What Went Wrong in Macro - Historical Details
Warning: What follows is inside baseball, and oldster inside baseball at that. Unless you received a Ph.D. degree before 1990, you may find it hard to keep track of the players.
Before delving into the details here, it might help to read a previous post that gives an overview.
My sense of what went wrong in macro is based partly on what I observed as a Ph.D. student in the economics department at MIT during the 1977-78 and 1978-79 academic years. For reasons that have no bearing on the discussion here, I ended up transferring to Chicago and finishing my Ph.D. there, with Robert Lucas as one of my thesis advisors. This matters only in the sense that by chance, I was on the inside looking out from both sides of the biggest divide in economics.
During my time at MIT, Robert Solow was harshly critical of the new classical macro models pioneered by Robert Lucas, dismissive in a way that seemed to me to skirt uncomfortably close contempt. I recall hearing the same type of criticism from Frank Hahn, who must have been visiting MIT. Looking back, perhaps I misinterpreted them because I was not familiar with the sarcasm and put-downs that were a part of British intellectual life that Solow had to confront in his exchanges with Joan Robinson. But if it sounded like contempt to me, others may have heard it the same way.
Solow also seemed to be motivated to attack harshly because he was concerned that the type of model Lucas was developing might undermine political support for active countercyclical policy. To his credit, there was a legitimate basis for this concern. The new Chicago school of macro eventually did oppose an active response to the financial crisis and its aftermath. But the type of response that Solow exemplified may actually have contributed to the emergence of this new Chicago school. In retrospect, if the goal was to maintain support for active macro policy, the better course would have been to take seriously what the rebel group that was forming around Lucas was saying. This might have kept the rebels from cutting off contact with all outsiders, even those who were taking seriously the issues they were raising.
Once they cut off contact with the outside, these rebels developed a sub-culture that was more like what you’d expect to find among members of a platoon on the battlefield than among scientists parsing logic and weighing evidence. Loyalty and group cohesion took priority, so models that were illogical or inconsistent with the evidence went unchallenged. These values might have developed in any department, but they found support and encouragement at the University of Chicago, which was already committed to Stigler conviction instead of Feynman integrity.
As a result, real business cycle models, which were introduced by Prescott and Kydland in 1979, attracted support among the rebels, even though these models did an even worse job of matching the empirical evidence than the model that Lucas first proposed. In particular, they assumed away the possibility that monetary policy and inflation could have any interaction with output and employment. In a telling shift, the rebels also decided that econometrics was getting in the way. Word came down from the top that they were abandoning the cutting edge econometric work that Sargent had specialized in and using calibration in its place. Their mathematical theory turned opaque and misleading, exhibiting a pattern that I’ve called mathiness. So even if I’m right that there was still a window of opportunity in 1978, it didn’t remain open for long.
The alternative to derision would have been for skeptics to embrace and extend. This was what Stan Fischer and Rudi Dornbusch, who were supervising almost all of the Ph.D. students at MIT doing anything related to macro, were quietly doing at this time. Fischer, Dornbusch, and their students absorbed the rebel critique of traditional macro, saw what something was missing in the first generation of rebel models, and set about extending them. As a result, Fischer and Dornbusch trained a cohort of Ph.D. students at MIT who put the tools of modern macro to work and as Krugman has observed, turned out to be unusually influential. If Dornbusch and Fischer had set the tone for the response to Lucas and his followers, things might have turned out differently. But because of the inherent instability of acrimony, grievance, and factionalism, they and their students could not undo the effect of the more hostile response.
What’s sad is that the rebels could simply have declared victory. They had won the battle for mindshare among the next generation of macroeconomists. They could have taken credit for killing off the large multi-equation models that had previously dominated macro policy analysis.
From the MIT side, Lucas could have been embraced as a leading contributor to the larger intellectual program launched by Paul Samuelson, the founder of the department there. As Lucas solidified his position as the new intellectual leader at Chicago, he and Samuelson could then have healed the divide that had separated MIT and Chicago for decades. This divide had its roots in Stigler’s political agenda, but it played out in economics at the more damaging methodological level of Samuelson versus Marshall. At least for the young Robert Lucas, it was clear which side to root for in this fight.
To explain what I mean by the Samuelson program, you have to appreciate the sorry state of economics when Samuelson was a graduate student. Over the five decades from 1890 to 1940 (a time when physicists developed mathematical theories of statistical mechanics, quantum mechanics and both special and general relativity) economists avoided the use even of calculus and spent 50 years mired in the confusion spawned by the talky, market-by-market, supply-and-demand-ish approach to economic analysis codified in 1890 in Alfred Marshall’s Principles of Economics.
Samuelson saw that recovery for economics would require both the precision of mathematics and a commitment to models that could handle more than two variables at the same time. A supporting line of general equilibrium theory developed by Arrow, Debreu, and McKenzie developed mathematical tools that other economists could use. But it was Samuelson, with the help of his colleagues and students, who took the lead in putting general equilibrium theory to work answering real questions about how economies behaved.
