Wednesday, February 12, 2014

Peter Dorman, bluntly, on the crappiness of today's microfoundations



This is a bit old, but I just happened onto this interesting post at Econospeak contributed by Peter Dorman. It pulls no punches, and suggests that the stamp above might be appropriately applied to most of the highly "insightful" and "sophisticated" papers of modern micro-founded macroeconomics:

... Microfoundations for macroeconomics are fine in principle—not indispensable, but useful.  The problem is that what passes for microfoundations in the universe of orthodox macro is crap.

There.  I said it.  I used the “c” word.  But not the “s” word.

It’s nothing more than robotic imitation of teaching exercises to improve math skills, without any consideration for such mundane matters as empirical verisimilitude.  I will mention three crushing faults, each sufficient by itself to blow a wide hole in a supposedly useful model.

1. Utility theory.  Andrew Gelman calls this “folk psychology”; that may be generous.  It is rife with anomalies (see “behavioral economics”), and, most important, it is oblivious to the last several decades of work in psychology, evolutionary biology, neuropsychology, organization theory—all the disciplines where people study behavior in a scientific way.

2. Mono-equilibrium assumptions.  There are no interaction effects to generate multiple equilibria in the microfoundations macro theorists use.  Every individual, firm and product is an isolated atom, floating uninterrupted through space until it bumps into another such atom in the marketplace.  Social psychology, ecology, nonconvex production and consumption spaces?  Forget about it.  In evolutionary biology, by contrast, fitness surfaces are assumed nonconvex from the get-go; it’s central to the discipline.  Failure to recognize the interactive character of economic life leads economists to ask fundamentally wrong questions, like “what’s the equilibrium?” and “what’s the optimum?”  If this isn’t obvious to you already, you can get a longer version of the argument here.  (Note for those who are wondering: no, nonconvexity stemming from interaction effects has nothing to do with market failure.  The existence of externalities is neither necessary  nor sufficient for these effects.  See for yourself.)

3. Path dependence.  Microfoundations means general equilibrium theory, but the flavor it uses is from the mid-1950s.  The Sonnenschein-Debreu-Mantel demonstration (update to the 1970s)  that initial conditions and out-of-equilibrium trades alter the equilibrium itself (they assume away problem #2) has turned GET upside down.

Notice that I haven’t mentioned the standard heterodox criticisms of representative agents and ergodicity.  You can add those if you want.

Now here’s the clincher.  As Krugman points out, faced with the choice between addressing the evidence or maintaining consistency with their microfounded models, macroeconomists as a herd have gone for the second.  This is because they believe that the micro theory they use is really, really, really true, and that no model that cannot be yoked to it can be considered scientific.  And if we actually knew with certainty that mid-50s general equilibrium theory with optimizing agents and no interactions outside the market was the only acceptable framework for thinking coherently about economics, they’d be right.  But they’re not.

Like I said, their microfoundations are crap.

Damned with faint praise


I posted yesterday on the new book by Kartik Athreya, Big ideas in Macroeconomics: a non-technical view. There's a blurb on the back by Herb Gintis, who I think is a very smart guy and worth listening to. I noticed he also has written a review on Amazon, which is titled: At Last! A Serious Presentation and Defense of Modern Macroeconomic Theory. That sounds really positive, doesn't it? I thought, wow, Gintis thinks the theory is in great shape? I'm shocked, and I have to think hard about what he says. The review starts out saying nice things about the book. But then look at the parts I've highlighted below:


It is easy to find excellent, accurate, accessible, and entertaining books that present the current theories and ideas in many fields, including law, physics, and biology. It is virtually impossible to find such books in economics. Sadly, when someone writes a book about economics, it almost always is an attempt to convince the reader that some politically motivated partial truth is the whole truth.

There are some books that present basic economic theory in an unbiased manner, and I review the ones I have found in an entry on my web site (http://people.umass.edu/gintis). Click on You Must Read This! and look for "Books on Economics for Serious Beginners: Very Introductory Readings."

But for the sort of advanced macroeconomic that guided policy makers and central bankers leading up to the financial meltdown of 2008, there has been virtually no serious accessible exposition without a political bone to pick. I love this book because it treats the reader as intelligent and discerning, presents the theory ably and in great detail, but avoids the mathematical detail that makes the material quite impenetrable to all but the expert who spends the bulk of his time devoted to the subject.

The reader who even cursorily inspects this book might consider me a biased observer because I contributed a blurb to the book jacket and the author graciously thanks me for my support in the introductory pages. The fact is that I consider this an exemplary exposition of modern macroeconomics, and I think his defense of the theory is as good as one can find anywhere, the theory is in fact so weak that nothing can save it. People continue doing it not because it is good theory, but because it is the only game in town. I believe a complete revolution in macroeconomic theory is in the process of being born, although it will take some years to take over as the mainstream theory.

