NBER Reporter 2009 Number 4: Research Summary
The Effects of Medicaid Expansions and Welfare Reform on Fertility and the Health of Women and Children
Expansion of publicly-financed health insurance programs has been an integral feature of recent efforts to reduce the number of uninsured, non-elderly persons in the United States. The origin of this approach was the creation of Medicaid in 1965. The expansion of Medicaid eligibility to near-poor, pregnant women who were not eligible for Aid to Families with Dependent Children, which took place between 1986 and 1989, solidified the use of this approach for reducing the number of non-elderly uninsured.
The expansion of Medicaid and the creation of the State Children's Health Insurance Program (SCHIP) resulted in a large increase in the number of women and children with publicly financed health insurance. For example, between 1987 and 2008 the proportion of children under age 18 (under age 3) with publicly financed health insurance increased from 16 to 30 percent (20 to 38 percent). Surprisingly, though, the proportion of children without health insurance coverage did not decline commensurately with these changes. Between 1987 and 2008, the proportion of children under age 18 (age 3) without any health insurance coverage decreased from 13 percent to 10 percent (12 percent to 9 percent). The large increase in publicly provided health insurance coverage and the relatively small decrease in rates of uninsured between 1987 and 2008 suggest that part of the increase in participation in public programs came at the expense of private insurance coverage. This substitution of public for private health insurance coverage -- "crowd out" -- continues to be an important part of the debate over how to reduce the number of uninsured. 1
While estimates of the extent of crowd out vary, most studies find some level of crowd out, which differs by age of child and by family income. For example, Yazici and I 2 estimate that approximately 15 percent of the increase in Medicaid enrollment among children up to age 9 between 1988 and 1992 was at the expense of private insurance. Two other important findings emerge from this analysis. First, the expansions were associated with an increase in enrollment for those who were always eligible for Medicaid. This might be because the expansions increased the proportion of persons enrolled in Medicaid and, as a result, decreased the stigma associated with participation. Or, it could be because the expansions increased the number of providers who serve Medicaid patients, thereby making it easier to find a provider and increasing the benefit of Medicaid participation. Second, much of the switching from private to public insurance occurred among families that suffered employment and income losses. For this group, publicly provided insurance was a much needed backstop, not simply a desirable alternative to private insurance.
Given that the Medicaid expansions initially were targeted at pregnant women, it is notable that there has been very little study of the issue of crowd out for that group. In a recent paper, some colleagues and I address this research gap using confidential data from the National Hospital Discharge Survey, which provides information about the health insurance coverage of women giving birth. 3 This study too has several notable findings. First, the effects of the Medicaid expansions differed significantly by the level of eligibility. Eligibility expansions that occurred in the late 1980s and were targeted at the poorest women resulted in significantly larger declines in the proportion of uninsured pregnant women than later expansions that targeted higher income groups. The differential declines in the proportion uninsured are attributable to much smaller relative declines in private insurance among the poorest women, which is reasonable given that they were least likely to be covered by private insurance prior to the expansion. However, because the expansions increasingly focused on women from higher income groups who were more at risk of switching from private insurance, the extent of crowd out grew over time. Indeed, our estimates suggest that up to 80 percent of the growth in Medicaid enrollment among women in the highest income eligibility groups came at the expense of private insurance.
While Medicaid and other publicly provided health insurance programs focus on health insurance coverage, the ultimate purpose of these programs is to improve the health of previously uninsured persons by giving them the financial means to obtain the care required to maintain good health. This purpose is explicit with regard to the initial expansions of Medicaid that were focused on pregnant women and that were partly motivated by the 1985 Institute of Medicine Report entitled Preventing Low Birthweight, which concluded that more and timelier prenatal care could reduce infant mortality by preventing low birth weight.
Surprisingly, after twenty years of study, the jury is still out on the efficacy of Medicaid in improving infant and child health. In a recent paper, my colleagues and I re-examine the relationship between the Medicaid expansions and infant health using the National Hospital Discharge Survey. 4 We ask whether the Medicaid expansions for pregnant women were associated with shorter stays in the hospital at the time of delivery for both the mother and child; with deliveries by cesarean section; and with deliveries in a public versus private hospital. Our results indicate that the Medicaid expansions were not significantly associated with these outcomes except in the case of cesarean section: the rates of cesarean section increased significantly as Medicaid eligibility expansion increased, but that was probably attributable to factors other than just Medicaid expansions.
