Research Proposal The Interaction of Health, Education and Employment in Western China Principal Investigators

Download 325.71 Kb.
Size325.71 Kb.
  1   2   3   4   5   6   7
Research Proposal

The Interaction of Health, Education and Employment in Western China

Principal Investigators
Guo Li (EASRD)

Alan Piazza (EASRD)

International Consultants
Paul Glewwe (University of Minnesota)

Emily Hannum (University of Pennsylvania)

Albert Park (University of Michigan)

Chinese Consultants and Collaborators
Professor Chunming Chen (China Center for Preventative Medicine)

Dr. Hongwei Meng (China National Institute for Educational Research)

Professor Jiayi Wang (Northwest Normal University)

Professor Jianxin Zhang (Chinese Academy of Sciences)

May 3, 2002
Table of Contents

  1. Motivation and Context 1

    1. The General Issue 1

    2. Objectives of the Proposed Research 5

    3. Literature Review 10

    4. Bankwide Consultation 15

    5. Intended Audience 16

  1. Analytic Design 17

    1. Analytical Framework 17

    2. Specific Research Tasks 23

    3. Data Issues 25

  1. Organization 25

    1. Work Program 25

    2. Consultants 26

    3. Local Research Capacity Building 29

    4. Preliminary Dissemination 31

  1. Resource Requirements 31

  1. Attachments 39

1. The 2000 Gansu Survey of Living Conditions (GSCF) 39

2. Supporting Memorandum from within the World Bank 43

3. Curriculum Vitae of Main Consultants 44

4. Survey Questionnaires from the 2000 GSCF 79

Executive Summary
This research proposal is concerned with the linkages between health, education and labor market outcomes in developing countries. Past research has provided some information on relationships between these three socioeconomic outcomes, but much remains to be learned. The setting for this study is Gansu province in China, one of the poorest provinces in China. Detailed data were collected for Gansu province in 2000 on a sample of two thousand children age 9-12. The research proposed here will be based on a comprehensive re-survey of those children in 2003.

The research has the following four objectives:

  • Document the disparities in the provision of health and education services, and in health and education outcomes, in Gansu province, and how these disparities changed from 2000 to 2003;

  • Estimate the determinants of educational performance (measured in many different ways), with special attention given to the impact of child health status on educational performance;

  • Estimate the impact of educational performance on the transition to the labor market and on labor market performance; and

  • Estimate the direct and indirect (via education outcomes) impacts of children’s health on their the transition to the labor market and on labor market performance.

While much research has been done on these issues, the literature suffers from three problems. First, almost all of the existing research focuses on the general population of a given country instead of focusing on subpopulations that are more likely to be poor. Second, virtually all of the research to date uses a small number of very simple measures of human capital, with respect to both health and education, and as such it cannot explore the roles played by different kinds of human capital, or their interactions. Third, most of the existing studies have serious methodological weaknesses that raise doubts about the accuracy of the estimated relationships.

The data collected in 2000, when combined with new data to be collected in 2003 (and follow-up data to be collected in 2004 and 2005), will permit research that can overcome all of these problems. First, the data will come from one of the poorest provinces in China. Second, the data collected in 2000 and 2003 will measure many dimensions of both health and education outcomes. Third, the data collected will allow for the use of a broad range of empirical methods; for example, two randomized evaluations will be conducted and panel data will be available for all children.

This research proposal requests $271,000 from the World Bank’s Research Committee to pay for both data collection and analysis. These funds will combined with $135,000 that has already been obtained elsewhere (Fogarty Foundation and Spencer Foundation) to finance the total costs of the project. The work will begin in the summer of 2002 and be completed in December 2005.

The audiences for this project include the international public policy community (of which the World Bank is one member), the government of China (and the governments of other low income countries), and academic researchers. The results will be disseminated in the form of seminars (in China, at the World Bank, and elsewhere), working papers, and published journal articles.

I. The Motivation and Context
A. The General Issue

A growing body of research in the social sciences demonstrates the close relationship between human capital and incomes in developing countries. While human capital it sometimes narrowly defined in terms of education, in recent years health has increasingly been recognized as a key component of human capital (UNDP 2000). Researchers have examined the impacts of health and education on labor productivity, as well as the effects of income on health and education (see Behrman and Deolalikar, 1988, and Strauss and Thomas, 1995, for reviews). Studies have also examined the effects of health on education and of education on health (Behrman and Wolfe 1989; Glewwe, 1999a; Glewwe, Jacoby and King 2001; Kjellstrom, Koplan, and Rothenberg 1992).

The role played by human capital is of particular interest to the World Bank, because the Bank has financed, and continues to finance, a large number of projects in developing countries that aim at improving health and education outcomes. In education, the Bank has provided over $30 billion in loans since the 1960’s, and is currently financing 164 education projects in 82 countries. Its lending on health is on a similar scale, with new lending in health, nutritional and population of about $1.3 billion per year. In China, the Bank lent $414 million in the 1990’s for four health and education projects (Education and Development in Poor Provinces, Basic Education in Poor and Minority Areas, Rural Health and Preventive Medicine, and Rural Health Workers), and new lending is being planned for the first decade of the new millennium (e.g. $104 million for a tuberculosis control project). Given the Bank’s resource constraints, and the resource constraints faced by other donors, a better understanding of the impact of these human capital investments on health and education outcomes, and on employment outcomes is of critical importance to the World Bank.

Human capital outcomes are of particular importance in rural China, where over 800 million Chinese live. Disparities are growing in access to health and education services in rural China. Fiscal decentralization and the diminishing role of rural collectives have resulted in an increasing private share in health expenditures and growing inequities in access to health services (Henderson et al., 1994; West and Wong, 1995; Bloom and Gu 1997; Yu, et al. 1997). Use of preventive health care services has declined as costs for those services have risen, and outbreaks of preventable diseases have occurred with greater frequency as fees for immunizations have increased (Gu, et al. 1995; World Bank 1997). Moreover, evidence indicates that child malnutrition increased in rural areas in the late 1980s and early 1990s (World Bank 1997), and recent estimates suggest that over 40 percent of children in poor rural areas are stunted in their physical growth (Park and Zhang, 2000). This slippage in performance has led the World Bank and the UNDP (2000) to identify health and nutrition as priority areas for China’s poverty alleviation strategies in the 21st century.

