An Interview with Philip Mirowski

Philip Mirowski's work is difficult to classify as it stands at the crossroads between the history of science, economics and politics. Throughout his writings he consistently draws from a diversity of philosophical, economic and scientific traditions. This interview attempts to provide a brief overview of Mirowski's complex intellectual trajectory.

His interest in the complex relationship between economic thought and the natural sciences dates back to a 1984 article entitled Physics and the ‘marginalist revolution'. Subsequently, in the context of More Heat than Light (1989), he argues that the economic orthodoxy was born by imitating the framework of 19^th century physics. Neoclassical economists utilized the framework of the energetics of the time, leading them to model utility as potential energy, but without really understanding all of the consequences of such a move. As he argues in this interview, economists have never come to terms with the initial origin of their theories, something that haunts the profession to this day. Moreover, this was by no means an isolated incident. He identifies another major turning point in the 20^th century, when economics gradually transitions from the reliance on energetic metaphors to the adoption of the language of information and the framework of cybernetics and the theory of computation. In Machine Dreams (2002), he provides an extensive account of the impact of cybernetics and information theory on post World War II economics. That is to say, the advent of the computer in the 20^th century engendered a gradual transformation of economic theory. From the early days of Operations Research and Game Theory to contemporary experimental economics, cyborgs have been slowly, but surely, entering the mainstream of economic theory. In this new paradigm, the economic agent starts to resemble a computer and the market becomes something akin to a giant information processor. While the informational turn apparently reinforces the economic orthodoxy, it also has the potential to undermine most of this tradition. As Mirowski argues, a computational view of markets reduces the importance of the cognitive capacities of economic agents while emphasizing the institutional framework of markets. Thus, rather than remaining fixated on maximizing agents, economics can now embark on a novel research program, the understanding of the wide diversity of market forms (Mirowski, 2007).

Nevertheless, this informational turn also has broader socio-political implications, particularly in the context of the rise of neoliberalism. Mirowski, similar to Foucault, understands neoliberalism as a system of knowledge/power, or as he puts it: 'a set of epistemic commitments' (Mirowski, 2009). The idea of the computational superiority of the market is not confined to the world of academia and think tanks. As Mirowski argues in Never Let a Serious Crisis Go to Waste (2013), while neoliberalism started as a thought collective it is now a deeply ingrained socio-political and cultural movement. This perhaps accounts for its surprising resilience even in the face of major shocks such as the recent financial crisis.

TCS: In the beginning of More Heat than Light (1989) you go to great lengths to explain the concept of energy and you argue that ‘neoclassical economic theory boldly copied the reigning physical theories in the 1870s’. Could you talk about the relationship between the energetic metaphor and the transition from classical political economy to neoclassical economics?

PM: I think most people understand that there was some kind of epistemic break between the classical economists and the neoclassical economists. Although it is also interesting that the term ‘neoclassical’ is sometimes used to evoke much more of a continuity between Adam Smith or David Ricardo and neoclassical economists than is warranted. However, I think most people understand that there’s a really substantive break consisting of some obvious sharp turns, such as moving from a stress on the centrality of production to a primacy of exchange, the elevation of equilibrium over tendency laws, and the shift from a largely narrative based structure to an obviously mathematical structure. Moreover, there was a limiting of ambitions, from a classical political economy that talked about large historical movements through time to a neoclassical economics praising statics as if that was some sort of advance in precision.

I thought no one had bothered to try to explain all these things in a remotely satisfactory fashion and it seemed pretty clear, from my having done a fair amount of history of natural science when I was a graduate student, that a lot of this had to do with imitating the natural sciences. The thing that surprised even me was that the more I looked into it, the more it became clear that the main players actually openly admitted mimicking physics. The main players were people like Jevons, Walras, Edgeworth (but not the Austrians, which is significant as they get thrown in with the marginalist revolution but actually aren’t a coherent part of that world). A crucial point in clarifying the history is that they were not very imaginative about how to create this new economics they so fervently wanted, so they ended up pretty much copying energy physics of the later 19^th century. The other thing that I would point out is that they didn’t pull that option out of thin air. Energy physics was a cultural phenomenon in the late 19^th century, something that crossed language communities and spread across nations as the ‘Energetics Movement’. In its most basic form, this was a belief that these new laws of energy recently enunciated (conservation, optimization, and the like) promised to give us great insight into humanity generally, and how the mind works and so forth. So here are these people aware of a time in which this grand cultural movement is happening. In this context, it’s not weird or implausible for them to access that intellectual movement and maybe imitate some of its mathematics. I think this is clearly the explanation of the origins of neoclassical economics, which also explains a lot of its curious peccadillos as well.

