Although that thesis has received wide exposure through a number of influential articles, the evidence to support it has not been systematic. The present paper looks at the evolution of the full population of manufacturing joint ventures with U.S. firms established by Japanese investors in the United States. We track every one of these 57 U.S.-Japanese joint ventures to determine if there is a clear trend toward acquisition or dissolution by their Japanese parents. The results indicate that the majority of joint ventures maintained their joint venture status. Only in a minority of cases did the Japanese acquire total control. Hence, although they are subject to some limitations, our results do not support the Reich/Mankin and Hamel view that the Japanese use joint ventures with American firms as Trojan horses to penetrate the U.S. market. 
1. Introduction 

Japanese foreign direct investment in the United States grew dramatically during the 1980s. While holdings of all foreign direct investors increased substantially from $220 billion in 1986 to $403 billion in 1990, Japanese investments more than tripled from $26.8 to $83.5 billion. The growing Japanese presence in the United States has attracted considerable attention. The U.S. Department of Commerce Survey of Current Business (July, 1991) reports that manufacturing and service investment by Japan overtook European investments in the middle of the 1980s when measured by all but the historical cost method. Japanese manufacturing investment, however, was still behind the U.K. and Canada in 1989 (Noble, 1992, p. 135). 

The rapid growth of Japanese investment in the U.S. and the fact that it constitutes the first significant investment by a non-Western nation has triggered research into its determinants and patterns--see for example Hennart and Park (1993; 1994), Kenney and Florida (1993), Kogut and Chang (1991), Martin, Mitchell and Swaminathan (1993), and Yamawaki (1994)--and has led observers to ask whether Japanese investment is fundamentally different from previous foreign investment in the United States--see Caves (1993) for a summary of the evidence. Some have raised the suspicion that Japanese firms behave in a way that might disadvantage their partners. That distinct behavior is supposed to be in evidence when Japanese firms enter into joint ventures with non-Japanese firms. A number of authors (e.g., Reich/Mankin, 1986; Hamel, 1991; Pucik, 1988) have argued that the Japanese are particularly good at learning from their partners. When the learning is complete, Japanese partners see no reason to continue the ventures. They then acquire or liquidate them. 

Although that thesis has received wide exposure through a number of influential articles, the evidence to support it has not been systematic. The present paper attempts to discover the outcome of the full population of U.S.-Japanese joint ventures set up by Japanese investors and operating in the United States in 1980. We track each one of these 57 U.S.-Japanese manufacturing joint ventures to determine if there is a clear trend toward acquisition or dissolution by their Japanese parents. The results indicate that the majority of joint ventures maintained their joint venture status. Only in a minority of cases did the Japanese acquire total control. Hence, while they are subject to some limitations, our results do not support the Reich/Mankin and Hamel view that the Japanese use joint ventures with American firms as Trojan horses to penetrate the U.S. market. 

2.Japanese Joint Ventures as Trojan Horses

As shown by the Japanese-dominated consumer electronics industry, these [joint venture] agreements can act like a Trojan horse: the U.S. company provides the Japanese company access to its customers only to see the Japanese decide to go it alone and set up a distribution network on the basis of a reputation gained with the help of the U.S. partner. Even if the Japanese do not terminate the agreement after establishing a presence in the United States, Japanese manufacturers are in a position to squeeze their U.S. distributors' profit margins precisely because sales and distribution functions are so vulnerable to replacement (pp. 83-84).

Value is appropriated in joint ventures when the venture is used to absorb the skills of the partner. The party that learns the fastest gets the upper hand in the venture and is able to renegotiate the terms of the venture in its favor.1 As one of the interviewed managers put it: 
If they [our partners] learn what we know before we learn what they know, we become redundant. We've got to try to learn faster than they do (Hamel, 1991, p. 88). 

Hamel further argues that a firm's ability to learn from its partner hinges on three main factors. The first one is intent: does the firm view the joint venture as an opportunity to learn, or as a source of cheap products? Hamel suggests an asymmetry between Western and Japanese partners: the Japanese show explicit learning intents not shared by their Western counterparts. Second, the ability to learn depends on the transparency of the partner. Transparency is the degree to which the knowledge possessed by the firm's employees is open and accessible to the other joint venture partner. Agreeing with Reich/Mankin, all Western managers interviewed believed that the Japanese were inherently less transparent than themselves. Lastly, learning is facilitated by receptivity. Receptivity is found to be greater in Japanese firms than in their Western counterparts. Since Hamel found that Japanese firms tend to possess more of these three factors than their American partners, he concluded that U.S.-Japanese joint ventures will tend to end up fully acquired by their Japanese partners. 

