The Imagined Market
- Machine Dreams: Economics Becomes a Cyborg Science by Philip Mirowski
Cambridge, 670 pp, £24.95, February 2002, ISBN 0 521 77526 4
I’ve started giving my students money. Not to bribe my way to favourable teaching reviews, but to provoke reflection about the relations between economic and sociological views of human beings. The money is used to play the ‘ultimatum game’. A large class divides itself into pairs, who must not be friends or acquaintances. Each pair collects ten 5p coins. One of the pair proposes a division of the coins, and the other either accepts or rejects the proposal. The game is then over: no negotiation is allowed. If the offer is accepted, the students keep the coins, splitting them as agreed. If it is rejected, the coins are returned to me.
If everyone is a rational egoist – a homo economicus, the dominant construction of the individual in economics – proposers should offer their partners 5p, aiming to keep 45p for themselves. They can be sure respondents will accept, for rejecting the offer means leaving with nothing. In practice, however, proposers mostly offer 25p, with a minority offering 20p. If 5p offers are made (to get a decent number of them I have to infiltrate confederates into the class), the majority are rejected. There’s nothing idiosyncratic about these results: they are typical outcomes of the ultimatum game. Admittedly, 5p or perhaps even 50p is a trivial sum. When the ultimatum game has been played with much larger stakes, however – students in Indonesia have played it for amounts up to three months’ typical expenditure – the results do not differ fundamentally. Even substantial rewards do not provoke rational egoism.
Of course, nuances of interpretation are possible. A homo economicus who knows that his partner might not be a rational egoist can be expected to offer more than 5p: the optimum offer will depend on what he believes the distribution of probabilities of rejection to be (and, when larger sums are involved, the extent of his risk aversion). The microdynamics of the social interaction may be significant: even when playing with strangers, the figure one cuts in their eyes may matter. Nevertheless, a simple conclusion is that players of the ultimatum game are oriented to a norm that enjoins ‘fairness’, and perceived violations of the norm are punished, even at personal cost, by rejection of unfair offers.
If people are not homines economici, should economic theory be rejected as a fantasy based on unsound foundations? The classic riposte came from Milton Friedman, who argued that the test of a theory should not be the realism of its assumptions but the precision and success of its predictions. To assume that people are egoists who each maximise what economists call their ‘utility’ (roughly speaking, their individual satisfaction or pleasure), and that firms are rational entities that seek to maximise profit, allows a sophisticated mathematical apparatus to be brought to bear and testable predictions to be generated. While there is much room for debate about the success of those predictions, other social-science disciplines are certainly in no position to sneer.
A different, more troubling argument for taking economics seriously comes from the French sociologist of science Michel Callon, in The Laws of the Market (1998). Callon argues that economics does not describe an already existing ‘economy’, but helps bring that economy into being. Economics is not a descriptive but a performative endeavour. (Performative utterances – ‘you are now husband and wife’; ‘I name this ship Queen Elizabeth’ – make themselves true by being uttered.) In part, Callon has in mind loosely ‘economic’ practices such as marketing and accountancy, and the Enron scandal has certainly focused attention on how key data such as firms’ earnings are constructed through the practice of accountancy. However, Callon also claims a strong performative role for academic economics.
Friedman, for example, has not simply analysed markets but helped create them, starting in 1972 with the world’s first currency futures exchange, the International Monetary Market. Those influenced by Friedman – most famously the ‘Chicago boys’ in Pinochet’s Chile, but also US, Japanese and British economic policy-makers in the 1980s – reshaped entire national economic systems. And the influence of economics didn’t cease with the arrival of Clinton and Blair, though it changed form. Gordon Brown has been a particularly important channel. His decision to delegate to the Bank of England the power to set interest rates – arguably the single most crucial and most successful decision Labour has made – was underpinned by academic work applying game theory to expectations of inflation (game theory is the mathematical analysis of strategic, rational interactions in which the pay-off to a participant depends both on what he or she does and on what others do).