That thesis seemed to lend insight to some age-old questions: Why do people think economics is successful as a science? Why does economics believe it enjoys a privileged status amongst the social sciences? The answer could be that it has a lot to do with the imitation of science, precisely because for a relatively general populace, appearances count for a lot. The similarities jump out at you, if you have a basic acquaintance with physics. At least in the 19^th century, the early neoclassicals made what are recognizably similar moves. They claim they possess laws, they use similar mathematics (the calculus), they claim they have access to some kind of scientific empiricism. So a lot of this is just status by emulation more than anything else. Now that doesn't mean that it’s the only thing that was going on back then, and in fact that complaint coloured a lot of the pushback that I got against the book.

TCS: Throughout the book, you present us with a very compelling argument for the crucial influence of physics in the development of economic theory. Do mainstream economists recognize this narrative and, if not, how can we explain their lack of awareness?

PM: Let’s take a recent example of a popular contemporary economist blogger, insisting that economists don’t suffer from physics envy. ¹ Instead of looking at the history, or the technical issues involved, he just blurts out some random impressions: economists make more money than physicists (I cannot make this stuff up); an economist can talk about how much progress physicists are likely to make, and get taken seriously, but not vice versa (ditto); economic theorists, traditionally, have been free from the constraints of empirical validation (note the total innocence of the history of empiricism in economics); and that equilibrium means something different in economics than in physics. The level of arrogance combined with parochial ignorance is pretty stunning, but not unusual. He has no conception of the historical track record of the disciplines of economics and physics evolving through time, with earlier points of interaction being masked by later developments, and further waves of strange action at a distance. The result would therefore never appear to contemporaries as strict identity.

Economists with a tad more sophistication may admit something happened in the 1870s, but they go on to insist that a century and a half of further development has produced a set of doctrines that doesn’t necessarily have anything to do with that heritage. So in a weird sort of way, the more sophisticated modern line is that intellectual origins don’t matter to contemporary doctrines. Your average modern economist thinks history has entirely been banished by the activities of subsequent generations. Actually, I would argue against that; it is more plausible to think that it’s a path-dependent process. Neoclassical economists can’t entirely wish away their origins, and there are a number of times where I try to point that out in the book: for example, the eccentric ways in which the early neoclassicals are totally confused about how to deal with production. Moreover, I think the physics origins really instantiates the central metaphor of a market as if it were a kind of machine that takes stuff from a place it’s not supposed to be and puts it in a place that it deserves to be. That leads to this whole idea of allocation as a special phenomenon, which captures the essence of economics. That’s the way the first three generations of neoclassical economists think, in terms of movement in a commodity space. So for them, trade (exchange) is motion in a commodity space. All of these points comprise a deep inheritance of the early appropriation from physics that is really hard to get away from.

TCS: Finally, does this mean that neoclassical economics should pay more attention to its own history and its epistemic presuppositions?

PM: My intention has been not to focus on the motives or conceptual integrity of any one thinker, which is the way a lot of history is written. Rather, I wanted to explain what they have in common and why they could recognize each other as part of the same project back when there was yet no stable orthodoxy congealed around the optimization of utility. But more importantly, I highlight that the appropriators said they were doing it too. I find it just amazing when people like Robert Solow (in Bender and Schorske, 1997: 74) actually say that they don’t believe that’s really true, that they imitated physics. This issue can easily be settled by simply looking at the evidence, which others and I have provided (De Marchi, 1993); so there must be something else going on here.

I have eventually come round to the position that orthodox economists cannot take a long hard look at their own history, because they have proven incapable of situating themselves in their own sequential images of how markets and societies work. In other words, the denial of intellectual history is not simply Orwellian, it is a structural consequence of the theories they hold dear. Reflexivity is their wolfsbane. For instance, who really needs economists in Walrasian general equilibrium, or in the modern economics of information? This tension is slowly building to a head in the modern profession, and I have begun to write about it in my most recent work.

