Strategic Research Partnerships: Results of the Workshop
Albert N. Link
University of North Carolina at Greensboro
Nicholas S. Vonortas
The George Washington University
A strategic research partnership (SRP) can be broadly defined as an innovation-based relationship that involves, at least partly, a significant effort in research and development. The objectives of this workshop on strategic research partnerships were to evaluate:
- What are the policy needs for indicators related to the formation, activities, and economic consequences of alliances and SRPs? What data and indicators are currently available about alliances and SRPs? What are their strengths and weaknesses?
- How should the Science Resources Studies (SRS) Division of NSF proceed to develop SRP indicators?
This paper summarizes the salient points of the background papers that follow in this report, as well as the general discussion during the workshop. To advance NSF's interest in the possibility of developing indicators related to SRPs, the papers and the discussion appraised the potential use and the nature of such indicators. The approach was essentially based on two questions:
- Are SRP's important to the economic system? More specifically, do SRPs deserve policy attention? Are SRPs important from an economic or policy perspective?
- What data initiatives are needed with regard to understanding the economic importance or policy relevance of SRPs?
There is little doubt as to the answer to the first question. Prior research has shown that SRPs constitute an important-and probably increasing-component of the innovation system, and the papers presented and discussed at the workshop confirm this. There is a long list of reasons why this is so. In a few words, it can be argued that SRPs are socially useful because they expand the effective R&D resources applied to innovative investment.
Second, the background papers and the discussion during the workshop clearly indicated that existing data on SRPs suffer from various shortcomings. There is a need to develop systematic tracking of the incidence of the inputs and outputs associated with various types of SRPs.
Analytical and policy needs have traditionally driven the construction of new statistical indicators. Specific broad policy questions apply pressure and "pull" such indicators. The literature and, more generally, the understanding of a phenomenon is mature enough to be able to support methodologically the construction of efficient indicators. It was the consensus of opinion at the workshop that this is exactly where we now are with respect to SRPs; SRS should begin a systematic collection of statistical information related to various dimensions of SRP activity.
Section II of this paper summarizes the chronology of innovation-related indicators that NSF has developed. Section III presents our interpretative summary of the salient points from the background papers presented at the workshop and the discussion that followed as related to the first of the two objectives of this workshop: What are the policy needs for indicators related to the formation, activities, and economic consequences of alliances and SRPs? What data and indicators are currently available about alliances and SRPs? What are their strengths and weaknesses? Section IV summarizes the workshop discussion that was related to the second of the two objectives: How should the Science Resources Studies (SRS) Division of NSF proceed to develop SRP indicators?
II. Chronology of Related Indicators by NSF
NSF has had significant experience with indicators related to strategic research partnerships, R&D collaboration, and technology transfer. The chronology of the introduction of these indicators is:
- R&D expenditure flows across sectors (1972)
- Bibliometric co-authorship indicators (1980)
- University patents (1982)
- Federal technology transfer indicators (1991)
(CRADAs, patents, licenses and invention disclosures)
- Industrial joint research ventures (1991)
- R&D expenditure flows across countries (1991)
- International strategic alliances (1993)
- University-industry research centers (1993)
- Foreign location of R&D facilities (1993)
- Federal cooperative technology program dollar support (1996)
- University technology transfer indicators (1996)
(AUTM licenses, startups, revenue)
- Patent/article citation indicators (1996)
- Defense dual use programs (1996, 1998)
- Advanced Technology Program data (1998)
III. Overview of Background Papers
Nine background papers were commissioned and presented at the workshop. The salient points from these papers are:
- The mainstream theoretical economic literature may not be very useful in isolation in indicating the best approach to constructing SRP indicators. While industrial organization economists were pioneers in the early 1980s in getting a better understanding of the spreading phenomenon of inter-organizational cooperation, they have paid less attention to formally modeling the important understandings about the variety of forms of SRPs and the variety of incentives to join that have been gained since. Even though transaction costs theory and contract theory have made significant steps to explain the formation of inter-firm collaboration, it cannot be argued that formal mainstream economic theory has been able to approach the richness of concepts arising in the fields of strategic management and evolutionary economics. A combinatorial approach that would reach across narrow disciplinary approaches is considered more appropriate for the methodological approach to SRP indicators.
- A number of definitions, taxonomies, and analytical approaches to SRPs have been posited in the literature. Although academically interesting, such diversity also introduces difficulties in comparing analytical results across studies.
- The broadest possible differentiation of SRPs is between formal and informal relationships. They are both very valuable to supporting innovation. Informal relationships seem to be the larger but also the most amorphous of these two sets, presenting great difficulty in data collection. All existing systematic databases exclude informal SRPs.
- Extant databases (CATI, CORE, NCRA-RJV) are as diverse as the definition of an SRP. Even though the basic intellectual interests that motivated their construction were fairly similar, there is a significant difference between the first and the other two due to the difference in sources of original material and data collection approach.
