Effects of Industrial Policy on Productivity: The case of import quota removal during postwar Japan
Kozo KIYOTA (Keio University and RIETI) and Tetsuji OKAZAKI (University of Tokyo and RIETI)
Abstract This paper attempts to provide a systematic analysis on the effects of industrial policy in postwar Japan. Among the various types of Japanese industrial policy, this paper focuses on the removal of de facto import quotas through the foreign exchange allocation system. Analyzing a panel of 100 Japanese manufacturing industries in the 1960s, we find that the effects of the quota removal on productivity were limited—the effects were significantly positive, but time was required before they appeared. On the other hand, the effects of tariffs on labor productivity were negative although insignificant. One possible reason for this is that the Japanese government increased tariff rates before removing the import quotas and maintained high tariff rates afterward. As a result, the effects of the Japanese industrial policy in the 1960s might be smaller than widely believed in the Japanese economic history literature.
Reviewed by Sebastian Fleitas
“I haven’t got anything against open competition. If they can build a better car and sell it for less money, let ‘em do it. But what burns me up is that I can’t go into Japan. We can’t build, we can’t sell, we can’t service, we can’t do a damn thing over there … I think this country ought to have the guts to stand up to unfair competition” Henry Ford II (1969)
People used to say that a miracle happened in Japan during the sixties. By 1960, the Gross Domestic Product per capita (GDPpc) of the US was 2.8 times that of Japan. In the same year, the GDP per capita of Chile was the same of the Japanese while Argentinian was 40% higher. One decade later the situation had dramatically changed. By 1970, US GDPpc was only 1.5 times greater than the Japanese. In addition, Japan GDP pc was 85% higher than the Chilean and 33% higher compared to the Argentinian. While comparison of GDPpc actually raise more questions than answers, the comparison with these Latin-American countries can be appealing because Japan and these countries had very aggressive currency controls and industrial policies during this period. The difference of results makes us think that Japan must have done something different, something better. To find these differences it is needed to evaluate separately the effects of each of the policies applied during those times, understanding the incentives that they provided to the firms. As Lars Peter Hansen – recent Nobel Prize in Economics- suggested, one key important thing in Economics is that we can do something without doing everything.
This paper, circulated in NEP-HIS 2013-11-09, focuses on the removal of de facto import quotas through the foreign exchange allocation system during the sixties in Japan. This system was used as a powerful tool for industrial policy in the 1950s, and hence their removal was supposed to have a substantial impact on industries. After direct control of international trade by the government ceased in 1949 as a part of the “Dodge Plan,” the Japanese government regulated trade indirectly through the allocation of foreign exchange. This implies that, given the prices, there was a de facto import quota for some goods, since the upper limit of the import quantity was determined by the foreign exchange budget. Under continuing pressure from the IMF, the Japanese government swiftly removed the de facto import quotas. However, this process was different from what the literature in economics refers to as trade liberalization. The removal of import quotas did not necessarily constitute trade liberalization because tariff protection was substituted for import quotas. Therefore, to correctly quantify the effects of the quota removal, it is needed to control for the effects of the tariff protection.
In order to estimate the effect of quota removal, this paper utilizes detailed industry-level data from the Census of Manufactures (100 Japanese manufacturing industries in the 1960s) and data on trade protection. This enables them to control for industry (not firm) heterogeneity while covering the majority of manufacturing industries. Based on governmental information, the authors precisely identify the timing of the quota removal for each commodity, using original documents of the Ministry of International Trade and Industry (MITI). The authors estimate the parameters of interest (effect of the quota removal and the tariffs) using least square estimation including industry and time fixed effects. In this sense, the identification strategy of the effect of the quota removal is based on the variation in the timing of the quota removal across industries.
The authors find that the effects of the quota removal on productivity were limited. None of the industry performances are systematically related to the removal of the import quotas. Additionally, an increase in tariffs generally has significantly negative effects on the number of firms, output per establishment, and industry value added. The concern about reverse causality (higher tariffs were imposed on small industries in terms of the number of establishments and value added) is addressed using leads of the tariff and quota variables. The authors also check the effects on the growth rate of the result variables, finding that the quota removal had significantly positive effects, but time was required before they appeared. One explanation they provide for this is that the Japanese government increased tariff rates before removing the import quotas and maintained high tariff rates afterward.
I think that the main takeaway from the paper is that it suggests that the effects of the Japanese industrial policy in the 1960s might be smaller than widely believed in the Japanese economic history literature. However, I think the paper will benefit if the authors discuss more clearly some aspects. First, it is important to clarify what are the intended effects of the policy and what are the mechanisms for the effect of the quota removal on productivity. A clear discussion about mechanisms and intended effects could help the reader to understand the evaluation of the policy and what are the expected results. For example, is it a good or a bad result to see increases in productivity along with a decrease in the number of establishments? It seems natural to think that the government could impose de facto quotas to limit external competition and provide a handicap for the firms during the learning process. However, it is not clear what the intention of the government was when they removed the quota. Sometimes, the quota removal could be the result of the government thinking that some firms of the industries already have an appropriate level of productivity and that the less productive firms need to exit to allocate the resource to more productive production. But sometimes, the quota removal compensated with an increase in tariffs could be just a way to update the protectionism against the lobby of the new world financial institutions.
Second, I think the paper would benefit from a more detailed discussion about the identification strategy used and its suitability. A relevant challenge to the identification is the potential endogeneity of the timing of the quota removal. Since the Outline of the Plan for Trade and Foreign Exchange Liberalization was announced before the actual liberalization took place, the firms should have had incentives and time to adjust their behavior. Additionally, as mentioned above the criteria of the government could have been based on the observed trends of the industries. Suppose that the government decided to increase more the tariffs in those sectors that already have the lowest increases in productivity and that they suppose would be the most affected from the quota removal. Since the authors do not control for the pre-existing trends of the productivity of the industries, this issue can undermine the identification strategy, which is based on the idea that the timing of the quota removal varied exogenously across industries. Controlling for time trends per industry could help to capture these potential trends, and help to control for at least this potential source of endogeneity.
Finally, a third issue is related to the identification of the coefficients for tariffs and quota removal. Even assuming that the timing of the quota removal was exogenous, an issue raises from the fact that while the tariff rate is a continuous variable the quota removal is a binary variable. However, this quota removal binary variable tries to represent a treatment effect that is potentially different by industry. In this sense, the dummy variable is only a proxy for the actual severity of the removed protection. At the same time, as it was discussed before, the loss of protection via quota removal could be correlated with the tariff increases since the authorities would have tried to compensate the affected industries. If this is the case, the tariff effect is not precisely identified since it can be capturing the unobserved heterogeneity on the severity of removed protection. In this sense, maybe the use of a continuous variable that represents the magnitude of the removed protection via the quota removal could help to better identify the effects of those variables separately.
To sum up, I think this and other papers from the same authors are making important contributions to better understand the effects of the industrial policy during postwar Japan. In this paper the authors point out that the effects of quota removal might be smaller than widely believed in the Japanese economic history literature. Even more, they point out that the effects of different policies generally overlap and that any assessment of these effects needs to take care of this fact. I cannot stress enough how important industrial policy was for postwar Japan, but if you still have doubts, you should have asked Henry Ford II.