Wednesday, March 21, 2012

Two great points from stickman

Here.

I'll put the first point in my own words: My skepticism of you increases exponentially with the number occasions on which you choose to buck consensus. Stickman is commenting at first on the popularity of weird diets among libertarians, but he goes on to note Lew Rockwell's promotion of a variety of kooky ideas. This looks really bad, and I feel this way about climate skeptics a lot. OK, skepticism is a good impulse - I'm not going to dismiss you as crazy just because you reject this scientific consensus... but if you reject the climatologists' consensus, and then you also reject the monetary macro consensus (let's generously allow that reasonable people can disagree on fiscal policy), and then you also reject empirical research in favor of a priorism, and it looks very bad. Add creationism to the mix and I wonder why we're even talking. But it doesn't stop with scientific consensus. Often these are the people who hem and haw about the Civil Rights Act, who think FDR was a fascist, and who think that the Progressive movement was the worst thing that ever happened to America. These sorts of people aren't interested in getting to the bottom of things so much as bucking consensus whenever and wherever they come across it.

It should come across as very odd to people that climate change skepticism is almost exclusively a preoccupation of the right. This is not to say everyone on the right rejects climate change - it's to say that if you reject climate change you are almost certainly on the right. What is happening in the Earth's atmosphere should not be correlated at all with political views, and yet it's highly correlated. That alone - leaving all the climate science itself to one side - should lead people to steeply discount climate change skepticism. In other words, if climate change skepticism were evenly distributed across the political spectrum, I would probably take it a lot more seriously than I do.
*****

The second point that stickman makes is one I've made on here many times before: Selgin, Lastrapes, and White (2010) is very good history, but very bad policy evaluation. Stickman points out that they acknowledge at the outset of the paper that they lack a counter-factual and they're simply doing history. That's true - but then one wonders why they conclude with this:

"(Available research does not support the view that the Federal Reserve System has lived up to its original promise. Early in its career, it presided over both the most severe inflation and the most severe (demand-induced) deflations in post-Civil War U.S. history. Since then, it has tended to err on the side of inflation, allowing the purchasing power of the U.S. dollar to deteriorate considerably. That deterioration has not been compensated for, to any substantial degree, by enhanced stability of real output... Finally,the Fed cannot be credited with having reduced the frequency of banking panics orwith having wielded its last-resort lending powers responsibly."

The bolded lines are not the sort of statements that are made about historical research - they are statements made about evaluation research. And that's certainly the way this paper is promoted. And while Selgin, Lastrapes, and White may have reeled in their fans at some point, I'm not aware of any case where they have said "oh no - you misunderstand - you can't take this to be an evaluation of the Federal Resere - it can't accomplish that".

A good rough-and-ready test is to just think about all the ups and downs that they cite and ask "did similar things happen in economies that did not change their central banking system in 1914?" The answer is "yes!". If you look up banking panics on Wikipedia and look at their list for the 19th century, you see it flip flopping between panics originating in England and the United States. 1819 - U.S., 1825 - Britain, 1837 - U.S., 1847 - Britain. Britain had a central bank long before the 19th century so it offers a good opportunity to hold institutions constant here. It seems like banking panics in the 19th century were just something that happened in credit-based industrializing economies. Flash forward into the early 20th century and you see the same thing - the elephant in the room in the post-Fed record, the Great Depression, happened in Britain and the U.S.. If we treat this like a difference-in-differences test, a cursory look at the evidence suggests that the pre-post work done in Selgin, Lastrapes, and White is probably going to be a very misleading way to infer anything at all about Fed policy. Stickman's right - the introduction seems to say that. I just wish the title, the conclusion, and the promotion said it as well!

8 comments:

  1. Those guys are anti-central-banking. They seem to view Canada (supposedly no bank failures during the Depression) & old Scotland as the closest to their ideal. It would seem odd to use another country as a central bank facing crises as if that was evidence against them!

    ReplyDelete
    Replies
    1. Wonks Anonymous -
      The empirical task at hand is identifying the impact of the Fed. That means extracting variation from the data over this period that is not attributable to institutional changes. As a plausible solution, that can mean either differencing out the change in a country that had central banking the whole time or differencing out a place that didn't have central banking the whole time. The point is you need to compare a case of institutional change to a case of institutional stability to identify the treatment effect.

      As for Canada - the structure of the banking system might have helped prevent bank failures (I'm taking your word on this - I don't know much about it), but you might want to check out Canada's data on unemployment, GDP, and price level during the 1930s before you go any further down that road... besides, the Bank of Canada was established during the depression so that muddies the waters.

      These are still going to be very rough estimates, so I would agree - we should have as many comparison cases as possible - with and without central banks. But the point is, to identify the impact of the Fed you want to compare institutional change to no institutional change.