Similarly, decentralization in educational finance has increased the private costs of schooling. This has been accompanied by increased educational disparities between urban and rural areas and between coastal and interior provinces (Davis 1989; Tsang 1994; Hannum 1999b). Many rural children in poor areas fall short of the national goal that each child complete at least nine years of education (Hannum 1999a, 2002; Brown and Park 2002). These policy shifts in health and education raise important questions about rural children's welfare, particularly in the rural areas of the underdeveloped interior provinces.

Despite the fact that the majority of health and education access problems are in rural areas, and that the majority of China’s children reside in rural settings, detailed studies of rural children’s health and schooling in China are rare. Health studies have been primarily descriptive, and none have systematically examined the joint interactions of health, education, and income. Surveys sponsored by the Ministry of Health have facilitated the monitoring of the nutrition and health status of the population (e.g., China Center for Preventative Medicine and State Statistical Bureau, 1998), and the well-known China Health and Nutrition Survey, a panel survey of households in eight provinces conducted in 1989, 1991, 1993, 1997, and 2000, has documented health outcomes and some of their socio-economic correlates (Popkin, 1994; Popkin et al, 1993, Zhai et al., 1995). However, these datasets contain little information on educational outcomes. Moreover, they do not target a specific cohort of children, and they lack randomized trials to identify more convincingly the effects of specific policies on outcomes of interest.

Large-scale studies of education in China have been even more limited. Analyses to date do little more than describe disparities in enrollment or attainment by location of residence, poverty status, gender, and/or ethnicity. Examples of this literature are Hannum and Xie 1994; Zhang 1998; Brown and Park 2002; Hannum 2002.

Much research on human capital has been done in other developing countries, but the research to date suffers from several limitations. First, while much of the interest in this relationship centers on the role that human capital could play for poor and otherwise disadvantaged households, almost all of the current research focuses on the general population of a given country instead of focusing on subpopulations that are more likely to be poor. Second, almost all of the research in this area uses a small number of very simple measures of human capital, both in terms of health and education, and as such that research cannot explore the roles played by different kinds of human capital, or their interactions. Third, most of the existing studies have serious methodological weaknesses that raise doubts about the accuracy of the estimated relationships.

The research proposed here addresses each of these three deficiencies in the literature. First, it focuses on one of the poorest areas of China, Gansu province. In 2001 Gansu ranked 29th out of 31 provinces in terms of per capita income. A recent World Bank report (World Bank and UNDP, 2000) shows that 23% of the population in Gansu is poor, compared to 6.5% for China as a whole. That report also states that “the educational, health and nutritional status of China’s remaining absolute poor is deplorable” (p.70). Second, the data used will have multiple measures of human capital, including both health and education outcomes, many of which have not been collected in past household surveys of poor populations in developing countries. Third, the panel data from Gansu province will be extremely rich, allowing for the use of a much larger set of econometric tools to analyze various relationships. Moreover, two specific questions (the impact of current vision and current nutritional status on health and education outcomes) will be addressed using prospective randomized trials.

The knowledge gained from the research proposed here will be important not only for China but for other countries as well (although bear in mind that China contains one fourth of the world’s population living in developing countries). It will have immediate implications for policy in several other developing countries. For example, Vietnam faces many of the same issues found in China. Rural areas are lagging behind urban areas, and financing for health and education has become more precarious as Vietnam has decentralized that financing. While China has a much higher income than Vietnam, Gansu province has an income level and a poverty rate comparable to that of Vietnam. Moreover, China and Vietnam are only two of several countries in Asia that have made the transition from a planned economy to a market economy in the last two decades; the findings here should also be relevant for Cambodia, Kazakhstan, Laos and Mongolia, and perhaps in the future for North Korea. One interesting similarity of Gansu province with Mongolia and Kazakhstan is that all three are relatively dry areas with harsh climates.

This research is also relevant for developing countries that have never had planned economies. For example, the research on the impact of the provision of eyeglasses (see below) is relevant for many East Asian countries, whose populations are particularly prone to poor visual acuity. A case in the point is the Philippines; data collected in the Cebu metro area show that about 20% of 12-year-old children have serious vision problems but only about 1% wear glasses.
B. Objectives of the Proposed Research
The proposed research will use a panel data set that will be generated by reinterviewing all households that participated in the 2000 Gansu Survey of Children and Families (GSCF), a very detailed survey of 2000 children aged 9-12 in 20 rural counties in Gansu province. The survey instruments included questionnaires for children, mothers, households, teachers, school administrators, and village leaders, as well as tests of child cognitive ability and achievement in mathematics and language skills. The 2000 GSCF is probably the most detailed education-focused household survey ever conducted in China. It also includes several health measurements. See subsection II.A below for a detailed description of the education and health data collected in the 2000 GSCF.

With the 2000 GSCF and the data collected in 2003, the research will have the following four objectives:

  • Document the disparities in the provision of health and education services, and in health and education outcomes, in Gansu province, and how these disparities changed from 2000 to 2003;

  • Estimate the determinants of educational performance (measured in several different ways), with particular attention given to the impact of child health status on educational performance;

  • Estimate the impact of educational performance on the transition to the labor market and on labor market performance; and

  • Estimate the direct and indirect (via education outcomes) impacts of child health on the transition to the labor market and on labor market performance.

The specific research questions to be analyzed are:

  1. What disparities exist in the provision of health and education services in Gansu province, and how have these disparities changed over time?

