By William K. Black
November 21, 2016 Bloomington, MN
Bloomberg has written an article about the origins of Paul Romer’s increasingly famous critique of modern macroeconomics.
His intention actually had been to write a paper that would celebrate advances in the understanding of what drives economic growth. But when he sat down to write it in the months before taking over as the World Bank’s chief economist, Romer quickly found his heart wasn’t in it. The world economy wasn’t growing much anyway; and the math that many colleagues were using to model it seemed unrealistic. He watched a documentary about the Church of Scientology, and was struck by how groupthink can operate.
So, Romer said in an interview at the Bank’s Washington headquarters, “I just thought, OK, I’m going to say what I think. I don’t know if I’m the right person, but no one else is going to say it. So I said it.”
The upshot was “The Trouble With Macroeconomics,” a scathing critique that landed among Romer’s peers like a grenade.
A bit of background makes the first paragraph more understandable. Romer’s specialty is developmental economics.
There are many economists who have said for years that modern macroeconomics is an abject failure. But all economists are not equal, and Romer is both an extremely distinguished economist and the World Bank’s chief economist. When he writes that macroeconomics is absurd his position gets vastly more attention from the field.
The Bloomberg article humorously summarizes Romer’s article.
“For more than three decades, macroeconomics has gone backwards,” the paper began. Romer closed out his argument, some 20 pages later, by accusing a cohort of economists of drifting away from science, more interested in preserving reputations than testing their theories against reality, “more committed to friends than facts.” In between, he offers a wicked parody of a modern macro argument: “Assume A, assume B, … blah blah blah … and so we have proven that P is true.”
The idea that consumers and businesses always make rational choices pervades mainstream economics. Romer thinks that’s not only wrong, but may lead to the misleading conclusion that government action can’t fix big problems.
There is no better place to be writing this than from (nearly) Minneapolis, for the University of Minnesota’s economics department is the most devoted coven worshipping the most extreme form of “rational expectations.” The most famous cultists have now relocated, but the U. Minnesota economics department remains fanatical in its devotion to rational expectations theory.
A belief that consumers and businesses always make rational choices does not “pervade mainstream economics.” Mainstream economics is increasingly influenced by reality, particularly in the form of behavioral economics. Behavioral economics, which has led to multiple Nobel awards, has many currents, but each of them agrees that consumers and business people typically do not make rational decisions even in simpler tasks, much less demonstrate the ability to predict the future required by rational expectations theory. Similarly, even the proponents of modern macroeconomics admit that its predictive ability – and predictive ability is supposed to be their holy grail of legitimacy – is beyond pathetic. What is true is that mainstream economics’ most egregious errors have come from assuming contrary to reality in a wide range of contexts that corporate officers, consumers, and investors make optimal decisions that maximize the firm or the household’s utility.
In any real scientific field modern macro would, decades ago, have been abandoned as an abject failure. Romer, therefore, is not storming some impregnable bulwark of economics. He is calling an obvious, abject failure an obvious, abject failure. Private sector finance participants typically believe the academic proponents of rational expectations theory are delusional. Romer is calling out elites in his profession who have ignored these failures and doubled and tripled-down on their failed dogmas for decades. This makes the Bloomberg article’s title deeply misleading: “The Rebel Economist Who Blew Up Macroeconomics.” Romer is not a rebel. He did not blow up academic, mainstream macroeconomics – the academic proponents of modern macroeconomics blew it up decades ago. Romer is mainstream, and he is sympathetic on personal and ideological grounds to the theoclasscial economist most famous for developing rational expectations theory. Romer has strongly libertarian views and did his doctoral work under Robert Lucas. Romer has long been appreciative of Lucas. All of this means that Romer’s denunciations were sure to hit home far harder with mainstream and theoclassical economists than anything a heterodox economist could write.
The same Bloomberg article made a key factual claim that is literally true but misleading.
What’s at stake far exceeds hurt feelings in the ivory tower. Central banks and other policy makers use the models that Romer says are flawed.
Central banks and private economic forecasters rarely use modern macro models, though they have begun to use New Keynesian models that are hybrids. They do not do use “freshwater” models because they are known to have terrible predictive ability and because alternative models not based on rational expectations have far superior predictive ability. The private financial sector typically does not rely on modern macro models, even the New Keynesian hybrids. Romer is not saying that the models are “flawed” – he is explaining that they are inherently failed models. Worse, he is saying that the designers of the models know they are failed and respond by gimmicking the models by littering them with myriad assumptions that have no empirical or theoretical basis and are designed to try to make the models produce less absurd results.
