I come to bury industrial policy, not to praise it (pt II—Rodrik)

[continuing from previous post "I come to bury industrial policy, not to praise it (pt I)"]

... Now on to Dani Rodrik's contribution to the 2010 The Economist debate concerning industrial policy. Rodrik's role was to argue against the following motion:

"This house believes that industrial policy always fails." 

Rodrik begins his argument by pointing out—correctly—that there is, in a technical sense, no way he can lose. "... always fails"? Give me a break. How could an idea tried repeatedly over the period of decades in much of the world have "always" failed? All that's required to win the debate is to find a single success. I pointed out one in my previous post. So he's won the debate before he even gets started.

But then, unfortunately for his case, Rodrik goes further.
He builds his argument bit by bit, reaching a crescendo at the end with a statement that comes perilously close to substituting "always fails" with "always can work":

Fostering structural transformation and innovation is a central public purpose. Governments cannot evade the challenge. The only debatable question about industrial policy is not "whether" but "how."

Now, even though I'm a third of the way through part II of a blog post on "industrial policy," it's not too late to share a bit of elaboration, in Rodrik's own words, about what that term means (at least, to him):

The conventional approach to industrial policy consists of enumerating technological and other externalities and then targeting policy interventions on these market failures. The discussion then revolves around the administrative and fiscal feasibility of these policy interventions, their informational requirements, their political-economy consequences, and so on. We start also from generic market failures, but then we take it as a given that the location and magnitude of these market failures is highly uncertain.

... The task of industrial policy is as much about eliciting information from the private sector on significant externalities and their remedies as it is about implementing appropriate policies. The right model for industrial policy is [one] of strategic collaboration between the private sector and the government with the aim of uncovering where the most significant obstacles to restructuring lie and what type of interventions are most likely to remove them. Correspondingly, the analysis of industrial policy needs to focus not on the policy outcomes—which are inherently unknowable ex ante—but on getting the policy process right. (pp. 2-3)

Rodrik focuses on specific "market failures" that he assert is of particular significance in early- to mid-stages of national economic development: “information externalities entailed in discovering the cost structure of an economy, and coordination externalities in the presence of scale economies.” (p. 5) In other words, a key obstacle to development according to Rodrik is the fact that entrepreneurs systematically have inadequate incentives to search for potentially profitable, but untried, economic activities. The reason is entrepreneurs who successfully engage in such search will soon find their new ventures copies by later entrants, dissipating the quasi-rents that they have earned. As a consequence of this appropriability problem—more or less exactly analogous to that which is generally characteristic of the search for new ideas—the effort devoted to the search for fundamentally new economic activities will be inadequate to get under-developed economies out of a low-diversification, low-income, and low-growth trap.

The way out of that trap—again, according to Rodrik—is industrial policy. What this means is, as Rodik emphasizes repeatedly in the paper just cited and as he has written over and over since, is undertaking a process by which government seeks information from industry, then uses policy mechanisms at its disposal to catalyze entrepreneurial activity in the generation of new, profitable products. Once the new products are part of the economy, and early entrants are suitably rewarded to ensure (second-best) dynamic efficiency, copy-cats will enter and a new, sustainable industrial activity will be added to the country’s portfolio.

Of course, the devil in getting the process right is most assuredly in the details. A brilliantly argued paper written in 1995 describing the successful industrial policies implemented by the Korean government a half-century ago has this to say about the conditions that were required for success in that environment:

… [W]hat was required was a competent, honest and efficient bureaucracy to administer the interventions, and a clear-sighted political leadership that consistently placed high priority on economic performance [along with] an exceptionally high degree of equality in income and wealth—wealth distribution played an important role in shaping the political landscape in both countries. This is probably the single most important reason why extensive government intervention could be carried out effectively, without giving rise to rampant rent seeking.

This paragraph is, in my view, among the most important ones written on the topic of industrial policy—one that rather seriously calls into question generalizability of the Korean experience. Who authored this massive caveat to the industrial policy paeans that followed? Gene Grossman, Victor Norman, and (actually first among them), Dani Rodrik.

I agree with this paper of Rodrik's. He and his co-authors were absolutely correct about the exceptional conditions required to make industrial policy work. Even if the particular characteristics that held for Korea in the 1960s and 1970s also held in Taiwan at the same time, or for Chile in another era, or ... well, it doesn't matter much: one can say with great confidence that they do not hold much of anywhere else—particularly not in the places in which the process of development remains stuck in low gear. Most persistently poor places do not possess a competent, honest and efficient bureaucracy. Most persistently poor places do not possess a clear-sighted political leadership that consistently places high priority on economic performance. Most persistently poor places do not possess an exceptionally high degree of equality in income and wealth. In fact, there are real questions as to where these conditions hold in middle-income and wealthy countries either. As a consequence, there is little reason to believe that a strategy that was effective in Korea in the 1960s is going to work in most other places.

Now, it is worth pointing out along the way that at least three other objections can be made with respect to the framework advanced by Rodrik:

• Limits to appropriability aren’t the primary problem faced by entrepreneurs
• Cluster-building strategies have been tried and mostly haven't worked
• Imitation of successful entrepreneurial strategies is not easy

Elaboration of each of these points would take another three blog posts. However I can offer a sketch of an argument that could be advanced in a more complete critique.

Limits to appropriability aren’t the problem. Limits to appropriability from innovation are not the fundamental impediments to entrepreneurial initiative in the context of technology-based innovation. A parallel argument can be made about the search for new economic activities in an under-developed country.
Cluster-building strategies have been tried and mostly don’t work: Cluster building strategies are mostly closely associated with Harvard's Michael Porter. While Porter's work on clusters has received a great deal of attention, it has the deficiency that, to put it bluntly, it doesn’t seem to work.

Imitation of successful entrepreneurial strategies is not easy. A key premise of the argument in Rodrik is that imitation of successful entrepreneurial strategies is easy. Implicit in this view is the notion that new-to-the-world technological innovations are relatively difficult to copy, but that the sort of relatively simple, new-to-a-particular-region economic activities that would generate new products in a developing country are relatively simple. This is a very questionable claim. What is likely true is that low efficiency, low-effectiveness approaches are easily copied to producing a new product are likely to be easily copies in a developing country—just as they are in a developed country. However, in both settings, there is every reason to believe that the underlying problem solved by an entrepreneur will be a complex one, and consequently that any solution found to the problem will not be easily copied. If this is true, then the implied policy strategy is very different: rather than seek to build product-based clusters, focus on enhancing the management capabilities of firm.

Overall assessment: While Rodrik wins the debate on a technicality, the argument he advances is so overstated that it's an effective draw.