Comment: One consequence of China’s economic reforms is that both the health and education sectors increasingly rely on user fees to cover costs (Tsang 1994; Bloom and Gu 1997). Health care researchers have observed declining use of health services with this shift toward linking access with households’ ability to pay (Bloom and Gu 1997; Yu, Cao, and Lucas 1997). In 1997, the percentage of poor counties in China with lower secondary school enrollment rates above 85 percent was only 40 percent, compared to 70 percent of all counties (World Bank, 1999). The World Bank’s poverty alleviation projects in rural China have put priority on improving the quality of basic health and education services (Dahlman, Carl, and Jean-Eric Aubert. 2001), so the Bank has a stake in monitoring recent changes.

  1. What disparities exist in health and education outcomes in Gansu province, and how have these disparities changed over time?

Comment: According a recent World Bank study (1997), child malnutrition increased in rural areas between 1987 and 1992; other evidence generally confirms a lack of improvement in health indicators in China in the early 1990s (Bloom and Gu 1997). Preventative care has declined with rising costs for services, and outbreaks of immunizable diseases have reportedly increased as immunization fees have risen (Gu et al. 1995; World Bank 1997). In education, new direct costs present significant barriers to the education of poor children in rural China, particularly above the primary level (Hannum 2002; Hannum and Park 2002).

  1. What is the impact of early childhood nutrition, as measured by current height-for-age and other indicators of chronic malnutrition, on school performance in terms of scores on academic tests and tests of “life skills”.

Comment: Evidence from developing countries shows a strong link between early childhood nutritional status and subsequent school performance (Glewwe, Jacoby and King, 2001). In Gansu province, child malnutrition is quite common; according to data collected in 1997 in Tongwei County, approximately 32% of children age 12 and under are stunted (calculations by Albert Park using data from the 1997 China Rural Poverty Survey).

  1. What is the impact of early childhood nutrition, as measured by current height for age and other indicators of chronic malnutrition, on indicators of psychological well-being?

Comment: The 2000 GSCF was the first large-scale survey to measure psychological well-being among rural children in China, and piloted items for scales tapping anxiety, depression, aggression, stress, locus of control, self-concept, and resilience. The data from the 2000 GSCF indicate potentially serious problems regarding psychological well-being. For example, of all the children in the sample, 12% fully agreed that they “feel inferior to others”, 7% fully agreed that they “sometimes threaten or hurt others” and 13% fully agreed that they are “always worried”. In the GSCF for 2003, additional emphasis will be placed on developing batteries of items related to adolescent risk-taking behavior. Multiple-item scales will be used in analyses of psychological well-being.

  1. What is the impact of current health and nutritional status on school performance?

Comment: In China the most common physical health problems among school-age children include myopia, malnutrition, iron deficient anemia, and ascariasis (Meng and Bloom 1990; Zhang, et al. 1995; Shi, Zhu, and Zhou 1990). Estimates suggest that 60% of children age 6-10 have low intake of Vitamin A (Dr. Geng Peng-fei, Deputy Director, Gansu Center for Preventive Medicine, personal communication). As explained below, this analysis will be based on a randomized trial.

  1. What is the impact of poor vision on school performance?

Comment: It is estimated that 20-30% of middle age school children in China suffer from serious vision problems (Shi et al, 1990), yet the GSCF data collected in 2000 show that only 8% of children age 9-12 in Gansu wore glasses, even though 20% reported vision problems. As explained below, this question will also be addressed using a randomized trial.

  1. What is the correlation between psychological well-being status and school performance?

Comment: A more ambitious question could be asked, namely: what is the impact of psychological well-being on school performance? However, there are serious methodological difficulties with trying to estimate this causal relationship. We will attempt to overcome them, starting with a simple difference-in-differences estimator, but at this point we can only promise to measure the correlation between psychological well-being indicators and school performance. This will provide a first approximation of the magnitude of the relationship.

  1. How do different aspects of school performance in early adolescence (age 9-12) predict schooling and employment outcomes in early teenage years (age 12-15)?

Comment: It may be possible to estimate the causal impact of school performance in early adolescence on later school and employment outcomes, but there are serious simultaneity problems with such estimation. At minimum we will be able to predict outcomes in teenage years (primarily school continuation or dropping out, and employment prospects for those who drop out), based on school performance in the pre-teen years. This should provide useful “early warning signals” of which students are “at risk” of dropping out of school.

  1. Which skills learned in school affect the earnings of teenage workers?

Comment: The 2000 GSCF administered tests in mathematics, reading comprehension, and “cognitive development”. The 2003 GSCF will administer these same tests, along with specialized tests that focus on different kinds of literacy skills (see below). In 2003 the children will be 12-15 years old, and about one third to one half will have left school. Many of those who leave school will be working.

  1. What is the impact of early childhood malnutrition and current health status on probability of employment in early teenage years?

  2. What is the impact of early childhood malnutrition and current health status on income from employment?

Comment on questions 10 and 11: The causal link from health to employment outcomes has received increasing attention in recent years (Fogel 1994, Strauss and Thomas 1998), yet almost all studies have only very limited data, often measuring both health and employment outcomes at one point in time. None of the children in the 2000 GSCF were working in 2000, so their health in that year cannot be attributed to their working conditions in 2003.
C. Literature Review

A growing body of empirical research on adults finds a positive impact of human capital (both health and education) on labor productivity in developing countries (Strauss, 1986, in Sierra Leone; Sahn and Alderman, 1988, for rural Sri Lanka; Thomas and Strauss, 1997, for urban Brazil; Croppenstedt and Muller, 2000, for rural Ethiopia; Behrman and Deolalikar, 1989, for rural India; Fafchamps and Quisumbing, 1999, for rural Pakistan; Haddad and Bouis, 1991, and Foster and Rosenzweig, 1992 and 1994, for the Philippines; and Glewwe, 1999b, for Ghana). The importance of education to labor productivity has received great policy emphasis, but there are many methodological problems that could lead to bias in estimated returns (Schultz 1988; Barro 1991; Behrman and Deolalikar 1993; Glewwe, 1996). Moreover, the process by which childhood health and educational attainment translate into adult labor productivity is poorly understood. There is a literature that examines the effect of child health on schooling outcomes (for China, see Jamison 1986), but these studies have been inconclusive, largely because of challenging identification problems (Behrman 1996). A recent study by Glewwe, Jacoby, and King (2001) exploits longitudinal data from the Philippines to identify more convincingly the effects of child height on test scores, and finds significant positive effects. The study does not, however, relate health and learning to labor market outcomes and restricts itself to narrowly defined measures of both health and learning.

