Medicaid Expansions and Participation in Supplemental Security Income by Noncitizens

Abstract

Objectives. To estimate the effect of Medicaid expansion on noncitizens’ and citizens’ participation in the Supplemental Security Income (SSI) program. The Affordable Care Act (ACA) expanded Medicaid eligibility to cover low-income nonelderly adults without children, thus delinking their Medicaid participation from participation in the SSI program.

Methods. Using data from the Social Security Administration for 2009 through 2018 (n = 1020 state-year observations) and the Current Population Survey for 2009 through 2019 (n = 78 776 respondents), we employed a difference-in-differences approach comparing SSI participation rates in US states that adopted Medicaid expansion with participation rates in nonexpansion states before and after ACA implementation.

Results. Medicaid expansion reduced the SSI (disability) participation of nonelderly noncitizens by 12% and of nonelderly citizens by 2%. Estimates remained robust with administrative and survey data.

Conclusions. Medicaid expansion caused a substantially larger decline in the SSI participation of noncitizens, who face more restrictive SSI eligibility criteria, than of citizens. Our estimates suggest an annual savings of $619 million in the federal SSI cost because of the decline in SSI participation among noncitizens and citizens.

Immigrants’ eligibility for and participation in Medicaid are highly controversial. They have evoked concerns relating to the fiscal consequences of providing public health insurance to noncitizens, as well as policy actions to classify certain groups of immigrants as public charges if deemed to become eligible for Medicaid.^1–3 Noncitizen’s eligibility for Medicaid, however, may lower their participation in other means-tested programs, which will reduce Medicaid’s net fiscal impact. In this article, we studied the impact of Medicaid eligibility on participation in the Supplemental Security Income (SSI) program by noncitizens and citizens.

The Affordable Care Act (ACA) expanded Medicaid eligibility to cover low-income nonelderly adults without children. Prior to the ACA, their Medicaid eligibility was linked to participation in the SSI program, which required an arduous and lengthy disability application process. People with disabilities, who have higher levels of medical need, were effectively locked into poverty to maintain Medicaid eligibility because of the low-income and assets limits of SSI.^4–7

The SSI eligibility criteria are considerably more restrictive for noncitizens who face additional eligibility requirements, including work experience in the United States of at least 40 quarters.^8,9 Thus, if Medicaid expansions under ACA caused lower SSI participation, the decline should be much higher among noncitizens. There is, however, no systematic research on the effect of Medicaid expansions on the SSI participation of noncitizens, a highly vulnerable group with relatively low incomes that also experienced a substantial rise in Medicaid eligibility after the ACA expansions. Previous research documents that immigrants in the United States are more likely than similarly placed natives to be low income and to work in jobs that do not offer employer-sponsored insurance. From 2011 through 2013 (the 3 years prior to ACA implementation), foreign-born adults aged 19 to 64 years with incomes less than 300% of the federal poverty line had an uninsurance rate of 49% compared with a 28% uninsurance rate among similar US-born adults; of those with incomes less than 150% of the federal poverty line, the uninsurance rate was 55% among the foreign-born and 33% among the US-born.¹⁰

Two previous studies estimated the effect of Medicaid expansions on SSI use among all US residents and found a small decline in SSI participation associated with Medicaid expansions.^11,12 One study used 1 year of post-ACA data and the other used 2 years of post-ACA data. Arguably, these studies did not have sufficient post-ACA expansion data to yield the full impact of Medicaid expansions on SSI participation. In this study, we used 6 years of post-ACA data and estimated change in SSI participation among noncitizens and citizens separately. Specifically, we compared SSI participation in states that adopted ACA Medicaid expansion before and after ACA implementation and compared it with the corresponding change in nonexpansion states to study how Medicaid expansions affected SSI participation among noncitizens and citizens.

