Navigation – Plan du site
Revue des livres
Essais critiques

The Quest for Microfoundations

James E. Hartley
p. 237-247
Référence(s) :

Roger E. Backhouse and Mauro Boianovsky, Transforming Modern Macroeconomics: Exploring Disequilibrium Microfoundations, 1956–2003, Cambridge University Press, 2012, 215 pages, ISBN: 978-1-107-02319-2

Pedro Garcia Duarte and Gilberto Tadeu Lima (eds), Microfoundations Reconsidered: The Relationship of Micro and Macroeconomics in Historical Perspective, Edward Elgar, 2012, 246 pages, ISBN: 978-7-78100-409-8

Texte intégral

1 The Fairy Tale History of Macroeconomics is, like all such stories, a tale of triumph. Once upon a time, there were the Dark Days of Macroeconomics when Keynesian IS-LM curves roamed the land. Then came the Troubled Times when the Keynesian models floundered in the Great Stagflation of the 1970s. But, along came an intrepid set of warriors, bearing new and better models, sorting through the wreckage of the Great Destruction of the Keynesian model. After Keynesian Economics came the promised land. And so, macroeconomists now live in a better country, where New Classical and New Keynesian economists live in harmony, building Walrasian general equilibrium models, the only debate being whether there are sufficient frictions in the economy to prevent Pareto Optimality from prevailing all the time.

2 The defining feature of the macroeconomic models being used in the last three decades which both unites these models into one large group and differentiates these models from the preceding generation of models is microfoundations. Most clearly articulated by Thomas Sargent (1982) in “Beyond Supply and Demand Curves in Macroeconomics,” the microfoundations argument is that macroeconomic models, to be of any value at all, must begin with the objective functions and constraints of individual decision makers. While there are modern macroeconomic models which abandon market clearing and rational expectations (indeed, Sargent himself has published such models), there are few modern macroeconomic models which reject the need for microfoundations. Nowadays, to say that a model lacks rigorous microfoundations is tantamount to saying the model does not need to be taken seriously.

3 It is thus not without interest that a pair of books have recently been published which critically examine exactly this feature of modern macroeconomics. Both books are fundamentally asking the same question: was the rise of modern microfoundations as the most important criterion for evaluating macroeconomic models inevitable? Roger E. Backhouse and Mauro Boianovsky’s Transforming Modern Macroeconomics: Exploring Disequilibrium Microfoundations, 1956–2003 is an encyclopedic methodological history of the quest for microfoundations. Pedro Garcia Duarte and Gilberto Tadeu Lima’s edited volume, Microfoundations Reconsidered: The Relationship of Micro and Macroeconomics in Historical Perspective, is a hodgepodge of papers, some more successful than others, circling around the question. Taken together, the two books offer up the prospect of an interesting historical counterfactual: What would modern macroeconomics look like if a different view of microfoundations had won the day in the mid-1970s?

4 It is worth noting at the outset that nothing in either of these two books and nothing in the literature being discussed therein disputes the need for microfoundations broadly defined. In all of this literature, it is an article of faith that macroeconomic models should have some explanation of how macroeconomic activity relates to the decisions being made by individuals. Similarly, the literature being surveyed here does not dispute the need to build models of some sort. This is not, in other words, some sort of dispute in which people standing outside the discipline of economics are lobbing bombs into the crowd. This is a technical dispute. Given the assumption that the best way to understand the macroeconomy is to write down models with microfoundations, what type of microfoundations will make for the best model? As these books amply demonstrate, that is a question which is vastly more complicated and involved than it is commonly believed to be.

1. Disequilibrium Microfoundations

5 Sometimes a book covering the history of economic thought gets it just right. Transforming Modern Macroeconomics is such a book. It makes a worthy sequel to David Laidler’s Fabricating the Keynesian Revolution (Laidler, 1999); indeed, if one wanted a thorough history of 20th century macroeconomic thought, these two volumes together are the best place to start.

