and

1Corresponding author. This paper seeks to develop our understanding of the somewhat diffuse nature of techno-logical externalities in technological space by associating a geographical dimension with the sectoral dimension. Using a panel data set containing French patents granted as well as private and public research expenditures by industry and region over the period from 1992 to 2000, this paper estimates a knowledge production function. The region- and industry-specific nature of the sample data allows us to model spatial spillovers associated with public and private research expenditures in own- and other-industry sectors for our sample of 94 French regions. This allows us to empirically examine Jacobs externalities thought to arise from cross-industry fertilization that occurs in the innovation process, ver-sus Marshall-Arrow-Romer (MAR) externalities arising from within-industry diffusion of innovation. Since industry-specific patents are zero in some regions and time periods, we rely on a Bayesian spatial Tobit regression model that accommodates these zero observations. Zer