Education and Health Relationships

Introduction

Most people would not be surprised to learn that education is positively associated with health. This seems intuitive, and consistent with what is observed in society. However, many would be surprised by the strength and pervasiveness of the link between education and health across different contexts and different indicators of health. More educated people live longer than those who are less educated, and the importance of education as a determinant of mortality is only growing over time. Chronic diseases, such as asthma and diabetes, are more prevalent among lower educated groups compared to higher educated groups. Even among those with chronic disease, education is positively associated with timely disease diagnosis, effective self-management, and better disease outcomes. Education is positively correlated with healthy behaviors such as exercise and use of preventive care and it is negatively associated with virtually all the risky health behaviors such as poor eating habits, lack of exercise, problem drinking, illegal drug use, and smoking. Maternal education plays a similar role as a determinant of children’s health. Maternal education is positively associated with a broad range of children’s health and developmental outcomes, ranging from children’s preventive health care to mental health outcomes.

Some people argue that is it not education per se, but rather factors correlated with education, such as income, that lead to better health. It may be observed that educated people, for example, exercise more than the less educated. But this may be the case not because of education but rather because educated people earn higher incomes and can afford, say, gym memberships. To some extent, this is true – factors correlated with education, especially income and ability, do account for some portion of the association between education and health. But, in general, the strong and pervasive association between health and education persists and remains policy-significant in magnitude even when researchers take into account a broad range of other factors that are correlated with both education and health, such as income, family background, and demographic characteristics. Does this mean that education truly improves health, or are there factors that cannot be measured well that underlie this relationship? If education does indeed cause better health, what makes education so crucial to health? These questions have intrigued economists for the past four decades.

Education and Health Relationships Figure 1

The existence of a robust, positive association between education and health does not necessarily mean that more education causes better health. The reverse causal pathway is also plausible. Better health early in life may lead individuals to complete more schooling, because longer life expectancy increases the benefits of educational investments, and/or because better health improves school attendance and helps students to learn better. There is a growing body of evidence suggesting that early health – even health in utero – can have profound implications for future, adult health, and wellbeing. Thus, an observed association between education and health among adults may result not from education casually affecting health but rather from early health affecting both health and education in adulthood.

Education and Health Relationships Figure 2

Also possible is a noncausal explanation for the correlation between education and health. The correlation may come from unmeasured variables that are associated with both health and education, such as ability, genetics, or family socioeconomic status (SES). Some have suggested that individuals with strong preferences for present versus future outcomes – that is, individuals with high discount rates – will not make long-term investments in health or education. If this trait is hard to measure in data, it may appear that education is positively associated with health, but in reality individuals’ time preferences , which is unmeasured, determines both health and education. In this case, a strong, positive association between education and health may exist, but it does not reflect a causal relationship. It is also possible that more schooling causes individuals to be more future-oriented. In this case, education may affect health causally through its effect on the rate of time preferences .

Education and Health Relationships Figure 3

In recent literature, economists employ innovative econometric methods to determine whether the association between education and health is causal. Although most studies are based on data from the US and the UK, increasingly data from around the world are being used to examine this relationship. The economics literature on education and health is very large, and it is expanding rapidly. This short review does not cover all health economics literature in this area. Instead, the focus is on empirical research on education and health in developed countries published in the past 10 years in economics journals. The goal is to highlight some provocative papers, synthesize results, and draw conclusions from recent studies that have attempted to distinguish causal relationships from associations between health and education.

The Grossman Model

Most empirical literature on education and health is motivated by the Grossman Model (Grossman, 1972a,b; 2000). The Grossman Model is a model of the demand for the commodity ‘good health,’ which is treated as a durable good, or a type of capital. Health has both direct consumption value as well as investment value in this model. Health has consumption value because individuals derive utility from being in good health. Health has investment value because it determines the total amount of time that is available to work in the market and nonmarket sectors. Briefly, in the Grossman Model, individuals maximize a utility function which includes health and other commodities with respect to investments in health, given budget and time constraints. Optimal gross investment in health determines the optimal amount of health, because the depreciation rate (e.g., wear and tear on health capital) and initial health are given.

Grossman analyzes the effect of education on health in his pure investment model. In this version of the model, the consumption value of health is not considered. Education is viewed as the technology of the health production function. More education makes individuals better producers of health. In other words, an increase in education would allow an individual to obtain more health from a given set of inputs, decreasing the marginal cost of an investment in health. The decrease in the marginal cost of investment increases the returns on health capital, and the optimal level of health is higher than before. Thus, according to the Grossman Model, more education leads people to choose higher levels of health because education increases individuals’ productive efficiency in producing health (Grossman, 1972a,b; 2000).

The mechanism through which education increases productive efficiency is hard to pinpoint. One can argue, in fact, that it is more likely that education causes better health by improving individuals’ allocative efficiency in producing health (Grossman, 2008, 1972a). For example, more education may cause individuals to understand better how to combine inputs to produce health; thus, individuals may make more efficient choices about how much to exercise, what to eat, how to adhere to medical treatments, and what health behaviors to avoid. The distinction between the productive and allocative efficiency arguments can be important from an empirical perspective. If education increases health primarily through improvements in allocative efficiency, if one is estimating a health production function, then there should not be an association between education and health if all inputs are included in the model as well (Grossman, 2008). This is not the case if education improves productive efficiency, because more education leads to individuals directly obtaining more health from a given set of inputs.

Grossman (1972a) emphasizes that one’s stock of health is an endogenous choice variable. Current health depends on initial health, depreciation of the health stock in all previous periods, and gross investment (and thus inputs used to produce investments) in all previous periods (Grossman, 1972a). Therefore, when researchers estimate the effect of early health on subsequent education outcomes, it is important to include controls for factors that may affect education directly and also may affect early health through prior health investments, such as family background. However, it is possible that the controls included do not completely account for prior health investments, and that these factors remain in the error term of the equation. Thus, endogeneity resulting from omitted variable bias is a concern to researchers when estimating the effects of early health on later education outcomes.

