Expongo aquí dos visiones distintas sobre el evidente fracaso de la microfundation en la Macroeconomía. La primera es de Krugman, la segunda de Farmer Lo hago porque simple complaciencia en las historias brillantemente contadas. Ah! Y además estoy de acuerdo con ellos, aunque discrepan en algún punto.
Microfoundations and the Parting of the Waters
The blogospheric debate about microfoundations, saltwater/freshwater and all that has, I think, been illuminating. Among other things it’s serving almost as an oral history of What Really Happened – minus the oral part, but not mediated by the usual slowness and overthinking of formal publication.
And I think the intellectual history is useful, because it gives you some idea of how people came to make the choice of which side to be on. It’s certainly possible to make the case for an eclectic, fairly salty approach on general principles, as Simon Wren-Lewis, Noah Smith, and Nick Rowe do. But the abstract logic gains force when you recall how it actually happened.
Oh, and I was there – not as a participant in the growing macro war, but as a student at the time the great divide was taking place. I felt the seduction of the microfoundations-uber-alles doctrine, but also got to watch as the demand for microfoundations, originally grounded in appeals to empirical power, became free-floating, a dogma to be defended in the teeth of the evidence.
So, if you had to choose a beginning, it would be the famous Phelps volume. The papers in that volume all started with two observations, of which the first was that there was overwhelming evidence for some kind of short-run non-neutrality of money. None of the papers in that volume questioned the proposition that nominal shocks had large real effects. You can see why if you look at annual changes in nominal versus real GDP between 1950 and 1970:
Obviously there was a near one-to-one correspondence. Obviously, too, it was really hard in that era, with its lack of major supply shocks, to tell a story in which real GDP was driving nominal spending rather than vice versa. So the Phelps volume began with the stylized fact that in the short run nominal demand, driven for example by changes in monetary policy, gets reflected largely in quantities rather than prices.
But as the papers also observed, it was hard to explain that fact in terms of standard microeconomics: with everyone acting rationally, money should have been neutral even in the short run. Traditional Keynesian analyses simply said that people aren’t completely rational, that they have money illusion – or maybe that contracts are focal points in which nominal wages or prices matter because of salience, even though they should be arbitrary. But these were ex post rationalizations rather than being derived from some kind of fundamentals.
So the Phelps crowd came up with a lovely story: you see, it was all about information. Individuals and firms couldn’t tell, in the very short run, whether a rise in the price they were being offered represented a shock specific to them – people for some reason wanted more of their widgets — or a general change in demand. It was rational to respond to an idiosyncratic rise in demand by producing more, so confusion could explain why short-run aggregate supply seemed upward-sloping.
As Phelps and others (including Milton Friedman, who was thinking along similar lines) realized, this meant that the apparent tradeoff between unemployment and inflation would be unstable: sustained inflation would get built into expectations, and would no longer produce low unemployment. The stagflation of the 70s seemed to confirm this prediction, and brought the microfoundations project immense prestige. Encouraged by all this, freshwater economists gleefully proclaimed Keynes dead, the subject of nothing but “giggles and whispers”.
But here’s the thing: after that initial success, Phelps-Lucas/type microfoundations quickly collapsed both intellectually and empirically. Intellectually, the problem was that rational individuals simply should not have been confused in the way the models demanded; there’s too much information out there, whether in newspapers or in asset prices. You just couldn’t get a Lucas supply curve out of a model looking even vaguely like the real economy.
Empirically, the problem was that slumps last too long. Even if you wave away the information problem, confusion about aggregate versus idiosyncratic shocks can last for quarters, maybe, but not years.
So the truth was that microfoundations in macroeconomics had its moment, but failed utterly at the one thing it was sold, above all, as being able to do – namely, give a better explanation of why nominal shocks have real effects. Time, you might think, to reconsider the project.
And some did. There was a revival of Keynesian thinking in the late 70s and early 80s, albeit one that tried to cram as many microfoundations into the models as possible without being grossly unrealistic.
But many economists had so committed themselves to the idea that Keynes was dead and rationality roolz that they simply dug in deeper. Rationality-based microfoundations must be right; if their microfoundations couldn’t explain why nominal shocks have real effects, then nominal shocks must not have real effects – it’s all real shocks. And so real business cycle theory was born.
So now we have people debating whether models with microfoundations lead to better predictions, both of the future and of policy impacts, than models with ad hoc elements; as Wren-Lewis and Smith say, this is by no means obvious if the microfoundations are wrong, as they often clearly are. But what you want to realize is that this isn’t going to convince the microfoundations crowd. After all, more than thirty years ago they decided that the joy of microfoundations trumped the utter failure of microfounded models to work in practice, and they have now trained successive cohorts of students in this view.
