You are here

New Claims by Opponents of Medicaid Expansion Rest on Faulty Analysis

September 18, 2019

Some opponents of the Affordable Care Act’s (ACA) Medicaid expansion who seek to dissuade states that haven’t adopted it from doing so are using inaccurate claims and misleading arguments to make their case. In particular, in a recent paper and Wall Street Journal op-ed, former Trump White House health official Brian Blase claimed that large numbers of “middle-class” people are illegitimately enrolled in Medicaid through the ACA’s Medicaid expansion.[1] Yet the data do not show that large numbers of ineligible people are enrolling.

Blase drew mainly on one study that finds that significant numbers of people who reported in Census surveys that they have annual income above the Medicaid expansion eligibility cutoff (138 percent of the poverty line) appear to have gained coverage through the expansion.[2] The op-ed, which Blase co-authored with Aaron Yelowitz (an author of the study in question), concluded that the expansion has become “an entitlement program for the middle class” and that, as a result, the expansion isn’t achieving its goals.

Those reporting survey income above 138 percent of poverty, however, could be eligible for the Medicaid expansion for many legitimate reasons. They could, for example, be eligible for part of the year because they had low income in some months due to temporary unemployment or unstable hours — Medicaid eligibility is generally based on monthly, not annual, income. Or, they could have income from child support or other sources that don’t count toward Medicaid eligibility. Or, in their responses to the Census questions, they could have provided rough estimates of their incomes rather than precise answers. The Census surveys don’t verify income, while Medicaid does.

In addition, some higher-income people whom surveys record as enrolled in Medicaid may be enrolled in other coverage (such as marketplace coverage), either because they responded incorrectly to the survey questions or because of the way that Census studies infer Medicaid enrollment for those who don’t answer the relevant survey question. A separate study by Yelowitz corrects for some, though not all, of these factors — and that alone cuts in half the number of enrollees who appear to have income above the Medicaid limits.

In their op-ed, Blase and Yelowitz implied that Medicaid enrollment is based on self-reported income but, in fact, state Medicaid programs verify an applicant’s income when he or she applies and then periodically during the year. To make their case about “massive” numbers of people improperly or fraudulently enrolling, the authors also misrepresented the findings of audits by the federal government and the state of Louisiana, as explained below.

"In reality, the Medicaid expansion is serving the people that it’s supposed to serve — low-income people who are eligible and need it. States that have adopted the expansion have seen far larger coverage gains, particularly for low-income people who are eligible and need it."In reality, the Medicaid expansion is serving the people that it’s supposed to serve — low-income people who are eligible and need it. States that have adopted the expansion have seen far larger coverage gains, particularly for low-income people, than non-expansion states since the expansion took effect. And across the states, uninsured rates have fallen by about the same amount that Medicaid’s enrollment has risen, indicating that the vast majority of those gaining coverage would otherwise be uninsured. Moreover, research shows that the Medicaid expansion is improving access to medications, check-ups, screenings, and other needed care, reducing medical debt and evictions, and reducing premature deaths among low-income people. If there is a scandal here, it is that millions of low-income people in non-expansion states remain uninsured.

Claims Rest on Problematic Presentation of Data on the Medicaid Expansion

Blase’s claims, detailed in a Galen Institute paper and in the Wall Street Journal op-ed coauthored with Yelowitz, are largely based on a study by Yelowitz and other researchers that uses the Census Bureau’s American Community Survey (ACS) data to examine Medicaid enrollment by income across states. It finds that Medicaid enrollment rose at a greater rate among people with incomes above 138 percent of the poverty line in nine states that expanded Medicaid than in 12 states that didn’t expand. [3] The op-ed erroneously concludes that the study shows that “massive” numbers of “middle-class” people are improperly enrolled in Medicaid. It also references — and misrepresents — studies of Medicaid expansion enrollment by the Department of Health and Human Services (HHS) Inspector General and the Louisiana Inspector General’s office.

Many People Who Report Somewhat Higher Income in Surveys Are Eligible for Medicaid Expansion

Blase and Yelowitz’s claims don’t take into account that the income that people report on a Census survey doesn’t always align with how Medicaid treats income in determining eligibility. There are a number of reasons why people who report incomes somewhat above 138 percent of the poverty line in a survey may be eligible for the Medicaid expansion.