The argument that the Japanese seem to get the upper hand in joint ventures with Western partners is also made by Pucik (1988). Pucik studied 23 joint ventures by U.S. and European firms in Japan. He noted that the Western partner in these joint ventures often relies on the Japanese partner for staffing. This saves the Western firm from having to recruit employees in Japan, and it reduces exit costs since the laid-off employees can be transferred back to their Japanese parents. 

To quote Pucik: 

What is often overlooked, however, is that, in the long run, lack of control over staffing puts the Western firm into a position of extreme vulnerability.

When the time comes to renegotiate the joint venture agreement, be it in five, ten, or twenty years, it becomes very tempting for the Japanese to walk away from the joint venture and pull the business back into the parent company or into a wholly owned subsidiary. After all, they have already learned all current technology and have not much else to gain by continuing the partnership (p. 490).

Reich/Mankin, Hamel, and Pucik present a pessimistic scenario for Japanese-American joint ventures. In their view, the Japanese enter the castle of the American firm and research its defenses (namely the resources the firm is bringing to the venture). The Japanese parent uses this knowledge to neutralize the advantages which allowed the American firm to initially enter the joint venture. Thereafter, the castle can be attacked from within. Occupied via a buyout, the venture can even serve as a source of future attacks. For the Japanese, the joint venture, therefore, is a Trojan horse. 

3.Potential Evolution of Joint Ventures

Gomes-Casseres (1987) looked at changes in the ownership structure of over five thousand foreign subsidiaries established between 1900 and 1975 by 180 U.S. multinational firms. He identified three possible types of change in joint ventures. First, the venture could be entirely liquidated, with its assets sold piecemeal or scrapped. Second, the venture could be sold to a local partner or to outsiders, remaining in operation, but under different ownership. Third, the venture could be transformed into a wholly-owned affiliate. 

Gomes-Casseres found that about two percent of all U.S. subsidiaries abroad were liquidated. That rate was slightly higher for wholly-owned subsidiaries than for joint ventures. Joint ventures, however, were more likely to be sold. Ownership changes were often at the margin: U.S. multinational firms were more likely to buy out their foreign partners when the U.S. side already controlled a majority of the shares of the joint venture than otherwise. Also, they were more likely to sell off the subsidiary if they were minority owners than otherwise. Altogether, majority-owned subsidiaries were the most unstable, followed by 50-50 joint ventures, minority joint ventures, and wholly-owned subsidiaries. Blodgett (1992), using a different sample and an event history methodology, found, however, that 50-50 joint ventures were more stable than ventures where ownership was unequally divided. 

Balakrishnan and Koza (1993) argue that joint ventures are often phased acquisitions: an American firm who wants to sell its assets to a Japanese investor may find it difficult to persuade the prospective acquirer that the assets are worth the asking price. Buying only part of the firm gives the Japanese buyer the opportunity to learn the true value of the assets, and to then complete the acquisition or sell his share.2 More generally, joint venture stakes may be seen as options, to be exercised when conditions warrant it: hence acquisition by one of the partners may be triggered by unexpected departures from industry sales trends (Kogut 1991). 

From this cursory survey of the literature on the evolution of joint ventures we draw two conclusions. First, some joint venture dissolutions can be explained by a baseline failure rate that affects all affiliates, both joint venture and wholly-owned subsidiaries. That failure is due to conditions in the overall environment. 

Second, the acquisition of a local partner by a foreign investor needs not result from the expropriation of the assets contributed by this partner a la Reich/Mankin. The joint venture may be a way to alleviate the information asymmetry between buyer and seller. It may also be a mechanism by which both parties place bets on the future. 

4.Some Testable Hypotheses

In the first one, the Japanese partner absorbs the knowledge assets of the American partner. Japanese bargaining power grows to a point where the Japanese partner is able to persuade its American counterpart to sell its share of the joint venture. Therefore the joint venture is transformed into a wholly-owned subsidiary of the Japanese partner. Another possible outcome of faster learning by the Japanese is the dissolution of the venture and its replacement by a new Japanese wholly-owned subsidiary. 

Hence if the Japanese have an inherently superior ability to capture the contribution of their joint venture partner, then the typical evolution of U.S.-Japanese joint ventures should be their acquisition by their Japanese partners, or perhaps their dissolution. By contrast, the continuation of the venture should be interpreted, in a Hamel world, as evidence that the Japanese are not able to expropriate the contribution of the American partner.3 

The strong version of the Trojan horse hypothesis thus implies that: 

H1.The number of cases of U.S.-Japanese joint ventures in the United States where the Japanese partner buys full ownership from its U.S. partner is higher than those where the Japanese partner divests its share to its U.S. partner or where ownership shares remain unchanged.

H2.The number of cases of U.S.-Japanese joint ventures in the United States where the Japanese partner buys full ownership from its U.S. partner or where the joint venture is dissolved is higher than those where the Japanese partner divests its share to its U.S. partner or where the ownership shares remain unchanged.