Change in macroeconomic theory can be extremely rapid. Keynesianism displaced classical macroeconomics in just a few years after the end of WWII, and the reigning "rational expectations" macro displaced Keynesianism in just a few breathtaking years. This is a credit to economics as a discipline--given new evidence and given new economic conditions, young Ph.D. economists are quite willing to throw over the past, and in the best graduate schools, hiring of new faculty is based on how productive they are as researchers, not whether or not they agree with the reigning orthodoxy.

Macroeconomics has always been deeply political. After WWII in Europe and the US, the growth in organized labor led to cost-push inflation (higher wages --> higher prices) and unemployment caused by wages above the market-clearing level. Keynesianism blamed the unemployment on the market system itself and suggested deficit spending as the way to restore full employment and accommodating higher prices by increasing the money supply. Of course, this is a stupid theory because it leads to chronic deficit spending and chronic inflation. With the decline in the power of organized labor (caused mainly by increased international competition eliminating domestic monopolies in internationally traded goods such as steel, mining, and automobiles), a resurgent right-wing macroeconomics, called rational expectations theory, displaced Keynesian macro by recognizing the fatal flaws in its reasoning, which contradicted economic rationality. Keynesianism remains today in an attenuated form that recognizes coordination failures and price inertia, but is mainly a liberal profession of faith.

The only virtue of rational expectations macro (RA macro), which Athreya explains so nicely in this book, is that it killed Keynesian macro. The theory itself is nothing but smoke and mirrors. It purports to be solidly based on widely-accepted microeconomic principles that accurately describe the market economy (the "rational" in rational expectations), but this is simply false. The market economy is in fact a complex dynamic system whose behavior can be simulated, much as the weather is simulated by supercomputers, but cannot be captured in a few recursive equations, as the RA macro supporters claim. Instead of a large number of economic actors, RA macro assumes there is one "representative agent," and instead of large numbers of firms and industries, RA macro assumes there is one "production function." The only source of volatility in such a world is technology shocks and the vagaries of "expectations" (which of course cannot be measured, because they don't really exist). All we end up with is smoke and mirrors, plus lots of abstruse equations.

Among the more bizarre modeling choices of RA macro is to assume that all markets clear instantaneously. This of course assumes away the coordination failures that really underlie macroeconomic fluctuations. Especially exotic is the assumption that there is always full employment! "Unemployed" workers are simply people who temporarily prefer not to work (they prefer "leisure," in the parlance of macro theory). In fact what happens in a recession is that millions of jobs disappear. The displaced workers prefer their old or equivalent jobs at their accustomed wages, but these are gone. Of course, many could find work at a lower wage, but that is not always the rational thing to do because the worker may get locked into the lower wage occupational level.

This absurd sort of macro modeling would be okay if the resulting models predicted well, but they do not. RA macro long ago gave up econometric testing their equations in favor of "calibrating" them, which means just get the best fit you can, however poor. An poor they are.

Of course, prediction is not everything. Engineers cannot predict when a car will go over a bump, but they can model the effects of such a shock on the car and proposed mechanisms (shock absorbers) that allow the system (the car) to survive the shock with minimal damage. The same is true, to a much more limited extent in macroeconomics, where "automatic stabilizers" can lessen the effects of shocks to the economy.

The real problem with RA macro (and similarly of Keynesian macro) is the theory cannot deal with finance at all, and financial markets lie at the heart of contemporary economic instability. The Walrasian micro model on which RA macro is built simply has no place for money or finance. Graduate students in economics do not even study finance---that is left to the business schools.

Of course, all of that is now changing. Economists around the world are revamping their theories and developing the empirical data on finance so that a more useful theory is likely to be forthcoming. It is a very exciting time for macroeconomics. Athreya's defense of the current theory is quite brilliant and I urge the reader to learn from him. However, we all know the story of the silk purse and the sow's ear.

In short, nice attempt on an impossible task: giving a coherent defense of modern rational expectations macroeconomics.

Tuesday, February 11, 2014

Economists with hurt feelings



Kartik Athreya is an economist at the Federal Reserve Bank of Richmond. He is a true believer in the modern methods of macroeconomics, and has written a new book entitled Big ideas in Macroeconomics: a non-technical view. I've just had a partial read of the first 50 pages or so on google books.

Athreya is really irritated at all the criticism that macroeconomists have had to endure since the onset of the financial crisis. He famously expressed his irritation a few year ago, demanding that bloggers and non-economists in general shut up and leave discussion of economics to himself and other experts in the field. It seems that the purpose of the new book is to correct all the confusion and explain to everyone why modern macro is so wonderful and right and beyond criticism. I don't think it is going to succeed. Anyone who reads his section on the Walrasian Clearinghouse -- an imaginary mechanism to explain how an economy reaches Walrasian equilibrium -- will pretty quickly come to the conclusion that Athreya's beloved picture of how an economy works is mostly mathematical fantasy.

