Spillovers from industrial R&D activity: On the importance of geographic and technological proximity

Public returns to private research and development (R&D) expenditures are a consequence of the imperfect appropriability condition that typifies knowledge generating activity. This feature recommends a prominent role for R&D and spillovers in theories of industry structure, city formation, and endogenous economic growth. Moreover, R&D spillovers may provide justification for policy intervention if a decentralized system of decision making underprovides R&D services. Industrial agglomeration implies a correlation between inter-firm measures of geographic and technological proximity that complicates identification of R&D spillovers. Parameter estimates obtained by methods that do not account for this colinearity may overstate the importance of one of these factors for spillover transmission. This study presents empirical evidence of inter-firm spillovers from R&D activity using a firm-level production function framework. The spillover-enhancing effects of geographic and technological proximity are quantified for a panel of machinery and computer equipment manufacturing firms. Results from a preliminary analysis are consistent with the stylized fact that geographically and technologically proximate stocks are the most important source of R&D spillovers. The final analysis controls for the degree of industrial agglomeration in these data. These parameter estimates indicate that technologically proximate stocks are the most important source of R&D spillovers. However, the magnitude of such spillovers is insensitive to distance between spillover sending and receiving firms. In contrast, R&D spillovers between technologically dissimilar (but not too dissimilar) firms are attenuated by distance. The results presented here indicate that while R&D spillovers are greatest among firms in the same industry these spillovers do not appear to serve a localization function. The propensity for similar industrial activity to form into geographic clusters may instead reflect the importance of agglomerating forces exclusive of the geographic dependence of knowledge spillovers. While economically less important, spillovers across narrowly defined industry boundaries do appear to be sensitive to inter-firm distance and may therefore play a direct role in the process of urbanization.