Thursday, November 18, 2010

What exactly do Selgin, Lastrapes, and White expect me to think about this?

George Selgin, William Lastrapes, and Larry White recently released a Cato Institute working paper looking at economic performance before and after the Fed which concludes that the Fed has been bad for the macroeconomy on all kinds of measures. The sort of people you'd expect to blog on a Cato working paper have fulfilled their duties in praising the piece: Cafe Hayek, Coordination Problem, and Econ Log. I'm not sure exactly what there is to be so impressed about. It's when I see things like this that I'm glad I came up through the labor economics/econometrics side of economics first before developing an interest in macroeconomics, because methodologically this paper has very little to offer and I'm not sure why that's not being pointed out by more people.

The paper is essentially a century and a half long pre-post test of the Federal Reserve, an approach to performance evaluation that simply wouldn't fly in, say, the labor economics literature. The question of how macroeconomic performance and volatility have changed over time may be an interesting question to answer, but simply comparing a century with the Fed to a century without the Fed isn't really a test of the impact of the Fed at all.

The problem is (as Arnold Kling does point out in his post) that things change over time that have nothing to do with the Fed, and those changes are being picked up and attributed to the Fed by Selgin, Lastrapes, and White. What's worse is that we are likely to have substantial endogeneity in this case. It's not simply that other changes that are going on may be falsely attributed to the Fed - it's that the Fed was created precisely to address these changes in the economy that were occuring! It's the same problem that you have when you regress output or employment on federal spending to get at the impact of fiscal policy - it doesn't work because you implement fiscal policy precisely when the economy is weak. The pre-Fed period was an agrarian period where the United States largely followed other technological leaders, where we still had a frontier, and when growth was extensive: applying standard production techniques to more land, more people, and more capital. The Fed was established at a time of transition, precisely because of the volatility that that transition was expected to usher in: we emerged as an industrial economy that was a technological leader, not a follower. The frontier was closed and the low hanging fruit of extensive growth was no longer available. Instead, we grew intensively - by innovating on the production processes that we had been using. This road is inherently rockier, which is precisely why you saw a growth in macroeconomic management at this time. Selgin, Lastrapes, and White see a more volatile 20th century and attribute it to the Fed - I see a more volatile 20th century and say "well isn't that why we made the Fed in the first place? Wasn't this precisely what we were expecting when we were having these discussions about another central bank between 1907 and 1913?".

I've been doing program analysis at the Urban Institute for four and a half years now, and I haven't once done a pre-post test like the one Selgin, Lastrapes, and White present here. We could never get that kind of assessment published, and if we were producing it for a client, we would probably get it sent back to us with an angry note and lots of red ink. It's not like these points should be lost on monetary or macroeconomists. It's precisely these concerns with identification problems that lead Barro and Romer to do their innovative work trying to isolate the effect of fiscal policy. Macroeconomists do know about this stuff. But sometimes I feel like they are less attuned to the problem than labor economists and econometricians. A lot of people looked at the Selgin, Lastrapes, and White paper and thought it was pretty interesting and compelling. I just thought "What the hell do they expect me to think of this? This is fluff. I can't make heads or tails of this."

Read critically, people.

UPDATE: Looking at what happens before and after a policy change can be improved by a lot of different methods. One is to compare the pre-post change to a pre-post change from a comparison group that did not experience the policy change (this is called difference-in-differences). This allows you to subtract out the change over time that wasn't associated with the policy. People also trust pre-post tests if they look at change in a very short time period as a result of a very sharp policy change where nothing else in the system has changed (this is called a natural experiment or a regression discontinuity design, depending on how it's implemented). Intuitively we can also put more weight on pre-post observations in a shorter time period (for example, in my 1920-21 paper I point out a rise in economic activity after the Fed finally cut rates and argue that this is consistent with the idea that monetary policy is stimulative - it's not a rigorous test, but the tight time frame makes it more plausible as an illustration). A dicier method that I'm usually skeptical of is called "instrumental variables", which uses an exogenous proxy to measure the impact of an endogenous policy change. These have to be very, very well justified.

