母親健康知識、產前檢查與健康生產函數

行政院國家科學委員會專題研究計畫 成果報告

母親健康知識、產前檢查與健康生產函數

中 華 民 國 93 年 12 月 20 日

Parental Education and Child Health: Evidence from a Natural Experiment

in Taiwan

Shin-Yi Chou Lehigh University and NBER

Jin-Tan Liu

National Taiwan University and NBER Michael Grossman

City University of New York Graduate Center and NBER Theodore Joyce

Baruch College and NBER

April 2004

(Preliminary: Do not cite)

Abstract:

Many studies have documented that parental education, especially maternal education, has a significant positive impact on child health. This study exploits a natural experiment to estimate the causal impact of education in Taiwan. In 1968, the Taiwan government extended compulsory education from six to nine years, which required all school-age children (between six and fifteen) to attend elementary school for six years and junior high school for three years. To accommodate the expected increase in enrollment in junior high schools, the government opened 140 new junior high schools, a seventy-percent increase, in 1968. This education reform created the largest expansion in junior high school constructions and student enrollment in Taiwan. Our natural experiment exploits variations across cohorts in exposure to compulsory education reform and across regions in newly established school density. We estimate the impact of mother’s

education on child health by using cohort and newly established school density interactions as instruments for parents’ education. Our main data are annual birth and death certificates from 1978 to 1999. Our 2SLS estimates suggest that mother’s schooling has larger effects on child health outcomes than father’s schooling. Parental schooling reduces the probability of low birthweight, very low birthweight and prematurity, but has no significant impact in lowering neonatal, infant and postneonatal mortality.

I. Introduction

Improvements in child well-being are widely accepted public policy goals in developing and developed countries. Not only are these improvements viewed as desirable in their own right, but there is mounting evidence that schooling and health--two key components of well-being and of the stock of human capital--are crucial

determinants of economic growth (Topel 1999; Bloom, Canning, and Sevilla 2001, 2002; Krueger and Lindahl 2001). Many studies point to the importance of family background in general and mother’s schooling in particular in child health (Haveman and Wolfe 1995; Grossman and Kaestner 1997 and contained in the Appendix of this application; Behrman 1999). This finding is related to a larger literature that shows that an individual’s own schooling is positively related to his or her own health (Grossman 1972a, 1972b, 1975; Grossman and Kaestner 1997; Grossman 2000 and contained in the Appendix of this application). The positive correlation between mother’s schooling and child health in numerous studies was one factor behind the World Bank’s campaign in the 1990s to encourage increases in maternal education in developing countries (World Bank 1993). In a 2002 issue of Health Affairs devoted primarily to the nonmedical determinants of health, Deaton (2002) argues that policies to increase education in the U.S. and to

increase income in developing countries are very likely to have larger payoffs in terms of health than those that focus on health care, even if inequalities in health rise. Since more education typically leads to higher income, policies to increase the former appear to have large returns for more than one generation throughout the world.

Efforts to improve the health of an individual by increasing the amount of formal schooling that he or she acquires or that try to improve child well-being by raising

maternal schooling assume that the schooling effects reported in the literature are causal. A number of investigators have argued, however, that omitted “third variables” may cause schooling and health or well-being to vary in the same direction. For example, Fuchs (1982) argues that persons who are more future oriented (who have a high degree of time preference for the future or discount it at a modest rate) attend school for longer periods of time and make larger investments in their own health and in the well-being of their children. Thus, the effects of schooling on these outcomes are biased if one fails to control for time preference. The time preference hypothesis is analogous to the hypothesis that the positive effect of schooling on earnings, explored in detail by Mincer (1974) and in hundreds of studies since his seminal work (see Card 1999, 2001 for reviews of these studies), is biased upward by the omission of ability. Behrman and Rosenzweig (2002) present an argument that is even more closely related to ability bias in the earnings-schooling literature. In their model parents with favorable heritable endowments obtain more schooling for themselves, are more likely to marry each other, and raise children with higher levels of well-being. In turn these endowments reflect ability in the market to convert hours of work into earnings and childrearing talents in the nonmarket or household sector.