Medicaid and SSI eligibility are restricted to citizens, lawful permanent residents, and certain other noncitizens, a category that includes refugees and asylees. Undocumented immigrants and temporary residents are ineligible for SSI and Medicaid. Further, the ACA, following the guiding principle of the Personal Responsibility and Work Opportunity Reconciliation Act of 1996, restricted Medicaid eligibility to lawful permanent residents who have been in the United States for more than 5 years, but a few states have exemptions to cover certain excluded populations.

The postulated mechanism we studied is the ACA’s introduction of a pathway to Medicaid eligibility based on income alone (Medicaid) versus a pathway that combines a restrictive employment, financial means, and disability test (SSI). Because noncitizens face a more restrictive process of SSI eligibility, we hypothesized that the Medicaid expansions would cause a larger reduction in their SSI participation.

We performed empirical analyses with 2 sets of data: the reports of the Social Security Administration on SSI¹³ for 2009 through 2018 and the Annual Social and Economic Supplement of the Current Population Survey (CPS)¹⁰ data for 2009 through 2019, covering a period of 5 years before and 6 years after implementation of the ACA.

Our analysis of the Social Security Administration data provided evidence that Medicaid expansions indeed lowered SSI participation among noncitizens by a much larger proportion than among citizens (US-born and naturalized). We found that between 2013 and 2018, in Medicaid expansion states, the proportion of noncitizens receiving SSI fell 22%; among citizens the decline in SSI participation was a mere 4%. The declines in SSI participation of citizens and noncitizens in Medicaid nonexpansion states were much smaller in magnitude. Our back-of-the-envelope estimates suggest an annual savings of $619 million in SSI costs because of the decline in SSI participation among noncitizens and citizens in Medicaid expansion states.

METHODS

Our primary analysis was based on data from Social Security Administration reports. These reports include counts of SSI recipients by age (0–17, 18–64, and ≥ 65 years), citizenship status (citizens and noncitizens), state of residence, and year. We used the SSI data for nonelderly adults (18–64 years), yielding 510 observations for noncitizens and citizens each. Our outcome variable was SSI participation rate (number of SSI recipients divided by total population of each group). We computed nonelderly adult population size by citizenship status, state, and year from the American Community Surveys of the US Census Bureau. The advantage of using administrative data is that our analysis was not affected by underreporting of benefit receipt in government-administered surveys, including the CPS, as reported in recent studies.¹⁴

In supplementary analysis, we tested our findings from the administrative data using the CPS survey data. The CPS includes data on respondents’ age, number of children, and income-to-poverty level ratio, which we used to restrict the sample to nonelderly childless adults with incomes below the federal poverty level. The CPS data are rich in individual characteristics, such as citizenship status, year of immigration, gender, educational attainment, marital status, household size, and state of residence, which were used as control variables or to stratify the sample. In some analyses, we used information on whether the respondent had a serious physical or cognitive limitation to further restrict the sample to adults reporting having a disability. We excluded noncitizens who immigrated to the United States less than 5 years prior to the CPS interview, because immigrants who have lived as permanent residents in the United States for less than 5 are ineligible for SSI. This yielded a sample of 73 001 citizens and 5775 noncitizens. Our outcome variable was a binary indicator for receiving any SSI income payments in the previous calendar year.

We used a difference-in-differences¹⁵ method to compare SSI recipiency in Medicaid expansion states and in states that did not expand, before and after 2014. Medicaid expansion states were Alaska, Arizona, Arkansas, California, Colorado, Connecticut, Delaware, the District of Columbia, Hawaii, Illinois, Indiana, Iowa, Kentucky, Louisiana, Maryland, Massachusetts, Michigan, Minnesota, Montana, Nevada, New Hampshire, New Jersey, New Mexico, New York, North Dakota, Ohio, Oregon, Pennsylvania, Rhode Island, Vermont, Washington, West Virginia, and Wisconsin. Nonexpansion states were Alabama, Florida, Georgia, Idaho, Kansas, Maine, Mississippi, Missouri, Nebraska, North Carolina, Oklahoma, South Carolina, South Dakota, Tennessee, Texas, Utah, Virginia, and Wyoming. In sensitivity analyses, we excluded Louisiana and Montana because they expanded Medicaid in 2016 but the results were unaffected. Excluding Wisconsin, which we categorized as an expansion state because it received approval to provide coverage to childless adults with incomes up to 100% of the federal poverty line, also did not affect results.

