The Coverage Gap: Uninsured Poor Adults in States that Do Not Expand Medicaid

Issue Brief

While millions of people have gained coverage through the expansion of Medicaid under the Affordable Care Act (ACA), state decisions not to implement the expansion leave many without an affordable coverage option. Under the ACA, Medicaid eligibility is extended to nearly all low-income individuals with incomes at or below 138 percent of poverty ($17,236 for an individual in 2019).1 This expansion fills in historical gaps in Medicaid eligibility for adults and was envisioned as the vehicle for extending insurance coverage to low-income individuals, with premium tax credits for Marketplace coverage serving as the vehicle for covering people with moderate incomes. While the Medicaid expansion was intended to be national, the June 2012 Supreme Court ruling essentially made it optional for states. As of January 2020, 14 states had not expanded their programs.2

Medicaid eligibility for adults in states that did not expand their programs is quite limited: the median income limit for parents in these states is just 40% of poverty, or an annual income of $8,532 for a family of three in 2019, and in nearly all states not expanding, childless adults remain ineligible.3 Further, because the ACA envisioned low-income people receiving coverage through Medicaid, it does not provide financial assistance to people below poverty for other coverage options. As a result, in states that do not expand Medicaid, many adults, including all childless adults, fall into a “coverage gap” of having incomes above Medicaid eligibility limits but below the poverty level, which is the lower limit for Marketplace premium tax credits (Figure 1).

Figure 1: Gap in Coverage for Adults in States that Do Not Expand Medicaid Under the ACA

This brief presents estimates of the number of people in non-expansion states who could be reached by Medicaid if their states adopted the expansion, and discusses the implications of them being left out of ACA coverage expansions. An overview of the methodology underlying the analysis can be found in the Data and Methods, and more detail is available in the Technical Appendices.

How Many Uninsured People Who Could Have Been Eligible for Medicaid Are in the Coverage Gap?

Nationally, more than two million4 poor uninsured adults fall into the “coverage gap” that results from state decisions not to expand Medicaid (Table 1), meaning their income is above current Medicaid eligibility but below the lower limit for Marketplace premium tax credits. These individuals would be eligible for Medicaid had their state chosen to expand coverage. Reflecting limits on Medicaid eligibility outside ACA pathways, most people in the coverage gap (76%) are adults without dependent children.5

Adults left in the coverage gap are spread across the states not expanding their Medicaid programs but are concentrated in states with the largest uninsured populations. A third of people in the coverage gap reside in Texas, which has both a large uninsured population and very limited Medicaid eligibility (Figure 2). Seventeen percent live in Florida, eleven percent in Georgia, and eight percent in North Carolina. There are no uninsured adults in the coverage gap in Wisconsin because the state is providing Medicaid eligibility to adults up to the poverty level under a Medicaid waiver.

The geographic distribution of the population in the coverage gap reflects both population distribution and regional variation in state take-up of the ACA Medicaid expansion. The South has relatively higher numbers of poor uninsured adults than in other regions, has higher uninsured rates and more limited Medicaid eligibility than other regions, and accounts for the majority (9 out of 14) of states that opted not to expand Medicaid.6 As a result, more than nine in ten people in the coverage gap reside in the South (Figure 2).

Figure 2: Distribution of Adults in the Coverage Gap, by State and Region, 2018

What Would Happen if All States Expanded Medicaid?

If states that are currently not expanding their programs adopt the Medicaid expansion, all of the 2.3 million adults in the coverage gap would gain Medicaid eligibility. In addition, 2.1 million uninsured adults with incomes between 100 and 138% of poverty7 (most of whom are currently eligible for Marketplace coverage) would also gain Medicaid eligibility (Figure 3 and Table 1). Though most of these adults are eligible for substantial tax credits to purchase Marketplace coverage,8 Medicaid coverage would likely provide more comprehensive benefits and lower premiums or cost-sharing than they would face under Marketplace coverage. For example, research from early implementation of the ACA showed that coverage of behavioral health services, prescription drugs, rehabilitative and habilitative services, and long-term services and supports may be more limited in the Marketplace compared to Medicaid.9,10 In addition, research examining the population with incomes between 100-138% FPL in expansion and non-expansion states finds that Medicaid expansion coverage produced far greater reductions than subsidized Marketplace coverage in average total out-of-pocket spending, average out-of-pocket premium spending, and average cost-sharing spending.11

