post-autistic economics review
Issue no. 21,  13 September 2003
article 1



      issue 21 contents                            PAE Review index                               home page




Seven Theses for a Theory of Realist Economics*
Part I: Theses One to Four
(Part II: Theses Five to Seven will appear in the next issue)

Jacques Sapir1   (L'École des Hautes Études en Sciences Sociales, Paris)  

© Copyright 2003 Jacques Sapir



The issue of realism has been central to the PAE movement from its beginning.  As I have previously stated in this journal and elsewhere, for me realism is not the opposition between a "factual" world and a "theoretical" one, between reality and abstraction.  Instead for me realism is both a methodological stance and the definition of a theoretical research program.

Realism however can give rise to different interpretations.  Uskali Mäki has made an important distinction between world realism and truth realism.2  This distinction nevertheless raises the issue of what we understand as being the "real world", and there is here a kind of fast-lane to positivism.

I agree with Tony Lawson’s distinction between events and processes.3  A process, a notion central to the works of Marx and Keynes,4 is understood here not as a sequence of events but as "...the genesis, reproduction and decline of some structural mechanism or thing, the formation, reformation and decay of some entity in time".5  This realism is completely different from empirical realism, which takes for granted the notion that any human agent can have a direct, non-mediated access to reality.

‘Realism’ as I use the word is both procedural and subjectivist.  Subjectivism does not mean that human subjectivity is the only possible reality, a fallacy commonly found in some post-modern authors, but that subjective views of reality, as far as they shape human decisions, are part of reality.  Realism will then define methodological constraints for economists.  That does not mean that economics must have a specific methodology, which is the position of mainstream economists defending Friedman's instrumentalism, but rather that the methodological requirements for social science can have distinct applications for economics, with specific methodological rules for conducting enquiries or for story-telling.6  Elsewhere I have described what such applications in the methodology of economics could be.7  Realist economics does not bear kindly theoretical tinkering or ad-hoc arguments.  There are, as I have explained before in this journal, limits to pluralism.8

A coherent research program needs to be developed for a realist economics.  To this end I offer the following seven theoretical theses.

Thesis 1: The central issue in economics is the co-ordination of decisions and interactions generated by decentralised, heterogeneous and interdependent agents whose decision-making abilities are constrained by limited cognitive capacities.

In the real world, in the theoretical sense of this word, decision-making is done in a decentralised way.  Not to acknowledge this fact is to reduce human agents to the status of mere parts of a giant machine, the issue then being who is the power behind such a machine, God Almighty, the market auctioneer (pace Walras), the Party general secretary or the mainstream economist himself.

But human agents are not only decentralised, they also are heterogeneous.  Not to acknowledge heterogeneity, as when one assumes identical decision-making patterns and initial positions or a single commonly shared rationality principle, transforms the community of human agents’ into a world of clones.  If this were really the case there would be no sense in talking about decentralisation even in a politically free society.

The decentralisation principle is then largely grounded on the refutation of the possibility of a single rationality principle which could be shared by all agents, everywhere, always and under every possible condition.  Daniel Kahneman and his colleagues, the late Amos Tversky especially but also Richard Thaler, Paul Slovic and Sarah Lichtenstein to name just a few, have made this refutation.9  The reluctance of mainstream economists to acknowledge these scientific results - a paradoxical position for a group professing fondness for the Popperian legacy - betrays their unwillingness to accept true decentralisation, whatever they may say about possible different initial human and material resources allocations to individuals.  Heterogeneity is a necessary concept for understanding decentralisation.  Ultimately heterogeneity means not just that situations can be different and thus also the social positions from where decisions are made.  This is heterogeneity in its descriptive sense.  In a more analytical sense heterogeneity derives from the fact that patterns of decision-making, models of rationality - here to be understood as the simple fact of having a reason for doing something - are different. Heterogeneity is not exogenous to the decision-making process, something that a dedicated policy could eradicate, but instead something at the very heart of this process.

The interdependency of decentralised and heterogeneous agents must be understood.  The standard economics theoretical tradition emphasises the Robinson Crusoe metaphor, negating the interdependency issue, and envisions the social process from the point of view of a completely isolated individual.  Against this tradition, realist economists conscientiously put the issue of possible unintentional effects of individual decisions on other agents at the very centre of economic activity and as part of social life.  Here they reclaim both Hayek's legacy, at least the one coming from The Constitution of Liberty,10 and the Durkheimian one with its concept of social density.11  This last, that the web of intentional and unintentional relations and the perceptions related to them is the real place where decisions are made, was developed by Emile Durkheim in his seminal work on the social impact of the division of labour.12

To jointly acknowledge decentralisation and interdependency implies a switch from the allocation paradigm to the co-ordination one.  Co-ordination can be achieved through intentional processes (networks and hierarchies) as well as through unintentional ones (markets).  But whatever the process one thinks fits best at a given time and for a given problem, decentralisation is the central issue.

