نابرابری و رشد: نقش دوگانه سرمایه انسانی در توسعه

To examine how human capital accumulation influences both economic growth and income inequality, we carefully endogenize the demand and supply of skills. We explicitly introduce the costs and externalities in education, and examine how both relate to learning-by-doing and R&D intensity. In addition, we endogenize the determinants of the skill-bias of labor demand: the complementarity between technology and skilled and unskilled labor. Our results identify parameters that are central to the evolution of inequality during the development process. We characterise development thresholds when countries switch endogenously from pure learning to deliberate R&D, and we show that technical change can generate multiple steady states that are consistent with the cross-country data on inequality and skill-premia.

The relationship between equality and growth is of fundamental interest not only to economists, but also to policymakers. Despite the fact that the empirical literature on the subject dates back to Kuznets (1955), no definitive stylized facts have emerged that indicate how inequality and growth interact during the development process.1 This lack of a clear relation between inequality and growth may be partly due to the dual role of human capital in development. In his analysis of the relationship between human capital and inequality, Tinbergen (1975) suggested that inequality is ultimately determined by the opposing effects that technology and education exert, respectively, on the demand for and supply of skilled labor, and hence on the relative wage. He stipulated that the relationship between growth and inequality is determined by the “race between technological development and education” (1975, p. 97). Tinbergen's statement has not been formalized and the task of this paper is to model both effects explicitly to shed light on the remarkably diverse empirical relationship between human capital and inequality. In this paper, we explicitly model endogenous human capital and technology to highlight that the relationship between growth and inequality is indeed complex due to offsetting supply and demand effects. In our model, the direct effect of greater supplies of human capital is to lower the relative wage, and hence, inequality. However, we also highlight that human capital accumulation indirectly generates more innovations, which in turn increase the demand for skilled workers to absorb new technologies into production. As a result, our model predicts a non-monotonic relationship between educational attainment and inequality, which may help to explain why cross-country analyses of income inequality tend to find that school enrollment rates have little explanatory power.2 In modeling human capital accumulation and technical change, we focus on three essential features. First, we explicitly introduce costs and externalities associated with skilled labor supply, and examine how both relate to learning-by-doing and R&D intensity.3 Second, we endogenize the determinants of the degree of skill-bias in labor demand. Specifically, we endogenize the complementarity between technology, and skilled and unskilled labor. Third, we differentiate between deliberate R&D and serendipitous learning-by-doing. R&D is potentially the more productive means of innovation, but it is also more costly and it requires an adequate number of researchers to yield sufficient returns and justify R&D investment. As a result, the model determines endogenously which countries rely only on learning or also on R&D in order to provide a possible rational for the threshold effects that abound in the development literature. The results derived in the paper attract attention to parameters that are central to the supply and demand of human capital, but which have been absent from previous empirical analyses of the inequality and growth relationship. Our results mirror Tinbergen's (1975) hypothesis that the pattern of relative wages along the development path depends on the strength of the demand for skills exerted by technology relative to the supply of skills generated by education. Our model emphasizes that whether a country experiences rising or falling inequality as the growth rate increases depends crucially on (a) the externality (social returns) in education, (b) the evolution of the direct cost of education, (c) the elasticity of substitution in production between skilled and unskilled, and (d) the relative productivities and costs of learning-by-doing versus R&D. Neither of these variables has thus far been included in the regressions that attempt to untangle the relationship between inequality and growth across or within countries. For example, since it is likely that the use of advanced technologies lowers the elasticity of substitution of skilled and unskilled in production, the product mix of a country should also enter as an important determinant into growth and inequality regressions. Our second result is the endogenous emergence of poverty traps. Poverty traps are usually explained by the presence of positive externalities, such as threshold effects in education (e.g., Azariadis and Drazen, 1990), endogenous fertility decisions (e.g., Becker et al., 1990), coordination failures (e.g., Acemoglu, 1996), or external effects associated with human capital investment (e.g., Tamura, 1996).4 Our model provides another possible mechanism: the interdependence of the supply and demand for skilled labor when technical change is skill-biased. In contrast to previous work, where agents are assumed to be identical, our framework shows that being in a poverty trap has implications about both aggregate income and the distribution of income. In our model, development thresholds are endogenously generated, since the move from learning to R&D occurs only when the income generated from R&D justifies the factor payments to the sector. With few skilled workers, research output is small, costs of implementing new technology are high, and the benefits to the final good production are reduced. In terms of skill complementarity, the benefits generated by a small R&D sector are even further eroded as unskilled labor can easily adapt to the slow rates of technical change. The previous literature on inequality and growth falls into several different classes. The early literature stipulates imperfect mobility between distinct sectors (from agriculture to industry), which affects the distribution of labor incomes.5 In the last decade, the impact of initial inequality on growth rates has been examined by papers that emphasize the constraints to investment (in human or physical capital) imposed by capital market imperfections, or the impact of inequality on socio-political (in)stability and redistribution.6 More recently, and motivated by the experience of the US and a number of other OECD countries, interest has shifted towards the effect of growth on distribution. In particular, the rate (or direction) of technical change is seen as a major determinant of the relative wage between educated and non-educated workers.7 However, the joint determination the rate of technological change and the stock of educated workers has received little attention. One exception is Galor and Tsiddon (1996), who argue that in the presence of externalities in the accumulation of human capital, both at the aggregate and the family level, the initial distribution of income determines both aggregate income and the distribution of human capital—and therefore of earnings. Closest to our work are Galor and Moav (2000a) and Gould et al. (2000), who also maintain that it is not the ex ante distribution of income, but rather the technology-driven demand for labor that is central to explaining inequality and growth. As in our model, they assume that technological change erodes the aggregate stock of human capital due to its effect on the productivity of uneducated workers. As the number of educated workers increases, the rate of technical change accelerates due to externalities associated with human capital. Since skilled and unskilled labor are perfect substitutes, this process implies that the rate of technological progress, the number of educated workers, and the skill premium grow together during the convergence to the (unique) steady state. Although this may be a suitable description of recent trends in certain industrial countries, in a broader sample, the relationship between educational attainments and wage inequality is ambiguous, as countries with significantly different levels of human capital exhibit similar returns to education.8 Our paper explores precisely the reasons why a non-monotonic relationship between the supply of skilled workers and the skill premium may be observed.9 The structure of the paper is as follows. Section 2 presents the model, specifying the relative demand for labor, the supply of skills, and technological change. Section 3 describes the general equilibrium and stationary states, and Section 4 shows the possible patterns of inequality and growth and how they depend on parameter values. We then discuss our results in the light of empirical evidence, and examine possible policy options. Section 5 concludes.