Probably most important, surveys measure annual income, while Medicaid eligibility is primarily based on monthly income. Someone whose annual income is above 138 percent of the poverty line, but who experiences a period of unemployment or underemployment during the year, may well be eligible for Medicaid for part of the year.

Income changes over the course of a year are especially prevalent among low-income people. Low-wage jobs are often unstable, with frequent job losses and work hours that can fluctuate from month to month. Many Medicaid enrollees also work seasonal jobs in industries such as retail or tourism. A study looking at participation of working-age adults in the Supplemental Nutrition Assistance Program (SNAP), which has federal income limits close to those of the Medicaid expansion, found that workers earning low wages are frequently in and out of work and on and off SNAP as their earnings fall and rise.[4] A similar study looking at Medicaid showed similar income volatility.[5]

The majority of people with income below 138 percent of the poverty line at the beginning of a 12-month period had income above 138 percent of the poverty line at some point during those 12 months, another study found.[6] Conversely, about 40 percent of people with incomes between 138 and 200 percent of the poverty line saw their incomes fall below 138 percent of the poverty line at some point over the course of a year. This degree of income volatility means that large numbers of people with annual income above 138 percent of the poverty line might be appropriately enrolled in Medicaid at some point during the year.

There are additional reasons people with survey income above 138 percent of the poverty line could be eligible for Medicaid, including:

  • Mismeasurement of income in surveys. People don’t necessarily answer surveys precisely, while they must be precise in a Medicaid application, which is signed under penalty of perjury. In their study, Yelowitz and his coauthors acknowledge that their findings could be attributable in part to survey measurement error in either insurance coverage or income.
  • Differences in who counts as part of a household. The ACS asks about the income of all related people who live together; Medicaid uses different household rules that generally only take the income of immediate family members into account. In general, parents’ income counts in determining eligibility for children under 19 (or 21 at state election) and income of each spouse counts in determining the other’s eligibility whenever the family members live together. But, for example, a grandparent under 65 with earnings below 138 percent of the poverty line could be eligible for Medicaid in an expansion state when she lives with her adult children, even when the extended family’s total income exceeds the eligibility limit. The same would be true for a 30-year-old low-wage worker living with his or her parents.
  • Medicaid rules that exclude certain types of income in determining eligibility. The ACA changed how Medicaid counts income to align with how the federal income tax system treats income in determining eligibility for premium tax credits, which help pay for health coverage in the ACA marketplaces. The tax rules exclude child support, Supplemental Security Income (SSI) benefits, and certain other forms of income, which can make eligible Medicaid enrollees appear ineligible based on their Census survey responses. Child support, a child’s SSI payment or Social Security survivor’s benefit, some veterans’ payments, and worker’s compensation payments generally wouldn’t be counted in determining Medicaid eligibility but would likely be reported in Census surveys.[7]

The study does not adjust for the impact of these differences between how Medicaid calculates eligibility and how people may report their income on the ACS. Some of these adjustments aren’t possible in the ACS data (which don’t capture monthly income, for example). A prior study conducted by Yelowitz, however, did adjust for some of the differences. These limited adjustments cut in half the number of people who appeared ineligible.[8]

Census Data May Also Exaggerate Number of Higher-Income People Enrolled in Medicaid

In addition, some higher-income people whom Census data record as enrolled in the Medicaid expansion may actually be enrolled in other forms of coverage, due to mistakes about insurance coverage in answers to the surveys and due to ways in which Census infers Medicaid enrollment for people who don’t answer the relevant survey question.

On the ACS, insurance coverage options include “insurance purchased directly from an insurance company” and “Medicaid, Medical Assistance, or any kind of government-assistance plan for those with low incomes or a disability.” This question wording may lead some people purchasing subsidized coverage through the ACA marketplaces to mistakenly place themselves in the Medicaid category, rather than responding that they have private insurance. For these and other reasons, a meaningful share of respondents appear to misreport their source of insurance coverage, as significant differences between survey-based estimates and administrative data show.[9]

For some people who don’t respond to the ACS question about Medicaid receipt, the Census Bureau imputes a response based on various characteristics that do not include income. This approach somewhat exaggerates the number of Medicaid enrollees with higher incomes, including the number of Medicaid expansion enrollees, a Congressional Budget Office analysis concluded.[10]

Medicaid Programs Have Stringent Verification Procedures

Blase and Yelowitz imply in their Wall Street Journal op-ed that Medicaid enrollment is based purely on self-reported income. In particular, they incorrectly assert that people using the HealthCare.gov website who “enter no income simply to explore their options” can enroll in Medicaid with no further eligibility checks. But that’s not how Medicaid works.