But actually, I think the book may in this way provide a great service. Anyone who reads it carefully will be able to see for themselves just how fragile and artificial the modern rational expectations approach to macroeconomics really is. I'm going to order the book because, from what I have seen, there will be much to learn, although possibly not about how a real economy works. Readers will experience a close encounter with the reality-defying attitude and arrogance of mainstream macroeconomics. Reading from one who holds it in such high esteem, the reader can also trust that he or she is not being deceived by some ignorant blogger who is setting up straw man arguments (a common response from wounded macroeconomists).

David Glasner has read the book and offers an informative review. One section deserves highlighting, regarding Athreya's tendency to dismiss whole realms of real economic phenomena as irrelevant simply because they don't fit into his preferred modelling methodology:

As Athreya acknowledges in chapter 5, an important issue separating certain older macroeconomic traditions (both Keynesian and Austrian among others) is the idea that macroeconomic dysfunction is a manifestation of coordination failure. It is a property – a remarkable property – of Walrasian general equilibrium that it achieves perfect (i.e., Pareto-optimal) coordination of disparate, self-interested, competitive individual agents, fully reconciling their plans in a way that might have been achieved by an omniscient and benevolent central planner. Walrasian general equilibrium fully solves the coordination problem. Insofar as important results of modern macroeconomics depend on the assumption that a real-life economy can be realistically characterized as a Walrasian equilibrium, modern macroeconomics is assuming that coordination failures are irrelevant to macroeconomics. It is only after coordination failures have been excluded from the purview of macroeconomics that it became legitimate (for the sake of mathematical tractability) to deploy representative-agent models in macroeconomics, a coordination failure being tantamount, in the context of a representative agent model, to a form of irrationality on the part of the representative agent. Athreya characterizes choices about the level of aggregation as a trade-off between realism and tractability, but it seems to me that, rather than making a trade-off between realism and tractability, modern macroeconomics has simply made an a priori decision that coordination problems are not a relevant macroeconomic concern.

A similar argument applies to Athreya’s defense of rational expectations and the use of equilibrium in modern macroeconomic models. I would not deny that there are good reasons to adopt rational expectations and full equilibrium in some modeling situations, depending on the problem that theorist is trying to address. The question is whether it can be appropriate to deviate from the assumption of a full rational-expectations equilibrium for the purposes of modeling fluctuations over the course of a business cycle, especially a deep cyclical downturn. In particular, the idea of a Hicksian temporary equilibrium in which agents hold divergent expectations about future prices, but markets clear period by period given those divergent expectations, seems to offer (as in, e.g., Thompson’s “Reformulation of Macroeconomic Theory“) more realism and richer empirical content than modern macromodels of rational expectations.

Athreya offers the following explanation and defense of rational expectations:
[Rational expectations] purports to explain the expectations people actually have about the relevant items in their own futures. It does so by asking that their expectations lead to economy-wide outcomes that do not contradict their views. By imposing the requirement that expectations not be systematically contradicted by outcomes, economists keep an unobservable object from becoming a source of “free parameters” through which we can cheaply claim to have “explained” some phenomenon. In other words, in rational-expectations models, expectations are part of what is solved for, and so they are not left to the discretion of the modeler to impose willy-nilly. In so doing, the assumption of rational expectations protects the public from economists.

This defense of rational expectations plainly belies the methodological arrogance of modern macroeconomics. I am all in favor of solving a model for equilibrium expectations, but solving for equilibrium expectations is certainly not the same as insisting that the only interesting or relevant result of a model is the one generated by the assumption of full equilibrium under rational expectations. (Again see Thompson’s “Reformulation of Macroeconomic Theory” as well as the classic paper by Foley and Sidrauski, and this post by Rajiv Sethi on his blog.) It may be relevant and useful to look at a model and examine its properties in a state in which agents hold inconsistent expectations about future prices; the temporary equilibrium existing at a point in time does not correspond to a steady state. Why is such an equilibrium uninteresting and uninformative about what happens in a business cycle? But evidently modern macroeconomists such as Athreya consider it their duty to ban such models from polite discourse — certainly from the leading economics journals — lest the public be tainted by economists who might otherwise dare to abuse their models by making illicit assumptions about expectations formation and equilibrium concepts.

See also Noah Smith's comments on the book.

Wednesday, December 18, 2013

Microfoundations: PLEASE tell a better story!