UPDATE 2: So I asked Don Boudreaux to humor me and let me know what he thought of my critique. He writes: "I'll satisfy your curiosity. First, you ought to pay more attention to public choice; your 'the Fed was created with good intentions' assumption is naive. Second, given your reasons for dismissing as 'fluff' Selgin's, Lastrapes's, and White's conclusions, what evidence have you that the Fed succeeded? Third, I see no reason why the economy of the twentieth century - or the economy that those mythical wise and well-meaning technocrats of a century ago thought they foresaw - was destined to be inherently more volatile than was the economy of the 19th century. Why, for example, you presume that the closing of the geographic frontier is economically significant is beyond me. Were not, say, the economic frontiers pushed outward by electrification, by telecommunications, by inexpensive transoceanic steamships, by the Internet less important than the Wm. Jackson Turner's geographic frontier? If your criticism of Selgin-Lastrapes-White is the harshest that there is, their paper is destined to become a celebrated classic." And odd and telling reply, isn't it? The public choice point is strange - I don't deny the incentives associated with the founding of the Fed, but there's nothing in public choice theory that requires an entirely myopic perspective on public actions either. I also never said that there was evidence succeeded. In fact I said quite the opposite, didn't I? He makes a big fuss over the frontier point too, which I re-explain in the comment section
here. Notice anything missing in Don's response? Oh yeah - no mention at all of the massive methodological problems with this paper that formed the crux of my criticism. Now Don's not dumb. He understands the argument. He makes the same argument when cautioning people against being too sanguine about empirical fiscal multiplier estimates. So he understands the concept perfectly. Which is why when he fails to mention it, it's pretty transparent what it means - he knows I'm right, that he was overzealous in praising the paper, and he doesn't want to admit it.

4 comments:

  1. Reluctantly, yes.

    Even though I agree with the conclusion of the paper, I wasn't much impressed by the actual content. I was kind of disappointed. The "tests" were often woefully inadequate from a methodological perspective. The authors occasionally recognise this, and then plough on anyway. Each historical episode really needs a full-length paper alone, otherwise they're just preaching to the choir. The whole paper feels like the summary and conclusion of something much longer (like Selgin and White's life's work!)

    Also, even the abstract feels a bit disingenuous:

    "As the one-hundredth anniversary of the 1913 Federal Reserve Act approaches, we assess whether the nation‘s experiment with the Federal Reserve has been a success or a failure. Drawing on a wide range of recent empirical research, we find the following ... We
    conclude that the need for a systematic exploration of alternatives to the
    established monetary system is as pressing today as it was a century ago."

    I mean, come on! I don't know much about Lastrapes, but Selgin and White are ardent opponents of central banking. They write as though when beginning research for this paper they didn't already know what the conclusion would be. Perhaps that's just proper form in a paper like this, but it seems ridiculous when you have read what these scholars have written elsewhere.

    Having said all that, I do like the paper, I am just better at making critical comments than praising comments :P

    ReplyDelete
  2. "Each historical episode really needs a full-length paper alone, otherwise they're just preaching to the choir. The whole paper feels like the summary and conclusion of something much longer (like Selgin and White's life's work!)"

    ^This. In short, I don't think they care about what you think Daniel. ;)

    ReplyDelete
  3. Drat! I could have sworn they were hanging on my every word :)

    ReplyDelete
  4. The problem is...that things change over time that have nothing to do with the Fed, and those changes are being picked up and attributed to the Fed...

    I'm not an economist, but I have noticed this sort of logic among free-banking, libertarian, Austrian Econ types for some time. I think that they have a prejudged position on things - the state, the market, etc. - and they go looking for facts to back it up. Their arguments seem plausible until you follow the thread for a while.

    You could easily substitute 'language' for 'central banking' in their arguments. Sure, if humans had never developed language we would not have these economic problems...no inflation, depression, bank runs, etc. Like those who long for the prelapsarian paradise in Eden (before Man had knowledge) these people seem to me to be utopians, and completely ahistorical in their outlooks, despite their 'comparisons' across time of the economy.

    ReplyDelete

All anonymous comments will be deleted. Consistent pseudonyms are fine.