Figure 3: Nonelderly Uninsured Adults in Non-Expansion States Who Would Be Eligible for Medicaid if Their States Expanded, 2018

A smaller number (about 418,000) of uninsured adults in non-expansion states are already eligible for Medicaid under eligibility pathways in place before the ACA. If all states expanded Medicaid, those in the coverage gap and those who are instead eligible for Marketplace coverage would bring the number of nonelderly uninsured adults eligible for Medicaid to more than 4.8 million people in the fourteen current non-expansion states. The potential scope of Medicaid varies by state (Table 1).

Discussion

The ACA Medicaid expansion was designed to address the high uninsured rates among low-income adults, providing a coverage option for people with limited access to employer coverage and limited income to purchase coverage on their own. In states that expanded Medicaid, millions of people gained coverage, and the uninsured rate dropped significantly as a result of the expansion.12 However, with many states opting not to implement the Medicaid expansion, millions of uninsured adults remain outside the reach of the ACA and continue to have limited options for affordable health coverage. From 2017 to 2018, non-expansion states saw a significant increase in their uninsured rate, while expansion states did not.13

By definition, people in the coverage gap have limited family income and live below the poverty level. They are likely in families employed in very low-wage jobs, employed part-time, or with a fragile or unpredictable connection to the workforce. Given limited offer rates of employer-based coverage for employees with these work characteristics, it is likely that they will continue to fall between the cracks in the employer-based system.

It is unlikely that people who fall into the coverage gap will be able to afford ACA coverage, as they are not eligible for premium subsidies: in 2020, the national average unsubsidized premium for a 40-year-old non-smoking individual purchasing coverage through the Marketplace was $442 per month for the lowest-cost silver plan and $331 per month for a bronze plan,14 which equates to nearly eighty percent of income for those at the lower income range of people in the gap (below 40% FPL) and nearly a third of income for those at the higher income range of people in the gap.

If they remain uninsured, adults in the coverage gap are likely to face barriers to needed health services or, if they do require and receive medical care, potentially serious financial consequences. While the safety net of clinics and hospitals that has traditionally served the uninsured population will continue to be an important source of care for the remaining uninsured under the ACA, this system has been stretched in recent years due to increasing demand and limited resources.

Most people in the coverage gap live in the South, leading state decisions about Medicaid expansion to exacerbate geographic disparities in health coverage. In addition, because several states that have not expanded Medicaid have large populations of people of color, state decisions not to expand their programs disproportionately affect people of color, particularly Black Americans.15 As a result, state decisions about whether to expand Medicaid have implications for efforts to address disparities in health coverage, access, and outcomes among people of color.

There is no deadline for states to opt to expand Medicaid under the ACA, and debate continues in some states about whether to expand. For example, legislatures in Kansas and Wyoming are likely to take up the issue in the upcoming 2020 session.16 Further, initiatives in several states, including Missouri, Oklahoma, and South Dakota, may put the question of Medicaid expansion on the ballot in upcoming elections. The three states (Idaho, Nebraska, and Utah) that adopted the Medicaid expansion via ballot initiative in the November 2018 election all plan to implement expansion in 2020 with state Medicaid waiver proposals that condition the scope and structure of expansion. The Trump Administration has indicated to states that it is open to these types of proposals, which may lead additional states to consider extending coverage. However, some proposed waivers that could expand coverage for some people in the coverage gap also place new restrictions or requirements on that coverage.17 Thus, it is uncertain what insurance options, if any, adults in the coverage gap may have in the future, and these adults are likely to remain uninsured without policy action to develop affordable coverage options.

