Evan Soltas
May 8, 2012

Towards a New Industrial Policy

Yale economist Chris Blattman has a thought-provoking post on the need for more academic consideration of industrial policy in development economics:

[I]ndustrialization is the most important and essential process of development. Everything from lower poverty, reduced inequality, and tax bases to support education and health and welfare systems will (and must) spring from high value-added production. Anything policymakers can do to hasten the process will have unparalleled benefits. The problem? We have little to no idea how to do that...[but] the fact that we know so little, and the tools are so poor, suggests (to me) that the marginal gains from more research are huge. there is no more important place for scholars to spend their time...What is the next frontier? I would put money on industrial development and, with it, a new breed of industrial policy.
I agree with Blattman in that development economics needs more and better tools, and that its toolkit has disappointed where it has been put into practice in the real world. Meanwhile in Asia, there seems to be a refined version of "industrial policy" toolkit which has brought development, if not success, to South Korea and Taiwan, to name two examples which are not really in dispute.

His is not an endorsement of industrial policy. It is a call for further investigation, as there is clearly something to learn here in the development successes of East Asia, which have striking connections to other development stories.

This post, therefore, is a survey of successes of such "new industrial policies." What I mean by "new" is that I chose in particular to examine accidental industrial policy, or circumstances under which the planning mechanism was weak -- i.e. not Soviet -- but there was still a conscious organizational effort at the regional or city level.

That said, I want to note the limitations of such a historical (i.e. anecdotal) rather than data-driven and scientific approach. It is the species of faulty logic seen in CEO autobiographies to assume that because policy A or circumstances B existed in successful examples X, Y, and Z, then therefore A and B are good industrial policies. Rather, the purpose of such a survey is to do a "first-pass" identification of similarities in the historical record, and from this we may derive the broad outlines of what determines the success of such "new industrial policy."

I was inspired by Jane Jacobs' famous consideration of Detroit, Michigan. Jacobs, examining the history of Detroit's shipbuilding industry in the 19th century, connected that to its subsequent development as a center for automotive design and manufacturing. Jacobs posited that when firms in different industries operated in close proximity, then there was a knowledge spillover externality as firms drew on a common and diversified pool of ideas in innovating their products. (This is different from the MAR externality, which suggests that spillovers exist within the same industry because of such sharing of ideas. See here for a good summary; I can't find Jacobs' original work online.) This was very much an accidental industrial policy, encouraged weakly by local organization -- not applied from on high as some sort of five-year plan, but the idea that this pattern of development, this agglomeration of related but distinct industries, happened by accident strikes me as unlikely.

The five examples I could think of where I saw successful industrial policy all seemed to work from Jacobs externalities and had strikingly similar stories: (1) South Korea in the 1970s and 1980s, (2) Stanford and Silicon Valley in the 1950s, (3) the emergence of the Parisian suburb La Défense in the 1960s, and (4) New Jersey in the 1960s.

South Korea's history of economic development repeats exactly Jane Jacobs' story of turn-of-the-century Detroit -- and surprisingly enough, although there were some studies which examined agglomeration externalities in Korea, I couldn't find any which pointed out this most obvious fact. What was the first industry encouraged by government policy? Shipbuilding. The country gained a major competitive advantage in supertankers and the sort of large cargoships which are needed for high volumes of international trade. The first docks went up in the late 1970s and early 1980s -- by 2008, South Korea had over a 50 percent share in the global shipbuilding industry. You know the companies' names, but not as shipbuilders: Hyundai, Samsung, Daewoo. And South Korea's next industry? Automobiles, as you most likely know Hyundai/Kia. Daewoo, similarly, makes buses and construction vehicles. The shipbuilding-to-automobiles transition is a story South Korea shares with Detroit. Samsung, too, reapplied its industrial expertise, in particular from the interior wiring of ships, in consumer electronics -- their first electronics product was a black-and-white TV set for the domestic market, and then telephone switchboards. Most of the technology for the first decade appears to be inspired from what they put in ships; now, of course, Samsung's biggest products are LCD televisions and smartphones. South Korea's industrial development is a history of the systematic exploitation of Jacobs externalities.

The story of Stanford University and its role in the development of Silicon Valley hits many of the same notes: the initial formation of a cluster of firms in distinct industries but whose technologies were transferrable from one to the other. After World War II, Stanford's Dean of the School of Engineering, Frederick Terman (later provost), cooperated closely with the Dept. of Defense and his graduates to organize firms based outside of Stanford -- the results were Varian Associates, Hewlett-Packard, and others in the "Stanford Industrial Park." (Fittingly, that had been until recently the headquarters of Facebook.) But in the interim, Terman's informal industrial planning brought William Shockley, the inventor of the transistor, to the Park. By the 1960s, Terman had assembled the collection of firms who were producing transistors, vacuum tubes, and semiconductors -- but it wasn't until the 1970s and 1980s, with the formation of Kleiner Perkins and the IPO of Apple Computer that the success of Terman's industrial policy caught the attention of the world.

The development of the Parisian suburb La Défense demonstrates the same successful traits of "new industrial policies" in Stanford. Today, it is a wealthy and highly commercial area, filled with the corporate headquarters of French corporations. Yet before it was as we know it today, La Défense began as an industrial park organized by the national government, which established EPAD, La Défense's industrial planning agency, in 1958. In that same year, the de Gaulle government built a major convention center, the CNIT (Center for New Industries and Technologies, pictured) to serve as the focal point for regional development, much as did Stanford serve for Silicon Valley. After this the French government worked to relocate a waves of corporations from the metropolitan center, green-lighting the construction of skyscrapers, which Paris did not see elsewhere.

Which area of the world has the most scientists and engineers per square mile? The answer is New Jersey -- my home state, incidentally -- which took advantage of its fortuitous location at the center of the East Coast of the United States and also a residential and corporate exodus from New York City during the 1960s and 1970s. Benefiting from the massive container port infrastructure and a preexisting heavy-industrial base, New Jersey effectively became one large high-tech industrial cluster. Bell Labs, located in Murray Hill, NJ, was an industrial park of a different order of magnitude, fostering the development of telecommunications industry and much, much more. After that, New Jersey brought in pharmaceuticals and chemicals. The presence of a shared high-tech, high-skill labor pool and the proximity of major firms -- there are large stretches of highway in my state where there is one corporate headquarters or office park after another -- and the regional success in these fields strikes me as no accident.

The running theme I found in all of these was that Jacobs externalities appear to be highly determinative of the success of high-tech industry at the regional level, and that any "new industrial policy" going forward should explore the potential gains from coordination which stem from these spillovers.