R&D Spillovers and the Geography of Innovation and Production

R&D Spillovers and the Geography of Innovation and Production Author(s): David B. Audretsch and Maryann P. Feldman Source: The American Economic Review , Jun., 1996, Vol. 86, No. 3 (Jun., 1996), pp. 630- 640 Published by: American Economic Association Stable URL: https://www.jstor.org/stable/2118216 JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at https://about.jstor.org/terms is collaborating with JSTOR to digitize, preserve and extend access to The American Economic Review This content downloaded from 107.15.27.206 on Thu, 08 Apr 2021 18:59:02 UTC All use subject to https://about.jstor.org/terms R&D Spillovers and the Geography of Innovation and Production By DAVID B. AUDRETSCH AND MARYANN P. FELDMAN * More than most other economic activities, as Jaffe et al. (1993) point out, one obvious innovation and technological change depend explanation why innovative activity in some upon new economic knowledge. Thus, Paul industries tends to cluster geographically more Romer (1986), Paul Krugman (1991a, b), than in other industries is that the location of and Gene Grossman and Elhanan Helpman production is more concentrated spatially. (1991), among others, have focused on the Thus, in explaining why the propensity for in- role that spillovers of economic knowledge novative activity to cluster geographically var- across agents and firms play in generating in- ies across industries, we need first to explain, creasing returns and ultimately economic and then to control for, the geographic con- growth. In fact, several recent studies have centration of the location of production. identified the existence of spatially-mediated As Alfred Marshall (1920) and, later knowledge spillovers. An important finding of Krugman (1991b) argue, there may be geo- Adam B. Jaffe (1989), Zoltan Acs et al. graphic boundaries to information flows or (1992, 1994), and Feldman (1994a, b) is that knowledge spillovers, particularly tacit knowl- investment in R&D by private corporations edge, among the firms in an industry. Al- and universities "spills over" for third-party though the cost of transmitting information firms to exploit. If the ability to receive knowl- may be invariant to distance, presumably the edge spillovers is influenced by distance from cost of transmitting knowledge rises with dis- the knowledge source, then geographic con- tance. That is, proximity and location matter. centration should be observed, especially in While there is considerable evidence support- industries where knowledge spillovers are ing the existence of knowledge spillovers, nei- likely to play a more important role. The pur- ther Jaffe (1989), Jaffe et al. (1993), nor Acs pose of this paper is to examine the extent to et al. (1992, 1994), and Feldman (1994a) ac- which industrial activity clusters spatially and tually examine the propensity for innovative to link this geographic concentration to the ex- activity to cluster spatially. But implicit in the istence of knowledge externalities. Of course, knowledge production function model is the assumption that innovative activity should concentrate geographically in those industries where the direct knowledge-generating inputs * Audretsch: Wissenschaftszentrum Berlin ftir Sozial- are the greatest and where knowledge spill- forschung and the Centre for Economic Policy Research, overs are the most prevalent. No one, to date, Reichpietschufer 50, D- 10785 Berlin, Germany; Feldman: has examined the underlying propensity for in- Institute for Policy Studies, Johns Hopkins University, Baltimore, MD 21218. This article was written while dustrial activity to cluster spatially. While one Maryann Feldman was visiting at the Heinz School of of the central themes in the industrial organi- Public Policy and Management, Carnegie Mellon Univer- zation literature is to explain the degree of con- sity. We thank Richard Baldwin, Paul Krugman, James centration of economic activity within an Markusen, and participants at the CEPR Conference on industry (F. M. Scherer and David Ross, the "Location of Economic Activity: New Theories and New Evidence," 17-20 December, 1993, Vigo, Spain, for 1990), the focus has typically been on the ex- their useful comments. We also thank Jim Adams, Zvi tent of dispersion across different enterprises Griliches, Bronwyn Hall, Frank Lichtenberg, Richard and establishments within a single spatial Nelson and Mike Scherer and the participants in the dis- unit-the country. The emerging importance cussion at the 1995 AEA Meetings. We would also like to thank the anonymous referees for useful comments and of location as a unit of observation argues for suggestions. Gail Cohen Shaivitz provided invaluable re- examining both production and innovation search assistance. within a geographic context. We empirically 630 This content downloaded from 107.15.27.206 on Thu, 08 Apr 2021 18:59:02 UTC All use subject to https://about.jstor.org/terms VOL. 86 NO. 3 AUDRETSCH AND FELDMAN: GEOGRAPHY OF PRODUCTION 631 test for the importance of geographic location used by Acs and Audretsch (1988, 1990) to to different types of industries by linking the analyze the relationships between firm size geographic concentration in manufacturing in- and technological change, and market struc- dustries to industry specific characteristics, ture and technological change, and by Acs et most notably the relative importance of knowl- al. (1992, 1994), Feldman (1994a, b), and edge spillovers. Feldman and Richard Florida (1994) to ex- In the following section of this paper, we amine the geography of innovation. examine the spatial distribution of innovative We adopt the state as the spatial unit of ob- activity as well as the geographic concentra- servation. While this is at best a crude proxy tion of production. An empirical model is of the relevant economic market,2 it does have specified in Section II, and the results are one obvious appeal other than that it conforms presented in Section HII. In the final section, to a number of data sources-the most rele- we provide a summary and conclusion. The vant unit of policy-making is at the level of the empirical evidence suggests that, even after state. Still, states are certainly not an entirely controlling for the degree of geographic con- satisfactory unit of observation for the analysis centration in production, innovative activity of spatial phenomena. The analyses of spatial tends to cluster more in industries where processes are handicapped by a lack of data knowledge spillovers play a decisive role. Al- for what might be considered to be the ideal though such industries also tend to exhibit a observation. Certainly considerable progress greater geographic concentration of produc- would be made if data sources identifying in- tion, the results suggest that the propensity for novation activity at the city or county level innovative activity to cluster is more attribut- were made available. able to the role of knowledge spillovers and Using the citation data base described not merely the geographic concentration of above, an innovation is attributed to the state production. in which the establishment responsible for the development of that innovation is located. I. The Spatial Distribution of Innovation Some innovations are, in fact, developed by and Production subsidiaries or divisions of companies with headquarters in other states. Since headquar- To measure the spatial distribution of in- ters may announce new product innovations, novative activity we rely on the most recent the data base discriminates between the lo- and most ambitious major data base that pro- cation of the innovating establishment and vides a direct measure of innovative activity. the location of the larger, innovating entity The United States Small Business Administra- (Edwards and Gordon, 1984). For our pur- tion (the Small Business Administration's In- poses, the state identifier of the establishment novation Data Base or the SBIDB) compiled is used to investigate the spatial distribution of a data base of 8,074 commercial innovations innovation. Of the total number of innovations introduced in the United States in 1982. A pri- recorded in the data base, 4,200 were manu- vate firm, The Futures Group, compiled the facturing innovations with information speci- data and performed quality control analyses fying location.3 for the United States Small Business Admin- Figure 1 shows the distribution of innova- istration. A data base consisting of innovations tions by states. California is the state in which by four-digit standard industrial classification (SIC) industries was formed from the new product announcement sections in over 100 2As Krugman (1991b p. 57) emphasizes, "States technology, engineering and trade journals aren't really the right geographical units," because of the that span every industry.' These data were lack of concordance between economic markets and po- litical units. 'The SBIDB contains a total of 4,476 innovations in 'A detailed description of the U.S. Small Business Ad- manufacturing industries. Of these, there are 276 innova- ministration's Innovation Citation Data Base can be found tions which are not used because

Load more