We estimated separate multivariable linear models on noncitizen SSI recipiency and citizen SSI recipiency, and controlled for state unemployment rate, population size of the group (citizens and noncitizens) in the state, and state and year effects. During the period of our study, the noncitizen population in the United States was subject to a number of policies that may have affected their health, well-being, and participation in SSI.¹⁶ To account for these changes in policies, for the noncitizen SSI recipiency, we estimated a second model that additionally controlled for a range of time-varying state-level policies toward immigrants—namely, enforcement of Section 287(g) of the Illegal Immigration Reform and Immigrant Responsibility Act, the Secure Communities Program, state Dream Act policy, eligibility of undocumented immigrants to obtain a driver’s license, and state implementation of E-Verify.

For the supplementary analyses based on CPS data, we estimated linear probability models and also controlled for the following demographic characteristics: age, gender, educational attainment, marital status, household size, period of arrival to the United States, and years since migration. Linear probability models give consistent estimates of average causal effects and have an intuitive interpretation.¹⁷

Our analytical approach yields unbiased estimates of the causal effect of Medicaid expansion on the assumption that the trend in SSI recipiency in expansion and nonexpansion states would have been similar if Medicaid expansion had not occurred. We examined this parallel trends assumption first visually and then statistically, as in previous research on the effects of Medicaid expansion.¹⁸ The statistical approach used data for the pre-ACA period (2009–2013) and tested whether the difference in SSI recipiency between expansion and nonexpansion states was constant over time. We used models similar to those in our main analysis, except that state Medicaid expansion status was interacted with year of observation. If the parallel trends assumption holds, there should be no statistically significant interaction between state Medicaid expansion status and year in the pre-ACA period. Estimates presented in Figure A (available as a supplement to the online version of this article at http://www.ajph.org) suggest that to be the case. Estimated coefficients of the interactions in the Medicaid expansion dummy variable and year dummy variables were negligible and statistically insignificant for both citizens and noncitizens.

We conducted all analyses using Stata version 16 (StataCorp LP, College Station, TX). We used robust standard errors clustered on state of residence to account for arbitrary correlation of observations within each state.

RESULTS

In this section, we present results from descriptive analysis followed by results of the multivariable analysis.

Descriptive Results

The SSI receipt rate among noncitizens was substantially lower than that among citizens even prior to the ACA. Between 2009 and 2013, 0.8% of noncitizens aged 18 to 64 years received SSI, which was less than a third of the receipt rate of 2.5% among citizens (Table 1). The former also experienced a decline in SSI receipt rate after the ACA. Overall, between the pre-ACA period (2009–2013) and the post-ACA period (2014–2018), nationally, the SSI receipt rate increased 2% among citizens but decreased 13% among noncitizens. Between those same 2 periods, the SSI receipt rate among citizens fell 0.03 percentage points in Medicaid expansion states compared with nonexpansion states, whereas the corresponding decline among noncitizens was 0.1 percentage points. These statistics do not adjust for changes in demographics. The multivariable analysis, presented in the next subsection, adjusts for a rich set of demographics.