While these studies have provided valuable insights about the impact of health on schooling and the impact of health and schooling on productivity, several limitations have prevented them from fully illuminating the questions of interest. The first is that the populations studied tend to be “national” populations that include both poor and better off households. Sample size limitations often force researchers to estimate effects that are averages over poor and non-poor households. If non-linear effects exist, such as larger impacts of child malnutrition on education outcomes among the poor (relative to the non-poor), they will not be captured in estimates of average effects.

Second, the majority of studies of health and schooling have employed a very limited set of measures of health and education outcomes (see Strauss and Thomas, 1995). This point is important, so the following paragraphs discuss it in detail.

Economic studies typically use narrow definitions of health and education outcomes. To more fully understand the relationships between health, schooling, and employment, studies should identify and measure the health outcomes most directly related to adolescents’ lives, and those educational outcomes likely to have the greatest impact on productivity. We propose a broader conceptualization of health that adds psycho-social (mental health) indicators, health behaviors, and non-traditional direct assessments of health to the standard repertoire of health measures. We will also use newly-developed “life skills” measures of literacy, numeracy, and analytic ability designed to relate more directly to productivity outcomes than traditional academic tests.

Consider first the health measures. Evidence suggests that psycho-social factors, such as level of personal control, sense of agency, self-concept, and stress, as well as health behaviors are particularly significant in adolescence. Adolescents in developed countries are susceptible to influences in the social environment that encourage risky behaviors such as smoking, violence, alcohol and drug use, or risky sexual behavior (Smedley and Syme 1999).

The example of smoking is illustrative. Studies from developed countries show that most smokers start smoking during their adolescent or early adulthood years (CDC 1994). Adolescent smoking may impede human capital acquisition (Galambos and Silberseisen 1987, Kleinman, et al. 1988, Eckert 1983), and thus may indirectly affect adult productivity and health outcomes. Recent evidence from China suggests a parallel set of issues. The age of smoking initiation fell by approximately 3 years for both men and women between 1984 to 1996 (Yang et al. 1999) and high rates of smoking initiation occur as early as elementary school among boys (Sun and Cui 1999).1 Smoking among Chinese adolescents is associated with poor school performance and problem behaviors such as truancy from school, running away from home, destructiveness and fighting (Li et al. 1996, 1999). While measures of psychosocial well-being and risk and protective factors for adolescents have been widely used in developed country contexts, only a few studies of health, human capital acquisition, or productivity in developing countries have examined psycho-social factors and/or health behaviors.

The majority of economic studies of children’s health and schooling have employed anthropometric data, especially child height, as the primary indicator of health. Yet anthropometric measures alone may miss entirely some health problems, such as micronutrient deficiencies, and for other health problems height is just a rough aggregate indicator. Further progress in understanding which dimensions of health matter most requires simultaneous use of multiple health measures (Strauss and Thomas 1998).

Similarly, there is a need for better conceptualization and measurement of the dimensions of learning that are most fundamentally linked to labor productivity. Academic achievement tests or years of schooling are the most commonly used measures of learning, but both are highly imperfect measures of the development of skills that are relevant for productivity outcomes (OECD and Statistics Canada 2000). Literacy and numeracy, on the other hand, have a more demonstrable effect on economic and social well-being. The importance of such skills increases with the process of economic development as the importance of information and technology increases. In many surveys, however, these skills are ignored or at best respondents are asked to rate their own ability rather than directly testing them.

Increasingly, the need for standardized measures of literacy or life skills has been recognized by researchers and the policy community. The recent International Adult Literacy Survey (IALS) initiative (OECD and Statistics Canada, 2000) represents an important step in the direction of developing international standards for measuring productivity-related skills. The International Adult Literacy Surveys (IALS), begun in the 1990s, have been conducted in 14 industrialized countries. Recent studies using IALS data from OECD countries have found that important differences in literacy skills exist across and within nations, and that literacy skill deficits affect large portions of the adult population. Literacy is strongly correlated with success and the use of opportunities, but is not synonymous with educational attainment (OECD and Statistics Canada 2000).

In 1998, UNESCO sponsored a related project to explore how adult literacy, numeracy, and life skills could best be measured in developing country settings. The added focus on life skills, which include knowledge, attitudes/values and actions, critical awareness, and problem solving, was thought to be particularly important in developing countries where large fractions of the population have low educational levels. The project was headed by Dr. Daniel Wagner, Director of the International Literacy Institute at the University of Pennsylvania, and core members included Dr. Bruce Murray of Statistics Canada and Dr. Hongwei Meng of the China National Institute for Education Research. (Dr. Meng is one of the main Chinese collaborators in our project.) Background reports have been prepared for four developing countries, including China (an initiative led by Dr. Meng). Like the IALS, the UNESCO instruments include a very carefully designed set of questions that measure different aspects of literacy, numeracy, and other basic skills.

The third problem with the current literature is that almost all studies have serious methodological problems. For example, adult labor productivity is commonly modeled as a function of education, primarily accumulated in childhood, and current health. However, the directions of causation are not so simple. Productivity can affect health through income and occupational choice, and adult health is the outcome of early childhood educational and health experiences. Moreover, years of schooling is often an imprecise measure of educational human capital; although one could replace it with measures of current cognitive skills, those skills may be partly determined by the type of work done and thus would be endogenous.