- A major limitation of all existing databases on partnerships is the lack of information concerning the termination of partnerships.
- Another major limitation of existing databases is the lack of performance measures for partnerships. This seems to be primarily the result of continuing uncertainty among experts concerning the level at which performance is judged (e.g., at the level of the partnership or the individual member) and the yardsticks that can be used.
- Yet another major limitation may be that small firms may be underrepresented in most existing SRP databases. Possible reasons for that may be that small firms often do not formalize their R&D and probably their partnerships too. One implication is that an inference that SRPs are relatively less important for small firms than they are for large firms could well be mistaken. Another implication is that we may need to take a more serious look at informal partnerships than has been done in the past. There is thus some ground for arguing the necessity of a survey that would carefully reach small as well as larger firms. Some experts do, however, doubt the reliability of survey answers by a significant cohort of small firms, particularly those in non-R&D intensive activities.
- Formal SRPs are just one of the ways firms may use to link to each other and to other organizations. Consequently, it is argued that SRP indicators will underestimate the extent of collaboration. To get a more complete picture, it may be useful to combine SRP indicators to others such as bibliometric and patent indicators.
- CRADAs is an important form of SRPs with potentially a lot of information, which has yet to be collected extensively and systematically. SRS has been reporting aggregate data on CRADAs, but there is a sense that much broader coverage is warranted. Output indicators are also needed.
- A second important differentiation of SRPs is between public-private and private-private. Which one of these is of interest crucially depends on the question. The presumption that public-private SRPs may be more important for public policy makes the implicit strong assumption that public policy is limited to contributing resources to SRPs. Policy decision-makers are clearly interested in other policies as well that affect the economic and regulatory environment in which SRPs operate. Knowledge about both sets of SRPs is useful for policy.
- It would be useful to researchers and policy analysts if existing and to-be-created SRP indicators could be linked to other data related to economic and technological performance of individual organizations that other government agencies collect (e.g., market performance of individual firms and patent portfolios). Such bridging of data is expected to produce direct benefits to SRP analysis, not least because it will allow creating control groups of non-SRP participants.
- One often meets in the literature efforts to approximate the performance of SRPs with patents. A suggestion has been that collaboration in R&D should result in co-inventing and co-patenting. Unfortunately, co-patenting is much too little to be relied upon as such an indicator. It appears that firms have traditionally avoided co-patenting, probably due to expectations of increased legal problems. Although co-patenting numbers are increasing, they are still too low relative to the overall level of patents issued annually in the United States.
- One other set of indicators that may be naturally linked to SRP indicators is the so-called innovation indicators. Such indicators are being constructed in the European Union and a few individual country members. NSF is also considering them and currently fielding a pilot survey in the information technology area. For example, one possibility could be to use the same survey instrument for both general innovation indicators and SRP indicators. There is some skepticism among experts concerning the feasibility and reliability of innovation indicators - having to do particularly with the definition of an innovation and the willingness of respondents to identify failed innovations. Still, in the lack of better alternatives, many consider this worth the try; it is expected that there is a learning curve in raising the quality of collected statistics.
- NSF is clearly not the only agency considering SRP indicators. The Organization for Economic Cooperation and Development has also developed a strong interest in the subject, in relation to several other concurrent activities. The latter include: the revision of the Frascati manual to better define R&D, partnerships, outsourcing, etc.; the creation of a Globalization manual; the revision of the Oslo manual on innovation surveys. There is a strong sense at the OECD that this is the right time in the life cycle of the phenomenon for public agencies to get involved in defining and building statistical information on SRP indicators.
- One of the more interesting international phenomena in SRPs during the past couple of decades has been the relative increase in the proportion of contractual agreements vis-a-vis traditional, equity-based joint ventures that create new organizational entities. This is an important development from both the business strategy and public policy points of view. It is also a major complicating factor in collecting data for SRP indicators. Many of these contractual agreements are synthetic. That is, these agreements include various activities, thus introducing a problem of classification, and apparently are of more informal nature than required for accurate reporting.
- It is now clear that activity in three technological areas has been responsible for the vast majority of observed SRPs during the past couple of decades. The largest by far is information technology. Biotechnology comes next. Advanced materials follow.
- SRP intensity and characteristics may vary with the field and the technological life cycle. A possible consideration is to link SRP data to some kind of a life cycle model.
- Specifically to biotech, SRPs are extremely useful in the research, production, and delivery stages of products. Biotechnology is, of course, an area with particular characteristics, including greater than average reliance on academic research, proliferation of small firms, rapid turnover of firms, a strong effort by the big pharmaceuticals to avoid falling behind, extensive patentability of research outcomes, significant availability of venture capital. There are already four proprietary databases relating to biotechnology SRPs, with reported limitations in terms of scope of data, cost, and quality.
