Kenneth Rogoff's Hooverismo…
Laboring in Vain to Raise the Level of the Debate Over the Debt and the Deficit: Low-Morale Weblogging

You Don't Need a Rigorous Microfoundationeer to Know Which Way the--Well, to Know Much of Anything, Really


Memory is always shaky. And my undergraduate transcript is in a box somewhere in the basement. But I believe that the semester I audited Tom Sargent's evening Monetary Economics class was also the semester I took (P/F, with lots of help from high-powered physics major roommates) Edward Purcell's Introduction to Quantum Mechanics class.

At the end of the physics class Purcell had us set up a proton--not a real proton, with its finite mass and complex internal structure, but a fake proton that was infinitely massive, possessed one simple unit of positive electric charge, and had no magnetic moment (rather ironic given what Purcell's Nobel Prize in Physics 1952 was for)--and its associated electromagnetic field, or, rather, not the relativistic QED electromagnetic interaction with its Feynman diagrams and virtual particles, but just a static energy potential. We then used the Schroedinger Equation (and where does that come from?) to calculate the position distributions associated with the first three energy eigenvalues of a single electron--but not a real electron: an electron-like object without any internal angular momentum. And I did it! I had built a full microfounded model of where the hydrogen atom's electron is in its 1s, 2s, and 2p energy eigenstates, and thus of why those eigenstates had the energy values they did, and why the hydrogen atom could emit photons of energies corresponding to 2s->1s, 2s->1s, 3s->2s, and 3s->1s transitions (and absorb the same photons) and not photons of other energies. And then it was a simple matter of programming to generalize this microfounded construct to create a fully microfounded discipline of chemistry, right?

Well. Except for the fact that the electron was not an electron. Except for the fact that the proton was not a proton. Except for the fact that the model was not relativistic. Except for the fact that the static electric potential was not the actual quantum-electro-dynamic field. All these matter for the real electron: in my model the 2s energy level was the same as the 2p energy level--there was no Lamb shift and hence no Lamb-structure lines in my hydrogen spectrum, because there was neither the self-interaction of the electron with its own electromagnetic field nor interference by virtual electrons and positrons with the photons carrying the electromagnetic forces between the proton and the electron.

And while Richard Feynman may have whispered about the Black Arts of "renormalization" and "measurement apparatus" and "sum-over-histories", plus warned his students against the "mumbo-jumbo of the collapse of the waveform" when he did his four-semester introductory course, Edward Purcell did not in his one-semester quantum mechanics intro. And while we could do our approximate calculation for one fake-electron around one fake-proton with one fake-interaction, I do not believe to this day that there are any first-principles microfounded calculations of something as basic as the electron affinities of the different elements:

File Electron affinity of the elements svg Wikipedia the free encyclopedia

Indeed, we first made astatine at the far right of the chart, just to the left of radon, 73 years ago. Yet we do not know--either by theory of by observation--many of the properties of astatine: "astatine may be dark, or it may have a metallic appearance and be a semiconductor, or it may even be a metal". And I have never met anybody who can tell me why chlorine has a higher electron affinity than fluorine while bromine has a lower electron affinity than chlorine only by frantic handwaving about "partial shielding of the nuclear charge by the electrons of the 3d subshell". And nobody has even tried to explain to me why one might from quantum-mechanical theory expect to see the electron-affinity pattern of the nine-element grouping cobalt-rhodium-iridium-nickel-palladium-platinum-copper-silver-gold.

So I was rather insulated from Tom Sargent's laying down the theoretical principle that economic models needed truly rigorous microfoundations. I listened to him present microfounded overlapping-generations models of the value of money, and learned an enormous amount--but not because the models were rigorously microfounded. The cash I carry around in its pocket does not have value because I want to buy things from people who will die before I have made anything they want to buy from me and I will also want to sell to people who will only produce things I want to buy after my death. That just isn't going on. Overlapping-generations models of money are metaphors--tractable and useful metaphors invented by the brilliant Paul Samuelson, but metaphors. They are not models that are accurate descriptions of reality at a micro level out of which we can construct accurate descriptions of reality at a macro level by simply examining their large-scale emergent properties.