There were many steps in the implementation of the Samuelson program. Here are the ones that seem most relevant for macro:
1. The first was to show that general equilibrium analysis could yield insights that Marshall’s partial equilibrium analysis could not. In one illustration, Samuelson, together with Wolfgang Stolper, started with two markets that Marshall would have recognized, each characterized by a horizontal supply curve implied by a constant returns to scale production function with capital and labor as inputs. This setup left Marshallians with nothing to say about how prices would respond to a policy change. By adding the economy-wide adding-up condition on capital and labor, Stolper and Samuelson derived a smooth production possibility frontier that showed how prices in the two markets responded together to any policy change. This would have been no surprise to Edgeworth, but it a sign of the damage that Marshall had done that this result was beyond the comprehension of economists trained using Marshall’s masterwork.
2. Next, Samuelson showed how to add time and money to an applied general equilibrium model, using a structure that he called overlapping generations. Technically it was a dynamic model, but a stationary one where nothing really changed.
3. Robert Solow (a close colleague of Samuelson’s at MIT) and Peter Diamond (who got his Ph.D. at MIT) showed how to describe the behavior of an economy in which things did change. By restricting attention to a single type of output, Solow developed a workable framework for talking about changes in wages, the return to capital, and total output. Diamond linked this model more explicitly to the rest of applied general equilibrium theory by adding Samuelson’s overlapping generations.
4. Then Robert Lucas showed how to add uncertainty to a version of the Samuelson and Diamond models. This let him pin down loose conjectures from Keynes about the role of expectations. (Keynes picked a title for his book “The General Theory …” that invoked the work of Einstein, but his talky arguments, like Marshall’s, would more accurately be described as pre-Newtonian.) By showing how to introduce imperfect information, Lucas offered an intellectually serious explanation of the empirical regularity known as the Philips curve. Moreover, it explained why the curve would shift when the inflation rate moves permanently higher. At a deeper level, Lucas signaled that the neoclassical synthesis, which provisionally exempted macro fluctuations from the Samuelson program, was no longer necessary. Macro fluctuations could be treated as just another application of general equilibrium theory.
5. Finally, Dixit and Stiglitz (who both did their Ph.D. work at MIT) showed how to extend general equilibrium models from a small, fixed set of goods to one with a potentially unlimited range of goods. This turned out to be important for macro fluctuations because it offered a better way to think about how prices are determined. It also meant that an economy could respond to a change in policy not just by changing the quantities of existing goods, but also by developing entirely new goods. As Dupuit had shown long ago, an adjustment along the the margin of new goods was very different from an adjustment that held the set of goods fixed. In Krugman’s hands, this extension finally let economists reason about the increasing returns implicit in Adam Smith’s discussion of the gains from an increase in the “extent of the market,” without getting tripped up by Marshall’s ill-conceived notion of of “external increasing returns to scale.”
The surprising person is this account is Lucas, especially given that he earned his Ph.D. at the University of Chicago, where George Stigler had enshrined Marshall as the symbol of a Victorian past to which economic theory and policy should return. (I’ll say later why I think that Stigler took exactly the opposite stance toward Marshall from the one that Samuelson adopted.)
In a “professional memoir” available here or here, Lucas explains that as an undergraduate, he had taken no economics, so he prepared for the Ph.D. program in economics at Chicago by working on his own through the book that Samuelson wrote as his graduate thesis, The Foundations of Economics.
I loved the Foundations. Like so many others in my cohort, I internalized its view that if I couldn’t formulate a problem in economic theory mathematically, I didn’t know what I was doing. I came to the position that mathematical analysis is not one of many ways of doing economic theory: It is the only way. Economic theory is mathematical analysis. Everything else is just pictures and talk. …
Lucas’s account, written long after the divisions had hardened, gives little other detail about how he ended up playing such a central role in the Samuelson program. I’ve heard a story second hand that when Lucas was a young faculty member at Carnegie-Mellon, it was a revelation to hear Diamond present his paper, “National Debt in a Neoclassical Growth Model,” which combined Samuelson’s overlapping generations model with Solow’s growth model. Whether or not account this story is accurate, in “Expectations and the Neutrality of Money”, the paper where Lucas lays out a both a tractable general equilibrium model with rational expectations and a theory of the Philips curve based on incomplete information, Samuelson’s overlapping generations model is the framework that Lucas uses.
Lucas appreciated Samuelson’s impatience with vague, verbal hand waving. He notes with approval Samuelson’s brash criticism of Marshall, who was, Lucas notes, “the god of Chicago economics”:
Here is a graduate student [Samuelson] in his 20s, reorganizing all of economics in four or five chapters, right before your eyes, and let Marshall, Hicks, Friedman, and everyone else get out of the way!
It is interesting to compare Samuelson on Marshall (albeit later in life) with Lucas and Sargent on prevailing Keynesian macro-theory:
Marshall’s crime is to pretend to handle imperfect competition with only the tools of perfect competition. … [A]ll of his prattle about the biological method in economics … cannot change this fact: any price taker who can sell more at the going price … and who has falling marginal cost will not be in equilibrium. Talk of birds, bees, giant trees, and declining entrepreneurial dynasties is all very well but why blink at such an elementary point?
