پراکندگی سرمایه انسانی و رشد اقتصادی

Based on a theoretical consideration of human capital production technology, this study empirically investigates the growth implication of dispersion of population distribution in terms of educational attainment levels. Based on a pooled 5-year interval time-series data set of 94 developed and developing countries for 1960–1995, the study finds that dispersion index as well as average index of human capital positively influences productivity growth. Given limited social resources for human capital investment, the finding implies that education policy that creates more dispersion in the human capital will promote growth.

Models of endogenous growth have pursued theoretical framework where persistent economic growth is conditioned on human capital accumulation, as in Lucas, 1988 and Romer, 1990. The implication of these models is that human capital is the driving force in the growth process of an economy. Barro, 1991, Benhabib and Spiegel, 1994 and Dinopoulos and Thompson, 2000, and many others find empirical evidence supporting these models, relating human capital with productivity growth. In these studies, human capital stocks are represented by average schooling years or enrollment ratio of the population which, however, may not appropriately reflect the actual capacity of aggregate human capital in an economy. It may well be that given the same average index of schooling, the economies with different distributions of population in terms of educational attainment levels may perform differently. Education policy and structure of social incentive system regarding human capital accumulation will determine the distribution of population across various educational attainment levels. It is interesting to note that not only the average schooling years but also distribution of population in terms of different educational attainment levels varies widely across economies and across time. For example, although both Canada and Norway have similar average schooling years for their population, Canada’s population is vastly more diverse in terms of its individuals’ educational attainment levels than those of Norway.1 Whether the structure of each country’s distribution has any growth effect is an intriguing issue. This study defines the human capital dispersion of population as the degree of dispersion in population distribution in terms of educational attainment levels and empirically investigates the productivity growth implication of human capital dispersion based on a theoretical consideration regarding the form of human capital production technology. The findings will have a strong implication on education policy. Theoretical consideration behind this inquiry stems from the generalization of the human capital production technology and the consideration of the measurement issue of human capital index. The common use of average index of schooling years to proxy human capital relies on the assumption that human capitals obtained from different levels of education are perfectly substitutable and that each additional educational year contributes to the productivity growth at a constant rate regardless of the education levels. Considering the fact that different types of knowledge are obtained at each level of education, a linear return function of education may be a crude approximation. If the return function exhibits convexity, implying that higher education contributes to productivity growth at a higher rate, a greater dispersion of a country’s population distribution in terms of educational attainment levels would lead to a greater stock of aggregate human capital, given the same average schooling years in the population. Recent empirical studies provide mixed assessments on the magnitude of social returns to human capital. When human capital is considered as a factor of input in a production function, empirical studies find that its impact on growth is insignificant as in Benhabib and Spiegel (1994), or even if it exists, that the returns beyond private returns are minimal or negative as in Pritchett (2001).2 On the other hand, when human capital is considered as a factor influencing productivity growth as suggested in endogenous growth models, its impact is found to be significant as shown in Benhabib and Spiegel, 1994, Dinopoulos and Thompson, 2000 and Bils and Klenow, 2000. These findings suggest an existence of strong externality arising from human capital investments. This study contributes to the literature by considering the growth effect of human capital dispersion of population. Based on a pooled time-series data set of 94 developed and developing countries for the period of 1960–1995, the findings show that the human capital dispersion of population proxied by variance of educational attainment is strongly and positively correlated with productivity growth controlling for the average schooling years of population. Only a handful of studies have looked into relationship between human capital dispersion and growth. Birdsall and Londoño (1997) find that the standard deviation of human capital has a negative relationship with per capita growth based on a cross-section data set of 43 countries. Lopez et al. (1998) develop a model where dispersion of human capital may influence the productivity level. They find that the standard deviation and coefficient of variability of schooling contribute negatively to per capita output in 12 developing economies. The current study differs from Lopez et al. (1998) in that it investigates the influence of human capital dispersion on output growth per labor rather than the level of output and extends the work of Birdsall and Londoño (1997) with an empirical study driven by a theoretical consideration of endogenous growth with human capital spillovers based on much richer set of data. Thomas et al., 2000 and Castelló and Doménech, 2002 find negative influence of human capital inequality on growth based on human capital Gini coefficient. Gini coefficient, a measure of human capital inequality, accounts for the degree of disproportionate distribution of education by quintiles of population and was intended to proxy the income inequality. However, as the current study focuses on the degree of dispersion rather than the inequality in human capital, variance measure of educational attainment is utilized rather than Gini coefficient.3 The paper is organized as follows. Based on the discussion regarding the non-linearity in human capital production technology, Section 2 reasons why both average index and dispersion index of human capital should have growth implications. Section 3 provides data description. In Section 4, empirical model and econometric method used in the estimation are presented. Section 5 presents the empirical results and draws policy implications. Section 6 discusses measurement issues regarding human capital dispersion relating with the current literature. Section 7 provides concluding remarks.

The study examines the significance of human capital dispersion on the productivity growth based on production function regressions against a 5-year interval pooled time-series data set of 94 developed and developing economies. Controlling for catch-up effect, regional and time dummies, the empirical results provide strong evidence that both average index and dispersion index of human capital contribute positively to the productivity growth. The findings on the average index of human capital support earlier theoretical literature on human capital effect on productivity growth. Most interesting result in this study is that more dispersion of population distribution in terms of educational levels adds to the productivity growth, contrary to the previous findings. This evidence is consistent with the theoretical discussion regarding the existence of non-linearity in human capital production technology in Section 2. The results of this study support the convexity hypothesis where each additional education year at a higher level raises the human capital of individual at an increasing rate. The findings in this study imply that given the limited social resources for education, an economy would perform better when it allocates resources to support all levels of education, rather than when it focuses on promoting one particular level of education, for example, basic education. The implication may be especially strong for the underdeveloped economies when they optimize their education policy strategy targeting growth. A more detailed data set on the distribution of educational attainment would help derive a more accurate measure of dispersion to differentiate the degrees of dispersion amongst the advanced economies. Furthermore, it would be interestOperational Researching to investigate the degree of complementarity amongst different types of labors distinguished by their educational attainments and by area of study. These issues remain for future research.