Journal Issue: Children with Disabilities Volume 22 Number 1 Spring 2012
Childhood Disability, Future Human Capital, and Economic Success
The second strain of the literature we review seeks to determine whether and how childhood disability affects the accumulation of education, skills, and other human capital and consequently economic well-being in adulthood. We look first at disability at birth and then at childhood disability.
Disability at Birth
An extensive literature examines the future economic cost of being born prematurely or with low birth weight or low Apgar scores (standardized evaluations of a newborn's health condition). While these conditions are not measures of disability themselves, they are associated with higher rates of disability and thus can be regarded as marker conditions. The goal of this literature is to determine whether children born with one of these conditions suffer adverse health and economic consequences later in life.
A key empirical challenge for these studies is the strong correlation between being born with one of these markers and other disadvantages such as low socioeconomic status. Therefore, separating the causal effect of being born with a marker condition from the effect of being born, say, into a family in poverty, has been a focus of the more recent work in this area. We focus here on the relationship between markers of poor health at birth, future disability, and future economic outcomes.
The most recent social science literature in this area has used a combination of large administrative data sets and samples of twins and siblings to examine the longer-term effects of health at birth on both education and labor market success. As noted, the most common measures found in the literature are birth weight, Apgar scores, and length of gestation. In general these measures are considered more objective than survey measures of infant health. Weight at birth is considered low if it is below 2,500 grams, and very low if it is below 1,500 grams. Gestational periods are considered premature if they are below thirty-seven weeks. Apgar scores are based on five items and scored on a scale of ten. Scores below seven are considered poor.76
Jere Behrman and Mark Rosenzweig used data on twins from the Minnesota Twins Registry to examine the effects of low birth weight on the educational attainment and adult health of women.77 They found that increasing birth weight by one pound (454 grams) increased schooling attainment by about one-third of a year and that the difference in schooling attainment was larger between twins with different birth weights than across families with children of different birth weights. Using the Panel Study of Income Dynamics (PSID), Dalton Conley and Neil Bennett found that low birth weight had a more pronounced influence on timely high school graduation among siblings with different birth weights than between families.78 These findings suggest that differences in birth weight between siblings account for much of the observed relationship between birth weight and educational attainment. Differences in birth weight between families account for less of this relationship.
Many of the findings in the United States can be extended by using evidence from other nations where the data are much richer and permit more robust studies of the long-term effects of disability at birth and in childhood. One study showed that, conditional on many measures of family background and circumstances, low-birth-weight children from the 1958 British birth cohort (the National Child Development Study, or NCDS) had lower test scores, educational attainments, wages, and probabilities of being employed at age thirty-three than those with healthy birth weights.79 Another study of a sample of Norwegian twins found that low birth weight was associated with lower height, IQ, educational attainment, and earnings.80 A third study used administrative data from the Canadian province of Manitoba and found both low birth weight and low Apgar scores to be strong predictors of lower rates of high school completion and greater use of welfare for longer periods of time.81
The evidence over the past few years strongly indicates that even when other factors associated with health at birth are accounted for, children born with less than optimal health suffer from lower educational outcomes and poorer labor market outcomes on average.
Disability in Early Childhood
The development of physical or mental disabilities in early childhood can have both immediate and longer-term consequences for human capital accumulation and economic well-being. Most research in this area tends to focus on general measures of physical disability in early childhood, measures of childhood mental health, or specific physical conditions such as asthma (a recent exception is a study by Janet Currie and others, which examined all three of these groupings using administrative data82). The literature has explored a range of health measures from subjective self-assessments of health to reported chronic conditions to administrative records of health problems. While the ideal set of health measures is open to some debate, the findings across these measures are mainly consistent with one another. We review the main findings in each of these areas.
Measures of Physical Disability. The literature on chronic physical disability finds a consistent relationship between early childhood health and longer-term outcomes. Anne Case and her colleagues used data from the 1958 British birth cohort study, which allowed them to track children from childhood into middle age.83 They examined childhood chronic conditions reported at ages seven and sixteen and found that children with such conditions had lower educational attainment, wages, and employment probabilities at age thirty-three than other children. Using the 1958 study as well as one other British survey (the Whitehall II study of British civil servants), and two American surveys (the PSID and the Health and Retirement Study), Anne Case and Christina Paxson found that childhood health, measured using height as a proxy, was associated with a number of later life outcomes: taller children tended to attain more schooling, employment, earnings, and health.84 Case and Paxson also drew on the British Whitehall II study to show the long-term effects of early health on occupational attainment, with health proxied by a report of hospitalization for more than four weeks before age sixteen.85 They found that adults who had better childhood health were more likely to start at higher grades within the civil service and were more likely to be promoted once they entered the civil service.
