In 2011, Medicaid covered one quarter of all children and was the most common way that poor families paid for medical care (Kaiser Family Foundation 2013). Yet, almost 50 years after its implementation, evidence on Medicaid’s health effects is sparse, especially for its core population of disadvantaged children. Estimates for adults vary (Finkelstein et al. 2012, Sommers et al. 2012), but studies that examine eligibility expansions in the 1980s for infants and children find that Medicaid coverage reduces mortality (Currie and Gruber 1996a,b).1 My Job Market Paper develops a new research design to quantify the effects of Medicaid on the health of children at the time of the programme’s implementation (Goodman-Bacon 2014). The results suggest that by the end of the 1970s, Medicaid saved about 25,000 lives among nonwhite children, reduced their mortality rates by 8%, and narrowed the racial gap in mortality rates by 15%.
Who was eligible for Medicaid?
The empirical design exploits the sharp increases in public insurance eligibility that accompanied Medicaid’s state-by-state implementation between 1966 and 1970, and large variation across states in the size of those increases. Medicaid sought to usher in “a new era in medical care for low income families” (Department of Health, Education and Welfare 1967). It greatly expanded federal funding for public insurance, required coverage of basic medical services, and mandated eligibility for cash welfare recipients. The statutory requirement that welfare recipients be eligible for Medicaid had the biggest effect on nonwhite children, and it also induced substantial variation across states in their eligibility rates. Nationally, about 18% of nonwhite children received welfare and, therefore, suddenly gained insurance eligibility when their states began running Medicaid programmes (compared to 3% of white children and 1% of adults). The increases in public insurance eligibility and utilisation were much larger in states with higher shares of their population on welfare.
The dashed lines in Figure 1 show that prior to Medicaid, both high- and low-welfare states (split into two equally sized groups) had similar public insurance use among children, but that after Medicaid was put in place, higher-welfare states had about 5.5 percentage points more children on public insurance than in lower-welfare states. This difference is about half as large as the average level of Medicaid use –10% (solid line).2 Survey data suggest that public insurance was about three times higher for nonwhite children.
Figure 1 The share of children using public health insurance before and after Medicaid
Notes: The figure plots the share of children ages 0-19 who received medical services paid for by a means-tested public insurance programme in the years before and after states implemented Mediciad. Source: AFDC cases are from Health and Human Services Caseload Data 1960-1999 (HHS 2012); population data are from 1960 population estimates (Haines and ICPSR 2005), and the Survey of Epidemiological End Results (SEER 2009); data on public insurance use are collected from various editions of “Recipients of Medical Vendor Payments Under Public Assistance Programs” and “Medicaid State Tables” (DHEW 1963-1976).
These differences are unlikely to represent other factors that themselves influence health.
- First, cross-state differences in welfare participation predates Medicaid by at least several decades, suggesting that they are not due to contemporaneous factors in the 1960s that also determined health outcomes.
- Second, welfare rates are uncorrelated with a range of state characteristics, including pre-Medicaid mortality trends, suggesting that states with higher and lower welfare rates are not systematically different.
- Third, there is no evidence of differential expansions in higher- and lower-welfare states in other new programmes from the 1960s that have been shown to improve health: Food Stamps (Almond et al. 2011), Head Start (Ludwig and Miller 2007), Community Health Centers (Bailey and Goodman-Bacon 2013), or welfare itself.3
Thus, comparing changes in health before and after Medicaid between states with different welfare rates constitutes a test of whether or not public insurance affects the health of children.
Medicaid reduced mortality for nonwhite children
My analysis implements this test using a difference-in-difference design that compares child mortality in higher- and lower-welfare states before and after they implemented Medicaid.
- The results show that mortality trends were indistinguishable in the two groups before Medicaid, but diverged soon after Medicaid implementation for nonwhite children. These mortality differences persist for nine years and are only present for the group with the greatest exposure to Medicaid (nonwhite children) and not for groups who used the programme less (white children).
The results for nonwhite children are driven by mortality reductions in two especially vulnerable age groups: neonatal infants (under 28 days), and younger children (ages 1 to 4). I also find some auxiliary evidence that Medicaid was associated with increases in the probability that poor mothers gave birth in a hospital, which has a natural connection to the earliest (neonatal) deaths. For children, the mortality reductions are most likely due to reductions in infectious disease (such as pneumonia) that were common among poor nonwhite children, and easily treatable with antibiotics.
These results are the first to document that Medicaid’s implementation had important health effects. Between 1966 and 1979 Medicaid saved 25,000 lives among nonwhite children, reduced their aggregate mortality rate by 8%, and reduced the racial gap in mortality rates by 15%. Medicaid spent $38 billion (in 2012 dollars) on children through 1976, which suggests that the cost per life saved over this time period was about $1.9 million dollars. The comparable estimate for children from research on the 1980s eligibility expansions is about $1.7 million dollars (Currie and Gruber 1996a). Furthermore, the estimated effects from the 1980s are very large relative to the associated changes in insurance coverage, suggesting that they affected health through channels other than insurance coverage itself.4 To the extent that these channels represent unobserved costs, the benefits of Medicaid implementation may have actually come at significantly lower cost than in subsequent expansions.