To determine eligibility, state Medicaid programs verify an applicant’s reported income against electronic data sources at the time of application and then periodically during the year. When verifying income, state Medicaid agencies compare the sworn attestations that clients make on their application and renewal forms to available electronic data. No further documentation is needed if the income reported by the applicant and the electronic data sources both show income below, at, or above the relevant Medicaid eligibility threshold. If there is a difference — for example, if the applicant claims income below the eligibility level and the data source shows income above it — states request documentation to resolve the difference.[11] Similarly states must ask enrollees for documentation if the state receives information from electronic data sources suggesting the enrollee may no longer be eligible for coverage.[12]

This verification process occurs regardless of whether the application process starts at the federal marketplace or the state Medicaid agency. When someone applies at the federal marketplace and appears eligible for Medicaid, depending on the state, the marketplace either determines eligibility itself or transfers the case to the state Medicaid agency for a final determination. In no case does the federal marketplace simply accept the applicant’s assertions. In the nine states that choose to have the federal marketplace determine eligibility for Medicaid, it applies the state’s own eligibility rules and standard Medicaid verification procedures.[13] Regardless of whether the state or the marketplace determines eligibility, the applicant’s self-reported income is verified.[14]

People With Modestly Higher Incomes Who Remain Enrolled in Medicaid Likely Do So Due to Income Volatility, Not Fraud

Some people reporting higher income in the ACS likely experienced a change in income after their eligibility was determined and then remained enrolled in Medicaid for some period after the change.

As noted, low-income people are particularly likely to experience income changes. As a result, Medicaid eligibility rules can require them to shift back and forth between Medicaid and marketplace coverage multiple times during a year (often referred to as “churn”), requiring them to navigate complex processes to maintain coverage.

Insurance transitions often disrupt the continuity of care and coverage. A study of 2015 churn rates in three states found that about 25 percent of low-income adults changed coverage over the course of a year. Over half of those experiencing a change had a gap in coverage. Coverage changes were associated with changes in physicians, increased use of the emergency room, and decreased medication adherence even for many who didn’t have gaps in coverage.[15] Churn also creates problems for health care providers and Medicaid managed care organizations, limiting their ability to provide effective care and increasing their administrative costs as people cycle in and out of coverage. People who churn in and out of coverage have higher health care costs, some studies suggest.[16]

Some states have decided this churn is so counterproductive that they have changed their eligibility rules to limit the frequency with which households need to change coverage. States have the option to provide children enrolled in Medicaid and the Children’s Health Insurance Program (CHIP) with “continuous eligibility” — a full year of coverage regardless of changes in their family’s income. States can also elect to provide continuous eligibility to adults through a Medicaid waiver. To date, 24 states have adopted continuous eligibility for children in Medicaid, and 26 have adopted it for CHIP. So far, Montana and New York are the only states with continuous eligibility for adults. Utah has a proposal pending.

In states that have not adopted continuous eligibility, it is likely that some people still remain enrolled in Medicaid for a period after their income rises; similarly, it is likely that some people remain enrolled in the marketplaces for a period after their income falls. But while Blase’s paper suggests that the federal government is spending billions on people who are inappropriately enrolled in Medicaid, the reality is that the fiscal impact of these mistakes is limited. Medicaid expansion enrollees whose incomes rise modestly above 138 percent of the poverty line are generally eligible for subsidized marketplace coverage. And for people with low incomes, the federal cost for subsidized marketplace coverage is similar to (or sometimes greater than) the federal cost for Medicaid.[17]

Authors Do Not Accurately Present the Results of Government Audits

The audits conducted by Louisiana and by federal investigatory officials that the Blase and Yelowitz op-ed cites do not support the authors’ claim of “massive improper and fraudulent expansion enrollment.”

While Blase and Yelowitz claim that a 2018 Louisiana audit “found 82 percent of expansion enrollees were ineligible,” it actually raised eligibility questions about one-tenth that number of people.[18] The audit examined only the roughly 10 percent of enrollees who appeared most likely to be ineligible based on the state’s quarterly wage database. It found that 82 percent of that 10 percent had income above the eligibility threshold at some point during the year, not 82 percent of the expansion population.

