Murphy and Bayes’ Rule

Two horrible posts on Austrian economics, one by Paul Krugman and the other by Brad DeLong. The only thing I’ll write with respect to Krugman’s post is that he’s, maybe unintentionally, insulting his peers in the profession with Austrian tendencies. This is understandable because Krugman’s knowledge of what the Austrian School stands for is virtually zero — he only knows two predictions that some Austrians made on the eve of the Great Recession. But, given that his knowledge on this subject is close to zero, his posts should say as much (explicitly or implicitly).

DeLong’s post is a footnote to Krugman’s, where he chastises Bob Murphy for not amending his priors given the failure of his prediction (of high inflation). The problem is that the only option for change, acording to DeLong, is to embrace a non-Austrian worldview. In reality, this isn’t the only option. Murphy’s prediction of high inflation is based on a heuristic that represents a set of “Austrian” theories (because it’s not as if Austrian and non-Austrian theories are always incompatible) as he interprets them. There were many Austrians, including myself, who were not convinced the prediction was right — there were some which predicted deflation. The point is, whatever Murphy’s heuristic is, it can represent different interpretations of “Austrian” theory. So, a revision of his priors could lead him to reviewing his understanding of the underlying theory. Another way to put it is: there are two options, replace or revise. DeLong pretends as if the former (replace) is the only choice.

I can change the case to make my point clearer and more agreeable. Consider a Bizarro world where the dominant school is the Austrian one, and that DeLong belongs to a loud minority known as the “Keynesian School.” Following an important financial crisis in 2007–09 a deep recession sets in. Based on his interpretation of Keynesian doctrine — I don’t mean “doctrine” disparagingly, although it fits with Krugman’s accusation of cultism —, DeLong predicts that a fiscal multiplier of two, such that the radically austerian program of negative fiscal expenditure implies a contraction in GDP by X by 2012. It turns out that the economy contracted by less than X. Bob Murphy, columnist and blogger for The New York Times, notes the failure of DeLong’s prediction and publishes: man, this guy was really wrong, if he weren’t part of the cult then he would have replaced his priors, by which I mean he’d adopt Austrianism. But, this statement is wrong, because to (our hypothetical) DeLong the failure induced him to revise his priors by fixing the perceived mistakes in his understanding of the Keynesian worldview.

To Krugman and DeLong remaining an Austrian after the failure of your prediction is fanatical because they mistake the prediction for the theory. They simply don’t know the actual theory well enough to concede the possibility of revision. This is understandable since they are great economists who are too busy with other topics that are more important to them. I’m only asking that they they accept the limits to their understanding. Don’t insult somebody by essentially accusing him of being part of a cult knowing that his behavior doesn’t reflect that at all.

8 thoughts on “Murphy and Bayes’ Rule

  1. Roberto Severino

    In other words, why can’t we have a better economics blogging community where people don’t have to result to ad hominem insults constantly because someone adheres to a particular economic school of thought? Everyone is so caught up in trying to be right and dismissing what everyone else says.

    Reply
  2. Brad DeLong

    I *really* don’t think you understand. Back in 2007, I was roughly 65% monetarist, 33% Keynesian, 2% Austrian. Now I’m 0.5% Austrian, 21.5% monetarist, 78% Keynesian. The past five years have been a huge shock that has carried a lot of information, and rational and reasonable people adjust their beliefs in response. Murphy finds shifting your beliefs in response to new information to be blameworthy. That is just silly. You too, apparently, think the job is to (a) maintain your theory but (b) “save the phenomena” by somehow adding another epicycle. That is not a good game to play. You should not play it.

    That’s just weird.

    Brad DeLong

    Reply
    1. JCatalan

      All I’m saying is that Austrians have a certain body of theory that all interpret in their own way, and in some cases even disagree on. A revision of his priors doesn’t necessarily require Murphy to change from 95% Austrian to 50% Austrian. He could remain 95% Austrian and just change his understanding of the theory (or develop the theory further). But saying 95% Austrian doesn’t mean not changing one’s priors at all.