When estimating effects of education on health, omitted variables bias is still a possibility, because unmeasured factors such as ability may exist that are correlated with both education and health. But in this case structural endogeneity is potentially a problem as well because in a full model of education and health, education and health may be determined simultaneously. Moreover, when estimating the effects of education on health, a reverse causal pathway, with current health affecting current education, is plausible. Thus, when estimating the effects of education on health, health is considered to be endogenous in a structural as well as in a statistical sense.

Econometric Methods Used To Test For Causal Effects

In recent literature, two main empirical approaches have been used to distinguish causal relationships between education and health from associations. The first approach is to rely on a natural experiment. Some examples of natural experiments that have been used to identify effects of early health on later outcomes are famines (Chen and Zhou, 2007), periods of religious fasting (Almond and Mazumder, 2011), outbreaks of illness (Bleakley, 2010; Almond, 2006), rainfall (Maccini and Yang, 2009), nuclear accidents (Almond et al., 2009), and crop infestation (Banerjee et al., 2010). These events are treated as exogenous shocks to early health. One drawback of examining the effects of these events is that the results sometimes may not be readily generalizable to other settings.

In studies of the effects of education on health, researchers have taken advantage of the natural experiments induced by variation in educational policies across time and place. Some examples of natural experiments that have been used to identify effects of education on health are variation in policies that affect school entry (Braakmann, 2011), access to secondary education (Arendt, 2005), and variation in county-level access to college (Currie and Moretti, 2003). Frequently, researchers have drawn on these natural experiments to implement instrumental variables (IV) methods (Eide and Showalter, 2011). The primary advantage of using IV methods in this context is that this approach addresses both the statistical and structural endogeneity. A drawback of this approach, however, is the possible low predictive power of the instruments and its associated problems (Staiger and Stock, 1997). Also, IV findings cannot be generalized to individuals whose educational decisions are not ‘at the margin’ or, in other words, individuals whose educational decisions are not influenced by the policy that is being used as an instrument.

The second approach used to test for causal relationships in this literature are sibling/twin fixed effects models. This method involves estimating the correlation between withintwin (or within-sibling) differences in birth outcomes and within-twin (or within-sibling) differences in later educational outcomes. This approach essentially ‘differences out’ family specific fixed characteristics that may confound an observed association between early health and later education outcomes. Sibling/twin fixed effects models address a specific form of statistical endogeneity – confounding by unmeasured, fixed family-specific characteristics.

There are some advantages in using twins rather than siblings to implement these models. In studies on the educational consequences of birth weight, within-sibling birth weight can vary because of differences in intrauterine growth retardation (IUGR) and/or differences in gestational length, whereas between twins, variation must come from a single source, IUGR (Oreopoulos et al., 2008; Almond et al., 2005). Also, sibling fixed effects models do not address time-varying family characteristics. Maternal health behaviors may vary by birth order, or SES could change between births. These changes may be unmeasured (Oreopoulos et al., 2008; Royer, 2009) and confound an observed relationship between early health and later education outcomes. This issue does not arise in the case of twins, who are born at the same time. Also, unobserved individual heterogeneity, such as genetic differences, may exist within siblings and within fraternal twins (Almond et al., 2005).

In these ways, sibling fixed effects models implemented using data on twins, particularly monozygotic twins, are subject to fewer biases. However, an important advantage of using siblings instead of twins is that the results are more easily generalized to the population. Moreover, even analyses based on within-twin differences can suffer from problems related to measurement error, unstable estimates (Royer, 2009), and selection problems caused by mortality at birth within-twin pairs (Black et al., 2007; Royer, 2009).

Effect Of Health On Education

Health At Birth

Malnutrition and poor health in utero or early in childhood is a predictor of later health outcomes, including infant mortality, height, cognitive function, chronic disease, and disability (Barker et al., 1989; Banerjee et al., 2010; Case and Paxson, 2009; Van Ewijk, 2011; Delaney et al., 2011; Chay and Greenstone, 2003). Economic conditions measured at birth have been found to be related to adult mortality (Van Den Berg et al., 2006). These findings, which demonstrate the importance of the early health environment for later health, imply that poor health environment early in life may affect economic outcomes as well. In estimating long-term effects of health at birth, the challenge is in determining whether poor early health is the cause of later problems, or whether it is instead a correlate of such problems (Oreopoulos et al., 2008; Black et al., 2007). There is a burgeoning health economics literature in this area, focusing on education as an outcome, with many innovative identification strategies being used. Numerous studies focus on estimating the long-term effects of birth weight, a single aspect of early health. However, increasingly other measures of early health are being considered. In fact, in many studies, researchers estimate reduced-form models in which the health environment early in life is linked directly to later educational outcomes. In these papers, the mechanism through which early health detracts from later education is not always well-understood.

In a landmark study, Almond et al. (2005) examined the long-term consequences of low birth weight (LBW) using data on twins born in the US between 1983 and 2000. They examined the correlation between within-twin differences in birth weight and within-twin differences in (1) hospital charges, (2) other measures of health at birth, and (3) infant mortality rates. The authors also estimate the effect of prenatal smoking on a variety of infant health outcomes using singleton births, controlling for sociodemographic variables available on birth certificates. In these analyses, they attribute the entire effect of smoking on infant health to the effect of smoking on birth weight, which is probably an overstatement. The authors cannot fully control for unobserved heterogeneity using this approach – but they can gauge whether the magnitudes of the effects generated using the sample of twins are reasonable.