Anti-realism as a methodological strategy supported by mainstream economists does not stop with rejection of heterogeneity and/or interdependency.  Perfect information, as in the initial Walrasian model or as in the rational expectations theory, is part of such a strategy.  Refutation of the perfect information assumption can be epistemic.  Simon and de Groot have shown that even if a perfect information structure could exist, our cognitive capacities preclude us from computing in a time short enough for this structure to be of actual use for our decision-making process.13  But refuting the perfect information assumption can also be ontological.  Perfect information could be an unreachable goal because the real world is too complex to be understood - the classical Hayekian understanding of uncertainty - or because our own attempts to gather more information are generating endogenous modifications of the information structure (Stiglitz, Akerlof).  Uncertainty is then not an exogenous addition but is endogenously generated.  This understanding of uncertainty puts the asymmetrical information school on the right side of the methodological realism border when compared to the information search school (Stigler).

One has to add that if we agree with the fact that there can not be a single and common rationality principle then the rational expectations theory is devoid of any logical basis.  Whatever the reason for endogenous uncertainty, this assumption is another defining characteristic of mainstream or non-realist economics.  It is so as to deny uncertainty that neo-classical economics pretends to give to profit and price a natural law dimension.14

A common attribute of varieties of non-realist economics, whether because they refuse to acknowledge heterogeneity or interdependency or endogenous uncertainty, is their denial of the relevance of time and money.  Realist economics, on the other hand, stresses time and money relevance.  Time is relevant as a causal factor,15 something which was understood quite early by Gunnar Myrdal who pointed to the relevance of the ex-ante / ex-post divide in the perceptions of economic agents,16 and by the classical institutionalist school with its first mover / second mover paradigm.17  Time is also relevant, as a delay between decision and effects or between a demand and a supply response, as clearly understood by Mordecai Ezeckiel a long time ago.18

Money is a necessary institution for co-ordination.  It generates the illusion of homogeneity that agents need to make complex decisions on the basis of their limited cognitive capabilities and because, by allowing for the separation between income formation and income utilisation, money makes possible a better use of time.

Thesis 2: If money is a necessity in an uncertain world, money also introduces a specific form of uncertainty, casting doubts on the market’s ability to efficiently process information.

In a world devoid of uncertainty money would not matter.  But money gives to every agent in an uncertain world the ability to shelter himself in liquidity.  Liquidity in turn allows every agent to defect from the long and continuous chain of interdependencies generated by the division of labour.  This very possibility of defection introduces a new strategic uncertainty which is at the heart of economic decision-making in money-based economic systems.  Actually there is a deep interaction between uncertainty and the flight to liquidity, which in turn generates this strategic uncertainty.  This was perfectly described years ago by G.L.S. Shackle:

"When knowledge seems especially elusive, we desire money rather than specialised, vulnerable assets.  We sell the assets, their prices fall and it becomes no longer worthwhile to produce them, no longer worthwhile to invest, to give employment.  Had Keynes attended to Cantillon, he could have freed himself from the proposition that an employer will always offer a wage equal to marginal product of value of his body of employed people.  For since he must employ people first and sell their product later, he cannot know for sure what their marginal product is going to be".19

Hyman Minsky has shown how financial innovation, as burgeoned during the second half of the XXth century, could be deeply destabilising.20 From Marx to Keynes, realist economists have analysed how the flight to liquidity should put crisis - not equilibrium - at the centre of economic thinking.  Crisis is the permanent horizon of a capitalist economy because either liquidity is too much in demand or is not wanted at all.  The specific uncertainty generated by liquidity pushes economic systems toward under-investment and under-employment.  This uncertainty can not be managed by economic computation and can be called radical uncertainty.

Here we are facing the first paradox of money.  As an institution money pretends to solve the heterogeneity problem by setting monetary prices as a common norm for decision-making, something which makes the deepening of the division of labour possible.  However by doing so money generates the radical uncertainty which constrains the expansion of the division of labour.

A second paradox of money is that as an institution it would seem to unify time through interest rates and its function as a reserve of value.  But money, through its liquidity function, contributes greatly to making the future even more uncertain.