Governments can employ a variety of policies to raise the educational levels of their citizens. These include compulsory schooling laws, new school construction, and targeted subsidies to parents and students. If proponents of the third variable hypothesis are correct, evaluations of these policies should not be based on studies that relate adult health or child well-being to actual measures of schooling because these measures may be correlated with unmeasured determinants of the outcomes at issue. In this paper we

propose to use techniques that correct for third-variable bias to evaluate the effects of a policy initiative that radically altered the school system in Taiwan and led to a dramatic increase in the amount of formal schooling acquired by the citizens of that country during a period of very rapid economic growth.

II. Background

A. 1968 Education Reform in Taiwan

In 1968, the Taiwan government extended compulsory education from six to nine years, which required all school-age children (between 6 and 15) to attend elementary school for six years and junior high school for three years. To accommodate the expected increase in enrollment in junior high schools, the government opened 140 new junior high schools, a seventy-percent increase, at the beginning of the academic year 1968-69. This education reform created the largest expansion in junior high school construction and student enrollment in Taiwan’s history (Chang 1991; Clark and Hsieh 2000. Unless otherwise noted the material in this section is drawn from these sources, especially from Clark and Hsieh. Note, however, that we have revised some of their data.).

Primary school education in Taiwan was nearly universal by the mid-1960s, but approximately one-half of primary school graduates did not obtain additional education because enrollment in junior high school was restricted by a competitive national examination and by the limited number of junior high schools, especially in rural areas. The 1968 reform abolished the junior high school entrance examination and made it possible for all primary school graduates to continue their education. Children who had previously ended their education after primary school also were allowed to continue their education as long as they were under the age of 15 in 1968

but were unlikely to do so.

The large number of new junior high schools that opened in 1968 increased the number of these schools from 0.8 schools per thousand primary school graduates in the academic year 1967-1968 to 1.3 schools per thousand graduates in the

academic year 1968-1969. The immediate impact was to increase the percentage of primary school graduates who entered junior high school from 56 percent in 1967 to 77 percent in 1968.

Chang (1991) estimates that the 1968 compulsory schooling law increased the mean number of years of formal schooling completed by members of the labor force by 1.3 years by 1978 and by 3.4 years by 1988. Concurrently, there was a two-fold increase in real per capita gross domestic product in U.S. dollars from 1968 to 1978 and a four-fold increase from 1968 to 1988 (Heston, Summers, and Aten 2002). Hence, growth in the two decades following the nine-year compulsory schooling law was much more rapid than the two-fold expansion in real per capita gross domestic product in the two decades prior to its enactment.

A notable aspect of the school construction program was that its intensity varied across regions of Taiwan. Data shows the number of new junior high schools that opened in 1968 per thousand children between the ages of 12 and 14 in 1967 in each of the 21 cities or counties of Taiwan.1 Program intensity varied from 0.02 in Kaohsiung City to 0.76 in Penghu County. Hence, the nine-year compulsory schooling legislation provides a “natural experiment” to evaluate the impacts of parents’ schooling on the health of their

1

In Taiwan large cities are separate local entities. Hence Taipei County pertains to the region outside of Taipei City. Unlike Clark and Hsieh (2000), we present separate program intensity data for Taipei City, Kaohsiung City, Taichung City, and Tainan City. This is appropriate

children. In particular, those over the age of 11 in 1968 were unlikely to be affected by school reform and constitute a control group. On the other hand, those 11 years of age and under in 1968 were very likely to have been affected by school reform and constitute a treatment group. Moreover, the effects of school reform on the number of years of formal schooling completed in the treatment group should be larger the larger is the program intensity measure in city or county of birth. Clark and Hsieh (2000) present strong evidence in support of this proposition for men, and we present similar evidence for women in Section V. Thus, we employ the products of cohort indicators and the program intensity measure as instruments for schooling. Greater intensity among younger cohorts should lead to more schooling but should be uncorrelated with unmeasured determinants of the well-being of the offspring of these cohorts. Unlike Clark and Hsieh (2000) who are forced to predict schooling based on county of current residence of males between the ages of 30 and 50, we have information on county of birth. Based on our computations from the Taiwan Panel Survey of Family Dynamics, less than 10 percent of the population attended junior high school in a county that differed from their county of birth in the period after 1968.

Finally, we present some evidence that underscores the validity of our instrument. Taiwanese authorities planned to allocate more new junior high schools in regions where initial enrollment in junior high school was low (Clark and Hsieh 2000). This suggests that unmeasured determinants of schooling might be correlated with program intensity.