It seemed to me that if Tom Sargent were correct in his methodological declarations, we might as well go home--and that the physicists and chemists should go home as well. But I knew that the physicists and chemists did not need to go home. Even though they lacked a rigorous micrfounded model of the chemical properties of astatine, when they said thing would go "boom" they tended to go BOOM!!!!

All this is a very long-winded wind-up to why I am not nearly as upset as Nick Rowe is about the New Keynesian model as studied by Jordi Gali and others. Jordi simply wants to study those equilibria of the model that converge over time to a classical full-employment long run, and that seems to me a fine thing to do. Nick Rowe, however, is very unhappy and complains:

The "canonical" New Keynesian model… has replaced ISLM as the new "workhorse" model. It has three equations. The new "IS curve"… the ratio of current demand to expected future demand is a negative function of the real rate of interest… the expectations-augmented Phillips Curve, which both defines "full-employment" and tells you the relationship between deviations of output from full employment and deviations of actual from expected inflation… the monetary policy reaction function…. This model has zero tendency towards full employment, even if you assume a sensible monetary policy. New Keynesian modelers just assume a tendency towards "long run" full employment, even though it's not there in the model.

But, at least as I see it, the purpose of the New Keynesian model is to serve as a counterexample: Robert Lucas and company claimed that models with what they called "rigorous microfoundations" could not have equilibria with broadly Keynesian conclusions. New Keynesian models are demonstrations that this is not so. Whether and to what degree actual modern economies are self-equilibrating to anything like full employment in a reasonable medium run is a very important question, but is not something that these models would be very likely to be able to tell us.

And I would disagree with Nick Rowe's claims that the Old Keynesian IS-LM-PA (price adjustment) model got us back to full employment in the medium term (as long as monetary policy was not insane). It didn't get you back to full-employment if you were in a liquidity trap. Then, as Paul Krugman says, you have to go outside the model--either to a Pigou real-balance effect of outside money on aggregate demand via consumption or to Keynes's point that eventually "use, decay, and obsolescence" of capital (plus deleveraging) would shift the IS curve out and to the right.

And I do think that once you recognize that we are no more in the genuine "microfoundations" business than even Peter Higgs and company, Paul Krugman gets the recent intellectual history right:

There was… an undercurrent of dissatisfaction…. In “normal” economics we assume that prices rise or fall to match supply with demand. In Keynesian macroeconomics, however, one simply assumes that wages and perhaps prices too don’t…. Why make this assumption? Well, because it’s what we see in reality…. But that’s an unsatisfying answer…. The trouble is that finding that deeper explanation is hard…. Modern “New Keynesians” have come up with stories…. But… it’s more a matter of offering excuses, or if you prefer, possible rationales, for an empirical observation that we probably wouldn’t have predicted if we didn’t know it was there. This, understandably, wasn’t satisfying to many economists. So there developed an alternative school of thought, which basically argued that the apparent “stickiness” of wages and prices in the face of unemployment was an optical illusion. Initially the story ran in terms of imperfect information; later it became a story about “real” shocks, in which unemployment was actually voluntary; that was the real business cycle approach.

And so we got the division of macroeconomics. On one side there was “saltwater” economics…. On the other side was “freshwater” – people… who went for logically complete models even if they seemed very much at odds with lived experience…. What’s not OK is… that freshwater economics became a… cult, ignoring and ridiculing any ideas that didn’t fit its paradigm. This started very early; by 1980 Robert Lucas… wrote approvingly of how people would giggle and whisper when facing a Keynesian. What’s remarkable about that is that this was all based on the presumption that freshwater logic would provide a plausible, workable alternative to Keynes--a presumption that was not borne out by anything that had happened in the 1970s. And in fact it never happened: over time, freshwater economics kept failing the test of empirical validity, and responded by downgrading the importance of evidence…. And this inward turning had what can now be seen as a fateful consequence: freshwater macro, basically something like half or more the macroeconomics field, stopped teaching not only new Keynesian research but the past as well. And what that meant was that when crisis struck, we had half a generation of economists who not only had no model that could make sense of the crisis, but who blithely reproduced classic errors of the past…. In 2008 and 2009 we had well-known professors from Chicago and elsewhere opposing stimulus because… income must be spent, so government spending cannot increase demand. Intellectually, much of the profession had unknowingly regressed 75 years.