      You write, “Murphy finds shifting your beliefs in response to new information to be blameworthy.” I guess I don’t see this at all. Murphy, in what you excerpt in your post, is making the same point I am. Whatever mistakes in prediction you and Krugman made didn’t cause you to totally change your world view, it just pushed you guys to amend it. That you amended it in the way you did doesn’t mean that others have to do exactly the same.

      What I’m saying is that it should be acknowledged that there are an infinite number of possible changes that could occur, based on one’s interpretation of the evidence, one’s interpretation of the theory, and the knowledge someone holds. Where you see dogmatism I see the product of a genuine disagreement in interpretation. It doesn’t make sense to me to compare this to being part of a cult. I just think the fact that you’re .5–2% Austrian induces a bias where you don’t see any merit in any Austrian position that you don’t already hold.

      It’s not just you. Austrians do it the other way around (as Daniel Kuehn notes), and they’re equally in error.

      Reply
  3. komeil

    Infected body is not going to be cured by injecting ever increasing morphine.Doing so will lead to death of body. This is the essence of Austrian. I do not see any innovations in Keynesian except the various way of injecting this morphine.This is the essence of Keynesian.

    Reply
  4. Blue Aurora

    Although I’m more inclined to Professor J.B. DeLong and D.P. Kuehn’s intellectual positions…

    Keynesians’ knowledgability (sp?) of opposing intellectual sides and civility toward said opposition (and the applying the same things to Austrians) is indeed far from ideal. Then again, I’ve noticed that people tend to practice confirmation bias in many domains and avenues of life (including the blogosphere for politics and economics). While I don’t deny that things can be improved, I do believe that we all should acknowledge this: realistically-speaking, it’s going to be very hard to change it.

    That stated, it seems that Professor J.B. DeLong has been going on his views and knowledge of probability theory and statistics as of late. Not too long ago, DeLong got into a debate with the statistician Cosma Shalizi over something not too far away from the matter of Bayes’ Rule.

    Although this isn’t Professor DeLong’s blog, I’d like to ask him this:

    What is your view on the standard decision theory in academic economics: Subjective Expected Utility?

    It seems to me that Professor DeLong – like Nobel Memorial Laureates Joseph E. Stiglitz and Paul R. Krugman – is inclined to support the tenets of S.E.U. decision theory as a pedagogical device and as a hypothesis. See the following link below for reference.

    http://factsandotherstubbornthings.blogspot.com/2012/11/is-ellsberg-paradox-really-paradox.html

    Reply
  5. Neodoxy

    Jonathan,
    Do you think that Murphy’s interpretation theory was necessarily wrong? While I am ignorant as to the exact nature of Murphy’s original statement, I don’t understand why his prediction of hyperinflation was necessarily unreasonable at the time. Given how much money has piled up in the vaults of banks and in the pockets/portfolios of businesses rather than actually circulating, can we really say Murphy was wrong as such? This seems like it could be about the real nature of the exact economic conditions of the day and how policies were really implemented, rather than the inherent theory.

    Reply
    1. JCatalan

      It does have a lot to do with conditions of time and place, but that should be part of a prediction. But, even apart from that, the main tool the Fed has been using has been asset swaps. A bank’s lending is a function of its capital reserves. But, the banks’ assets remained the same, except that instead of holding illiquid mortgage backed securities they exchanged these for cash. The main purpose of the Fed’s actions were to, at first restore, and then mimic wholesale credit channels to maintain short-term liquidity so that banks could keep up with their liabilities in the face of collapsing asset values. This isn’t a simple helicopter drop, so the effects are going to be different — much less inflationary. I agree that the Fed probably avoided a much greater deflation, but I think everyone agrees with respect to that.

      Reply
  6. Pingback: Revisiting Priors | Economic Thought

Leave a Reply