (absent from Arrow-Debreu type economies) of anonymous one-time bilateral exchange. This is a literature which, while attracting some notice, has had quite limited impact on monetary economics in practice, as few find the assumption that trade is limited to anonymous one-time relationships to be a reasonable description of current practice, whether with money or not. In other words, what Graeber claims is a central claim of neoclassical economics is in fact a fringe position within the field. More generally, other theories of the role for valuation of money highlight the role of networks of trust (see Kiyotaki and Moore “Evil is the root of all money,” for a much more eloquent argument on these lines) and the role money has in recording and formalizing implicit debt contracts (see Kocherlakota “Money is Memory”). Generally speaking, it is quite accepted in mainstream economics that exchange can be facilitated by debt contracts without explicit recourse to money, and that while money facilitates certain kinds of anonymous multilateral exchange, such exchange can exist even without it. For a striking example, see Woodford “How important is money in the conduct of monetary policy?” Journal of Money, Credit, and Banking 2008, for an argument by probably the most influential living monetary economist that “money” as such is of negligible importance even in monetary policy.

    Personally speaking, I did find Graeber’s discussion of the role of indirect exchange fascinating and it did cause me to reassess the weight placed on anonymity as a practically important limitation in historical economies, but to say that it refutes a claim which is central to neoclassical economics seems to me to be something of an overstatement.



    December 17, 2011 at 2:55 pm

  21. I have very little to add to this enlightening post and comment stream save to note that it is a commonplace in economics that ‘money is debt’ (which by do means implies the reverse) and that debt is a socially constructed artifact.


    Fred Thompson

    December 30, 2011 at 10:33 pm

Comments are closed.

%d bloggers like this: