The Affordable Care Act Increased Medicaid Coverage Among Former Foster Youth

Abstract

Youth aging out of the foster care system are a vulnerable population. When in foster care, youth are eligible for their state’s Medicaid program, but lose eligibility on emancipation. The Affordable Care Act (ACA) has the potential to address some of the health care needs of former foster youth through the Medicaid expansion to low-income adults and by extending Medicaid eligibility up to age twenty-six for former foster youth. Using the 2011–18 National Youth in Transition Database, we found that Medicaid expansions increased Medicaid coverage among former foster youth by approximately 10.1 percentage points, and the age extension increased coverage by 3.4 percentage points. There is suggestive evidence of positive spillovers for both policies. Our findings imply that the ACA improved Medicaid coverage among former foster youth, with larger effects from Medicaid expansions. Modest effects of the Medicaid age extension may imply a need to revise enrollment, recertification, outreach, and eligibility determination processes to further increase Medicaid coverage among former foster youth.

In 2018 there were nearly 440,000 children in the US foster care system. Each year about 57 percent of these children (~250,000) exit the system, most of whom return to their families through reunification or establish another permanent connection through adoption or legal guardianship.¹ Some of these children, however, become too old for child welfare services. This historically occurs at age eighteen, when they become legal adults; approximately 18,000 children age out of foster care annually.¹

Youth aging out of foster care experience substantial challenges to becoming independent. Relative to their peers in the general population, former foster youth are more likely to have a disability that limits daily activities, and they have higher rates of developmental, medical, and behavioral disorders.² Former foster youth also have lower education attainment, are less likely to be employed, earn lower wages, and are more likely to be teen parents, incarcerated, and homeless.²

When in foster care, youth are categorically eligible for their state’s Medicaid program. Before the Affordable Care Act (ACA) became law, many youth lost their Medicaid eligibility on turning nineteen.³ In the thirty states that exercised the Chafee option through the Foster Care Independence Act of 1999, youth in care at age eighteen could preserve Medicaid eligibility and other services up to ages twenty, twenty-one, or twenty-two, depending on their income, resources, and other requirements established by state law.⁴ States opted for the Chafee option between 2000 and 2010.

Beginning January 1, 2014, the ACA mandated all states to provide full Medicaid coverage to foster youth aging out of the system until their twenty-sixth birthday, regardless of their income or resources. This provision (henceforth referred to as Medicaid age extension) applies to individuals who are enrolled in both Medicaid and foster care on attaining either age eighteen or the age in which the state terminates federal foster care assistance under Title IV-E of the Social Security Act.^3,4 Former foster youth are eligible for Medicaid only in the state in which they age out, although thirteen states have opted to cover youth who aged out in other states.

Another ACA provision offering former foster youth a pathway to Medicaid eligibility is the Medicaid expansion to low-income adults ages 18–64 with household incomes up to 133 percent of the federal poverty level. This provision (henceforth referred to as Medicaid expansion), however, was not implemented by all states, as adoption of the expansion became optional pursuant a 2012 Supreme Court decision.⁵ Former foster youth who are eligible for Medicaid through the ACA age extension can receive full Medicaid benefits, as opposed to the “alternative benefit plan” that states choosing Medicaid expansion may define for newly eligible low-income adults.⁶ Eligibility under the Medicaid age extension takes precedence over eligibility under Medicaid expansions. Thus, foster youth meeting eligibility requirements through both pathways are enrolled under the Medicaid age extension.⁷ Both ACA provisions were expected to increase health care access for former foster youth, but the extent to which this has materialized remains unknown.

Previous studies of ACA Medicaid expansions have documented increases in Medicaid coverage, health care use, and improved health and socioeconomic outcomes.^8–11 Although there are no studies of the Medicaid age extension, this provision is similar to the ACA’s young adult mandate, which requires parents’ private health insurance plans to cover adult dependents up to age twenty-six. Studies suggest that the young adult mandate increased private health insurance coverage¹² and health care use.^13,14 These studies, however, may not generalize to the former foster youth population, given their heightened health, social, and behavioral needs and differences between Medicaid and private health insurance.