The data to be collected as a part of this research proposal will provide alternative estimation methods that can overcome these methodological problems. For example, we will have data on health and education status (including cognitive skills) for the children when they were 9-12 years old, before they entered the labor market. These measurements of human capital are not affected by subsequent experience in the labor market, and thus can serve as instrumental variables for human capital in the teenage years. Another innovative aspect of this research proposal is that two of the questions to be addressed (the impact of vision and of current nutritional status on education and employment outcomes) will be analyzed using randomized trials, as explained below.
D. Bankwide Consultation

This research proposal has been circulated and presented in several ways among staff at the World Bank. First, it was circulated in draft form to Bank staff working on China, and comments were received from those Bank staff. Second, it has been circulated to and discussed with Bank staff in the HD network, including Shanta Devarajan. Third, a seminar explaining the proposal was held on April 19, 2002, at the Bank, inviting a wide variety of Bank specialists who work on health and education, including staff in the HD network anchor and from DECRG. Many useful comments were received, almost all of which have been incorporated into this report. We have received strong support from the region and from staff in the HD network and from DECRG. One memo expressing support, from Emmanuel Jimemez (the Sector Director for Human Development in the East Asia Vice Presidency) is included as Attachment 2.
E. Intended Audience

There are three audiences for the findings produced by this research proposal. The first is the international public policy community, especially those concerned with poverty, health and education in China and other low-income countries. This includes the staff of international development agencies, such as the World Bank, the UNDP and the Asian Development Bank, and many non-governmental organizations (NGOs) that work in low-income countries. Initial results will be presented at seminars for World Bank staff in Washington, and aid agencies with offices in China will be invited to a workshop in Beijing for Chinese policymakers (see below).

A second target audience is Chinese policymakers and researchers. Preliminary results will be presented in Beijing and in Lanzhou (the capital of Gansu province). The research team also has close contacts with health and education officials in Gansu. Our main local collaborator, Jiayi Wang, has close ties to provincial education officials, and is considered a national expert on education problems facing poor and minority areas. We are also working directly with research staff in the Ministries of Education and Health to help us design appropriate tests for academic achievement and physical health. For example, Ge Pengfei, Deputy Director of Gansu’s Center for Preventative Medicine under the provincial health bureau, will supervise the nutrition and vision interventions and the physical health measurements. These experts will effectively communicate relevant research findings to local policy makers.

The research team is also working closely with national organizations and ministries in China. Within the national Ministry of Education, Professors Li Zhang and Mansheng Zhou, Director and Vice Director of the National Center for Research on Educational Development, were consultants on the 2000 GSCF. Our collaborator Chunming Chen, professor and former director of the Chinese Academy of Preventive Medicine under the Ministry of Health, is a well-known national expert on nutrition and health issues who has direct contact with national Ministry of Health officials. Two other collaborators, Professor Jianxin Zhang, Director of the Child Development Institute in the Chinese Academy of Sciences, and Meng Hongwei, Director of the Research Center on Educational Measurement and Monitoring at the China National Institute for Educational Research, also actively provide national policy advice.

The third target audience is academic researchers. China’s recent performance in poverty reduction and human capital development is of great interest to these researchers. Results will be disseminated to this audience in the form of journal articles, and perhaps a book. Dissemination and final products are discussed further in subsection III.D below.
II. Analytic Design

A. Analytical Framework

As explained above, the research proposed here will overcome the three deficiencies in the literature. First, by undertaking the research in a very poor province of China, the research will focus on a population that is of particular concern regarding the impact of human capital on labor incomes.

Second, the data used in the research are unusually rich in terms of the variety of health and education outcomes measured. The health data collected in the 2000 GSCF include self-reported (by the child or by his/her mother): current weight, current general health status, birth weight, vision problems, psycho-social indicators (self-esteem, depression, anxiety), recent illnesses (including respiratory problems and diarrhea), food frequencies, and availability and use of health care. Data were also collected on health facilities in the village and on household members’ use of those facilities. The data to be collected in 2003 will include all of this information plus physical measurements (using appropriate equipment) of: height, weight, lung capacity, iron intake (from pin-prick), Vitamin A deficiency (via “dark adaptation” tests), zinc and magnesium deficiency (from hair samples), visual acuity, and blood pressure. In addition, detailed information will be collected on the child’s history of smoking, exposure to passive smoking, alcohol consumption, reproductive health knowledge, eating habits, and recent use of health care services.

The education data will be even richer than the health data. In the 2000 GSCF the following data were collected for the target child: current years in school, daily attendance, years repeated, year entered school, and test scores on mathematics, reading and abstract thinking ability. A full set of items regarding children’s educational aptitudes, aspirations and attitudes, study habits and time allocation, and classroom environment were also asked in questionnaires administered to the child, his or her mother, and his or her main teacher. Data were also collected on all teachers and principals in primary schools serving the sampled villages. In 2003, the same data will be collected, not only for the 2000 children in that survey but also for one younger school-age sibling. In addition, for both the original children and the younger children tests will be administered on “life skills”, which measure the application of academic skills to everyday situations. More specifically, tests will be given on prose literacy (reading comprehension of everyday materials such as newspaper stories and information pamphlets), document literacy (the ability to understand maps, diagrams, charts, etc.) and numeracy (adding numbers on bank account slip, calculating interest charges on loans, etc.). Note finally that the data to be collected in 2003 may be supplemented with additional information. The research team will travel to Gansu in the summer of 2002 to discuss in detail the kinds of data to be collected in 2003.

Third, the data collection and analysis plans are designed to overcome many of the methodological deficiencies of past research. The following paragraphs briefly explain how the data to be collected in Gansu can be used to answer each of the 11 specific research questions presented above in subsection I.B.

  1. What disparities exist in the provision of health and education services in Gansu province, and how have these disparities changed over time?

  1. What disparities exist in health and education outcomes in Gansu province, and how have these disparities changed over time?

These first two questions can be answered using descriptive data from the 2000 GSCF and the data to be collected in 2003. The data on education in the 2000 GSCF has sufficient detail to allow us to detect significant changes in education variables between 2000 and 2003; it is less clear whether changes can be detected for health over that time period because the data on health in the 2000 GSCF are less detailed.