Table 1: Uninsured Adults in Non-Expansion States Who Would Be Eligible for
Medicaid if Their States Expanded, by Current Eligibility for Coverage, 2018
State Total Currently Eligible
for Medicaid
Currently in the Coverage Gap
(<100% FPL)
Currently May Be Eligible for Marketplace Coverage
(100%-138% FPL**)
All States Not Expanding Medicaid 4,850,000 418,000 2,324,000 2,108,000
Alabama 242,000 17,000 134,000 91,000
Florida 846,000 42,000 391,000 414,000
Georgia 518,000 44,000 255,000 219,000
Kansas 87,000 7,000 40,000 40,000
Mississippi 186,000 16,000 100,000 70,000
Missouri 217,000 13,000 113,000 92,000
North Carolina 389,000 32,000 194,000 163,000
Oklahoma 197,000 20,000 95,000 82,000
South Carolina 214,000 20,000 101,000 93,000
South Dakota 35,000 5,000 14,000 16,000
Tennessee 260,000 39,000 117,000 103,000
Texas 1,553,000 99,000 761,000 693,000
Wisconsin* 88,000 64,000 0 24,000
Wyoming 18,000 N/A 9,000 7,000
NOTES: * Wisconsin provides Medicaid eligibility to adults up the poverty level under a Medicaid waiver. As a result, there is no one in the coverage gap in Wisconsin. ** The “100%-138% FPL” category presented here uses a Marketplace eligibility determination for the lower bound (100% FPL) and a Medicaid eligibility determination for the upper bound (138% FPL) in order to appropriately isolate individuals within the range of potential Medicaid expansions but also with sufficient resources to avoid the coverage gap. Totals may not sum due to rounding. N/A: Sample size too small for reliable estimate.
SOURCE: KFF analysis based on 2019 Medicaid eligibility levels and 2018 American Community Survey.

Data and Methods

This analysis uses data from the 2018 American Community Survey (ACS). The ACS provides socioeconomic and demographic information for the United States population and specific subpopulations. Importantly, the ACS provides detailed data on families and households, which we use to determine income and household composition for ACA eligibility purposes.

Medicaid and Marketplaces have different rules about household composition and income for eligibility. The ACS questionnaire captures the relationship between each household resident and one household reference person, but not necessarily each individual to all others. Therefore, prior to estimating eligibility, we implement a series of logical rules based on each person’s relationship to that household reference person in order to estimate the person-to-person relationships of all individuals within a respondent household to one another. We then assess income eligibility for both Medicaid and Marketplace subsidies by grouping individuals into household insurance units (HIUs) and calculate HIU income using the rules for each program. For more detail on how we construct person-to-person relationships, aggregate Medicaid and Marketplace households, and then count income, see the detailed Technical Appendix A.

Undocumented immigrants are ineligible for federally-funded Medicaid and Marketplace coverage. Since ACS data do not directly indicate whether an immigrant is lawfully present, we draw on the methods underlying the 2013 analysis by the State Health Access Data Assistance Center (SHADAC) and the recommendations made by Van Hook et. al.1,2 This approach uses the Survey of Income and Program Participation (SIPP) to develop a model that predicts immigration status; it then applies the model to ACS, controlling to state-level estimates of total undocumented population from Pew Research Center. For more detail on the immigration imputation used in this analysis, see the Technical Appendix B.

Individuals in tax-filing units with access to an affordable offer of Employer-Sponsored Insurance (ESI) are still potentially MAGI-eligible for Medicaid coverage, but they are ineligible for advance premium tax credits in the Health Insurance Exchanges. Since ACS data do not designate policyholders of employment-based coverage nor indicate whether workers hold an offer of ESI, we developed a model that predicts both the policyholder and the offer of ESI based on the Current Population Survey (CPS). Additionally, for families with a Marketplace eligibility level below 250% FPL, we assume any reported worker offer does not meet affordability requirements and therefore does not disqualify the family from Tax Credit eligibility on the Exchanges. For more detail on the offer imputation used in this analysis, see the Technical Appendix C.

