Sunday, October 4, 2009

Alex Rosenberg debates economist David Levine

It's not often that you see a scientist (especially an economist) seriously engaging with philosophers of science (especially philosophers of economics). So I really appreciated this debate:

Alex Rosenberg's analysis is as always sharp and stimulating. Many thanks to him for representing philosophy of science!

In my view, however, Levine has done a good job using empirical examples from experimental and behavioral economics to counter the skeptical claim Rosenberg makes about economics - that it has failed to improve the quality of its predictions. Eventually the debate seems to shiftto whether or not Levine represents mainstream economics. I think it's hard to deny that he does. So as much as I agree with Rosenberg's skeptical analysis of neoclassical micro- and macroeconomics, it no longer seems viable to extend it to the discipline as a whole.


  1. Anna, I also enjoyed the conversation. I found Levine a little cagey at some important points, but then, I'm not that well aware of economics or the philosophy thereof. One question that I didn't get clear on, though, and which Rosenberg seemed to be pressing successfully, was whether there had been much uptake of experimental and behavioral economics work in central work on macroeconomics. That is, when we're interested in things like the financial crisis, how much do the main macroeconomic models incorporate work like Levine's?

  2. Okay, I am going to watch the whole show. But let's make some distinctions. Micro and macro neo-classical economics are very different beasts and have very different empirical implications.
    The understanding of the economics of rent-control (and associated predictions) is far more solid than macro-economic or stock-market forecasting.

  3. To Matthew's question: experimental and behavioral research can generally only generate local 'engineering' solutions. (This is the view of Vernon Smith, a Nobel laureate who pioneered research in the area.) So, it is a bit strange to expect that they can be incorporated in macro-models. (I know that some 'nudge' theorists think they can have macro policy implications, but that is a different question.)

  4. The main problem with the exchange is that there is a conflation between economics as a professional research activity (Levine) and economics as a policy science with policy implications (why did it fail us?). The second is really what bothers people, but answering that requires us not to merely look at how economics is researched and taught, but how it is sold to and bought by policy-makers and the public.

    A propos earlier point (Matthew's question): One point Levine makes is that behavioral (micro) models will have little impact on (macro) business cycle models. (The stuff washes out.)

    Levine makes a compelling case that economic models are very sophisticated (and nothing like the carricature that is often portrayed by economists' critics) and often empirically very well confirmed. But Levine ignores something important. It is way too easy to *test* economic models. What is *hard* to figure out is what influences parameter changes. (This is one reason why current behavioral only leads to local knowledge.)