  1. What is the impact of early childhood nutrition, as measured by current height-for-age and other indicators of chronic malnutrition, on school performance in terms of scores on academic tests and tests of “life skills”.

The methodology of Glewwe, Jacoby and King (2001) will be employed. In particular, the analysis will compare within-household differences between siblings who attend the same school. The dependent variable will be the difference in the siblings’ test scores, and the key explanatory variable will be differences in height-for-age. Since the latter difference is likely to be endogenous, birth weight information on the older sibling will be used as an instrumental variable.

  1. What is the impact of early childhood nutrition, as measured by current height for age and other indicators of chronic malnutrition, on indicators of psychological well-being?

The methodology of Glewwe, Jacoby and King (2001) can again be used. The only difference is that the dependent variable will be sibling differences in measures of psychological well-being.

  1. What is the impact of current health and nutritional status on school performance?

The key methodological problem here is that current health and nutritional status are likely to be endogenous with respect to school performance. We will use two approaches. First, we plan to implement a randomized intervention in which nutrient-enriched wheat flour will be provided to a randomly selected subsample of the households in the survey. This will start immediately after the 2003 survey. Results will be measured by brief follow-up surveys in 2004 and 2005 that will collect data only on the main outcome measures of interest (school enrollment, scores on achievement tests, height, weight, Vitamin A, iron, visual acuity, blood pressure, employment and wage income). Second, a standard instrumental variables approach will be used. The main instruments for current health and nutritional status will be prices of food items, source of water, distance to local health facilities, type of toilet, recent weather and local outbreaks of illnesses. It may also be possible to estimate sibling difference outcomes using a subset of these instrumental variables (those that vary over time); that is, one may compare the academic performance of the older sibling in 2000 with that of the younger sibling in 2003.

  1. What is the impact of poor vision on school performance?

Visual acuity of children is likely to be exogenous. The main exception to this is that those children who have glasses will have better functional acuity, and purchase of glasses is endogenous. Two approaches will be used to avoid potential problems of endogeneity bias. First, a randomized trial will be conducted in 2003 whereby a random sample of children with poor vision will be provided eyeglasses. Special surveys in 2004 and 2005 will collect information on school performance of all the children in the sample.

Second, instrumental variable methods will be used. Possible instrumental variables for use of eyeglasses are availability of optical services, price of glasses, household income (perhaps lagged to 2000) and Vitamin A intake.

  1. What is the correlation between psychological well-being and school performance?

As pointed out in subsection I.B, it is simple to obtain correlation between any variables of interest. The analysis will include raw correlations and partial correlations (i.e. correlation in a regression framework after controlling for other covariates). It may be possible to find methods to estimate causal relationships, but this is uncertain.

  1. How do different aspects of school performance in early adolescence (age 9-12) predict schooling and employment outcomes in the early teenage years (age 12-15)?

Prediction does not necessarily imply causation, so predictive relationships can be estimated without concern for bias. This information will be useful for identifying children “at risk” of dropping out school. Attempts will be made to tease out a causal relationship, but it is unclear whether a credible identification strategy can be found.

  1. Which skills learned in school affect the earnings of teenage workers?

Simple OLS estimates of the impact of current skills on earnings could be biased because the type of work may affect skill retention and development. The best way to overcome this problem is to use skills acquired by 2000 to predict current skills (2003). None of the children were working in 2000, so skills in 2000 are not influenced by work experience in 2003. Another advantage of using instrumental variables is that, since skills are likely to be measured with error, it will remove measurement error bias. Finally, since the number of children working for wages in 2003 may be low, brief follow-up surveys will be conducted in 2004 and 2005, at which time a larger percentage of the sampled children will be working for wages.

  1. What is the impact of early childhood malnutrition and current health status on probability of employment in early teenage years?

  2. What is the impact of early childhood malnutrition and current health status on income from employment?

The answers to questions 10 and 11 can be obtained using estimates derived to answer earlier questions. In particular, estimates of the impact of early and current child nutrition on school performance (questions 1 and 3) can be combined with estimates of the impact of academic skills on income and the probability of employment (questions 8 and 9). Still, some additional estimates will be needed to calculate the impact of current health status on employment and income. Since the indirect impact of health on employment outcomes is already accounted for in the current educational characteristics, past values of health variables (i.e. their values in 2000) can be used as instruments for current health (which measures the direct effect of health on those outcomes).

B. Specific Research Tasks

This project will begin in the summer of 2002 and end in December 2005. There are three main tasks: finalize plans for collection of new data; collect new data in 2003 (with brief follow-up surveys in 2004 and 2005), and analyze the data in 2004 and 2005.

Pre-survey work, including designing the survey instruments and detailed planning of the survey design, will occur beginning the summer of 2002. Hannum, Park, and Glewwe will travel to Gansu and complete the following tasks: discuss and test preliminary versions of the survey instruments, focusing in particular on health-related and other measures not included in the 2000 GSCF; investigate the feasibility of different options for designing the randomized intervention (taking into consideration financial costs, intrusiveness of design, receptiveness to treatment by individuals and communities, etc.); learn from face-to-face interviews with children, teachers, principals, health officials, and collaborators to better refine questions and research design; explore possibilities for collecting complementary secondary data (e.g., on health and education services in previous years, rainfall data, etc.). Chinese collaborators will conduct additional pre-tests of survey instruments and protocols as necessary throughout 2002.

The main survey will be undertaken in June 2003, just before children and teachers leave for summer vacation. At this time, or in September (the beginning of the school year), the vision and nutrition interventions will be initiated by provincial collaborators at Gansu’s Academy of Preventive Medicine. Based on judgments about the receptivity of respondents, some parts of the survey (e.g., physical health measurements, attainment and ability tests, literacy tests) may be administered at staggered times, but all during the summer of 2003. The following two summers, in 2004 and 2005, each child will be resurveyed on a limited set of health, education and employment outcomes, including most health measurements, achievement, ability, and literacy tests, and educational experience in the past year, and employment and wages.

