This paper estimates the nature and magnitude of local externalities from own industry scale, as envisioned by Marshall. Using census panel data on individual plants in high-tech and machinery industries across up to 487 countries, the study quantifies the direct effects of local external environments on plant productivity. The paper addresses endogeneity issues in estimation and investigates the magnitudes of scale externalities for corporate versus single plant firms, as well as the sources and extent of externalities (e.g., employment, number of plants, births). It also explores whether externalities are static or dynamic and whether they depend on local own industry activity or overall local urban scale and diversity. The findings are related to the extent of agglomeration and the degree of mobility of industries across cities. The paper finds that high-tech industries are more concentrated than machinery industries, with the difference increasing over time. High-tech industries are also more mobile than machinery industries. The study estimates that scale externalities are very local, derived from the number of own industry plants in the own county, rather than from industry total employment or activity in surrounding counties. Single plant firms and corporate plants benefit equally from static externalities, but in industries with dynamic externalities, single plant firms derive greater benefits. Corporate plants may use internal information networks to substitute for local knowledge spillovers. The paper also finds that high-tech industries do not benefit from Jacobs-urbanization externalities, suggesting that the search for such externalities may more productively focus on service and R&D sectors. The extent of spatial agglomeration of individual industries is closely related to the extent of scale economies for the industries. However, the decline in industrial geographic concentrations over time seems unrelated to scale externality magnitudes, which have not changed. The paper also finds that even industries without scale economies agglomerate to some extent, possibly due to trade linkages with industries that do experience scale economies. The paper uses plant and city-industry level data from the Census of Manufacturers for 1963-1992 and information from the Annual Survey of Manufacturers for certain non-Census years. It examines the extent of agglomeration and the evolution and mobility of industries across metropolitan areas. The study finds that high-tech industries are more concentrated at the extreme high end of the distributions, while in all industries, employment has spread out into middle employment centers. The paper also finds that high-tech industries have larger plant sizes, which could be a potential source of immobility. The paper estimates the nature and extent of agglomeration economies by examining production functions at the plant level. It finds that high-tech industries have scale elasticities of 0.08, indicating that an increase in the number of plants in a county raises plant output by 18.5%. The study also finds that high-tech industries have significantly larger scale effects than machinery industries. The paper concludes that high-tech industries are more concentrated and more mobile than machinery industries,This paper estimates the nature and magnitude of local externalities from own industry scale, as envisioned by Marshall. Using census panel data on individual plants in high-tech and machinery industries across up to 487 countries, the study quantifies the direct effects of local external environments on plant productivity. The paper addresses endogeneity issues in estimation and investigates the magnitudes of scale externalities for corporate versus single plant firms, as well as the sources and extent of externalities (e.g., employment, number of plants, births). It also explores whether externalities are static or dynamic and whether they depend on local own industry activity or overall local urban scale and diversity. The findings are related to the extent of agglomeration and the degree of mobility of industries across cities. The paper finds that high-tech industries are more concentrated than machinery industries, with the difference increasing over time. High-tech industries are also more mobile than machinery industries. The study estimates that scale externalities are very local, derived from the number of own industry plants in the own county, rather than from industry total employment or activity in surrounding counties. Single plant firms and corporate plants benefit equally from static externalities, but in industries with dynamic externalities, single plant firms derive greater benefits. Corporate plants may use internal information networks to substitute for local knowledge spillovers. The paper also finds that high-tech industries do not benefit from Jacobs-urbanization externalities, suggesting that the search for such externalities may more productively focus on service and R&D sectors. The extent of spatial agglomeration of individual industries is closely related to the extent of scale economies for the industries. However, the decline in industrial geographic concentrations over time seems unrelated to scale externality magnitudes, which have not changed. The paper also finds that even industries without scale economies agglomerate to some extent, possibly due to trade linkages with industries that do experience scale economies. The paper uses plant and city-industry level data from the Census of Manufacturers for 1963-1992 and information from the Annual Survey of Manufacturers for certain non-Census years. It examines the extent of agglomeration and the evolution and mobility of industries across metropolitan areas. The study finds that high-tech industries are more concentrated at the extreme high end of the distributions, while in all industries, employment has spread out into middle employment centers. The paper also finds that high-tech industries have larger plant sizes, which could be a potential source of immobility. The paper estimates the nature and extent of agglomeration economies by examining production functions at the plant level. It finds that high-tech industries have scale elasticities of 0.08, indicating that an increase in the number of plants in a county raises plant output by 18.5%. The study also finds that high-tech industries have significantly larger scale effects than machinery industries. The paper concludes that high-tech industries are more concentrated and more mobile than machinery industries,