among actors in a collective learning process; (3) the availability of a common set of resources,
such as universities, research centers, technology centers, and a pool of specialized and skilled
labor, all of which help reduce the costs and uncertainties associated with innovative activities.
These key components of the different approaches serve in the present paper as a
foundation for the elaboration of a set of indicators for the geographical distribution of innovation
activities. It should be stressed that these indicators are not designed to measure innovative
activities or knowledge flows in regional terms, topics discussed by most empirical studies in the
field of the geography of innovation,3 especially with regard to the U.S. economy and based on
statistics for patent citations and the location of R&D activities. The purpose of these indicators
here is simply to map the geographical distribution of innovative activities by using as a proxy
the geographical distribution of their inputs (knowledge and technological infrastructure) and
outcomes (patents and trademarks).
In Brazil, previous research along these lines has been published by Albuquerque et al.
(2002). Working with data on patents, scientific articles and researchers by municipality, the
authors discuss indicators that describe the spatial distribution of scientific and technological
activities in Brazil based on the hypothesis that spatial determinants are as important as factors
relating to the innovation system in mediating the relationship between scientific production and
technological production.
In this paper we set out to make progress by extending the scope of the indicators used.
We recognize the limitations of patent statistics as indicators of innovative activities, since not
everything that is patented is an innovation just as not all innovations are patented. Assuming that
the relationship between geography and innovation is empirically proven, and therefore that some
of the key elements of innovation have geographical determinants, we set out to map these
geographical determinants represented by: (1) the tacit knowledge held by workers and
specialists; (2) the number of innovative firms; (3) patent and trademark registrations; (4)
scientific production; and (5) institutions providing scientific, technological and service
infrastructure for innovative firms. In this sense the present paper differs from the literature in
two ways: first, by proposing the use of multiple indicators,4 albeit without taking the additional
step of aggregating them into a single synthetic indicator; and second, by using these indicators
not to measure innovative performance but to map the geographical distribution of innovative
activities. The next section summarizes the methodology used to produce the indicators and
describes their respective databases.
2. Regionalized quantitative indicators: methodology, databases, and application to Sao
Paulo
Five types of regionalized quantitative indicators were produced: (1) the number of skilled
workers, based on data from the Labor Ministry’s employment survey (RAIS - Relaçao Annual
de Informaçôes Sociais) for 2002; (2) the number of innovative firms, based on regionalized
tabulations of Brazilian Institute of Geography and Statistics (IBGE)’s Technological Innovation
Survey (PINTEC) for 2000; (3) the number of patents and (4) marks registered with INPI
(Brazil’s industrial property office) and USPTO (United States Patent & Trademark Office); and
3 See especially Clark, Feldman & Gertler (eds.) (2000), The Oxford Handbook of Economic Geography, Part IV -
“The Geography of Innovation”, with an excellent review of the literature on this subject by Maryann Feldman.
4 Hagedoorn & Cloodt (2004) is an important benchmark study in the use of multiple indicators, albeit applied to the
evaluation of firms’ innovative performances.