TCS: In a more recent book, Machine Dreams: Economics Becomes a Cyborg Science (2002), you have focused on the impact of a different metaphor, that of the computer. Here you look at the influence of cybernetics and computer science on post-Second World War economic theory. Could you talk about the process that took you from More Heat to Machine Dreams, from energy to the computer?

PM: I knew even when I was still writing the first book that something really odd happened in the Second World War that had changed the orthodoxy once again. Now, the first temptation was to notice that a lot of the most influential people were actually trained as natural scientists, such as Kenneth Arrow, Gerard Debreu, Tjalling Koopmans or Leonid Hurwicz. When you look at their backgrounds, you can interpret it as another wave of natural scientists coming over into economics. But actually the more I looked into it, the more interesting the story became. Yes, it happened that a lot of the reasons these people could come into economics with little prior preparation was because the existing simplistic mathematical models were really easy for them, mostly another heritage of the initial physics appropriation. But over time what happens is that they got caught up, in particular in the United States, with what a lot of the natural scientists were doing during the war.

Part of the story I stumbled upon is that Operations Research becomes one more mediation through which economics becomes inflected with natural science themes after the Second World War. The economists themselves in that era aren’t terribly imaginative, and they aren’t responsible for inventing most of it, instead appropriating much of it from John von Neumann, David Blackwell, John Nash and a whole host of others. Another curious aspect of the influence of Operations Research is that it was bound up with the early development of both electronic computers but also programming at RAND. Linear programming, statistical optimization, matrix methods, formal logic, information theory, and a whole raft of techniques were imported into economics at that stage. Familiarity with early digital computers gave economists a head start relative to many other disciplines.

Game theory stood as another good example, in that the military and the economists were all initially enthusiastic about Von Neumann’s version of game theory, thinking it would constitute a science of conflict. Further, the famous Arrow-Debreu existence proofs were a direct consequence of early exposure to Nash game theory. Nash equilibrium then comes to represent an anti-Von Neumann approach to solving games. But it’s apparent that the people who are involved in early Operations Research back away from game theory very quickly, as does the military, interestingly enough. Consequently, this whole fascination with game theory in economics really takes another generation to incubate; it doesn’t really take root in microeconomics until the 1960s and 1970s. Long story short, because the computer is increasingly important in many of the areas in which economists are interested, they become inadvertently recruited into this whole trend of the intellectual development of the computer. This ranges from cybernetics as a theory of certain kinds of human/automaton metaphors, to the incorporation of stochastic models in decision theory, to the actual use of computers in econometrics and simulation, which happens earlier than in some of the other social sciences. In many ways, the use of the computer obviously jump starts econometric empiricism, which is also one of their many post-war inventions.

What Machine Dreams does is to try to tell the story of how these 20^th-century sciences slowly transform what the orthodoxy means in economics. I argue that their increasing commitment to a vision of the computer after the Second World War slowly changes their image of what the market is, and that’s actually situated at the heart of the matter. What I mean by this is that, once you change your metaphor of what the market consists of, then everything else changes. There is a sense in which neoclassical economics began to lose its quiddity as ‘the study of the allocation of scarce resources to given ends’. The interpretation of trade as motion in a commodity space is going to change, the interpretation of what it means to be a human being is also going to change, and so on. Thus in Machine Dreams I stumbled on this suppressed epistemic rupture, namely, that in their quest to imitate what they thought was the most up-to-date science, they end up inadvertently swapping metaphors. This in turn bequeaths us, at least in part, the shape of the orthodoxy that we have today. But economics is not alone in this respect, and I don’t think anyone has done a good job describing this trend, even in the history of science. I think now it’s uncontroversial to say it doesn’t matter what natural science you look at, in its modern instantiation, it pretty much has a computational flavor to it. So computation has changed the ways in which some of the very basic concepts are framed in all the sciences.

However, this also explains some of the reactions to More Heat than Light. There were people commenting that they didn’t recognize this ancient ‘physics’ doctrine anymore; that modern neoclassicism doesn’t look like the models of Jevons or Walras. As I show in Machine Dreams, there has been an element of truth to that, as the post-war orthodoxy actually looks somewhat different from the 19^th-century version. No one believes in Energetics any longer; but a lot of people seem to conflate the mind with a computer. I think one role of the historian is to identify precisely these shifts, these ruptures, namely that big cultural ideas often change in important ways that aren’t always conscious, but ultimately transform an entire discipline.

