It's time for a third phase of development thinking focused on structural change, driven by changes in endowment structure and comparative advantages. The market will be the fundamental institution for resource allocation and the state would play a proactive facilitating role in the process. I make this case because, in two earlier waves of development economics had mixed records. The first emerged after World War II with a focus on market failures and an embrace of traditional structuralist, state-led development policies; the second adopted a largely neo-liberal view that targeted government failures and recommended Washington Consensus-types of policies.
I lay out this argument in the most recent issue of the World Bank Research Observer  (subscription required), which synthesizes half a century of various approaches proposed by development economics, and suggested a way forward. My WBRO paper, New Structural Economics: A Framework for Rethinking Development, is critically discussed in the same issue of the journal by Joe Stiglitz, Anne Krueger, and Dani Rodrik
While Krueger, Stiglitz, Dani and I all agree on the need to reignite the debate on development recipes — especially in light of the current global financial and economic crisis — we also part ways in several important respects.
Anne argues that "Only later in the development process does upgrading become a major part of industrial growth once there has been significant absorption of rural labor —much of it happening in existing firms in response to rising real wages, lower capital costs, and learning through exposure to the international market" (p. 223). My view is that the migration of unskilled rural labor to unskilled labor-intensive industries may not occur spontaneously and needs to be nudged along, for example with the government facilitating the emergence of new unskilled labor-intensive industries.
While Anne agrees that "the market should be used to determine comparative advantage, and that governments have responsibilities for insuring an appropriate incentive framework and provision of infrastructure, both hard and soft," she objects to government intervention aimed at fostering the development of specific industries.
In fact, identification of new industries and prioritization of government's limited resources to facilitate the development of those industries are both essential for successful growth strategies in developing countries. Why? The required infrastructure improvements are often industry-specific. One has to simply look at the list of recent success stories in African countries to understand: textile in Mauritius, apparel in Lesotho, cotton in Burkina Faso, cut-flowers in Ethiopia, mango in Mali or gorilla tourism in Rwanda all required that governments provide different types of infrastructure.
Identification of new industries and prioritization of infrastructure investment are necessary also because, for a new industry to be competitive in the globalized world, not only must the industry align with the country's comparative advantage so that its factor costs of production can be at the lowest possible level but also the industry needs to have the lowest possible transaction related costs, which depend on the improvement of hard and soft infrastructure and the formation of clusters. Without the government's identification and facilitation, the formation of clusters by trial and error is likely to be a long and costly process.
I agree with Anne that cost-benefit analysis is an excellent tool for evaluating the potential merits of every single infrastructure project (p. 223). Such analysis forces policymakers to provide quantitative data to back up qualitative arguments and is therefore an invaluable technique for increasing social welfare. But it is microeconomic by nature. Without the identification of potential industries, their likely location and needed infrastructure, there are just too many possible projects to be feasible for careful analysis. Moreover, for every public investment project, there are many benefits and costs which are intangible and therefore difficult to value. It is also well known that the results of that analysis can be very sensitive to the choice of the discount rate, and that the information used to determine future benefits and costs are limited by current knowledge.
Anne's skepticism of any industry specific interventions results from pervasive past failures by governments to pick winners. Those failures were mostly due to the misguided attempts by many governments to develop industries that were inconsistent with their countries' comparative advantages.
Anne also worries that identification of any new industry for upgrading "would inevitably favor larger, established firms and hence encounter the same sorts of problems as did the older import-substitution strategy" (pp. 224-5). Her worry is valid for the old structuralist import-substitution strategy, because those industries went against the country's comparative advantages, were too capital intensive and thus only a few rich and politically well connected firms could enter. However, if the identified new industries are consistent with the country's comparative advantages — capital intensive or not — many firms will be able to enter and contest the dominance of large firms as exemplified by the auto industry in Japan in the 1960s, the textile industry in Mauritius and electronics in Taiwan, China in the 1970s, and the garment in Bangladesh and the salmon-farming in Chile in the 1980s.
