Strategic research partnerships (SRPs) have caught much attention from many perspectives. Firms are concerned because forming SRPs has become an important complement to their in-house R&D. It has been documented that firms increasingly rely on collaboration with other firms to conduct R&D activities (Gulati, 1995; Powell, Koput and Smith-Doerr, 1996; Osborn and Hagedoorn, 1997). Governments are concerned because they consider cooperative R&D as a tool for enhancing industry competitiveness. Japan is regarded as a forerunner in the practice of cooperative R&D. The most celebrated example is the VLSI (Very Large Scale Integrated circuit) project, designed to help Japan catch up in semiconductor technology. The project, conducted between 1975 and 1985 with a budget of 130 billion yen ($591 million) of which 22% was financed by the government, developed state-of-the-art semiconductor manufacturing technology. All of the major Japanese semiconductor producers participated in this project, and Japanese semiconductor companies gained world leadership after the project. It is widely believed that this success story is only one of many.
The perceived success of the VLSI project has motivated other countries to emulate "Japanese style" collaboration. The 1984 U.S. National Cooperative Research Act was enacted to relax antitrust regulations in order to allow the formation of research joint ventures. Major cooperative R&D projects followed. SEMATECH was established in 1987 to develop semiconductor production technology with a $1.7 billion budget as of 1996, half of which was financed by the government. The Department of Defense sponsored cooperative ventures on the development of flat-panel displays which will spend an estimated $1 billion over five years beginning in 1994. A successor bill of the 1984 law, The National Cooperative Research and Production Act, was passed in 1993 to extend the 1984 law to not just research and development, but to production of new technologies as well.
In Europe, the block exemption from Article 85 of the Treaty of Rome, which determines EEC competition rules for certain categories of R&D agreements, was introduced in 1985. Even earlier, many cooperative R&D projects were organized, including the $5.6 billion European Strategic Programme for Research and Development of Information Technology (ESPRIT) project in 1984, and the UK Alvey project in 1984, both for the development of computers and information technology. These projects were in response to another famous Japanese cooperative R&D project, the Fifth Generation Computer Project. Other European efforts include programs under the European Research Coordination Agency (EURECA) started in 1985.
In developing countries, there are similar efforts. The Korean government has launched a series of cooperative R&D projects whose scheme is very close to the Japanese one, and in Taiwan there is its own version of R&D consortia.
Given the importance of SRPs in general, and the Japanese SRPs in particular, this article focuses on SRPs in Japan. Section 2 argues the role they play in Japan, and some data are introduced. Empirical evidence is presented in Section 3. In Section 4, the strength and limitation of the data used in these empirical analyses are discussed, and the future direction of the data collection and empirical evaluation are presented.
SRPs play a very important role in Japan because of its distinctive institutional settings. First, the importance of SRP as a vehicle to share knowledge with other firms becomes more important where imperfections of factor markets are severe (Sakakibara, 1997b). The lifetime employment system prevalent among large corporations is a cause of low mobility of researchers among companies. Companies are oriented to maintain a stable number of researchers, and so even if they recognize new technological opportunities, it is hard for them to suddenly increase hiring. Also, though the situation is changing recently, we seldom observe researchers move from one company to another, especially to a competitor. Saxonhouse (1985) pointed out that the Japanese government's cooperative R&D projects are viewed as a substitute for the unusually high degree of informal interfirm communication which takes place among the more professionally oriented, potentially mobile R&D personnel in the United States. American researchers might be implicitly disclosing potentially proprietary information in order to enhance their employment prospects, and also in order to receive in exchange proprietary information of commensurable value. Without having spillover channels through recruiting, Japanese companies are motivated to use SRPs as a means of information exchange.
Second, mergers and acquisitions (M&As) as an alternative instrument to access research inputs tend to be cumbersome. Compared with the United States, M&As are still uncommon in Japan because the dominant owners of corporations are institutional shareholdersoften motivated to solidify relationships not to seek immediate returns. Rules which facilitate M&As are less developed than the United States. Moreover, R&D knowledge and capability belong to individuals (von Hippel, 1988), not to firms, and so acquisitions intending to capture R&D capability can turn out to be purchases of "empty shells" due to the departure of all key personnel with the "crown jewels." Under these circumstances, cooperation with other companies becomes a practical alternative (Sakakibara, forthcoming-a).
