An exam of the spatial patterns of innovation in Brazilian industry : an empirical analysis : paper presented at ERSA 2012 52nd European Congress of the Regional Science Association International - Bratislava, Slovakia - 21st - 25th August

The aim of this paper is to exam the spatial patterns of innovative performance in Brazilian industry, taking into account its regional interdependencies, and the impact of the main innovative inputs. There is a huge literature concerning regional innovation and the importance of local inputs in innovative performance. However, most of the studies use data from developed countries. This paper verifies if the role played by innovative inputs in developed countries remain important in developing ones, in which patents are proportionally rare. In this sense, it's applied an empirical model based in the Jaffe's (1989) knowledge production function to Brazilian regions. The model uses patents as a proxy for the innovative output and includes regional variables of local industrialand academic R&D, agglomeration characteristics and some spatial elements such as neighborhood's innovative activities. The main results show the importance of local industrial R&D to regional innovation measured by patents, and, similarly, a relation between patenting activity of the firms and local academic research. With the purpose of evaluate which externality is more important to innovation in Brazilian regions, marshallian or jacobian externalities; the Krugman specialization-diversification index of industrial employment is adopted in the model. The importance of been close to the most innovative regions is assessed with the commonly used spatial lagged variables and the estimation results corroborates the relevance of technological spillovers spatial mediated. Finally, some efforts are made to exam other kinds of proximity as proposed by Boschma (2005) and a network weight matrix based on university-industry collaborative links, such as Ponds et al (2010), is added to the model to test the importance of non spatial proximity. The overall conclusion suggests that in Brazilian case, main innovative inputs seemed in developed countries remain important, but presents also some specificity such as a strong concentration of innovative activities in the Southeast related with the industrial agglomeration and different relative magnitude importance in some local determinants of innovation.