Moreover, few of these cases were likely the result of deliberate fraud. All of these individuals had their eligibility verified at the time they enrolled, and they likely experienced a subsequent change in income. The audit also does not reflect Louisiana’s current eligibility procedures: Louisiana’s new eligibility system, which went live late in 2018 — after the audit was completed — now regularly reviews the eligibility of enrollees whose incomes may have risen.[19]

Similarly, both the Wall Street Journal op-ed and the Galen Institute paper cite 2018 HHS Inspector General audits of California and New York, which Blase claims found that 25 percent of expansion enrollees were ineligible. In reality, both audits mostly identified cases where the states miscategorized people who were eligible for Medicaid, but under a different eligibility category, as expansion enrollees. That’s still concerning, because the federal government pays a higher share of costs for expansion enrollees than other enrollees (with states paying the remainder). But it is inaccurate to suggest, as Blase does, that these studies show that a quarter of individuals were ineligible for the Medicaid program.[20]

Data Show That Medicaid Expansion Benefits Low-Income People Who Would Otherwise Be Uninsured

The broader thesis of both the Wall Street Journal op-ed and the Galen Institute paper is that the Medicaid expansion is not achieving its goals, and money spent on Medicaid expansion isn’t well-spent. A large body of evidence refutes these claims.

Figure 1
Expansion States' Uninsured Rates Fell As Medicaid Coverage Rose

Since expansion took effect in 2014, uninsured rates have fallen far more in states that have expanded Medicaid than in non-expansion states, particularly for low-income people. Since 2013, Medicaid coverage among adults with incomes under 138 percent of poverty in Medicaid expansion states has increased about 18 percentage points, while uninsurance has dropped about 19 percentage points. In non-expansion states, by contrast, low-income adults haven’t fared as well — with uninsurance falling by only about 11 percentage points, or only a little more than half as much as in expansion states.[21] Uninsured rates have fallen in all states due to provisions of the ACA that apply nationwide, such as its subsidized coverage through the health insurance marketplaces.

In the op-ed, Blase and Yelowitz also imply that large numbers of people have signed up for Medicaid expansion despite having other coverage options, arguing that Medicaid expansion has worked against the preservation of private coverage. But, as Figure 1 shows, there is a strong linear and nearly one-to-one relationship between the increase in the share of a state’s population that’s covered by Medicaid and the decrease in the share that’s uninsured. This analysis, which incorporates people at all income levels, suggests that the large majority of those gaining coverage under Medicaid expansion would otherwise have been uninsured.

That is the same conclusion reached by several studies that have directly investigated whether Medicaid expansion has “crowded out” private coverage. Most of these studies find “no evidence of ‘crowd-out,’” while others show “slight declines in private coverage” from Medicaid expansion, a comprehensive Kaiser Family Foundation literature review concludes.[22]

Meanwhile, a growing body of research finds that expansion is yielding substantial gains for beneficiaries.[23] Expanding coverage has improved financial security, reducing medical debt and evictions.[24] It has increased the share of low-income adults who have a personal physician, get check-ups, and receive recommended preventive care such as cholesterol and cancer screenings, and it has decreased the share delaying needed care or skipping medications due to cost.[25] Evidence is building that these gains in access to care are translating into improved health outcomes, including significant reductions in premature deaths among older adults.[26]

Had the uninsured rate in non-expansion states fallen since 2013 to the same degree that it fell in Medicaid expansion states, 4.7 million fewer Americans would have been uninsured last year.[27] Misleading claims about who is enrolled in the Medicaid expansion should not deter additional states from extending these benefits to the millions of people who continue to be left out.

End Notes

[1] Brian Blase, “Health Reform Progress: Beyond Repeal and Replace,” Galen Institute, September 2019, https://galen.org/assets/Health-Reform-Progress-Brian_Blase.pdf and Brian Blase and Aaron Yelowitz, “ObamaCare’s Medicaid Deception,” Wall Street Journal, August 14, 2019, https://www.wsj.com/articles/obamacares-medicaid-deception-11565822360.

[2] Charles Courtemanche, James Marton, and Aaron Yelowitz, “Medicaid Coverage Across the Income Distribution Under the Affordable Care Act,” National Bureau of Economic Research Working Paper No. 26145, August 2019, https://www.nber.org/papers/w26145

[3]Ibid.

[4] Brynne Keith-Jennings and Raheem Chaudry, “Most Working-Age SNAP Participants Work, But Often in Unstable Jobs,” Center on Budget and Policy Priorities, March 15, 2018, https://www.cbpp.org/research/food-assistance/most-working-age-snap-participants-work-but-often-in-unstable-jobs.

