Thursday, 9 August 2012

Giving Economics a Bad Name


Greg Mankiw is known to every economist and economics student, if only because of his best selling textbook. John Taylor is known to every macroeconomist, if only because of the large number of bits of macro with his name on it (Taylor rule, Taylor contracts etc). Both are respected by other academics because of the quality and influence of their academic work.

With two others, they recently wrote this about the Obama administration’s attempts to stimulate the economy through fiscal policy after the recession: “The negative effect of the administration’s ‘stimulus’ policies has been documented in a number of empirical studies.” They then quote from two studies. The first looks at a minor aspect of the stimulus packages, the Cash for Clunkers attempt to bring forward car purchases. There are other studies of this programme which are more favourable. The second study is co-authored by John Taylor, and others have interpreted his findings differently.

No other studies are directly referred to. That might just be because the overwhelming majority suggest that the stimulus package worked. Dylan Matthews on Ezra Klein's blog documents them here. As I wrote in a recent post, the evidence is about as clear as it ever is in macro. Which is not too surprising, as it is what Mankiw’s textbook suggests, and it is what the New Keynesian theory both authors have contributed to suggests.

Now the quote comes from a paper prepared for the Romney presidential campaign. It is clearly political in tone and intent. As both academics are Republican supporters, it may therefore seem par for the course. But it should not be. The Romney campaign publicised this paper because it was written by academics – experts in their field. It allows those who oppose fiscal stimulus to continue to claim that the evidence is on their side – look, these distinguished academics say so.

It is one thing for economists to disagree about policy. It would also be fine to say I know the evidence is mixed, but I think some evidence is more reliable. It is not fine to imply that the evidence points in one direction when it points in the other. I say here imply, because the authors do not explicitly say that the majority of studies suggest stimulus is ineffective. If they chose their words carefully, then you have to ask whether ‘intending to mislead’ is any better than ‘misrepresenting the facts’. Was that the intent, or just an isolated unfortunate piece of bad phrasing? All I can say is read the paper and judge for yourself, or this post from Brad DeLong.

This is sad, because it tells us as much about economics as an academic discipline as it does about the individuals concerned. In the past I have imagined something similar happening in physics. It actually stretches the imagination to do so, but if it did, the academics concerned would immediately lose their academic reputation. The credibility of their work would be questioned.  Responding to evidence rather than ignoring it is what distinguishes real science from pseudo science, and doctors from snake oil salesmen.

What can economics as a discipline do about this sad state of affairs? The answer is pretty obvious, to economists in particular, and that is changing the incentives where we can. However we cannot do much about the incentives provided by politics and the media. I have been pretty pessimistic about this in the past, but in a future post I will try and be more positive and talk about one possible way forward. 

31 comments:

  1. On the pessimistic front:

    http://www.urban.org/publications/901502.html

    ReplyDelete
  2. It seems that behaviour like Mankiw' and Taylor's, which is not uncommon in economics, holds the discipline back significantly. Surely the natural sciences progress at a more rapid pace.

    ReplyDelete
  3. You are being naive.

    Physicists are not god: they cannot change the universe, merely seek to understand it. Economists do play god: they seek to change the economy, not just understand it. Physicists can manipulate nature. They cannot create it. Economists seek to do both.

    Of course, physics was also inherently ideological in a world in which it was taken for granted that some authority (Aristotle, the Bible or something else) had determined how nature works. But the last time that was true in Europe was the 17th century. Even in biology, it is possible to imagine a world in which evolution is simply accepted as true. But economics is always going to be ideological, because it is always going to be about how societies should be run. It is inherently ideological in a way that physics or biology is not.

    Economists need to accept this.

    ReplyDelete
    Replies
    1. My dear Martin. From my perspective, the primary thing that economists create is hubris.

      Physics manipulates the world indirectly. Through engineering. Physics seeks to understand how the world works while engineering takes that understanding and uses it to manipulate the world. There is a rather elegant but sometimes strained relationship between physics and engineering.

      Economists could learn something by examining it in a less than dismissive manner.

      Delete
    2. Economists may need to accept this, but they will also have to accept their irrelevance to the real world. To the general public, all that comes across from disputes like this is that economists disagree.

      Most people are not going to make the effort to decide which side of an economics controversy they are on, rather will tune out both sides. I suspect the paper that triggered Wren-Lewis' post has finished off the debate whether the stimulus worked: It didn't, because the economy is still weak and economists disagree on whether it worked. That's wrong, but it is what the voters will conclude.

      Delete
  4. You misspelled Ezra Klein. His friend Yglesias had a very good take on the matter: "An interesting piece of background here is that back in March 2009 it was the Obama administration saying we should expect rapid catch-up growth, and Mankiw arguing that this is an unfounded assumption. The Obama administration really has changed its tune on this "unit root" controversy, and it's interesting that Mankiw has also changed his mind. Normally the problem in intellectual life is that people hesitate to confess to error even when refuted by events. Changing your mind after having been apparently vindicated by events requires a staggering level of open-mindedness."

    http://www.slate.com/content/slate/blogs/moneybox.2.html

    And Mankiw was rewarded with the chairmanship of the Harvard econ department. So yeah it's a problem.