The cross-sectional estimates suggest that a 1 standard deviation increase in birth weight leads to reduction in hospital costs, reduction in infant mortality, increase in Apgar score, and reduction of assisted ventilator use after birth of .51, .41, .51 and .25 standard deviations, respectively. Based on the twins analysis, however, these magnitudes fall to .08, .03, .03, and .01. The smoking analysis shows that smoking affects birth weight appreciably, but smoking is not related to most infant health outcomes – as a result, cost savings of smoking cessation during pregnancy are modest. Either the true effect of birth weight on infant health has been overstated in prior work; and/or each analysis isolates a different set of determinants of birth weight.

Using a similar approach to that of Almond et al. (2005), there have been several studies based on samples of twins which examine the effects of birth weight on long-term educational outcomes. All these studies support the idea that birth weight has long-term consequences for adult education and health outcomes. Black et al. (2007), for example, draw on administrative data on twins born between 1967 and 1981 in Norway and study the consequences of birth weight. They build on Almond et al. (2005) in that they are able to examine the effects of birth weight not just on short-run health outcomes (infant mortality and 5 min Apgar score) but also on long-run outcomes including adult height, intelligence quotient (IQ), employment, earnings, education, and birth weight of the first child. Like Almond et al. (2005), Black et al. (2007) found that within-twin differences in birth weight are associated with smaller effects on short-run outcomes compared to cross-sectional, ordinary least squares (OLS) estimates. However, Black et al. (2007) report that there are long-term effects of birth weight on adult height, body mass index, IQ, education, earnings, and birth weight of the first born child. For these outcomes, OLS and within-twin estimates are similar in magnitude.

Royer (2009) studies the effects of birth weight on educational attainment, later pregnancy complications, and birth weight among offspring using data on same-sex, female twins born in California between 1960 and 1982. Among these twins, long-term outcomes can be studied for those who survive to adulthood, remain in California, and give birth to infants between 1989 and 2002. Consistent with other research, Royer finds that cross-sectional estimates of the effect of birth weight on short-run health are overstated. The estimated within-twin effect of birth weight on 1-year mortality is similar to that of Almond et al. (2005) in magnitude. Royer finds small, long-term effects of birth weight on women’s educational attainment. It is interesting and unexpected that Royer finds that the positive effect of birth weight on education is largest for infants who are of normal birth weight (42500 g). Royer also finds that within-twin differences in birth weight are correlated with women’s later pregnancy complications and birth weight of their own children. Currie and Moretti (2003), also using data from California, report a similar finding. They find that birth weight differences within pairs of sisters are correlated with within-sister variation in subsequent birth of an LBW infant. This effect is stronger for women living in low SES neighborhoods.

Behrman and Rosenzweig (2004) also estimate twin fixed effects models to study the association between birth weight and adult outcomes, including educational attainment. They use a sample of monozygotic twins born in Minnesota between 1936 and 1955. The findings show that fetal growth (weight divided by length squared) is positively associated with both height and educational attainment in adulthood.

Other researchers have examined effects of health at birth using data that include siblings as well as twins. Oreopoulos et al. (2008), for example, test whether within sibling differences in health at birth are correlated with within-sibling differences in later outcomes. The sample includes more than 96% of all children born in Manitoba, Canada between 1978–82 and 1984–85. They examine the effects of infant health not just on infant mortality, but also on long-term educational and employment outcomes, including childhood mortality, language scores in grade 12, physician services utilization during adolescence, reaching grade 12 by age 17, and social assistance receipt. Notably, they use multiple measures of infant health including birth weight, Apgar score, and gestational length. The findings from this paper based on twins are consistent with those from Almond et al. (2005) – the effect of poor infant health on mortality rates diminishes when twin differences are examined. However, infant health – especially birth weight and Apgar score – are associated with educational attainment at age 17 and public assistance receipt, suggesting that there are long-run effects of infant health on human capital accumulation.

Johnson and Schoeni (2010) also find long-term effects of LBW and early economic disadvantage on educational attainment, labor market, and health outcomes measured in adulthood. They use data from the Panel Study of Income Dynamics (PSID) and sibling fixed effects models. Similarly, Fletcher (2011), using data from the National Longitudinal Study of Adolescent Health (Add Health), estimates the effects of LBW on education outcomes using siblings fixed effects models. He finds that LBW is associated with early grade repetition, special education placement, and diagnosis of learning disability. However, unlike Oreopoulos et al. (2008) and Johnson and Schoeni (2011) does not find effects of LBW on longer term educational outcomes such as educational attainment.

In addition to examining effects of health at birth, there are many papers examining the effects of prenatal shocks to health, including inter-uterine exposure to famines, religious fasting, illness, adverse economic conditions, and toxins. Chen and Zhou (2007), for example, test for causal effects between exposure to the 1959–61 famine in China and health and labor market outcomes in adulthood among those who survived. They find that children born in 1962 (who were in utero during the famine) became shorter adults than they would have been had they not been exposed to the famine. Among those exposed during early childhood, famine exposure is associated with reduced labor supply and earnings in adulthood. Almond et al. (2009) study effects of prenatal exposure to radiation stemming from the 1986 Chernobyl nuclear accident in the Ukraine. These authors study effects on academic outcomes among children in Sweden who were exposed 8–25 weeks post-conception to varying degrees of fallout from the accident. The findings show that low levels of prenatal exposure to radiation has no discernible effects on children’s health, but it is associated with worse academic performance in high school. The effects are stronger for children from more disadvantaged backgrounds.