TABLE 1—

Percentage of Citizens and Noncitizens Aged 18–64 Years Who Were Supplemental Security Income (SSI) Recipients, Before and After Medicaid Expansion: United States, 2009–2018

	Citizens, % (95% CI)		Noncitizens, % (95% CI)
	2009–2013	2014–2018	2009–2013	2014–2018
Medicaid nonexpansion states	2.42 (2.04, 2.80)	2.50 (2.12, 2.88)	0.64 (0.37, 0.92)	0.59 (0.33, 0.86)
Medicaid expansion states	2.53 (2.19, 2.87)	2.58 (2.25, 2.90)	0.94 (0.74, 1.13)	0.79 (0.61, 0.96)
All states	2.49 (2.23, 2.75)	2.55 (2.30, 2.80)	0.83 (0.67, 1.00)	0.72 (0.57, 0.86)

Note. CI =  confidence interval.

Source. SSI receipt data are from the Social Security Administration SSI Annual Statistical Reports, 2009–2018. Population sizes are from the American Community Survey, 2009–2018.

Figure 1 shows the trends in the SSI receipt rate of citizens and noncitizens by Medicaid expansion versus Medicaid nonexpansion states. For citizens, the SSI receipt rate peaked in 2013, followed by a modest fall in both expansion and nonexpansion states. For noncitizens, SSI receipt also peaked in 2013, followed by a sharp decline in expansion states (from 0.9% to 0.7%) compared with nonexpansion states (0.7% to 0.6%).

FIGURE 1—

Trends in Average Percentage of Citizens and Noncitizens Aged 18–64 Years Who Were Supplemental Security Income (SSI) Recipients, by State Medicaid Expansion Status: United States, 2009–2018

Source. Authors’ analysis of data from the Social Security Administration SSI Annual Statistical Reports, 2009–2018. Population sizes are from the American Community Survey, 2009–2018.

Figure 1 also shows that trends in SSI recipiency rates in expansion states were similar to those in nonexpansion states in the pre-ACA period, which is consistent with the parallel trends assumption. Statistical tests also found no statistical difference in pre-ACA trends between expansion and nonexpansion states. Results are available in Figure A.

Difference-in-Differences Regression Results

Next, we used a multivariable regression framework to compute the difference-in-differences estimates of the effect of Medicaid expansion on SSI receipt. This involved comparing change in SSI receipt before and after ACA implementation in Medicaid expansion and nonexpansion states. We did these analyses using administrative SSA data as well as survey data; the results are presented in Tables 2 and 3, respectively.

TABLE 2—

Effect of Medicaid Expansion on Percentage of Adults Aged 18–64 Years Receiving Supplemental Security Income (SSI), by Citizenship Status: United States, 2009–2018

	Citizens	Noncitizens
	Model 1, b (95% CI)	Model 1, b (95% CI)	Model 2, b (95% CI)
Medicaid expansion state	−1.72 (−1.79, −1.65)	0.79 (0.72, 0.86)	0.81 (0.70, 0.93)
Post-ACA	0.10 (0.00, 0.19)	−0.19 (−0.35, −0.04)	−0.19 (−0.34, −0.03)
Medicaid expansion state × post-ACA	−0.05 (−0.11, 0.00)	−0.10 (−0.18, −0.02)	−0.09 (−0.17, −0.01)
No. of observations	510	510	510

Note. ACA = Affordable Care Act; CI = confidence interval. The dependent variable is percentage of adults aged 18–64 years receiving SSI. All models controlled for number of noncitizens aged 18–64 years in the state, state unemployment rate, state fixed effects, and year fixed effects. Model 2 (for noncitizens) additionally controlled for state policies toward immigrants—namely, enforcement of Section 287(g) of the Illegal Immigration Reform and Immigrant Responsibility Act, the Secure Communities Program, state Dream Act policy, eligibility of undocumented immigrants to obtain driver’s license, and state implementation of E-Verify.

Source. SSI receipt data are from the Social Security Administration SSI Annual Statistical Reports, 2009–2018. Population sizes are from the American Community Survey, 2009–2018.