6 What unites the Laidler book and the Backhouse and Boianovsky book is that they transcend Whig history and venture into what we might call Frostian history. Constantly peering down roads less traveled, these books offer perspectives on not just what has been, but also what might have been had economics taken a slightly different path. Early in their book, Backhouse and Boianovsky cite a passage from a letter Leijonhufvud wrote Patinkin in 1974:

In theory I had been taught general equilibrium models on the one hand and Keynesian models (by Modigliani) on the other; I was struck mostly by the incompatibilities, offended by the “Neoclassical synthesis”, and saw no palatable way out. It was in searching for a way out from this cul-de-sac that I eventually started to “backtrack” into doctrine-history, trying to identify, as it were, the forks in the road where the major decisions on what conceptual experiment to pursue were made. (56, citing Leijonhufvud to Patinkin, June 15, 1974, Don Patinkin Papers, Box 65, David M Robertson Rare Book and Manuscript Library, Duke University)

That description is a good summary of the idea underlying Transforming Modern Macroeconomics, a history of forks in the road to modern macroeconomics.

7In the traditional telling of the development of macroeconomics, the triumphant theory wins because it beats the alternatives in explaining the economy. But, left unstated in the triumphalist account are a host of questions. How do we know when the economy has been explained? What are the questions which the theory sought to answer? How do we know the set of answers provided by this model are better than the set provided by that model? How do we know the criteria being used to answer the last question are the right criteria? These sorts of questions are rarely addressed, let alone answered, in the modern macroeconomics literature. Backhouse and Boianovsky are also implicitly raising yet another question: What are the problems in the previous generation of models which the next set of models are trying to solve?

8Modern macroeconomic theory starts with the assumption that the normal state of the economy is equilibrium. In its New Classical version, equilibrium is assumed throughout; in its New Keynesian version, equilibrium is thwarted by some explicitly introduced imperfection in the economy. All of this is very much in line with a Walrasian view of the economy in which the model is akin to the theory of the triangle and the goal of economists (or geometers) is to study the pure theory of the economy (or triangle) and note why real economies (or drawn triangles) fail to fit the theory. New Classical economists do the former; New Keynesian economists do the latter.

9The story of disequilibrium models starts, curiously enough, with Patinkin’s Money, Interest, and Prices (Patinkin, 1956), which is also generally listed as one of the most important building blocks of the equilibrium macroeconomic models. Buried in the latter part of the book, there is a sketch of a model, not nearly as fully developed as the model in the rest of the book, in which involuntary unemployment arises because prices and wages do not respond immediately to economic shocks. A firm faced with a decline in demand for its product at the prevailing price responds by cutting employment; a fall in the real wage is not sufficient to induce the firm to increase its demand for labor because it is concerned with insufficient demand for its product. This chapter on involuntary unemployment is less important for the immediate effect it had on the development of macroeconomics (it had little immediate effect), but rather because it is the first inkling of a type of macroeconomic model which would soon be developed by others.

10Clower and Leijonhufvud, in work which was not immediately seen as linked to Patinkin’s story of involuntary unemployment, step onto the stage to provide further tentative glimpses of what a fully articulated disequilibrium model might show. Clower (1965) distinguishes between the notional demands which appear in a Walrasian equilibrium model and the effective demand which is realized in a Keynesian model. The economy could get stuck in a disequilibrium state if the effective demand for products is low and firms thus have no incentive to hire additional workers to produce more products they cannot sell. Leijonhufvud (1968) provides a similar story of coordination failure centering on the interest rate.

11There is a direct line from the work of Patinkin, Clower, and Leijonhufvud to the more complete fixed price models of the 1970s, the most prominent of which is the work of Barro and Grossman (1971). (So close is the connection, Leijonhufvud categorized Barro and Grossman’s work as simply a more teachable version of what had previously been developed.) The Barro and Grossman model, along with similar models by Solow and Stiglitz, Benassy, Muellbauer and Portes, and Neary and Stiglitz, were built up from the notion that prices do not automatically adjust to the market clearing level. Obviously, if prices are stuck at a place where the markets do not clear, then the economy will behave in a very different manner than would be predicted if prices were continuously at market clearing levels. Instead of asking why markets do not clear, these models, in effect, ask what happens if they do not.