    ReplyDelete
  5. Yep, nicely put. It's a bad situation.

    ReplyDelete
  6. Economists: Experts at designing incentives for every industry but our own?

    ReplyDelete
    Replies
    1. +1000

      But I have hard time to see what you could do about it. Sometimes, when I hear politicians speak, I wish for more technocratic governance. However, it usually takes me about three seconds to remember that most social sciences have large minorities of mad men as well – and that those mad men often have a very oversized influence. Mankiw is a case in point.
      I really look forward to Simon Wren-Lewis next post.

      PS: Why isn’t “academic choice” a major subfield?
      http://www.nakedcapitalism.com/2011/04/blacklisted-economics-professor-found-dead-nc-publishes-his-last-letter.html

      Delete
  7. I have no idea why you and Prof Krugman persist, nay insist, on using the term "snake oil". If you would go to your own excellent library and search the terms snake oil and Echinaea you would see that the original salesmen were selling an Echinacea tonic that was effective for snakebites and other ailments. What happened is that colonial doctors decided that the evidence of the Echinacea tonic pointed to "no efficacy" when the evidence in fact pointed in the other direction. The majority of current studies suggest that Echinacea is effective, although there are always a minority of studies finding no statistical difference. There are very few studies on the oils of actual snakes but they also showed efficacy for the treatments they were used for. Therefore colonial/modern medicine was ‘intending to mislead’ by using the term snake oil to refer to something ineffective and they were also ‘misrepresenting the facts’. I know a little economics and have always kept up with it because I used to work in development and I knew that I would have to learn enough to argue with the economists. Development economists have to earn their keep in the "real world" so their theories tend to be better.

    ReplyDelete
    Replies
    1. They persist because in common usage today (and as far back in my time on Earth as I can remember) the term "snake oil" has meant exactly what they are using it represent. I have no idea if your explanation is correct as it is very obscure and very irrelevant. Whatever "snake oil" meant 250 years ago today it means a false cure, a slickly packaged poorly designed, probably harmful diversion from a useful medicine.

      So snake oil is a perfect substitute for the less polite but more accurate "damned lies"

      Delete
    2. Nope. If the original snake oil term was a deliberate defamation then continuing to use it continues the original defamation.

      Delete
  8. These four are ex-economists. To refer to them as economists is to perpetuate myth. Please desist.

    ReplyDelete
  9. I put together a little piece on my view on the Lesser Depression:

    http://socialmacro.blogspot.com/2012/08/the-balance-sheet-recession-hypothesis.html

    ReplyDelete
  10. Economics does not need Mankiv and Taylor to give it a bad name. The entire field is a wreck. You need only to read the books by David Colander to realize it.

    Economics is the religion of our age. One day, it will be relegated to the backwater of the university, like theology has been in the past, and nobody will be worse off for it.

    ReplyDelete
  11. Nick Rowe urges those I-am-not-an-economist contributors to economics blogs (particularly those commenters who promote heterodox views) to read a first year economics textbook.

    http://worthwhile.typepad.com/worthwhile_canadian_initi/2012/07/can-you-please-read-a-first-year-textbook.html

    But he won't recommend any.

    Given that Mankiw gives "economics a bad name" it's clear that his best selling (and very expensive) book is not worth reading.

    Do you recommend any introductory macro economics texts?

    ReplyDelete
    Replies
    1. 1. Olivier Blanchard ( IS-LM, expectaions)
      2. Carlin and Soskice (IS-LM, IS-PC-MR NK model)
      3. Gardner Ackley (2nd ed. 1978) (IS-LM, "classical model")

      Blanchard is an excellent introductory text.

      Carlin and Soskice is at a slightly more advanced, but still ug level, with bits of calculus. Also thicker and much more comprehensive than Blanchard. It also comes highly recommended from Olivier Blanchard himself. IIRC SW-L recommended it too sometime ago.

      Ackley is what Robert Gordon calls '78 macro. A synthesis of Old Keynesian, Monetarist, and Friedman & Phelps. Many people (eg Gordon, Krugman, Summers, DeLong) have argued that this sort of macro is best suited for the current crisis.

      Digression and rant:

      One issue I have with C&S and Blanchard is that they are too uncritical in regurgitating the std, but very dubious, case for the existing microfoundations and ratex.

      However, in their defence, they take external consistency and empirical evidence very seriously, unlike the RBC crowd. C&S, for example, in most of the book, drop rational expectations in the IS and Phillips curves on empirical grounds. Quoting from ch 15:
      "... we have chosen to prioritize the ability of the model to better match the stylized facts over the desirability of satisfying the requirements for fully rational and optimizing behaviour"

      This seems to me the way to go. Yet the authors feel it necessary to go on and develop a model with "fully rational and optimizing behaviour", admit it violently disagrees with data, and make many dubious modifications to the model to get it to disagree less. Why the authors feel this approach, rather than their earlier approach is better is something that leaves me mystified.

      End of digression and rant.

