Rodrik has written a characteristically provocative post:

Ignore the intermediate period of debt crisis (1975-90) and focus on the differences between the periods of import-substitution (IS, 1950-1975) and Washington Consensus (WC, 1990-2005). What do we see?
Labor productivity under IS grew at a rate that is double the rate under WC.The rate of labor productivity growth within sectors (the component identified as “within” in the chart) was comparable under the two policy regimes.The worse overall performance under WC is accounted entirely by the fact that there was much less desirable structural change — labor moving from low to high-productivity activities — under WC than under IS.
We already knew 1, and I have long suspected points 2 and 3 to be true as well. But I had never seen the facts laid out so clearly.

First, I wonder if it is fair to set aside 1975-90 as an anomaly. I’m reminded of William Easterly’s brilliant review of Ha-Joon Chang’s Bad Samaritans. We could be seeing a bad case of confirmation bias.

But Rodrik does raise an interesting and important point:

For all its faults, IS promoted rapid structural change. Labor moved from agriculture to industry, and within industry from lower-productivity activities to higher-productivity ones. So much for the inherent inefficiency of IS policies!
Under WC, firms and industries were able to accomplish a comparable rate of productivity growth, but they did so by shedding (rather than hiring) labor. The displaced labor went not to higher-productivity activities, but to less productive lines of work such as informality and various services. In other words, the WC ended up promoting the wrong kind of structural change.

Focusing only on manufacturing productivity, Rodrik explains, can obscure broader patterns that are ultimately more important. This is clearly true. But I wonder if it’s fair to blame the Washington Consensus policy mix — one wonders how many of the Latin American countries surveyed saw the Walmartization of their retail sectors during the WC.

Part of the problem is that countries like Brazil, memorably described by Brazilian economist Edmar Bacha as “Belindia,” or a mix of Belgium and India, are awkward units for purposes of economic planning. For example, the bolsa familia transfers are barely adequate for sustaining incomes in affluent regions while they are high enough to make nonemployment in high-poverty regions like the Northeast a viable option. And I think that’s what Rodrik is driving at — he believes that you need industrial policy to actively transform these “backward” regions.

I’m not quite sure what to think. I have very little faith in bureaucrats, yet the experience of caste oppression in the segregated South suggests that there are circumstances in which certain patterns have to be disrupted through external shocks.