By Philip Mirowski, Carl Koch Professor of Economics and the History and Philosophy of Science University of Notre Dame. Professor Mirowski has written numerous books including More Heat than Light, Machine Dreams and, most recently Science-Mart
Edited and with an introduction by Philip Pilkington, a journalist and writer living in Dublin, Ireland
The debates surrounding the Dynamic Stochastic General Equilibrium [DSGE] models are perhaps some of the most interesting and important to have surfaced in the wake of the crisis. Of course, they, too many debates within the economics profession after the crisis, are deployed in order to insulate the research program from any fundamental criticism. But it is in the nature of the material that the critical observer can see something more interesting going on. And that is the contradiction at the heart of economics: the dichotomy, the abyss that opens up by necessity between macroeconomics and microeconomics.
Mirowski puts it as such:
A methodologist would point out, as I have done, that the canonical DSGE assumes its canonical outlandish format in order to ‘save’ it’s microfoundations, viz., the non-negotiable prescription that macro and neoclassical microeconomics are one big unified theory. All these current fragmentary amendments to render the DSGE model more ‘realistic’, or perhaps more politically acceptable to the ‘New Keynesians’, are self-contradictory, since they attempt to mitigate or ‘undo’ the microfoundations which had been imposed by decree from the outset. It ends up being one more instance of economists asserting both A and not-A simultaneously.
To truly understand the significance of this debate is to see that, were neoclassical economics ever to be amended in a manner that allowed it to be both logically consistent and empirically realistic, the whole thing would simply fall apart. As the great Cambridge economist Joan Robinson put it in her essay ‘Spring Cleaning’: the proper approach might be to scrap the lot and start again. Perhaps that is too much to ask, but surely it is reasonable to say, as we enter into a second Great Depression, that any analysis based on equilibrium is inherently flawed and should be done away with as quickly as possible.
– Philip Pilkington
Part IV: DSGE and the Threatened Unravelling of the Whole Damn Thing
A third reaction to the crisis is to refrain from indictment of the global orthodoxy, and instead suggest that since the crisis was eminently a ‘macroeconomic’ event, the onus for failure must be narrowly restricted to that subset of the profession tasked with study of the macroeconomy; and furthermore, the correct response is simply to jettison the paradigmatic model found in contemporary macroeconomic textbooks, the so-called ‘dynamic stochastic general equilibrium’ [DGSE] model. The crisis, for this cadre, does not portend the ‘death of economics’, but just a garden variety ‘model failure’: so replace the model. We might think of this as the ‘ounce of prevention’ response. Now, I can imagine my audience rolling their eyes – even those willing to put up with a modicum of technical issues raised so far are not going to countenance a tedious discussion of a specific mathematical model, no matter how crucial to the self-image of the economics profession. And it is true that there is almost no commentary in the general press on the DSGE model, compared with breathless denunciations of ‘rational economic man’ and the EMH. But this option does even more directly call into question the commonplace notion that economists can learn from their mistakes.
This is exemplified by an event in 2010 that was literally unprecedented in the history of economic thought in America. Congressional testimony is regularly convened on all manner of issues of applied economics; and economists are regularly enjoined to testify. But never before, to my knowledge, has an entire session been convened to hold public hearings on criticism of a mathematical model produced by economic theory, not on its purported applications. Yet, on 20 July 2010 a kind of Star Chamber was convened to pillory the DSGE model.  The basic stance of the hearings was defined in the opening comments by Chairman Brad Miller:
According to the model’s most devoted acolytes, the model’s insights rival the perfect knowledge Paul described in the First Letter to the Corinthians; but unlike the knowledge Paul described, DSGE’s insights are available in the here and now. To be fair, DGSE and similar macroeconomic models were first conceived as theorists’ tools. But why, then, are they being relied on as the platform upon which so much practical policy advice is formulated? And what has caused them to become, and to stay, so firmly entrenched? And, finally, the most important question of all: What do we get when we apply the various tools at our disposal to the urgent economic problems we’re facing today?
This is how the committee staff described the DSGE model for a lay audience:
The dominant macro model has for some time been the Dynamic Stochastic General Equilibrium model, or DSGE, whose name points to some of its outstanding characteristics. ‘General’ indicates that the model includes all markets in the economy. ‘Equilibrium’ points to the assumptions that supply and demand balance out rapidly and unfailingly, and that competition reigns in markets that are undisturbed by shortages, surpluses, or involuntary unemployment. ‘Dynamic’ means that the model looks at the economy over time rather than at an isolated moment. ‘Stochastic’ corresponds to a specific type of manageable randomness built into the model that allows for unexpected events, such as oil shocks or technological changes, but assumes that the model’s agents can assign a correct mathematical probability to such events, thereby making them insurable. Events to which one cannot assign a probability, and that are thus truly uncertain, are ruled out.