B. Data and Sample

Our data collection consists of all birth certificates and infant and child death certificates for the years 1978 through 1999. There were more than 300,000 births a year

in Taiwan during this period. Birth and death certificates will be linked through national identification numbers received by each person born in Taiwan. We consider the

following outcomes from these data: the probabilities of low- (less than 2,500 grams) and very low-weight (less than 1,500 grams) births; the probabilities of neonatal, postneonatal, and infant, and preterm births.

Low and very-low birthweights have extremely strong associations with infant morbidity and mortality. In a proximate sense, birthweight and low birthweight are caused by prematurity and slow growth in utero. Premature births have gestational ages (the difference between the date of birth and the date of the mother’s last menstrual cycle) of less than 37 weeks. The general consensus in the clinical literature is that relatively little is known about the causes of preterm delivery (for example, Hack and Merkatz 1995) and that most interventions designed to prevent these deliveries are not successful (for example, Goldenberg and Rouse 1998). On the other hand, fetal growth offers more scope for intervention since it is known to be linked to maternal smoking and maternal weight gain (see Joyce 1999 for a review of this evidence). Neonatal deaths pertain to deaths within the first month of life, while postneonatal deaths pertain to deaths between the ages of one month and one year. Infant deaths are the sum of those occurring in the neonatal and postneonatal periods. We distinguish between neonatal and postneonatal mortality because their causes are very different. Most neonatal deaths are caused by congenital anomalies, prematurity, and complications of delivery, while most

postneonatal deaths are caused by infectious diseases and accidents. Infants who die within the first month of life will be excluded from estimation for the probability of postneonatal death.

In addition to birthweight and gestational age, birth certificates contain the following information that is relevant for our research: gender; parity; mother’s city or county of birth; father’s city or county of birth; mother’s marital status; mother’s age; mother’s schooling; father’s age; and father’s schooling.

Throughout our study, we consider the women (or men) who were between the ages of less than 1 and 20 in 1968 and between the ages of 22 and 45 when they (or their wives) gave birth in the period from 1978 through 1999. We will use all births for estimation.

C. Basic Approach

As indicated in the previous section, we propose to use aspects of the nine-year compulsory schooling law enacted in Taiwan in 1968 as an instrument for schooling. Consider students from families who would only send them to school for six years in the absence of the law. Enactment of the legislation essentially reduces the price of three additional years of schooling for these students because it forces their parents “to become more ‘generous’” (Becker 1993, p. 140). Of course the price may not fall to zero because the law may not be fully enforced. But the presumption is that the law should encourage some students who otherwise would not have attended junior high school to do so. The opening of many new junior high schools in Taiwan in 1968 reinforces this effect. Following Duflo (2001), Clark and Hsieh (2000), and Breierova and Duflo (2002), we will incorporate variations in the intensity of the school program in Taiwan across

counties, measured by the county-specific number of new junior high schools that opened in 1968 per thousand children between the ages of 12 and 14 in 1967, in the instrument for schooling. Note that the legislation also may have had effects on the probabilities of

completing schooling levels beyond junior high school. Thus, the number of years of formal schooling completed is the most comprehensive measure of its impact.

To construct our instrument, we first form control and treatment groups using women or men between the ages of less than 1 and 20 in 1968. Those between the ages of 15 and 20 were very unlikely to have been affected by school reform and constitute the control group. On the other hand, those under the age of 12 were very likely to have been affected by school reform and constitute the treatment group. Children between the ages of 12 and 14 in 1968 who had previously ended their education after primary school were allowed to return to school but were unlikely to do so. Hence, we also distinguish

between this group and those under the age of 12 in some analyses.