As of January 2014, Medicaid financial eligibility for most nonelderly adults is based on modified adjusted gross income (MAGI). To determine whether each individual is eligible for Medicaid, we use each state’s reported eligibility levels as of January 1, 2019, updated to reflect state Medicaid expansion decisions as of January 2020 and 2018 Federal Poverty Levels.3 Some nonelderly adults with incomes above MAGI levels may be eligible for Medicaid through other pathways; however, we only assess eligibility through the MAGI pathway.4

An individual’s income is likely to fluctuate throughout the year, impacting his or her eligibility for Medicaid. Our estimates are based on annual income and thus represent a snapshot of the number of people in the coverage gap at a given point in time. Over the course of the year, a larger number of people are likely to move and out of the coverage gap as their income fluctuates.

Starting with our of estimates of ACA eligibility in 2017, we transferred our core modeling approach from relying on the Current Population Survey (CPS) Annual Social and Economic Supplement (ASEC) to the American Community Survey (ACS). ACS includes a 1% sample of the US population and allows for precise state-level estimates as well as longer trend analyses. Since our methodology excludes a small number of individuals whose poverty status could not be determined, our ACS-based population totals appear slightly below CPS-based totals and some ACS population totals published by the Census Bureau. This difference is in large part attributable to students who reside in college dormitories. Comparing the two survey designs, CPS counts more of these individuals in the household of their parent(s) than ACS does.

Technical Appendix A: Household Construction

In KFF’s estimates of eligibility for ACA coverage, income eligibility for both Medicaid and Marketplace subsidies is assessed by grouping people into “health insurance units” (HIUs) and calculating HIU income according to Medicaid and Marketplace program rules. HIUs group people according to how they are counted for eligibility for health insurance, versus grouping people according to who they live with (e.g., “households”) or are related to (e.g., “families”). HIU construction is an important step in assessing income as a share of the federal poverty line (FPL) because it impacts whose income is counted (and thus the total income for the unit) and how many people share that income (and thus the corresponding FPL to use for comparison, since FPL varies by family size). Our HIUs are designed to match ACA eligibility rules for both Medicaid and Marketplaces. Below we describe how we construct HIUs for this analysis. The programming code, written using the statistical computing package R v.3.6.1, is available upon request for people interested in replicating this approach for their own analysis.

Person to Person Relationships

We construct spousal and parent-to-child person-to-person linkage variables within each household of the microdata. The American Community Survey (ACS) includes only the relationship of each person in a household to one central reference person. Using the household reference person’s known relationships to all other individuals within each household, we iterate through every pair of individuals present in each household to determine probable person to person links for possible mother, father, and spousal pairs. Our approach to determining probable family interrelationship linkages closely follows the construction documented by IPUMS-USA with the notable exception of unmarried partner relationships.1 We intentionally diverge from IPUMS-USA because the presence of an unmarried partner relationship does not impact federal program eligibility. Among individuals designated as married with a spouse present in the household, our constructed spousal pointer matches the IPUMS SPLOC variable 99% of the time in the 2013 microdata. Our construction of mother and father pointers match the IPUMS MOMLOC, POPLOC, MOMLOC2, and POPLOC2 variables for more than 99% of all person-records.

Family Aggregation

Separate from person-to-person linkage variables, we assemble individual records into family units reproducing the Census Bureau’s Family Poverty Ratio (POVPIP) variable. Although the Census Bureau does not include a unique family identifier on the ACS microdata, we approximate the groupings used to generate the ACS income-to-poverty ratio variable with the following steps:

  1. Both non-relatives of the household reference person (RELP of 11-17) and all individuals in non-family households (HHT of 4-7) are categorized as single-person families.
  2. Married couple and other family households without subfamilies (PSF of 0) are categorized into single-family households.
  3. Married couple and other family households with subfamilies (PSF of 1) are categorized based on their subfamily number (SFN).