5.Methodology

The basic unit of analysis is the plant. However, when multiple plants manufacturing a given product are established or acquired in one single transaction, this counts as one observation. We deviated from this rule only once, in the case of Mitsui and Co. and Nippon Steel's partial acquisition of Amax's U.S. aluminum operations. Because of the size of this acquisition, and the possibility for bias it represents, it is worth explaining this case in detail. 

In 1974 Mitsui acquired a 45 percent (and Nippon a 5 percent) equity stake into 39 aluminum fabricating plants owned by Amax and a 22.5 and 2.5 stake into the Intalco aluminum smelter (which was a joint venture of Amax and Pechiney-Ugine-Kuhlmann). The new joint venture of Amax, Mitsui, and Nippon Steel was named Alumax. Alumax subsequently built two new aluminum smelters (one wholly-owned in 1980 and one in joint venture with Pechiney in 1975), two new architectural aluminum products plants (both in 1974), one new prepainted aluminum plant (in 1979), one new irrigation tubing plant (in 1975), two new secondary aluminum plants (in 1975 and 1978), and two new extruded shapes plants (in 1976 and 1980). Even if all plants acquired in 1974 are counted as one entry, our coding rules would have had us count the new plants established after 1974 as nine new joint venture entries, for a total of ten Alumax observations. Because this would have made up almost one-fifth of all our population of U.S.- Japanese joint ventures, we decided instead to represent these ten potential entries by only three observations (all the 1974 acquisitions counting as one entry, and the two new smelters, for a total of three observations). To refute the Trojan hypothesis, we need to show that there are relatively few firms which behave as hypothesized by Reich et al. To include all ten entries would have given too large a weight to the behavior of Mitsui and Nippon Steel. Note that our decision to count Alumax as only three cases biases our results towards support of the Trojan Horse hypothesis, since Nippon and Mitsui sold back their stake in all Alumax plants to Amax in 1986. 

Given this exception to our coding rule, we ended up with 57 U.S.-Japanese joint ventures for which we have complete information. Table 1 lists the initial distribution of these 57 ventures according to their combined Japanese ownership levels. In other words, when there were multiple partners, the share of each partner was aggregated according to nationality. For example, in the Alumax case described above where Mitsui, Nippon Steel, and AMAX owned 45%, 5%, and 50% respectively, the Japanese share was considered to be 50%. Table 1 shows that in 19 cases the combined share of the Japanese parent was over 50%, in 14 cases it was 50%, and in 24 cases the Japanese were minority investors. 

Note that combining the Japanese share has the effect of stacking the deck for support of the Trojan horse hypothesis, for it assumes that, in the case of three partners, two Japanese and one American, an increase in the combined Japanese share means that they jointly have expropriated the contributions of the American partner. While this may be true in some cases, this ignores the possibility that the two Japanese partners may in fact behave independently, and that they may use the venture to expropriate each other's contributions as well. 

Table 2 lists the sectors in which our joint ventures were operating. While it is risky to extrapolate the technology level of the venture from its SIC code (since some high-technology ventures may be in low-technology sectors and vice-versa), the general impression, confirmed later in the paper by a detailed analysis, is that most of the ventures were not in high-technology industries. Using the criteria described in Section 6 below, we find that only 9% of the ventures involved high-technology. 

Next, we determined if any of these U.S.-Japanese joint ventures experienced a change in status between December 31, 1980 and December 31, 1989. In addition to the sources listed above, information on changes in status were obtained from Mergers and Acquisitions, the F and S Directory of Corporate Change, the Lexis-Nexis database, and numerous phone and mail inquiries to the subsidiaries and their parents. In the case of discontinued affiliates, the City Hall, Post Office, or Chamber of Commerce of the last known location was contacted. Affiliates in which Japanese ownership goes down to zero are those which have been sold to the partner, sold to a third party, liquidated or gone bankrupt over the period.7 

We chose a nine year (Dec. 31, 1980 - Dec. 31, 1989) window for our study because, according to Hamel's view of joint ventures as learning races, one would expect a few years to elapse before partners would consider changes in the organizational form of the investment. We believe that our nine year period allows sufficient time for learning and for the subsequent re-evaluation of the original arrangement to take place.8 We chose 1989 as an end period because our goal is to separate strategic considerations at the U.S. subsidiary level from the impact of the Japanese parent's domestic business. Nineteen eighty-nine marks the end of the so-called ;bubble economy,; and Japan then entered a period of domestic adjustment in which it still finds itself today. There is some evidence that adverse conditions in Japan have, since 1989, forced Japanese parents to reduce or liquidate the ownership of their U.S. ventures. Hence a 1989 end date has the advantage of preventing our results from being unduly influenced by Japanese home-market conditions. Besides, a 1989 cutoff tends to bias the results toward support of the Trojan hypothesis, since we would expect to see a greater number of increases in equity, and a greater tendency not to liquidate or divest unprofitable ventures before 1989, when both low rates of interest in Japan and high levels of domestic profit reduced the cost of such strategies. 