The agents populating DSGE models, functioning as individuals or firms, are endowed with a kind of clairvoyance. Immortal, they see to the end of time and are aware of anything that might possibly ever occur, as well as the likelihood of its occurring; their decisions are always instantaneous yet never in error, and no decision depends on a previous decision or influences a subsequent decision. Also assumed in the core DSGE model is that all agents of the same type – that is, individuals or firms – have identical needs and identical tastes, which, as ‘optimizers’, they pursue with unbounded self-interest and full knowledge of what their wants are. By employing what is called the ‘representative agent’ and assigning it these standardized features, the DSGE model excludes from the model economy almost all consequential diversity and uncertainty – characteristics that in many ways make the actual economy what it is. The DSGE universe makes no distinction between system equilibrium, in which balancing agent-level disequilibrium forces maintains the macroeconomy in equilibrium, and full agent equilibrium, in which every individual in the economy is in equilibrium. In so doing, it assumes away phenomena that are commonplace in the economy: involuntary unemployment and the failure of prices or wages to adjust instantaneously to changes in the relation of supply and demand. These phenomena are seen as exceptional and call for special explanation.
While skepticism concerning the DSGE is worn openly in this précis, it was nowhere near as scathing as the disparagement of the model that one hears in private and reads on blogs. Incredulity often focuses upon the presumption of a single immortal representative agent capturing the entire economy. For instance, at the Institute for New Economic Thinking (INET), I witnessed one famous economist compare coordination failures in DSGE models to the right hand losing track of what the left hand was doing, and the treatment of uncertainty in DSGE as tantamount to diagnosing the onset of Alzheimer’s. You can just imagine the bizarre shapes that ‘information’ assumes in this solipsistic portrait: what can it mean for this god-like agent to learn anything? A good DSGE joke current on the blogs is: ‘Based on all available information, I rationally expect DSGE models to suck for an infinite number of future periods; and because I am a representative agent, everybody agrees with me.’
So maybe economist jokes are not all that funny, but there are a few philosophical points to be extracted from the imbroglio. The first is that, within the profession, seeking out the Golden Mean does not guarantee intellectual credibility. The DSGE model was the product of a long series of compromises resulting in what was conceived as best-practice consensus, following a period in which participants had endured what they felt was three decades of bickering, discord and wrangling over the correct way to theorize in macroeconomics. In the middle of the Noughties, embrace of the Great Moderation was coupled with declaration of the Great Macro Accord, and the DSGE model was its offspring. Complacency in the world of ideas replicated complacency in the world of policy. In another instance of bad timing, Olivier Blanchard, Chief Economist of the IMF, decreed: ‘The state of macro is good . . . macroeconomics is going through a period of great progress and excitement, and there has been, over the past two decades, convergence in both vision and methodology’ (2008, pp. 2, 26). ‘DSGE models have become ubiquitous. Dozens of researchers are involved in their construction. Nearly every central bank has one, or wants to have one’ (2008, p. 24). But perhaps the fact that the DSGE model was an attempt to be all things to all sides, a détente imposed from above, emitted from a very few ‘top-ranked’ economics departments rather than a voluntary truce taking hold organically. This had something to do with its clueless set-up for its vertiginous fall.
While there are some good historical summaries of how the rational expectations movement and the so-called ‘Lucas critique’ killed off the previous Keynesianism of the 1960s/1970s, there are very few sociological meditations on how economics got from there to the DSGE model. Starting out under the banner of ‘consistency’, it was insisted that neoclassical microeconomics and macroeconomics be fully interchangeable. Second, orthodox macroeconomists came to conflate ‘being rational’ with thinking like an orthodox economist. What this implied was that agents knew the one and only ‘true model’ of the economy (which conveniently was stipulated as identical with neoclassical microeconomics); and since they all knew the same thing, for practical purposes of the model, they were all alike in most relevant respects. Hence, far from congealing an intellectual travesty, it seemed plausible (not to mention mathematically convenient) to portray the entire economy as playing out between the ears of a single person. Thus the ‘representative agent’ fiction in fact constitutes a projection of deep commitments of the existing elite of the orthodox economics profession. That is why it became a shared presumption of neoliberals who believe in the natural healing powers of the market, as well as ‘New Keynesians’ looking for reasons why the economy falters.
