U.S.-Japan Strategic Alliances in the Semiconductor Industry: Technology Transfer, Competition, and Public Policy (1992)

For semiconductor companies shopping for external technology, the mechanisms of acquisition range from OEM (original equipment manufacturing) licensing, at the simple end of the spectrum, to joint ventures and the outright purchase of innovative companies, at the more complex and expensive end. Virtually every linkage imaginable along this spectrum falls into the catchall category of "strategic alliances." Indeed, the category is so encompassing that it runs the risk of failing to reveal meaningful distinctions.

For purposes of this report, all alliance linkages that do not serve as a direct conduit for knowledge transmission can be omitted. Specifically, this means the exclusion of strategic alliances dealing exclusively with pure marketing, short-term, spot-market buyer-seller transactions, and arms distance equity investments (made solely to extract financial profits, with no attempt to monitor or obtain technology). Depending on the nature of the exchanges, some instances of these transactions may lead to a transfer of technology. For example, marketing and sales agreements that involve after-service can diffuse technology as an unintended by-product. Nevertheless, certain types of strategic alliances—particularly the simple, one-shot transaction variety—are omitted from consideration here.

A wide range of linkages that involve technology transfer still remains. Dataquest has broken American-Japanese semiconductor alliances down into 12 types. These alliance types can be modified, expanded, and grouped under four different headings, each dealing with different phases of busi-

ness activity: (1) research and development, (2) manufacturing, (3) marketing and services, and (4) general purpose (see Table 1).

Categorizing alliance types can help us analyze their structure and effects. However, areas of ambiguity remain that defy clarification and precise categorization. In the case of product-or technology-oriented equity

investments, for example, boundary questions arise with respect to the level of capital invested. If the percentage of investment is significant (40 or 50 percent), some might call it a strategic alliance. Conversely, if the percentage falls below a certain threshold (25 percent for example), some might conclude (incorrectly) that the level of investment is too trivial to qualify for inclusion as a strategic alliance. Because the main focus of this report is on the transfer of technology, the establishment of an arbitrary capital investment cut-off point—whether 25 percent or 33 percent—is irrelevant and potentially misleading. Key technologies can be obtained with only 5 percent equity investment; conversely, technology might not be transferred even when there is majority ownership. Arbitrary thresholds cannot be established as meaningful guidelines.

This analysis of alliance types focuses on those strategic alliances that serve as conduits for the transmission of technology in one or more directions, either on a one-time basis or as a mechanism for repeated and regular transmission. It would, of course, be useful if the various types of alliances could be correlated with the commercial importance of the technology transferred but unfortunately no discernible pattern of correlation emerges from the data. Thus, we must be content with an operational definition of strategic alliances based on the construction of a typology that highlights the nature (iterative or non-iterative) and direction (one-or two-way flow) of technology transfer.

Over the postwar period the scope and nature of American-Japanese alliances have changed, reflecting changes in technology, commercial competition, and the positions of the U.S. and Japanese semiconductor industries. In the earliest phase, covering roughly three decades from 1950 to 1980, very few strategic alliances were concluded. The few that were consummated took the form of simple licensing agreements, involving the sale of basic U.S. patents to latecomer companies in Japan. Until the 1970s, most American-Japanese alliances fell into the category of licensing agreements aimed at transferring basic R&D know-how to Japanese semiconductor producers.

Among the notable licensing agreements were RCA-Hitachi (1961), TRW-Mitsubishi (1962), Honeywell and Fairchild with NEC (1962), Sperry Rand-Oki (1963), and General Electric-Toshiba (1964). Such alliances occurred at a snail's pace of roughly one per year. There were almost no alliances in the areas of manufacturing, marketing and servicing, or general-purpose cooperation, three of the four alliance categories mentioned earlier.