The twin paradoxes of money stress the fact that if monetary prices are a necessary fiction, from the realist economics point of view, they nonetheless are a fiction. That was precisely what Max Weber tried to show when explaining that monetary prices are necessary in a decentralised economy but are not the result of demand and supply equilibrium - as pretends capitalist spontaneous collective thinking.  Monetary prices actually reflect the balance of power between social or individual forces and interests.21  Keynes, in one of his first works, wrote something very similar.  He explained that inflation and deflation translated into the monetary world social conflicts opposing large, structured social groups.22

However if monetary prices are a necessary fiction they also are an uncompleted one.23 They are unable to carry the whole range of information needed for decision-making.  Because we need information which can not be conveyed through monetary prices and which belong then to different information spaces, our decisions are situated and embedded in multidimensional worlds.  One consequence is that the transitivity of  individual preferences is broken in a systematic way.24  Then the Allais' Paradox holds true,25 and we can forget the subjective expected utility theory and every device invented by mainstream economics to transform the static Walrasian world into a dynamic one and to cope with uncertainty (even in a Bayesian form).  A second consequence, as demonstrated by Grossman and Stiglitz, is that in such a situation, where prices do not convey all needed information, competitive markets are not informational efficient.26

Thesis 3: Time and money are at the very heart of the interchange between the individual and collective levels.

Time matters, inter-allia, because of the time constraint: the more we wait before making a decision the more we lose even if our decision is the perfect one.  However the time constraint has not the same meaning for individuals and groups.  Our decision tempo is largely shaped by our more or less deep insertion into collective groups, from the family to the enterprise, including social and political organisations.  In turn, the way collective groups are institutionalised shapes also their impact on our individual use of time and our sensibility to the time constraint.

The power that money gives, particularly as liquidity, is not used in a vacuum of representations.  Kahneman and his colleagues have demonstrated that our individual preferences are shaped, or more precisely "framed" by collective contexts.27  But the way I use my liquidity power could affect decisively some collective groups to which I belong, even if I have no idea of this fact.  A bank-run, even if induced by misguided collective representations, is a movement of thousands of individuals who try to protect their savings but, by doing so, usually destroy most of the economic context supporting collective groups (enterprises) from where their income is generated.

Any attempt to seriously make time and money relevant, from a theoretical point of view, amounts to repudiating methodological individualism.  But because time and money relevance comes from interdependency and from social density, we also have to repudiate the idea of a single dominating collective context.  If realist economics embraces methodological holism it is a non-deterministic holism. 

Thesis 4: Any attempt to negate the theoretical status of time and money leads to non-scientific assumptions and transforms the economist himself into a producer of ideology.

Being serious about time and money places economics in the very middle of the social sciences.  If statistical regularities and stabilities are to be found, they are not the products of intemporal laws but of social systems of institutions.  The stability of these systems is itself a local and temporary phenomenon.  On the other hand if one wants to ground economics on laws similar to ones found in natural sciences, in physics or mechanics, one has to negate time and money relevance.  Such a strategy is logically coherent if and only if one negates either decentralisation or interdependency.  Both are radical retreats from realism.

Here we have one of the most fanciful paradoxes of mainstream economics.  To reject realism for axiomatics, mainstream thinkers have to invoke ergodicity.28  But to pretend that economic processes could be in any sense a kind of ergodic process, one has to demonstrate that they are subject to a determination which is non-human (thereby violating the initial assumption of decentralised decision-making) and non-social.  Obviously the standard theory of individual preferences and its conclusion, the closed and universal model of rationality, fit nicely here.


Traditional assumptions about individual preferences (transitivity, continuity, reflexivity, independence and time monotony) are then just not ad-hoc assumptions but the logical core for any instrumentalist methodology grounded on preference utilitarianism.29  They provide the stable, non-social, reference point needed to pretend that observable local economic stabilities are like the exposed tips of yet unknown "natural" laws of economics.

It happens to be the case, however, that all these axioms can be tested and when they are they are invalidated.30  Facing such results most mainstream thinkers pretend they are irrelevant.  They dismiss the very idea of confronting an economic theory with real life experiments.31  By doing so they fail to understand that they can claim legitimacy for the  axiomatic approach if and only if they can find empirical grounds for their ergodic assumption.  What psychology has done is no less than to destroy the only substantial argument for ergodicity, that is the universality and stability of the neoclassical model of rationality.

The willingness to integrate into economic theory the findings of applied psychology versus the refusal to do so is the true borderline between economics as a scientific activity and economics as production of ideology.  The defence of axiomatism clearly no longer belongs to any kind of scientific approach to economic phenomenon but instead is a form of religious thinking.