Lucas and Sargent from a paper they presented in the summer of 1978:
On on recent failures of Keynesian models in general and of the traditional stable Phillips curve in particular:
That these predictions were wildly incorrect, and that the doctrine on which they were based is fundamentally flawed, are now simple matters of fact, involving no novelties in economic theory. The task which faces contemporary students of the business cycle is that of sorting through the wreckage, determining which features of that remarkable intellectual event called the Keynesian Revolution can be salvaged and put to good use, and which others must be discarded.
On the econometrics of the Keynesian multi-equation models:
This Keynesian solution to the problem of identifying a structural model has become increasingly suspect as a result of developments of both a theoretical and statistical nature. Many of these developments are due to efforts to researchers sympathetic to the Keynesian tradition, and many were well-advanced well before the spectacular failure of the Keynesian models in the 1970s.
Like Samuelson, Lucas and Sargent are blunt, but it is the bluntness that comes from seeing clearly that they are right, as they were.
But when they turned to the “new classical models” that they proposed as alternatives, they were more tentative:
There are, of course, legitimate issues involving the ability of equilibrium theories to fit the facts of the business cycle. Indeed, this is the reason for our insistence on the preliminary and tentative character of the particular models we now have.
In their discussion of the existing criticisms of these models, they express openness to the models that Stan Fischer and John Taylor were developing of fixed nominal wages specified in long-term contracts, noting only that that such models will not justify a return to the traditional Phillips curve because the contract terms will adjust if the inflation rate shifted permanently higher. (This of course is now the conventional wisdom.) They were solidly in favor of models of imperfect information that could no doubt have been extended to include recently proposed models of sticky information. They also supported the use of search models to capture the behavior of the labor market. So in 1978, when I was hearing the harsh critique from Solow, it is hard to find evidence of a closed minded intransigence among the people he was criticizing.
That mindset did develop soon thereafter and it seems to me that the frustration that would encourage it was already evident. Lucas and Sargent also wrote:
To a large extent, criticism of equilibrium models is simply a reaction to these implications for policy. So wide is (or was) the consensus that the task of macroeconomics is the discovery of the particular monetary and fiscal policies which can eliminate fluctuations by reacting to private sector instability that the assertion that this task either should not, or cannot be performed is regarded as frivolous independently of whatever reasoning and evidence may support it. Certainly one must have some sympathy with this reaction: an unfounded faith in the curability of a particular ill has served often enough as a stimulus to the finding of genuine cures. Yet to confuse a possibly functional faith in the existence of efficacious, reactive monetary and fiscal policies with scientific evidence that such policies are known is clearly dangerous, and to use such faith as a criterion for judging the extent to which particular theories “fit the facts” is worse still.
The Great Depression was a traumatic experience for both Solow and Stigler. They reacted very differently. For Solow, the human cost of mass unemployment was so high that to him, it seemed obvious that the government had to do something to bring unemployment down quickly in the wake of a recession. Stigler saw the many harmful and ineffective things that the government tried during the 1930s (including price floors and official support for cartels) and resolved that the guide to government policy must be “do no harm.”
Stigler wanted halt to progress in economic theory because he feared that it would lead to more theories like those of Keynes and Chamberlin (who provided the foundation for Dixit and Stiglitz). For him, there was apparently too much risk that such theories might lend political support for government policies that should not be tried. Under his division of labor with Milton Friedman, Friedman took on Keynes and Stigler took on Chamberlin. Marshall, they agreed, was safe. They turned Chicago in the last bastion of opposition to the Samuelson program and thereby prolonged for decades the confusion that Marshall had spawned. For Stigler, the logical implication of “do no harm” was “do nothing at all,” so what good could come from the Samuelson program anyway? Samuelson, like Lucas, must have found it infuriating to have his life’s work dismissed by someone who already knew all the policy answers.
From the other side, Solow and like-minded macroeconomists also tried to stop the development of the new type of theory that Lucas and Sargent were developing because they also feared that it might have harmful political repercussions. But in this case, the strategy backfired. Macroeconomists would almost surely have made faster progress toward consensus about the efficiency improving steps that the government could take to stabilize economic output if they had put more trust in science.
Faced with macro fluctuations, there are some policies, such as providing support for cartels, that the government should not try. There are other policies, such as countercyclical purchases of construction services, that it could safely try. After all, it is hard to see the harm that could come from buying more when the price is low. If economists stick to the well established methods of science, it shouldn’t be that hard to tell the difference.
There is no reason to stifle the development of new lines of theoretical or empirical inquiry out of fear that some terrible harm will ensue. If economists stick to science, the insights that are wrong will fade away. Only the ones that are right will having lasting influence. Sticking to science means both paying attention to scholars who are exploring new paths and refusing to pay attention to breakaway groups. No matter what they say, when they stop engaging with outsiders who disagree, they stop doing science.