TABLE 3—

Effect of Medicaid Expansion on Probability of Low-Income Adults Aged 18–64 Years Receiving Supplemental Security Income, by Citizenship Status: Current Population Survey, United States, 2009–2019

	Citizens	Noncitizens
	Model 1, b (95% CI)	Model 2, b (95% CI)	Model 3, b (95% CI)
Low-income and childless
Medicaid expansion state	0.0868 (0.0761, 0.0975)	−0.0022 (−0.0228, 0.0184)	−0.0140 (−0.0354, 0.0075)
Post-ACA	−0.0093 (−0.0314, 0.0128)	−0.0458 (−0.0855, −0.0061)	−0.0450 (−0.0964, 0.0063)
Medicaid expansion state × post-ACA	−0.0060 (−0.0166, 0.0047)	−0.0258 (−0.0466, −0.0051)	−0.0284 (−0.0566, −0.0002)
Pre-ACA mean of outcome variable	0.118	0.036	0.036
No. of observations	73 001	5 775	5 775
Low-income, childless, and disabled
Medicaid expansion state	0.1504 (0.1152, 0.1856)	−0.6217 (−0.8145, −0.4289)	−0.4555 (−0.7442, −0.1669)
Post-ACA	−0.0202 (−0.0869, 0.0465)	−0.3351 (−0.6583, −0.0119)	−0.3286 (−0.7657, 0.1085)
Medicaid expansion state × post-ACA	−0.0305 (−0.0614, 0.0004)	−0.1536 (−0.2720, −0.0352)	−0.1960 (−0.3909, −0.0011)
Pre-ACA mean of outcome variable	0.343	0.247	0.247
No. of observations	17 418	513	513

Note. ACA = Affordable Care Act; CI = confidence interval. Sample is restricted to childless adults aged 18–64 years in households with incomes below the federal poverty threshold.¹⁰ All models controlled for age (categories: 18–26 [ref], 27–34, 35–42, 43–49, 50–57, and 58–64 years), gender (female [ref], and male), educational attainment (categories: high school or lower [ref], some college, and associate degree or higher), marital status (categories: married [ref], married but spouse absent, separated, divorced, widowed, and never married or single), household size (categories: 1 [ref], 2, 3, and 4 or more), number of (non)citizens aged 18–64 years in the state, state unemployment rate, state fixed effects, and year fixed effects. Models 2 and 3 additionally controlled for period of arrival to the United States and years since migration. Model 3 additionally controlled for state policies toward immigrants—namely, enforcement of Section 287(g) of the Illegal Immigration Reform and Immigrant Responsibility Act, the Secure Communities Program, state Dream Act policy, eligibility of undocumented immigrants to obtain driver’s license, and state implementation of E-Verify.

Source. Authors’ analysis of data from the Current Population Survey, 2009–2019.

Analyses based on SSA data adjusted for state-level characteristics suggested that the SSI receipt rate among citizens fell 0.05 percentage points (or 2% of the preexpansion mean for citizens) in Medicaid expansion states compared with nonexpansion states (Table 2). The corresponding decline in the SSI receipt rate among noncitizens was of a much higher magnitude: 0.1 percentage points, or 12% of the preexpansion mean for noncitizens. Estimates remained robust in models that controlled for a rich set of policies toward noncitizens that can potentially affect the utilization of means-tested programs (model 2).

We conducted regression analysis with survey data on 2 samples of adults with incomes below the poverty threshold: nonelderly childless adults and nonelderly childless adults with disabilities. Estimates based on the sample of nonelderly childless adults suggest that, compared with nonexpansion states, SSI receipt among citizens fell 0.6 percentage points (or 5% of the pre-ACA mean) in Medicaid expansion states; the decline was 2.6 percentage points (or 72% of the pre-ACA mean) for noncitizens (Table 3).

Narrowing the sample to nonelderly childless adults with disability—the population whose SSI recipiency was most affected by the Medicaid expansion—increased the size of the effects: compared with nonexpansion states, SSI receipt among citizens fell 3.1 percentage points (or 9% of the pre-ACA mean) in Medicaid expansion states; the decline was 15.4 percentage points (or 62% of the pre-ACA mean) for noncitizens.