Worse yet… economists responded to the bitter warfare between schools of thought by running away from business cycle issues in general. I know whereof I speak: when Robin Wells and I began writing our principles of economics textbook, the general view was that you should focus on long-run growth, and relegate things like recessions and recoveries to a brief section at the end. Why? Because focusing on the long run was safer, less likely to get the committees that choose textbooks riled up….

Finally, all was not well even in saltwater economics… they suffered from rigor envy. And so New Keynesian models tried to have as few deviations from perfect markets as possible…. The result was DSGE…. DSGE models have three aspects that make them unsuited to times like these. First, they’re unwieldy…. Second, they assume that… shocks are more or less normal, not involving large, low probability events…. Finally, the desire to make the things tractable tends to favor linearity… not a modeling style that leaves you ready to deal with sudden financial crisis…. What we really needed, I’d submit, was a large number of economists ready and willing to go for good first approximations…. Good old-fashioned IS-LM fits the bill, and as I see it the economists who did best in this crisis began with IS-LM, then backed it up later with simplified versions of New Keynesian analysis. But knowledge of IS-LM has become surprisingly rare, and comfort with it… rarer. And this has had terrible consequences.

In the years after 1980, and even more so, the years after 2000, the foundations for crisis were laid. The banking system became, de facto, largely unregulated and unsecured. Leverage rose…. The conditions for disaster became ever better; and the disaster came…. The answer should have been simple, and backed by an overwhelming consensus. The immediate problem was a huge shortfall of demand… we needed not just a rescue of the financial system but also strong government action to support demand while the wreckage was cleared…. There was and is a case for large-scale unconventional monetary policy, which in a zero-bound economy has to work largely through inflation expectations. But the more proximate tool… was fiscal policy, especially increased government purchases of goods and services. Anyone who knew the IS-LM model understood that. But too much of the economic profession had lost the hard-won understanding of earlier generations. So instead of a common call for action, we got acrimonious argument, with quite a few economists essentially acting as spoilers, undermining the credibility of those trying to get governments to do the right thing. And as I said, to a remarkable extent the “learned” arguments against government action were actually repeating fallacies like Say’s Law and the Treasury View that had been thoroughly refuted in the 1930s.

Should we be surprised, then, that economic policy makers, after responding fairly effectively to the banking crisis, proceeded to lose the thread?… And so our response to the crisis has been utterly inadequate. The best you can say about economic policy in this slump is that we have for the most part avoided a full repeat of the Great Depression…. And I blame economists, who were incoherent in our hour of need. Far from contributing useful guidance, many members of my profession threw up dust, fostered confusion, and actually degraded the quality of the discussion. And this mattered. The political scientist Henry Farrell has carefully studied policy responses in the crisis, and has found that the near-consensus of economists that the banks must be rescued, and the semi-consensus in favor of stimulus in the initial months (mainly because the freshwater economists were caught by surprise, and took time to mobilize) was crucial in driving initial policy. The profession’s descent into uninformed quarreling undid all that, and left us where we are today.

And this is a terrible thing for those who want to think of economics as useful. This kind of situation is what we’re here for. In normal times, when things are going pretty well, the world can function reasonably well without professional economic advice. It’s in times of crisis, when practical experience suddenly proves useless and events are beyond anyone’s normal experience, that we need professors with their models to light the path forward. And when the moment came, we failed.


Feed You can follow this conversation by subscribing to the comment feed for this post.


This reminds me of the quote by the statistician George Box: "Essentially all models are wrong, but some are useful."

The comments to this entry are closed.