      Delete
    2. Actually in the very comments of that post Nick Rowe says:

      "I really like Mankiw too, for the same reason as you. (And my undergrad was analytic philosophy too!!) I was one of the co-authors on the Canadian edition of Mankiw. Left it a couple of editions ago."

      Yes Mankiw has gone off the rails, and seems lacking in intellectual honesty. But his textbooks are top notch, and completely unbiased. They are fine to learn from.

      Delete
    3. As an I-am-not-an-economist contributor with a human and mathematical background I would not recommend any macro text-book based on micro foundations which themselves are based on disproven macro assumptions, in particular human rationality (look up Jung's "shadow") and the applicability of Gibbsian matrices of partial differentials (look up Hamilton's quaternions). Those good at reasoning verbally tend to leave facts in the shadow, while a patchwork of flat maps will never reveal the fact that the earth is spherical and thus limited.

      Delete
  12. A boycott of Mankiw's textbook seems called for.

    ReplyDelete
    Replies
    1. Agreed. I teach micro, and currently use it, and would change if classes didn't start next Monday. But I will certainly change it next semester.

      Delete
  13. Yeah, don't enrich Mankiw.

    His textbook also falls prey to a tendency that's screwing up all college textbooks: ever more wordy, ever more empty margins. I'd say they're more or less getting worse with every edition.

    I like Bernanke's book. If you want to introduce modern macro concepts from the start, Williamson's book is the only one that actually has microfoundations.

    ReplyDelete
  14. Simon - Try to get your hands on the 'In Focus' Sunday Times report yesterday, on ethical malpractice amongst natural scientists. Falsification and fabrication of results are extensive and well-documented, much more so than one might assume. Obviously, it's depressing. But it might lift you insofar as I think you wrongly assume dishonesty is a problem peculiar to your discipline.

    ReplyDelete
  15. I had a long comment exchange with Mark Thoma on this topic in the past - around 2006-7.

    Carola's comment in this thread "Economists: Experts at designing incentives for every industry but our own?" is very apt - but I'd go further - what academic economists have is no concept of their profession. As such, there's no mechanism for reducing the influence of someone like Mankiw when they go bad. Further, the publishing regime is very skewed and people in influence get to play favourites in the US economics field to quite a large extent.

    ReplyDelete
  16. "what academic economists have is no concept of their profession."

    Could you expand on this?

    ReplyDelete
  17. Model-based evaluations of the Obama stimulus are fatally flawed as they essentially assume the answer (see Taylor here, from Noah's link: http://johnbtaylorsblog.blogspot.co.uk/2011/06/how-to-resolve-stimulus-debate-use-data.html)

    Econometric-based evaluations of the Obama stimulus cited on Ezra's blog almost all look at correlations between the variation on stimulus spending over states and output/job growth. If you seriously believe in Say's law (as most stimulus skeptics seem to) then you don't deny that the government can alter the composition of output. Rather you argue that the total level of output won't be any different. If the government spends more in Nevada than South Dakota, then you will see higher growth in the former than the latter. But if Say's law holds, then aggregate growth over the two will be unchanged.

    I'm *not* saying that this is necessarily correct. That's another debate. But these econometric studies definitely are not a test of a Say's law type argument. At best, they are a test of whether the government can move things around, which economists skepitcal of stimulus do not in general deny.

    The effects of fiscal policy, like a lot of stuff in macro, is hard to empirically test. There are some persuasive studies that it does matter (e.g. the IMF paper on the effect of fiscal consolidations). But Obama's stimulus is not a natural experiment, and so identifying the effect of it may not be possible.

    ReplyDelete
  18. I disagree with Martin Wolf in one respect that what makes physics a science is that the theory is confronted with the data. much of macro seems to be to be theology in that it is only weakly related to real world phenomena. Theory will progress once, as in chemistry, it is confronted by hard tests of the (micro) data
    So the salt water and fresh water views are wrong as they haven't confronted the muddy water of the data
    Danny Blanchflower

    ReplyDelete
    Replies
    1. It is hard to do controlled experiments in economics. Some people are trying to do so. But the situations are artificial.

      Delete
  19. This comment has been removed by the author.

    ReplyDelete
  20. In physics: look at String Theory. The string theorists survived for decades despite their theories making no new predictions and repeatedly being disproven by experiment.

    However, physics departments are finally correcting themselves, and string theory is being thrown by the wayside as new students look for theories which fit the real world.

    In economics, "professors" are reviving doctrines proven false 200 years ago. So yeah, it is a lot worse.

    I threw my old copy of Mankiw's textbook into the garbage when I started reading the garbage Mankiwu's been pushing lately. The edition I had wasn't a very good undergraduate textbook anyway; it didn't discuss deflation AT ALL. (Seriously?)

    As for micro, where's the discussion of advertising, creation of consumer preferences, price setters vs. price takers, and other such issues which are of prime importance to businesses?

    You might call it a "supply-side textbook". A boycott of Mankiw's *crappy* textbook is definitely called for.

    ReplyDelete
  21. thanks for posting.

    ReplyDelete