Third, relatively weak research capability in universities and national research laboratories, and the weak linkage between these public research organizations and corporations in Japan, make knowledge transfer among firms through SRPs important. In the United States, strong university-based efforts and university-firm linkages work as a substitute for knowledge sharing through SRPs (Sakakibara 1997b).
Fourth, firms are often motivated to form SRPs as a means of internal diversification. SRPs are directly connected with diversification in Japan because, as Porter (1992) points out, entry into new businesses is typically conducted by established firms through internal diversification. Japanese companies tend to face weaker pressure from shareholders than U.S. firms to realize short-term returns, and so a goal of them is their own perpetuation. Due to the underdevelopment of the market for corporate control, resource reallocation from mature and/or declining businesses to emerging businesses is conducted internally. Through the participation in SRPs, firms can test the possibility to diversify into new businesses. Sakakibara (forthcoming-a) empirically finds that the motives for cooperative R&D are analogous to the motives for diversification.
Because of these critical roles SRPs play, we expect to observe many SRPs in Japan. The exact number of SRPs is difficult to obtain because "pure-private" SRPs are not often announced, and so the journal-article database can be biased. Corporate executives have noted that the existence of private SRPs itself could be a signal to rival companies regarding which research direction companies try to seek. This is the primary reason that the rest of the paper is based on the data of government-sponsored R&D consortia in Japan.
The promotion of cooperative R&D by the Japanese government started in 1959, when the Ministry of International Trade and Industry (MITI) and aircraft makers launched the YS-11 turboprop aircraft development project. In 1961, a formal scheme to promote cooperative R&D efforts was established as the Act of the Mining and Manufacturing Industry Technological Research Association. Under the Act, which was modeled after the British Research Associations initiated in 1917 and later adopted by Germany, France and Sweden, firms can pool researchers and funds into nonprofit Mining and Manufacturing Technological Research Associations (TRAs hereinafter). The formation of TRAs was intended to promote R&D consortia as a means of coping with trade liberalization and to enhance the productivity of Japanese industries. At that time, Japan faced the task of abolishing protective policies for domestic industries following these industries' recovery from the devastation of the Second World War.
Under this scheme, participating companies enjoyed several tax benefits on their research expenses. Typical tax benefits included accelerated depreciation for expenses on machinery and equipment, instant depreciation of fixed assets for R&D, and discounts of property taxes on fixed assets used for R&D (the Council of the Mining and Industry Technology Research Association, 1991). The TRA system was introduced as a substitute for direct R&D subsidies to individual companies, which the Japanese government had to phase out as Japan prepared to join the league of developed countries and to abolish protective policies. After the scheme of TRAs was introduced, the amount of R&D subsidies to individual companies considerably declined, and in order for firms to receive significant amounts of R&D subsidies, they needed to form R&D consortia.
TRAs are not the only form of cooperative R&D in Japan. Other organizational forms for cooperative R&D include foundations and corporations. These forms are chosen by participants on the basis of each form's financial and organizational benefits (for details of different types of cooperative R&D, see Sakakibara, 1997b). It is not only MITI, but also many other ministries that are involved in the formation and operation of these consortia.
The most comprehensive data on SRPs have been collected and documented in Sakakibara (1994, 1997a,b), which include 237 government-sponsored R&D consortia which occurred between 1959 and 1992. 1171 companies participated in these consortia during this period and many were involved in multiple projects. Inclusion of these multiple projects yields a data set with 3021 company-project pairs. They cover all the identifiable government-sponsored R&D consortia during that period including all the TRAs as well as other forms of cooperation.
Figure 1 illustrates the overall trend of Japanese government-sponsored R&D consortia in terms of the budget allocated in each project, aggregated by sector. This figure illustrates that the efforts peaked in the late 1970s and 1980s. This figure also shows that while the electronics and machinery sector caught much attention, consortia are observed in many other sectors as well.
Source: Sakakibara (1994).
The previous section establishes a critical role SRPs play in the Japanese institutional setting. This is a necessary but not sufficient condition to call for the government's support for this particular policy instrument. If the government supports SRPs that would have been formed without government sponsorship, there is no need for government intervention. Also, if government-sponsored SRPs do not achieve intended goals to stimulate private innovative activities, the validity of their existence becomes doubtful.
In order to determine whether the government should promote SRPs, empirical examinations on the existing SRPs are informative. A natural question for policy makers is whether SRPs they promote attract the right kind of firms from a public-policy perspective, and whether the SRPs they support are an effective means to stimulate private R&D efforts rather than crowding out private spending. Also, the research productivity of SRPs is an important consideration.