We also examine the association between birth weight and the Medicaid expansions. We include arguably more complete controls than earlier studies did for the non-random nature of the Medicaid expansions, and we allow for different effects of the expansions depending upon the level of eligibility -- higher levels of income eligibility should have had smaller effects, given our previous findings that higher levels of Medicaid eligibility were associated with smaller decreases in the proportion of women uninsured. Our results largely confirm our earlier findings and those of other researchers in suggesting that Medicaid expansions had little effect on infant health. 5 However, our study, like most of the previous studies, did not actually examine the effect of being covered by Medicaid on infant health, but rather the effect of being eligible for Medicaid. The former effect remains largely unknown because of the absence of adequate data, although in one study I examine the association between Medicaid participation (versus being uninsured) on infant health, as proxied by birth weight. I find little evidence that Medicaid participation improved birth weight. 6
There are few studies of the effect of Medicaid expansions on child health, presumably because of the lack of suitable data for conducting them. In one study, colleagues and I use data from hospital discharge records to assess whether Medicaid expansions were associated with a decrease in children's admission to the hospital for ambulatory-care-sensitive conditions, which are illnesses that arguably are can be avoided by adequate primary care. 7 We find that the Medicaid expansions were associated with a relatively large decrease, on the order of 10 to 20 percent, in the incidence of ambulatory-care-sensitive discharges among young children in low-income families, although the results were not uniform.
The Medicaid income-eligibility expansions of the late 1980s also may have affected fertility decisions of newly eligible groups of women, because the availability of Medicaid reduces the cost of giving birth and the costs of medical care for children in families eligible to participate. My colleagues and I examine this possibility in two papers. 8 Using information from several states, we find that the Medicaid expansions were associated with an increase in the birthrate among white women, but did not influence that rate among black women. This increase in birthrate was consistent with our findings related to abortion in a subsample of states. In these states, the Medicaid expansions were associated with a significant decline in the abortion rate among white women. Based on these results, we conclude that, in the states examined, subsidized health care for low-income pregnant women may have encouraged some groups of women to have more children (and fewer abortions) than they would have had without such coverage. These results and this issue are again relevant to current health care reform efforts because these efforts include expansions in publicly financed health insurance, although to groups with higher incomes than those that we studied.
One of the most important social policy changes in the last 40 years was the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA), which has become known as welfare reform. A major goal of welfare reform was to change the provisions in the welfare system that were believed to encourage behaviors such as out-of-wedlock birth that make welfare receipt more likely. Accordingly, PRWORA placed time limits on benefits, withheld cash assistance for increases in family size (that is, family cap), and required recipients to work. In short, welfare reform eliminated the AFDC program's entitlement to long-term income support for unmarried women with children. The objective was to reduce the benefits of welfare and to encourage women to avoid non-marital childbearing by either reducing fertility or increasing marriage.
In several papers, my colleagues and I investigate whether welfare reform changed fertility and marriage decisions. The most comprehensive of these studies uses data from all 50 states to investigate whether state and federal welfare reform in the 1990s reduced rates of non-marital childbearing among women aged 19 to 39 who were at highest risk of welfare use. 9 We find little consistent evidence for an effect of welfare reform on marriage or on non-marital childbearing. This is similar to the results in an earlier literature of little, or mixed, evidence that more generous levels of cash assistance in the AFDC program affected fertility. If anything, we find that AFDC waivers, which predated welfare reform, were associated with a negative effect of 7 to 9 percent on Hispanic women's fertility, and that federal welfare reform was associated with a small positive effect of between 3 and 4 percent for white and black women. However, the absence of a consistent set of findings -- for example, a similar effect of a given policy for women of all race/ethnic groups -- and the predominance of insignificant effects lead us to conclude that, in general, state and federal reform did not affect fertility. In another study, we use birth and abortion information for 24 states to estimate the effects of the family cap provisions of welfare reform on birth and abortion rates. 10 In family cap states, birth rates fell more and abortion rates increased more among high-risk women with at least one previous live birth than for similar childless women, which is consistent with an effect of the family cap. However, this parity-specific pattern of births and abortions also occurred in states that implemented welfare reform without a family cap. Thus, the effects of welfare reform may have differed between mothers and childless women, but there is little evidence of an independent effect of the family cap.