Another study that examined the long-term effects of child health used a retrospective health measure with data from the PSID.86 In 1999 PSID respondents aged twenty-five to forty-seven were asked whether their health when they were less than sixteen was excellent, very good, good, fair, or poor. In models with sibling comparisons, the adults who had suffered poorer health in childhood not only started at a lower level of earnings but experienced slower earnings growth over time than their healthier siblings.
Janet Currie and her colleagues used administrative data from Canada to track physical and mental health of children at various points in childhood (ages zero to three through ages fourteen to eighteen).87 Using sibling comparisons (family fixed-effects models), they examined the relationship between health at different points in childhood and various outcomes including educational attainment and welfare take-up. They found that both poor health at birth and early mental health disabilities were associated with poorer long-term outcomes, even when one accounts for the health status of the child later in life. Physical disabilities in early childhood were also associated with poorer outcomes, consistent with the findings in other studies, but apparently because they predict future disabilities rather than leading directly to the poorer outcomes. Unless they persisted over time, physical disabilities in childhood had little effect on future educational outcomes and welfare take-up.
Mental Health Disabilities. According to the U.S. surgeon general's report in 1999, approximately one in five children and adolescents in the United States exhibits signs or symptoms of mental or behavioral disorders.88 This high prevalence of mental health problems among children and the potential for these problems to hinder the accumulation of human capital are worrisome. While the body of literature examining the effects of mental health disabilities is considerably smaller than that examining physical health, an increasing number of studies have explored the effects of common mental health conditions such as ADHD.
Studies seeking to examine the effects of mental health disabilities on child outcomes encounter several challenges. To begin with, definitive tests that allow for a conclusive diagnosis do not exist for most mental health disorders. Diagnoses are often made through a series of questions that are asked of parents and teachers, combined with observation of the child. The "threshold" for having a mental health disability is thus not entirely clear. Second, society's acknowledgment of mental health problems as health disorders rather than poor behavior on the part of children has changed over time and continues to differ across cultures. Third, treatment for mental health problems, particularly for ADHD, has increased fairly rapidly, making it difficult to assess the effect of these problems with and without treatment.89 Finally, as with other measures of health, there are large differences in mental health by socioeconomic status: one study, for example, reports that the prevalence rate of ADHD is almost twice as high for families in the United States with incomes below $20,000 as for those with higher incomes. Observed differences in outcomes across children with and without a mental health problem may therefore partially reflect these other observable and unobservable differences across children.
Three strands of literature have attempted to address these empirical challenges. First, several studies focus on particular "externalizing" mental health conditions (for example, ADHD, conduct disorder, and oppositional-defiant disorder). Salvatore Mannuzza and Rachel Klein reviewed three studies of the long-term outcomes of children with ADHD.90 In one study, ADHD children were matched to controls from the same school who had never exhibited any behavior problems and had never failed a grade; in a second study, controls were recruited at the nine-year follow-up from nonpsychiatric patients in the same medical center who had never had behavior problems; and in a third study, ADHD children sampled from a range of San Francisco schools were compared to non-ADHD children from the same group of schools. These comparisons consistently show that the ADHD children had worse outcomes in adolescence and young adulthood than control children. For example, they had completed less schooling and were more likely to have continuing mental health problems. By excluding children with any behavior problems from the control groups, however, the studies may have overstated the effects of ADHD.
A second set of studies looked at the longer-term consequences of behavior problems in relatively large samples. One examined adolescents who met diagnostic criteria for four types of disorders: anxiety, depression, hyperactivity, and conduct disorders when they were evaluated at age fifteen and who were followed up to age twenty.91 Those in the sample with hyperactivity and conduct disorders completed fewer grades, while anxiety and depression had little effect on schooling levels. Another study used the NLSY data to show that children who had behavior problems at ages six to eight were less likely to graduate from high school or to attend college, even after accounting for differences among the mothers of these children.92 Like the first study, these researchers found that externalizing behavior problems were significant predictors of future outcomes, whereas internalizing problems were not. One limitation of this study is its focus on a relatively small number of children, who, given the design of the NLSY, were born primarily to young mothers. Several slightly older studies have found similar results. For example, children with early onset psychiatric problems were less likely to have graduated from high school or attended college.93
Elizabeth Farmer used data from the 1958 British birth cohort study to examine the consequences of childhood externalizing behavioral problems on men's outcomes at age twenty-three. She found that boys who fell into the top decile of an aggregate behavior problems score at ages seven, eleven, or sixteen had lower educational attainment, earnings, and probabilities of employment at age twenty-three.94 A separate study that used the NCDS data found that behavioral problems at age seven were related to poorer educational attainment at age sixteen, which in turn was associated with poor labor market outcomes at ages twenty-three and thirty-three.95 A study of a cohort of all children born between 1971 and 1973 in Dunedin, New Zealand, found that those with behavior problems at age seven to nine were more likely to be unemployed at age fifteen to twenty-one than those without such problems.96
Taken together, this research consistently shows that the children with ADHD and other behavior problems have worse outcomes in adolescence and young adulthood than control children, but the studies do not address the possibility that the negative outcomes might be caused by other factors related to a diagnosis of ADHD, such as poverty, the presence of other learning disabilities, or the fact that many people diagnosed with ADHD end up in special education.