Data will be entered in Lanzhou (the capital of Gansu province) by the provincial State Statistical Bureau, and supervised by graduate students of Northwest Normal University. The data should be ready for analysis by the fall of 2003. During 2004 and 2005, the consultants will take primary responsibility for the analysis and production of written academic papers, and will collaborate on specific projects with the Chinese investigators. During this time they will also prepare for conducting a third wave of the survey in 2006.

C. Data Issues

The data will be maintained and archived at the Population Studies Center of the University of Pennsylvania. Data will be made publicly available after project investigators have had time to undertake initial analysis of the data, and no longer than three years after the data are entered. An archivist at the Population Studies Center will provide documentation and web access to the data. For a detailed description of the 2000 Gansu Survey of Families and Children, see Attachment 1.

III. Organization
A. Work Program

The research team will be overseen by Guo Li and Alan Piazza, both of EASRD. The proposed project brings together a multidisciplinary group of scholars from leading institutions in both the U.S. and China. The research team has extensive and complementary experience studying the sociological, economic, psychological, educational, and public health aspects of economic development, especially in rural China. The main investigators on both the U.S. and Chinese sides have a wealth of experience collecting primary data in China, and many members of the group successfully collaborated in the GSCF 2000 survey of two thousand children and their mothers, households, teachers, principals, and community leaders. There will be three American academic researchers and a larger number of Chinese researchers. Summary information on the main consultants is provided in the following paragraphs. Their c.v.’s are included in Attachment 3.

Most of the work on this research project will be done by the consultants. A project office will be established at University of Pennsylvania to organize research, training, and conference activities. The University of Pennsylvania, the University of Michigan, and the University of Minnesota will each provide faculty and graduate students to analyze the data.

Many Chinese organizations and personnel will assist in the collection and the analysis of the data. The State Statistics Bureau (SSB) of China will oversee the overall data collection, in collaboration with the following institutions. The Chinese Academy of Preventive Medicine (CAPM) will develop the physical health components of the project. The Chinese Academy of Sciences (CAS) will develop the psychosocial health components of the project. The China National Institute for Educational Research (NIER) will develop the literacy measures. Finally, the Northwest Normal University (NWNU) will assist in the school-based elements of the survey. Each of these institutions will conduct analyses of the data, with cross-institution collaborations emerging according to research interests. Researchers from these institutions will give feedback on others’ ongoing studies and share research results via project workshops and a project Web site.

Guo and Piazza will keep operational staff at the Bank informed about the progress of the research. Results that are useful for specific operational work will be conveyed immediately after they are found.
B. Consultants

Paul Glewwe is Associate Professor of Applied Economics at the University of Minnesota. He also worked in the research vice presidency of the World Bank for over 13 years. His research focuses on poverty, inequality, education and nutrition in developing countries. He also has extensive experience in implementing household surveys in developing countries, including a survey in Northeast China in 1995.

Emily Hannum is Assistant Professor in Sociology and Population Studies at the University of Pennsylvania. Her research has focused on human capital investment, in particular educational attainment and its relationship to income level, occupational attainment, and inequality, as well as poverty and children’s social welfare. Her work in China includes papers on trends in gender and urban-rural inequality in primary and secondary education (Hannum and Xie 1994; Hannum 1999a), changes in the impact of education on income and occupational attainment (Xie and Hannum 1996; Hannum and Xie 1998), and inequities in access and participation in rural primary and secondary education under market reforms (Hannum 1999b, 2002; Hannum and Park 2002). She was principal investigator for the 2000 GSCF, funded by the Spencer Foundation, and is co-directing (with Albert Park and Jiayi Wang) the Gansu Rural School Networks Project, a randomized experiment that combines an IT-based school networking intervention with a survey and ethnographic study of rural school culture, management, and performance.

Albert Park is Associate Professor of Economics at the University of Michigan. He is a development economist whose research has focused on rural development issues in China. Current projects focusing on health and educational outcomes in China include papers on mother’s schooling and child health (Park and Zhang 2000), gender bias in household nutrient allocation (Park and Rukumnuaykit 2000), educational attainment in poor areas (Brown and Park 2002), and urban labor market development. He was the principal investigator for a large collaborative research and training project on rural poverty in China funded by the Ford and Luce Foundations (1996-1999), and has directed rural household surveys in northwest China as part of his dissertation field research in 1993, funded by Fulbright-Hays and Ford Foundation. He also played an important roles in the implementation and analysis of the 2000 GSCF.

Professor Chen Chunming is based at the China Center for Preventative Medicine (CAPM). The CAPM is an influential research institute within the Ministry of Health. Dr. Chen is one of China’s leading health and nutrition experts (two recent publications include Chen 1996, 1997) and directed CAPM’s participation in the University of North Carolina’s China Health and Nutrition Survey, which has been one of the best sources of recent information on physical health outcomes in China.

Dr. Meng Hongwei is Director of the Research Center for Educational Measurement and Monitoring, China National Institute for Educational Research (NIER). Dr. Meng has a doctorate in educational evaluation and has extensive experience directing national and international literacy projects, as well as rural education projects. Most recently, he was appointed to head the Drafting Committee for the China National Education for All Action Plan. He is also on the Governing Board of the UNESCO Institute of Statistics. He has written numerous scholarly articles and policy documents on issues of literacy assessment, rural education, and educational evaluation and monitoring in China.

Professor Jiayi Wang is Vice-President of Northwest Normal University and an advisor to provincial and national policy makers on rural poverty and education issues. He has a doctorate in teaching and learning and works in the areas of educational measurement and research methodology (e.g., Wang 1993, 1995, 1998; Wang and Pei 1997). He directs research programs in schools in rural Gansu, and has directed numerous large-scale research and development projects related to poverty and education in western China supported by Chinese, British, and Canadian development organizations. He was a key investigator for the 2000 GSCF, and is the Chinese principal investigator for the Gansu Rural School Networks Project (described above).