Almond (2006) use US data to test for long-term effects of prenatal exposure to the 1918 influenza pandemic on economic outcomes including education. They find that such exposure is associated with about a 15% reduction in the likelihood of graduation from high school and a 5–9% fall in men’s wages, as well as with increases in physical disability and receipt of public assistance. Maccini and Yang (2009) estimate reduced-form models to examine the effect of rainfall around the time of birth on Indonesian adults’ socioeconomic and health outcomes. They find that rainfall in utero does not affect adult outcomes. However, rainfall in the first year of life is positively associated with health and educational attainment among women, presumably because higher rainfall increases agricultural yields and household resources. Almond and Mazumder (2011) study long-term effects of prenatal exposure to Ramadan, a period of religious fasting. Using data from Michigan, they find that prenatal exposure is associated with lower birth weight. Using data from Uganda and Iraq, these authors report that exposure to Ramadan in utero is associated with large increases in the likelihood of adult disability. Case and Paxson (2009) use data from the Health and Retirement Study and find region-level infant mortality and disease rates in the first 2 years of life are associated with cognitive function in old age (Case and Paxson, 2009). In sum, there is a convincing body of evidence that prenatal health conditions and health at birth have long-term effects on later educational attainment and other adult outcomes. In some cases, the causal mechanism appears to be adult health, but in other cases, mechanisms linking early health to later outcomes are not clear.

Health During Youth

There also is a small but growing literature on the effects of health during childhood on educational outcomes in developed countries. Case et al. (2005), for example, examine this relationship using the 1958 National Child Development Study. This survey includes data collected from birth until age 42 on all children born in the UK during the week of 3 March 1958. The results show that chronic health conditions in childhood, as well as LBW, are associated with reductions in educational attainment, employment, social status, and adult health. Although this study draws on unusually rich data which should minimize problems of unobserved heterogeneity, the methods do not directly address the problem of disentangling causality from correlation.

Some researchers, however, have used sibling fixed effects models to difference out family-specific factors that may drive both children’s health and educational outcomes. These studies generally support the idea that health during childhood affects educational attainment. Some studies have used self-rated overall health rankings to measure child health. Smith (2009), for example, estimate sibling fixed effects models using data from the PSID to examine the effect of child health on adult labor market outcomes. Child health is measured using a retrospective self-report of overall health before age 17. The sibling fixed effects model findings do not show a statistically significant relationship between health in childhood and educational attainment. However, there are positive effects of child health on family income, household wealth, individual earnings, and labor supply.

Chay et al. (2009) focus on how access to and quality of health care early in life affects later educational outcomes. They examine the effects of desegregation and forced integration of hospitals in the US during the 1960s and 1970s on racial disparities in test scores in the 1980s. They find that access to better health care in early childhood reduced African-American/ white disparities in achievement test scores later in life.

Other studies estimate effects of specific chronic health conditions during childhood on later educational and labor market outcomes. Fletcher et al. (2010), for example, use data from the National Longitudinal Study of Adolescent Health (Add Health) to examine the effect of childhood asthma on missed days from school and work, obesity, and adult health. They use sibling fixed effects models and find large, detrimental effects of childhood asthma on absenteeism. Rees and Sabia (2011), also using Add Health, find that migraine headaches detract from educational outcomes. Sabia (2007), using data from Add Health, finds a negative association between body weight and grades for white females, but not for other sociodemographic groups. Grossman and Kaestner (2009), however, using data from the NLSY79, do not find any statistically significant association between body weight and children’s achievement test scores.

There is also evidence that exposure to tropical disease in childhood affects later educational outcomes. Bleakley (2007) studies the effect of hookworm on long-term educational outcomes in the US, taking advantage of a natural experiment in which a public health campaign was instituted in the early 1900s to eradicate the disease. Bleakley finds that childhood hookworm has very large effects on adult wages, mostly through reducing the returns to schooling. In another paper, Bleakley (2010) finds that childhood malaria reduces income in adulthood. In this study, to identify effects of malaria on outcomes, he takes advantage of malaria eradication campaigns instituted in the US and in Latin America.

Results from several studies highlight the importance of mental health for educational and other human capital outcomes. Currie et al. (2010) draw on administrative data from Manitoba, Canada, and examine whether childhood health problems are associated with adult educational attainment, test scores, and social assistance receipt. The primary estimation strategy is sibling fixed effects models. The results show that childhood health problems, especially mental health problems, detract from adult educational attainment and other outcomes. These findings are consistent with those of Currie and Stabile (2006). They employ sibling fixed effects models and use national survey data from the US and Canada and find that hyperactivity symptoms during childhood are associated with worse educational outcomes, such as grade repetition and special education placement. Fletcher and Wolfe (2008) are able to replicate these findings of the effects of hyperactivity on short run educational outcomes using a different data source (Add Health). However, Fletcher and Wolfe find that hyperactivity does not affect longer term educational outcomes, such as educational attainment.

In addition to these studies that focus on hyperactivity, other economics studies show that depressive symptoms during youth are associated with lower grades and lower educational attainment (Eisenberg et al., 2009; Fletcher, 2010). In addition, a few new studies using data from the US show that having genetic markers for depression and attention deficit hyperactivity disorder are associated with adverse educational outcomes (Ding et al., 2009; Fletcher and Lehrer, 2009). However, Contoyannis and Dooley (2010), using data from the Ontario Child Health Study, examined the association between child health (measured by conduct or emotional disorder, and by chronic condition or functional limitation) on a range of educational attainment and labor market outcomes measured in adulthood. They find that child health is negatively associated with educational attainment and labor market outcomes, but these findings do not persist when sibling fixed effects are included in the models.

Effect Of Education On Health

Maternal Education And Child Health

Maternal education is a powerful correlate of children’s health outcomes, but whether this relationship is causal remains an open question. Several recent papers focus on testing whether a causal relationship exists between maternal education and child health. Currie and Moretti (2003) make important progress in this area by examining the effect of maternal education on infant health at birth using data from US individual birth certificates from 1970 to 1999. They hypothesize four potential causal pathways linking maternal education to infants’ health: (1) effects of maternal education on prenatal care; (2) effects of maternal education on spousal earnings; (3) effects of maternal education on health behaviors (prenatal smoking); and (4) effects of maternal education on fertility (quality/quantity tradeoff). They use an IV method with availability of colleges at the county level as an instrument for maternal education. Currie and Moretti find that higher maternal education improves children’s birth weight and gestational age at birth. This is a large effect – an additional year of college is estimated to reduce the incidence of LBW by 10%. Their results show that maternal education increases the probability of marriage, increases husband’s education, reduces parity, increases use of prenatal care, and reduces smoking. These pathways, therefore, may be mechanisms through which maternal education affects infants’ health.