TCS: In your recent lectures at INET, in collaboration with Edward Nik-Khah, you develop a history of the relationships between information and economic theory. This is a very original attempt to write an alternative narrative of the history of economic theory in the 20^th century. Why does information become such a crucial notion in modern economics after the Second World War?

PM: This is the book coming out from Oxford University Press called The Knowledge We Have Lost in Information (2016). It follows on from the series of lectures we did at INET, where we focus on what we think are some of the main determinants of the shape of orthodox economics in the later 20^th century. In a nutshell, we take off from the notion of information, which was already present in Machine Dreams, except now we’re actually trying to follow the various transmutations of the economics of information from their origins in the 1930s and 1940s all the way up to the present. In comparison, I would say that Machine Dreams kind of stops in the 1990s, more or less.

The central insight is that markets in the current orthodoxy no longer resemble those implicit in the early phase of neoclassical economics. Markets are re-imagined in dramatic ways, and the most dramatic is the rise to pre-eminence of the idea of a market as an information processor. In other words, the supposed virtue of markets has shifted from the mechanical task of bringing about superior ‘allocations’ to the epistemic task of serving as the primary mechanism for the validation of truth. Certainly there was some recognition of that in passing in Machine Dreams, but I have now come to think it has proceeded to dominate the evolution of the whole of microeconomics. I keep insisting that almost all natural sciences are now sciences of information in one form or another (biology, physics, etc.). Accepting the metaphor of information processing totally transforms the nature of what counts as orthodoxy there, and in economics. We would even argue now that one of the reasons game theory becomes acceptable in the 1970s in a way it was not in the 1940s is because it gets incorporated in this central thematic of information processing. Likewise, the fascination with so-called ‘behavioral economics’ is just an acknowledgement that it has recently become one of the ten commandments of orthodoxy that people are just more stupid than the market. We also argue that the most important development in this transformation is working out the issues of information in the rise of market design, experimental economics and the market microstructure literature.

TCS: In your lectures ² you also identify a cultural and political tendency underpinning the marriage of information and economics. Economists who made extensive use of the notion of information in their work, such as Friedrich Hayek, Ronald Coase, Vernon Smith, George Stigler and Gary Becker, were all members of the Mont Pelèrin Society that acted as the catalyst for neoliberal thinking. Could you elaborate on the importance of neoliberalism for the informational turn in economics?

PM: What is interesting about the informational turn in economics is that it’s partly due to wider trends in the natural sciences, but it’s also partly attributable to a political phenomenon. What I would like to highlight is that you can’t understand the spread of the idea of a market as an information processor without understanding the concomitant rise of neoliberalism. That movement originates within the political/cultural group centred around Hayek and the Mont Pelèrin Society (Mirowski and Plehwe, 2009). In a sense, the market as an information processor was the neoliberals’ central argument against socialism. That is, if socialism is about the ambition to plan and shape the economy, neoliberals retort that markets know more than any single human being, and therefore anyone attempting to plan the economy is deeply misled, and so consequently, socialism can’t work. What’s interesting about this doctrine is that it has basically become orthodox wisdom in the modern economics profession, even though this idea was situated at the fringe in the 1940s and 1950s. What we do in the forthcoming book The Knowledge We Have Lost in Information and in the INET lectures is to start off from Hayek and the Socialist Calculation Controversy. We trace how the image of the market changes from Ludwig von Mises’ notion that without markets you can’t calculate to Hayek’s notion that basically without markets you can’t think, which is a much more dramatic statement than anything found in the earlier classical liberal doctrine.

The next stage of the narrative notices that many people in the United States who thought of themselves as ‘market socialists’ felt that they were impelled to answer and refute this whole nascent neoliberal movement. It just so happens that those people tended also to be the very same people who brought the new sciences into economics after the Second World War: members of the Cowles Commission, by and large. In fact, everyone in the Cowles Commission during its Chicago period regarded themselves as refuting Hayek. It turns out these were many of the same economists often credited with the genesis of the orthodox economics of information – hence the neoliberal sting in the tail.

