chapter  9
30 Pages

The Economic Performance of Regions

ByMICHAEL E. PORTER

Studies of competitiveness and economic development To the CBP data we matched patent data from the US Patent and Trademark Office and CHI Research,have tended to focus on the nation as the unit of

analysis, and on national attributes and policies as the which is allocated to SIC codes using an algorithm developed by Silverman (SILVERMAN, 1999). Patentsdrivers. As regional scientists and economic geo-

graphers have long understood, however, there are are the best available measure of innovative activity across all regions, and we explore the patterns ofsubstantial differences in economic performance across

regions in virtually every nation. This suggests that patenting across geography and its relationship with industry location.5 All our data covers the 1990 to 2000many of the essential determinants of economic perfor-

mance are to be found at the regional level. time period. The primary geographic unit used in the analysis isThere is a substantial theoretical literature on regional

economic development, and numerous case studies the Economic Area (EA) as defined by the Bureau of Economic Analysis. There are 172 EAs covering thehave explored the influences on economic development

and performance in particular regions. SCOTT, 2000, entire US, which are generally smaller than states but larger than most metropolitan statistical areas or MSAsprovides a comprehensive review of the economic geo-

graphy literature over the past half century. FELDMAN, (see Appendix A). We utilize EAs, rather than MSAs which have been the focus of much of the statistical2000; GLAESER, 2000; and HANSON, 2000, provide

additional literature review. Despite this rich tradition, literature, because EAs cover the entire US, have stable definitions over time and, most importantly, betterempirical studies of large samples of regions have been

comparatively rare. A recent body of work has exam-reflect true economic boundaries of regions because they capture the actual patterns of market exchange thatined various hypotheses about regional performance in

large samples of cities, most notably the respective often cross arbitrary MSA borders. We utilized states (51 including Washington, DC) as the geographic unitinfluence of economic specialization and diversity.1 In

this paper, we aim to contribute to this empirical litera-for some analyses due to less data suppression. All of the analyses here have been replicated using all threeture with a complementary approach. Using a newly

assembled dataset covering every metropolitan area, geographic units and, by and large, the results are similar. The first section of this paper focuses on differenceseconomic area and state in the US, and new statistical

methods to derive the composition of regional econo-in overall regional economic performance in terms of wages, wage growth, employment growth, andmies and the boundaries of clusters of linked industries,

we seek to explore the basic facts about regional econo-patenting. The next section uses the actual patterns of industry employment across geography to decomposemies in the US. In particular, we explore the overall

economic performance of regions, the composition of regional economies into traded, local, and resourcedependent industries, and we explore their respectiveregional economies, and the role of clusters in composi-

tion and performance. roles in economic performance. We then employ statistical methods to derive clusters of traded industries thatOur primary aim here is not to test a particular

theory, but to examine facts and relationships that have co-locate.6 We explore the attributes, overlap, and distribution of clusters across the US economy and thebeen implicit or explicit in many theories.2 How much

do regions vary in wages, employment growth and relationship between the mix of clusters in a region and its performance. A final section provides a summarypatenting rates? How important is size or industry

specialization in performance? Does the particular and conclusions. composition of industries in a region matter? What are the groups of industries that are linked in geographically concentrated clusters, and how does cluster position and mix relate to a region’s performance. Those and

DIFFERENCES IN REGIONAL many other questions are examined, employing basic

ECONOMIC PERFORMANCE statistical tests. In-depth analyses of particular hypotheses are the subject of other papers.3 A region’s overall average wage7 is perhaps the most

basic measure of its economic performance and mostThe core dataset is the annual County Business Patterns (CBP) data, covering employment, establish-associated with its standard of living. In 2000, the

average wage in US EAs was $27 533. There is aments and wages by county at the four-digit SIC (Standard Industrial Classification) level.4 The newer striking variation in average wages among EAs, ranging

from $19 228 in North Platte, NE-CO to $52 213 inNAICS system offers some improvements because it is less aggregated, but the changes affect a modest San Francisco-Oakland-San Jose, CA (see Fig. 1).