رشد اقتصادی، مزیت نسبی، وتفاوت جنسیت در نتایج مدرسه: شواهدی از تفاوت وزن دوقلوها چینی

Data from two surveys of twins in China are used to contribute to an improved understanding of the role of economic development in affecting gender differences in the trends in, levels of, and returns to schooling observed in China and in many developing countries in recent decades. In particular, we explore the hypothesis that these phenomena reflect differences in comparative advantage with respect to skill and brawn between men and women in the context of changes in incomes, returns to skill, and/or nutritional improvements that are the result of economic development and growth.

An emerging worldwide phenomenon is the rise in the schooling attainment of women relative to men, resulting in the level of schooling being higher for women than men in many countries of the world. China is a prominent example. Fig. 1, based on data from the 2005 Chinese Mini-census, displays the mean number of years of schooling, by gender and rural–urban categories, across different birth cohorts by the year each reached the age of 22. As can be seen, at least since 1965, women's schooling has risen faster than that of men in both rural and urban areas, and by 2002 in urban areas women's schooling attainment is higher than that of men. In rural areas, men and women's schooling in the most recent cohort is almost at parity by 2002, despite women's schooling being half of that of men in the 1960s.Another common finding is that the estimated rate of return to schooling for women, conventionally estimated using log-linear wage functions, is higher than that of men. This is true in almost all developed countries (Trostel et al., 2002), and is true also for a majority of all countries of the world (Psacharopoulos and Patrinos, 2004). Here again, at least in urban areas where there are superior data on earnings, China is not an exception. Zhang et al. (2005) used successive annual urban surveys from six provinces of China from 1988 through 2001 to estimate the rates of return to schooling separately for men and women by year. Fig. 2, produced from the reported annual estimates from their study, shows three phenomena: (i) a higher rate of return for women in every year, (ii) rising rates of return for both men and women, and (iii) a faster rise in the return for women than for men.In this paper, data from two surveys of twins in China, the 2002 Adult Twins Survey and the 2002 Child Twins Survey, supplemented with data from the 2005 Chinese Mini-census are used to contribute to an improved understanding of the role of economic development in affecting gender differences in the trends in, levels of, and returns to schooling observed in China and in many developing countries in recent decades. In particular, we extend the framework set out in Pitt et al. (2012) to explore the hypothesis that these phenomena reflect differences in comparative advantage with respect to skill and brawn between men and women in the context of changes in incomes, returns to skill, and/or nutritional improvements that are the result of economic development and growth.1 The framework describes optimal human capital investments in an economy in which brawn and skill contribute to production and workers sort among occupations according to their comparative advantage in the two attributes (Roy, 1951). The model incorporates two biological differences between men and women established in the medical literature — that men have substantially more brawn than women (e.g., Günther et al., 2008 and Mathiowetz et al., 1985) and that increases in nutritional inputs that augment body mass increase brawn substantially more for males than for females (e.g., Round et al., 1999). The comparative advantage explanation for the phenomena exhibited in Fig. 1 and Fig. 2 is that a rise in the skill-intensity of production (a decline in the value of brawn) leads to an increase in schooling investment overall, higher levels of schooling for workers with a comparative advantage in skill (women), and higher measured returns to schooling overall. Occupational sorting by comparative advantage means that women will be disproportionally represented in skill-intensive occupations so that the average productivity of schooling for women will be higher than that of men, and increasingly so as the occupational division of labor by gender increases.2 It is difficult to test directly this explanation given that all of these characteristics of an economy are endogenous equilibrium outcomes. Instead, we test the predictions of the model for how exogenous variation in body mass differentially affects schooling investment and wages for males and females, as these reflect both the operation of comparative advantage in occupational choice and the differential effects by gender of nutrition on brawn. We do so by obtaining gender-specific estimates of the effects of differences in birthweight within same-sex twin pairs on schooling, performance in school, and wages. These estimates by themselves are also useful in assessing directly how nutritional improvements in a population will affect schooling levels and returns by gender.3 Birthweight is known to reflect nutritional intake in the womb and to have substantial effects on child and adult health. Differences in birthweight across individuals, however, may reflect parental preferences for investments in human capital and thus any correlation between birthweight and subsequent (post-birth) investments in human capital in the general population is not