Rodrik: almost there, but not quite

Somehow this Project Syndicate piece from Dani Rodrik really irritated me. I think it’s because he’s half right  — he’s one of the few contemporary neoclassical development economists who I think bark up the right tree — but also scarily misleading.

Rodrik’s main point is that disagreement among economists is healthy; that it’s harmful for economists to coalesce around a narrow model of the economy. Rodrik rightly criticises the apparently near-unanimous agreement in a 2009 survey among ‘top’ economists (whoever decides who they are) on at least five policies: trade, rent controls, exchange rates, outsourcing and fiscal policy.

Rodrik says that trade policy shouldn’t always involve lower tariffs and an absence of quotas because some models have shown that in a situation of increasing returns to scale or externalities trade restrictions can raise welfare. There’s a huge neoclassical literature in this area, including famous papers by James Brander and by Paul Krugman, who won the Nobel prize for his work broadly on this stuff. Political economists like Ha-Joon Chang convincingly argue the case for infant-industry protection.

Rent controls, Rodrik says, are similarly debatable in the case of imperfect competition (which is always the case, in my view).

Exchange rate policy is also open to discussion and should differ according to circumstances. “[T]he proposition that floating exchange rates are an effective system relies on assumptions about the workings of the monetary and financial system that have proved problematic”.

No single model works anywhere. One size doesn’t fit all, which Rodrik has argued before and which I argue in my book.

So far, so good.

But when Rodrik describes economics as a “discipline [which] comprises a diverse collection of models, and that matching reality to model is an imperfect science with a lot of room for error” I think he goes badly wrong. Modelling, especially of the mathematical sort, isn’t always the route to the truth.

Loads of well-known economists (and some lesser names) have unearthed important concepts without models and have explicitly said that models shouldn’t be the only method going. Adam Smith, for example, wrote about the division of labour after physically observing the operation of the pin factory in Pathhead. He didn’t build a model. The whole Scottish tradition of political economy grew out of a combination of empirical observation and reasoning, combining the universal and the particular.

None of Keynes, Ricardo or Marx were ‘modellers’ as such, and Keynes famously rubbished Tinbergen’s attempt to apply econometrics to his theory. One of my favourite development economists, Albert Hirchsman, preferred to observe and draw conclusions rather than go to poor countries armed with a model. Michal Kalecki was similarly suspicious of generalisations. Chang doesn’t model but he’s really useful.

It’s certainly been my own experience that very few policy prescriptions or models are universally applicable in developing countries. People don’t all behave the same way; they tend to be influenced by different things (even that touchstone of all safety-conscious economists, ‘incentives’, means varying things in various places); and institutions and values vary. Modelling, with its inevitable reductionism and falsity of assumption simply isn’t the best way to understand other economies. This is one of the themes of a great book, Postmodernism, Economics and Knowledge, by Cullenberg, Amariglio and Ruccio.

So economics isn’t a discipline characterised by models. Even if most mainstream economists now perform modelling, the existence of a few non-modellers or economists who don’t fetishise the model proves Rodrik wrong.

It’s therefore despairingly narrow to imply that disagreement between advocates of various types of model is enough. Proper disagreement needs to take place on a much wider plane — and it has done so in the past, between competing schools of thought: Austrian, Marxist, Post-Keynesian, neoclassical, green, whatever. Real dialectic is necessary for genuinely new thinking.

I’ve always been dubious about the stale joke about the ambivalent economist which Rodrik prods into life like a seaside donkey halfway through his piece. President Eisenhower is supposed to have wished for a one-handed economist because they kept saying “on the one hand, on the other”. The joke’s not that funny, and in my experience mainstream economists tend to be rather unanimous in their worldview and policy prescriptions. If only they’d been a bit more two-handed in the run-up to the 2007/8 crisis.

[Update: here’s a slightly fresher quip from @elianalorch: Have you heard the joke about the economist? …Well, if it was funny, you would’ve heard it already.]

I just don’t think that economists try to “match reality to model”. They rigidly stick to an approach which says that modelling is the only way to discover things about economies. Many economists fight to prove their model correct even in the face of contrary evidence. Look at the so-called ‘debate’ between freshwater and saltwater economists in the United States. I can think of a few pubs in Edinburgh where that sort of bust-up would be sniggered at as a lovers’ tiff.

To describe mainstream economics as an “imperfect science with a lot of room for error” suggests that we’re almost on the right track but we need to tweak things. How can economics be almost correct when most of its prominent practitioners so patently failed to predict the crisis and when the discipline remains so far from useful in many important situations, such as in the developing — ie. majority — world?

I think the Rodrik piece is so irksome because he’s the upstart, the leftfield critic wheeled out to satisfy the doubters. In reality his view is quite conventional and serves to delude the non-specialist that the discipline is vigorously searching for answers. I don’t want to criticise Rodrik too much because he sounds like a good bloke, but something’s wrong when the claims of an ersatz radical are broadcast as leftfield to such a wide audience.