Supporting SRPs is only one of many policy tools government can choose to stimulate innovative efforts. For example, public procurement, funding of research in national laboratories and universities, tax incentives, subsidies to individual firms can be chosen as a policy means. The evaluation of the effectiveness of SRPs by measuring their productivity thus gives the government a useful guidance regarding whether they should choose SRPs over other policy means.
This section focuses on three issues regarding SRPstheir formation, their effect on R&D spending of participating firms, and their productivity. Empirical studies on Japanese SRPs for these issues are limited. The paucity of empirical research is largely due to the insurmountable task of obtaining data on government-sponsored SRPs. There is no central clearing house for such data, and even within MITI, the largest sponsor of SRPs, no single place from which one can obtain the whole data of MITI-sponsored SRPs.
In the analysis of the formation of SRPs, an important issue for policy makers is the firm- and industry-level determinants of the formation of R&D consortia. At the industry level, there are two important considerations. The first is the degree of competition of the industry in which member firms of a SRP belong. The degree of industry competition will affect a firm's propensity to participate in SRPs in two opposite ways. As Katz (1986) discussed, firms in competitive industries might be more motivated to form R&D consortia to ease the subsequent product market competition. Also, SRPs allow firms to access complementary technology that enables firms to develop their R&D capabilities and improve their strategic position. These needs might be greater in competitive industries (Baum and Oliver, 1991; Eisenhardt and Schoonhoven, 1996). On the other hand, organizational economics and organizational theory document the difficulties involved in organizing cooperative ventures in general (Killing, 1983; Harrigan, 1985, 1986; Pucik, 1988; Borys and Jemison, 1989), and SRPs in particular (Doz, 1987; Hladik, 1988; Osborn and Baughn; 1990; Jorde and Teece, 1990). These studies emphasize the organization costs associated with complex ventures, including costs to monitor opportunistic behavior of participants, and to align interests among participants. If firms are in highly concentrated industries, they might find the cooperation (or collusion) easier to achieve because these difficulties can be resolved in an oligopolistic understanding among rivals.
SRPs consortia can be used as vehicles to internalize the externality created through spillovers of research outcomes (Spence, 1984). Firms can agree to share the costs and outputs of an R&D project before its execution, so they can restore the incentive to conduct R&D. Cohen et al. (1998) found that, on average, intra-industry R&D spillovers are more extensive in Japan than in the United States. One major channel that facilitates spillovers is the patent system. In Japan, patent applications are automatically published 18 months after their initial filing. In the United States, in contrast, the content of the patent applications will be published only if they are granted, which is typically more than two years after the application. Under these conditions, spillovers can be major issues determining the participation in SRPs in Japan.
Firm-level factors can also influence the rate of participation in SRPs. The first factor that plays an important role is R&D capabilities of participants. Firms can use SRPs to gain access to technological capabilities of other firms to create next-generation technological competencies. This might imply that a firm that currently has disadvantageous R&D capabilities is motivated to form R&D consortia more than R&D-capable firms. However, Cohen and Levinthal (1989) demonstrated the possibility that a company's own R&D increases its learning capability from others. Firms that already invested in R&D, therefore, can benefit more from R&D consortia than less R&D-capable firms, and so they might be more motivated to learn from others.
The second factor of consideration is the experience of past participation in SRPs. The network of prior alliances provides information of new alliance opportunities, potential partners and their quality (Kogut et al., 1992). With the formation of new alliances this network updates, it is the new network that becomes influential for subsequent firm behavior (Gulati, 1999). In the case of SRPs, the experience of participation in past consortia can create technological network through which a firm can gain access to technological resources of other firms. Furthermore, Baumol (1993) argued that cheating in the cooperative R&D game can be easily detected in a repeated game situation, and punishment to exclude a cheater from the following projects is very costly for a cheater. Therefore, firms that have repeatedly participated in R&D consortia can benefit from the sustained cooperation, which further motivate them to participate.
The third factor is the encounter with other firms in product markets. The literature on networks stressed a firm's access to external networks as an important source of capabilities that the firm can draw upon (Gulati, 1999; McEvily and Zaheer, 1999). When a firm is diversified, it is likely that the firm has a better knowledge on potential partners in SRPs through contact with a large number of firms in many product markets. In other words, contact with other firms in product markets constitutes a network through which the firm can obtain superior information on future consortia. Also, when a firm is diversified into many product markets, the firm might wish to draw on outside knowledge with a greater extent by combining in-house technological competencies and external technological acquisitions to serve these markets (Gtanstrand, Patel and Pavitt, 1997). This desire further motivates the firm to form SRPs.
