گروه بندی های صنعتی و سرمایه گذاری مستقیم خارجی

We explore worldwide foreign direct investment (FDI) location decisions by Japanese manufacturing firms from 1985 through 1991. Our conditional logit estimates provide evidence that firms' location decisions are affected by membership in either vertical or horizontal keiretsu. Consistent with previous studies that stress agglomeration effects on firms' location decisions, we find that the stock of investment in a region by a firm's vertical keiretsu partners increases the probability of location. Further, we find that the recent flow of investment into a region by a firm's horizontal keiretsu partners increases the probability of investment to the region, providing evidence of networking effects.

It has frequently been suggested that firms in the large industrial groupings of Japan and Korea, known, respectively, as keiretsu and chaebol, may behave differently than their US or European counterparts in a number of dimensions. A small set of papers in recent years have examined whether membership in such groupings affects firms' foreign direct investment (FDI) decisions, with a particular focus on the effect of keiretsu groupings on Japanese FDI activity. The majority of papers on this topic have examined the effect of vertical keiretsu groupings, where upstream suppliers are centered around a large downstream manufacturer. As detailed in Head et al. (1995), significant agglomeration externalities are likely to be present for firms in vertical keiretsu groupings. Suppliers work closely with downstream firms on designing specialized components and often provide just-in-time delivery of supplies. Thus, geographic proximity enhances the efficiency of such arrangements, giving firms incentives to locate their FDI in the same region. Head, Ries and Swenson (HRS) empirically examine Japanese FDI in the US and find that a greater existing stock of a Japanese firm's own vertical keiretsu partners in a particular US state makes that firm more likely to locate in that state as well. However, they find that this vertical keiretsu effect is largely due to the automobile sector and is no longer present when observations from this sector are excluded from their sample. 1 The other major form of industrial grouping in Japan is the horizontal keiretsu. These are groupings of firms in often-unrelated industries that are centered on a large bank. There are three features of horizontal keiretsu that observers have pointed to as potentially important for economic behavior: (1) cross-ownership between partner firms, where a group bank holds a significant share of the group-member firms' equity, and these member firms hold substantial equity positions in each other; (2) potentially easier terms of credit for members from the keiretsu-affiliated bank, both due to a reduction in the costs of monitoring and the discipline of repeat financing; (3) organized meetings among major keiretsu firms often called Presidential Council meetings which potentially facilitate the exchange of information and the coordination of actions. With respect to FDI activity, researchers have focused on easier credit terms as the feature of horizontal keiretsu that potentially leads to greater firm investment, including FDI. The employed tests examine whether membership in any horizontal keiretsu increases a firm's likelihood of FDI, since such membership should give the firm cheaper financing of such investment, ceteris paribus. 2 The evidence for these effects of horizontal keiretsu membership is much more mixed than for vertical keiretsu. 3 In addition to mixed empirical results, a recent paper by Miwa and Ramseyer (2002) argues that economic effects stemming from the features of horizontal keiretsu described above are more myth than reality. They point out that the vast majority of financing by Japanese firms in keiretsu groupings comes from non-keiretsu financial institutions, and this share has been increasing over time. This runs counter to the notion that horizontal keiretsu firms rely on their keiretsu bank for easier credit terms and may explain why previous studies have only found mixed evidence that horizontal keiretsu membership increases (foreign) investment. 4 Likewise, Miwa and Ramseyer show that cross-shareholding arrangements are much lower than one would expect, particularly between the non-financial members of these horizontal keiretsu. Finally, Miwa and Ramseyer dispute the notion that the Presidential Council meetings by executives of the keiretsu's main members have any influence on behavior by members whatsoever, contending that “scholars who stress their importance have yet to produce a lunch club decision that much mattered” (p. 193). Of the points raised by Miwa and Ramseyer, their argument against the effect of Presidential Council meetings is likely the weakest. Even if important collusive agreements among firms are not forthcoming from such meetings,5 there is likely a high level of networking and information exchange that occurs when heads of member firms meet. Such networking connections may have a very substantial role in economic decisions and behavior, as evidenced by recent papers documenting a strong and significant effect of networking connections on international trade flows.6 We find these papers compelling and suggest they serve to provide the obvious response to Miwa and Ramseyer (2002) criticism, which is to ask why such meetings take place if they serve no purpose. Additionally, a 2001 report by Japan Fair Trade Commission, 2001 based on surveys of horizontal keiretsu firms found that “The backbone of corporate groups is … [the] meeting of presidents of the companies in the same group. What are the benefits of such meetings? ‘Exchanges of information' topped the list, followed by ‘utilization of accumulated business expertise and know-how of affiliates.’” (p. 8) In this paper, we examine for the first time the potential effect of networking connections through keiretsu relationships on the FDI decisions made by Japanese firms. We hypothesize that if network effects are present then previous FDI activity by a firm's horizontal keiretsu partners into particular foreign regions will make FDI by the firm into these same regions more likely because this lowers the necessary costs of information acquisition. Thus, our focus is on whether region-specific horizontal keiretsu activity affects a firm's FDI location decisions. This contrasts with previous studies of FDI by horizontal keiretsu that has examined whether keiretsu membership per se affects a firm's decision to engage in FDI at all. Similar to HRS, we estimate a Japanese firm's FDI location decision using a conditional logit framework. However, we introduce two important innovations. First, we sample FDI location decisions by Japanese manufacturing firms in the late 1980s and early 1990s across all foreign regions of the world. Second, our sample includes both vertical and horizontal keiretsu. This allows us to examine the potential networking effects of horizontal keiretsu, while controlling for agglomeration benefits that are a strong feature of vertical keiretsu linkages. We find that agglomeration externalities from vertical keiretsu are important considerations for Japanese investment location decisions worldwide. The stock of investment in a region by a firm's vertical keiretsu members increases the probability of the firm's location in that region. For example, our estimates suggest that the existence of investment by vertical keiretsu members of 500 employees or more in a region increases the firm's probability of location in that region by over 50%. The agglomeration impact of unrelated firm location by Japanese manufacturing firms in the same industry is also statistically significant, though less than half the magnitude of the vertical own-keiretsu effects. These agglomeration results are consistent with HRS, but importantly generalize these results for Japanese location decisions worldwide, not just for regions in the US. Additionally, our vertical keiretsu effects are robust to the exclusion of automobile-centered vertical keiretsu, which was not true for HRS. We also explore and find supportive evidence for the idea that investment activity by related horizontal keiretsu members affects worldwide investment location decisions of Japanese firms. We find that recent (previous-year) investment activity by horizontal keiretsu partners leads to a greater probability of location in the same region by a firm. 7 For example, previous-year investment activity of horizontal partners of at least 100 employees increases a firm's likelihood of location in that region by over 20% on average. We attribute these effects to the networking and information sharing that occurs between members of horizontal keiretsu, as such information sharing can yield cost-savings to a newly locating firm. Consistent with this interpretation, we find such horizontal keiretsu effects are substantially larger for the Presidential Council firms in the most closely knit horizontal keiretsu—precisely where one would expect such information sharing to be the greatest. We further find that the horizontal keiretsu effects only occur with respect to recent investment activity, suggesting that information relevant for entry decisions depreciates with the age of the previous investments. These results provide the first systematic evidence for the networking effect of business groups on FDI decisions of which we are aware. 8 By examining the effect of recent investment activity by horizontal keiretsu members across regions, we better identify this networking effect, in contrast to previous studies that examined only whether membership in a horizontal keiretsu per se affects investment decisions. The rest of the paper proceeds as follows. Section 2 outlines our empirical methodology that starts with the framework of HRS and modifies it to allow for the effects of networking and information-sharing on FDI location decisions. Section 3 describes our sample of investment decisions by Japanese manufacturing firms from 1985 through 1991 and construction of our agglomeration and networking/information variables. Section 4 then presents out conditional logit estimates and sensitivity tests before a final section concludes.