informative with respect to how an exogenous early increase in nutrition affects post-birth human capital. In contrast, within-twin-pair differences in birthweight cannot reflect parental preferences. A number of studies have estimated the effects of birthweight on longer-term human capital and health outcomes using within-twin-pair birthweight variation, exploiting the fact that this variation is orthogonal to parental preferences and constraints (Behrman and Rosenzweig, 2004, Black et al., 2007, Currie and Moretti, 2007, Oreopoulos et al., 2008 and Royer, 2009). There are two limitations to these studies, however. First, none compares birthweight effects by gender.4 Yet, recent studies of the effects of randomized interventions improving the nutrition of children have found that such interventions increase schooling investment significantly more for girls (Bobonis et al., 2006, Maluccio et al., 2009 and Miguel and Kremer, 2004) and wage rates significantly more for boys (Hoddinott et al., 2008), results consistent with the framework we employ here. Second, none of these studies provides a theoretical framework linking early nutritional advantages to human capital investments.5 The Pitt, Rosenzweig and Hassan study was the first to estimate the relationships between nutrition endowments and schooling by gender and between schooling and wages using a specification embodying brawn and skill and that is consistent with the Roy model. The study has two limitations, however. First, the empirical implementation of the model was based on estimates of nutrition production functions to identify body mass endowments. Any biases in the estimates of the production functions, from incorrect functional-form assumptions, for example, will be carried over to the endowment measures used to assess the model. Birthweight differences within twin pairs are less sensitive to functional form assumptions. Second, the estimates were obtained from rural Bangladesh, which has experienced little structural change or wage growth since the early 1980s. As a consequence, it is not possible to assess to what extent changes in the returns to skill or incomes alter gender-specific human capital investment decisions, and these were not derived from their model. We explore these effects within the context of the model and empirically by exploiting the multiple birth cohorts represented in our data, the changing urban occupational structure in China, rural-urban differences in occupational structures, and variation in household incomes across households making schooling decisions. We are thus better able to assess how economic growth affects gender differences in schooling acquisition as a consequence of gender differences in comparative advantage. In Section 2 we set out the model. Parents choose the optimal amount of schooling for children who differ in gender and are heterogeneous in endowed body mass in a labor market in which brawn and skill are differentially productive across occupations. The body mass endowment affects the returns to schooling and, for males, brawn and thus male wages directly. Implications of the model are derived for how variation in the body mass endowment differentially affects schooling for males and females and how changes in the overall productivity of schooling in the economy due to shifts in the occupational structure differentially affect gender-specific schooling investments. The effects of parent income on gender differentials in the relationship between body mass and schooling are also derived. Section 3 describes the data and constructs a new times-series of a measure of the brawn-intensity of urban occupations in five cities in China from the 1970s to 2002 based on unique information from the survey data. These show a monotonic rise in the overall skill-intensity of occupations since the start of reforms in the early 1980s, consistent with the general rise in schooling levels in urban China. In accord with the comparative advantage hypothesis, the skill-intensity of occupations for women is higher than that of men, and the difference in skill-intensity increases during the period. The increase in the average skill-intensity in the occupations of women relative to that of men thus can account for the rising estimated urban “returns” to schooling for women compared with men depicted in Fig. 2. In Section 4 the methods for using twin pairs to estimate birthweight effects under varying occupation distributions in the labor market and by parental income are described. In Section 5 we report the estimates by gender of the effects of within-twin-pair differences in birthweight on schooling attainment and wages for adult (same-sex) twin aged 18-29 in 2002 and on schooling performance, homework time and parental expectations of schooling attainment and health for same-sex child twin aged 12-15. In accord with the predictions of the model, increased birthweight increases attained schooling, schooling performance, parents' expected schooling attainment and homework time significantly more for females and than for males, while having no differential effect on body mass. Increased birthweight also significantly increases the adult wage rates of men, despite having no effect on their schooling levels, indicating the importance of brawn in the economy. We also find that the difference in the effect of birthweight by gender on attained schooling increases as the skill-intensity of the occupational mix increases but is smaller at higher levels of parental income, as is consistent with the model. Our results, which