One explanation for these findings is that there are strong cohort patterns in non-marital fertility; exposure during the early teen years to a policy regime without a welfare entitlement might produce large behavioral changes for such "entering" cohorts but little change among older cohorts. To investigate this possibility, we use data from the National Longitudinal Surveys of Youth 1979 and 1997 to compare welfare use, fertility, and marriage among teenage women in the years before and immediately following welfare reform. 11 We find significant differences between cohorts in welfare use and in outcomes related to welfare use. Welfare reform is associated with reduced welfare receipt, reduced fertility, and reduced marriage among young women who, because of a disadvantaged family background, are at high risk of welfare receipt.
The goal of PRWORA was to change behavior, to decrease dependence on government assistance and to increase economic self-sufficiency. While the evidence suggests that welfare reform did not have a significant effect on marriage and fertility, it was very successful at reducing the welfare rolls and increasing work. Notably, the decline in the welfare caseload and the increase in employment among low-income, unmarried women may have significantly reduced the prevalence of health insurance among this group. Health insurance coverage for these families may have been adversely affected, because those who left or were deterred from entering welfare may have found it difficult to obtain Medicaid coverage because of administrative hurdles, and because many of the jobs that low-skilled women typically obtain after leaving welfare did not offer private health insurance. In turn, the loss of health insurance may have adversely affected these women's ability to obtain health care for themselves and their children, and may have adversely affected their health. In a series of papers, colleagues and I study these possible consequences of welfare reform.
In the first paper, we examine whether welfare reform affected health insurance coverage. 12 Our results indicate that changes in the welfare caseload were associated with an increase in the proportion of low-educated women and their children who are without health insurance. Our estimates also suggest that the 42 percent decrease in the caseload between 1996 and 1999 was associated with: a decrease in Medicaid participation of between 3 and 4 percentage points (between 7 and 9 percent); an increase in employer-sponsored insurance coverage of 2 percentage points (6 percent); and an increase in the proportion uninsured of between 0.5 and 2.5 percentage points (2-9 percent). For children in these families, the decline in the caseload between 1996 and 1999 was associated with similar, but smaller effects. We also estimate the effect of changes in the caseload attributable to state and federal welfare reform policy. Because welfare policy was responsible for only part, perhaps one third, of the decline in the caseload, welfare reform per se had significantly smaller effects on the health insurance status of low-income families. However, we find that changes in the caseload due to state and federal welfare policy had fewer adverse consequences on insurance status than changes in the caseload due to other factors. This latter finding is plausible, because women induced to leave or not enter welfare because of government policy may be much more likely to take advantage of transitional Medicaid benefits and to find jobs that provide health insurance than women induced to leave the program because of a strong economy.
Given that we find that welfare reform was associated with some loss of health insurance, my colleagues and I examine whether this affected health care use and health of these women and their children. We first study pregnant women and ask whether welfare reform affected their use of prenatal care and infant health. 13 Our findings indicate that welfare reform had at most relatively small effects on the prenatal care use and infant health of low-educated unmarried women. Among unmarried women with less than 12 years of education, decreases in the welfare caseload were associated with less prenatal care and lower weight infants. Decreasing the welfare caseload over the 1990s was associated with a 2 percent decrease in first trimester care; a 10 percent increase in last trimester care; a 1 percent decrease in the number of prenatal care visits; and virtually no change in birth weight. Among unmarried women with 12 years of education, our estimates indicate similarly small effects. The relatively small effects of changes in welfare policy and welfare caseload on the prenatal care use and birth weight of low-educated women are consistent with the relatively small effect of welfare reform on health insurance.
In another study, Elizabeth Tarlov and I examine the effect of welfare reform on the health behaviors and health of low-educated women. 14 The motivation for this study is that many other single mothers who, in the absence of reform, would have entered welfare were deterred from doing so. The switch from subsidized household work (welfare) to paid employment, and other transitions experienced by women for whom welfare was no longer an option, may have affected health insurance, financial resources, time constraints, daily activities and responsibilities, and levels of psychological distress. All of these may have effects on women's health behaviors and health. Accordingly, we estimate the association between the welfare caseload and welfare policies and four health behaviors-smoking, binge drinking, diet, and exercise-and four self-reported measures of health-body mass and obesity, days in poor mental health, days in poor physical health, and general health status. The results of our study suggest that the decline in welfare participation as a result of welfare reform was associated with a decline in the incidence of binge drinking of 25 percent or more. Welfare reform does not appear to be related to other health behaviors such as smoking, diet, and exercise, or with other measures of health such as days in poor mental and physical health, body mass and obesity, and general health status, although there is some limited evidence that welfare reform was associated with a decrease in smoking prevalence.