The following regression model incorporates the notion that the impact of reform on the treatment group should be larger the larger are the number of new junior high schools that opened in the city or county of residence at the beginning of the 1968 school year:

∑∑

∑

∑

∑

Here e Sijt is the number of years of formal schooling completed by mother (or father) i

born in city/county j with her/his child born in year t. We also include cohort dummies (Ci), where the number of cohort dummies depends on the definitions of treatment and

control groups described below. Region (city or county) of birth dummies (Rj) are

included to capture the regional fixed effects. To control for trend effects, we include dichotomous indicators (Yt) for each year from 1978 to 1999, where 1978 is the omitted

year. The regression also includes interactions between region of birth dummies and year dummies to capture different trend effects in different regions. Finally, Ti is a dummy

indicating whether mother i belongs to the treatment group as described in more details below, and Pj denotes the program intensity in her region of birth. The coefficient β

estimates the impact of an additional junior high school (per thousand children aged 12-14) on mother’s education in the treatment group. We estimate the equation separately for mothers and fathers.

To evaluate our identification strategy, we form multiple treatment and control groups based on differential impacts of the policy on different cohorts. The percentage of primary school graduates who entered junior high school rose from 56% in the academic year 1967-68 to 77% in academic year 1968-69, to 87% in academic 1975-76, and to 99% in 1978-79. Thus, the law was not fully effective until the youngest members of the treatment group were about to enter junior high school. Since the 1968 legislation

undoubtedly had important lagged effects, it is plausible to expect that the reform had the largest effect on the youngest members of the treatment group. We first estimate models that are limited to 0-5 year olds in 1968 (the treatment group) and 15-20 year olds in 1968 (the control group) and alternatively 6-11 year olds and 15-20 year olds. Our identification strategy will be validated if the increase in schooling is larger among 0-5 year olds than among 6-11 year olds.

We also form our control group using women (or men) aged 12-14 in 1968. As mentioned above, this group was allowed to return for free junior high school education, but was less likely to do so since they may have begun to work. We compare this control group to a treatment group which is restricted to those between the ages of 9-11 in 1968. This comparison (9-11 versus 12-14) yields similar number of births in each group and also minimizes the number of years that must elapse before a treatment group woman

gives birth at the same age as a control group woman. Finally, we compare those aged 0-11 in 1968 (treatment group) with those aged 12-20 and 15-20 in 1968 (control groups).

We further add the variables of the interactions between the treatment dummy and program intensity using all births to these women (or men) who were between the ages of 22 and 45 when they (or their wives) gave birth in the period from 1978 through 1999. Program effects for mother’s and father’s educational attainment are discussed separately. In the estimation, the control group is the cohort aged 15-20 in 1968 and the treatment group is a cohort aged 0-5 in 1968. The estimate suggests that an additional junior high school (per thousand children aged 12-14) increases mother’s education by 1.03 years. In other specification, the estimate for the treatment group of mothers aged 6-11 in 1968 of 0.45 years is smaller. These results strengthen our identification strategy, since the education reform not only has a positive impact on the educational attainment of the treatment groups, but also has a larger impact for the younger women. The program effect is not statistically significant when the treatment group consists of the cohort aged 9-11 and the control group consists of the cohort aged 12-14. This result implies that the 12-14 year olds may not be a pure control group because some individuals in this group might go back to junior high schools after the reform. A comparison of the estimates in columns 4 and 5 yields a similar implication. The impact of the reform on the treatment group is smaller when the control group contains women aged 12-14 (0.332 compared with 0.461).

The estimates in male sample, are larger and more significant in a statistical sense than the corresponding estimates in female sample. This suggests that the education reform has a bigger impact on father’s educational attainment than on mother’s

educational attainment. For example, the estimation indicates that an additional junior high school per thousand children aged 12-14 increases the number of years of schooling completed by fathers aged 0-11 by 0.78 years compared to 0.46 years for mothers. In addition, the results suggest that the education reform had a larger impact on the education of younger fathers (i.e. 1.09 years for ages 0-5 and 0.77 years for ages 6-11) which is consistent with our expectation.

D. Full Specification

We can generalize eq. (1) to allow for a full set of program intensity and cohort interactions as follows:

∑ ∑

∑

∑

∑

∑

We estimate the above equation separately for mothers and fathers. Parental years of schooling is regressed on 20 cohort dummies, 20 region of birth dummies and interactions between cohort dummies and program intensity for mothers (or fathers) between the ages of less than 1 and 20 in 1968. Mothers (or fathers) aged 20 in 1968 form the control group, and this dummy is omitted from the regression. Each coefficient of βk can be interpreted as an estimate of the impact of the education reform on a given

cohort k. We also include 21 years of child birth year dummies and 420 interactions between region of birth dummies and year of birth dummies.