6.Results

Japanese joint ventures show a higher rate of failure than that found by Gomes-Casseres in the case of U.S. investments abroad: 28%, compared to 13% in the case of overseas joint ventures of U.S. firms. Gomes-Casseres; findings that ownership changes seem to be at the margin is verified in our case: none of the minority or majority joint ventures evolved into a majority or wholly-owned affiliates and only one 50-50 subsidiary became a minority-owned affiliate. If minority-owned affiliates changed status, most ended up sold or liquidated. On the other hand, one-third of the 50-50 joint ventures and 42% of majority-owned ones had been transformed into wholly-owned subsidiaries of the Japanese parent. Table 4 shows changes over the 1980-89 period in terms of 1980 Japanese equity share. Note, however, that our figures are not strictly comparable with those of Gomes-Casseres, since he did not look, as we do, at changes over a specified time period. 

Table 4 summarizes the evolution of U.S.-Japanese joint ventures in terms of the categories we need to test the Trojan hypothesis. There are 13 cases where the Japanese increased their combined share to 100%, and in all cases this was obtained by buying out their American partners' stakes. In 16 cases combined Japanese ownership decreased to zero: in seven of these cases the Japanese sold their stake to their 1980 American joint venture partner. In three cases the joint venture went bankrupt or was liquidated. In two cases the Japanese share of the venture was sold to another U.S. company, in three cases it was sold to a third country company, and in one case it was sold to another Japanese company. What these tables show is that U.S.-Japanese joint ventures are more likely to be dissolved, liquidated, or sold than they are to be fully acquired by the Japanese partner. Yet, the striking result is perhaps not the nature of change, but the lack of change. In 21 of the 57 cases in our population, or 37%, the participants chose not to make a major change in the organization of their venture. 

The results from our data set then cast doubt on the hypothesis of Reich/ Mankin, Hamel, and Pucik that the Japanese have a strong tendency to take the upper hand in their joint ventures with foreigners. Recall that the strong version of the Trojan horse hypothesis is supported if the number of Japanese-U.S. joint ventures acquired by the Japanese partner is higher than the sum of the number of cases where the U.S. partner buys out its Japanese partner and those where the division of equity remains unchanged. The weak version is supported if the number of Japanese acquisitions plus the number of liquidations is superior to that of U.S. takeovers and that of unchanged equity stakes. 

In our case, the number of joint ventures fully acquired by the Japanese is 13 and there are three liquidations, while there are seven cases of U.S. firms buying out their Japanese partners, and 21 cases of unchanged equity shares. Hence, looking at the stronghypothesis, 13 cases are compatible with the Trojan horse scenario, while 28 contradict it. Adding the three cases of dissolution to those of full acquisition of the venture by the Japanese partner (the weak version of the hypothesis) does not change our results: neither the strong nor the weak version of the Trojan horse hypothesis is supported. 

Note that, as mentioned above, our weak version of the Trojan horse hypothesis considers the dissolution of a U.S.-Japanese joint venture to be akin to its full acquisition by a Japanese partner. We are implicitly assuming that the dissolved venture is being replaced by a parallel venture, now fully-owned by the erstwhile Japanese partner. In fact, as we mentioned earlier, dissolutions could result from other causes unrelated to our argument. 

The Hamel argument, based on learning, might suggest that the Japanese strategy in joint ventures would vary, depending on the potential in the joint venture for learning from the American partner. That rationale for Trojan horse joint ventures would be strongest in industries in which there is significant technology to be learned. We may therefore want to look at a subset of high- technology joint ventures. 

To identify these ventures, we used the following two step process. First we selected all firms in our population operating in high-technology industries. In our population, this included firms in pharmaceuticals, electronics, precision instruments and special industry machinery. Within these high-technology sectors, we then reviewed the stated objective of each venture to see if it exhibited a high level of technology. The process yielded only five ventures we could label as high- technology. 

Table 5 shows that the Japanese partner increased its share in one of these five ventures, but not to 100%. One venture was sold to the U.S. partner, one was sold to a third country company, one was liquidated, and one was unchanged. Within this high-technology subgroup, the strong version of the Trojan horse hypothesis was not supported. Only one of the cases (that where the Japanese partner bought out its U.S. partner) is compatible with the hypothesis, while two cases (that of the affiliate sold to the U.S. partner and that of the unchanged ownership) contradict it. 

The weak form of the Trojan horse hypothesis receives stronger support. Adding the case of the liquidated affiliate to the case of the affiliate where the Japanese parent increased its equity share gives two cases compatible with the Trojan horse hypothesis, while there are two cases that contradict it (the case of the affiliate with unchanged ownership and that of the affiliate sold to the U.S. partner). Although the numbers are small, the proportion of Japanese buyouts in the high- technology subset is clearly higher than that in the population as a whole. 