Performativity is a stronger claim than influence, and economic policy-making has yet to be analysed systematically from this viewpoint. The work of some financial economists has been more clearly performative. Fischer Black, Myron Scholes and Robert C. Merton’s Nobel Prize-winning option pricing equation of 1973 – which I described two years ago as ‘the single most important breakthrough in the modern mathematical theory of finance’[*] – didn’t originally describe the market prices of options very closely. (An option is a contract giving the right, but not the obligation, to buy, or to sell, an asset – such as a block of shares – at a given price on, or up to, a given future date. It is an example of a ‘derivative’, an asset the value of which depends on the price of another, ‘underlying’, asset.) Over subsequent years, however, the fit between the mathematical model and observed prices improved, in part because traders started to use the model to identify, exploit and thus reduce discrepancies between it and market prices. Black, Scholes and Merton’s mathematics, and the many developments of it, became a performative tool: it taught practitioners how to price derivatives and how to hedge their risks. Without it, it is hard to imagine the global financial derivatives market growing, as it has, from close to zero in 1970 to an astonishing $134.7 trillion in December 2001. The latter figure, calculated by the Bank for International Settlements, is equivalent to $22,000 for every person on earth.
Francesco Guala of the University of Exeter has described another instance of performativity: the way in which game theory was drawn on by Government agencies to design the auctions of mobile phone licenses. With firms responding by hiring their own game theorists to plan their bidding strategies, the theory was no longer an external description of the auction, but had become – as Callon would have predicted – a constitutive, performative part of the process. The public purse benefited handsomely (the third generation spectrum auction netted £22.5 billion in Britain alone), though the cost may be a financially crippled industry.
Economics is a world-shaping, not just a world-describing, endeavour. The discipline thus matters not just to economists but to the rest of us, whose world is being shaped. Philip Mirowski is the most stimulating, provocative present-day historian of economics. His book’s abrasive account of the processes forging economic orthodoxy over the past sixty years will annoy his orthodox colleagues, delight dissidents (except the followers of those about whom he is scathing), educate outsiders whose view of the discipline is often decades out of date, and pull the history of economics firmly into the mainstream of the history and social studies of 20th-century science. One of the many virtues of the book is the case it makes for economics without homo economicus.
In previous work, Mirowski examined the original formulation of modern economic orthodoxy: the ‘neoclassical revolution’ of the 1870s and 1880s, led by economists such as Stanley Jevons, Léon Walras and Vilfredo Pareto. ‘Neoclassical economic theory,’ Mirowski claimed in the Cambridge Journal of Economics in 1984, ‘is bowdlerised 19th-century physics.’ It was profoundly shaped by its adoption – sometimes explicit, sometimes implicit – of metaphors and mathematical apparatus arising from the 19th-century reformulation of classical mechanics. The economists’ ‘utility’ took the place of the physicists’ ‘potential energy’. In the later period examined in Machine Dreams, the key external impetus is no longer classical physics. It is the digital computer and ‘cyborg sciences’: new fields (such as cybernetics, artificial intelligence and theoretical linguistics) that in the middle of the 20th century, often under military patronage, blurred the boundary between organisms and machines.
Machine Dreams is not, however, a simple story of emulation. Mirowski’s thesis is that potentially more radical outcomes of the encounter between economics and the computer influenced but were eventually sidelined by the discipline’s mainstream. Economics preserved neoclassicism’s homo economicus, albeit with some cyborg features, rather than taking forward what Mirowski reads as the vision of John von Neumann (mathematician, polymath, weaponeer, Cold War hawk, pioneer computer scientist, game theorist) of a fully cyborg economics, one which ‘appreciated the significance of the computer as an information processor for a reconceptualisation of the very nature of the economy’. The modern orthodoxy of economics was forged in its encounter with the cyborg sciences. It was and is an orthodoxy that prized mathematical rigour, which it took to previously unscaled heights; that rejected the historically and sociologically informed ‘institutionalist’ current in earlier American economics; and that expressed itself in a powerful disciplinary hierarchy which has no real analogue in looser disciplines such as sociology. There is for example a clear-cut pecking order of economics journals: at the top, nearly all are American. An economics department in a British university whose staff don’t publish in these journals cannot hope for a good rating in the Research Assessment Exercise.