Finally, Anne questions the uncertainty surrounding the scope, depth and length of government protection and points out the risk of political capture and rent-seeking if a government adopts a dual-track approach in its transition from a heavily distorted economy to a well-functioning market economy. She argues that "a major challenge for liberalizing reform is for it to be credible that the altered policies are not reversible. Lin's prescription would greatly increase the challenge of creating credibility, and a slower transition would be a longer period during which growth was slow and political pressures opposing liberalization at all were mounting" (p. 225). The credibility argument was used to support the shock therapy in the East European and Formal Soviet Union's transition in the early 1990s. However, to ward off large unemployment and subsequent social/political instability, governments in transition economies were very often forced to provide other disguised and less efficient forms of subsidies and protection to firms in the old priority sectors even though those firms were privatized. As a result most transition economies encountered the awkward situation of "shock without therapy".
Joe Stiglitz and Dani Rodrik diverge from me more in terms of emphasis and style than substance.
Beyond the traditional need for regulation, Joe sees a catalytic role for governments "in promoting entrepreneurship, providing the social and physical infrastructure, ensuring access to education and finance, and supporting technology and innovation" (p. 231). He challenges the view of the efficiency and stability of unfettered markets, and stresses the need for advances in technology as a key condition for increases in per capita income. Consequently, he favors public action to create a "learning society" (p. 231).
I concur with Joe on the importance of learning as a means of climbing the ladder toward industrial development. However, countries that are still at the early phase of their development generally do not have not only the necessary human or physical capital to leapfrog into capital-intensive, high-tech industries. The more effective route for their learning and development is to exploit the advantages of backwardness and upgrade and diversify to new industries according to the changing comparative advantages determined by their changing endowment structure. In other words, the absorption of knowledge and the enhancement of human capital should be commensurate with the level of economic development.
Joe may be a bit too optimistic when in suggesting that full capital mobility in a globalized world allows countries to free themselves from patterns dictated by endowments, as conventionally defined. He postulates that "With fully mobile capital, outside of agriculture, natural resource endowments need not provide the basis for explaining patterns of production and specialization" (p. 232). However, short-term capital flows are too volatile to be a reliable source for long-term productive investments in developing countries, as shown in East Asian financial crisis in the late 1990s. The more reliable FDI is motivated by investors' seeking of profits. They mostly go to tradable sectors or production activities that are consistent with a host country's comparative advantage. An exception may be when they are driven by occasional cases of privatization of large non-tradable sectors such utilities and telecommunications.
Dani assumes that coordination and externality issues only exist in situations where markets send entrepreneurs the wrong signals. He therefore suggests that I may be arguing "both for and against comparative advantage." (p. 228). Let me clarify: Comparative advantage is determined by factor endowment. If an industry is consistent with a country's comparative advantage, the factor cost of production will be lower than otherwise. But for that industry to be competitive in its domestic and international market, transaction-related costs should also be reduced to their lowest possible level. Yet individual firms cannot internalize the reduction of many of the transaction-related costs arising from issues such as infrastructure provision, logistics, finance, educated labor and so forth. That's why government coordination and facilitation are vital.
The differences between Dani's and my understanding of the government's role arise to a large extent from our different interpretations of experiences in successful countries in Japan, Republic of Korea and China. He regards the successful catching up in Japan and Korea as evidence for the need to defy a country's comparative advantages (p. 228). In my debate with Hajoon Chang, published in Development Policy Review in 2009 , I have pointed out that the quick industrial upgrading in Korea and other East Asian success stories were in fact consistent with the changes in their endowment structures.
Dani questions the differences between my recommendation for gradual trade liberalization and the old structuralist policies. The latter approach advocates protection and subsidies to build new industries that were not the country's comparative advantages, whereas the dual-track gradual approach for trade liberalization advises the government in transition economy to provide transitory protection/subsidies to old industries which were not viable in an open, competitive market but were established under the misguided old structuralist strategy. The pragmatic dual-track approach helps a transition economy avoid unnecessary and costly economic and social disruption, and eventually leads to market based prices and resource allocation.
Overall, Anne's questions about the practicality of my framework arise mostly from the issues of how to identify new industries that are the country's latent comparative advantages and how to administrate the coordination and incentives for the first movers (p. 223). Joe's and Dani's advocacy of broad based intervention such as undervalued real exchange rate to support tradable sectors but reluctance to embrace the idea of sector-specific policies are also related to how to identify industries where countries enjoy latent comparative advantages. The question has been addressed in a companion paper entitled "Growth Identification and Facilitation ", co-authored with Célestin Monga and published in Development Policy Review.
You can read my full rejoinder to Kreuger, Stiglitz and Rodrik on my website .