* Kaestner is a Research Associates in the NBER's Program on Health Economics and the Economic Well-being of Children. He is also a Professor of Economics at the University of Illinois at Chicago.
1. Crowd out was first studied by NBER researchers David M. Cutler and Jonathan Gruber. See D. M. Cutler and J. Gruber, "Does Public Insurance Crowd Out Private Insurance?" NBER Working Paper No. 5082, April 1995, published in Quarterly Journal of Economics, May 1996, 111(2), pp.391-430.
5. See for example L. Dubay, T. Joyce, R. Kaestner, and G. Kenney, "Changes in Prenatal Care Timing and Low Birth Weight by Race and Socioeconomic Status: Implications for the Medicaid Expansions for Pregnant Women," Health Services Research, 2001, 36(2): pp.373-98, and R. Kaestner, A. Racine, and T. Joyce, "Did Recent Expansions in Medicaid Narrow Socioeconomic Differences in Hospitalization Rates of Infants?", Medical Care, 2000, 38(2): pp. 195-206.
6. R. Kaestner, "Health Insurance, the Quantity and Quality of Prenatal Care, and Infant Health," Inquiry, 1999, 36 (2): pp. 162-75.
7. R. Kaestner, T. Joyce, and A. Racine, "Does Publicly Provided Health Insurance Improve the Health of Low-income Children in the United States?" NBER Working Paper No. 6887, January 1999, published as "Medicaid Eligibility and the Incidence of Ambulatory Care Sensitive Hospitalizations for Children," Social Science and Medicine, 2000, 52: pp. 305-13.
8. T. Joyce and R. Kaestner, "The Effect of Expansions in Medicaid Income Eligibility on Abortion" Demography, 1996, 33(2): pp.181-92. Also T. Joyce, R. Kaestner, and F.Kwan, "Is Medicaid Pronatalist? Effects of the Medicaid Eligibility Expansions on Abortions and Births," Family Planning Perspectives, 1998, 30(3): pp. 108-13.
9. T. Joyce, R. Kaestner, and S. Korenman, "Welfare Reform and Non-marital Fertility in the 1990s: Evidence from Birth Records," NBER Working Paper No. 9406, December 2002, published in Advances in Economic Analysis and Policy, 2003, 3(1): Article Number 6. Also see N. Kaushal and R. Kaestner, "From Welfare to Work: Has Welfare Reform Worked?" Journal of Policy Analysis and Management, 2001, 20(4): pp. 740-61.
10. T. Joyce, R. Kaestner, S. Korenman, and S. Henshaw, "Family Cap Provisions and Changes in Births and Abortions," NBER Working Paper No. 10214, January 2004, published in Population Research and Policy Review 23, 2004, (5): pp. 475-511. Also see S. Korenman, T. Joyce, R. Kaestner, and J. Walper, "What Did the 'Illegitimacy Bonus' Reward?" NBER Working Paper No. 10699, August 2004, published in Topics in Economic Analysis and Policy, 2006, 6(1): Article Number 3.
11. R. Kaestner, S. Korenman, and J. O'Neill, "Has Welfare Reform Changed Teenage Behaviors?" NBER Working Paper No. 8932, May 2002, published in Journal of Policy Analysis and Management, Spring 2003, 22(2): pp. 225-48.
12. R. Kaestner and N. Kaushal, "The Effect of Welfare Reform on Health Insurance Coverage of Low-income Women and Children." NBER Working Paper No. 10033, October 2003, published in Journal of Health Economic, November 2003, 22(6): pp. 959-81.
14. R. Kaestner and E. Tarlov, "Changes in the Welfare Caseload and Health of Low-educated Mothers," NBER Working Paper No. 10034, October 2003, published in Journal of Policy Analysis and Management, Summer 2006, 25(3): pp. 623-44.