To address some of these concerns around selection into diagnosis and biases from omitted variables, Janet Currie and Mark Stabile used data from the NLSY and the Canadian National Longitudinal Survey of Children and Youth to examine the experience of children with symptoms of mental health problems as reported by parents and teachers. They compared affected children to their own siblings (within a sibling fixed-effects context similar to the models used to examine low birth weight reviewed earlier).97 An advantage of using survey data is that questions about symptoms of mental health problems were asked of all children, whereas only children who are brought in for treatment receive a diagnosis. The survey questions are similar to those that would be used as part of a medical diagnosis, and because all of the children surveyed are asked the same questions, a "mental health score" can be constructed for all children in the sample, including those with potentially mild disabilities that would not result in a diagnosis. This feature allows researchers to examine the effect of both high and low levels of mental health disability on outcomes. Finally, because children are compared with their own siblings, the estimates control for both observed and unobserved family characteristics that are shared by siblings.
Currie and Stabile found that in both data sets children with symptoms of ADHD had a higher probability of future grade repetition and lower test scores in math and reading. These probabilities were large relative to those of physical health problems in these same samples of children and appear even among children with symptoms of ADHD that would generally be considered too low to warrant a diagnosis. For example, the results suggest that the effect of moving from the mean to the lowest hyperactivity score in the United States on the probability of repeating a grade is similar to the effect of an additional $50,000 in family income. The results are strikingly similar across children in the United States and Canada despite the significant differences in the health insurance systems across the two countries. The authors also found that socioeconomic status made surprisingly little difference; outcomes for poorer siblings were about the same as those for better-off siblings. Boys with higher levels of ADHD symptoms do worse than girls with the same levels of symptoms, however, particularly in the United States. The U.S. results were replicated and extended by Jason Fletcher and Barbara Wolfe, who found similar short-term effects but also found that these effects dissipated over time, meaning that there was little difference in educational outcomes between children with and without ADHD.98 Fletcher and Wolfe also showed that having a sibling with ADHD was detrimental to educational outcomes for the non-ADHD sibling over the longer run. This finding may lead to the smaller estimated effects in models that rely on sibling comparisons over time, because the sibling without ADHD is also negatively affected.
In a related paper, Currie and Stabile examined a variety of mental health problems, including depression and conduct disorders, as well as ADHD and a general index of behavioral problems.99 While ADHD remained the mental health disorder most strongly associated with poor educational outcomes in the future, conduct disorders and depression had some effect on grade repetition. Consistent with other studies, the effects of early mental health disorders persist into the future even when Currie and Stabile controlled for contemporaneous mental health problems, suggesting that the effects of these problems may be cumulative and costly.
James Smith and Gillian Smith used retrospective health questions in the 2007 PSID wave to show that depression, substance abuse, and other psychological problems experienced in childhood significantly reduced the number of weeks worked a year and the level of earnings in adulthood, even after they adjusted for fourteen childhood physical illnesses and controlled for within-sibling differences.100 Like Currie and Stabile, they suggest that the effects of mental health problems are much greater than those of most physical health problems (see the article by Delaney and Smith in this volume for further discussion of this point).101
Reviewing the literature on a wide variety of individual physical health problems and their effects on children is too broad a task for this article, but we do examine the literature on the relationship between childhood asthma and future outcomes given the large numbers of children who suffer from asthma. Estimates in the United States suggest that one in ten children has asthma and that the prevalence of asthma among children has doubled over the past twenty-five years.102 Asthma also tends to be more prevalent in lower-income households than in better-off ones. As with the other disabilities examined here, understanding the longer-term consequences of asthma in childhood is complicated by this correlation with socioeconomic status, treatment effects, and other omitted variables that may be correlated with all of these.
Point-in-time comparisons support a correlation between asthma in childhood and poor future health.103 A study that used sibling comparisons from the Study of Adolescent Health found that having childhood asthma increased the number of missing school or work days in young adults by 10 percentage points—a considerable loss in human capital and productivity.104