Remember that the weak test assumes that a liquidated affiliate is the functional equivalent to a Japanese buyout because the Japanese are supposed to re-establish a similar, wholly-owned venture to replace the former joint venture. Given the small number of cases, we were able to look for evidence of such a pattern. We found no instances where the audio equipment and pharmaceutical firms which dissolved or sold their joint ventures subsequently established new wholly-owned subsidiaries manufacturing similar products. This indicates, at best, lukewarm support for even the weak form of the hypothesis. 

7. Conclusion

Reich/Mankin, Hamel, and Pucik argue that Japanese firms tend to learn faster from their U.S. partners than U.S. firms do from their Japanese partners. Hence, they predict that the Japanese will eventually buy out the equity stake of their American partner or perhaps liquidate the venture and set up a competing wholly-owned subsidiary. In this paper we call this view the "Trojan horse" hypothesis. Our operationalization of this hypothesis is straightforward: we consider it to be verified if the proportion of all U.S.-Japanese joint ventures in the U.S. taken over by the Japanese partner is higher than that in which the Japanese partner sell its share to its U.S. partner or ownership shares remain unchanged. A weaker version counts dissolved ventures as equivalent to those taken over by the Japanese, on the assumption that the Japanese then recreate parallel, wholly-owned substitutes. 

Two specifications in our operationalization of the Trojan horse hypothesis bias our results toward support of the hypothesis. First, we combine the Japanese share in joint ventures with more than one Japanese partner, implicitly assuming that the Japanese always collude between themselves to expropriate their American partners. Second, we assume, in the weak version of the hypothesis, that dissolutions are always instigated by the Japanese parent who then replaces the venture with a fully-owned affiliate. This is clearly unwarranted: dissolutions may only mean that the prospects for the venture fall short of expectations. 

Inversely, we have assumed that successful use of the joint venture as a Trojan horse by the Japanese partner would lead to the purchase of the U.S. stake or the dissolution of the venture. This need not always be the case. Consider the following: the Japanese partner has now learned the substance of the contribution of its American partner. It is now in a better bargaining position. This may allow it to establish rules that increase its share of returns without increasing its equity share. For example, the Japanese parent could negotiate specific up-front payments for its technology before the joint-venture profits are calculated. The U.S. partner may decide to stay in the venture as long as it still makes normal profits. 

Keeping in mind these limitations, ours is the first test of the Reich/Mankin hypothesis based on essentially the full population of U.S.-Japanese manufacturing joint ventures active in the United States in 1980. Following the evolution of the full population of these joint ventures between 1980 and 1989, we found that neither the strong nor the weak version of the Trojan horse hypothesis was supported. Recall that the strong version is supported if the proportion of U.S.-Japanese joint ventures fully acquired by the Japanese partner is higher than that fully acquired by the U.S. partner or in which the shares remain unchanged. We find that there were 13 cases of the former, and 28 cases of the latter. The weak version of the hypothesis assumes that dissolutions of the venture are functionally equivalent to full purchase of the U.S. stake by the Japanese. There were, however, only three liquidations, so this is not enough to tip the scales towards acceptance of the Trojan horse hypothesis. 

When we focus on the subgroup of high-technology ventures, the numbers, although slightly more favorable to the Trojan horse view, still fail to support at least the strong form, with one case of increased ownership by the Japanese partner vs. one case in the "no change" category and one case where the Japanese stake was sold to the U.S. partner. There was one dissolution. 

There are many possible reasons for our findings. The first, and most obvious one, is that the Japanese learning advantage postulated by Reich/Mankin, Hamel, and Pucik, while certainly in evidence in the examples they cite, is not characteristic of the universe of all Japanese parents of Japanese-U.S. joint ventures in the United States in 1980. Our results are inconsistent with the view that the Japanese seek to gain equity control of the joint ventures they establish with U.S. partners because they have absorbed all their skills. 

One explanation for the large number of ventures with unchanged status may be that, contrary to Hamel's argument, the American partners are able to learn from their Japanese partners and to protect themselves against Japanese opportunism. Then we would observe long-lasting ventures, but ventures having all the supposedly destabilizing goals--competition for learning and attempts to use the venture to improve one's competitive position--mentioned by Reich/Mankin and by Hamel. Roehl and Truitt (1987) describe such dynamics in the case of aerospace ventures. 