12Fixed prices immediately became the center of the Great Divide in macroeconomics. Partly because the 1970s was not a particularly good time to be talking about fixed price models, the idea of microfoundations soon became synonymous with the idea that for a disequilibrium model to be rigorous, it would have to explain why rational agents would not move prices to a market clearing level when it was beneficial to everyone to move prices to that point. Failing to see any reason to assume utility maximizing agents would ever act in unprofitable ways, New Classical economics was born. New Keynesian economics quickly followed by showing that sometimes there will be a friction which will prohibit the immediate movement of prices.

13But, another answer was possible, and it was an answer explored at great length by a set of economists not normally thought of as macroeconomists. And it is this set of models which posits the most intriguing question raised by Backhouse and Boianovsky: why were these models abandoned? Embedded in the assumption of a market clearing model is the tatonnement process in which a Walrasian auctioneer moves prices to their market clearing level. Now obviously the Walrasian auctioneer is a pure theoretical construct, not only having no counterpart in the real world, but also not having even an approximate counterpart. In the real world, who moves prices to the market clearing level? The Walrasian model with its perfectly competitive price takers has nobody with the ability to set prices. What happens, then, when a model is built in which the agents in the model not only can set prices, but any move from disequilibrium prices to equilibrium prices involves these agents moving the prices? What happens if there is no tatonnement process?

14In a remarkable set of models built from the 1950s through the 1970s, economists tried to answer this question. Dreze, Benassy, Negishi and Hahn all worked at building a general equilibrium model in which agents have price setting power. These models end up with imperfect competition because that is the only way to explain price setting behavior. And most importantly, the end result of these models is a story of disequilibrium, not because of some friction preventing the natural movement to equilibrium, but because there is no inherent tendency to move to equilibrium in the first place. Disequilibrium is the default state of the macroeconomy; a macroeconomy in equilibrium would be a bizarre coincidence.

15 Backhouse and Boianovsky argue that this set of disequilibrium models was not abandoned because they did a poorer job of answering the questions asked by the equilibrium models which triumphed. The two sets of models weren’t even asking the same question. One set posits equilibrium as the default and works out the implications; the other set asks how prices are determined. The disequilibrium model’s implications for macroeconomics were never fully developed. But, the problem with the Walrasian equilibrium model remains. In the absence of the auctioneer, how do prices move to the equilibrium level? A good microfoundations, in this argument, cannot simply assume that they magically do so, but instead that movement must be explained in terms of microeconomic agents with the ability to set prices. And once we have posited that set of agents, it follows that there is no reason to assume that those agents will move prices to equilibrium levels. What if, for example, it is in the best interest of a firm with market power to set non-market-clearing prices? What if a microeconomic theory of price adjustment implies that disequilibrium is the normal state of the economy?

16Backhouse and Boianovsky illuminate two important threads in the quest for microfoundations: the attempt to ground Keynes in microeconomic theory and the attempt to explain price setting in the Walrasian model. Both lines of research fell by the wayside as equilibrium Walrasian models rose to prominence by showing such promise in the 1970s. Yet, nearly a half a century later, it is not clear that the equilibrium macroeconomic models have made any progress in solving the questions which motivated the early disequilibrium models. Backhouse and Boianovsky point clearly to an unfinished research agenda.

2. Reconsidered Microfoundations

17 That unfinished research agenda, the quest for a good set of microfoundations, is exactly what is promised by the title of Duarte and Lima’s volume, Microfoundations Reconsidered: The Relationship of Micro and Macroeconomics in Historical Perspective. It is not hard to imagine the genesis of this book. Gather a half-dozen methodologically inclined economists at a conference in Brazil, have them give papers on the topic promised in the title, and then publish a volume with those papers. The time seemed ripe to revisit the idea of microfoundations, a catch-phrase much used by economists with surprisingly little discussion about what that word might mean. As an added bonus, the book could be framed as a response to the financial crisis of 2007-2008, though like many books framed in such a manner, it is hard to see much connection between the topic of this book and the mechanics of banking regulation of mortgage markets in the United States. Nonetheless, the idea had promise, the economists gathered were all top-notch, and presumably the conference itself was fun and interesting.