This family identifier then informs estimates of family-wide statistics, such as the percent of the uninsured Americans in a family below poverty or the count of Medicaid-enrollees with one or more workers in their family. This family aggregation matches the groupings used to determine the income-to-poverty ratio variable, and estimates of health insurance presented by family poverty categories align with Census Bureau publications based on the ACS.2 Since many family members obtain health coverage separately from one another (for example, an elderly parent cohabiting with their working-age child might hold Medicare coverage and Employer Sponsored Insurance, respectively), descriptive statistics focused on family attributes rely on this family identifier but Medicaid and Marketplace eligibility determinations do not.

Overview of KFF-HIUS

We construct two different HIUs for everyone in the sample: a Medicaid HIU and a Marketplace HIU. We use two HIUs because the rules for counting families and income differ between the two programs. For example, in Medicaid, children with unmarried parents have both parents’ income counted toward their income, whereas under Marketplace rules, only the income of the parent who claims the child on his/her taxes counts. In another example, certain tax dependents (e.g., a parent) are treated differently for Medicaid eligibility than they are for Marketplace eligibility. To account for these rules, we developed an algorithm for sorting people into HIUs. We construct HIUs and HIU incomes separately for each person in a household and take into account the family relationships and income of the other people in the person’s household. People in the same household or in the same family may not have the same HIU composition or income for determining either Medicaid eligibility or eligibility for tax credits.

In simplest terms, the HIU algorithm sorts people into tax filing units. For all people in the data set, the algorithm assesses whether they are likely to be a tax filer themselves and, if so, who they are likely to claim or, if not, who is likely to claim them. It also captures whether someone is neither a tax filer nor claimed as a dependent by someone else. Importantly, the HIU construction considers all relationships for each person within the household. This step is particularly important in correctly classifying people in non-nuclear families (e.g., households with more than one generation, with unmarried partners, or with relatives outside the nuclear family such as an aunt or uncle), which may contain either one or multiple tax filing units.

In counting income for both Medicaid and Marketplace HIUs, we use modified adjusted gross income (MAGI), corresponding to the ACA rules. MAGI differs from total income in that some sources of income (e.g., cash assistance payments from TANF or SSI) do not count toward MAGI. We calculate HIU income as a share of poverty using the Health and Human Services Poverty Guidelines.3

For a small number of people, Medicaid HIU income as a share of poverty does not match Marketplace HIU income as a share of poverty due to the different rules between the programs. This analysis first calculates Medicaid HIU and classifies anyone who meets Medicaid eligibility into that category (including most individuals below 138% FPL in the Medicaid expansion states). We then calculate Marketplace HIU; anyone meeting subsidy eligibility is grouped into that category (above Medicaid and also above 100% FPL up to 400% FPL for most individuals). This approach follows the eligibility rules in the ACA, which specify that people are eligible for tax credits only if they are ineligible for Medicaid.

Steps in Calculating KFF-HIUS

Before we group people into HIUs, we first calculate annual MAGI for each respondent. We compare each person’s income to IRS filing requirements for being a tax filer4 and for being a qualifying relative claimed by someone else.5

We then group people into HIUs. We begin this process by grouping everyone within a household who is related into “cohabitating families.” Cohabitating families include all family relations; they also include unmarried cohabitating partners and relatives of each cohabitating partner.

Within each cohabitating family, we assess whether any individual is eligible to claim any other individual as a tax dependent. People are eligible to claim others as tax dependents if their income is above the IRS filing threshold for a head of household or, if married, for a married couple. People are eligible to be claimed by others if (a) they are a child (under age 19 or, for tax credits, 23 if a full-time student), and someone else in the cohabitating family has at least twice their income, or (b) they are below the limit to be a tax filer, have income below the qualifying relative limit, and someone else in the cohabitating family has at least twice their income. Within each cohabitating family, we assess who is likely to claim whom, using the assumptions that:

  • People who are claimed by others are more likely to be claimed by close relatives (e.g., a parent) than by others (e.g., a grandparent).
  • Married couples (who file) file jointly
  • If more than one person in a cohabitating family is eligible to claim others within that cohabitating family, the wealthiest person claims the eligible dependents.