Consensus is often mistaken for groupthink within the DSGE model, and this tends to mirror a sociological characteristic of the economics profession. Both the ‘agent’ in the DSGE model and in the American profession could not imagine an effective search for truth emerging out of substantial persistent disagreement over fundamentals. Agents in orthodox models are enjoined from ‘agreeing to disagree’; and economists in good standing must knuckle under as well. The utter revulsion for anything smacking of real heterodoxy, combined with a fear of appearing ‘unscientific’ to outsiders, eventually led to a ‘donnybrook’ far more drastic than any embarrassment or compromised legitimacy that might have otherwise previously arisen from strident disagreement in the court of public opinion. Indeed, the effect of the crisis has been to bring those repressed disputes out into the open.
Once the presumption of omniscience broke down, then the consequences of the banishment of methodological self-scrutiny began to be felt. Both the criticisms and defenses of DSGE, at the Congressional hearings and elsewhere after the crisis onset, were distressingly unsophisticated, as one might expect as fallout from the ostracism of methodological thought. Robert Solow testified in Congress that DSGE models ‘didn’t pass the smell test’, introducing a novel olfactory standard for scientific model choice. The defense of DSGE at the hearings by V.V. Chari equally reveals the paucity of resources (and rhetorical skills) possessed by contemporary economists:
So, any interesting model must be a dynamic stochastic general equilibrium model. From this perspective, there is no other game in town. Modern macroeconomic models, often called DSGE models in macro share common additional features. All of them make sure that they are consistent with the National Income and Product Accounts. That is, things must add up. All of them lay out clearly how people make decisions. All of them are explicit about the constraints imposed by nature, the structure of markets and available information on choices to households, firms and the government. From this perspective DSGE land is a very big tent. The only alternatives are models in which the modeler does not clearly spell out how people make decisions. Why should we prefer obfuscation to clarity? My description of the style of modern macroeconomics makes it clear that modern macroeconomists use a common language to formulate their ideas and the style allows for substantial disagreement on the substance of the ideas. A useful aphorism in macroeconomics is: ‘If you have an interesting and coherent story to tell, you can tell it in a DSGE model. If you cannot, your story is incoherent’. 
It is one thing to assert that you personally cannot imagine any other possible way to discuss the macroeconomy than the DSGE; it is quite another (in public, before a tribunal) to insist that no one else can either, without babbling incoherently. But of course there were alternatives studded throughout the literature, which had been proposed repeatedly by those seeking to exit the ‘big tent’ of orthodox macroeconomics. This is not the appropriate venue to examine those proposals; rather, it is to ask – why are so many economists so loathe to let go of DSGE? All the usual considerations of inertia and sunk costs of intellectual commitment come into play; but there is something else as well. One way to understand such intransigence is to explore the possibility that excision of the DSGE cannot staunch the bleeding of the American economics profession; for these economists, renunciation of the DSGE is a slippery slope to the dissolution of the entire economic orthodoxy. Après DSGE, la déluge.
There were a few economists who had proposed that the monolithic coherence of macroeconomics and neoclassical microeconomics was a sham, but there were not accorded much respect, and were notably absent from the Congressional hearings. Perhaps the most prominent was the European economist Alan Kirman. He headed a group of scholars who issued the ‘Dahlem report’ (Colander et al., 2009) excoriating the economics profession early on in the crisis, and explained his own position in a widely read blog post (Kirman, 2009). Kirman suggested that the root problem with macroeconomics was really philosophical: the vaunted foundations of the DSGE model in full neoclassical general equilibrium were illusory. First, he cited some technical results dating from the 1970s stating that full general equilibrium analysis does not allow one to make much of any aggregate generalizations from the behaviour of a diverse group of neoclassical agents; and furthermore, except under some strained special circumstances, one cannot guarantee the existence of a unique or stable general equilibrium. The reason that DSGE models could pretend that there was a full macroeconomic equilibrium was that the fiction of a one-person economy was one of the few cases where (obviously) the individual is identical to the aggregate economy, and that existence proofs were available in that case for a unique stable equilibrium. To put it more bluntly, DSGE models were predicated upon the only arbitrary special case where neoclassical microeconomics and macro could be logically reconciled. Instead of drawing the conclusion that the marriage of micro and macro was doomed, and the DSGE a stillbirth, the profession had chosen to pledge its troth to an outré mutant case and call it the whole world. It would be as though a religious fanatic arranged to live in a hermetic world comprised only of Christian Scientist cyborgs, so that he need never encounter anyone who might call his faith in natural healing into question. As Kirman (2009) wrote: ‘both the development of the DSGE model and the efficient markets hypothesis share a common feature – despite the empirical evidence and despite their theoretical weaknesses, their development proceeded as if the criticism did not exist’.
