6.4Localization of Information Technology Activity and Productivity

This section describes some literature on the relationships between the localization patterns of IT activity across space and regional productivity.

First, Malecki (1991) applied the method of Location Quotients at the county level in the state of Florida to study the effects of changes in employment profiles and demographic trends on regional economic growth between 1982 and 1987. He used a cross-sectional econometric model that relates the change in total employment to various demographic and occupational variables. He found that half of the counties studied had real economic growth between 1982-1987. These counties exhibited the highest concentration of their basic employment in the secondary sector. Their location quotients for manufacturing were either greater than one or were increasing during the period studied. This implies that even though Florida’s economy is service-oriented and a small fraction of the labor force is employed in the secondary sector, manufacturing remains a catalyst for economic growth.

Malecki (1987) elaborated on the issue of geographic localization of high tech industry. He noticed the efforts from communities and all 50 states to reproduce the success of Silicon Valley and Route 128 as leading technological clusters. He emphasized the necessity of strong governmental support in order to do so. However, each state or community must also understand that its unique local conditions are important. Being part of a large urban region, having abundant air transportation and strong universities constitute great advantages that will attract high tech firms. Still, as Malecki argued, “encouraging and nurturing new companies bears more fruit than trying to lure firms from elsewhere. (...) the hope is that rapidly growing local high-tech firms might replace declining industries.’ Finally, Silicon Valley or Route 128 are a proof of the significant advantages to be acquired through investment in human capital.

Zucker et al. (1998) studied empirically localized knowledge spillovers, using data on California biotechnology. They argued that the output that results from R&D investment in this industry is not a public good because it is neither nonrivalrous nor partially excludable, which is contradictory with the traditional definition of knowledge spillovers as given by Romer (1990). Indeed, Zucker et al. (1998) found that the positive impact of university research on nearby firms comes from identifiable market exchanges between two parties that both benefit from:

‘For an average firm, five articles co-authored by academic stars and the firm’s scientists imply about five more products in development, 3.5 more products on the market, and 860 more employees. Stars collaborating with or employed by firms, or who patent, have significantly higher citation rates than pure academic stars.’

Beardsell and Henderson (1999) examined the spatial evolution of the computer industry and its impact on productivity across 317 metropolitan areas in the USA from 1970 to 1992. First, they studied the evolution of employment to see if it concentrates in fewer locations or if patterns appear relatively fluid. They also emphasized the importance of locational characteristics (such as labor pooling, state taxes, intermediate product diversity) as determinants of the location behavior of computer firms. Finally, they found strong evidence of localization economies (own industry externalities) as determinants of productivity growth, and little evidence of urbanization economies. Pollard and Storper (1996) studied the growth in three growth-generating sectors: industries handling information and advanced management functions (“intellectual capital”), high technology industries (“innovation based”) and “variety-based” industries, which represents industries with high levels of product differentiation, relatively short production runs, and lower level of mechanization than mass-production industries. They focused on twelve metropolitan areas across the United States between 1977 and 1987. Their findings suggest that the determinants of regional employment growth of the 1980s might no longer be the ones of the 1990s. “Variety-based” industries are no longer a motor of growth, but intellectual capital and innovation-based industries exhibit high growth in all areas studied. Therefore, it could be possible that these industries have a low propensity to agglomerate, as asked by the authors. One reason could be the telecommunication revolution, which might have reduced the importance of localization economies.

This chapter presented findings in the field of regional economics, regarding the externality effects of location patterns of employment. Agglomeration effects explain in part why productivity differs across regional units (regions, states, counties). A study from Ciccone and Hall (1996) showed that the density of economic activity at the county level could increase labor productivity at the state level. This finding goes against principles of neoclassical theory, which stipulate that congestion effects dominate when employment density increase. The next chapter describes a methodology for measuring the externality effects of the location patterns of information technology employment.