18 But, what about the book? It starts off well enough with an essay by Kevin Hoover, “Microfoundational Programs”, in which he notes that while there were three distinct microfoundations programs in the post-war era, the winning program was methodologically the least promising. But first he adds a prehistory, looking at Frisch and Keynes, neither of whom explicitly discuss the microfoundations program. In Lucas and Sargent’s telling, Frisch and Keynes were the source of the problem, creating an entity called “macroeconomics” because of an inability to solve a mathematical problem which can now be solved by the newest generation of economists. However, going back and actually reading what Frisch and Keynes were doing reveals that the divorce of microeconomics and macroeconomics was not a step taken because of an inability to see a possible reconciliation. Rather, the creation of macroeconomics was a deliberate move to face the realization that the microeconomy was too big and too heterogeneous to assume that the macroeconomy would be a simple replication of microeconomic behavior. Rather, macroeconomic behavior emerges, in interesting and distinct ways, from the interaction of microeconomic agents. The whole is greater than the sum of its parts.

19 After Keynes, the quest for microfoundations began. The first incarnation described by Hoover is what he calls the “general equilibrium program,” which is the story told by Backhouse and Boianovsky. Beginning with Hicks, the general equilibrium program attempted to reattach macroeconomics and microeconomics by creating a sufficiently complicated model. The end result was the highly abstract models currently in the realm of microeconomic theorists and the “historical dead-end” of disequilibrium models. Hoover’s explanation for the failure of this program is that it was too distant from reality, offering little ability to construct econometric tests of the models or their implications.

20 The second microfoundational program, the “aggregation program,” is econometric in nature. With its roots in Klein’s work, this approach generated the mammoth empirical macroeconometric models of the 1960s. These models consist of an ever-increasing number of macroeconomic equations, beginning with the assumption that microeconomic models can give us an idea of the parameters needed in the equation and macroeconomic data can be used to verify the relationships. Particularly in the wake of the Lucas critique, these models have been viewed as a relic of an earlier age, but it has been a curious death. Lingering in the shadows like some wraith, they continue to live on in the policy analysis being done not only by the government, but also by private forecasters who have a large financial stake in getting accurate forecasts of macroeconomic behavior. This should be, but is not, a troubling sign for the dominant microfoundations view; if macroeconometric models are of as little use as we say they are, then why do they seem so, well, useful?

21 Hoover’s third microfoundations program, the “representative-agent program,” won the day, but in a curious fashion. There was never an explicit theoretical argument proving a rationale for this approach to microfoundations. There was also never an empirical justification for this approach. Indeed, its victory is a pure example of the triumph of faith. By repeated assertion of its truth, this program has become the essential part of the modern macroeconomics toolbox. Hoover implicitly argues that the failure of this program to adequately pay attention to the problems of the programs which came before it will in the end render it to increasing levels of empirical irrelevancy.

22 Thus, by the end of the first chapter, the book in which Hoover’s essay is found seems quite promising. But, again like many a conference volume, things go downhill. (This confirms the unwritten rule for both conference volumes and pop albums: always lead with the best item.) Consider the trio of papers by Wade Hands, Philip Mirowski and Michel De Vroey.

23 Hands’ paper, “The Rise and Fall of Walrasian Microeconomics: The Keynesian Effect,” sets out to document that what was called Keynesian macroeconomics and what was called Walrasian economics are not nearly as opposed as the conventional history would suggest. Indeed, the two theories evolved together, with the prominent developers of Walrasian models being quite concerned with incorporating macroeconomic Keynesian ideas into the microeconomic model being crafted. Hands cites four specific examples. First, in building aggregate demand, market demand loomed as more important than individual-based choice-theoretic demand. Second, Walrasian sequential tatonnement was abandoned in favor of tatonnement as a means of showing stability. Third, path-dependency is not incorporated into the model. Fourth, nominal income looms large as an explanatory variable. Hands argues that all of these features of the Walrasian model as it was developed in the post-war era were explicitly the result of the influence of Keynesian macroeconomics.

