Recent theories of economic growth suggest that increasing returns to knowledge within regions will cause their growth rates to diverge. If knowledge spillovers are as important as much of the theoretical literature assumes and several recent empirical studies have found, innovative activity should be concentrated geographically. In Discussion Paper No. 953, Research Fellow David Audretsch and Maryann Feldman explain and control for the geographic concentration of production and then examine the linkages between geographic concentration of innovative activity in the US and knowledge externalities. They calculate Gini coefficients for the geographic concentration of innovative activity and manufacturing value added for specific industries from the Small Business Administration's Innovation Data Base, which covers 8,074 commercial innovations introduced in 1982. Measuring spatial clustering with the unit of observation at the state level provides a crude proxy for the relevant market but corresponds to the most important level for policy-making. They then regress their Gini coefficients for 168 four-digit SIC industries on a number of variables to show that production tends to be geographically more concentrated in industries that make greater use of natural resources and skilled labour or have higher transport costs or R&D intensities.

Even after controlling for variations in the concentration of production, the evidence strongly suggests that innovative activity clusters more in industries in which knowledge spillovers are of greatest importance those involving industry R&D, university research and skilled labour. Indeed, while knowledge spillovers foster the clustering of both innovative activity and production, greater concentration of production may even increase the dispersion of innovative activity, but the positive impact of university research is unequivocal. Policy-makers concerned to attract and create innovative firms should therefore concentrate efforts to promote university research, which clearly serves as an important input to the innovative activity of private firms.

Knowledge Spillovers and The Geography of Innovation and Production
David B Audretsch and Maryann P Feldman

Discussion Paper No. 953, May 1994 (IO/IT)