Perhaps a more fundamental reason why we find that most U.S.-Japanese joint ventures do not undertake major changes in ownership over our nine year period is that the Reich/Mankin and Hamel focus on joint ventures as mechanisms to appropriate the contributions of the partners is too limited. As pointed out in the transaction cost literature, joint ventures can also be seen as institutional devices to create value for both partners. Hennart (1988, 1991), for example sees joint ventures as institutional arrangements that allow parties with complementary, but hard to transact assets, to combine these assets and to reap the profits by commercializing goods and services incorporating these assets (link joint ventures) or by pulling together similar assets in order to obtain economies of scale and to minimize risk (scale joint ventures). In both these cases rewarding the firms that provide the inputs by giving them a share of the profits of the venture reduces (but does not fully eliminate) incentives for opportunism. Hence it can be argued that the Reich/Mankin and Hamel view is unduly limiting in at least several dimensions. 

First, the focus on partners trying to exploit each other is probably too pessimistic. Crude attempts to capture the intangibles contributed by one's partner can be expected to lead to defensive counter-measures that are likely to raise the costs and reduce the benefits of cooperation. At some point the damage to the joint venture profits is likely to exceed the firm's gain from appropriating its partner's knowledge. Consider Fujitsu-Amdahl, one of the joint ventures in our population. Fujitsu has been careful not to be too closely involved in the management of Amdahl. According to Amdahl's CEO, "Fujitsu likes us to be independent because it recognizes that we understand the marketplace and that we are good designers (Keough, 1986)." When Fujitsu acquired Amdahl, it agreed to keep its ownership stake in the firm to 49.5 percent for 10 years. It publicly announced in April 1994, at the expiration of this agreement, that it would continue to do so in the foreseeable future.9 

Second, the Reich/Mankin and Hamel learning scenario may be unduly static, focusing as it does on the stock of resources both firms bring to the joint venture at its inception. If this is all the resources that participation in the venture can offer, then indeed when one party fully appropriates this initial pool, there are no longer any reasons for collaboration. Additional resources can, however, become available through two routes. First, one or both parents can continue to bring new resources into the venture structured so that these resources are damaged if a partner leaves. Alternatively, the venture can be structured so that it generates, on its own, a pool of resources available to both partners to use inside and outside the venture. In this alternative scenario, both sides may well consider continued participation to be attractive. 

Third, not all joint ventures are link joint ventures where the parties bring to the table intangible assets. Our database also contains scale joint ventures used to pool demand to capture economies of scale. For example, North Pacific Paper Corporation, a joint venture of Weyerhaeuser and Jujo Paper located in the Pacific Northwest, pools newsprint demand from Japan and the United States. As in the case of many such scale joint ventures (Stuckey 1983), ownership stakes have remained stable over the period. 

In short, the dynamics of joint ventures are more complex than suggested by Reich/Mankin, Hamel, and Pucik. Joint ventures arise from a wider set of motives, and hence their history depends on a larger set of factors. Even if we stay within the "learning race" view of joint ventures, we must keep in mind that the contribution of the partners may be constantly replenished. The insightful findings of Reich/Mankin, Hamel, and Pucik may arise from focusing on a particular type of joint venture, a learning race where partners contribute a fixed sum of knowledge. That all the cases involve Japanese firms may be a coincidence. It seems that a more useful focus for our research is the specific strategies followed by the partners rather than their nationalities. 

Table 1

Japanese Ventures in the U.S. in 1980

Table 2

Industry Distribution of U.S.-Japanese

Joint Ventures in the United States

on December 31, 1980

Note: The category of increase includes three cases where Japanese parents increased their ownership share, but still retained only a minority ownership. 

Table 3

Table 4

Changes in the Ownership Status of U.S.-Japanese

Manufacturing Joint Ventures in the United States

(Dec 31, 1980-Dec 31, 1989)

(by initial ownership level of the Japanese partner)

Table 5

Changes in the Ownership Status of U.S.-Japanese High Technology

Manufacturing Joint Ventures in the United States

(Dec 31, 1980-Dec 31, 1989)

(by initial ownership level of the Japanese partner)

Endnotes

2. Note that the venture is likely to be acquired when the Japanese complete their learning, but in this case the Japanese learn about the value of the assets contributed and embedded in the venture, and not about their substance, as in the Reich/Mankin scenario. 

3. It is possible that the venture could continue nominally as a joint venture, but with the Japanese having the upper hand. For example, the Japanese parent could choose to take the profits from its increased bargaining position via a contract with the joint venture. The Japanese parent could, for instance, supply parts or other services at above market prices to the joint venture, while sharing the profits and losses of the venture with the American partner. The American partner may stay in the venture as long as it earns normal profits. One may wonder why, however, if the joint venture makes profit, the Japanese partner would agree to share the profit of an operation in which the American partner does not contribute much. 

4. A dissolution of the venture could also be interpreted as support for the fact that the American partners have expropriated the knowledge of their Japanese partners, but Hamel assumes that this does not take place. 