As history, Machine Dreams is a remarkable achievement. It is hard to imagine a historian who was not an economist (as Mirowski is) being able to encompass the economics of the second half of the 20th century in its diversity and technicality. The paradigmatic formulation at the centre of much postwar economics – Arrow-Debreu equilibrium – is, for example, dauntingly abstract, and the mathematics that establishes the existence of such an equilibrium is advanced. (‘Equilibrium’ in economics is a complex and multi-faceted notion. Perhaps simplest is the notion of the ‘equilibrium price’ of a commodity: the price at which the quantity of the commodity that suppliers will sell equals the quantity that purchasers will buy. Equilibrium can be investigated for a single commodity, holding influences from elsewhere in the economy constant: this is called ‘partial equilibrium analysis’, and was the approach pursued by the British economist Alfred Marshall. In contrast, ‘general equilibrium analysis’ investigates the possibility of an equilibrium for the economy as a whole. Kenneth Arrow and Gerard Debreu’s work involved developing an abstract model of a competitive economy and proving mathematically that the economy – as modelled – must have a general equilibrium.)
The account in Machine Dreams of the Cowles Commission; of Arrow, its most celebrated member; and of Arrow-Debreu equilibrium is my favourite of the book’s many historical threads. The Commission was an economic think-tank set up in 1932 by Alfred Cowles III. It was based originally in Colorado Springs, where Cowles spent a decade recuperating from tuberculosis. From 1939 its home was in Chicago, until a rancorous dispute with Friedman and his colleagues in the University economics department led it to decamp to Yale.
More than any other single group, the Cowles Commission set the tone of postwar economics. Its politics, interestingly, were not free-market. It had a commitment to economic planning, in particular a belief that sufficiently sophisticated planning could mimic the workings of a competitive economy. If it could be demonstrated that such mimicking was possible, it would refute the arguments made famous by Ludwig von Mises and Friedrich Hayek that socialism was doomed to inefficiency by the computational infeasibility of replicating the information-signalling role of free-market prices.
In the postwar US, those who argued in favour of central planning could all too easily find themselves on a McCarthyite blacklist. The Cowles group’s other affiliations, however, were more in tune with the surrounding culture. It enjoyed the patronage of the US Air Force’s think-tank, the Rand Corporation. It had a commitment to the new mathematical techniques of operational research that emerged in the Second World War. There was a useful link to von Neumann, who fortuitously had to change trains in Chicago as he shuttled between the Princeton Institute for Advanced Study and the nuclear weapons laboratory at Los Alamos. Above all, there was a commitment at Cowles to the advanced mathematics that von Neumann and John Nash were making available to economists. From the emerging Cowles viewpoint, much of what passed as mathematical economics, including the work of their Chicago colleagues, was mere empirical statistics. The Commission aimed at the rigour and generality with which the French ‘Bourbaki’ school of mathematicians (to which Arrow’s collaborator Debreu was linked) was seeking to reconstruct pure mathematics. ‘The practical effect of the Cowles programme,’ Mirowski writes, ‘was to toughen up the mathematical training of economists and thus repel anyone trying to trespass from another social science – to eject history from the curriculum and to renounce loudly any psychological commitments.’ If you couldn’t understand the deep mathematical proof that showed that a competitive economy had to have an Arrow-Debreu equilibrium, or couldn’t reason formally about a Nash equilibrium (a situation in which no participant in a game can improve his or her pay-off, given the strategies pursued by the other participants), you weren’t a real player.