This paper finds evidence that Japanese business groups (keiretsu) provide networking and information externalities that affect the FDI location decisions of partner firms. Previous work on Japanese business groups and FDI decisions has focused on vertical keiretsu, showing that location of FDI by a firm in a particular region is positively correlated with the existing FDI in that region by members of the firm's vertical keiretsu. The results from these studies provide evidence of agglomeration externalities. We extend this literature to examine more closely the effect of horizontal keiretsu, large firms in primarily unrelated fields that are tied to a major bank and whose leaders systematically meet in Presidential Council or “lunch club” meetings. Such meetings provide opportunities for information-sharing and networking, which can affect these firms' economic activity, including their foreign investment decisions. In particular, such information-sharing may allow a firm to better navigate a foreign country's tax/regulatory environment and better site its plant, lowering initial setup costs, as well as operating costs. This would then make the firm's location in the region more likely. Using a sample of all Japanese manufacturing investments from 1985 through 1991, we find evidence that recent FDI activity in a region by a firm's horizontal keiretsu members increases the probability of location in that region by the firm, even after controlling for other factors including sources of agglomeration externalities. As one would expect, this effect is much stronger for horizontal keiretsu firms that regularly participate in the Presidential Council meetings in the most closely knit horizontal keiretsu groups. A by-product of our analysis is that we confirm agglomeration externalities of vertical keiretsu relationships found by previous studies, but for a choice set that encompasses locations across the world, not just within a particular country, such as the United States or China.