24 Mirowski’s paper, “The Cowles Commission as an Anti-Keynesian Stronghold 1943–54,” covers the same ground as that surveyed by Hands. Mirowski’s conclusions are worth repeating verbatim. “[S]ubsequent history demonstrates beyond a doubt that the Keynesian system is inherently incompatible with neoclassical microeconomic theory in its most rigorous manifestation, viz., the theory of Walrasian general equilibrium” (133). Moreover, “In the meantime, historians can demonstrate that Keynesian ideas were hobbled from birth, in the sense that two (and maybe even all three) of the postwar schools of American neoclassical economics were in a very palpable sense hostile to the theoretical content of the General TheoryThere is nothing ‘radical’ about this proposition; it is simply historical fact” (134, italics in original). And finally, Mirowski describes the “Cowles Creed”: “everyone who is said to count as a real economist necessarily believes what our crowd believes, viz., that Walrasian general equilibrium theory is the only game in town….There is but one God, its name is Walras, and Arrow-Debreu is his prophet. In this Passion Play, Keynes was just a minor Simeon Stylites” (160).

25Enter Michel De Vroey. His paper, “Microfoundations: A Decisive Dividing Line Between Keynesian and New Classical Macroeconomics” looks at the same question under different labels. The translation is a bit rough, but as soon as one realizes that De Vroey’s Lucas is simply the inheritor of the Walrasian general equilibrium models discussed by Hands and Mirowski, then we have reentered the same debate: what is the connection between Keynesian microfoundations and those found in modern macroeconomic models? To answer that, De Vroey again sends us back to the terminological drawing board. The microfoundations provided by Keynesian macroeconomics are dubbed Hayek-Patinkin microfoundations, which are then contrasted to Lucasian microfoundations. Furthering the exercise in taxonomy, De Vroey argues that the former is actually a more general version of the latter. (With appropriate translation of terms, De Vroey’s paper is the closest of these three papers to the discussion in Backhouse and Boianovsky.)

26De Vroey’s “Hayek-Patinkin microfoundations” are an argument that a model has microfoundations as long as the agents in the model are all assumed to have optimizing plans. Whether those plans are realized is irrelevant. These are the microfoundations provided by Keynesian macroeconomics. Lucas and the Walrasian general equilibrium modelers came along and insisted that optimizing plans are not sufficient, that the plans must also be realized. Microfoundations in this definition is simply the market clearing assumption. De Vroey concludes that New Classical (or Walrasian) microfoundations are just a subset of the more general theory of microfoundations found in Keynesian economics.

27We now have three stories covering the same ground: 1) the Walrasian model was heavily influenced in its development by the implications of the Keynesian model, 2) the Walrasian model is antithetical to the Keynesian model, and 3) the Walrasian model is just a special case of the Keynesian model. The fact that papers by Hands, Mirowski and De Vroey would argue opposing things is not, in and of itself, a failing. After all, debate is healthy. However, the fact that neither the editors nor the authors seem willing to acknowledge that these papers are in fact arguing opposing things is curious. Even more curious is the fact that Mirowski added a short section to his paper post-conference which begins by noting that several people noticed his paper and Hands’ paper directly contradict one another and then he proceeds to insist that they do not really contradict each other because if we define “Keynes” in different ways then somehow the papers don’t really contradict one another and (move along folks, nothing to see here) maybe we should just hurry along to the conclusion. (Nobody seems to have noticed that De Vroey’s paper was covering the same ground as the other two papers with yet another conclusion.)

28Why this desire to ignore at best and evade at worst the quite obvious contradictions between the papers by Hands, Mirowski and De Vroey? One possible explanation: this contradiction points right at the rather sizeable criticism of this volume as a whole. Despite the promise of the title of the volume, the editors’ introduction arguing that the conference produced insight, and the promise of the first paper which traces some interesting themes, the book quickly devolves into Just So stories. How did the microfoundational Walrasian general equilibrium model become the way it is? Hands argues Keynes had a deep and lasting influence on its development. Just So. Mirowski argues Keynes had zero influence on its development. Just so. De Vroey argues that it is a special case of the Keynesian model. Just so. And just how did the rhinoceros get his skin?