Once we determine who within the cohabitating family is likely to claim each other, we know the HIU size and are able to apply income rules for the HIU. We apply Medicaid and Marketplace rules for whose income counts in calculating Medicaid HIUs and Marketplace HIUs, respectively.6 People who are filers but are not eligible to claim someone else or to be claimed by someone else are an HIU of 1. People who are not filers and are not claimed by filers have their HIU size and income counted according to Medicaid non-filer rules.7

Inflation Factors

In order to determine ACA eligibility during calendar year 2018, we compared tax filing unit income against the most current premiums available, for open enrollment 2020.8 We relied on the Bureau of Labor Statistics Employment Cost Index (ECI), Private Wages and Salaries to inflate the income of each HIU by approximately 6.7% to align 2018 incomes to 2020 premiums.9 Since most state Medicaid eligibility determinations through the MAGI pathway are calculated as a percent of HHS Poverty Guidelines for that year and not a fixed dollar amount, inflation was not necessary to assess the Medicaid eligibility of individuals.

After inflating 2018 tax filing unit incomes to match 2020 premiums, we similarly inflated 2018 IRS thresholds for both filing requirements10 and for qualifying relative tests11 by the same factor so that these thresholds aligned with the inflated income amounts.

Limitations

As with any analysis, there are some limitations to our approach due to the level of detail that we can obtain from available survey data. Key limitations to bear in mind include:

  • We currently are not able to appropriately group anyone who lives outside the household with a household that claims them as a tax dependent. For example, we are not able to connect students living away from home or children with a non-custodial parent with the people who may be claiming them (and whose income should count to their HIU). We are also not able to determine married people who file separately.
  • To group people into tax filing units, we have to make assumptions about how people are likely to file their taxes. We assume that tax filers claim qualifying relatives they are able to claim. We make this assumption based on the fact that Medicaid and Marketplace eligibility rules are determined not by who is actually claimed on the tax return but by who is allowed to be claimed. However, people may sort themselves into different tax filing units than we estimate.

Technical Appendix B: Immigration Status Imputation

To impute documentation status, we draw on the methods underlying the 2013 analysis by the State Health Access Data Assistance Center (SHADAC) and the recommendations made by Van Hook et. al..1,2  This approach uses the Survey of Income and Program Participation (SIPP) to develop a model that predicts immigration status for each person in the sample; it then applies the model to a second data source, controlling to state-level estimates of total undocumented population as well as the undocumented population in the labor force from the Pew Research Center.3 Below we describe how we developed the regression model and applied it to the American Community Survey (ACS). We also describe how the model may be applied to other data sets. The programming code, written using the statistical computing package R v.3.6.1, is available upon request for people interested in replicating this approach for their own analysis.

Data Sources

We used the second wave of the 2008 Survey of Income and Program Participation (SIPP) panel data to build the regression model. The SIPP Wave Two dataset contains questions on migration history at the person level.

The regression model is designed to be applied to other datasets in order to impute legal immigration status in surveys that do not ask about migration status. The code mentioned above includes programming to apply the model to either the SIPP Core file, ACS, or the Current Population Survey (CPS). Because the SIPP Core file contains different survey questions and variable specifications from the ACS and CPS, we create unique regression models to apply the model to each dataset. For the analysis underlying this brief and other KFF estimates of eligibility for ACA coverage, we apply the regression model to the 2013 ACS and then each subsequent year of the ACS.

