Intra-industry externalities and regional growth: A panel data analysis.

摘要

This research work uses a panel data model to analyze regional wage variation in the Argentine manufacturing sector during the period 1946-1984. During this interval, Argentina experienced a process of industrialization through import substitution (ISI) that left profound effects on the regional structure of wages and employment. Manufacturing wage dispersion, measured as the standard deviation of real regional wages with respect to the country's average, increased steadily during this period, indicating that the real wage gap between the provinces and the manufacturing core widened. This fact, which contradicts both Solow's neo-classical model and the Heckscher-Ohlin theorem of factor price equalization, left the country with a distorted manufacturing sector concentrated mostly in three urban centers with the rest of the provinces mostly agricultural and poor.;A panel data analysis permits separation of agglomeration effects due to "Napa Valley" types of concentration mainly caused by static comparative advantages such as climate, geographic location (coastal or Mediterranean region), soil characteristics and other time invariant effects that prompt industry to agglomerate in certain places (the wine industry in Mendoza), from "Silicon Valley" types of agglomeration due to dynamic comparative advantages, such as technological spillovers among firms, agglomeration of related services, learning by doing or skilled labor pooling. This model shows external economies as stemming from firms' investment in workers' human capital. This process, as every externality-generating one, causes firms to get productivity gains by locating near one another.;This model is tested using data at the 2-digit level and provinces as the unit of measurement. After differencing out fixed factors of production, the results support the hypothesis that dynamic external economies exert a strong attraction of manufacturing employment towards places where manufacturing already exists, forming a sort of circular causation. External and industry-specific types of scale economies, coupled with a government-promoted industrialization via import substitution are key factors explaining regional dispersion of wages during the period under study.