There are, it’s true, some hints of a guilty conscience – as Matt Yglesias points out, there’s the odd tendency of freshwater types to immediately accuse anyone with saltwater ideas of being dishonest. (I’m not a nice guy, but if look at what I said about, say, Cochrane, it was that he was ignorant, not corrupt.)
Oh, and the notion that there had been a convergence of views by 2007, which was then ruptured by the crisis, was a saltwater delusion. People like Olivier Blanchard convinced themselves that the other side was listening; it wasn’t. The hysterical reaction to the notion that fiscal policy is effective at the zero lower bound demonstrated that the freshwater types had never bothered to learn the least thing about how New Keynesian models worked.
So there’s a lot of history here; but the main driver behind this history was, I believe, the inability of many economists to accept the fact that they took a wrong turn.
Neo-Paleo-Keynesianism: A suggested definition
There has been a lot written on the blogosphere in recent weeks about the microfoundations of macroeconomics. Tony Yates argues in favour of micro-founded structural models. Adam Posen is sceptical of micro-foundations and Simon-Wren Lewis, Noah Smith and Nick Rowe call for a more eclectic approach. For those looking for a neat summary of these debates, Paul Krugman traces the history of macroeconomic ideas. Responding to a piece by Brad Delong, he argues that there has been a recent resurgence of what he calls “neo-paleo-Keynesianism”. This is very useful concept and I have much in common with the ideas expressed in Paul's piece. This essay offers a novel definition of the term that Paul coined and an invitation to fellow academics to join me in pursuing an agenda based on this definition.
I agree with Paul Krugman: macroeconomics has taken the wrong path. I disagree with Paul’s reading of when that happened. The error has nothing to do with new-classical versus new-Keynesian approaches; it is a more fundamental error that pervades both new-classical and new-Keynesian schools of thought. Macroeconomics took a wrong turn in Cambridge Massachusetts in 1955 when Paul Samuelson, in the third edition of his textbook, introduced the idea of the “neoclassical synthesis” (see Pearce and Hoover for a discussion of the influence of Samuelson’s textbook). Everything since then has been the economic equivalent of the scientific theory of phlogiston.
Many economists are exposed to the philosophy of science through Milton Friedman’s book, Essays in Positive Economics. Friedman promoted the views of Karl Popper who argued (here) that science progresses when theorists make bold conjectures that are confronted by facts. Those conjectures stand until they are refuted by the evidence. Occasionally, economics students are exposed to the ideas of Thomas Kuhn who talks (here ) of paradigm shifts and scientific revolutions. Rarely does the economics curriculum of a Ph.D. program have time to push much further into the methodology of science. That’s a pity since there is much to be learned from alternative philosophies.
Axel Leijonhufvud has argued persuasively (here) that we have much to learn from Imré Lakatos, a philosopher of science who spent much of his career at the London School of Economics. Lakatos, (here) in contrast to Popper and Kuhn, sees science as a set of competing scientific research programs. His approach is a useful one for understanding the current debate amongst practicing macroeconomists who are facing a series of natural experiments that provide serious challenges to both new-classical and new-Keynesian agendas.
According to Lakatos, all tests of scientific theories are necessarily tests of joint hypotheses. The sciences, both physical and social, are best characterized as interacting communities of scholars. Those scholars adhere to research programs that interpret the evidence through different lenses.
Each research program has a ‘hard core’ and a ‘protective belt’. When an event in nature appears to refute a theory, the scientist must decide which of the possible components of his theory should be rejected in order to reconcile his worldview with the outcome he observed. Assumptions that make up the hard core of a research program will never be rejected; instead, the scientist will amend one of the assumptions in its protective belt.
The new-Keynesian research program is the descendent of Samuelson’s ‘neoclassical synthesis’. According to that synthesis, the economy is ‘Keynesian’ in the short-run when not all markets have had time to clear: it is ‘classical’ in the long run when all price adjustment has run its course. It is these twin propositions that form the hard core of the new-Keynesian program. According to that program, market economies are self-correcting, and although the adjustment to the long-run equilibrium may take time, that adjustment will, eventually, occur.
Despite its name, the new-Keynesian research program is neither new, nor Keynesian. The idea, that real economic activity may be different from its long run steady state as a consequence of sticky prices, is firmly rooted in monetarist tradition. It originated in the eighteenth century and is summarised by David Hume in his delightful essay, Of Money. Keynes argued, in contrast, in the opening chapters of The General Theory, that high unemployment of the kind that persisted in the Great Depression is one of many possible steady state equilibria.
How can we recover this idea, without discarding three hundred years of microeconomic principles? I will refer to a research agenda that maintains the notion of unemployment as a steady-state equilibrium as paleo-Keynesianism. When combined with general equilibrium theory in a way that provides a micro-foundation to this key idea, I will refer to the resulting synthesis as neo-paleo-Keynesianism.