We generate the first estimates of the impact of the ACA’s former foster youth age extension and Medicaid expansions on foster youth, using longitudinal data from the 2011–18 National Youth in Transition Database. Our primary outcomes include Medicaid and other health insurance coverage. In secondary analyses, we explore potential spillover benefits of these policies on socioeconomic outcomes. We used time, age, and state variation generated by the policy choices of the states in which former foster youth emancipated, including the decision to expand Medicaid under the ACA and the decision to extend Medicaid coverage before the ACA under the Chafee option.

Understanding the impact of health care reform on marginalized populations such as former foster youth is essential for improving their well-being and reducing persistent inequalities resulting from family background. Our study provides policy makers with important and timely information regarding the implications of the ACA on a vulnerable yet understudied population for whom health insurance may be especially critical.

Study Data And Methods

Data

The National Youth in Transition Database is a cohort-based, longitudinal survey that collects demographic, health care, and socioeconomic information on foster youth who turn seventeen in certain years.¹⁵ All fifty states; Washington, D.C.; and Puerto Rico submit National Youth in Transition Database data to the Children’s Bureau in the Department of Health and Human Services’ Administration for Children and Families. Starting with the 2011 federal fiscal year, and every three years thereafter, all youth who reach their seventeenth birthday and are in foster care within the forty-five-day period after their birthday are eligible for the National Youth in Transition Database. Youth in each cohort are interviewed in three waves. The first wave is conducted at age seventeen. Follow-ups are conducted at ages nineteen and twenty-one.

We analyzed three waves of the National Youth in Transition Database for two cohorts, where the first cohort of seventeen-year-olds was initially surveyed in fiscal year 2011, with follow-up in FY 2013 and FY 2015. The second cohort was first surveyed in FY 2014, with follow-up in FY 2016 and FY 2018. We limited the sample to individuals with nonmissing data for at least two of the three waves. The analytic sample was composed of 16,946 youth (8,120 in cohort 1 and 8,826 in cohort 2), yielding 45,173 observations.

Measures

Outcome Variables:

The primary outcomes consisted of two separate indicator variables of health insurance coverage: Medicaid and other health insurance. Other health insurance may pay for all or part of medical care costs and reflect group coverage offered by employers, schools or associations, individual health plans, self-employed plans, or parents’ insurance plans. This also could include access to free health care through a college, Indian Health Service, or other source. Secondary outcome variables consisted of socioeconomic measures indicating whether the youth was employed (either part-time or full-time), had biological children, was currently enrolled in school (high school, GED, vocational, college), experienced homelessness (no regular or adequate place to live), received a referral for substance use disorder treatment (including self-referral or referral by a social worker, school staff, medical personnel, foster parent, or other adult), or was incarcerated (confined in a jail, prison, correctional facility, or detention facility). Employment and current school enrollment were measured contemporaneously, whereas other measures were equal to one if the individual responded affirmatively for “ever” at age seventeen, and “since the previous wave” at ages nineteen and twenty-one. We focused on all of these secondary outcomes because they are sensitive to health insurance eligibility.^{12,14,16–20}

Policy Variables:

We considered two main policy variables. First was whether a state expanded Medicaid to low-income adults under the ACA. This binary variable was equal to one if a youth’s state of residence expanded Medicaid at the time of the interview and zero otherwise. This variable accounts for states that expanded early (before 2014) and late (after 2014). The second policy variable was whether a state chose not to extend Medicaid coverage beyond age eighteen by not exercising the Chafee option through the Foster Care Independence Act of 1999. No states exercised this option during the study period. Therefore, it is a constant binary variable representing one if a state was a “non-Chafee” state before the ACA and zero otherwise. Non-Chafee states were the treatment group, as they should be most strongly affected by the ACA’s Medicaid age extension. Importantly, the group of states that did not exercise the Chafee option was different from the group of states that expanded Medicaid under the ACA (see online appendix exhibit 1).²¹

Control Variables:

Because we had longitudinal data, we were able to adjust for time-invariant characteristics of a person (that is, sex, race/ethnicity) by including individual-level fixed effects. In addition, we included several state-level variables that may be correlated with both the ACA policy variables and outcomes. These include the unemployment rate, the maximum Temporary Assistance for Needy Families and Supplemental Nutrition Assistance Program combined benefit for a family of three, whether a state’s governor was a Democrat, a state’s Earned Income Tax Credit rate and whether it was refundable, and the state’s real minimum wage. Finally, the Fostering Connections to Success and Increasing Adoptions Act of 2008 allowed states to extend foster care and other services beyond age eighteen. Some states opted into receiving federal funds for this extension, but federally funded youth must meet at least one of five established employment or education conditions. Alternatively, state-funded extended foster care programs do not need to meet federal requirements. For these reasons, we also included two binary measures of whether states extended their foster care program through federal funding or their own funding.²² We also adjusted for the opposite policy in each analysis. For example, in analyses of the Medicaid age extension, we controlled for Medicaid expansions.

Statistical Analysis

The first analysis examined the impact of the ACA’s Medicaid expansions. We conducted a difference-in-differences-differences analysis by comparing youth in Medicaid expansion states with those in nonexpansion states at ages nineteen and twenty-one relative to age seventeen, before and after implementation. Specifically, our main policy variable was an interaction of a binary variable measuring whether a former foster youth’s state had expanded Medicaid during the month and year in which they were interviewed and whether they were nineteen or older. We regressed outcomes on this interaction, the state-level covariates described above, and individual and year fixed-effects, using ordinary least squares. Additional details regarding the data, policy variation, and statistical analysis are provided in appendixes A and B.²¹

The second analysis evaluated the impact of the ACA’s Medicaid age extension. We conducted a separate difference-in-differences-differences approach by comparing outcomes among youth in cohort 2 (eligible for Medicaid age extension at emancipation) with youth in cohort 1 (ineligible for Medicaid age extension at emancipation), ages nineteen and twenty-one (the ages at which ACA’s Medicaid age extension coverage confers) relative to age seventeen, and by whether a youth resided in a non-Chafee state (and therefore became eligible for Medicaid when the ACA Medicaid age extension became effective). Specifically, our main policy variable was an interaction of three binary variables each equaling one if a youth was in cohort 2, age nineteen or twenty-one, and emancipated in a non-Chafee state.

This difference-in-differences-differences approach is important to disentangle the effects of the Medicaid age extension from both the Medicaid expansions and the Chafee program. In particular, the Medicaid expansions occurred at the same time as the Medicaid age extension. In addition, Medicaid coverage gained through the ACA Medicaid age extension for former foster youth was a close substitute to Medicaid coverage gained through the Chafee option, in terms of both eligibility requirements and benefits. This was not necessarily the case for Medicaid gained through ACA Medicaid expansions, which did not require foster children to “age out” or be enrolled in Medicaid at the time of “age out” for eligibility, and offered alternative benefit plans. Therefore, from a policy perspective, it is important to discover whether the ACA effects on former foster youth vary on the basis of income-based eligibility or categorical eligibility.

As a result of attrition and nonresponse in the National Youth in Transition Database, we also performed a bounding exercise, whereby we estimated the upper and lower bounds for all missing observations. Specifically, by assuming all missing observations responded both affirmatively and negatively, we created lower and upper bound estimates that account for attrition. We then compared our main estimates with these upper and lower bound estimates.

Finally, we tested whether the effects of the ACA’s age extension vary by adult age. Members of cohort 1 were eligible for Medicaid under the ACA’s Medicaid age extension at age twenty-one, but not at age nineteen, whereas members of cohort 2 were eligible at both ages nineteen and twenty-one as a result of the policy. To identify potential differences, we tested two different subsamples: respondents at ages seventeen and nineteen and respondents at ages seventeen and twenty-one.

Limitations

Our study has some limitations. First, the National Youth in Transition Database is voluntary and is therefore subject to nonresponse issues that affect sample selection (see appendix B).²¹ The response rate at wave 1 for cohort 1 is 53 percent of the baseline population and 69 percent for cohort 2; nonresponse in wave 1 is an exclusion criterion for follow-up. Nonresponse in wave 1 does not necessarily mean that estimates are biased, but raises an issue of external validity. If youth who respond at age seventeen are systematically different from youth who do not respond at age seventeen, and there is treatment effect heterogeneity between included and excluded populations, then estimates may not generalize to the baseline population. In appendix exhibit 3, we characterize the demographic characteristics of excluded and included populations.²¹ Another reason for nonresponse is that in some states, a subset of wave 1 respondents was randomly not chosen for follow-up, further reducing who is observed longitudinally. In addition, the National Youth in Transition Database experiences attrition. We addressed these limitations by first requiring individuals to be observed in at least two of the three waves and with individual fixed effects. Descriptive statistics show that despite some degree of attrition, the demographic composition of each cohort remains relatively stable over time. We also performed the bounding exercise previously described.

Other potential limitations include social desirability bias and reporting errors. States choose how to administer the survey (for example, in person, phone, Internet), which may influence reporting errors, as respondents may report differently to a person compared with how they complete an online form. In addition, concerns of social desirability may lead to underreporting of stigmatizing experiences (for example, substance use disorder treatment referrals, incarceration). Moreover, we cannot assess the impact of the ACA among youth ages twenty-two and older, as the National Youth in Transition Database ends at age twenty-one.

Last, we were unable to perfectly assess eligibility for either ACA coverage provision because of a lack of information on foster care and Medicaid status at age eighteen and on other eligibility criteria (that is, income). As such, our findings should be interpreted as capturing the impact of the ACA on youth who were in foster care at age seventeen. Despite these limitations, the National Youth in Transition Database data remain the only data available to answer questions about health insurance coverage and socioeconomic outcomes of youth aging out of foster care nationwide.

Study Results

Exhibit 1 shows descriptive statistics of the analytic sample by wave and cohort. The cohorts share similar demographic characteristics. Males were more likely to drop out of the sample by age twenty-one. Both cohorts experience nearly universal Medicaid coverage at age seventeen (92 percent and 93 percent, respectively). Medicaid coverage drops more substantially between ages seventeen and nineteen than between ages nineteen and twenty-one. Youth in cohort 2 were more likely to be covered by Medicaid as adults than youth in cohort 1. Our analysis of secondary outcomes found that youth in cohort 2 are less likely to have a child, more likely to be employed, and less likely to be incarcerated. They also experienced a better macroeconomic environment as young adults, as unemployment rates were lower in 2014 than in 2011.

Exhibit 1:

Descriptive statistics of the sample

	Cohort 1			Cohort 2
	Wave 1 (age 17)	Wave 2 (age 19)	Wave 3 (age 21)	Wave 1 (age 17)	Wave 2 (age 19)	Wave 3 (age 21)
Individual covariates (%)
Female	53	54	56	54	54	57
Male	47	46	44	46	46	43
White	58	59	59	62	62	62
Black	34	34	36	34	34	34
Hispanic	17	17	17	22	23	23
Outcomes (%)
Medicaid	92	78	72	93	84	77
Other health insurance	19	16	17	17	16	18
Had a child	7	12	28	5	11	24
Employed	14	34	51	15	41	58
Currently enrolled in school	96	56	32	94	54	30
Homeless	17	20	29	17	21	29
Substance use disorder referral	26	15	11	26	15	11
Incarcerated	33	22	21	29	20	19
State covariates
Did not extend Medicaid under the Chafee option (%)	26	24	26	26	26	25
ACA Medicaid expansion (%)	24	29	60	58	69	69
Unemployment rate (%)	9	7	6	7	5	4
Maximum TANF and SNAP benefit for a family of three ($)	974	978	979	985	1,013	1,023
Governor is a Democrat (%)	52	49	49	52	46	47
State EITC rate (%)	7	7	7	7	21	26
State EITC refundable (%)	39	41	38	41	55	60
State minimum wage ($)	8.24	8.03	8.35	8.21	8.67	8.83
Received federal funding to extend foster care under the Fostering Connections Act (%)	21	47	53	59	65	67
Extended foster care without federal funding under the Fostering Connections Act (%)	10	23	32	24	31	35
N	8120	7084	6388	8826	7749	7006

SOURCE Authors’ calculations using the 2011 to 2018 National Youth in Transition Database, Cohort 1-2. NOTES The National Youth in Transition Database captures health insurance, socioeconomic, and demographic outcomes at ages 17 (wave 1), 19 (wave 2), and 21 (wave 3) on foster youth who turned 17 in 2011 (cohort 1) and in 2014 (cohort 2). N = 45,173. Medicaid and other health insurance were the primary outcomes in our analysis. Socioeconomic outcomes (having a child, being employed, enrolled in school, homeless, incarcerated, or having a substance use disorder referral) were secondary outcomes. Secondary outcomes are exploratory and should be interpreted with caution. ACA is Affordable Care Act. TANF is Temporary Assistance for Needy Families. SNAP is Supplemental Nutrition Assistance Program. EITC is Earned Income Tax Credit. Fostering Connections Act is the Fostering Connections to Success and Increasing Adoptions Act of 2008.

Exhibits 2 and 3 show trends in Medicaid coverage by age, cohort, and state policy status. Exhibit 2 documents that for both cohorts 1 and 2, youth in states that expanded Medicaid had higher Medicaid coverage at ages nineteen and twenty-one than youth in nonexpansion states. Exhibit 3 shows differences in Medicaid coverage across Chafee and non-Chafee states. Similarly, regardless of cohort, youth in Chafee states were more likely to be covered by Medicaid at ages nineteen and twenty-one than youth in non-Chafee states.

Exhibit 2:

Trends in Medicaid coverage of former foster youth by age, cohort, and state Affordable Care Act Medicaid expansion status, 2011–18

SOURCE Authors’ calculations using the 2011–18 National Youth in Transition Database, cohorts 1–2. NOTES The National Youth in Transition Database captures outcomes at ages seventeen (wave 1), nineteen (wave 2), and twenty-one (wave 3) on foster youth who turned seventeen in 2011 (cohort 1) and 2014 (cohort 2). N = 45,173. States that opted to expand Medicaid under the Affordable Care Act expanded Medicaid eligibility to adults ages 18–64 with incomes up to 133% of the federal poverty line.

Exhibit 3:

Trends in Medicaid coverage of former foster youth by age, cohort, and pre–Affordable Care Act state policy on age extension, 2011–18

SOURCE Authors’ calculations using the 2011–18 National Youth in Transition Database, cohorts 1–2. NOTES National Youth in Transition Database captures outcomes at ages seventeen (wave 1), nineteen (wave 2), and twenty-one (wave 3) on foster youth who turned seventeen in 2011 (cohort 1) and in 2014 (cohort 2). N = 45,173. The Affordable Care Act extended Medicaid eligibility to foster youth until their twenty-sixth birthday. Chafee states are the thirty states that, before the Affordable Care Act’s enactment, adopted the Chafee option (pursuant to the Foster Care Independence Act of 1999) allowing former foster youth to preserve Medicaid eligibility up to ages twenty, twenty-one, or twenty-two, depending on the state. Non-Chafee states are those that did not adopt the Chafee option.

Exhibit 4 displays difference-in-differences-differences estimates of the impact of the ACA Medicaid expansions and the ACA Medicaid age extension on health insurance coverage and socioeconomic outcomes among former foster youth. Full regression results are in appendix exhibits 4 and 5.²¹

Exhibit 4:

Effects of Affordable Care Act Medicaid low-income adult expansion and age extension on Medicaid, other health insurance, and socioeconomic outcomes for former foster youth, 2011–18.

SOURCE Authors’ calculations using the 2011–18 National Youth in Transition Database, cohorts 1–2. NOTES N = 45,173. Medicaid and other health insurance were the primary outcomes in our analysis. Socioeconomic outcomes (having a child or being employed, enrolled in school, homeless, incarcerated, or having a substance use disorder referral) were secondary outcomes. Secondary outcomes are exploratory and should be interpreted with caution. Lines represent the 95% confidence intervals.

Medicaid expansions increased Medicaid coverage by approximately 10.1 percentage points (p < 0.01; appendix exhibit 4). Among cohort 1, between 72 and 78 percent of youth were covered at ages nineteen and twenty-one (see exhibit 1). This increase of 10.1 percentage points translates to an increase of roughly 13 to 14 percent. Other health insurance decreased by 3 percentage points (p < 0.01), or about 18 percent. Our exploratory analysis of secondary outcomes found that Medicaid expansions increased the likelihood of having a child by 1.4 percentage points (p < 0.05) and of being referred to substance use disorder treatment by 2.4 percentage points (p < 0.01). We find no statistically significant association between Medicaid expansions and whether a youth was employed, in school, homeless, or incarcerated (appendix exhibit 4).

The Medicaid age extension increased Medicaid coverage among former foster youth by 3.4 percentage points (p < 0.05; appendix exhibit 5), or about 5 percent. In contrast to Medicaid expansions, we do not detect significant effects of the Medicaid age extension on other health insurance coverage, although the direction and magnitude of the coefficient is similar to that of Medicaid expansions. We also find that the ACA age extension increased employment by approximately 4.0 percentage points (p < 0.05) and reduced incarceration of 4.7 percentage points (p < 0.01). We find no significant effect for the other secondary outcomes.

We next examine the sensitivity of the main results to attrition and nonresponse in the survey. The results from the bounding exercise described earlier for both policies are provided in appendix D.²¹ Appendix exhibit 6 shows that even under a scenario in which the true response to missing values would have been either all “yes” (that is, equal to 1) or all “no” (that is, equal to 0), having all observations in the sample for all waves would still show that Medicaid expansions increased Medicaid coverage.²¹ This finding implies the results are not due to sample attrition. The spillover effects onto secondary socioeconomic outcomes, however, are sensitive to attrition. Appendix exhibit 7 reports the findings from the bounding exercise for the Medicaid age extension.²¹ In contrast, these results are sensitive to attrition.

Finally, we test whether adult age matters for the effects of the ACA Medicaid age extension. Appendix exhibit 8 shows more statistically precise estimates for Medicaid coverage among youth at age nineteen than those at age twenty-one, but the magnitude of the estimated coefficients is comparable.²¹

Discussion

The ACA’s Medicaid expansion to low-income adults and Medicaid age extension to former foster youth had the potential to address the health care needs of foster youth, a vulnerable population at high risk for physical, mental, and behavioral disorders, homelessness, substance use, and unemployment. We generated the first estimates of the effects of these ACA provisions on key health insurance and socioeconomic outcomes of foster youth. Using longitudinal data and causal empirical methods, we documented several findings.

First, we found that ACA’s Medicaid expansions increased Medicaid coverage by 10.1 percentage points among former foster youth, whereas the ACA Medicaid age extension increased Medicaid coverage in this population by 3.4 percentage points. Our findings of the impact of ACA Medicaid expansions on Medicaid coverage among former foster youth coincide with previous studies estimating this effect in the broader newly eligible population. In particular, other studies have documented increases in Medicaid coverage and produced estimates of comparable magnitude.^8,23–26 Similarly, our findings of the impact of the ACA Medicaid age extension on Medicaid coverage align with previous studies examining the impact of the ACA’s young adult mandate requiring parents’ private health insurance plans to cover dependents up to age twenty-six, which have shown increases in private health insurance coverage.^12,14,27,28