A scuffle has broken out among some economists over the touchy topic of microfoundations, i.e. the usual formal requirement that macroeconomic models, to be considered legitimate, must be based on things like households and firms optimizing their intertemporal utility, having rational expectations, and so on. Apparently, things got kicked off by economist Tony Yates, who offered a spirited defense of microfoundations of this kind; he's really irritated that people keep criticizing them and worries that, My God!, this might cause DSGE macroeconomists to lose some credibility! In response, Simon Wren-Lewis came back with an equally spirited argument about why modelling should be more flexible and "eclectic." Noah Smith's summary of the whole disagreement puts everything in context.

Noah makes the most important point right at the end. Simply put, no one (I think) is against the authentic spirit of microfoundations, i.e. the idea that macroeconomic models ought to be based on plausible stories of how the real actors in an economy behave. If you get that right, then obviously your model might stand a chance of getting larger aggregate things right too. The problem we have today is that the microfoundations you find in DSGE models aren't like this in the least. So macromodels are actually based on things we know to be wrong. It's very strange indeed. As Noah puts it:

Yates says I just want to get rid of all the microfoundations. But that is precisely, exactly, 180 degrees wrong! I think microfoundations are a great idea! I think they're the dog's bollocks! I think that macro time-series data is so uninformative that microfoundations are our only hope for really figuring out the macroeconomy. I think Robert Lucas was 100% on the right track when he called for us to use microfounded models.

But that's precisely why I want us to get the microfoundations right. Many of microfoundations we use now (not all, but many) are just wrong. Obviously, clearly wrong. Lots of microeconomists I talk to agree with me about that. And lately I've been talking to some pretty prominent macroeconomists who agree as well. 

So I applaud the macroeconomists who are working on trying to develop models with better microfoundations (here is a good example). Hopefully the humble stuff I'm doing in finance can lead to some better microfoundations too. And in the meantime I'm also happy to sit here and toss bombs at people who think the microfoundations we have are good enough!

I couldn't agree more.

In fact, before coming across this debate this morning, I had intended to make a short post linking to the very informative lecture (below, courtesy of Mark Thoma) by macroeconomist George Evans. Lord knows I spend enough time criticizing economists -- and this recent post discussed the limitations of the learning literature, in which Evans has been a key player -- so I want to make clear that I do admire the things he does.




He tells an interesting story about an economic model (a standard New Keynesian model) that -- when the agents in the model learn in a particular constrained way -- has two different equilibria. One is locally stable and the economy has inflation right around a targeted value. Start out with inflation and consumption and expectations close to that equilibrium and you'll move toward that point over time. The second equilibrium is, however, unstable. If you start out sufficiently far away from the stable equilibrium, you won't go there at all, but will wander down into a deflationary zone (and what happens then I don't know).

This model for aggregate behaviour is based on some fairly simple low-dimensional equations for how current consumption and inflation feed, via expectations, into future values and a trajectory for the economy. I don't know how plausible these equations are. I'm guessing that someone can make a good argument about why they should have the form they do (or a similar form). That story would involve references to how things happening now in the economy would influence peoples' behaviour and their expectations, and then how these would cause certain kinds of changes. To really believe this you'd want to see some evidence that this story is correct, i.e. that people, firms, etc., really do tend to behave like this.

The point I want to make is that -- for someone like myself who has not been socialized to accept the necessity of what currently counts as "microfoundations" -- nothing about the story becomes more plausible when I wade into the equations of the New Keynesian model and see how households and firms independently optimize their intertemporal utilities subject to certain budget constraints. If anything, seeing all this dubious stuff makes me less likely to believe in the plausibility of the low dimensional equations for aggregate variables. And this is precisely the problem with microfoundations of this kind. They don't give a good argument for why the aggregate variables should satisfy these equations. They give a very bad. An unconvincing argument. At least for me.



Tuesday, December 17, 2013

Can you spell FRAUD?

A great article in the New York Review of Books by US District Court Judge Jed Rakoff, asking pointed questions about why no executives of major financial institutions have been convicted of fraud for their actions in the lead up to the financial crisis. As he argues, the Commission set up to explore the crisis found plenty of evidence of widespread fraud; the Justice Department has simply failed to act. Why?

...the stated opinion of those government entities asked to examine the financial crisis overall is not that no fraud was committed. Quite the contrary. For example, the Financial Crisis Inquiry Commission, in its final report, uses variants of the word “fraud” no fewer than 157 times in describing what led to the crisis, concluding that there was a “systemic breakdown,” not just in accountability, but also in ethical behavior.

As the commission found, the signs of fraud were everywhere to be seen, with the number of reports of suspected mortgage fraud rising twenty-fold between 1996 and 2005 and then doubling again in the next four years. As early as 2004, FBI Assistant Director Chris Swecker was publicly warning of the “pervasive problem” of mortgage fraud, driven by the voracious demand for mortgage-backed securities. Similar warnings, many from within the financial community, were disregarded, not because they were viewed as inaccurate, but because, as one high-level banker put it, “A decision was made that ‘We’re going to have to hold our nose and start buying the stated product if we want to stay in business.’”

