The Importance of Being Exogenous

I usually don't respond to blogosphere commentary on my work because, after all, time is scarce. But this critique by Nate Silver is noteworthy because the error it makes is so fundamental. It offers a teachable moment.

Silver questions my citation of the Romer tax study in my recent NY Times piece. He writes:

The paper (.pdf) Mankiw refers to, written by Berkley Economists Christina and David Romer, is the sort of thing that will make your head spin. But the gist of it is that (i) It is very important to differentiate the motivation for different types of tax cuts or tax increases, and (ii) a certain type of tax cut or tax increase may have a much larger effect on growth than is generally acknowledged. The type of tax cut that Romer and Romer think falls into this category is what they call an "exogenous" tax cut -- one designed not to counter business cycles, but rather a "spontaneous" tax cut under relatively healthy economic circumstances. This is very much not the type of tax cut that we are contemplating right now.

Essentially, Silver says that the Romers study exogenous tax changes, the tax changes now being contemplated in DC are not exogenous, and therefore the multipliers estimated by the Romers do not apply to thinking about current policy.

This argument raises the question: Why did the Romers focus on exogenous policy changes? The reason is that these are the only changes that can be used to reliability identify the effects of tax policy. If a tax change is made in response to some event, call it X, that influences the economy, it is hard to disentangle the effects of the tax change from the the direct effects of X. The Romers focus on exogenous tax changes for the same reason doctors conduct randomized drugs trials--not because they are interested in randomization as a prescriptive tool, but because randomization solves a statistical identification problem.

Imagine if a clinical doctor reasoned the same way as Silver did. He would say, "All the evidence on the effects of this drug are from randomized drug tests. In my practice, I never randomize treatment of my patients. Therefore, I can safely ignore the results from the randomized experiments."

That is, of course, fallacious. We need the randomized experiments to inform us about the effects of medical interventions, even though interventions in practice are rarely randomized. Similarly, we need to consider the effects of exogenous tax changes, even though many actual tax changes are not at all exogenous.