5. This is the case for Mount Pleasant Chemical, one of the ventures in our sample. This joint venture was established by Sumitomo Chemical and Stauffer Chemical Co. to manufacture Sumitomo;s brand of insecticide for forests. After nearly nine years, the venture was disbanded, as it was unable to obtain sufficient market share against a well established domestic brand See Japanese Investments in the US are Riskier than many First Support, Japan Economic Journal. July 5, 1986, p. 13. 

6. This is the case of the Alumax joint venture between Mitsui, Nippon Steel, and Amax mentioned below. The Japanese joint venturers found that they could get access to Amax's aluminum supplies through long-term contracts. 

7. Another possible reason why the affiliate may no longer be in existence is a reorganization by which the affiliate loses its own identity and is now merged with another entity. We investigated this possibility, but found no evidence of such a change. 

8.Kogut (1988, table 10-1) found that instability rates for international joint ventures peaked six years after the birth of the joint venture. Our nine year widow is therefore long enough to capture most of the changes that will affect the vinture. 

9. Telephone interview with Bill Stuart, Director of Finance and Public Relations, Amdahl, August 2, 1994. 

BALAKRISHNAN, S. and KOZA, Mitchell. "Information Asymmetry, Adverse Selection and Joint Ventures: Theory and Evidence," Journal of Economic Behavior and Organization 20, (1993): 99-117. 

BEZIRGANIAN, Steve D. "U.S. Affiliates of Foreign Companies: Operations in 1989." Survey of Current Business 73 (July 1991): 89-112. 

BLACK, J. Stewart and Mark MENDENHALL. "Resolving Conflicts with the Japanese: Mission Impossible?" Sloan Management Review 34 (Spring 1993): 49-59. 

BLODGETT, L.L. "Factors in the Instability of International Joint Ventures: An Event History Analysis." Strategic Management Journal 13 (September 1992): 475-481. 

CAVES, Richard. "Japanese Investment in the U.S." The World Economy 16 (May 1993): 279-300. 

GERLACH, Michael. "Trust is Not Enough: Cooperation and Conflict in Kikkoman's American Development." Journal of Japanese Studies 16 (Summer 1990): 389-425. 

GOMES-CASSERES, Benjamin. "Joint Venture Instability: Is It a Problem?" Columbia Journal of World Business 22 (Summer 1987): 97-102. 

GOMES-CASSERES, Benjamin. "Ownership Structures of Foreign Subsidiaries: Theory and Evidence." Journal of Economic Behavior and Organization 11 (January 1989): 1-25. 

HAIGH, Robert W. "Building a Strategic Alliance: The Hermosillo Experience as a Ford-Mazda Proving Ground." Columbia Journal of World Business 27 (Spring 1992): 60-74. 

HAMEL, Gary, Yves L. DOZ, and C.K. PRAHALAD. "Collaborate with Your Competitors--and Win." Harvard Business Review 67 (January-February 1989): 133-136. 

HAMEL, Gary. "Competition for Competence and Inter-Partner Learning Within International Strategic Alliances." Strategic Management Journal 12 (Summer 1991 Special Issue): 83-104. 

HENNART, Jean-Francois. "A Transaction Cost Theory of Equity Joint Ventures." Strategic Management Journal 9 (1988): 361-374. 

HENNART, Jean Francois. "The Transaction Costs Theory of Joint Ventures: An Empirical Study of Japanese Subsidiaries in the United States." Management Science37 (April 1991): 483-497. 

HENNART, Jean-Francois, and PARK, Young-Ryeol. "Greenfield vs. Acquisition: The Strategy of Japanese Investors in the United States," Management Science, vol. 39, no. 9, (September 1993): 1054-1070. 

HENNART, Jean-Francois, and PARK, Young-Ryeol, "Location, Governance and Strategic Determinants of Japanese Manufacturing Investment in the United States," Strategic Management Journal, vol. 15, no. 6, (1994): 419-436. 

HURRY, Dileep, Adam T. MILLER, and E.H. BOWMAN. "Calls on High-Technology: Japanese Exploration of Venture Capital Investments in the United States." Strategic Management Journal 13 (February 1992): 85-101. 

IMAI, Ken-ichi. "Japanese Corporate Strategies toward International Networking and Product Development." In International Adjustment and the Japanese Firm, pp. 140-159. Edited by Paul Sheard. St. Leonards, Australia: Allen & Unwin, 1992. 

INKPEN, Andrew. "The Characteristics and Performance of Japanese-North American Joint Ventures in North America." Advances in International Comparative Management 9 (1994): 83-108 
. 

JAPAN ECONOMIC INSTITUTE. Japanese Manufacturing Investment in the United States. Washington, D.C.: Japan Economic Institute (various years) 

JAPAN EXTERNAL TRADE ORGANIZATION. Directory of Japanese Affiliates in the United States. Tokyo: JETRO (various years) 

JOHANSON, Jan and VAHLNE, Jan-Erik. "The Internationalization Process of the Firm--A Model of Knowledge Development and Increasing Foreign Market Commitments." Journal of International Business Studies8 (October 1977): 23-32. 