It wasn’t, however, just a question of an increase in mathematical sophistication. Mirowski’s key argument is that economics faced an implicit choice in the 1950s between different ways of mathematising. The road not taken was that suggested by von Neumann’s later work: the mathematics of computation, especially of self-reproducing automata (robots that can make other robots), with its requirements for close attention to the computational limitations of different categories of abstract machine. Instead, led by Cowles, economics took Bourbaki’s route towards abstraction and away from any thoughts of such limitations. Mirowski’s historical narrative is here interwoven with an evaluative argument. His preferred route is the path he believes was outlined by von Neumann. The Bourbaki highway led, he argues, to ‘the utter incoherence of a half-century of information economics’ – to a mathematised economics rendered sterile by insufficient attention to computability.
An economist today would protest that since around 1980 the discipline has come down from the heights of Cowles abstraction, but Mirowski is not impressed. ‘The secret to achieving fame among the economic orthodoxy,’ he writes, ‘is to begin with a textbook neoclassical agent in his standard problem setting . . . and to tweak the model with some limited aspect of an information-processing characterisation that seems to “resolve” some isolated prior problem with the orthodox tradition.’ In other words, Mirowski’s colleagues are indulging in ad hoc fiddling, when radical reconstruction is necessary. Von Neumann’s vision was sidelined but cannot be permanently suppressed: economics will be incoherent until it is taken seriously.
There are, of course, notorious dangers in the yoking together of historiography and evaluation. Their most common form is Whig history: telling the story of the past as one of advance to today’s superior wisdom. To those accustomed to the modern history of science’s careful avoidance of evaluation, Machine Dreams will sometimes read as Whig history with the assessments swapped: the past told as chronicle of the path to what Mirowski considers the gigantic error that is mainstream modern economics. His alienation from the mainstream may also account for an interesting implicit methodological decision. Although many of the protagonists described in Machine Dreams are still alive, Mirowski hasn’t embarked on any systematic programme of interviews. Almost all historians prefer contemporary documents to retrospective recollection, but it may also be that he chose to avoid the entanglement that interviewing generates. Granting an interview is a form of hospitality, and the ethics of hospitality would then make the harsh criticisms of Machine Dreams difficult: perhaps Miroskwi even doubted his subjects would tell him the truth. While the economists about whom he writes have been happy to model others as motivated by self-interest, ‘when summoned to reflect on their personal successes, they regularly cite such lofty goals as the alleviation of pain, the augmentation of general welfare, the abolition of injustice and the advancement of human understanding.’
Are Mirowski’s frequent evaluative comments simply a gloss on a historically defensible narrative, or have his preferences led his narrative astray? In the area I know well enough to assess – the work of Herbert Simon, one of the most successful dissidents from orthodoxy – his views (strong here, as on most topics) don’t damage his historical account in any substantial way. His opinions are also worth considering in their own right. The more thoroughly cyborg economics that he recommends involves a shift away from homo economicus to the direct modelling of markets themselves as diverse computational entities. He notes with delight a successful simulation by Dhananjay Gode and Shyam Sunder of a sophisticated form of market: the double-auction market, in which participants simultaneously quote prices at which they will buy and at which they will sell the product traded. The agents in the model were not profit-maximising homines economici but ‘zero intelligence traders’, whose price quotations were produced by a random number generator subject to only the simplest of budget constraints. Yet the simulated market achieved highly efficient outcomes, on average as good as those achieved by business studies graduate students. Gode and Sunder concluded that there was no need to assume a rational, profit-oriented, homo economicus: outcomes were being shaped by the structure of the market, rather than by the nature of its participants.
Let’s treat ‘markets as machines’, Mirowski argues, ‘while provisionally reserving for their human environment all the privileges of consciousness, intentionality, nonalgorithmic thought, and the entire gamut of diversity of cognitive capacities and idiosyncratic behaviour’. This isn’t just a slogan, but a technical programme for economics. Different forms of market, Mirowski suggests, can be simulated by machines of different types, some having the capacity to simulate others. Those abstract machines can be arrayed according to the cyborg sciences’ ‘Chomsky hierarchy’, a classification of automata that maps directly onto a hierarchy of the complexity of the formal languages they can process.