McCrary and Royer (2011), however, use US birth certificate data and come to different conclusions. They test whether maternal education affects fertility and infant health (birth weight, prematurity, infant mortality) using large samples of birth records from Texas and California which include the exact date of birth. They rely on school entry cutoffs, which allow them to compare birth outcomes of women born just before and just after their states’ school entry cutoffs. Although women born just after the school entry date do complete less education than women born just before, their infants are as healthy as those of women born just before the school cutoff. These findings, then, suggest that for women whose educational decisions are affected by school cutoff policies, maternal education does not appear to play a causal role in infant health.

Carneiro et al. (2011) examine the effects of maternal education on children’s cognitive test scores, behavior problems, and the home environment using data from the National Longitudinal Survey of Youth 1979 (NLSY79). They instrument for maternal education using local labor market conditions, college tuition, and the existence of a 4-year college in the county where the mother lived at age 14. The findings show that maternal education is positively associated with test scores and negatively associated with behavioral problems among children.

Chou et al. (2010) estimate the effect of maternal and paternal education on LBW and infant mortality using birth certificate data on infants born in Taiwan between 1978 and 1999. They take advantage of a natural experiment related to educational attainment. In 1968, Taiwan extended compulsory schooling from 6 to 9 years and opened 150 new junior high schools. Before 1968, junior high enrollment was restricted by a difficult exam. The findings show that maternal education and paternal education both affect infant health, but maternal education appears to be more important.

Finally, Chen and Li (2009) use Chinese data to examine whether maternal education affects the health of adopted versus biological children. They find that maternal education is associated with better child health for both adopted and biological children. This finding does not definitely establish a causal relationship, but it is revealing that maternal education is strongly associated with child health, even when genetic explanations are eliminated.

Education And Health

There is a large literature on the effects of education on one’s own health. In this literature, economists have studied the effects of education on mortality, chronic health conditions, and a wide range of health behaviors. In an influential paper, Lleras-Muney (2005) uses a quasinatural experiment to determine whether the association between education and mortality represents a causal relationship. The natural experiment consists of states changing their compulsory schooling and child labor laws between 1915 and 1939, inducing some individuals to obtain more schooling than they would have otherwise. Data come from the US Censuses from 1960, 1970, and 1980. Her sample includes whites born in 48 states who were 14 years old between 1914 and 1939, with available data on education. She creates synthetic cohorts by aggregating Census data into groups by gender, cohort, and state-of-birth, calculates mortality rates for these groups, and examines direct effect of changes in compulsory schooling on mortality rates by comparing mortality rates of cohorts immediately before and after there was a change in legislation. This regression discontinuity approach offers only suggestive evidence of an effect of education on mortality. She then uses the compulsory education laws as instruments, and finds statistically significant negative effects of education on mortality. The effect is large in magnitude – a 10% increase in education lowers mortality by 11%.

Albouy and Lequien (2009) examine the effect of education on mortality in France and come to different conclusions. They rely on changes in compulsory schooling laws as a natural experiment and use regression discontinuity and IV methods, as was done by Lleras-Muney (2005). However, their findings show that while changes in schooling laws affected education, there was no effect on mortality.

Numerous studies examine the effect of education on health and health behaviors using variation in school policies to instrument for education. These studies have yielded mixed findings. Arendt (2005), for example, examines the relationship between education and health (measured by self-reported overall health, body mass index, and smoking) in Denmark. He instruments for education using school reforms intended to expand access to secondary school education. The findings suggest that better education is associated with better health, but the instruments do not perform well empirically in this study, making it hard to draw conclusions from the IV results. Kemptner et al. (2011) explore the relationship between education and health using German data, instrumenting for education using changes in compulsory schooling laws. They find evidence of causal effects of education on having a long-term illness for men, for results for other health outcomes are less consistent. Braakmann (2011) studies the effect of education and a range of health and health behaviors using data from the UK. He instruments for education using month of birth, because in the UK, school policies interact with the month of birth such that children born after 30 January are forced to attend school longer than those born before 30 January. The IV results show no effects of education on health. Other studies using compulsory schooling laws for identification show that additional schooling improves self-reported health (Oreopoulos, 2007; Silles, 2009), and may decrease the likelihood of having hypertension (although these findings are mixed) (Powdthavee, 2010).

In addition to school policies, researchers have drawn on other natural experiments to isolate the causal relationship between education and health. de Walque (2007), for example, tests whether the correlation between post-high school education and smoking behaviors (measured by the likelihood of current smoking and the likelihood of having quit smoking) is causal, using the risk of induction into Vietnam War as an instrument for education. He uses data from the smoking supplements of the NHIS between 1983 and 1995. The sample includes persons born between 1937 and 1956 with the age of 25 years and above at the time of the survey. The findings indicate that college education is causally related to a reduction in the likelihood of smoking. However, it can only be concluded that this effect occurs among individuals induced to attend college because of Vietnam draft. Grimard and Parent (2007) address the same question using a similar identification strategy, but different data. They find similar, but less consistent, evidence that education is causally related to smoking.

Siblings/twins fixed effects models also have been used to study the effect of education on health. Webbink et al. (2010), for example, use fixed effects models and data on identical twins from the Australian Twin Register to examine the causal effect of education on body weight. They find a strong association between education and overweight status, but this association only persists within twins for males (not for females). Fletcher and Frisvold (2009) estimate the association between college attendance and investments in preventive care using longitudinal data on a sample of individuals who graduated from high school in 1957 in Wisconsin. These individuals are followed for approximately 50 years. The findings show strong associations between college attendance and preventive care usage. These results persist when sibling fixed effects are included in the models. These findings are consistent with those of Lange (2011). Using data from the National Health Interview Survey (NHIS), he finds that more educated people are more likely to respond to individual risk factors for cancer by investing in preventive care than less educated people. This study suggests the mechanism through which education affects use of preventive care may be individuals’ understanding and processing of health information.

