« The power of competition, or should there by library fines? | Main | I, Dirt »
Real business cycle theory is not dead
Stephanie Schmitt-Grohe and Martin Uribe report (ungated here):
In this paper, we perform a structural Bayesian estimation of the contribution of anticipated shocks to business cycles in the postwar United States. Our theoretical framework is a real-business-cycle model augmented with four real rigidities: investment adjustment costs, variable capacity utilization, habit formation in consumption, and habit formation in leisure. Business cycles are assumed to be driven by permanent and stationary neutral productivity shocks, permanent investment-specific shocks, and government spending shocks. Each of these shocks is buffeted by four types of structural innovations: unanticipated innovations and innovations anticipated one, two, and three quarters in advance. We find that anticipated shocks account for more than two thirds of predicted aggregate fluctuations. This result is robust to estimating a variant of the model featuring a parametric wealth elasticity of labor supply.
Jerry-rigged, yes. But the principle of Occam's Razor isn't as strong as many people think. If the mechanisms governing real world outcomes actually are that complicated, and if the simpler theories have failed, than jerry-rigging it is. And, as Christina Romer has pointed out, deflationary money shocks are still bad for output and employment, which means the theory has at least an extra branch.
Posted by Tyler Cowen on August 18, 2008 at 05:04 AM in Economics | Permalink
Comments
I particularly like Menger's anti-razor, the Law of Miserliness; "Entities must not be reduced to the point of inadequacy."
Posted by: Jethro at Aug 18, 2008 7:57:42 AM
Jerry-rigged is wrong and a conflation of two rather different terms. Neither ancestral termm, jury-rigged or jerry-built, seems to fit. A jury rig is an temporary mast and yards built if the orginals are damaged and used to get to a port for repairs. While if something is jerry built it is built cheaply and shoddily. I suggest using kludged instead, as it can mean a rather inelegant initial working solution, without the connotation that it is a repair which jury rigged has.
Posted by: Brett Dunbar at Aug 18, 2008 8:00:08 AM
I have noticed that Tyler has a soft spot for RBCT. There is something about this theory that is quite appealing to him. Why? Is it because the idea of a single explanation for both growth and cyclicality appeals to him aesthetically? (Recall his interest in quantum mechanics as well; could be the same dynamic at work.) Is it because it was the first BC model that he worked from beginning to end, and thus he harbors fond memories of it? Is it because the moral neutrality of RBCT--the idea of recession as an optimal response--appeals to him ideologically?
Posted by: Rich at Aug 18, 2008 8:54:20 AM
"If the mechanisms governing real world outcomes actually are that complicated, and if the simpler theories have failed, than jerry-rigging it is."
Here here!
Posted by: goodnessOfFit at Aug 18, 2008 9:34:54 AM
Can someone give me a real world example of an innovation anticipated several quarters in advance.
Posted by: spencer at Aug 18, 2008 10:15:57 AM
In answer to spencer, here are three: pre-announced tax system changes; pre-announced changes to pension eligibility ages; large investment projects for which plans are drawn up well in advance.
But really, much as I admire Schmitt-Grohé and Uríbe's work, RBCs are all about putting all the economic behaviour of the model into the shocks -- ie the parts of the model that aren't explained. Just because these models are "estimated" doesn't make them "testable". They are methodologically vacuous, as wells as being ideologically non-neutral. This is not a comment on the authors, though. The profession as a whole has done this to itself.
Posted by: no surprise at Aug 18, 2008 4:14:34 PM
"If the mechanisms governing real world outcomes actually are that complicated, and if the simpler theories have failed, than jerry-rigging it is."
then
Posted by: Kevin H at Aug 18, 2008 4:45:09 PM
Tyler: If the mechanisms governing real world outcomes actually are that complicated, and if the simpler theories have failed, than jerry-rigging it is.
The problem is that when you jury-rig with formulae and parameters about as complex as the data against which they are tested, you have no way of knowing whether you have actually come up with a useful (i.e. predictive) model or whether you have simply recoded unique economic data in some fancy way with no predictive power.
Given the poor state of the art in the statistics of econometric time series (straight regression leads to spurious acceptance of false models, regressing the differences leads to spurious rejection of true models, cointegration depends on the usually false hope that you've accounted for every nonstationary price component, etc.) the problem is even worse: if the model has many parameters it's very easy to overfit the data, again with the result that the model can make few if any useful predictions.
Even worse, every parameter added to the model potentially adds to the confusion about what the model means and implies. There's much to be said for models in which logic and consilience, rather than mathematical precision, are prized. But mathematical or otherwise, simpler models that describe more accurately more economic phenomena are better: they are much more likely to be true rather than spurious, and it is usally easier for economists to agree about what the description of a simpler theory actually means.
Posted by: Nick at Aug 18, 2008 9:36:25 PM
P.S. the real antidote to the complexity of the real world is humility. Economists should stop pretending that we can come up with useful theories of everything, for example theories of phenomena like the so-called "business cycles" that consist of actions deriving from the unique thoughts of millions of human brains. By the pigeonhole principle, these phenomena are vastly more complex than one brain could possibly comphrehend.
Instead we should honor economists for describing small things well. Not even physicists have a theory of everything, despite physics being far simpler and more objective than the study of humankind.
Posted by: Nick at Aug 18, 2008 9:51:53 PM