In contrast to the we-do-not-want-to-know approach, George Akerlof has succeeded in integrating recent psychology results to a theory of inflation, which is clearly Keynesian.32 Akerlof’s writings are living proof that Kahneman and Tversky works can be solid ground for Keynesian assumptions, particularly when it comes to money and time.33


* This paper is a translation and adaptation of one that appeared in the French journal Alternatives Économiques (n° 57, 2003, hors série, pp. 54-56, see also and is published here with authorisation of the journal’s editorial board. The initial aim was to review assumptions developed in an earlier book, Les trous noirs de la science économique (Albin Michel, Paris, 2000) and to specify some details that could be of use for the PAE readership.

This book was published in the very middle of the battle following the French students appeal for more realism in the teaching of economics (spring 2000) and sold quickly, being re-printed twice before its forthcoming pocket edition,  September 2003. This coincidental publishing was a pure stroke of luck. The book was  written between 1995 and 1998 when I was teaching at the Vyshaya Shkola Ekonomiki (Higher School in Economics - Moscow). From lectures delivered in Moscow I wrote first a basic book for Russian students (K Ekonomicheskoy teorii neodnorodnykh sistem - opyt issledovaniya decentralizovannoy ekonomiki - Economic theory of heterogeneous systems; an essay on decentralised economies) which was published by Vyshaya Shkola Ekonomiki Press, Moscow, in 2001. At the same time I re-focused and expanded part of its content to write Les trous noirs, this time not as a basic book but as a critical essay on mainstream economics. This second book is not then the translation of the Russian one, although they are closely related.

I have adapted and developed here the arguments of the Alternatives Économiques paper for the sake of an English language readership not necessarily aware of debates currently raging in Paris.

1. Professor of economics, EHESS-Paris, director CEMI-EHESS.

2. U. Mäki, "How to combine rhetoric and realism in the methodology of economics" in Economics and Philosophy, vol.4, avril 1988, pp. 353-373.

3. T. Lawson, "Realism and instrumentalism in the development of econometrics", in Oxford Economic Papers, vol. 41, janvier 1989, pp. 236-258.

4. For the latter, A.M. Carabelli, On Keynes's Method, Macmillan, Londres, 1988.

5. T. Lawson, Economics & Reality, Routledge, London & New York, 1997, p. 34.

6. C. Lloyd, Explanations in Social History, Basil Blackwell, Oxford, 1986.

7. J. Sapir, "Calculer, comparer, discuter: apologie pour une méthodologie ouverte en économie", in Économies et Sociétés, série F, n°36, 1/1998, pp. 77-89.

8. J. Sapir, "Realism vs. Axiomatics" in Edward Fullbrook (ed.), The Crisis in economics, Routledge, London & New York, 2003, pp. 58-61.

9. For a now quite old review of this literature see J. Sapir, "Théorie de la régulation, conventions, institutions et approches hétérodoxes de l'interdépendance des niveaux de décision", in A. Vinokur (ed.), Décisions économiques,  Économica, Paris, 1998, pp. 169-215. Also: A. Tversky, "Rational Theory and Constructive Choice", in K.J. Arrow, E. Colombatto, M. Perlman et C. Schmidt (eds), The Rational Foundations of Economic Behaviour, Macmillan and St. Martin's Press, Basingstoke - New York, 1996, pp. 185-197.

10. F. Hayek, The Constitution of Liberty, University of Chicago Press, Chicago, 1960.

11.  E. Durkheim, Les règles de la méthode sociologique, PUF, coll. Quadriges, Paris, 1999 (1937).

12. E. Durkheim, De la division du travail social, PUF, coll "Quadrige", Paris, 1991 (1893).

13.  A. de Groot, Thought and Choice in Chess, Mouton, La Haye, 1965.  De Groot's work has been much used by Herbert Simon. See H.A. Simon, "Theories of bounded rationality", in C.B. Radner & R. Radner (eds.), Decision and Organization, North Holland, Amsterdam, 1972, pp. 161-176.

14.  G.L.S. Shackle, "The Origination of Choice", in I.M. Kirzner, (ed), Subjectivism, Intelligibility and Economic Understanding , Macmillan, Londres, 1986, pp. 281-287.

15.  M. Capek, The Philosophical Impact of Contemporary Physics, Van Nostrand, Princeton, 1961. G.P. O'Driscoll Jr. and M.J. Rizzo, Economics of Time and Ignorance, Basil Blackwell, Oxford, 1985, pp. 60-61.