Sakakibara (2000) finds from an event-history analysis that both industry and company factors affect the formation of R&D consortia. It is found that a firm in an industry with weak competition and appropriability conditions has a higher rate of consortia participation. A firm's R&D capabilities, network formation through past consortia, encounter with other firms in product markets, age, and past participation in large-scale consortia also positively affect its tendency of consortia formation. This indicates that firms, which frequently participate in SRPs, are the ones that will gain most from participation and have potential for effective cooperation. Policy makers need to recognize that they need to take both industry and company factors into account when deciding the target firms. Even if they want to attract a specific type of firms to government-sponsored SRPs, it might be difficult to do so if these conditions are not met.
The second issue of the interest is the effect of participation in SRPs on R&D spending of participating firms. Do SRP member firms increase or decrease their R&D spending? An answer to this question depends on the motives of participants and the organization of SRPs.
Sakakibara (1997a) makes a distinction between cost-sharing SRPs and skill-sharing SRPs. Cost-sharing SRPs refer to consortia formed to share fixed costs of R&D among participants, to realize economies of scale in R&D, and to divide tasks among members and avoid "wasteful" duplication. In contrast, objectives of skill-sharing SRPs include to learn from other participating firms. This type of SRPs can be viewed as opportunities for one partner to internalize the skills or competencies of the others to create next-generation competencies. This learning function of SRPs becomes especially important when firms try to enter a new business, to redefine their core industries, or when they respond to shifting industry boundaries.
The diversity of capabilities SRP participants possess can distinguish the different motives to participate in SRPs. In the case of skill-sharing or learning-based R&D cooperation, what is important is not only the outcome of the project, but also the process of resource accumulation, or learning in a SRP. Participants with a skill-sharing motive might find it easier to reach an agreement to cooperate without a clear end result in mind than firms whose primary motive for cooperation is cost-sharing. In addition, skill-sharing is an important means for a firm to enter a new business, implying that this motive is more likely in pre-competitive R&D where conflicts of interests are less apparent. Firms from different industries, therefore, might find it easy to cooperate when their motivation is skill-sharing. Also, the capabilities of participants in skill-sharing ventures are likely to be heterogeneous so as to best combine complementary resources and knowledge. This implies that participants are likely to come from a wide range of industries.
In contrast, cost-sharing, or scale-based R&D cooperation requires a relatively clearer understanding of the objective and configuration of a cooperative R&D project, because the benefits of cost-sharing and the realization of economies of scale has to be understood by member firms before the execution of the project. Participants in R&D consortia motivated by cost-sharing are likely to belong to a single industrial sector, because they are more likely to have similar prior knowledge, which makes the agreement easier to achieve. Their capabilities are, therefore, likely to be homogeneous.
The cost-sharing and skill-sharing motives are not necessarily mutually exclusive. An R&D consortium can pursue both motives simultaneously. The point here is that the relative importance of these motives can be distinguished by the diversity of capabilities among the consortium's participants. Sakakibara (1997a) finds that the relative importance of the cost-sharing motive in R&D consortia increases when participants' capabilities are homogeneous or projects are large, while the relative importance of the skill-sharing motive in R&D cooperation increases with heterogeneous capabilities, based on the survey data on Japanese government-sponsored SRPs.
The effects of SRPs on a participating firm's R&D spending will differ according to their motives for participation, and thus the diversity of member firms in SRPs. There are three ways that the diversity of R&D consortia participants affects the R&D expenditures of participating firms (Sakakibara, forthcoming-b). The first is the spillover effect of a firm's own R&D on others' R&D productivity. When the outcomes of SRPs are pooled and shared, firms find it best to increase their R&D efforts. The spillover effect is larger if a degree of knowledge complementarity among participants is higher, because firms can achieve better outcomes by combining their knowledge. Assuming that the diversity of participants increases the degree of knowledge complementarity, this diversity implies higher R&D expenditure.
The second effect of cooperative R&D on a firm's R&D spending is from learning, which is defined here as efforts by firms to assimilate and exploit knowledge or information generated by other firms (Cohen and Levinthal, 1989). Suppose that higher R&D expenditure facilitates better learning capability. Levin et al. (1987), for example, point out that independent in-house R&D is the most effective means to learn about rivals' technology. It is also documented that Japanese companies participating in consortia customarily set up in-house research groups to absorb and utilize the results of R&D consortia (Kodama, 1985). If there are better learning opportunities created by SRPs, the participants spend more in their R&D. Assuming that diversity of participants increases a degree of knowledge complementarity and thus learning opportunities, diversity implies higher R&D expenditure.