29While the volume taken as a whole gives little reason to trust that any of these three papers are onto something, at least they address the topic promised in the title. Robert Leonard’s paper on Oskar Morgenstern bears at best a tangential relationship to the topic of the volume. Pedro Garcia Duarte rounds out the volume by dropping the “historical” and “reconsideration” parts of the evaluation and functioning as a good tour guide of the contemporary methodological landscape of New Classical and New Keynesian models.

30What is one to make of a book like this? It is, again like many a conference volume, much better considered not as a book, but as an issue of a journal. All six of these papers would easily have found homes in an academic journal; all six papers will be of interest to someone. However, nobody ever sits down to write a review of Volume 32, issue 6 of a journal because we know and expect that the articles therein are united by nothing more than the broad classification of the subject matter of the journal. In an issue of the Journal of Economic Methodology or the Journal of the History of Economic Thought, all six of these papers are perfectly fine. But, before publishing such a journal issue as a book, it would seem to be incumbent on the editors to provide the connecting tissue between the parts, and on those grounds, this book needs more editorial work.

3. Looking Forward

31Economic Methodology and its cousin the History of Economic Thought are, to put it mildly, not held in high regard by the profession as a whole. This results in a curious feature of methodological debate: it always feels one-sided. Methodologists swoop down and criticize current practice. Historians of economic thought follow behind and show the poor foundations of current practice. Instead of fierce resistance, these forces are met with a vacant battlefield and deafening silence. And the result? Current practice stands on the throne, seemingly oblivious to the foes on the battlefield. Evaluating books like the ones being considered here thus poses a certain problem. Suppose the books win the methodological war, but the practices of the opponents are not modified in the least. Was the campaign a success?

32There is no doubt that Backhouse and Boianovsky have showcased a world of possibilities, some of which are quite intriguing. The contributions in the Duarte and Lima volume have also provided thought-provoking reflections on the meaning of microfoundations. But, who is the audience for these books? The best parts of these two books deserve a much wider readership than they will undoubtedly find. This is not necessarily a reason for despair. In many a good fairy tale, there is a hero who stumbles on a book of forgotten lore and uncovers a truth which brings an empire to its demise. There is some great economics lurking in these books, some economics that an enterprising graduate student, looking for a twist in building a new generation of models, could usefully exploit. The next revolution in macroeconomics may come from Big Data creating ubersophisticated macroeconometric models. But, it also may come from the simple realization that there are other forms of microfoundations than the one which the profession currently look at as if it is the only possibility.

Haut de page

Bibliographie

Barro, Robert J. and Grossman, Herschel I. 1971. A general disequilibrium model of income and employment. American Economic Review, 61(1): 82-93.

Clower, Robert. 1965. The Keynesian counterrevolution: A theoretical appraisal. In F. H hahn and F. P. R. Brechling (eds.) The Theory of Interest Rates. London: Macmillan, 103-125.

Laidler, David. 1999. Fabricating the Keynesian revolution: studies of the inter-war literature on money, the cycle, and unemployment. Cambridge: Cambridge University Press.

Leijonhufvud, Axel. 1968. On Keynesian economics and the economics of Keynes. Oxford: Oxford University Press.

Patinkin, Don. 1956. Money, Interest, and Prices. An Integration of Monetary and Value Theory. Evanston, IL: Row, Peterson and Company.

Sargent, Thomas J. 1982. Beyond demand and supply curves in macroeconomics. American Economic Review, 72(2): 382-389

Haut de page

Pour citer cet article

Référence papier

James E. Hartley, « The Quest for Microfoundations », Œconomia, 4-2 | 2014, 237-247.

Référence électronique

James E. Hartley, « The Quest for Microfoundations », Œconomia [En ligne], 4-2 | 2014, mis en ligne le 01 juin 2014, consulté le 26 avril 2017. URL : http://oeconomia.revues.org/781

Haut de page

Droits d’auteur

Licence Creative Commons
Les contenus d’Œconomia sont mis à disposition selon les termes de la Licence Creative Commons Attribution - Pas d'Utilisation Commerciale - Pas de Modification 4.0 International.

Haut de page
  • Logo Association Œconomia
  • Logo CNRS
  • Les cahiers de Revues.org