Without giving further examples, the point is that, in the aftermath of the financial crisis, the prevailing view of many government officials (as well as others) was that the crisis was in material respects the product of intentional fraud. In a nutshell, the fraud, they argued, was a simple one. Subprime mortgages, i.e., mortgages of dubious creditworthiness, increasingly provided the chief collateral for highly leveraged securities that were marketed as AAA, i.e., securities of very low risk. How could this transformation of a sow’s ear into a silk purse be accomplished unless someone dissembled along the way?

While officials of the Department of Justice have been more circumspect in describing the roots of the financial crisis than have the various commissions of inquiry and other government agencies, I have seen nothing to indicate their disagreement with the widespread conclusion that fraud at every level permeated the bubble in mortgage-backed securities. Rather, their position has been to excuse their failure to prosecute high-level individuals for fraud in connection with the financial crisis on one or more of three grounds:

Rakoff goes on the examine these grounds, and finds none of them convincing. So, then, why no prosecutions? He discounts the revolving door theory -- the prosecutors have avoided action because of former links to financial firms, or hope of future employment links -- because, in his experience, prosecutors are well motivated to get convictions. He suggests that there are a host of reasons: prosecutors simply had other priorities in the years after 9/11/2001; in many cases government regulators acquiesced early on to changing practices, where in increasingly lax demands on mortgage documentation became the norm. And, finally and most importantly, he points to changes in prosecuting practices over the past few decades:

The final factor I would mention is both the most subtle and the most systemic of the three, and arguably the most important. It is the shift that has occurred, over the past thirty years or more, from focusing on prosecuting high-level individuals to focusing on prosecuting companies and other institutions. It is true that prosecutors have brought criminal charges against companies for well over a hundred years, but until relatively recently, such prosecutions were the exception, and prosecutions of companies without simultaneous prosecutions of their managerial agents were even rarer.

The reasons were obvious. Companies do not commit crimes; only their agents do. And while a company might get the benefit of some such crimes, prosecuting the company would inevitably punish, directly or indirectly, the many employees and shareholders who were totally innocent. Moreover, under the law of most US jurisdictions, a company cannot be criminally liable unless at least one managerial agent has committed the crime in question; so why not prosecute the agent who actually committed the crime?

In recent decades, however, prosecutors have been increasingly attracted to prosecuting companies, often even without indicting a single person. This shift has often been rationalized as part of an attempt to transform “corporate cultures,” so as to prevent future such crimes; and as a result, government policy has taken the form of “deferred prosecution agreements” or even “nonprosecution agreements,” in which the company, under threat of criminal prosecution, agrees to take various prophylactic measures to prevent future wrongdoing. Such agreements have become, in the words of Lanny Breuer, the former head of the Department of Justice’s Criminal Division, “a mainstay of white-collar criminal law enforcement,” with the department entering into 233 such agreements over the last decade. But in practice, I suggest, this approach has led to some lax and dubious behavior on the part of prosecutors, with deleterious results.

If you are a prosecutor attempting to discover the individuals responsible for an apparent financial fraud, you go about your business in much the same way you go after mobsters or drug kingpins: you start at the bottom and, over many months or years, slowly work your way up. Specifically, you start by “flipping” some lower- or mid-level participant in the fraud who you can show was directly responsible for making one or more false material misrepresentations but who is willing to cooperate, and maybe even “wear a wire”—i.e., secretly record his colleagues—in order to reduce his sentence. With his help, and aided by the substantial prison penalties now available in white-collar cases, you go up the ladder.

But if your priority is prosecuting the company, a different scenario takes place. Early in the investigation, you invite in counsel to the company and explain to him or her why you suspect fraud. He or she responds by assuring you that the company wants to cooperate and do the right thing, and to that end the company has hired a former assistant US attorney, now a partner at a respected law firm, to do an internal investigation. The company’s counsel asks you to defer your investigation until the company’s own internal investigation is completed, on the condition that the company will share its results with you. In order to save time and resources, you agree.

Six months later the company’s counsel returns, with a detailed report showing that mistakes were made but that the company is now intent on correcting them. You and the company then agree that the company will enter into a deferred prosecution agreement that couples some immediate fines with the imposition of expensive but internal prophylactic measures. For all practical purposes the case is now over. You are happy because you believe that you have helped prevent future crimes; the company is happy because it has avoided a devastating indictment; and perhaps the happiest of all are the executives, or former executives, who actually committed the underlying misconduct, for they are left untouched.

