Chapter
7. CONCLUSIONS AND ISSUES FOR FUTURE RESEARCH
This study empirically
evaluates the impact of ATP-funded consortia on the research productivity
of participating firms. We find evidence that the impact of participation
on research productivity of participating firms is positive at all
levels. There is a positive association between the intensity
of participation in research consortia and the overall research productivity
of the participating firms. At the consortium level, we continue
to find a positive impact of consortia on the research productivity
of participating firms in the technological areas targeted by the
consortia. Furthermore, this positive impact of consortia is higher
when the average technological proximity (that is, the degree to
which patenting portfolios of participating firms are similar) of
participating firms is high, and when consortia participation stimulates
the participants’ creative thinking and contributes to their
R&D cost and time saving. Larger firms with higher R&D budgets
tend to benefit more from participation than other firms. The economic
significance of this finding is unclear due to the flaws of our data
set. Viewed together, however, our results demonstrate that consortia
participation is leading to verifiable, measurable increases in research
productivity.
We compared our U.S.
findings to results from Japanese government–sponsored R&D
consortia, which allowed us to examine the long-run impact of consortia
on patenting outcomes. Japanese data indicate that participation
in an additional consortium increases patenting between 4% and 8%,
similar to our finding in this study. Japanese data also suggest
that the benefits of research consortia are observed long after the
inception of the project, with a surge in patenting following the
official cessation of the consortia. This implies that the relatively
short time series of data available on firms that participated in
ATP-funded research consortia will tend to underestimate the total
impact of participation.
We also examined the
impact of two consortium characteristics using Japanese data: spillover
potential (as measured by average technological proximity of firms
within a consortium) and product market proximity. As expected, Japanese
data show a positive association between technological proximity
and research outcomes and a negative association between product
market proximity and research outcomes. These results have important
implications for our analysis of U.S. data. First, they confirm the
importance and robustness of technological proximity as a predictor
of consortium success in a data set with a much longer, more complete
time-series dimension. Second, the negative effect of product market
proximity on consortium outcomes suggests that bringing product market
rivals into a consortium is unlikely to produce a successful pattern.
Few ATP-funded consortia have this structure, and that is probably
beneficial for ATP.
Although Japanese data
confirm several of our U.S. findings, we are mindful of the limitations
of our analysis due to several data constraints. First, the lack
of information on a number of small, privately held firms that were
involved in ATP-funded consortia affects the cross-sectional dimension
of our data set. Second, the REI patent data effectively ends in
1995, just as the ATP was expanding its support of research consortia.
This affects the time series dimension of our data. Measurements
of consortium outcomes based on only a few years’ data are likely
to underestimate the full effect of consortia participation on member
firms. In addition, our analysis using ATP’s Business Reporting
System (BRS) survey data on firms’ perceptions of the impact
of participation in consortia on their own research and development
was limited because of confidentiality constraints.
The limitations in
our data set could be addressed in several ways. First, the USPTO’s
Automated Patent System is an on-line database that allows users
to download data on hundreds of thousands of recent patents. Using
this database, our patent series could be updated to include recent
granted patents.
Second, ATP staff economists
could analyze individual firm-level responses to BRS survey questions
without jeopardizing the confidentiality of these data. Combing the
qualitative BRS data with quantitative measures developed in this
study will create a richer, more complete picture of the impact of
consortia than we were able to achieve.
Third, future work
could extend the time-series dimension of our analysis by updating
firm R&D spending, capital investment, and industry affiliation
through Standard & Poor’s COMPUSTAT database. Firms’ innovative
inputs could be updated using the most recent version of this database.
Fourth, the consortium-level
and firm-level analyses discussed in this report depend on the construction
of a mapping from the stated technological goals of ATP-funded research
consortia to the relevant patent classes of the U.S. Patent and Trademark
Office’s patent classification system. As described in the Appendix
on data construction, we employed an outside consultant, Bailey Services,
Inc., to construct this mapping at a nominal price. We believe that
the accuracy of the mapping they provided is high enough for the
exploratory analysis conducted in this report. However, we strongly
suggest that ATP’s Economic Assessment Office consider investing
an additional $5,000–$10,000 in the creation of a truly comprehensive
patent mapping. Such a document could be a useful evaluation tool
for years to come.
In this study, we have
accomplished our mission of developing a framework that allows ATP
to quantitatively measure the impact of research consortia on the
research productivity of participating firms. We hope that ATP researchers
will be able to apply this framework, build upon our work, and extend
it in order to ensure that ATP maximizes the benefits of its investments
in new technology.