Notably, we find larger treatment effects from Medicaid expansions than from the Medicaid age extension. At least three factors may explain these findings. First, eligibility and verification requirements attached to each policy vary, which, in turn, may affect the number of youth newly eligible through each pathway. In particular, the Medicaid age extension requires states to cover individuals younger than twenty-six who are both under the responsibility of the state or tribe on attaining either age eighteen or the age in which the state terminates federal foster care assistance under Title IV-E and are enrolled in Medicaid at that age. Although Medicaid expansions do impose income thresholds, there are no foster care requirements nor a need for continued contact with child welfare agencies for verification of these requirements. The majority of former foster youth are likely eligible for Medicaid through this income-based provision.²

A second factor could be differences in awareness of eligibility and outreach. Former foster youth and the systems that serve them may have greater awareness of Medicaid expansions, given the considerable outreach and media attention that this provision received.²⁹

The third factor is data limitations. The ACA age extension offers Medicaid coverage to former foster youth up to age twenty-six, and the National Youth in Transition Database captures outcomes up to age twenty-one, thus restricting our ability to evaluate policy impacts on youth ages 22–26.⁴ Given alternative pathways for Medicaid eligibility and extended foster care services for foster youth ages 18–21, depending on the state and circumstances, even in non-Chafee states, it is possible that the Medicaid age extension would have greater effects on youth at later ages (for example, ages 22–26).^4,22 External validity issues from nonresponse in wave 1, as well as our inability to perfectly identify eligible youth for either policy, may also generate these differential findings. Future studies should consider novel approaches such as linkage of state child welfare and Medicaid data to overcome some of our data limitations and help elucidate differential impacts of Medicaid expansions and the Medicaid age extension.

A second key finding is a 3.0-percentage-point decrease in other health insurance after the Medicaid expansions, but no significant changes after the Medicaid age extension. Because our measure of other health insurance captures a variety of plans and other payment sources that may partially or fully cover medical care costs, it is unclear whether our findings are consistent with crowd-out or consolidation of services covered under one payer (that is, reductions in certain carve-out arrangements or in supplemental health insurance).

As for secondary socioeconomic outcomes, we found increases in substance use disorder treatment referrals after Medicaid expansions, which is consistent with prior research on the broader adult population.^16–18 We also found some evidence of increases in employment and reductions in incarceration after the Medicaid age extension. Together, estimates suggest that both policies may have induced spillovers benefiting former foster youth. However, as socioeconomic outcomes are more difficult to link to the impact of expanded Medicaid eligibility, we consider these estimates exploratory and interpret them with caution.

Our findings have important policy implications. Modest effects of the Medicaid age extension may imply a need to revise enrollment, recertification, outreach, and eligibility determination processes to further increase Medicaid coverage among former foster youth. In some states, Medicaid enrollment of former foster youth is not automatic, requiring awareness of the policy, a new application, and proof of eligibility.³⁰ Moreover, with the exception of thirteen states, the Medicaid age extension only applies to youth in the state in which they aged out of care.⁷ This is problematic, as former foster youth tend to be more mobile than their peers,³¹ and moving across states may mean losing access to health care coverage. States should consider a streamlined, automated approach for enrolling foster youth in Medicaid on aging out, and for keeping them enrolled through age twenty-six. This approach should place minimal burden on youth and should not be contingent on their state of residence.³

Conclusion

We show that the ACA increased Medicaid coverage for former foster youth, with stronger effects from Medicaid expansions than from the Medicaid age extension. These findings may indicate the value in broad eligibility (for example, based on income) over targeted eligibility (for example, based on foster care status) and have important implications for future policy and practice guidance. Given the heightened challenges facing former foster youth, streamlining the process of getting and staying enrolled in Medicaid on aging out of foster care should be an important policy goal.

Biographies

BIOS for 2021-00073 (Bullinger)

Lindsey Rose Bullinger (lrbullin@gatech.edu) is an Assistant Professor in the School of Public Policy, Georgia Tech, in Atlanta, Georgia.

Angélica Meinhofer is an Assistant Professor in the Department of Population Health Sciences, Weill Cornell Medicine, in New York, New York.