I suggest that this is not the best way to proceed. Although it is supposedly justified because it prevents future crimes, I suggest that the future deterrent value of successfully prosecuting individuals far outweighs the prophylactic benefits of imposing internal compliance measures that are often little more than window-dressing. Just going after the company is also both technically and morally suspect. It is technically suspect because, under the law, you should not indict or threaten to indict a company unless you can prove beyond a reasonable doubt that some managerial agent of the company committed the alleged crime; and if you can prove that, why not indict the manager? And from a moral standpoint, punishing a company and its many innocent employees and shareholders for the crimes committed by some unprosecuted individuals seems contrary to elementary notions of moral responsibility.

These criticisms take on special relevance, however, in the instance of investigations growing out of the financial crisis, because, as noted, the Department of Justice’s position, until at least recently, is that going after the suspect institutions poses too great a risk to the nation’s economic recovery. So you don’t go after the companies, at least not criminally, because they are too big to jail; and you don’t go after the individuals, because that would involve the kind of years-long investigations that you no longer have the experience or the resources to pursue.

In conclusion, I want to stress again that I do not claim that the financial crisis that is still causing so many of us so much pain and despondency was the product, in whole or in part, of fraudulent misconduct. But if it was—as various governmental authorities have asserted it was—then the failure of the government to bring to justice those responsible for such colossal fraud bespeaks weaknesses in our prosecutorial system that need to be addressed.

Friday, December 13, 2013

Macroeconomics: The illusion of the "learning literature"

For many economists, prevailing theories of macroeconomics based on the idea of rational expectations (RE) are things of elegance and beauty, somewhat akin to the pretty face you see below. This is especially their view in the light of two decades of research looking at learning as a foundation for macroeconomics -- something economists refer to as the "learning literature." You'll hear it said that these studies have shown that most of the RE conclusions also follow from much more plausible assumptions about how people form expectations and adjust them over time by learning and adapting. Sounds really impressive.


As with this apparently pretty face, however, things aren't actually so beautiful and elegant if you take the time to read some of the papers in the learning literature and see what has been done. For example, take this nice review article by Evans and Honkapohja from a few years ago. It is a nice article and reports on interesting research. If you study it, however, you'll find that the "learning" studied in this line of work is not at all what most of us would think of as learning as we know it in the real world. I've written about this before and might as well just quote something I said there:

What the paper does is explore what happens in some of the common rational expectations models if you suppose that agents' expectations aren't formed rationally but rather on the basis of some learning algorithm. The paper shows that learning algorithms of a certain kind lead to the same equilibrium outcome as the rational expectations viewpoint. This IS interesting and seems very impressive. However, I'm not sure it's as interesting as it seems at first.

The reason is that the learning algorithm is indeed of a rather special kind. Most of the models studied in the paper, if I understand correctly, suppose that agents in the market already know the right mathematical form they should use to form expectations about prices in the future. All they lack is knowledge of the values of some parameters in the equation. This is a little like assuming that people who start out trying to learn the equations for, say, electricity and magnetism, already know the right form of Maxwell's equations, with all the right space and time derivatives, though they are ignorant of the correct coefficients. The paper shows that, given this assumption in which the form of the expectations equation is already known, agents soon evolve to the correct rational expectations solution. In this sense, rational expectations emerges from adaptive behaviour.

I don't find this very convincing as it makes the problem far too easy. More plausible, it seems to me, would be to assume that people start out with not much knowledge at all of how future prices will most likely be linked by inflation to current prices, make guesses with all kinds of crazy ideas, and learn by trial and error. Given the difficulty of this problem, and the lack even among economists themselves of great predictive success, this would seem more reasonable. However, it is also likely to lead to far more complexity in the economy itself, because a broader class of expectations will lead to a broader class of dynamics for future prices. In this sense, the models in this paper assume away any kind of complexity from a diversity of views.

To be fair to the authors of the paper, they do spell out their assumptions clearly. They state in fact that they assume that people in their economy form views on likely future prices in the same way modern econometricians do (i.e. using the very same mathematical models). So the gist seems to be that in a world in which all people think like economists and use the equations of modern econometrics to form their expectations, then, even if they start out with some of the coefficients "mis-specified," their ability to learn to use the right coefficients can drive the economy to a rational expectations equilibrium. Does this tell us much?

My view is that NO, it doesn't tell us much. It's as if the point of the learning literature hasn't really been to explore what might happen in macroeconomics if people form expectations in psychologically realistic ways, but to see how far one can go in relaxing the assumptions of RE while STILL getting the same conclusions. Of course there's nothing wrong with that as an intellectual exercise, but it hardly a full bore effort to understand economic reality. It's more an exercise in theory preservation, examining the kinds of rhetoric RE theorists might be able to use to defend the continued use of their favorite ideas. "Yes, if we use the word 'learning' in a very special way, we can say that RE theories are fully consistent with human learning!"

