Human Capital and Economic Activity in Urban America

Abstract

Abel J. R. and Gabe T. M. Human capital and economic activity in urban America, Regional Studies. The relationship between human capital and economic activity in US metropolitan areas is examined, extending the literature in two ways. First, new data on metropolitan area gross domestic product are utilized to measure economic activity. Results show that a 1 percentage point increase in the proportion of residents with a college degree is associated with about a 2% increase in metropolitan area gross domestic product per capita. Second, measures of human capital are developed that reflect the types of knowledge within US metropolitan areas. Regional knowledge stocks related to the provision of producer services and information technology are important determinants of economic vitality.

Abel J. R. et Gabe T. M. Le capital humain et l'activité économique dans l'Amérique urbaine, Regional Studies. On examine le rapport entre le capital humain et l'activité économique dans les zones métropolitaines aux Etats-Unis, élargissant la documentation à deux temps. Primo, on emploie de nouvelles données sur le produit intérieur brut des zones métropolitaines afin d'estimer l'activité économique. Les résultats laissent voir qu'une augmentation de 1% de la proportion des habitants diplômés se rapporte à une augmentation de 2% du produit intérieur brut par tête dans les zones métropolitaines. Secundo, on développe des mesures du capital humain qui reflètent la connaissance qui caractérise les zones métropolitaines aux Etats-Unis. Le stock de connaissance régionale qui se rapporte à la fourniture des services aux entreprises et de la technologie de l'information s'avèrent un déterminant non-négligeable de la vitalité économique.

Capital humain Connaissance Nouvelle économie Productivité

Abel J. R. und Gabe T. M. Humankapital und Wirtschaftsaktivität in Städten der USA, Regional Studies. Wir untersuchen die Beziehung zwischen Humankapital und Wirtschaftsaktivität in Großstadtgebieten der USA und ergänzen die Literatur in diesem Gebiet auf zweierlei Weise. Erstens nutzen wir neue Daten über das Bruttoinlandsprodukt von Großstadtgebieten zur Messung der Wirtschaftsaktivität. Aus den Ergebnissen geht hervor, dass eine einprozentige Erhöhung des Anteils der Einwohner mit Hochschulabschluss mit einer etwa zweiprozentigen Erhöhung des Pro-Kopf-Bruttoinlandsprodukts in einem Großstadtgebiet einhergeht. Zweitens entwickeln wir Maßstäbe für das Humankapital, die den Arten von Wissen in Großstadtgebieten der USA entsprechen. Der regionale Wissensschatz im Zusammenhang mit der Bereitstellung von Wirtschaftsdienstleistungen und Informationstechnik ist ein wichtiger Determinant für die wirtschaftliche Vitalität.

Humankapital Wissen Neue Wirtschaft Produktivität

Abel J. R. y Gabe T. M. Capital humano y actividad económica en áreas urbanas de los Estados Unidos, Regional Studies. Examinamos la relación entre el capital humano y la actividad económica en las áreas metropolitanas de los Estados Unidos extendiendo la literatura de dos formas. Primero, utilizamos los nuevos datos del producto interno bruto de áreas metropolitanas para medir la actividad económica. Los resultados muestran que un aumento de un 1% en la proporción de residentes con un título universitario se asocia a aproximadamente un aumento de un 2% en el producto interno bruto de áreas metropolitanas per cápita. Segundo, desarrollamos mediciones del capital humano que reflejan los diferentes tipos de conocimiento en las áreas metropolitanas de los Estados Unidos. Los stocks regionales de conocimiento relacionados con el suministro de servicios para productores y la tecnología de la información son determinantes importantes de la vitalidad económica.

Capital humano Conocimiento Nueva economía Productividad

Keywords:

• Human capital
• Knowledge
• New economy
• Productivity

JEL classifications:

• J24
• O40
• R11

Acknowledgements

The authors are grateful to Richard Deitz, Isaac Ehrlich, Andrew Haughwout, Judith Hellerstein, Zhiqiang Liu, Ann Markusen, Yong Yin, and the seminar participants at the Federal Reserve Bank of New York, the National Bureau of Economic Research (NBER) Program on Technological Change and Productivity Measurement workshop, University at Buffalo Center of Excellence on Human Capital, and University of Toronto Martin Prosperity Institute for helpful comments on earlier drafts of this paper. Two anonymous referees also provided valuable suggestions that improved the paper; and Ishita Dey and Jonathan Hastings provided excellent research assistance. Gabe's participation in this research was supported, in part, by the Maine Agricultural and Forest Experiment Station (MAFES) and the US Department of Agriculture (USDA) Cooperative State Research, Education and Extension Service, Hatch Project Number ME 08214-08, multi-state Project Number NE 1029, MAFES External Publication Number 3100. The views and opinions expressed in this paper are solely those of the authors and do not necessarily reflect those of the Federal Reserve Bank of New York, the Federal Reserve System, or the University of Maine.