I explained that Romer was far from the first to call out modern academic macroeconomics as a failure but that he is a prominent mainstream economist. The Bloomberg article’s most interesting reveal was the response by the troika of economists must associated with rational expectations theory to Romer’s article decrying their dogmas.
Lucas and Prescott didn’t respond to requests for comments on Romer’s paper. Sargent did. He said he hadn’t read it, but suggested that Romer may be out of touch with the ways that rational-expectations economists have adapted their models to reflect how people and firms actually behave. Sargent said in an e-mail that Romer himself drew heavily on the school’s insights, back when he was “still doing scientific work in economics 25 or 30 years ago.”
What this paragraph reveals is the classic tactic of theoclassical economists – they simply ignore real criticism. Lucas, Prescott, and Sargent all care desperately about Romer’s criticism – but they all refuse to engage substantively with his critique. One has to love the arrogance of Sargent in “responding” – without reading – to Romer’s critique. Sargent cannot, of course, respond to a critique he has never read so he instead makes a crude attempt to insult Romer, asserting that Romer has not done any scientific work in three decades.
The rational expectations purists have been unable to come up with a response to their predictive failures and their false model of human behavior for thirty years. The Bloomberg article does not understand a subtle point about their non-defense defense, as shown in these key passages.
Allies of the three Nobelists have been more outspoken, and many of them point out that Romer — unlike Keynes in the 1930s — doesn’t offer a new framework to replace the one he says has failed.
“Burning down the edifice, and saying we’ll figure out what we’ll build on its foundations later, just does not seem like a constructive way to proceed,” said V. V. Chari, an economics professor at the University of Minnesota.
Romer’s heard that line often, and bristles at it: “I’m saying, ‘the car is broken.’ And everyone’s saying, ‘Romer’s a terrible guy, because he couldn’t fix the car’.”
What the rational expectations devotees are actually saying is their standard line, which is a radical departure from the scientific method. Their mantra is “it takes a model to beat a model.” That mantra violates the scientific method. Their models are designed to embody their rational expectations theory. Those models’ predictive ability is pathetic, which means that their theory and models are both falsified and should be rejected. The academic proponents of modern macro models, however, assert that their models are incapable of falsification by testing and predictive failure. This is not science, but theology.
V.V. Chari’s criticism of Romer is revealing. He complains that Romer does not want to “build on [rational expectation theory’s] foundations.” Why would Romer want to commit such a pointless act? Romer’s point is that rational expectations is a failed theory that needs to be rejected so macroeconomics can move on to useful endeavors.
A “foundation” in such a building metaphor is the deep, well-grounded stone or reinforced concrete beneath the visible building that is attached to solid bedrock. Rational expectations theory has no such empirical foundations. It was not based on testing that found that people behaved in accordance with the theory. Behavioral economics and finance, by contrast, is based on a growing empirical base – virtually all of which refutes the first three assumptions of the models. Similarly, the work of Akerlof (1970), Akerlof & Romer (1993), and the work of white-collar criminologists has falsified each of the first three assumptions of the models.
Further, the dynamic stochastic general equilibrium (DSGE) models routinely fail the predictive test and, as Romer details, fail despite the use of dozens of ways in which the models are “gamed” with arbitrary inputs and restrictions that have no theoretical or empirical basis. Chari is right to describe the modern macro model as an “edifice.” I would add that it is a baroque edifice top heavy with ornamental features designed to hide its lack of a foundation. Modern macro collapsed as soon as its devotees tried to build without an empirical foundation.
The rational expectation devotees respond that predictive failures – no matter how extreme or frequent – cannot falsify their models or their theories. The proponents claim that only a better model, with superior predictive ability can beat their model. That might sound acceptable to some, but there is a critical unstated twist. The many rival models actually used by the private sector and central banks that produce far superior predictive ability can never be treated as “better models” to these devotees because the models with far superior predictive powers reject rational expectations theory, rational decision-making, and the “budget constraint.” To the devotees, only DSGE models that accept this trio of market fictions are eligible to be acceptable “models.” Dr. Kocherlakota states that acceptable models “share five key features.” These five characteristics define DSGE models.
- They specify budget constraints for households, technologies for firms, and resource constraints for the overall economy.
- They specify household preferences and firm objectives.
- They assume forward-looking behavior for firms and households.
- They include the shocks that firms and households face.
- They are models of the entire macroeconomy.
Kocherlakota’s summary description is appropriately terse. He later explains the dogmatic gloss that devotees place on each of these five points. The “budget constraint,” for example, means that nations with sovereign currencies such as the U.S. cannot run deficits, even to fight severe recessions or depressions. Why? Because theoclassical economists are enormous believers in austerity. As Kocherlakota archly phrased the matter, “freshwater” DSGE models were so attractive to theoclassical macro types because their model perfectly tracked their ideology.