Anyway, I've revisited this idea in my most recent Bloomberg column, which should appear this weekend. I find it quite irritating that lots of economists go on repeating this idea that the learning literature shows that it's OK to use RE when in fact it does nothing of the sort. You often find this kind of argument when economists smack down their critics, implying that those critics "just don't know the literature." In an interview from a few years ago, for example, Thomas Sargent suggested that criticism of excessive reliance on rationality in macroeconomics reflects "... either woeful ignorance or intentional disregard for what much of modern macroeconomics is about and what it has accomplished."

On a more thoughtful note, Oxford economist Simon Wren-Lewis also recently defended the RE assumption (responding to a criticism by Lars Syll), mainly arguing that he hasn't seen any useful alternatives. He also refers to this allegedly deep learning literature as a source of wisdom, although he does acknowledge that it's aim has been fairly limited:

...If I really wanted to focus in detail on how expectations were formed and adjusted, I would look to the large mainstream literature on learning, to which Professor Syll does not refer. (Key figures in developing this literature included Tom Sargent, Albert Marcet, George Evans and Seppo Honkapohja: here is a nice interview involving three of them.) Macroeconomic ideas derived from rational expectations models should always be re-examined within realistic learning environments, as in this paper by Benhabib, Evans and Honkapohja for example. No doubt that literature may benefit from additional insights that behavioural economics and others can bring. However it is worth noting that a key organising device for much of the learning literature is the extent to which learning converges towards rational expectations.

However most of the time macroeconomists want to focus on something else, and so we need a simpler framework. In practice that seems to me to involve a binary choice. Either we assume that agents are very naive, and adopt something very simple like adaptive expectations (inflation tomorrow will be based on current and past inflation), or we assume rational expectations. My suspicion is that heterodox economists, when they do practical macroeconomics, adopt the assumption that expectations are naive, if they exist at all (e.g. here). So I want to explain why, most of the time, this is the wrong choice. My argument here is similar but complementary to a recent piece by Mark Thoma on rational expectations.

As I said above, Wren-Lewis is one of the more thoughtful defenders of RE, but he too here lapses into the use of "learning" without any qualification.

More importantly, however, I'm not sure why Wren-Lewis thinks there is only a binary choice. After all, there is a huge range of alternatives between simple naive expectations and RE and this is precisely the range inhabited by real people. So why not look there? Why not actually look to the psychology literature on how people learn and use some ideas from that? Or, why not do some experiments and see how people form expectations in plausible economic environments, then build theories in that way?

This kind of work can and is being done. My Bloomberg column touches briefly on this really fascinating paper from earlier this year by economist Tiziana Assenza and colleagues.  What they did, briefly, is to run experiments with volunteers who had to make predictions of inflation (and sometimes also the output gap) in a laboratory economy. The economy was simple: the volunteers' expectations fed into determining the economic future outcomes for inflation etc. by a simple low dimensional set of equations known perfectly to the experimenters, but NOT known by the volunteers. So the dynamics of the economy here were made artificially simple, and hence easier to learn than they would be in a real economy; but the volunteers weren't given any crutch to help them learn, such as full knowledge of the form of the equations. They had to, gasp, learn on their own! In a series of experiments, Assenza and colleagues then measured what happened in the economy -- did it settle into an equilibrium, did is oscillate, etc. -- and also could closely study how people formed expectations and whether their expectations converged to some homogeneous form or stayed heterogeneous. Did they eventually converge to rational expectations? Umm, NO.

From their conclusions:
In this paper we use laboratory experiments with human subjects to study individual expectations, their interactions and the aggregate behavior they co-create within a New Keynesian macroeconomic setup and we fit a heterogeneous expectations switching model to the experimental data. A novel feature of our experimental design is that realizations of aggregate variables depend on individual forecasts of two different variables, the output gap and inflation. We find that individuals tend to base their predictions on past observations, following simple forecasting heuristics, and individual learning takes the form of switching from one heuristic to another. We propose a simple model of evolutionary selection among forecasting rules based on past performance in order to explain individual forecasting behavior as well as the different aggregate outcomes observed in the laboratory experiments, namely convergence to some equilibrium level, persistent oscillatory behavior and oscillatory convergence. Our model is the first to describe aggregate behavior in a stylized macro economy as well as individual micro behavior of heterogeneous expectations about two different variables. A distinguishing feature of our heterogeneous expectations model is that evolutionary selection may lead to different dominating forecasting rules for different variables within the same economy, for example a weak trend following rule dominates inflation forecasting while adaptive expectations dominate output forecasting (see Figs. 9(c) and 9(d)).

