Friday, August 19, 2016

Noah Smith on heterodox models

As it is often the case when you've been blogging for a while, there is always a precedent, and one might have written about a particular topic. Noah Smith, which I think has been blogging for a shorter period of time than I've been, now comes up with another take down of heterodox economics (having difficulties of understanding the meaning of the mainstream, it's no wonder he gets heterodoxy wrong; on the definition of the latter go here).

He argues that "much of heterodox theory is non-quantitative." True, but so is the case with the mainstream. Not everything can be formalized. And there is qualitative knowledge. But, having said that, the point is that difference between the mainstream (marginalism, neoclassical economics) and heterodox approaches has nothing to do with the lack of formalization of the latter. His examples of heterodoxy are not the best, I would argue. He first deals with Hyman Minsky, which in my view accepted a good chunk of mainstream ideas in his interpretation of Keynes. And yes, he did not formalize his theory, but there are plenty of models (Lance Taylor, Alessandro Vercelli and Steve Keen have done it, to name a few).

The main critique of mainstream economics, or at least of its core theory of value and distribution, is presented in a very short, and formalized, little book called The Production of Commodities by Means of Commodities. The notion that supply and demand determine prices (long term equilibrium prices) and that the principle of substitution works is sketched there. And later Paul Samuelson (subscription required) admitted that neoclassical parables made little sense. I've dealt with this issues several times notably here. And the heterodoxy has formalized models of the determination of output, employment, inflation (see mine here), growth (in many varieties, but mainly demand-led) currency crises (see mine here), etc. So Noah is wrong when he suggests that "heterodox theory is non-quantitative."

Noah discusses essentially the so-called stock-flow models deriving from Wynne Godley's (my old mentor) work as examples of heterodox models that are formalized. His main critique is that there are problems with estimation of parameters. This is essentially one of the critiques raised against the Cowles Commission types of model that dominated the mainstream until the 1970s (before Lucas' critique). They are still raise about Ray Fair's model at Cowles. Fair provides a response and comparison with modern methodology here, which is worth reading.  But the defense that Wynne would provide would be different in my view.

As I noted before, Wynne "was more concerned with what he referred to as 'model architecture' than with parameter estimation. The architecture, which was careful about stock-flow consistency, showing that everything came from somewhere and went somewhere so to speak, also imposed a clear causality structure, which determined most of the results. In fact, Wynne believed that significant variations of the parameters might not greatly influence the end result of the model, which was used for simulations and scenarios that helped to understand how the economy functioned, rather than for strictly forecasting purposes."

Simpler models, which do not provide the full accounting, as I suggested here, but that separate income and price determination, where output is demand determined, are in that sense also useful. Or just stated simply, of course there is a formal alternative to the mainstream. It's one that does not emphasize microfoundations and individual behavior as Noah would like, but that's another story.


  1. Você não adicionou os links dos seus modelos no final do 3° parágrafo, Matías.

  2. My response to Noah Smith's blog post: