Tyler Cowen recently made this comment:
It is oddly fashionable in the economics blogosphere to insist that microfoundations do not matter or are not a worthy matter of study. Papers like this show that in fact they matter a great deal.
My own view is that microfoundations could be useful, but we are not making progress in that area, despite herculean efforts by hundreds of researchers that are much smarter than I am.
I recall back in the 1970s that Robert Lucas would tell us that we were making great progress in establishing the microfoundations of a new and much more scientifically rigorous macroeconomics. We now know his optimism was misplaced. Indeed we even knew this in 2007, before the Great Recession ruthlessly exposed the empty pretentions of modern macroeconomics.
The second sentence in the quotation above refers to a new paper by Christiano, Eichenbaum and Trabandt. Tyler provides the abstract:
We develop and estimate a general equilibrium model that accounts for key business cycle properties of macroeconomic aggregates, including labor market variables. In sharp contrast to leading New Keynesian models, wages are not subject to exogenous nominal rigidities. Instead we derive wage inertia from our specification of how firms and workers interact when negotiating wages. Our model outperforms the standard Diamond-Mortensen-Pissarides model both statistically and in terms of the plausibility of the estimated structural parameter values. Our model also outperforms an estimated sticky wage model.
That sounds really hopeful, doesn’t it? Perhaps it’s a great paper. But over the past 35 years I’ve read about 147 other similar abstracts that sound equally hopeful. My problem with this sort of literature is that it is not progressive. It ignores most other previous “successful” studies. Some would argue that this means economics is a not a “science.” When I read that, it just makes me think that the writer doesn’t have a good definition of ‘science.’ But I sympathize with the negativity, if not the epistemology. There is something wrong with a field where we have one explanation after another, and each new one ignores most of the existing literature.
So if I’m so smart what do I suggest? I don’t have a good answer. One solution would be to accept the other 147 papers on the microfoundations of wage and price stickiness as being “true” (in a Rortian sense that truth is what my t-statistics let me get away with) and then build on those. Thus Eichenbaum, Christiano and Trabandt could have said; “we already have 147 good models of wage inertia, we propose to add a 148th explanation to the ever growing “standard mega-model.”" Much like physicists add new particles to the standard model each time they are discovered. Unfortunately the macro mega-model would be far less elegant than the physics mega-model. Indeed journal referees would probably laugh if someone tried to do what I suggested, even if it could be done. And the math involved in integrating all these disparate approaches may prove to be an insurmountable obstacle.
I am also skeptical of Tyler’s claim that the paper shows that microfoundations matter a “great deal.” Consider the following comment by Tyler:
In this model money has a positive effect on output and employment, but only by lowering real interest rates and inducing more consumer spending. It does not in general appear to be a large effect, but there is a real positive effect. You will note that the underlying parameters of the labor matching model are defined in real terms.
If this were how money actually impacts output and employment, then I’d be inclined to agree with Tyler, the microfoundations in their paper would matter a great deal. But in fact the most powerful monetary shocks tend to have very little impact on real interest rates, and sometimes even move them in a perverse direction. Monetary policy works by changing NGDP against a backdrop of highly inertial nominal wages:
Y = f(W/NGDP)
We observe that nominal wages are highly inertial, even in the face of sudden NGDP shocks caused by monetary policy, but don’t quite know why they are so inertial. I actually agree with Tyler that macroeconomics might be improved a bit if we understood why nominal wages were so inertial, but I simply don’t see much progress in that area. Or perhaps I should say I see too much progress; 148 theories that have been “proven true,” with no understanding of how to fit them together into a coherent and useful mega-model.
That doesn’t mean that we can’t do good macro-policy. We have drugs that effectively treat mental illness, despite not knowing what’s going on at the neural level. We have good monetary policies for dealing with wage inertia, such as NGDPLT, even though we don’t know what causes wage stickiness. If we did know the cause we might be able to develop slightly better monetary rules, but I doubt the implications would completely overturn our current understanding of monetary policy. A good monetary policy is likely to look roughly like NGDP targeting, almost regardless of what causes nominal wage inertia.
PS. Perhaps I should add a bit to my dismissal of the debate over what is a “science.” One obvious problem is that it’s never clear whether what is being debated is the nature of economics, or the definition of ‘science.’ That alone makes it a really stupid discussion, roughly equivalent to arguing whether sandals are shoes. Even worse, the debate often proceeds with the implicit assumption that being designated a “science” is some sort of honor, like winning a Nobel Prize. Now you are a science! This shows a profound naivete about the nature of human beings’ attempts to better understand the world around them. The works of Plato, Shakespeare, Freud, Isaac Newton, Edward Gibbon, Adam Smith, and Dante are all very different from each other. And yet all of them help us to better understand the world we live in. To argue that one or another is superior by virtue of being or not being a “science” is utter nonsense.