If there had been a contingent of methodologists integrated into the profession, they might have insisted that all the brouhaha about jettisoning the DSGE model was a weary sideshow, since the gnawing problem that the economic orthodoxy was intent on avoiding was gauging to what extent the crisis had voided the legitimacy of neoclassical micro- economics. Legions of macroeconomists were mobilized into action by the crisis not to address its dire consequences, but instead to obscure this threatening conclusion through smoke, mirrors and legerdemain. No one who wanted to maintain their position in academia would countenance the possibility that amputation of the DSGE would result in the patient bleeding to death. So instead they promoted endless consultations over the health of the DSGE – and even Congress was snookered into the pointless game.
This argument would begin by characterizing the two options promoted by economists who thought of themselves as orthodox macroeconomists after the crisis hit. The first was to insist that all that ridicule of the DSGE model was simply ignorant: all those aspects of the crisis that critics said could not be accommodated by the model, had in fact been fully taken into account somewhere in the journal literature. You want heterogeneity of agents – we’ve done it. We’ve got models with frictions galore, and we have even coquetted with bounded irrationality. You claim there are big political divisions within macro and that DSGE only describes neoliberal fantasies of self-regulating markets; but the ‘freshwater–saltwater’ divide is just an illusion. We’ve got DSGE models to conform to all ideologies. We even have a version of the model here and there that mentions banks and credit.  All those nagging complaints are baseless, and mired in an outdated impression of real business cycle theory back in the 1980s.
This option, while commonplace, is utterly unavailing. A methodologist would point out, as I have done, that the canonical DSGE assumes its canonical outlandish format in order to ‘save’ its microfoundations, viz., the non-negotiable prescription that macro and neoclassical microeconomics are one big unified theory. All these current fragmentary amendments to render the DSGE model more ‘realistic’, or perhaps more politically acceptable to the ‘New Keynesians’, are self-contradictory, since they attempt to mitigate or ‘undo’ the microfoundations which had been imposed by decree from the outset. It ends up being one more instance of economists asserting both A and not-A simultaneously. By thrusting the rabbit into the hat, then pulling it back out with a different hand, the economist merely creates a model more awkward, arbitrary and unprepossessing than if they had just started out explicitly to incorporate confused heterogeneous agents, dodgy banks, consciously duplicitous CDOs, informationally challenged markets, and all the rest of the usual suspects for the crisis, minus the neoclassical window dressing. If you allowed freedom of amendment to the DSGE in this manner, you would end up with models that violated the Lucas critique in a more egregious fashion than the earlier Keynesian models these macroeconomists love to hate. Thus, a ‘more realistic DSGE’ ends up as a contradiction in terms.
The second option, the one favored by the really high-profile attackers of DSGE like Robert Solow and Paul Krugman, was to roll back the clock to 1969, and pretend that the whole sequence of sordid developments leading up to DSGE never happened. Sometimes this was portrayed as a ‘return to Keynes’, although a historian might aver that the American profession was never all that enamored of the actual Keynes and his writings. Nevertheless, this latter group was extrapolating from the heady days of late 2008, when all thoughts of DSGE were nowhere to be found. As the economic historian Greg Clark (2009) put it:
The debate about the bank bailout, and the stimulus package, has all revolved around issues that are entirely at the level of Econ I. What is the multiplier from government spending? Does government spending crowd out private spending?
…If you got an A in college Econ I, you are an expert in this debate: fully an equal of summers and Geithner.