If there are time-varying and region-specific unobservables that are correlated with the education reform, the estimates of coefficients βk are biased. For example, if the

allocation of new junior high schools is negatively related to the initial enrollment rate or positively related to skilled labor demand, the estimates βk will underestimate the effect

shown in Figures 4 and 5 that the program intensity and junior high school enrollment rate and the percentage of agricultural workers are not correlated. But we still want to control for these omitted effects by including interactions between cohort dummies (Ck, k

= 0, … 19) and the initial enrollment rate in junior high school in 1966 and interactions between cohort dummies and the percentage of agricultural workers in 1967 in our estimation.

We include the 20 interactions between age in 1968 and program intensity. Controlling for the initial enrollment rate and the percentage of agricultural workers does not affect the estimates of the coefficients significantly, but it improves the efficiency of these estimates. Since the cohort aged 15 and older in 1968 did not benefit from the reform, the cohort aged between 12 and 14 was allowed to return to school but may not do so, and the cohort aged 11 and under was completely exposed to the new policy with strong lagged effect as discussed above, we could expect the pattern of the coefficients βk

to be 0 for k≥15, to be significantly positive for k = 0 to 11 and decreasing from k = 0 to

11, and to be ambiguous for k = 12 to 14.

It is obvious that the coefficients βk start to decrease sharply when k=13, and

fluctuates near zero for k=14 to 19. The coefficients βk are all positive for k=0 to 11, and

decrease from 0 to 11. The pattern is consistent with our expectation that the reform had no impact on the cohort not exposed to it, and had the largest effect on the youngest cohort. These results further validate our identification strategy.

In the specification that controls for the enrollment rate and the share of the agricultural worker, the average program coefficient in the regression for mother’s schooling is 1.33 between the ages of 0 and 5, 1.01 between the ages of 6 and 11, 0.71

between the ages of 12 and 14, and 0.31 between the ages of 15 and 19. The

corresponding coefficients in the regressions for father’s schooling is 1.07 between the ages of 0 and 5, 0.95 between the ages of 6 and 11, 0.68 between the ages of 12 and 14, and 0.12 between the ages of 15 and 19. Those estimates confirm the patterns that the magnitude of the coefficients decrease with age in 1968. The sets of coefficients for the 0-5 and 6-11 year olds are significant for both the sample of mothers and fathers. For mothers, the F-ratios are 8.06 and 5.17, respectively. For fathers, the F-ratios are 7.86 and 6.80, respectively. The same tests applied to the 12-14 year olds and 15-20 year olds yield insignificant F-ratios for both mothers and fathers.

E. Restricted Estimation

Since the education reform had a smaller impact on individuals aged 12 and older in 1968, we impose the restriction that βk are equal to zero for and estimate the

following equation separately for mothers and fathers:

15 ≥ k

∑∑

∑

∑

∑

∑

The reference group comprises of mothers (or fathers) aged 15 to 20 in 1968. In general, the coefficients decrease with age in 1968. The estimated coefficients are all statistically significant, except for some coefficients reported in column (1). The F-statistics

presented at the bottom of the table test the hypothesis that the coefficients of the interaction terms are jointly zero. The F-ratios are 15.73 and 16.96 for mother’s and father’s samples, respectively, when the enrollment rate and the percentage of

agricultural share are employed as regressors. These numbers are larger than the critical values proposed by Bound, Jaeger and Baker (1995), Staiger and Stock (1997), and Stock and Yogo (2002) required to avoid biases in TSLS coefficients due to weak instruments.

For mothers, the estimates in column (2) suggest that on average the cohorts aged 0 and 5 and 6 to 11 in 1968 received 1.00 and 0.72 additional years of education for every junior high school constructed per 1000 children between the ages of 12 and 14,

respectively. For fathers, the estimates in column (4) suggest that on average the cohorts aged between 0 and 5 and 6 to 11 in 1968 received 0.84 and 0.77 additional years of education for every junior high school constructed per 1000 children between the ages of 12 and 14, respectively.

III. Effect of Parental Education on Child Health Outcomes

We have shown in the previous section that the education reform increased the years of formal schooling completed by parents. The next question is whether the higher education attainment of parents leads to better health outcomes of their children.