exploited the rapid development of the Chinese economy, suggest that early nutritional interventions in children, here measured by weight at birth, have effects on human capital investments that are heterogeneous, by gender and by level of development. The schooling investment responses to birthweight variation appear to reflect decisions by households to invest in schooling that are attentive to the differential net returns to such investments by gender but are also constrained by available resources. Thus, the results of studies assessing the efficacy of early childhood investments from particular settings are not easily generalizable, whether from RCT's or natural variations. The theoretical results by income moreover suggest that it is not possible to predict how economic growth will affect gender differentials in the effects of nutritional interventions on schooling investment, as income growth by itself shrinks the differential while the rise in the demand for skill increases the differential. However, the finding that in the adult urban sample as the occupational mix became more skill-intensive the differential effects became sharper by gender, despite the accompanying rise in incomes that occurred in the five cities over the twenty-year period, suggests that on net economic growth that is characterized by income growth, increased body mass, and increasing skill-intensity, as in China, will increase the gap in the levels of and returns to schooling between men and women.

In this paper, estimates of the gender-specific effects of birthweight on a variety of schooling and labor market outcomes obtained from data from two surveys of twins in China were used to contribute to an improved understanding of gender differences in the trends in, levels of and returns to schooling observed in many developing countries in recent decades. In China, these include the rise in the levels of female schooling relative to male schooling, the higher level of female schooling, and the higher and rising returns to female schooling compared to male schooling in the urban labor market. Using a simple model of schooling and occupational choice incorporating well-established biological differences in brawn between males and females, we showed that the comparative advantage of women in skill is reflected in their greater schooling investment and the selection of more skill-intensive occupations than men. We also showed that comparative advantage in skill is manifested in differences in the relationship between birthweight and schooling between males and females and that these differences reflect changes in the skill-intensity of the occupational structure in the aggregate economy as well as income changes. In the case of China, we showed, using unique information on the time-of-marriage occupations of birth-cohorts spanning the period between the early 1970s and 2002, that China's occupational structure since the 1980s has become more skill-intensive in urban areas. The model we used indicated that the rise in the relative value of skill to brawn would increase the schooling of females relative to males, increase specialization according to comparative advantage, with women entering the skill-intensive occupations at a faster rate than men, and increase the effects of early nutritional advantages on schooling for females relative to males. We showed that all of these phenomena are observed in China. The greater and rising participation of women in skill-intensive occupations is also consistent with the relative rise in the estimated returns to schooling of women in urban areas of China. Our framework and findings have implications for the effects of growth policy on the male–female wage gap. For example, economic growth policies that stress industrialization will alter the occupational mix in favor of less brawn-intensive jobs, which favors women, who have a comparative advantage in skill. If economic development increases the relative demand for skill-based work, the gender wage gap should diminish. However, our estimates also indicated that in urban China brawn is still valued in the labor market. Given that men have an absolute advantage in brawn, this means that despite the higher levels of schooling of women compared to men, men may continue to earn higher wages than women even where industrialization has taken place. However, it is important to note that not all economic growth strategies devalue brawn relative to skill. Growth facilitated by improvements in the productivity of seeds in agriculture, for example, in the absence of other changes, increases the demand for manual labor and thus may not reduce the male–female wage gap. Finally, a unique feature of our study was to use micro evidence on the relationship between birthweight, an early measure of nutritional advantage, and schooling outcomes in different contexts to make inferences about the relationships between specific aspects of economic growth and schooling investments and returns. Even our very simple model of schooling investment and occupational selection, consistent with our findings, indicates that estimates of the effects of childhood nutrition on measures of human capital investment, whether based on randomized interventions or natural experiments, will not only differ by gender but also by context. Simple linear estimates obtained from a particular country on the outcomes of early-childhood nutritional interventions will thus not be generalizable unless care is taken to understand both the level of development and the macro structure.