This proposal was, if anything, even more implausible than the revision of the DSGE. Most macroeconomists would rather abandon the field than admit that all their technical sophistication was superfluous, and purge the lessons they learned at the feet of Robert Lucas and Thomas Sargent. The entire field was populated by people drilled in contempt for reading Keynes, and confirmed in their convictions that those 1960s-era models, like the old-fashioned IS-LM and Phillips curve, fully deserved to be tossed on the trash-heap of history. Yet, even if some magic wand waved away generations of inertia, there was no guarantee that if you re-ran the tape of history over one more time, starting once more in 1969, the neoclassical orthodoxy would not just end up rejecting all those 1960s-era models all over again. For Lucas and Sargent had a point: the earlier ‘Keynesian’ macroeconomics as it existed back then was logically incompatible with neoclassical microeconomic theory, and if something had to give, it would be Keynes, and not the Arrow–Debreu theory of general equilibrium, at least in America. Hence, by a circuitous route I arrive once more at the lesson of this section: the real bone of contention is not the DSGE model per se, but rather the pre-eminence of legitimacy of neoclassical microeconomics. The DSGE model is a herring of the brightest red.
One of the places where the 2010 Congressional hearing missed an opportunity at gaining an understanding the true character of the path to dominance of the DSGE was in not inviting a historian and methodologist to provide meta-commentary upon the strange testimony offered by the invited participants. Not only would the missing witness have provided some context for the seemingly orthogonal positions voiced by Solow, Chari, Page and Winter, and pointed out that it was no accident that no substantial alternative to neoclassical theory had a place at the table; but she might have also suggested that the Congress (or its delegated agencies) itself deserved its own fair share of the blame for the rise to intellectual monopoly of the DSGE. To suggest where such testimony might have ventured, I here cite another occasion of testimony before the same House committee dating back to March 1981. Then the issue was a Reagan administration drive to cut the funding of economic research from the NSF. The speaker was Harvard economist Zvi Griliches:
It is ironic and sad that whoever came up with these cuts does not even recognize that most of the recent ‘conservative’ ideas in economics – the importance of ‘rational expectations’ and the impotency of conventional macroeconomic policy, the disincentive effects of various income-support programs, the magnitude of the regulatory burden, and the arguments for deregulation – all originated in, or were provided with quantitative backing by NSF supported studies.
Griliches was merely stating the obvious: economists produce the sorts of knowledge that its patrons desire, within the trajectory of its accumulated intellectual heritage; that list of patrons includes neoliberal elements within the government, with their allies in selected ranked economics departments. Congressmen today should not act as though the DSGE model and its precursors were somehow foisted upon unsuspecting regulators and an innocent public by imperious economists. Mostly, Americans just got what they paid for.
25. The Hearing Charter (quoted above) of the House Committee on Science and Technology and sworn testimony of economists Sidney Winter, Scott Page, Robert Solow, David Colander and V.V. Chari can be found at http://sciencedems.house.gov/ publications/hearings_markups_details.aspx?NewsID52876.
26. This is not the place to run through the tortured history of orthodox neoclassical macroeconomics, from the ‘neoclassical synthesis’ through Friedman’s monetarism to ‘New Classical’ to ‘real business cycles’ to ‘New Keynesians’, and thus finally to DSGE. Some good sources on this massive literature are Quiggin (2010) and Mehrling (2010). As usual, the history is frequently accompanied by utterly naive methodological statements: ‘the field looked like a battlefield. Researchers split in different directions, mostly ignoring each other, or else engaging in bitter fights or controversies. Over time however, largely because facts have a way of not going away, a largely shared vision of fluctuations and of methodology has emerged’ (Blanchard, 2008, p. 2).
27. V.V. Chari, Testimony before the Committee on Science and Technology, Subcommittee on Investigations and Oversight, US House of Representatives, 20 July 2010. For a similar argument, see Kocherlakota (2010).
28. See, for instance, Colander et al (2008), Meeusen (2010) and Howitt (2006, 2008).
29. For the curious, he was referencing the Sonnenschein–Mantel–Debreu theorems in microeconomics. See Rizvi (2006) for the relevant background.
30. This position is almost exactly repeated in the Chari testimony, and in Kocherlakota (2010), De Grauwe (2010) and Maskin (2009). All reactions in this paragraph are paraphrases of DSGE defenses found in these sources.
31. I shall indulge in just one example of how such protests were so misleading as to border on mendacity. The notion that DSGE models, which rarely incorporated money, much less a banking sector, could indeed handle a financial crisis, is often motivated by citation of the Diamond–Dybvig model (Diamond and Dybvig, 1983), which is a model of a run on a solvent bank. Since most of the main institutions in the current crisis were insolvent, and not merely illiquid, this model turns out to be utterly irrelevant. Furthermore, since most DSGE models encompass a presumption of the EMH at base, and accept the Modigliani–Miller theorem, there are no functions for finance to perform in the models.
32. See, for instance, Mirowski (forthcoming).
33. Quoted in Scheiding and Mata (2010).