16. G. Myrdal, Monetary Equilibrium, W. Hodge, London, 1939, pp. 43-44.

17. W. M. Dugger, “Transaction cost Economics and the State”, in C. Pitelis, (ed.), Transaction Costs, Markets and Hierarchies, Basil Blackwell, Oxford, 1993, pp. 188-216.

18.  M.Ezekiel, "The Cobweb Theorem", in Quarterly Journal of Economics , vol. LII, n°1, 1937-1938.

19. G.L.S. Shackle, Business, Time and Thought. Selected Papers of GLS Shackle, New York University Press, New York, 1988, p. 43.

20. H.P. Minsky, Stabilizing an unstable economy, Yale University Press, New Haven, Conn., 1986.

21.  M. Weber, Economy and Society: An Outline of Interpretative Sociology, University of California Press, Berkeley, 1948, p.108. See also about the nature of money in chapter II in the first part of Wirtschaft und Gesellschaft , translated as M. Weber, The Theory of Social and Economic Organization, the Free Press, New York, 1964.

22. J.M. Keynes, "A tract on Monetary reform", reprinted in J.M.Keynes, Essays in Persuasion, Rupert Hart-Davis, London, 1931

23. C. Deutschmann, "Money as a Social Construction: On the Actuality of Marx and Simmel", in Thesis Eleven, n°47, novembre 1996, pp. 1-19.

24.  This is clearly a consequence of the Endowment Effect. See, D. Kahneman, J. Knetsch et R. Thaler, "The Endowment Effect, Loss Aversion and Status Quo Bias" in Journal of Economic Perspectives , vol. 5/1991, n°1, pp. 193-206.

25. M. Allais, "Le comportement de l'homme rationnel devant le risque. Critique des postulats de l'école américaine" in Econométrica, vol. 21, 1953, pp. 503-546. Also: M. Allais & O. Hagen (eds.) Expected Utility Hypotheses and the Allais Paradox, Reidel, Dordrecht, 1979.

26. S. J. Grossman & J. Stiglitz, “Information and competitive price systems”, American Economic Review , Vol. 66, n° 3, May 1976, Papers and Proceedings of the Annual Meeting of the American Economic Association.

27. The framing effect is well described in D. Kahneman, "New Challenges to the Rationality Assumption" in K.J. Arrow, E. Colombatto, M. Perlman et C. Schmidt (eds.), The Rational Foundations of Economic Behaviour, St. Martin's Press, New York, 1996, pp. 203-219

28. For an enlightening analysis about how not to use the ergodicity concept see P. Mirowski, More Heat than Light, Cambridge University Press, London - New York, 1989. See also : B. Ingrao et G. Israel, "General Equilibrium Theory : A History of Ineffectual Paradigm Shifts" in Fundamenta Scientiae, vol. 6, 1985, pp. 1-45 et 89-125; P. Mirowski, "Energy and Energetics in Economy Theory" in Journal of Economic Issues, vol. 22, 1984, pp. 811-830.

29. C. Harsanyi, "Morality and the theory of rational behaviour", in A. Sen et B. Williams, Utilitarianism and Beyond, Cambridge University Press & Éditions de la Maison des Sciences de l'Homme, Cambridge & Paris, 1982, pp. 39-62.

30. D. Kahneman, "New Challenges to the Rationality Assumption" in K.J. Arrow, E. Colombatto, M. Perlman et C. Schmidt (edits.), The Rational Foundations of Economic Behaviour, St. Martin's Press, New York, 1996, pp. 203-219.

31. For a good example of such an argument, see M. Friedman, "The Methodology of Positive Economics", in M. Friedman, Essays in Positive Economics, Chicago University Press, 1953, p. 30-31.

32. G.A. Akerlof, W.T. Dickens, G.L. Perry, (1996), "The Macroeconomics of Low Inflation" in Brookings Papers on Economic Activity, pp. 1-76 ; Andersen T.M., (2001) "Can Inflation Be Too Low?" in Kyklos, vol. 54, Fasc.4, pp. 591-602.

33. G.A. Akerlof, "Behavioral Macroeconomics and Macroeconomic Behavior", in American Economic Review, vol. 92, n°3, juin 2002, pp. 411-433, p. 424 ; the source quoted here is D. Kahneman et A. Tversky, "Prospect Theory: An Analysis of Decision Under Risk" in Econometrica, vol. 47, n°2, mars 1979, pp. 263-292.

Part II: Theses Five to Seven will appear in the next issue.



Jacques Sapir, “Seven Theses for a Realist Economics; Part I: Theses One to Four”, post-autistic economics review, issue no. 21, 13 September 2003, article 1,