The third effect relates to the impact of R&D cooperation on product market competition. The more direct the product market competition among the participants, the less willing they will be to cooperate even if they own complementary knowledge. Katz (1986) argues that if higher levels of R&D make market competition more intense by lowering firms' production costs, then the resulting decline in profits will reduce their incentive to conduct R&D. Katz showed that R&D consortia could depress R&D as firms seek to lessen the severity of competition in the product market. In the case that participants are from more diverse industries (as opposed to coming from a single industry) in the product market, however, this argument implies higher R&D expenditure by consortia participants, since the market-competition effect is expected to be smaller in this case.
All three effects suggest the possibility that R&D consortia whose members have diverse backgrounds may increase participants' R&D spending, relative to consortia of single-industry participants. Note, these three effects are not necessarily mutually exclusive. The spillover effect and the learning effect are the results of technological diversity of consortia participants, while the product-market-competition effect is related with the degree of direct competition among participants in product markets. Sakakibara (forthcoming-b) finds from firm-level financial data and consortia data that when SRPs consist of firms with diverse technological knowledge, these firms offer learning opportunities and increase spillover productivity, which result in more intensified R&D efforts of participants. When R&D consortia participants have diverse business backgrounds, the expected product market competition is less intense, leading to higher R&D expenditures by participants. Sakakibara (1997a) also finds from the survey-data that when the skill-sharing motive for participating in cooperative R&D becomes relatively more important than the cost-sharing motive, a firm's R&D spending is likely to increase, consistent with results based on quantitative data.
The third, and perhaps most important issue, is the determinant of the performance of SRPs. There are multiple levels one can approach on this issue. The first level is the overall impact of the participation in SRPs on research productivity of participating firms. As explained earlier, there are many reasons we expect a positive relationship between the participation and the increase in research productivity of participants, including the cost- and skill-sharing effects. Branstetter and Sakakibara (1998) examined the data on Japanese government-sponsored R&D consortia. They found that if a firm participates in an additional project per year, it would raise its patenting per R&D dollar (i.e., its research productivity) by between 4% and 8%.
A more disaggregated approach is to identify the characteristics of consortia that are associated with the increase of research productivity of participating firms. Branstetter and Sakakibara (forthcoming) examined the same data. They focused on two major characteristics of SRPs: spillover potential and ex-post product market competition among participating firms. Theoretical literature, Katz (1986) and others, predicts that the greater the potential levels of R&D spillovers within consortium, the greater the level of R&D by member firms. This is because when a firm can benefit more from R&D outcomes by other member firms through higher spillovers, the firm is motivated to conduct more R&D, leading to better research outcomes as SRP members. On the other hand, some of the private benefits of cooperative R&D, in terms of raising firm profits, could be dissipated through product market competition. When the level of product market competition among participating firms is not intense, a participant can appropriate all the returns R&D outcomes, motivating member firms to conduct more R&D and achieve greater outcomes. Branstetter and Sakakibara (forthcoming) measure spillover potential as technological proximity among member firms in the technological space, calculated from member firms' patent portfolio, and the level of ex-post product market competition as the product market proximity of member firms. Their outcome measure is the number of patents taken by consortia participants in technological areas targeted by consortia. They find positive association between technological proximity and consortium outcomes, and a negative relationship between product-market proximity and consortium outcomes. In addition, they employ qualitative characteristics of consortia taken from survey results by Sakakibara (1997a, b), and find that these consortia are most effective when they focus on basic research.
Sakakibara and Branstetter (2000) apply the same methodology to the data of U.S. consortia, sponsored by the Advanced Technology Program of the Department of Commerce. They find similar results in the U.S. case: There is a positive association between the intensity of participation in research consortia and the overall research productivity of participants. There is also a positive impact of consortia on the research productivity of participants in the technological areas targeted by the consortia. This positive impact of consortia is higher when the average technological proximity of participants is high. In both Japanese and U.S. SRPs, there is evidence that R&D intensive firms tend to benefit more from the participation in consortia.