[A]lmost coincidentally—in these models, all government interventions (including all forms of stabilization policy) are undesirable.
Yes, “almost coincidentally.”
Specifying household preferences and firm objectives is equally erroneous, as Akerlof and Romer’s 1993 article on “Looting” demonstrated. “Firms” do not have “objectives.” Employees have “objectives,” and the controlling officers’ “objectives” are the most powerful drivers of employee behavior. As Akerlof and Romer (and every modern crisis) demonstrated, this frequently leads to firm practices that harm the firm, the consumer, and the shareholders. Such behavior, however, is impossible under the second assumption, so any model (such as control fraud or “looting”) that violates the assumption is not eligible to be a rival model because it these superior models do not produce “general equilibrium.” The “GE” in a “DSGE” model is general equilibrium, so rival models from economics and criminology that note that the economy is not a self-correcting apparatus that produces general equilibrium are ruled out as superior models even though they are superior in that they have an empirical and theoretical basis and demonstrate far superior predictive results.
Kocherlakota unintentionally highlighted modern macros’ inability to incorporate even massive frauds driving national scandals and banking crises, despite the efforts of Akerlof (1970) (a market for “lemons”) and Akerlof and Romer 1993: (“looting”) in this passage.
In the macro models of the 1980s, all mutually beneficial trades occur without delay. This assumption of frictionless exchange made solving these models easy. However, it also made the models less compelling.
He then goes on to a delighted description of macro economists now sometimes building in (arbitrary) lags (“frictions”) in the time required to accomplish “all mutually beneficial trades.” But what of the three great fraud epidemics that produced the U.S. financial crisis and the Great Recession? Sorry, that’s not allowed into the “friction” canon. The market model is still one of perfection (albeit slightly delayed). It does not matter how many massive financial scandals occur in which the largest UK banks and Wells Fargo deliberately abuse their customers by encouraging them to engage in transactions that will harm them and make the bankers rich. It doesn’t not matter that over ten million Americans were induced by bankers and their agents to pay excessive interest rates in return for yield spread premiums (YSP) to the bankers and brokers. None of these things are allowed to happen in these models. Your better model, which includes such frauds and abuses, is not allowed precisely because it (a) is better and (b) falsifies the theoclassical ideology underlying “rational expectations” theory.
The assumption of “forward looking behavior” produces “expectations,” which are assumed to be accurate and rational. Theoclassical proponents claim that we all have the ability to predict vast aspects of the financial future. While we are not perfect, we are optimal in our forecasts given the state of knowledge. If your rival model lacks rational expectations, it isn’t a real model. Romer rejects the rational expectations myth, so he is incapable of presenting a superior model to the devotees of rational expectations.
If macroeconomics, outside the cult of modern macro, were a car, it would not be “broken.” It would be episodically broken when the rational expectations devotees got hold of monetary or fiscal policy. The rational expectations model fails the most fundamental test of a financial model – people trying to make money by anticipating the macroeconomics consequences of changes in monetary and fiscal policy overwhelmingly do not use their models because they are known to have pathetic predictive ability. The alternative models that embrace Keynesian analysis and are not dependent on the fiction of rational expectations function pretty well. The real world macro car, when driven by real world drivers, works OK. Essentially, the rational expectations devotees say that we can never drive the macro “car” because the public will defeat any effort to drive the economy in any direction. Instead, the economy will lurch about n response to random technological “shocks” that cannot be predicted because they occur without any relationship to any public policy choices.
Romer takes particular delight in shredding the pretension to “science” in the model’s abuse of shocks. Again, however, the Bloomberg article seriously misleads in making it appear that his critique of shocks is novel. Then Minneapolis Fed Chair Dr. Kocherlakota (formerly chair of the U. Minnesota economics department, where he was a “rational expectations” devotee) forcefully owned up to the egregious predictive failures of the models. He acknowledged that “macro models are driven by patently unrealistic shocks.”
It is unfortunate that Bloomberg article about Romer’s article is weak. It is useful, however, because its journalistic inquiry allows us to know how deep in their bunker Sargent, Lucas, and Prescott remain. They still refuse to engage substantively with Romer’s critique of not only their failures but their intellectual dishonesty and cowardice. It is astonishing that multiple economists were awarded Nobel prizes for creating the increasingly baroque failure of modern macro. In any other field it would be a scandal that would shake the discipline to the core and cause it to reexamine how it conducted research and trained faculty. In economics, however, a huge proportion of Nobel awards have gone to theoclassical economists whose predictions have been routinely falsified and whose policy recommendations have proven disastrous. Theoclassical economists, with only a handful of exceptions, express no concern about these failures.