There is no clear empirical regularity that describes the evolution of inequality along the development process, either within or across countries. Recent dual-economy models have provided a possible explanation. Expanding the original Kuznets argument to allow for intra-sectoral inequality, yields the possibility of inequality patterns different from the traditional inverse-U shaped relationship (see Anand and Kanbur, 1993b and Fields, 1993). This paper has presented an alternative mechanism based on the dual role of human capital. We have argued that the stock of educated workers in an economy determines both the degree of income inequality and the rate of growth, and that the parameters of the demand for and the supply of labor are crucial determinants of whether inequality increases or decreases as an economy accumulates human capital. The driving force of the model is a production function where the relative productivity of skilled to unskilled labor changes with the rate of technical change. New technologies are in turn generated by skilled workers, which implies that the relative demand for labor, and hence the skill premium, are not monotonically decreasing in the stock of skills in the economy. When we pair this demand function with a supply of labor function, we find, first, that multiple equilibria are possible, and, second, that as a country accumulates skills, inequality may increase, decrease, or follow a U-shaped path. The trajectory followed by a particular country will depend on the direct cost of education, the extent of externalities in the education process, and the elasticity of substitution between skilled and unskilled workers in production. Of course, we make no claim that this is the only, or even the main, determinant of inequality. We simply believe that it is an important part of the story. This is particularly so, given that the “source” of growth seems to have changed substantially. In the words of Williamson (1991, p. 90) “the mode of accumulation in the nineteenth century appears to have been much more heavily directed towards conventional capital formation, while the mode of accumulation in the twentieth century seems to have been much more heavily directed towards human capital accumulation”.