Notes

For more information, see Panek et al. (2007).

The results are not sensitive to choice of year within this period or method of averaging when constructing the dependent variable.

Years of schooling, sometimes used in the labour economics literature to analyse an individual's private return to education, is an alternative measure of educational attainment. The correlation between the proportion of residents with a college degree and average years of schooling in a metropolitan area is 0.73. Standardized regression results using this alternative measure of educational attainment (which is available from the authors upon request) are nearly identical to results reported in the paper.

For more information, see Meade et al. (2003).

Economic growth models typically relate the long-run level of output to the steady-state stock of physical capital, rather than the investment rate. However, because data on the stock of physical capital are not available, flow measures are used to control for differences in physical capital across US metropolitan areas. Constructing these variables required the assumption that industry-level investment per worker is similar across US metropolitan areas as information that would allow one to vary the investment rate by metropolitan area is not available. Further, it is possible that the actual amount of investment per worker is related to the amount of human capital in the region, which would bias the results. To assess the robustness of the relationship between human capital and metropolitan area GDP per capita, results with these measures of physical capital omitted from the model are also reported.

Early empirical studies of urban agglomeration focused on city size rather than on density (Sveikauskas, 1975; Segal, 1976). The results (which are available from the authors upon request) remain unchanged if population size – rather than density – is used to control for urban agglomeration economies.

The reliance on a subset of the 363 metropolitan areas included in the US BEA metropolitan GDP data is due to differences in metropolitan area definitions between the US BEA and US Census. The data set is constructed using metropolitan area definitions utilized by the US BEA, which correspond to those issued by the Office of Management and Budget (OMB) in December 2006. Appropriate adjustments to the US Census data are then made to match, as closely as possible, the OMB metropolitan area definitions.

Evaluated at mean values, the results imply that a doubling of population density is associated with a 5.3% increase in economic activity, which is within the 4.5–6.0% range established by Ciccone and Hall (1996) and Ciccone (2002) for US states and European regions, respectively.

A version of the model omitting the capital structure variable, which does not have a significant effect on GDP per capita, was also estimated. The estimated coefficients corresponding to the other explanatory variables are nearly identical to those reported in column (1) of Table 2.

The Morrill Acts of 1862 and 1890 are credited for establishing the major land-grant universities that exist in the United States (Appleby, 2007). There were seventy-three land-grant universities created before 1890, located in places ranging from Boston, Massachusetts, and Orono, Maine, to Columbus, Ohio, and Corvallis, Oregon. The 1994 Land-Grant Act added a number of tribal institutions to the list of land-grant universities, which are not included in this analysis.

The data for the climate index are drawn from the County and City Data Book: 2007 (US Census Bureau, 2007), and correspond to the central city within each metropolitan area. The annual number of heating degree-days and annual amount of precipitation, averaged over the period 1971–2000, are used to construct the climate index. To develop relative measures of temperature and precipitation, each variable is first scaled by the average value and then each variable is normalized so the maximum value equals 100. The climate index is an evenly weighted sum of these two measures, re-normalized to a scale of 100. Higher values of the index indicate a relatively cold and wet and climate, while lower values of the index indicate a relatively warm and dry climate.

Stock and Yogo (2005) suggest a weak instrument test that compares an F-statistic from the two-stage regression model with a critical value that depends on the number of endogenous variables, the number of instruments, and the tolerance for the ‘size distortion’ of a test (α= 0.05) of the null hypothesis that the instruments are weak. The size distortion tolerance (for example, 10%) accounts for the idea that using the weakest combination of instruments might lead to a conclusion of biased second-stage estimates (from a Wald test), whereas using the entire group of instruments does not.

This test of over-identifying restrictions is computed as N × R ², where N is the number of observations; and R ² is computed from a regression of the residuals from the second-stage regression on all exogenous variables and the instruments. The test statistic is distributed χ ², with degrees of freedom equal to the number of over-identifying restrictions, in this case 1.

Because limited information maximum likelihood estimation (LIML) is more robust to the presence of weak instruments, the model was also estimated using this estimator and identical results to those presented in the paper using 2SLS were obtained.

The O*NET database is described in detail by Peterson et al. (2001) and Feser (2003).

The correlation between the share of residents with a college degree in a metropolitan area and the present aggregate knowledge-based measure of human capital is 0.83.

Florida et al. (2008) also suggest that a high regional share of educators may reflect a large population of students, which typically contribute less to regional economic activity.