Sakakibara (1997b) conducted an analysis of the performance of R&D consortia from a managerial perspective. The results show that there is no clear link between the existence of R&D consortia and industry competitiveness. This study also investigates the perceived benefits and costs of Japanese R&D consortia based on 398 responses to questionnaires distributed to high-level corporate R&D managers who have participated in R&D consortia. The perceived benefits of projects are rather intangible, such as researcher training and increased awareness of R&D in general, not the commercialization of project outcomes. The overall subjective evaluation of the typical project's success is modest, and participants do not perceive R&D consortia to be critical to the establishment of their competitive position. This finding of the positive but modest benefits of SRPs is consistent with the finding from econometric studies.
There is a fundamental issue that researchers have to cope with when they conduct evaluation studies of government-sponsored R&D. Participation in SRPs, or the selection of member firms by the government, is not a random event. To the extent that they could, governments seek to encourage firms with strong R&D programs to participate in their sponsored SRPs because they want to maximize the returns from government subsidies. As a result, if we observe good outcomes from certain types of SRPs, we cannot distinguish whether these types of SRPs are designed to yield good outcomes, or if only good firms participate in these particular SRPs. This selection problem is the single greatest limitation of past research to measure the impact of public-technology programs (Klette, Moen, and Griliches, 2000).
The data obtained by Sakakibara (1994, 1997a,b) make it possible to employ several techniques to deal with the selection problem. The data contain multiple dimensions of information (SRP, firm and time). At the SRP level, they include a description of each project, its period of operation, the total budget, the amount of government subsidy, and the names of participating firms. At the firm level, the data include all the financial information. They also contain not only input data (R&D expenditure) but also output data, measured as the number of patents taken by participating firms, both the overall patenting and patenting in the targeted area. These data are available over a long period of time; the most detailed data are available from the early 1980s to the mid 1990s. In addition to these quantitative data, qualitative evaluations of managers from participating firms are obtained through questionnaire survey.
The analyses of Branstetter and Sakakibara (1998, forthcoming), for example, demonstrate a way to address the selection problem by utilizing these data. By employing the data of patenting in the targeted technologies before, during, and after participation in a consortium by individual firms, they can control for pre-existing technological strength of a firm in the targeted technologies, which enable them to isolate an additional effect due to the participation.
Also, the data set includes observations on firms that did not participate in consortia. This dimension helps to highlight the effect of participation. Even if we observe any increase in R&D outputs by a participating firm during the period a SRP operated, we cannot conclude that increase is due to the SRP. It might be the case that technological opportunity in that field increased, or the overall economic condition was favorable. By having firms, which did not participate as a control, we can "extract" the pure-participation effect.
Finally, because we observe the same firms participating in multiple consortia, we are able to measure the marginal impact of different consortium characteristics and firm characteristics on research outcomes, controlling for consortium and firm fixed effects. A conceptual experiment this data set allows is, for example, to examine how the same firm would perform if we moved it from a consortium with a set of characteristics to one with a different set of characteristics. In a similar manner, we can estimate the impact of different firm characteristics; in other words, we can determine what kinds of firms benefit most from SRPs.
Qualitative and quantitative data offer different advantages. R&D output in this data set is measured by the number of patents generated by firms. There are limitations inherit in the patent data. For example, some innovations are not patentable. The impacts of learning by researchers in SRPs are not even codifiable. Survey data provide us with qualitative aspects of the outcomes of SRPs, and it is best, as discussed earlier, to utilize both types of data to evaluate SRPs.
Though we can learn a lot from existing data, certainly they are not perfect. First, we need multiple measures to evaluate the outcomes of SRPs. The ultimate goal of SRPs is the commercialization of research. It is very difficult to map from SRPs to the eventual commercialization of the targeted research project, however, because there is a time lag between a project and a commercialization: this time lag is project specific. Also, participating firms need to make their own efforts after the conclusion of SRPs. It is therefore difficult to quantify the exact contribution of SRPs on the eventual commercialization. Any data that help the mapping from SRPs to commercialization would be useful.
Also, it is very helpful for policy makers to obtain data of pure private SRPs. As already discussed, journal-article database might have a bias toward "hot" fields such as information technology, because they are frequently covered by media. Having data of pure private SRPs, policy makers can compare them with government-sponsored SRPs, and evaluate the marginal effect of government support. Given the increasing importance of SRPs, more data on them are helpful not only for policy makers but also for managers who consider participating in SRPs and are trying to maximize returns form participation.
 Based on the interview by the author.
 This section draws heavily on Sakakibara and Cho (2000).
 The arguments in this part draw heavily on Sakakibara (2000).