From Dark Age to Renaissance
By Arnold Kling
I’m feeling somewhat lonely these days. My understanding of macroeconomics is closer to that of Paul Krugman, Mark Thoma, and Brad DeLong than it is to that of Robert Barro, John Cochrane, or Eugene Fama. And yet I am a stimulus skeptic. I’ll reiterate why I am a skeptic. But what I really want to get into is my view of the intellectual history of macro since the 1970’s. I think that Krugman’s phrase Dark Age of Macroeconomics aptly describes the whole period.
First, about the stimulus. My objections are:
1. We have a heterogeneous labor force, and that will require Hayekian market adjustment, not central planning. Look at this data from page 96 of the Goldin-Katz book, which shows the percentage of the labor force with various levels of education:
|year||high school dropouts||high school graduates||some college||college graduates|
Can we ignore the implications of this for macroeconomics? I think not. See my lectures.
2. The stimulus plan is highly partisan and ideological in nature. Obama may claim to be Lincoln, but the content of the plan is Radical Reconstruction. The aim is to reconstruct the economy with a smaller private sector and a larger public sector. Any notion of “timely, targeted, and temporary” is out the window. The Radical Reconstructionists don’t much care that a lot of their “stimulus” will begin after the recession is more than three years old.
3. I have a Minsky-esque view of the nature of the crisis. That is, we have gone from being risk-loving (ponzi finance, in Minsky’s terminology) to ultraconservative (hedge finance, as he calls it). I don’t think we can (or should) put back together the Humpty-Dumpty of securitization and leverage that we had before. Businesses need to expand out of good, old-fashioned profits. Fiscal stimulus ought to be aimed at improving profitability.
Now, on to my personal history of macro.I spent the academic year 1975-1976 working as a research assistant at CBO and at the Fed. My job was to run simulations of macroeconometric models, most especially the Fed’s model, known as the Fed-MIT-Penn model. As of about 1970, those models were thought of as the empirical embodiment of macroeconomic theory. There was no difference between “scientists” and “engineers,” to use Greg Mankiw’s terminology.
When I began graduate school at MIT in the fall of 1976, I had the smug notion that I would be ahead of my classmates in my knowledge of macro, based on my close familiarity with the models. I was in for quite a setback. Not only were most of my classmates ahead of me in the mathematical aptitude that really mattered, but it turned out that I was arriving at a moment when macroeconometric models were in utter disrepute. Macro at MIT was dominated by Stanley Fischer and Rudi Dornbusch, the former influenced by Chicago, the latter influenced by Robert Mundell, and both absolutely in love with the mathematics of rational expectations–Euler equations, saddle points, and so on.
My second year at MIT, Ray Fair came from Yale for one quarter as a guest lecturer. He was and still is the last academic devotee of macroeconometric models. The class consisted of him going through his model. In the back of the room, John Huizinga and other students were soon laughing at him. By this point, what we knew about statistical methods was enough to make us realize that Fair’s techniques (and those of other macro modelers) were a joke.
The problem is that macro time series are nonstationary and have low statistical power. Go back to the table on labor force composition I gave earlier in this post. There is no way to estimate a model in a way that is robust to such dramatic structural changes.
Because the data are not powerful enough to reject any hypothesis against an interesting alternative, the macroeconomic “scientists” divorced themselves from the “engineers,” who continued to use macroeconometric models. In my view, neither the scientists nor the engineers accomplished anything. The scientists did mathematical masturbation, with some occasional, isolated empirical work that satisfied journal referees based on whatever fads were taking place at the moment but had no lasting persuasive impact. The engineers provided forecasts and policy simulations that were precise in form and totally unreliable in substance. In any case, the engineers were kept out of economics departments.
This generation of academic macro wankers managed to achieve consensus, primarily by avoiding discussing the fundamental issues of macro–how labor markets fail to clear and how financial markets affect real activity. So we get Olivier Blanchard’s smug pronouncement
a largely shared vision both of fluctuations and of methodology has emerged. Not everything is fine. Like all revolutions, this one has come with the destruction of some knowledge, and suffers from extremism, herding, and fashion. But none of this is deadly. The state of macro is good.
Now that we are in a financial crisis, this “shared vision” is turning out to be nothing of the sort. Christina Romer and Larry Summers have gone back to trusting macroeconometricians, treating model multipliers as if they were fact rather than fiction. On the other side, Fama and others seem to have gone back to some pre-Keynesian view that simply ignores the possibility that there can be such a thing as unemployment.
Unfortunately, just because mainstream macroeconomists wasted their efforts the past thirty years does not mean that the answer is to rediscover macroeconomics as it stood in 1970. The pre-Dark-Age macro is not some mystical treasure map. It has a lot of theoretical and empirical gaps that make it unsatisfying.
I find myself incorporating strands as disparate as Hayek’s information theory and Minsky’s risk preference cycles into my thinking. I wish I had more professional company. I agree that macroeconomics has been through a Dark Age. But I think it has a way to go in order to experience a Renaissance.