I don’t think many people are currently aware of that; but we’ve got evidence showing that, for example, Leonid Hurwicz actually studied with Hayek for a brief time, and explicitly said that he was trying to refute him. Jacob Marshak was another key player, and he was one of the early participants in the Socialist Calculation Controversy. Then we can just go down the list – Kenneth Arrow, Herbert Simon, Stan Reiter, and later Joseph Stiglitz and George Akerlof – and it’s truly amazing the extent to which they formulate knowledge as an economic entity. So this is a hidden history of 20^th-century orthodox microeconomics, namely, that it’s a bitter fight over this metaphor of market information. What does it really mean to think of markets as information processors? Basically, what happens after the 1980s is that the pro- and anti-socialists begin to converge.

TCS: Perhaps one of the most interesting aspects of your recent work is that you highlight the shift from a concern with the nature of the individual agent to the institutional framework of markets. In other words, what happens to the utility maximizing agent in a world where we focus more and more on market design and market structure?

PM: The crucial point to recognize is that if markets have really come to resemble these superior information processors, then the putative abilities of agents become less and less important for the analysis. This is true for Hayek, and for modern neoclassical economists as well. Agents can therefore become lumbered with all sorts of cognitive deficiencies, which constitutes the rise of the so-called ‘behavioural economics’ that everyone praises to the skies as a breakthrough in the history of neoclassical economics. We instead approach it as an epiphenomenon of the increasing contempt for the agent, and a growing anti-humanist stance, which is the flip side of market as information processor metaphor. It also reflects the general contempt for the masses (and democracy) inherent in neoliberal political economy. A further trend we notice is that the ontology of information slowly changes from being thought of as a thing to a kind of statistical inference, and now as some sort of a computation (which has serious implications for how a market is going to be modelled in the future).

Nevertheless, what is of paramount importance is that the information revolution changes the self-image of what economists are, and what they claim to be able to do. In a sense, if economists are specialists in information processing, they can claim to possess the capacity to be able to build boutique specialized markets from scratch. This turns out to be something really new; and it is called ‘market design’. There is no instance of any similar pretensions prior to the 1980s, and it has all kinds of further implications for the evolution of the discipline. For one thing, it’s great for economists to have a new way to sell themselves so that they become part of the new commercialized science in the neoliberal university, because what they’re selling now is their engineering expertise in building market structures. This underpins their general contempt for the humanistic disciplines, with their hapless lack of things to sell to the private sector. Then there’s another consequence, which is even scarier. Namely, there abides an intellectual question as to why economists should be able to build markets for clients at all, when in fact most markets, at least in neoclassical theory, operate entirely according to the same monolithic principles. The danger here is that by selling themselves as engineers, economists open up the conceptual possibility that markets don’t all operate the same ‘in the wild’ – something one observes in experimental economics and in the so-called market microstructure literature – and that markets are therefore better understood as an ecology. But I don’t think orthodox economists see themselves as entertaining this seriously, because many market designers don’t comprehend that if you undermine the very notion of the monolithic perfect market, you undermine almost every aspect of neoliberal political doctrine, and you certainly undermine formal neoclassical economics. So here we are, living in this weird situation, poised again on the cusp of a massive intellectual contretemps.

Alvin Roth has this interesting paper (2002) where he says that he didn’t learn very much about building markets from theoretical game theory, but took his cues from the anomalies. That’s a really important statement, and I don’t think people understand its consequences very clearly. If different market structures produce different effects, and if economists have less and less regard for the cognitive abilities of the agent, then this means that all the vaunted claims to ‘welfare superiority’ of markets are off. You can see how politically dangerous this might be, and I am not aware of anyone in the profession having thought this through with any sophistication. If any claims to welfare have become groundless, because it’s a dog eat dog world with the air filled with lies and deceit, then this poses serious challenge to the orthodoxy. In fact, markets can be promoted as ways to manage and limit the cognitive flaws [of the agent]; some market designers claim the ability to force agents to ‘tell the truth’, whatever that means in modern theory. This raises the issue whether ‘markets’ are epistemically omnipotent or just imperfect prostheses.