- Biotechnology SRPs frequently emerge among people with prior working relationships. If the observation is accurate, as it seems to be, human (scientist) mobility and SRP formation indicators should correlate strongly. It would be interesting to investigate the relationship between human mobility and partnerships in other technology areas as well.
- The proliferation of SRPs and the observation of a large failure rate of SRPs have naturally raised the question of performance. Early attempts to approximate SRP success by the longevity and stability of the relationship were more appropriate for equity-based partnerships than for most cases of contractual agreements. It is also clear that different partners have different objectives in the same partnership, making the definition of partnership performance problematic. This has obvious implications for data and indicators of performance. It is quite possible that there are no widely applicable objective indicators of success at the level of the partnership. SRS may have to make do with indicators of success at the level of the individual partner.
- A major complication in constructing partnership performance indicators reflects a selection problem. That is to say, if SRP participants tend to perform better than non-participants, is the difference the result of SRP participation or the result of SRP participants performing better irrespective of membership?
- Two important features of SRPs that have received a lot of attention by business analysts and much less so by all others is the multi-dimensionality of relationships and their organizational structure. As far as NSF's effort is concerned, the danger is with expanding the scope of the indicators very much to make them intractable and with crossing the borders between formal and informal SRPs, which increases the degree of fuzziness.
- The construction of SRP indicators must take into account the strong current interest in innovation environments and in the increasing role of networks as the reference point for individual innovations. Significant effort must be devoted to argue over the appropriate unit of analysis which could be the individual, the group, the business unit, or the firm. Ideally, one would like to have information at different levels.
IV. Developing SRP Indicators by SRS
It emerged from the background papers and the expert discussion during the workshop that:
- SRPs have important effects on innovative investment behavior and performance;
- Public policy can affect the performance of SRPs, both directly in certain circumstances and indirectly by changing the rules of the game;
- Currently available data on SRPs are limited in several respects, not systematically gathered or coordinated, and uneven in quality and degree of coverage.
Experts believe that there is a need for ongoing, systematic, and coordinated documentation and reporting of:
- Incidence of SRPs
- Types of SRPs
- Inputs into the formation of SRPs
- Output of SRPs
- Role of SRPs as inputs into the innovation process.
There is a clear sense that multiple measures of inputs and outputs, each reflecting the particular circumstances of a type of SRP, are useful and appropriate. Appropriate data could and should be integrated from a variety of sources in order to be useful in answering various, and quite different, policy- and strategy-oriented questions that analysts ask in relation to SRPs.
It would certainly prove useful to have information on:
Such lists are limited only by the imagination of the analyst, however, and are often beyond the means of responsible agencies to collect.
- number, industrial affiliation, and size of participating firms in SRPs
- organization structure of SRPs and types of members
- overall budget and individual contributions to SRPs, including government cost-share
- reasons for the termination of SRPs
- SRPs outputs, such as new products, patents, publications, licensing agreements, launching of startups, new jobs
- Impact of the SRP experience on the performance of individual participants.
Four important considerations for SRS were discussed:
- Is this the right time for direct SRS involvement in building SRP indicators?
- What would these indicators be?
- What indicators are also applicable in the international context?
- What are the policy issues we are trying to inform?
The discussion covered a wide range - it was meant to be brainstorming, after all. This was the first time that such a gathering took place in the United States. Nonetheless, several important indications were signaled to SRS representatives. They included:
- On the question of whether coverage should be limited to public/private SRPs, or to private-private SRPs, or be extended to both kinds, the outcome was balanced towards the latter.
- Public agency support of public/private SRPs should be collected.
- Questions could be added to existing NSF surveys to elicit information on the extent of private organizations' resources devoted to SRP activity.
- Additional survey questions should address the formal versus informal nature of SRPs.
- While the utility of a broad question asking respondents to estimate the share of formal and informal partnerships was felt to be significant, the difficulties in definitions and estimation seem to make it difficult to operationalize.
- The human mobility aspect of SRPs have not been addressed by any of the existing SRP databases.
- Data on international collaboration are important in spite of complicated definitional problems.
- An argument was made over the distinction between and value of understanding the extent of collaboration and understanding the factors driving it. Budget limitations may oblige SRS to follow a stepwise approach, however, limiting the coverage to the first leg as a start.
- Proceeding in parallel with survey data collection and with case studies of SRPs may reveal unanticipated trouble spots regarding definition, inputs, outputs, budgets, and so forth.
- Additional information should be collected on the experience of international agencies in developing SRP indicators and building appropriate survey instruments.
Although the workshop produced a useful dialogue, it was by no means conclusive. It clearly underlined the importance of a continuing, proactive discussion between experts and the SRS regarding the most appropriate procedure for developing SRP indicators.