DISCUSSION

In this quasi-experimental study, we found that Medicaid expansion led to meaningful reductions in SSI recipiency among noncitizens. Noncitizens experienced a sharper decline in SSI receipt after ACA implementation than citizens. For noncitizens, SSI receipt peaked in 2013—just prior to ACA implementation—followed by a 12% decline in Medicaid expansion states compared with nonexpansion states in the post-ACA period. SSI receipt among citizens similarly peaked in 2013, but the corresponding decline was a modest 2%.

In the United States, noncitizen participation in welfare programs is highly controversial. A number of recent studies have stressed the increase in immigrant participation in Medicaid and its fiscal implications.^19,20 Our analysis, however, suggested that by focusing entirely on Medicaid participation, these previous studies did not capture the full impact of Medicaid expansions on the exchequer.

Our analysis suggested that Medicaid expansion, by creating an avenue for public health insurance for low-income families, reduced their SSI receipt. Given these findings, inferences on the fiscal effects of ACA expansions, including the effects of immigrant insurance coverage, should be adjusted for the spillover effect of decline in SSI receipt. Our back-of-the-envelope estimates suggest an annual savings of $122 million in federal SSI costs because of the decline in SSI participation among noncitizens in Medicaid expansion states ($122 million = $260 [monthly benefit] × 12 × 39 311 [number receiving SSI in 2013 minus number receiving SSI in 2018 among noncitizens aged 18–64 years in Medicaid expansion states]), assuming that noncitizens who exited SSI received a third of the maximum SSI benefit of $771 in 2019. These savings are in addition to the corresponding savings due to the decline in SSI participation among citizens, a much larger population: $497 million ($260 [monthly benefit] × 12 × 159 528).

These estimates suggest that any discussion of the fiscal implications of Medicaid expansions should take into account savings from other means-tested programs, in particular declines in SSI participation. Our analysis contributes to other research that documents that access to public health insurance reduces health care expenditures by ensuring timely health care and thus avoiding expensive emergency public health care that is generally available to immigrants.^21–23

Although a reduction in SSI participation represents savings to the federal government, it is a loss of benefits to potential beneficiaries. Previous research documents higher employment rates in Medicaid expansion states than in nonexpansion states among adults aged 18 to 64 years with disabilities, which suggests that a reduction in SSI benefits may be offset, at least in part, by earned income.^4,5 Further research on the net financial gain to individuals is warranted.

Limitations

Our study has limitations. First, unobserved bias from time-varying factors that differentially affect expansion and nonexpansion states cannot be definitively excluded. We supported the validity of our findings by showing that our data are consistent with the parallel trends assumption in the pre-ACA period. We also performed robustness checks by fitting models with a rich set of controls, which produced similar estimates.

Second, our analyses based on SSA administrative data are at the state level, which could mask changes in the composition of the population within states. This would be a concern if changes in state composition were correlated with Medicaid expansion. To address this concern, we estimated models using individual-level survey data, which yielded similar conclusions. Previous research also finds that the level of generosity of a state’s social programs does not influence the residency patterns of immigrants.^24,25

Public Health Implications

The COVID-19 pandemic has revealed the serious implications of a large population without health insurance to overall public health. Note that because immigrants, on average, are younger and healthier than natives, they are likely to have lower health care utilization than natives; therefore, immigrant Medicaid eligibility is likely to be a cost-effective policy for society.^26–28

For the SSI population specifically—the focus of this study—the effect on Medicaid expansions on the economy is likely to be even greater, as these expansions also provided opportunities to low-income adults to increase work effort and accumulate savings and assets without the risk of losing public health insurance.^4,5,7,12 An increase in work effort, and therefore in income, would further increase fiscal windfalls from Medicaid expansions.