There is growing interest not just in the effect of the quantity of education on health, but also on the effects of school quality on health. Frisvold and Golberstein (2011) use data from the 1984 to 2007 NHIS, linking respondents to race-specific state-year of birth measures of school quality (such as pupil teacher ratios). A range of health outcomes are examined, including overall self-rated health, mortality, and obesity. Their findings show that higher quality schooling magnifies the effect of education on health. Similarly, Johnson (2010) uses data from the PSID and shows that within siblings, long-term childhood exposure to desegregated schools is associated with adult health, suggesting that school quality has long-run effects on health. Similarly, Kenkel et al. (2006) find that high school completion is associated with lower rates of smoking and higher rates of quitting smoking, but there are lower health returns to the GED versus the traditional high school diploma. These results also suggest there is some interaction between schooling quality and the effects of schooling on health.

Drawing Conclusions From Health Economics Research

From a policy perspective, it is critical to disentangle causal relationships between education and health from associations. If more and/or better education causes better health, then public policies that expand access to and/or improve the quality of education will also be effective in improving health. Similarly, if better health causes individuals to obtain more education, health policies can be used to increase education. If causal relationships do indeed exist, health policy and education policy are intertwined.

Economists have made important contributions in this area. There is now a convincing body of economics research supporting the idea that early health is causally related to long-term education and other economic outcomes. Health measured in utero, at birth, and during childhood and adolescence, affect outcomes such as educational attainment, labor supply, and wages in adulthood. There is also some evidence to support the idea that education causes better health, but these findings are inconsistent and vary by the health outcomes studied and the data used.

For research on education and health to be useful in shaping health and education policies, it is important not just to test for causality but also to identify causal mechanisms. Cutler and Lleras-Muney (2010) take an important step in this direction by examining the education gradient in health behaviors using data from a range of national data sets from the US and the UK. Their approach is to estimate a model in which education affects health behaviors and then include increasingly richer sets of controls in the model to see how inclusion of additional covariates affects the estimated coefficient on education. Overall, the authors conclude that material resources account for approximately 20% of the effect of education on health behaviors. Ability also accounts for a portion of the effect of education on health behaviors. This paper is an important addition to the literature because mechanisms through which education may affect health can now be understood.

Moreover, it is important to understand whether the effects of education on health, and the effects of health on education, are heterogeneous in the population. For example, some research suggests that the effects of education on health vary by individuals’ sociodemographic characteristics (Cutler and Lleras-Muney, 2006). Other studies support the idea that education causes better health, but the results are relevant to only subpopulations (often the lowest part of the education distribution), and, based on existing research, cannot be generalized to the entire population (Lleras-Muney, 2005). It is essential to know which groups are most likely to respond to changes in education, or to changes in health. Economics research has the potential to answer these questions about mechanisms and heterogeneity of effects, and thus help in shaping the development of effective health and education policies.

Bibliography:

  1. Albouy, V. and Lequien, L. (2009). Does compulsory education lower mortality? Journal of Health Economics 28, 155–168.
  2. Almond, D. (2006). Is the 1918 influenza pandemic over? Long-term effects of in utero influenza exposure in the post-1940 U.S. population. Journal of Political Economy 114, 672–712.
  3. Almond, D., Chay, K. and Lee, D. (2005). The costs of low birthweight. Quarterly Journal of Economics 120(3), 1031–1083.
  4. Almond, D., Edlund, L. and Palme, M. (2009). Chernobyl’s subclinical legacy: Prenatal exposure to radioactive fallout and school outcomes in Sweden. The Quarterly Journal of Economics 1729–1772.
  5. Almond, D. and Mazumder, B. (2011). Health capital and the prenatal environment: The effect of Ramadan observance during pregnancy. American Economic Journal: Applied Economics 3, 56–85.
  6. Arendt, J. N. (2005). Does education cause better health? A panel data analysis using school reforms for identification. Economics of Education Review 24, 149–160.
  7. Banerjee, A., Duflo, E., Postel-Vinay, G. and Watts, T. (2010). Long run health impacts of income shocks: Wine and phylloxera in nineteenth-century France. The Review of Economics and Statistics 92, 714–728.
  8. Barker, D. J. P., Winter, P. D., Osmond, C., Margetts, B. and Simmonds, S. J. (1989). Weight in infancy and death from ischaemic heart disease. Lancet 334, 577–580.
  9. Behrman, J. R. and Rosenzweig, M. R. (2004). Returns to birthweight. The Review of Economics and Statistics 86, 586–601.
  10. Black, S. E., Devereux, P. J. and Salvanes, K. G. (2007). From the cradle to the labor market? The effect of birth weight on adult outcomes. The Quarterly Journal of Economics 122(1), 409–439.
  11. Bleakley, H. (2007). Disease and development: Evidence from hookworm eradication in the south. The Quarterly Journal of Economics 122, 73–117.
  12. Bleakley, H. (2010). Malaria eradication in the Americas: A retrospective analysis of childhood exposure. American Economic Journal: Applied Economics 2, 1–45.
  13. Braakmann, N. (2011). The casual relationship between education, health and health related behaviour: Evidence from a natural experiment in England. Journal of Health Economics 30, 753–763.
  14. Carneiro, P., Meghir, C. and Parey, M. (2011). Maternal education, home environments and the development of children and adolescents. Institute for Fiscal Studies Working Paper. Cambridge, MA: National Bureau of Economic Research.
  15. Case, A., Fertig, A. and Paxson, C. (2005). The lasting impact of childhood health and circumstance. Journal of Health Economics 24, 365–389.
  16. Case, A. and Paxson, C. (2009). Early life health and cognitive function in old age. American Economic Review: Papers & Proceedings 99, 104–109.
  17. Chay, K. and Greenstone, M. (2003). The impact of air pollution on infant mortality: Evidence from geographic variation in pollution shocks induced by a recession. The Quarterly Journal of Economics 118, 1121–1167.
  18. Chay, K. Y., Guryan, J. and Mazumder, B. (2009). Birth cohort and the black-white achievement gap: The roles of access and health soon after birth. Working Paper 15078, National Bureau of Economics Research. Available at: http:// www.nber.org/papers/w15078
  19. Chen, Y. and Li, H. (2009). Mother’s education and child health: Is there a nurturing effect? Journal of Health Economics 28, 413–426.
  20. Chen, Y. and Zhou, L. (2007). The long-term health and economic consequences of the 1959–1961 famine in China. Journal of Health Economics 26, 659–681.
  21. Chou, S., Liu, J., Grossman, M. and Joyce, T. J. (2010). Parental Education and Child Health: Evidence from a Natural Experiment in Taiwan. American Economic Journal: Applied Economics 2, 33–61.
  22. Contoyannis, P. and Dooley, M. (2010). The role of child health and economic status in educational, health, and labour market outcomes in young adulthood. Canadian Journal of Economics 43, 323–346.
  23. Currie, J. and Moretti, E. (2003). Mother’s education and the intergenerational transmission of human capital: Evidence from college openings. The Quarterly Journal of Economics 118, 1495–1532.
  24. Currie, J. and Stabile, M. (2006). Child mental health and human capital accumulation: The case of ADHD. Journal of Health Economics 25, 1094–1118.
  25. Currie, J., Stabile, M., Manivong, P. and Roos, L. L. (2010). Child health and young adult outcomes. The Journal of Human Resources 45, 518–548.
  26. Cutler, D. M. and Lleras-Muney, A. (2006). Education and health: Evaluating theories and evidence. Working Paper 12352. Cambridge, MA: National Bureau of Economic Research.
  27. Cutler, D. M. and Lleras-Muney, A. (2010). Understanding differences in health behaviors by education. Journal of Health Economics 29, 1–28.
  28. Cutler, D. M., Meara, E. R. and Richards, S. (2008). The gap gets bigger: Changes in mortality and life expectancy by education, 1981–2000. Health Affairs 27, 350–360.
  29. Delaney, L., McGovern, M. and Smith, J. P. (2011). From Angela’s ashes to the Celtic tiger: Early life conditions and adult health in Ireland. Journal of Health Economics 30, 1–10.
  30. Ding, W., Leher, S. F., Rosenquist, J. N. and Audrain-McGovern, J. (2009). The impact of poor health on academic performance: New evidence using genetic markers. Journal of Health Economics 28, 578–597.
  31. Eide, E. R. and Showalter, M. H. (2011). Estimating the relation between health and education: What do we know and what do we need to know? Economics of Education Review 30, 778–791.
  32. Eisenberg, D., Golberstein, E. and Hunt, J. B. (2009). Mental health and academic success in college. The B.E. Journal of Economic Analysis and Policy 9, 1–35.
  33. Fletcher, J. M. (2010). Adolescent depression and educational attainment: Results using sibling fixed effects. Health Economics 19, 855–871.
  34. Fletcher, J. M. (2011). The medium term schooling and health effects of low birth weight: Evidence from siblings. Economics of Education Review 30, 517–527.
  35. Fletcher, J. M. and Frisvold, D. E. (2009). Higher education and health investments: Does more schooling affect preventive health care use? Journal of Human Capital 3, 144–176.
  36. Fletcher, J. M., Green, J. C. and Neidell, M. J. (2010). Long term effects of childhood asthma on adult health. Journal of Health Economics 29, 377–387.
  37. Fletcher, J. M. and Lehrer, S. F. (2009). The effects of adolescent health on educational outcomes: Casual evidence using genetic lotteries between siblings. Forum for Health Economics & Policy 12, 1–31.
  38. Fletcher, J. M. and Wolfe, B. (2008). Child mental health and human capital accumulation: The case of ADHD revisited. Journal of Health Economics 27, 794–800.
  39. Frisvold, D. and Golberstein, E. (2011). School quality and the education–health relationship: Evidence from Blacks in segregated schools. Journal of Health Economics 30, 1232–1245.
  40. Grimard, F. and Parent, D. (2007). Education and smoking: Were Vietnam war draft avoiders also more likely to avoid smoking? Journal of Health Economics 26, 896–926.