Professor Jianxin Zhang is Director of the Division of Developmental and Educational Psychology in the Institute of Psychology at the Chinese Academy of Sciences. He developed the Chinese Personality Assessment Inventory (CPAI), which includes indigenous personality constructs. He is also adapting a Chinese General Self-Efficacy Scale, and designed the cognitive tests for rural children in the 2000 GSFC.

The Gansu Province office of the State Statistical Bureau (SSB), led by Tiemin Tang, Senior Statistician and Head of the Gansu Survey Team for the Urban Societal Economy, and Ruijun Dong, Director of the Gansu Survey Team for the Urban Market Economy, will be subcontracted to collect the survey data. The Gansu SSB has extensive experience collecting urban and rural household income and expenditure data on an annual basis and have survey staff in all counties in the province. Directors Tang and Dong oversaw the data collection and data entry processes for the 2000 GSCF.

Other project participants bring valuable complementary expertise to the project. Dan Wagner, Director of the International Literacy Institute, University of Pennsylvania, and Bruce Murray, Director of the IALS project at Statistics Canada, will be resource persons for the project. Dr. Meng has collaborated closely with Dr. Wagner on previous projects, upon which our instruments will be based.

C. Local Research Capacity Building

The proposed activities build capacity in two ways. First, the collaboration with Northwest Normal University builds capacity for conducting research on rural poverty and education in a part of China where these issues are extremely pressing. Northwest Normal University is geographically well-situated to sustain intensive, longitudinal research

on rural educational development issues. It offers research and training centers that focus on rural and minority education, as well as graduate students and faculty who have extensive experience working, implementing development projects, and conducting research in impoverished rural and minority schools.

Yet, as at other universities in China’s interior, resources to support large-scale research projects at Northwest Normal University are limited. Opportunities for international exchange and access to the international research literature are also limited. The long-term collaborative relationships established through the 2000 and 2003 GSCF will contribute to developing NWNU’s research capacity, providing invaluable experience on the implementation of large, international research projects in the NWNU’s areas of expertise. Activities undertaken as a result of the 2000 GSCF include collaboration on research proposals, research conferences, and opportunities for short-term research visits to the US for faculty and graduate students at NWNU. We expect to continue and extend these activities with the 2003 GSCF.

Second, the 2003 GSCF project is establishing new, cross-institution ties that may lead to more research on education, poverty, and health in western China. For example, one benefit of the 2000 GSCF has been the establishment of a relationship between the Gansu Statistics Bureau and Northwest Normal University. This relationship has led to conversations about sharing

Statistics Bureau data with Northwest Normal University for purposes of policy research on educational access and outcomes. Similarly, with the 2003 GSCF we anticipate that the growing connections between Northwest Normal University, the Statistics Bureau, and the psychological and physical health institutes in Beijing will facilitate future collaborations linking education, poverty, and health issues.

D. Preliminary Dissemination

The project will produce several sets of papers, the preliminary results of which will be presented at workshops and conferences in China and the U.S. Academic research papers will be disseminated via a working paper series planned to run through the University of Michigan Population Studies Center Research Reports and the Chinese Academy of Social Sciences Institute of Population Studies Research Reports. Researchers will also submit papers to academic journals in the fields of economics, sociology, public health, psychology, education, and child development. Finally, policy and procedure reports will be developed in consultation with consultants from the Ministry of Education, Ministry of Health and State Statistics Bureau and disseminated to these agencies. As explained above in subsection I.E, preliminary results will also be disseminated in seminars at the World Bank and in China.

IV. Resource Requirements

The resource requirements can be divided into two main types, the time of World Bank staff working on the project and the expenses for data collection and analysis. Two Bank staff will manage this project, Li Guo and Alan Piazza (both of EASRD). The weeks required for the life of the project (from 2002 to December 2005) will be 7 for Li Guo and 7 for Alan Piazza.

The money costs for data collection and analysis will be $406,000. Of this amount, $35,000 has already been obtained from the Spencer Foundation and $100,000 has been provided by the U.S. National Institutes of Health (Fogarty International Center), so the request to the World Bank is $271,000. The overall data collection and analysis budget is summarized as follows:

Total Budget

Other Sources*

World Bank

U.S. Consultant Costs (travel, summer salary, research assistance)




Oversee survey design, instrument development, presurvey of Educational and General Socioeconomic Survey in 2002, monitor survey work, oversee data entry and data analysis (NWNU)




Education and General Socioeconomic Survey (Gansu SSB), 2003 (short pretest and main survey), includes achievement and ability tests, child, mother, household, teacher, and principal surveys.




Physical Health Surveys (CAPM):

2002 (pretest), 2003 (main), 2004 (follow-up), 2005 (follow-up)




Nutrition Intervention (2 years) and Eyeglass Intervention (one-time) (CAPM)




Child Cognitive Ability Test development and analysis (CAS)




Youth Literacy Test development, pretest, and analysis (CNIER)








Directory: sites
sites -> The United States and Post-Castro Cuba
sites -> Fighting Fit: Exploring Military Medicine (1850-1950)
sites -> 9. 5 Political Powers and Achievements Tom Burns- beacon High School
sites -> Indiana Academic Standards Resource Guide World History and Civilization Standards Approved March 2014
sites -> Penn State Harrisburg American Studies/Women Studies 104: Women and the American Experience Spring 2015 Instructor: Kathryn Holmes
sites -> Penn State Harrisburg am st/wmnst 104: Women and the American Experience Spring 2015 Instructor: Kathryn Holmes
sites -> Abolition and Women’s Rights Chap. 14 Se
sites -> In the years between the Seneca Falls Convention and the Civil War, powerful links existed between antislavery and women’s rights advocates. Virtually all women’s rights advocates supported abolition

Share with your friends:
  1   2   3   4   5   6   7

The database is protected by copyright © 2020
send message

    Main page