Moreover, I think these developments are also dangerous for neoliberalism; because how can the ‘market’ as a whole be a superior information processor if it’s made up of all these individual subcomponent wonky information processors, constantly misinterpreting events? So the whole system actually is evolving in a direction that could potentially undermine the bulk of its intellectual legitimacy; and Edward and I thought that was a fun story to tell. Market design might even erode the supposed centrality of prices to the economic system. In simple neoclassical theory, everything was collapsed into price, whereas now, in the brave new world of market design, economists should be concerned with all the possible component functions of the market outside of price discovery, especially ones notoriously ignored in the past. For instance, consider questions such as recording of trade data and dissemination of information. What is the size and the boundary of the market? Who knows what when? How is information conveyed? How do different market formats interact? How are property rights assigned when they characterize alternative formats of a particular market? I mean it’s amazing how much intellectual space is opened up. Again it ventures far from the previous neoclassical orthodoxy, and people just do not realize the extent to which this is happening. But this is me as a historian speaking.

TCS: In Machine Dreams and in other recent papers (2007, 2010) you argue for an algorithmic understanding of markets. You call this a theory of markomata, or a computational evolutionary economics. This account seems to gravitate around Vernon Smith’s experimental economics and Gode and Sunder’s research program on zero intelligence agents. Could you give us a brief introduction on these topics and their potential for a novel understanding of markets?

PM: Vernon Smith is a very important figure in modern history as he is one of the pioneers of experimental economics; and he was also a member of the Mount Pelèrin Society, and it shows up in his work. When people think about experimental economics they think it’s about finding out how people really behave, but for Vernon Smith it is different. He actually doesn’t believe in any of those refutations of so-called ‘rationality’ or of other neoclassical presumptions by experimentalists, and he explicitly says so. This is because he subscribes to the neoliberal precept that the market is smarter than everybody, however dumb they may be. This is why it is OK to use undergraduates for these experiments, because it doesn’t matter what they know or don’t know. It’s the market that produces the results; to this end, he coins what he calls the ‘Hayek hypothesis’, that basically, certain kinds of markets will produce a neoclassical equilibrium result no matter what. He allows some qualifications to this, which are not important for current purposes.

Vernon Smith exemplifies the attempt to marry modern market design with neoliberal doctrine, using computers as the primary tool. I’m not saying I endorse his program, but one needs to have a working understanding of it to grasp the significance of a secondary literature following up on it by Dan Gode and Shyam Sunder. What they seem to be saying is that if we believe Smith, we should want to know what it is precisely that causes this convergence to standard neoclassical equilibrium in the face of mentally challenged agents. So they have this brilliant idea that they would set up one version of the Smithian market experiment with students, and another set-up populated by zero-intelligence agents (which are basically random number generators). The first thing they found out is that both experimental set-ups give pretty much the same result, so essentially, what people think and do doesn’t matter to the outcome. This captures the essential anti-humanist stance of modern economics. They then looked into this symmetry in greater detail to try to figure out what about the computer algorithm caused this. Their answer basically comes down to a set of rules about how certain bidders and askers are pushed out of market participation as the process continues. The other causal factor is the imposition of budget constraints which operate as boundary conditions, such that simply imposing a budget constraint on these otherwise zero-intelligence traders is almost enough to produce these supposed ‘Hayakian’ results. So in a way, they took the neoliberal story that Vernon Smith proposed and they recast it into a story that turns neoliberalism on its head. Namely, Gode and Sunder demonstrate that markets produce their regularities because of their particular rules and algorithmic structures; markets are not themselves magic, but are limited information devices; and this has not proven to be a popular position in the economics profession. It is threatening because orthodox economists still want to hold on to this idea that markets in some sense are astounding because they give people what they want. Gode and Sunder say that it doesn’t much matter what people want or don’t want, that it doesn’t much matter what those people think or don’t think. Pretty much everything that we take to be natural regularities in markets is an artifact of their actual algorithmic design.

TCS: Another important consequence of adopting an algorithmic understanding of market structures, and you mention this in several instances, is that they are more computationally expensive as they become more complicated. In what sense can it be said that market algorithms evolve?

PM: Once we concede markets are actually algorithmic structures, and that they differ tremendously in structure and output, then it is natural to ask why we have the distribution and ecology of markets that we currently inhabit. The first place to look is to examine the distribution of markets in terms of their computational capacities and their computational complexities. Some neglected economic history suggests the vast bulk of existing markets are of rather low computational capacity and low complexities (think of posted price formats in supermarkets, or vending machines). Some markets support greater complexity, but they are generally of more recent provenance, and far less prevalent (modern art auctions or stock markets).