  41. Grossman, M. (1972a). On the concept of health capital and the demand for health. Journal of Political Economy 80, 233–255.
  42. Grossman, M. (1972b). The Demand for Health: A Theoretical and Empirical Investigation. New York: Columbia University Press, for the National Bureau of Economic Research.
  43. Grossman, M. (2000). The human capital model. In Culyer, A. and Newhouse, J. (eds.) Handbook of Health Economics, vol. 1A, pp. 348–408. Amsterdam: North Holland.
  44. Grossman, M. (2008). The relationship between health and schooling. Eastern Economic Journal 34, 281–292.
  45. Grossman, M. and Kaestner, R. (2009). Effect of weight on children’s educational achievement. Economics of Education Review 28, 651–661.
  46. Johnson, R. C. (2010). The health returns of education policies from preschool to high school and beyond. American Economic Review: Papers and Proceedings 100, 188–194.
  47. Johnson, R. C. and Schoeni, R. F. (2010). The influence of early-life events on human capital, health status, and labor market outcomes over the life course. The B.E. Journal of Economic Analysis and Policy 2, 188–194.
  48. Johnson, R. C. and Schoeni, R. (2011). The influence of early-Life events on human capital, health status, and labor market outcomes over the life course. The B.E. Journal of Economic Analysis & Policy: Advances 11,(3), article 3.
  49. Kemptner, D., Jurges, H. and Reinhold, S. (2011). Changes in compulsory schooling and the casual effect of education on health: Evidence from Germany. Journal of Health Economics 30, 340–354.
  50. Kenkel, D., Lillard, D. and Mathios, A. (2006). The roles of high school completion and GED receipt in smoking and obesity. Journal of Labor Economics 24, 635–660.
  51. Lange, F. (2011). The role of education in complex health decisions: Evidence from cancer screening. Journal of Health Economics 30, 43–54.
  52. Lleras-Muney, A. (2005). The relationship between education and adult mortality in the United States. Review of Economic Studies 72, 189–221.
  53. Maccini, S. and Yang, D. (2009). Under the weather: Health, schooling, and economic consequences of early-life rainfall. American Economic Review 99, 1006–1026.
  54. McCrary, J. and Royer, H. (2011). The effect of female education on fertility and infant health: Evidence from school entry policies using exact date of birth. American Economic Review 101, 158–195.
  55. Oreopoulos, P. (2007). Do dropouts drop out too soon? Wealth, health and happiness from compulsory schooling. Journal of Public Economics 91, 2213–2229.
  56. Oreopoulos, P., Stabile, M., Walld, R. and Roos, L. (2008). Short, medium, and long-term consequences of poor infant health: An analysis using siblings and twins. Journal of Human Resources 43.
  57. Powdthavee, N. (2010). Does education reduce the risk of hypertension? Estimating the biomarker effect of compulsory schooling in England. Journal of Human Capital 4, 173–202.
  58. Rees, D. I. and Sabia, J. J. (2011). The effect of migraine headache on educational attainment. The Journal of Human Resources 46, 317–332.
  59. Royer, H. (2009). Separated at girth: US twin estimates of the effects of birth weight. American Economic Journal: Applied Economics 1, 49–85.
  60. Sabia, J. J. (2007). The effect of body weight on adolescent academic performance. Southern Economic Journal 73, 871–900.
  61. Silles, M. A. (2009). The casual effect of education on health: Evidence from the United Kingdom. Economics of Education Review 28, 122–128.
  62. Smith, J. P. (2009). The impact of childhood health on adult labor market outcomes. The Review of Economics and Statistics 91, 478–489.
  63. Staiger, D. and Stock, J. H. (1997). Instrumental variables regression with weak instruments. Econometrica 65, 557–586.
  64. Van Ewijk, R. (2011). Long-term health effects on the next generation of Ramadan fasting during pregnancy. Journal of Health Economics 30, 1246–1260.
  65. de Walque, D. (2007). Does education affect smoking behaviors? Evidence using the Vietnam draft as an instrument for college education. Journal of Health Economics 26, 877–895.
  66. Webbink, D., Martin, N. G. and Visscher, P. M. (2010). Does education reduce the probability of being overweight? Journal of Health Economics 29, 29–38.
  67. Barker Theory (2010). Available at: http://www.thebarkertheory.org (accessed February 2012).
  68. Chatterji, P., Joo, H. and Lahiri K. (2012). Racial/ethnic and education-related disparities in control of risk factors for cardiovascular disease among diabetics. Diabetes Care 35, 305–312.
  69. Chay, K. and Greenstone, M. (2005). Does air quality matter? Evidence from the housing market. Journal of Political Economy 113, 376–424.
  70. Conti, G., Heckman, J. and Urzua, S. (2010). The education–health gradient. American Economic Review Papers and Proceedings 100, 234–238.
  71. Currie, J. (2009). Healthy, wealthy, and wise: Socioeconomic status, poor health in childhood, and human capital development. Journal of Economic Literature 47, 87–122.
  72. Currie, J. (2011). Inequality at birth: Some causes and consequences. American Economic Review: Papers & Proceedings 101, 1–22.
  73. Cutler, D. M., Lange, F., Meara, E., Richards-Shubik, S. and Ruhm, C. J. (2011). Rising educational gradients in mortality: The role of behavioral risk factors. Journal of Health Economics 30, 1174–1187.
  74. Cutler, D. M. and Lleras-Muney, A. (2012). Education and health: Insights from international comparisons. Working Paper 17738. Cambridge, MA: National Bureau of Economic Research.
  75. Fuchs, V. R. (1982). Time preferences and Health: An Exploratory Study. In Fuchs, V. R. (ed.) Economic Aspects of Health. Chicago: University of Chicago Press.
  76. Goldman, D. R. and Smith, J. P. (2002). Can patient self-management help explain the SES health gradient? Proceedings of the National Academy of Sciences 99, 10929–10934.
  77. Grossman, M. (2006). Education and non-market outcomes. In Eric, H. and Finis, W. (eds.) Handbook of the Economics of Education, vol. 1, pp. 577–633. Amsterdam: Elsevier.
  78. Kaestner, R. and Grossman, M. (2009). Effects of weight on children’s educational achievement. Economics of Education Review 28, 651–661.
  79. Mazumder, B. (2008). Does education improve health? A reexamination of the evidence from compulsory schooling laws. Federal Reserve Bank of Chicago Economic Perspectives 32, 2–16.
  80. Meara, E. (2001). Why is health related to socio-economic status? The case of pregnancy and low birth weight. Working Paper 8231. Cambridge, MA: National Bureau of Economic Research.
  81. Van Den Berg, G. J., Lindeboom, M. and Portrait, F. (2006). Economic conditions early in life and individual mortality. American Economic Review 96, 290–302.
  82. Van Der Pol, M. (2011). Health, education and time preferences . Health Economics 20, 917–929.
Demand for Insurance That Nudges Demand
Medical Decision Making and Demand