Consider the following equation which relates a child health outcome to parents’ schooling and other observable characteristics:

, 19 1 99 79 4 99 79 3 21 2 2 19 1 1

∑

∑

∑ ∑

∑

where Hijt represents child health outcome and the coefficient ω will measure the impact

of parents’ schooling on the outcome.

If ηijt in equation (4) and εijt in equation (1) or (2) are correlated, the application

of ordinary least squares (OLS) to equation (4) will produce inconsistent coefficient estimates. Consistent estimates of equation (4) and in particular of the causal effect of S on H can be obtained by the method of instrumental variables (IV). Estimates of equation (1) or (2) in the previous section provide the first stage of the IV estimation. The

dummies and program intensity--which represent the price of schooling--serve as the instrument or instruments for schooling in the IV procedure.

Under OLS estimation, regardless of the definition of treatment and control groups, higher parental educational attainments significantly reduce neonatal mortality, infant mortality, postneonatal mortality, the probability of low birth weight, very low birth weight, and the risk of prematurity. For example, the estimates suggest that an additional year of mother’s education reduces neonatal mortality, infant mortality and postneonatal mortality by 0.01 percentage points (0.1 deaths per thousand live births), 0.03 percentage points, and 0.02 percentage points for mothers aged 0 to 11 in 1968. It also reduces the probability of low birthweight, very low birthweight and prematurity by 0.18 percentage points, 0.10 percentage points and 0.09 percentage points, respectively. Under the 2SLS estimation, mother’s years of schooling shows no significant impacts on neonatal mortality, infant mortality, and postneonatal mortality. Mother’s educational attainment significantly reduces the probabilities of low birthweight, very low

birthweight, and prematurity. For mothers aged 0 to 11 in 1968, an additional year of mother’s schooling reduces the probability of low birthweight, very low birthweight, and prematurity by 1.66 percentage points, 1.13 percentage points, and 1.65 percentage points respectively. The 2SLS estimates are much bigger than the r the OLS estimates.

Under the OLS estimation, father’s years of schooling has similar effects on child health as mother’s years of schooling. However, under TSLS estimation, father’s years of schooling has smaller effects than mother’s years of schooling in reducing the

probability of low birthweight, very low birthweight, and prematurity. An additional year of father’s schooling reduces the probabilities of low birthweight, very low birthweight

and prematurity by 1.28 percentage points, 0.88 percentage points,, and 0.91percentage points. For neonatal, infant and postneonatal mortalities, father’s and mother’s years of schooling have similar impacts. Mother’s schooling may have a larger impact in the TSLS estimation than father’s schooling because her schooling has a greater impact on efficiency in the production of healthy fetuses and decisions with regard to medical and nonmedical prenatal inputs.

We further use the interactions between cohort dummies and program intensity as the IV. Similar to the previous findings, mother’s and father’s years of schooling have significant impacts on child health when these impacts are estimated by OLS. However, the TSLS estimates show that mother’s years of schooling has no impact on mortality but father’s years of schooling significantly reduces infant and postneonatal mortality. Both mother’s and father’s years of schooling significantly reduce the probabilities of low birthweight, very low birthweight, and prematurity. The mother’s schooling coefficients in these outcomes are larger in absolute value than the father’s schooling coefficients. These findings are not sensitive to the inclusion of initial enrollment and the percentage of agricultural employment

Reference

Adams, S. J. “Educational Attainment and Health: Evidence from a Sample of Older Adults.” Education Economics 10: 97-109, 97-109.

Alderman H., J. R. Behrman, V. Lavy, and R. Menon. “Child Health and School

Enrollment: A Longitudinal Analysis.” Journal of Human Resources 36: 185-205, 2001. Alexander, G. R., M. D. Kogan, and J. H. Himes. “1994-1996 U.S. Singleton Birth Weight Percentiles for Gestational Age by Race, Hispanic Origin, and Gender.” Maternal and Child Health Journal 3: 225-231, 1999.

Arendt, J. N. “Education Effects on Health: A Panel Data Analysis Using School Reform for Identification.” Ph.D. Dissertation, University of Copenhagen, 2002.

Arkes, J. “Does Schooling Improve Adult Health?” Working Paper, RAND Corporation, 2001.