I think what that suggests is that markets of increasing complexity need highly specialized support groups to make them work at all, because they are so complicated. They are more expensive to operate, but also computationally awkward, whereas, as you would expect in evolution, simple markets spread more widely because they are relatively uncomplicated, and are robust to a wider array of participants. Nevertheless, orthodox economists get this backward, and get far more excited about the most complex market forms. Moreover, I think people tend to conceptualize complexity in terms of technology, but that’s not the intellectual crux of the problem. The key to classification turns out to be the algorithmic structure of the market, which is an analytically valid way to gauge the increased complexity; and yet, I don’t see people entertaining that. This, by the way, is not at all the same as some version of technological determinism. People often write as though that's the only important dimension along which markets change. What little neoclassical economic history – cliometrics, mostly – has been conducted on the long view of markets tends to focus on issues such as the effect of the telegraph on markets, or else predicating evolution on ‘trust’ in small social groups. But this reflects a pure technological determinism that takes your eye off the actual evolution of algorithmic structures of markets.

TCS: You state repeatedly that your computational evolutionary theory of markets still needs a ‘natural history of markets’. Why is this an essential part of the puzzle for a revival of economic theory?

PM: If we accept Gode and Sunder’s results, this means that markets don’t all do the same thing, and therefore must occupy different niches. Vernon Smith doesn’t push that consequence of his own experiments; he doesn’t manage to go there even though it’s an obvious implication. Economists seem almost as bored with economic history as with the history of their own doctrines. This lack of curiosity can also be traced to the physics origins of neoclassical economics. They neglect the observation that ‘the market’ has not been a monolithic phenomenon perduring through time, revolving like some grand celestial mechanism, but instead a conglomerate of changing structures. Probably this happens because different markets are better suited for different situations. So the revival of a serious economics should imagine at any juncture there being an ecology of markets, with maybe some markets working better with smarter people, maybe others working better with people who devote little attention to them. Some scholars make the argument that posted-price markets are more robust to the kind of agents they encounter, which is possibly why they are the most common type of market we deal with in everyday life.

This opens up the possibility of an entirely different kind of economics, where you’re not any longer even searching for the ‘laws of the market’. In this approach, there are no laws of the market in the same sense that there’s no single ‘law’ of biological entities. The parallels are revealing. We humans are convinced we are so special, the pinnacle of evolution, but actually the terrestrial sphere is mostly populated by beetles and bacteria in terms of biomass. In the parallel case of markets we are talking of forms consisting of very simple rules, which seem to be historically far more common than the elaborate algorithms at the pinnacles of finance. This suggests a very different configuration of economics research competencies than currently exists. It would be comprised of a natural history of market forms, combined with population statistics of their evolution; and a mathematical theoretical component deriving from computer science, and not the residual fascination with calculus and deterministic dynamics left over from the origins of the neoclassical school.

TCS: More recently you have also addressed the issue of the recent financial crisis and neoliberalism. What insight can we gain by looking at the rules and algorithmic structures of financial markets? Does an algorithm and evolutionary perspective allow us to shed light onto the recent failures of certain derivatives markets?

PM: I have been working lately on some explanations for the crisis that one doesn’t encounter very often in the popular press. In fact, you are more likely to find it in the legal literature rather than economics journals. It has to do with the role of collateral in the rise of shadow banking, and the way it is integrated into the repo market, which has a very special market structure. Repos are derivatives instruments, which are essentially fake sales coupled to repurchases. They are really loans, but are effectively treated legally as sales, and for a long time have been situated outside of the normal purview of the standard banking regulatory structure. In fact, you can argue that the repo market and how collateral is handled has made all the difference to the precipitation of the 2008 crisis (Gabor, 2013).

In the case of repos, you get the impression that collateral is built in, because you supposedly sell an asset (high-quality, frequently sovereign bonds) and then buy it back (repurchase), usually overnight. But actually that is not true. As Annelise Riles (2011) points out, there is no such thing as complete collateralization, and I think everyone needs to really take that seriously. It costs money to keep track of collateral and it costs money to claim it back, etc. These are not simple ‘transactions costs’, but built-in consequences of the rules of transaction and clearing and bankruptcy. But as soon as banks started to get involved in the repo market, it became clear that it wasn’t just a costs centre, but that you could turn it into a profit centre. Simply put, collateral can be reused over and over (rehypothecated is the term of art), separate from the original repo, and then it becomes very hard to claim it back. Effectively, at any point in time no one knows where the collateral actually is any more. I mean there are organizations and structures in place that try to keep track of it, but the truth of the matter is that nobody really knows where the collateral rests. Part of the problem in the crisis was that there were collateral calls on the Lehman Brothers repo just prior to the collapse, and they couldn’t find it. Something that seems on the surface just a technicality, not important or just a minor side issue to the act of lending, turns out to be one of its central aspects, causing vulnerabilities in the entire system. This illustrates how the understanding of market formats and algorithms can illuminate something as large as macroeconomic collapse.

