Why do Japanese Exports Gain So Little from the Network Effect?

It is well known that business and social networks are important driving forces of international trade. Networks that were forged domestically in Japan are now being internationalized through cross-border migration of labor and foreign direct investment (FDI). These cross-border networks have an important role to play in eliminating informal trade barriers (such as the risk of breach of contract and lack of information about international commercial transactions) and promoting international trade.

Research into the network effect is well known in Japan as many researchers have focused on the relationship between keiretsu and Japan's trade structure. The earliest research started with the awareness of a problem, namely, that inter-firm networks forged in Japan perhaps were a trade barrier to non-Japanese enterprises (Fung, 1991; Qiu and Spencer, 2002). Analysis focusing on the network effect within Japan flourished in the 1980s and 1990s, when fierce trade friction occurred between Japan and the United States. Since 2000, the trade-promoting effect of cross-border keiretsu networks has gained more attention as Japanese businesses have become increasingly globalized. Baldwin and Ottaviano (2001) and Greaney (2003) showed theoretically that when buyers and sellers are in the same network, they can reduce their transaction costs and more easily enter foreign markets by exporting. Subsequent empirical research has supported this hypothesis (Head et al., 2004; Greaney, 2005).

The objective of the paper on which this column is based is to measure the network effect on the trade in intermediate goods. With the rapid spread of production fragmentation in recent years, it is becoming increasingly necessary to analyze the network effects on trade in intermediate goods. This research focuses on analyzing the auto industry, which, among the manufacturing industries, sees a particularly significant proportion of trade in intermediate goods. I estimate the network effects on auto parts exports from the six traditional auto-producing countries (TPCs): Japan, the United States, Germany, France, Italy, and Sweden. The network effect is defined as "the value of auto parts exports from the home country (that is, a TPC) resulting from an increase in overseas production by automakers headquartered in a TPC." I used an Anderson and Wincoop (2003)-type gravity equation with a fixed effects model to estimate the network effect. For the analysis, I used panel data covering six exporting countries (TPCs), 49 destination countries, and 31 auto parts over the 7-year period from 2002 to 2008.

Graph: International Comparison of Network Effects

Note: The graph shows how much the network effect in Japan, the United States, and Sweden deviates from the average network effect for the other TPCs.

Greaney (2005) demonstrated positively that Japan's network effect is relatively larger than that of other countries. Interestingly, however, in this research, Japan's network effect was smaller than that of the other TPCs. The graph gives an international comparison of network effects. Japan's network effect is 0.12% less than the average network effect of the other TPCs. In contrast, the United States and Sweden had greater network effects than the average of the other TPCs (0.32% and 0.24% greater, respectively). There is no evidence that the network effects of France, Italy, and Germany differ from the average effect of the other countries.

Why do Japanese exports have such a small network effect? One factor may be that, in Japan, the domestic keiretsu networks are strong, which restrain the network effect from promoting international trade. (On the other hand, the network effect readily asserts itself for U.S. and European automakers because they are set up for global sourcing.) One characteristic of the Japanese auto industry is that when Japanese automakers build production plants abroad, their strong inter-firm networks forged in Japan move with them to the host country. As mentioned previously, the existence of keiretsu networks in Japan increases the rate of local procurement of parts and lessens the dependence on imports. Therefore, even at the overseas subsidiaries of Japanese automakers, local procurement of parts is a major purchasing strategy. Global sourcing, including sourcing from Japan, plays a relatively small role.

A key point to consider in terms of the future of Japan's trade balance is the relationship between FDI and exporting in Japan. Generally speaking, the empirical research heretofore concludes that FDI increases exports from the home country. For example, when an upstream business (such as an automaker) expands overseas, it increases export demand at downstream businesses (parts suppliers). However, my research suggests that it is best not to be too optimistic about this export-inducing effect. It is a characteristic of Japanese businesses that when they expand overseas, their domestically forged inter-firm relationships move with them. In light of this, it is possible that FDI by downstream businesses takes the place of exports from the home country.

*This column is based on "Network Effects of Trade in Intermediate Goods: Evidence from the Automobile Industry" (Nishitateno, Shuhei, The Japanese Economic Review, forthcoming)