We also perform an exercise of empirical validation on the experimental data to test the model’s performance in terms of in-sample forecasting as well as out-of-sample predicting power. Our results show that the heterogeneous expectations model outperforms models with homogeneous expectations, including the rational expectations benchmark. [MB: In the paper they actually found that the RE benchmark provided the WORST fit of any of several possibilities considered.]

In the experiments, real learning behavior led to a range of interesting outcomes in this economy, including persistent oscillations in inflation and economic output without any equilibrium, or extended periods of recession driven by several distinct groups clinging to very different expectations of the future. Relaxing the assumption of rational expectations turns out not to be a minor thing at all. Include realistic learning behavior in your models, and you get a realistically complex economy that is very hard to predict and control, and subject to many kinds of natural instability.

One of the most important things here is that the best way to generate behavior like that observed in this experimental economy was in simulations in which agents formed their expectations through an evolutionary process, selecting from a set of heuristics and choosing whichever one happened to be working well in the recent past. This builds on earlier work of William Brock and Cars Hommes, the latter being one of the authors of the current paper. It also builds, of course, on the early work from the Santa Fe Institute on adaptive models of financial markets, which uses a similar modelling approach.

So, here we do have an alternative to rational expectations, one that is both far more realistic in psychological terms, and also more realistic in generating the kinds of outcomes one sees in real experiments and real economies. Wonderful. Economists are no longer stuck with their RE straitjacket, but can readily begin exploring the kinds of things we should expect to see in economies where people act like real people (of course, a few economists are doing this, and Assenza and colleagues give some recent references in their paper).

I think this kind of thing is much more deserving of being called a "learning literature." I don't know why it doesn't get more attention.

Wednesday, December 11, 2013

Secular stagnation.... just a poor excuse?


Larry Summers recently made some waves with his proposal that maybe we're in a new era of "secular stagnation," in which low growth is the norm, and much of it comes through temporary and artificial bubbles. Paul Krugman backed the idea here. At face value, it all seems somewhat plausible, but also sounds a lot like a "just so" story to cover up and explain why economic policy and low interest rates haven't been enough to encourage new growth, yet also haven't caused inflation. It also fits together quite well with the usual stories told about the failure of ordinary policy at the zero lower bound, which turns ordinary economics on its head.

Now, I don't want to say that story is completely wrong, but remember that it comes out of quite standard macro analyses based on representative agent models with individuals and firms optimizing over time, and where -- perhaps most importantly -- things like debt overhang do not enter in any way into explaining how people are behaving (and why they may be hugely risk averse). That should be enough to raise some major questions about the plausibility of the story, especially in the aftermath of the biggest financial crisis in a century. For a lot more on such doubts, see the illuminating recent paper Stable Growth in an Era of Crises by Joseph Stiglitz.

But also see this convincing counterargument by some analysts at Independent Strategy, as discussed in the Financial Times. From Izabella Kaminska's discussion:
From the note, their main points are:
• There is no shortage of high return investment projects in the world. And the dearth of global corporate investment, which drove the great recession, means that productive potential is shrinking despite corporate profitability, leverage and cash balances being sound.

• The three ingredients for growth are a) a stable macro environment; b) a sound banking system; c) economic reforms that encourage entrepreneurship. What is missing right now is private sector confidence in the ability of governments and central bankers to provide all three.

• Credit bubbles can boost growth only temporarily and incur heavy costs in terms of subsequent deleveraging and misallocation of resources.
And expanding a bit further, they add:
Secular stagnation is a myopic and short-term view for two reasons. First, it is based on the experience of the Anglo-Saxon economies and parts of Europe currently as well as Japan since the bursting of the bubble at the start of the 1990s. Krugman muses that interest rates should be set at the growth rate of populations, because they would then be equal to a society’s potential capital productivity (and the long-term return on it). But the change in population growth is less relevant than the rise in productivity of an expanding workforce.

Take Germany: its population is ageing and its net population growth is slowing to a trickle (although that may be improved by increased net immigration from southern and eastern Europe). But Germany’s productivity level and growth is high (as is total factor productivity, expressing the gains from technology). Italy has a similar stagnation in its working population, but its real GDP growth has disappeared because of the fall in total factor productivity — Figure 1.
 Read Kaminska's discussion here.

Actually, this push back isn't really surprising. It's what you get if you take the longer historical view, rather than trying to make excuses for why economic theory still can't make sense of things (the theory is poor, that's why!). As a couple of economists from Goldman Sachs noted just after Summers' speech:
"Our view of the recent weakness is more cyclical than secular... The slow rate of recovery in recent years is roughly in line with the performance of other economies following major financial crises, as shown by Reinhart and Rogoff, and the reasons for the weakness in aggregate demand over the last few years have now begun to diminish."
This refers to the great book by Reinhart and Rogoff, by the way, not their other discredited paper.