The size and cost of the Medicaid programme has always generated controversy, most recently reflected in the reluctance of some states to accept Medicaid eligibility expansions under the Affordable Care Act (ACA), yet Medicaid’s health effects have been difficult to estimate. This paper provides new evidence that Medicaid had tremendous benefits in terms of mortality in its first 15 years, suggesting that public health insurance is a very valuable and effective programme for poor children. The eligibility expansions set to take place in some states under the ACA will primarily affect childless adults and so these results may not be directly relevant to that policy debate. Nevertheless, calls to curtail Medicaid spending, for instance by making it a block grant programme, have the potential to damage the health of poor children to the extent that they lead states to restrict eligibility among this group.
Almond, Douglas, Kenneth Y. Chay, and Michael Greenstone (2006), "Civil Rights, the War on Poverty, and Black-White Convergence in Infant Mortality in the Rural South and Mississippi." MIT Department of Economics Working Paper no. No. 07-04. doi: http://dx.doi.org/10.2139/ssrn.961021.
Almond, Douglas, Hilary W. Hoynes, and Diane Whitmore Schanzenbach (2011), "Inside the War On Poverty: The Impact of Food Stamps on Birth Outcomes." The Review of Economics and Statistics no. 93 (2):387-403. doi: 10.2307/23015943.
Bailey, Martha J., and Andrew J. Goodman-Bacon (2013), "The War on Poverty’s Experiment in Public Medicine: Community Health Centers and The Mortality of Older Americans." Working Paper, University of Michigan Department of Economics.
Currie, Janet, and Jonathan Gruber (1996a), "Health Insurance Eligibility, Utilization of Medical Care, and Child Health." The Quarterly Journal of Economics no. 111 (2):431-466. doi: 10.2307/2946684.
Currie, Janet, and Jonathan Gruber (1996b), "Saving Babies: The Efficacy and Cost of Recent Changes in the Medicaid Eligibility of Pregnant Women." Journal of Political Economy no. 104 (6):1263-1296. doi: 10.2307/2138939.
Department of Health, Education and Welfare (1967), Handbook of Public Assistance Administration Supplement D, Medical Assistance Programs Under Title XIX of the Social Security Act, Medical Services Administration Social and Rehabilitation Service. Washington, D.C.
Finkelstein, Amy (2007), "The Aggregate Effects of Health Insurance: Evidence from the Introduction of Medicare." The Quarterly Journal of Economics no. 122 (1):1-37.
Finkelstein, Amy, Sarah Taubman, Bill Wright, Mira Bernstein, Jonathan Gruber, Joseph P. Newhouse, Heidi Allen, Katherine Baicker, and Group Oregon Health Study (2012), "The Oregon Health Insurance Experiment: Evidence from the First Year." The Quarterly Journal of Economics no. 127 (3):1057-1106.
Goodman-Bacon, Andrew J (2014), "Public Insurance and Mortality: Evidence from Medicaid Implementation." Working Paper, University of Michigan Department of Economics.
Kaiser Family Foundation (2013), The Medicaid Program at a Glance.
Leininger, Lindsey , Helen Levy, and Diane Schanzenbach (2012), "Consequences of SCHIP Expansions for Household Well-Being." Forum for Health Economics & Policy no. 13 (1).
Ludwig, Jens, and Douglas L. Miller (2007), "Does Head Start Improve Children's Life Chances? Evidence from a Regression Discontinuity Design." The Quarterly Journal of Economics no. 122 (1):159-208.
Sommers, Benjamin D., Katherine Baicker, and Arnold M. Epstein (2012), "Mortality and Access to Care among Adults after State Medicaid Expansions." New England Journal of Medicine no. 367 (11):1025-1034. doi: 10.1056/NEJMsa1202099.
1 Using a randomized expansion of eligibility, the Oregon Health Insurance Experiment (Finkelstein et al. 2012) finds no evidence that Medicaid reduced the one-year mortality rate of poor adults (although their sample sizes were only sufficient to detect very large effects). Sommers et al. (2012) do find large mortality effects using legislative expansions of eligibility in three states in the early 2000s.
2 In contrast, the 1980s expansions phased in gradually, were associated with about a 7 percentage point increase in Medicaid coverage (Currie and Gruber 1996b), and often crowded out private insurance (Cutler and Gruber 1996).
3 There is strong evidence that the desegregation of southern hospitals in response to the Civil Rights Act and the financial incentives provided by Medicare strongly reduced post-neonatal infant mortality in the South (Almond et al. 2006). My results are robust to excluding the south entirely, which suggests that my estimates for Medicaid are not picking up the effect of desegregation.
4 If Medicaid expansions increase providers’ revenue then they may invest in new life-saving technologies (Finkelstein 2007), or if families who switch from private to public coverage save money on premiums, copays and deductibles then their children’s health may improve even though their insurance status is unchanged (Leininger et al. 2012).