Dani Rodrik's latest book--One Economics, Many Recipes--is one of the best I've read recently. The nature of book-binding fits manuscripts to a certain page rage, regardless of the actual length needs of the discussion. Combined with poor writing and editing, a majority of books contain a high fat/content ratio. I've wondered about the empirical distribution of that ratio and the theoretical minimum. On the basis of no analysis, I conclude that the fat ratio follows a gamma distribution, with a mean at around 60% fat tailing to a 15% minimum. I imagine 20% fat is around optimal; The Economist has somewhat more fat than this but is one of the leanest publications around. Fat-avoidance has shaped my shift from consuming books to size-constrained blogs and articles.
Rodrik's book is written as a series of self-contained articles, greatly lowering the fat count as each chapter makes stand alone claims. These are grouped into three sets, the third of which, on global institutional frameworks and similar esoterica, I will ignore.
The first set of arguments concern the failure of successive iterations of development policy. Rodrick concludes that a basic framework--proper competitive incentives, a good macroeconomic situation, globalized outlook, and social insurance--is common to all successful economies, but that particular institutional designs vary. Instead of the Washington Consensus transplant approach of grafting overseas best-practices over every local organ, Rodrik adopts an "optimize what you optimize" mentality; use diagnostics to identify the functional hurdles preventing the rise of the productive sector of the economy, and improvise local solutions. This is great stuff, worth more than the cumulative total of all economic randomized trials ever conducted.
The second bit covers industrial policy. As Tyler Cowen notes, the substance of his proposals is actually fairly limited, while the rhetoric of railing against fundamentalist economists is overblown and against the humble spirit of the book. This is because Rodrik has a strong agenda of normalizing industrial policy as another piece of the government tool-box.
Rodrik presents a sophisticated case for viable government interventions in a complicated world, but I am skeptical. The historical examples rely heavily on East Asian examples. It's fair to say that the debate over industrial policy is basically a debate over government practices between 1950-1970 in East Asian economies. Of course, in this period virtually every other developing country attempted industrial policy in some form or another and all of them failed. The key counterfactual is to what degree the peninsulas and islands of Asia would have prospered without industrial policy, and the pressing policy question is to what degree other developing countries can replicate such planning institutions.
I suspect that the success of Asian economies is due more to government success in human and physical capital, oppenness to foreign trade, and competitive markets rather than successful industrial policy; and that more pluralistic, democratic countries will find it difficult to replicate their institutional structures (these countries find it hard enough to deliver basic government goods) . Here are some critical assessments of Eas Asian industrial policy. In several cases, the bulk of government subsidy was not directed at the most productive sectors but rather the more politically sensitive ones. So in Taiwan you have government subsidized heavy industry while the small and medium scale firms did the bulk of the production, and in Japan you see large subsidies for rural development and the postal banks. Korean policy was perhaps the most extensive, but some people disagree that it was effective relative to doing nothing.
It's also worth pointing out that all East Asian post-WWII successes were ethnically homogeneous, and were either city-states with strong British influence (Singapore, Hong Kong) or former Japanese colonies (Korea, Taiwan), giving these countries strong legacy institutions.
Rodrik also commends China and India's gradualism. But, given their large disparity in income relative to developed countries, convergence was inevitable and the question is how much did gradualism help economically, even if it was great politically. In China, much of the federal outlay was directed at Soviet-style large industry, while the small and medium size firms that drove the economy were more independent (though they maintained close ties to local Party officials). Indian industrial policy was absolutely catastrophic, and success in biotech and information technology is due greatly to the fact that these sectors did not exist when regulations were drawn up. The scope of Asian markets will continue to attract business regardless of failure or success of government policy.
I do see the potential for market failure, and Rodrik details some plausible scenarios. Much like how the path of immigrant migrations follows the intitial migrants rather than the fundamentals, corporate development tends to follow the path set by the first-movers. But the potential for government failure is also strong, and I am skeptical of the degree to which future governments can target the money properly. Broad-based, universally accessible policy such as research, investments in education and infrastructure, and a pro-growth, low tax environment lift all boats, but is perhaps better discussed under the label of "good governance" than "industrial policy."