We use this framework to guide our empirical
evaluation of the sources of cross-country differences
in the production of innovative output. We do so by
estimating the relationship between the production of
“international” patents and observable measures of
national innovative capacity. Our use of patent data to
evaluate the rate of technological innovation is subject
to several important (and well-known) limitations, including
differences in the propensity to patent across
different time periods, geographic regions, and technological
areas. We attempt to address these issues
by (a) using international patents; (b) establishing
the robustness of our results to controls through the
use of year and country dummies; and (c) carefully
interpreting our findings in light of the potential for
measurement error. 2 Also, since some elements of
national innovative capacity (such as the environment
for innovation in specific clusters) cannot be directly
observed at the aggregate level, we employ measures
reflecting more aggregate outcomes associated with
the presence of these drivers.
Our results suggest that the production function
for international patents is well-characterized by a
small number of observable factors that describe a
country’s national innovative capacity. We distinguish
between scale-based differences across countries (arising
from differences in population or the level of inputs
devoted to innovation) and productivity-based
differences (i.e. differences in innovative output per
unit of effort expended on the innovation process).
While scale-related measures, such as total population,
the size of the R&D workforce, or R&D spending
have important explanatory power, more nuanced
factors separately impact country-level R&D productivity.
We find robust and quantitatively important evidence
that R&D productivity varies with aggregate
policy choices such as the extent of IP protection and
2 In using international patenting data to understand the sources
and consequences of innovation, this paper builds on Evenson
(1984), Dosi et al. (1990), and recent work by Eaton and Kortum
(1996, 1999). We extend these prior analyses by linking our results
more closely to a range of theories about the determinants of national
innovative capacity and by exploring a relatively long panel
which allows us to incorporate both cross-sectional and time-series
variation. As well, we supplement the patent analysis by examining
alternative indicators of new-to-the-world innovative output,
including scientific articles and export shares in “high-technology”
industry segments.
openness to international trade, the shares of R&D
performed by the academic sector and funded by the
private sector, the degree of specialization by technology
area (a proxy for cluster specialization) and each
individual country’s knowledge stock. We also find
that the estimated level of national innovative capacity
affects total factor productivity (TFP) growth and
a nation’s share of high-technology exports.
Our results provide evidence on the sources of differences
in innovation intensity and R&D productivity
across countries and over time. We find that there has
been substantial convergence in the level of per capita
national innovative capacity across the OECD since
the mid-1970s. During the 1970s and early 1980s, the
predicted level of per capita international patenting by
United States and Switzerland substantially exceeded
that of other OECD members. Since that time, several
countries (most notably Japan, some Scandinavian
countries, and Germany) have achieved levels of
predicted per capita international patenting similar to
that of United States and Switzerland. Interestingly,
there are exceptions to the convergence pattern; for
example, UK and France have shown little change in
their measured level of national innovative capacity
over the past quarter century.
The paper proceeds by motivating and developing
our theoretical framework, outlining the relationship
between “visible” innovative output and the elements
of national innovative capacity, describing data and
methods, and discussing our empirical results.