In contrast to new-Keynesian ideas, neo-paleo-Keynesianism does not assume that ‘frictions’ prevent wages or prices from clearing markets. It does not deny that wages and prices move slowly, relative to quantities. But that observation does not mean that we must assume that there are menu-costs, contracting costs or any other artificial barrier to price adjustment. As I explain (here), sticky prices may simply be part of a rational expectations equilibrium. Robert Lucas was exactly right when he argued (here) that markets are always in equilibrium. But accepting that proposition does not require us to accept that equilibrium is unique; nor must we accept that equilibrium is optimal, or that unemployment is voluntary.
The neo-paleo-Keynesian (NPK) research program is unashamedly neo-classical. It seeks to reconcile Keynesian ideas with the microeconomics of general equilibrium theory; and it does so in a new way. As with new-classical and new-Keynesian economics, neo-paleo-Keynesianism constructs models of rational actors who interact in markets. In contrast to new classical and new-Keynesian programs, neo-paleo-Keynesianism contains two propositions that are absent from the hard core of these agendas: 1) there is a continuum of possible equilibrium unemployment rates and 2) the unemployment rate that prevails is determined by the ‘animal spirits’ of investors.
How might one accomplish this agenda? One approach that I describe (here) combines a search theory of unemployment with an asset-pricing model and (here) I develop a model driven by animal spirits that returns to paleo-Keynesian ideas without invoking sticky prices or wages. My survey paper (here) explains what is different about neo-paleo-Keynesianism, from the new-Keynesian alternative.[1]
What is wrong with new-Keynesian economics and why should we prefer the NPK approach? Lakatos provides us with an answer. Research programs are not refuted, as in Popper, nor are they dramatically overturned, as in Kuhn. They simply attract more new adherents than their competitors. In the language of Lakatos, research programs are progressive or degenerative.
In the normal course of events, a successful research program meets challenges to its hegemony by modifying hypotheses in its protective belt. A progressive research program is one that occasionally makes a prediction that is confirmed by experiment or, in the case of macroeconomics, by history. A degenerative research program is one that struggles with continued refutations by continually modifying its protective belt in ever more inelegant ways.
The new-Keynesian research program, like the new-classical program before it, is degenerative. Like Ptolemaician astronomy, it explains new data by adding ever increasing layers of complexity. And like that theory; new-Keynesian economics has not succeeded in making a single prediction that has been confirmed by fresh evidence that was unavailable when the theory was constructed.
So where does that leave the neo-paleo-Keynesian? What event is explained by this approach? First and foremost, there is the persistence of long-term unemployment in the wake of a financial crisis. For a new-Keynesian, it is hard to explain why wages and prices have been so sticky that employment still has not recovered more than five years after the onset of the stock market crash. For a neo-paleo-Keynesian it is an expected outcome for a theory in which high involuntary unemployment is an equilibrium state.
There are those who claim that we should return to the Keynes of the General Theory while rejecting the attempt to build microfoundations. That is the message, for example, of post-Keynesians like Paul Davidson. While there are attractive elements to that path, I do not believe that we should abandon all of classical economics. There is much to like in the ideas of demand and supply and several branches of economics have had notable successes by following the idea that actors are rational and goal oriented. Examples that come to mind are auction theory that was used successfully to sell the rights to the electromagnetic spectrum in the UK and matching theory that was used to develop kidney exchanges.
The path of combining multiple equilibria with ‘animal-spirits’ has the potential to explain many of the puzzles thrown out by recent experience. I showed (here) that it offers an alternative explanation of the monetary transmission mechanism, one that outperforms the new-Keynesian alternative, and I argue (here) that active stabilization of financial markets offers an alternative approach to traditional fiscal policy as a means of maintaining full employment.
If Keynes were alive today, one thing is certain; he would not be a Keynesian in the sense in which that term is used today. Keynes was notorious for changing his views on a daily basis and was said to be capable of holding several conflicting opinions at the same time. Would he be a neo-paleo-Keynesian? Who knows? What seems certain, to me, is that existing ideas have failed us. For me, that’s enough to try something new.
[1] I am not the only one that is seeking a return to paleo-Keynesian ideas. Others I would place in this category include Stephanie Schmitt Grohé and Martín Uribe (here) who drop classical principles of market clearing and return to the Keynesian concept of involuntary unemployment. Narayana Kocherlakota constructs (here) a model of incomplete factor markets, Greg Kaplan and Guido Menzio (here) combine search in the labor market with search in the product markets and Pascal Michaillat and Emmanuel Saez (here) takes prices as parametric. All of these papers are examples of multiple steady state equilibria.
No hay comentarios:
Publicar un comentario