Becker, G. S. Human Capital Third Edition. Chicago: University of Chicago Press, 1993 Becker, G. S. “A Theory of the Allocation of Time.” Economic Journal 75: 493-517, 1965. Becker, G. S. A Treatise on the Family. Cambridge, Massachusetts: Harvard University Press, 1981.

Becker, G. S., and H. G. Lewis. “On the Interaction between the Quantity and Quality of Children.” Journal of Political Economy 81: S279-S288, 1973.

Becker, G. S., and C. B. Mulligan. “The Endogenous Determination of Time Preference.” Quarterly Journal of Economics 112: 729-758, 1997.

Behrman, J. R. “Labor Markets in Developing Countries.” In Handbook of Labor Economics Vol. 3, edited by O. Ashenfelter and D. Card. Amsterdam: North-Holland, 2859-2939, 1999.

Behrman, J. R., and M. R. Rosenzweig. “Does Increasing Women’s Schooling Raise the Schooling of the Next Generation?” American Economic Review 92: 323-334, 2002. Berger, M. C., and J. P. Leigh. “Schooling, Self-Selection, and Health.” Journal of Human Resources 24: 433-455, 1989.

Bloom, D. E., D. Canning, and J. Sevilla. “The Effect of Health on Economic Growth: Theory and Evidence.” National Bureau of Economic Research Working Paper 8587, 2001.

Bloom, D. E., D. Canning, and J. Sevilla. “Health, Worker Productivity, and Economic Growth.” Working Paper, School of Public Health, Harvard University, 2002.

Bound, J., and G. Solon. “Double Trouble: On the Value of Twins-Based Estimation of the Return to Schooling.” Economics of Education Review 18, 169-182, 1999.

Breierova, L., and E. Duflo. “The Impact of Education on Fertility and Child Mortality: Do Fathers Really Matter Less Than Mothers?” Working Paper, Department of Economics, Massachusetts Institute of Technology, 2002.

Card, D. “The Causal Effect of Education on Earnings.” In Handbook of Labor Economics Vol. 3, edited by O. Ashenfelter and D. Card. Amsterdam: North-Holland, 1801-1863, 1999.

Card, D. “Estimating the Return to Schooling: Progress on Some Persistent Econometric Problems.” Econometrica 69, 127-1160, 2001.

Chaikind, S. and H. Corman. “The Impact of Low Birthweight on Special Education Costs.” Journal of Health Economics 10: 291-311, 1991.

Chang, C. C. J. “The Nine-Year Compulsory Education Policy and the Development of Human Resources in Taiwan (1950-1990). Ph.D. Dissertation, University of Maryland, 1991.

Clark, D. E. , and C.-T. Hsieh. “Schooling and Labor Market Impact of the 1968 Nine-Year Education Program in Taiwan.” Working Paper, Department of Economics, Princeton University, 2000.

Currie, J. “Child Health in Developed Countries.” In Handbook of Health Economics Vol. 1B, edited by A. J. Culyer and J. P. Newhouse. Amsterdam: North-Holland, 1054-1090, 2000.

Currie, J. and E. Moretti. “Mother’s Education and the Intergenerational Transmission of Human Capital: Evidence from College Openings and Longitudinal Data.” National Bureau of Economic Research Working Paper 9360, 1992.

Cutler, D. M. and E. Meara. “The Technology of Birth: Is It Worth It?” In Frontiers in Health Policy Research Vol. 3, edited by A. M. Garber. Cambridge, Massachusetts: MIT Press, 33-67, 2000.

Deaton, A. “Policy Implications of the Gradient of Health and Wealth.” Health Affairs 21: 13-30, 2002.

de Walque, Damien. “How Does Education Affect Health Decisions? The Cases of Smoking and HIV/AIDS.” Ph.D. Dissertation, University of Chicago, in progress. Duflo, E. “Schooling and Labor Market Consequences of School Construction in

Indonesia: Evidence from an Unusual Policy Experiment.” American Economic Review 91: 795-813, 2001.

Edwards, L. N., and M. Grossman. “Adolescent Health, Family Background, and Preventive Medical Care.” In Research in Human Capital and Development Vol. 3. edited by I. Sirageldin, D. Salkever, and A. Sorkin. Greenwich, Connecticut: JAI Press, Inc., 77-109, 1983.