[5] Aviva Aron-Dine, Raheem Chaudry, and Matt Broaddus, “Many Working People Could Lose Health Coverage Due to Medicaid Work Requirements,” Center on Budget and Policy Priorities, April 11, 2018, https://www.cbpp.org/research/health/many-working-people-could-lose-health-coverage-due-to-medicaid-work-requirements.

[6] Benjamin D. Sommers and Sara Rosenbaum, “Issues In Health Reform: How Changes In Eligibility May Move Millions Back And Forth Between Medicaid And Insurance Exchanges,” Health Affairs, February 2011, https://www.healthaffairs.org/doi/pdf/10.1377/hlthaff.2010.1000.

[7] Center on Budget and Policy Priorities, “Key Facts: Income Definitions for Marketplace and Medicaid Coverage,” September 26, 2018, http://www.healthreformbeyondthebasics.org/wp-content/uploads/2014/10/KeyFacts-Income-Definitions.pdf.

[8] Aaron Yelowitz, “How Did the ACA Affect Health Insurance Coverage in Kentucky?” Schattner Institute Working Paper, September 2016, http://isfe.uky.edu/sites/ISFE/files/research-pdfs/Schnatter.2017.Yelowitz.pdf.

[9] Joanne Pascale et al., “Validating Self-Reported Health Insurance Coverage: Preliminary Results on the CPS and ACS,” Annual Meeting of the American Association for Public Opinion Research, May 14, 2016, https://census.gov/content/dam/Census/newsroom/press-kits/2016/Validating%20Self-Reported%20Health%20Insurance%20Coverage-%20Pascale.pdf; Michel Boudreaux et al., “Accuracy of Medicaid Reporting in the ACS: Preliminary Results from Linked Data,” State Health Assistance Data Center, 2013, https://nces.ed.gov/FCSM/pdf/A4_Turner_2013FCSMr.pdf.

[10] Bilal Habib, “How CBO Adjusts for Survey Underreporting of Transfer Income in Its Distributional Analysis,” Congressional Budget Office Working Paper 2018-07, July 2018, https://www.cbo.gov/system/files/2018-07/54234-workingpaper.pdf. See particularly Figure 11.

[11] Jennifer Wagner, “Reasonable Compatibility Policy Presents an Opportunity to Streamline Medicaid Determinations,” Center on Budget and Policy Priorities, August 16, 2016, https://www.cbpp.org/research/reasonable-compatibility-policy-presents-an-opportunity-to-streamline-medicaid.

[12] 42 CFR §435.952.

[13] Center for Medicare & Medicaid Services, “Medicaid and CHIP FAQs: Coordination between Medicaid/CHIP and the Federally‐Facilitated Marketplace,” May 2012 and April 2013, https://www.medicaid.gov/State-Resource-Center/FAQ-Medicaid-and-CHIP-Affordable-Care-Act-Implementation/Downloads/FAQs-by-Topic-Coordination-with-Marketplace.pdf.

[14] States can opt to have the marketplace determine eligibility or make the determination themselves. 45 CFR §155.302. In the latter case, the marketplace verifies income using Medicaid rules. 45 CFR §155.305.

[15] Benjamin D. Sommers et al., “Insurance Churning Rates For Low-Income Adults Under Health Reform: Lower Than Expected But Still Harmful For Many,” Health Affairs, October 2016, https://www.healthaffairs.org/doi/pdf/10.1377/hlthaff.2016.0455.

[16] Anthem Public Policy Institute, “Continuity of Medicaid Coverage Improves Outcomes for Beneficiaries and States,” June 2018, https://www.communityplans.net/wp-content/uploads/2019/04/13_Report_Continuity-of-Medicaid-Coverage-Improves-Outcomes-for-Beneficiaries-and-States.pdf.