Mirowski’s vision of economics without homo economicus is so attractive partly because one disturbing aspect of performativity is that, by assuming egoism, economics may encourage it. A relevant experimental finding was published in 1981 by the sociologists Ruth Ames and Gerald Marwell. Their subjects played a ‘public goods game’ in which they had to choose between ‘investing’ tokens in an individual and a collective ‘exchange’. The optimum outcome for the group as a whole was for all tokens to be invested in the latter, but the pay-offs were set up so that, whatever the others did, any individual participant was better off by investing all his tokens in the individual exchange than putting them in the collective one. Many real-life situations involve such a trade-off between rational egoism and the collectively optimal outcome. Examples include co-operation in small groups such as university departments; problems of international politics such as control of greenhouse gas emissions; blood donation; even voting. Orthodox economic theory’s prediction of the outcome of Ames and Marwell’s public goods game was clear. Individual, not collective, rationality will prevail: nothing will be invested in the collective exchange. In most of their experimental groups, however, players invested around 40 per cent of their tokens in the collective exchange. Except, that is, when the game was played by graduate students in economics: they performed as rational egoists, with an outcome closer to the mutually impoverishing prediction of economic theory than that of any other group.
As Ames and Marwell admitted, the finding is ambiguous, possibly saying something about the kind of student attracted to economics, rather than the effects of learning. The economists I know, however, do not seem any less generous than other people. Perhaps a more general conclusion is more accurate: human beings are neither ingrained egoists nor stably the generous norm-followers that the ultimatum game at first seems to imply. One of the frustrations for those who look to experimental games to reveal the true nature of human beings is that apparently minor variations in experimental protocol can produce large changes in behaviour. One variant of the ultimatum game reported by the economist Samuel Bowles ranks subjects prior to playing the game on the basis of how they performed in a trivia quiz; the privilege of making the offer is then given to those who know they have scored highly. Even though the questions in the quiz are plainly trifling, offers become more unfair, and unfair offers are more likely to be accepted. It’s a finding that any educator should find terrifying as a metaphor of how educational success and failure can legitimise an unequal society.
If egoism and generosity are context-sensitive, we need, when making economic and social policy, to consider the effects on them. Interestingly, this is not in itself an argument against markets. Although markets can produce damaging outcomes of many kinds, they do not themselves seem to produce homines economici. Bowles and his colleagues have played the ultimatum game in a range of small-scale societies worldwide. In those societies more heavily involved in economic exchange, the ultimatum game tended to be played more fairly than in societies with little experience of markets. Economic anthropologists and sociologists, too, routinely observe gross differences between rational egoism and actual behaviour in real markets in both small-scale and industrial societies.
Homo economicus may thus be a product of the imagined market, so to speak, rather than of actual markets. However, the imagined market is now a strong force, for example in the public sector in Britain and many other countries, where it is favoured not just by conservatives but by many Third Way social democrats. It takes the form not only of clumsy efforts to introduce quasi-markets into public institutions, but of an implicit subordination of other kinds of human motivation to rational egoism. If the market cannot exercise its discipline, then it is assumed that the auditor must step in, for otherwise people cannot be trusted: hence the vast, expensive, expanding bureaucracy of surveillance that increasingly dominates the public sector.
Homo economicus, however, is only one way of being human – in general, a way that is to be discouraged. Unfortunately, his inverted twin, norm-governed homo sociologicus, is equally a construction. Co-operation, generosity, dedication, trust, a sense of service, orientation to work’s intrinsic satisfactions rather than to its extrinsic monetary rewards – such things need to be emphasised when designing institutions, not marginalised or taken for granted.
Societies like ours face delicate but necessary tasks in respect to markets. It’s unlikely that we will turn our back on them, but we need a better appreciation of their diversity, so we can weigh their advantages and disadvantages, rather than assuming them a priori to be either models of perfection or exploitative dystopias. Simultaneously, we need to ensure that imagined forms of human behaviour in markets do not achieve undue cultural influence, especially in the shaping of public institutions.
[*] LRB, 13 April 2000.