Spatial evolution of the computer industry in the USA

This paper examines the spatial evolution of computers across 317 metro areas in the USA since the introduction of the personal computer. We start by examining the relative distribution of employment across cities, examining how that distribution changes in 1977-1992 and how cities move through the distribution. For computers, transition matrices are stationary, with the industry exhibiting no tendency to settle down, nor any tendency of retrenchment during periods of national high-tech employment decline. There is no tendency of the relative size distribution of computer employment to collapse, go bimodal, etc. Overall computers exhibit some turbulence, with dramatic big winners and losers among cities, as well as persistence for some cities in employment shares. In attracting or repelling an industry, urban heterogeneity is important. Large, well educated cities near San Jose have a much greater chance of attracting high-tech employment (much lower mean first passage times moving up states) and less chance of losing it. In assessing the determinants of persistence in local employment patterns we examine sources of productivity growth. We find strong evidence of significant dynamic own industry externalities for single plant firms and little evidence of urbanization-Jacobs-knowledge type externalities. Corporate plants in computers seem to be self-reliant and not really influenced by externalities. (C) 1999 Published by Elsevier Science B.V. All rights reserved. JEL classification: L86; R39.