[17] The Congressional Budget Office (CBO) estimates the 2019 annual average federal cost of covering an individual in Medicaid or CHIP at $4,620, compared to $6,490 for covering an individual in the ACA marketplace. While these cost estimates are not directly comparable due to differences in the people who are eligible for coverage in these programs, they are suggestive evidence that coverage through Medicaid is not more costly to the federal government than coverage through the marketplace. Additional Treasury Department data show that ACA marketplace subsidies for those between 150 percent and 200 percent of poverty — those just above the Medicaid expansion level — are greater than the average subsidy, further suggesting that the federal government pays a similar amount, or perhaps less, for people with incomes modestly above 138 percent of the poverty line who remain enrolled in Medicaid. See, “Federal Subsidies for Health Insurance Coverage for People Under Age 65: 2019 to 2029,” Congressional Budget Office, May 2019, https://www.cbo.gov/system/files/2019-05/55085-HealthCoverageSubsidies_0.pdf; and, “Health Tax Provisions and Analysis, Table 3, Premium Tax Credit, 2018,” U.S. Department of the Treasury, accessed September 2019, https://home.treasury.gov/policy-issues/tax-policy/office-of-tax-analysis.

[18] Louisiana Department of Health, “Medicaid Eligibility: Wage Verification Process Of The Expansion Population,” November 8, 2018, http://app.lla.state.la.us/PublicReports.nsf/0/1CDD30D9C8286082862583400065E5F6/$FILE/0001ABC3.pdf

[19] Jennifer Wagner, “Louisiana Medicaid Audit Results Mischaracterized,” Center on Budget and Policy Priorities, February 25, 2019, https://www.cbpp.org/blog/louisiana-medicaid-audit-results-mischaracterized.

[20] The California audit found a small number of cases where an enrollee’s income appeared higher than when verified at enrollment. Other cases the auditors flagged were people who remained eligible but were placed in the wrong Medicaid coverage group and some for whom the state didn’t follow proper procedures to verify residency or citizenship. The New York audit primarily focused on whether New York was correctly claiming enhanced match for enrollees in the expansion group. New York had increased its eligibility levels for adults in Medicaid prior to the ACA, and as a result, the state can only claim enhanced match for enrollees with incomes above the poverty line. While the audit did find a small number of enrollees with income above the eligibility limit, most of the errors it found were cases where New York claimed enhanced matching funds for enrollees with income below the poverty line who were clearly eligible for Medicaid.

[21] CBPP calculations based on the Census Bureau’s 2013 and 2017 American Community Surveys. “Low-income adults” are defined as those aged 19 to 64 with income under 138 percent of the poverty line. Medicaid expansion states are the 31 states and the District of Columbia that expanded Medicaid to low-income adults under the ACA by January 1, 2017.

[22] Larisa Antonisse et al., “The Effects of Medicaid Expansion Under the ACA: Updated Findings from a Literature Review,” Kaiser Family Foundation, August 15, 2019, https://www.kff.org/medicaid/issue-brief/the-effects-of-medicaid-expansion-under-the-aca-updated-findings-from-a-literature-review-august-2019/.

[23] Robin Rudowitz and Larisa Antonisse, “Implications of the ACA Medicaid Expansion: A Look at the Data and Evidence,” Kaiser Family Foundation, May 23, 2018, https://www.kff.org/medicaid/issue-brief/implications-of-the-aca-medicaid-expansion-a-look-at-the-data-and-evidence/.

[24] Matt Broaddus, “Study: Medicaid Expansion Improves Low-Income Peoples’ Financial Health, Too,” Center on Budget and Policy Priorities, November 14, 2017, https://www.cbpp.org/blog/study-medicaid-expansion-improves-low-income-peoples-financial-health-too; Matt Broaddus, “Medicaid Improves Financial Well-Being, Research Finds,” Center on Budget and Policy Priorities, April 28, 2016, http://www.cbpp.org/blog/medicaid-improves-financial-well-being-research-finds; Heidi L. Allen et al., “Can Medicaid Expansion Prevent Housing Evictions?” Health Affairs, September 2019, https://www.healthaffairs.org/doi/10.1377/hlthaff.2018.05071.

[25] See Benjamin D. Sommers et al., “Changes in Utilization and Health Among Low-Income Adults After Medicaid Expansion or Expanded Private Insurance,” Journal of the American Medical Association, October 2016, http://jamanetwork.com/journals/jamainternalmedicine/article-abstract/2542420.

[26] Sarah Miller et al., “Medicaid and Mortality: New Evidence from Linked Survey and Administrative Data,” National Bureau of Economic Research Working Paper 2018-07, July 2019, https://www.nber.org/papers/w26081.

[27] Robert Greenstein, “Greenstein: More Americans Uninsured, Though Progress on Poverty Continues,” Center on Budget and Policy Priorities, September 10, 2019, https://www.cbpp.org/press/statements/greenstein-more-americans-uninsured-though-progress-on-poverty-continues.