Those U.S. companies wishing to pry open the Japanese market ran squarely into the dead end of formal barriers to trade and foreign direct investment. Unable to export or set up shop, several U.S. firms decided, as

a fallback position, to license their technology so as to gain at least some revenue stream. Others, such as Texas Instruments, which were determined to establish a foothold in Japan, had to agree, as the price of admission, not only to license their basic technology but also to enter a joint venture with a domestic producer. Texas Instruments' joint venture with Sony Corporation was dissolved as soon as the American partner was allowed by the Ministry of International Trade and Industry (MITI) to establish a wholly owned Japanese subsidiary. ⁸ Hence, the alliance pattern for the first 30 years consisted of a few scattered licensing agreements, which grew in numbers during the 1970s; several joint ventures (including the Texas Instruments-Sony venture encouraged by MITI and the unsuccessful Motorola-Alps venture described in the Motorola-Toshiba case study in Appendix A); and very little else.

One explanation for the paucity of alliances relates to the legal limits placed on foreign direct investments in Japan, the strict regulatory controls imposed over foreign exchange, and the requirement that Japanese corporations secure formal MITI approval for licensing and alliance formation.⁹ Most of these legal and regulatory constraints were not fully removed until around 1980. The passage of the Foreign Trade and Exchange Act in 1979 cleared the way for Japanese companies to enter freely into joint ventures (with only the obligation of prior notification), to move yen and dollars freely in and out of Japan, and to buy and barter technology without government approval.

This sweeping away of administrative obstacles also happened to occur at a time of very rapid development in the history of Japan's semiconductor industry. The co-occurrence of the two, along with other underlying forces at work, created the circumstances for an explosion of alliances from the mid-1980s onward. During the early 1980s, the frequency of alliance formation increased markedly, especially in the area of U.S. licensing of memory and microprocessor technology. In the mid-1980s, a number of agreements were signed in the area of semiconductor equipment, and by the late 1980s, the proliferation of U.S.-Japan alliances had reached a peak. Just as significantly, the nature of alliances underwent a change from simple licensing to more complicated and multifaceted arrangements (see Figure 2).

Aggregate data available from Dataquest and the American Electronics Association Japan Office confirm the boom in the number, type, and scope of strategic alliances involving U.S. and Japanese semiconductor companies

FIGURE 2 Types of technology alliances. Source: Sheridan Tatsuno, NeoConcepts, 1990.

(see Figure 3). Bear in mind that the aggregate data are incomplete because they cover only the strategic alliances that companies have made public. The number of publicly announced alliances may be only the tip of the iceberg, because many alliances are not publicly announced and do not enter the realm of public knowledge. For a variety of reasons, such as fears of sending the wrong signals to investors or of setting off strong competitive counterstrategies on the part of rival firms, American and Japanese companies often choose to keep their alliances quiet.

FIGURE 3 American-Japanese semiconductor alliances: 1984–1990. Source: Bruce Kogut and Dong-Jae Kim, "Strategic Alliances of Semiconductor Firms, "unpublished report to Dataquest, January 1991.

Therefore, it is almost impossible to estimate the total number of U.S.-Japan alliances with precision. Interviews with U.S. executives suggest that the available figures greatly understate the actual number; indeed, public data may represent less than half of the actual number of American-Japanese alliances. Although data are incomplete, it is clear that U.S.-Japan alliance activity in semiconductors has generally followed an upward trend over the past decade. That trend is not likely to fall off sharply.

One obvious factor explains the proliferation of alliances from the mid-to late 1980s: the growing and immense size of the Japanese semiconductor market, which had become the largest in the world by 1988. Seeing this trend clearly, U.S. semiconductor firms realized their need to find a way to ride the crest of Japanese growth. Strategic alliances with Japanese partners offered a vehicle for gaining a foothold in Japan's growing but, for many, still difficult to penetrate market. Conversely, Japan's rapidly growing semiconductor industry absorbed large amounts of foreign technology through alliances that might not have been developed by U.S. partners relying on their own resources. Joint ventures, marketing, sales, distribution, servicing, and standards coordination—relatively recent alliance types—can be traced to the dramatic growth of the Japanese market and the commercial imperative for U.S. companies to get involved.