      Delete
  2. 1) I'm reticent to argue by association -- whether that be guilt or honour -- but there certainly does come a point where wider credibility matters. On that note, Rockwell does promote some wacko ideas, doesn't he?! I don't follow him closely enough to know whether he gets called out on this, or if his audience genuinely views the world in as conspiratorial terms as he does. Seeing things like this, however, doesn't help dissuade me from the latter view... (Seriously though, those of you who follow LRC: Does his stuff on evolution, HIV/AIDS, etc bother you?)

    2) My general point about empirical (scientific) vs a prioristic approaches was set against the preceding post to the one Daniel discusses here. Gary Taubes, i.e. the guy writing about the paleo diet, was doing so in very scientifically cogent terms. I've actually listened to a few podcasts with his libertarian promoters as well, and he always makes the same careful points about hypothesis testing, experimental design, etc. Of course, his interviewers gleefully agree with everything he says during the podcast and, yet, this seems at complete odds with a lot of economic stuff that they push every other week!

    3) Hmmm... Yes, I suppose that LSW's conclusions do jar a bit with their tentative opening remarks. I would perhaps still give them some leeway, given their historical focus. (Strictly speaking, they are obviously correct and entitled to point out that the Fed presided over highest post-Civil War periods of inflation and deflation... much as they leave the all-important -- and unknown? -- counterfactuals unsaid.) What I do find slightly strange, however, is that they think a "deterioration in the purchasing power of the dollar" should necessarily be compensated by a "stability in real output". Surely real wages are the obvious "compensating" corollary to inflation?

    ReplyDelete
  3. I noted some odd correlations in political/scientific opinions (such as global warming) here.

    I recall a Taubes fan referring to "politically incorrect" nutrition study results and found the application of that phrase odd.

    ReplyDelete
    Replies
    1. Some interesting links and discussion, TGGP. There seems to be a fair deal of overlap in our preferred topics of discussion. (E.g. I've previously mentioned Al Roth myself.)

      On politically incorrect studies... Morally sensitive events (and even those characterized by outright tragedy) are often used productively by economists i.t.o. research settings. Given the thrust of my original post, this is particularly true for studies that utilize natural experiments to establish causality. Steven Levitt's abortion-crime link is an obvious example, but there are many others...

      On that note, you just have to shake your head when Krugman's more inept critics showcase his citation of WW2's pro-stimulus effects as evidence for an inherent war-mongering agenda. To be sure, there are reasonable positions that someone could take in arguing against Krugman's views, but the "warmonger!" accusation misses the point to a rather embarrassing degree.

      Delete
    2. Al Roth is interesting. I'm often critical of rationalism (in a perhaps strawman opposition to empiricism/pluralism), but Roth's work invokes in me a rationalist disgust with disgust a la Tabarrok.

      I don't know if the abortion-crime link is the example you'd want to give, since the original paper seems to have been the result of coding errors. I've heard David Card is one of the pioneers of the "natural experiment" method, with rivers determining boundaries of whether kids stay in high school or something. I recall there being an article attacking "cute-o-nomics" of that sort in place of "big questions", but I don't think the full version stayed available.

      Back to Roth, the Israeli daycare study cited in Freakonomics may be wrong. In terms of morality interfering with standard econ utility calculation, I might cite Scott Atran.

      I like older Krugman of slate columns in part because he was talking standard consensus stuff. In his recent arguments indicting the profession I've found plausible the claim that he isn't very familiar with a lot of the latest research (Arnold Kling, a more extreme critic of macro, definitely seems to be out of it even according to those unlike me still allowed to comment at EconLog), and also that said research hasn't been very helpful. I have a general heuristic of deferring to the expert consensus, even if they don't know terribly much, because it's unlikely that my own independent opinions are going to be any more accurate.

      Interesting point on Akerlof & Coase.

      Delete
    3. I thought I had a full post up on that heuristic I mentioned, but turns out I just laid it out in a comment. I also linked to Robin Hanson on contrarianism there, and in response to Daniel's argument about the number of ways one bucks the consensus, there's The Correct Contrarian Cluster and How to be Radical, both linked here (trying to avoid tripping any spam filter).

      Delete
    4. On the Levitt crime-abortion thing... You are quite correct about the coding errors, of course. I can't open the link that you provide, but recall discussing this with some classmates a while ago.[*] Still, I think the point about "economics not being a morality play" stands and is one that most people tend to agree with.

      The Card study that you describe sounds more like an straight-up IV than a natural experiment to me, but he certainly has been a pioneer in that area too. Personally, I'm quite fond of cute-o-nomics. I suspect it's hard not to enjoy that stuff. However, it's obviously understandable that those things must fall by the wayside in the face of our current economic situation... Cf. Noah Smith.

      The heuristic you describe dovetails nicely with my own views, though I do try to spend time on anything that strikes me as particularly interesting or relevant (e.g. climate)...

      [*] The move towards making data and code available is one of the most laudable movements in economics. Another example: I recently read this paper, which overturns a rather famous result by Leras-Muney (2005) on the positive effect that education has on long-term health; in part by uncovering an important coding error.

      Delete

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