KENNEY, M. and FLORIDA, R. Beyond Mass Production: The Japanese System and its Transfer to the U.S. . New York: Oxford University Press, 1993 
. 

KOGUT, Bruce. "A Study of the Life Cycle of Joint Ventures," in Cooperative Strategies in International Business. Edited by F. Contractor and P. Lorange. Lexington: Lexington Books, 1988. 

KOGUT, Bruce. "Joint Ventures and the Option to Expand and Acquire," Management Science 37 (1991) 19-33. 

KOGUT, Bruce and CHANG, Sea-Jin. "Technological Capabilities and Japanese Foreign Direct Investment in the United States," Review of Economics and Statistics, 73, (1991): 401-413. 

KEOUGH, Lee. "Amdahl steps out: the largest maker of plug-compatible mainframes is about to half-step from its well-heeled partner." Computer Decisions 18 (November 4, 1986): 34. 

MARTIN, Xavier, MITCHELL, Will, and SWAMINATHAN, Anand. "Rebuilding Toothpick Houses: Recreating and Extending Buyer-Supplier Links after Setting Up Foreign Manufacturing Facilities." Working paper, University of Michigan, June 1993. 

NOBLE, Gregory W. "Takeover or Makeover? Japanese Investment in America." California Management Review 34 (Summer 1992): 127-147 
. 

OHMAE, Kenichi. "The Global Logic of Strategic Alliances." Harvard Business Review 67 (March-April 1989): 143-154. 

OSBORN, R.N. and C. C. BAUGHN. "New Patterns in the Formation of U.S. / Japanese Cooperative Ventures: The Role of Technology. " Columbia Journal of World Business 22 (Summer 1987): 57-65. 

OUCHI, William G. and Michele Kremen BOLTON. "The Logic of Joint Research and Development." California Management 31 (Spring 1988): 9-33. 

PUCIK, Vladimir. "Strategic Alliances with the Japanese: Implications for Human Resource Management." In Cooperative Strategies in International Business, pp. 487-498. Edited by F.J. Contractor and P. Lorange. 487-498. Lexington, MA: Lexington Books, 1988. 

PUCIK, Vladimir. "Strategic Alliances, Organizational Learning, and Competitive Advantage: The HRM Agenda." Human Resource Management 27 (Spring 1988): 77-93. 

PUCIK, Vladimir. "Joint Ventures with the Japanese: The Key Role of HRM." Euro-Asia Business Review 6 (October 1987): 36-39. 

REICH, Robert B. and Eric D. MANKIN. "Joint Ventures with Japan Give Away our Future." Harvard Business Review (March/April 1986): 78-86. 

ROEHL, Thomas W. and J. Frederick TRUITT. "Stormy, Open Marriages are Better: Evidence from U.S. Japanese and French Cooperative Ventures in Commercial Aircraft." Columbia Journal of World Business 22 (Summer 1987): 87-95. 

SINHA, Deepak K. and Michael A. CUSUMANO. "Complementary Resources and Cooperative Research: A Model of Research Joint Ventures among Competitors." Management Science 37 (September 1991): 1091-1106. 

STUCKEY, John. Vertical Integration and Joint Ventures in the Aluminum Industry. Cambridge, Mass.: Harvard University Press, 1983. 

TOYO KEIZAI. Kaigai Shinshutsu Kigyo Soran (Japanese Multinationals, Facts and Figures). Tokyo: Toyo Keizai (various years). 

TYEBJEE, Tyzoon T. "A Typology of Joint Ventures: Japanese Strategies in the United States." California Management Review 31 (Fall 1988): 75-86. 

UNITED STATES DEPARTMENT OF COMMERCE. Survey of Current Business, July 1991. 

WAKASUGI, Ryuhei. "A Consideration of Innovative Organization: Joint R & D of Japanese Firms." In Evolving Technology and Market Structure: Studies in Schumpeterian Economics, pp. 209-226. Edited by Arnold Heertje and Mark Perlman. 209-226. Ann Arbor, MI: University of Michigan Press, 1990. 

WILKINS, Mira. "Japanese Multinationals in the United States: Continuity and Change." Business History Review 64 (Winter, 1990): 585-629. 

YAMAWAKI, Hideki. "Patterns of Entry of Japanese Multinationals into U.S. and European Manufacturing Industries," in D. Encarnation and M. Mason (eds.), Does Ownership Matter? Japanese Multinationals in Europe. Oxford: Oxford University Press, 1994. 

YOFFIE, D.B. "Foreign Investment in Semiconductors." In Foreign Direct Investment, pp. 197-228. Edited by Kenneth Froot. Chicago: University of Chicago Press, 1993.