It's also possibly central to the recent European sovereign debt crisis; and here’s why. Everyone wanted good quality collateral on the Continent, but there were not enough German bonds to sustain the amount of repo that was going on in European shadow banking. The ECB took it upon itself to do a number of things to unify Europe’s financial markets, so consequently it allowed for the treatment of the sovereign debt of all Eurozone member countries as effectively the same, for collateral purposes. Collateral became doubly evanescent. This is one of the reasons why the whole international shadow banking sector might have collapsed in 2011. That’s why the ECB was so willing to prop up Greek debt at that early juncture, that is to say, to sustain the whole system. The reason that they’re not so willing to do so in 2015 is that the ECB has pulled back from the earlier rules; almost nobody uses Greek debt anymore as collateral for repo, and so a default now wouldn’t be quite as destructive to the system as it would have been in 2011. I’m surprised that almost nobody tells this story (Daniela Gabor is the honourable exception) – that it isn’t merely that some banks ‘got into trouble’ though over-lending, but rather that the whole financial system depends upon minor aspects of rules and practices within particular markets.

Shadow banking is predicated on the neoliberal notion that borrowers and lenders can be trusted to police the rules of their own market constructs. Right up to the crash, the whole system was predicated upon a lie, which is that if any part of the system runs into trouble, there exists a subset of actors who can pull out by being first in the bankruptcy queue to claim their collateral, and thus not be harmed. The crude way of saying this is that crucial deformation of financial markets for the last two decades has been the consequence of legal expansion of the safe harbor principle. Certain classes of creditors, such as repo or other derivatives holders, are permitted to change the rules so that they don’t have to go to bankruptcy courts to supposedly claim back whatever can be claimed from certain failures of their counterparties. The effect of that has been to render the entire system more fragile. This whole idea that you can protect yourself and everyone else be damned seems to work when everything is working fine – the collateral works fine, the counterparties follow through, etc. It is only when the real crisis hits that you begin to see the fundamental falsity of the theory that lies behind these structures, and it’s the foundational neoliberal idea that the market knows more than any of its participants. So, for example, if a particular creditor and debtor get together and make a side arrangement to hide the extent of the debtor’s leverage, that’s fine. Markets will reveal all that needs to be revealed. Something like that is reputedly efficient because the market is the greatest information processor known to mankind. In many ways, this is the world that the economic orthodoxy has helped create.

TCS: It is interesting that you refer to this world that economic orthodoxy has helped to create, as it goes beyond the structure of certain derivatives markets and tells us something about society more broadly. This is also evident in your description of neoliberalism in that it extends towards much more than just an economic theory. In your book The Road to Mont Pelèrin (2009) you talk about the neoliberal thought collective and you describe this as a wider cultural phenomenon that transcends politics, philosophy or economics.

PM: I just want to be clear that I don’t think that many people fritter away their free time reading Hayek, Becker, Buchanan or Friedman. But the historian Jennifer Burns has this wonderful insight when she describes Ayn Rand as ‘the ultimate gateway drug to life on the right’ (Burns, 2009). I think that’s exactly right, namely that a certain picture of the world is learned at a very early age, usually by teenagers reading fiction or watching movies or going online. Some of them might go on and learn a little bit more about politics and economics, but in a weird sort of way that is unnecessary. Because the current culture gives them just enough to behave in ways that the neoliberals describe as being the ideal entrepreneur of the self, confusing freedom with imaginary lack of constraint, and so on and so forth. No one has to read Foucault. Just watch The Apprentice, or spend a little time on Facebook. So it’s weird that there is this sort of cultural gateway drug that allows you to kind of buy into this stuff without actually understanding it very well. I would argue that even people who may say they are sympathetic to neoliberalism don’t understand it very well. It’s not necessary, though, because you there’s just enough of this stuff around that they can pick it up as a worldview that denies the existence of ideology.