Edwards, L. N., and M. Grossman. “The Relationship between Children's Health and Intellectual Development.” In Health: What is it Worth, edited by S. J. Mushkin and D W. Dunlop. Elmsford, New York: Pergamon Press, 273-314, 1979.

Fuchs, V. R. “Time Preference and Health: An Exploratory Study.” In Economic Aspects of Health, edited by V. R. Fuchs. Chicago: University of Chicago Press, 93-120, 1982. Glewwe, P. “Why Does Mother’s Schooling Raise Child Health in Developing Countries?” Journal of Human Resources 34: 124-159, 1999.

Journal of Medicine 339: 313-320, 1998.

Griliches, Z. “Sibling Models and Data in Economics: Beginnings of a Survey.” Journal of Political Economy 87: S37-S64, 1979.

Grossman, M. “On the Concept of Health Capital and the Demand for Health.” Journal of Political Economy 80: 223-255, 1972a.

Grossman, M. “The Correlation Between Health and Schooling.” In Household Production and Consumption, edited by N. E. Terleckyj. New York: Columbia University Press for the National Bureau of Economic Research, 147-211, 1975.

Grossman, M. The Demand for Health: A Theoretical and Empirical Investigation. New York: Columbia University Press for the National Bureau of Economic Research, 1972b. Grossman, M. “The Human Capital Model.” In Handbook of Health Economics Vol. 1A, edited by A. J. Culyer and J. P. Newhouse. Amsterdam: North-Holland, 347-408, 2000. Grossman, M., and T. J. Joyce. “Unobservables, Pregnancy Resolutions, and Birth Weight Production Functions in New York City.” Journal of Political Economy 98: 983-1007, 1990.

Grossman, M., and R. Kaestner. “Effects of Education on Health.” In The Social Benefits of Education, edited by J. R. Behrman and N. Stacey. Ann Arbor, Michigan: University of Michigan Press, 69-123, 1997.

Hack, M., and I. Merkatz. “Preterm Delivery and Low Birth Weight -- A Dire Legacy.” New England Journal of Medicine 333: 1772-1773, 1995.

Haveman, R. H., and B. L. Wolfe. “The Determinants of Children’s Attainments: A Review of Methods and Findings.” Journal of Economic Literature 33: 1829-1878, 1995. Heston, A., R. Summers, and B. Aten. “Penn World Tables, Version 6.1.” Center for International Comparisons, University of Pennsylvania, 2002.

Joyce, T. “Impact of Augmented Prenatal Care on Birth Outcomes of Medicaid Recipients in New York City.” Journal of Health Economics 18: 31-67, 1999.

Krueger, A. B., and M. Lindahl. “Education for Growth: Why and for Whom?” Journal of Economic Literature 39: 1101-1136, 2001.

Leigh, J. P., and R. Dhir. “Schooling and Frailty among Seniors.” Economics of Education Review, 16: 45-57, 1997.

Lleras-Muney, A. “The Relationship between Education and Adult Mortality in the United States.” National Bureau of Economic Research Working Paper 8986, 2002. Mincer, J. Schooling, Experience, and Earnings. New York: Columbia University Press for the National Bureau of Economic Research, 1974.

Neumark, D. “Biases in Twin Estimates of the Return to Schooling. Economics of Education Review 18: 143-148, 1999.

Sander, W. “Schooling and Quitting Smoking.” Review of Economics and Statistics 77: 191-199, 1995a.

Sander, W. “Schooling and Smoking. Economics of Education Review 14: 23-33, 1995b.

Shakotko, R., L. Edwards, and M. Grossman. “An Exploration of the Dynamic

Relationship Between Health and Cognitive Development in Adolescence.” In Health, Economics, and Health Economics, edited by J. van der Gaag and M. Perlman.

Amsterdam: New York: North-Holland, 305-328, 1983.

Topel, R. “Labor Markets and Economic Growth.” In Handbook of Labor Economics Vol. 3, edited by O. Ashenfelter and D. Card. Amsterdam: North-Holland, 2943-2984, 1999.

Willis, R. J. “A New Approach to the Economic Theory of Fertility Behavior.” Journal of Political Economy 81: S14-S64, 1973.

World Bank. World Development Report 1993: Investing in Health. New York: Oxford University Press, 1993.