Tuesday, May 6, 2014

Good ad hoc vs bad ad hoc


This goes with this post and Tom Brown's comment asking what I meant by bad ad hoc models. Plastic, like expectations in economics, can be molded into many form factors.

15 comments:

  1. Hi Jason... OK, I'm feeling a bit thick today... so the point is the one on the left is "thrown together" with some general pieces... while the one on the right is a more polished looking complete unit... but about as realistic?

    ReplyDelete
    Replies
    1. This looks like bad ad hoc to me (perhaps the worst?):

      http://skepdic.com/adhoc.html

      And I read your link on Krugman's defense of ad hoc in some circumstances... so the old light bulb is just starting to flicker on... (but you still need to explain the Lego horses to me :)

      Delete
    2. Yes that is definitely bad ad hoc!

      A good ad hoc model is put together with using a limited set of pieces you have at hand that gets you something that looks at least a bit like reality.

      A bad ad hoc model is assuming the piece you need or implicitly assuming the result.

      In the paper Mark Sadowski linked to, they wanted to see if calling a recession affected low information consumers' spending (see if they could build a LEGO horse). Then they assumed there was a bit of the model that affected low information consumers' spending (the assumed a LEGO horse piece). It might be interesting if it was hard to think of a way that to make that work -- but it's apparently not because all that paper did was assume a function of the prior period and the current period: f(yt, yt-1). They didn't have to see if the function violated causality or caused the economy to collapse or made interest rates fluctuate wildly. In fact, they didn't have to check anything about it.

      As a side note, there's an effect called the EMC effect I worked on as a grad student. One of the issues was that it was so easy to come up with ad hoc models that could explain it (the "microfounded" theory QCD is intractable for nuclear systems). My advisor said that EMC stood for "Everyone's Model is Cool". I think this is a good analogy for the current state of macroeconomics, however I think it is more due to the uninformative macro data.

      Delete
    3. This is an example of bad ad hoc :)

      http://physicsoffinance.blogspot.com/2014/05/conservative-economists-assume-what.html

      Economists do this with uncanny regularity. Assume a can opener ... goes the joke. A kinder way to put it is that they're busy translating their hypotheses into math and showing that the math really is their hypothesis.

      Delete
    4. Funny... I have heard both of those jokes. Just to be absolutely clear, you are agreeing with Mark Buchanan there, right? Not commenting on *his* post.

      Delete
    5. Yeah, I'm agreeing with Buchanan -- you have to be vigilant of assuming the conclusions in your model.

      Delete
    6. BTW, if you didn't see this Sadowski comment, you might be interested:
      http://www.themoneyillusion.com/?p=26514#comment-327623

      Delete
    7. That's nice of him. I'm glad I'm not considered shrill or annoying :)

      Delete
  2. Also, I provided a link to your site in the comments to Glasner's new neo-Fisherite post:
    http://uneasymoney.com/2014/05/06/monetary-theory-on-the-neo-fisherite-edge/
    I'd love to see you get some more feedback from some of those guys.

    ReplyDelete
    Replies
    1. I always learn something from Glasner. I saw that post, but put it in the pending bin for now. I have a couple of "blog about" tags in my RSS feed. Right now I'm making visualizations for this Nick Rowe post:

      http://worthwhile.typepad.com/worthwhile_canadian_initi/2014/05/mrscc1r1mpkmrtcidmrtcidt.html

      Delete
  3. Jason, if you were to summarize your views on how expectations are used/misused in economics, what would you say? Would you say that the concept of expectations is generally abused by economists? I'm looking for something analogous to Nick Rowe / Brad DeLong's succinct summary of good monetary policy:

    "Good monetary policy attempts to make Say's Law true in practice, if not in theory."

    Would your one or two sentence summary mention bad "ad hoc" models? I do recall your phlogiston description. Would that be the one? (unfortunately if most people are like me, they'll have to look that one up). Would it be fair to simply say that you think that many economists abuse the concept of expectations?

    ReplyDelete
    Replies
    1. Economic models of expectations can work too well.

      (I'll see if I can do better.)

      Delete
  4. Jason, Noah uses the term "post hoc" in this article:

    http://noahpinionblog.blogspot.com/2014/05/academic-racism-has-kn-problem.html

    Do you see a connection with your point here?

    ReplyDelete
    Replies
    1. Yeah, there is definitely some overlap: each difference from some underlying theory is attributed to a particular effect of expectations.

      Delete

Comments are welcome. Please see the Moderation and comment policy.

Also, try to avoid the use of dollar signs as they interfere with my setup of mathjax. I left it set up that way because I think this is funny for an economics blog. You can use € or £ instead.