Due to underreporting of legal immigration status in the SIPP, in imputing immigration status we control to state and national-level estimates of the total undocumented population and also the undocumented population in the labor force from the Pew Research Center. Pew reports these estimates for all states and the District of Columbia.4

Construction of Regression Model

We use the SIPP Wave Two to create a binomial, dependent variable that identifies a respondent as a potential unauthorized immigrant. The dependent variable is constructed based on the following factors:

  1. Respondent was not a United States (US) citizen,
  2. Respondent did not have permanent resident status upon US entry,
  3. Respondent’s immigration status did not change to permanent resident since US entry, and
  4. Respondent does not have other indicators that imply legal status.5

We use the following independent variables to predict unauthorized immigrant status:

  1. Year of US entry,
  2. Job industry classification,
  3. State of residence,
  4. Family Poverty Level,
  5. Ownership or rental of residence,
  6. Presence of at least one citizen in household,
  7. Number of occupants in the household (< or >= six occupants),
  8. Whether all household occupants are related,
  9. Number of workers in household,
  10. Health insurance coverage status,
  11. Sex, and
  12. Ethnicity.

The regression model was sub-populated to remove respondents who could not be considered unauthorized. People who could not be considered unauthorized include people who 1) were born in the US, 2) are US citizens, or 3) have other indicators that imply legal status.

Imputing Unauthorized Immigrants in Other Datasets

We use the Pew estimates as targets for the total number of unauthorized immigrants that the imputation generates. We first apply this strategy to the 2013 ACS, which contains health insurance information prior to the ACA’s coverage expansions. We stratify the targets by state and the District of Columbia and by participation in the labor force. We impute immigration status within each of these 102 strata.6

To generate the imputed immigration status variable, we first calculated the probability that each person in the dataset was unauthorized based on the SIPP regression model. Next, we isolated the dataset to each individual stratum described above. Within each stratum, we sampled the data using the probability of being unauthorized for each person. After sampling, we summed the person weights until reaching the Pew population estimate for each stratum. The records that fell within the Pew population estimate were considered to be unauthorized immigrants. We repeated the process of sampling using the probability of being unauthorized and subsequently summing the person weights to reach Pew targets five times, creating five different unauthorized variables per record. These five imputed authorization status variables were then incorporated into a standard multiple imputation algorithm, closely matching the imputed variable analysis techniques used by the Centers for Disease Control and Prevention for the National Health Interview Survey.7

We used this first pass on the ACS 2013 to inform our sampling targets for the latest available microdata (ACS 2018). Looking at the results of our undocumented imputation on the ACS 2013, we calculated the share of undocumented immigrants lacking health insurance within each of those 102 strata prior to the ACA’s coverage expansions and transferred that information into a new dimension of sampling strata for the ACS 2018. We split each of the 102 sampling strata used on the pre-ACA ACS 2013 into uninsured versus insured categories, resulting in 204 sampling strata for subsequent years. We then repeated our imputation on the ACS 2018 with the newly-divided strata, allowing for a small decline in the undocumented uninsured rate based off of the percent drop in the uninsured rate we see in the Kaiser Family Foundation’s Survey of the Low-Income Population and the ACA.8

To easily apply the regression model to other data sets, we created a function that applies this approach to a chosen data set. The function first loads the dataset of choice, then standardizes the data to match the independent variables from the SIPP regression model, and finally applies the multiple imputation to generate a variable for legal immigration status.

Technical Appendix C: Imputation of Offer of Employer-Sponsored Insurance

An integral part of determining ACA eligibility is assessing whether workers without employer-sponsored insurance (ESI) hold an offer through their workplace that they decline to take up. In most cases, an affordable offer of ESI disqualifies members of the tax filing unit of the worker from receiving subsidized coverage on the ACA Health Insurance Marketplace. The American Community Survey (ACS) does not ask about employer offers of ESI; however, the Current Population Survey Annual Social and Economic Supplement (CPS-ASEC) includes questions about whether each worker received an offer of ESI from his or her employer at the time of interview. We use the CPS-ASEC offer of ESI variable to inform a regression-based multiple imputation of whether each tax filing unit constructed in the ACS had at least one affordable offer at work.

The September 2019 release of the CPS-ASEC incorrectly removed the variable indicating whether workers holding dependent ESI (most often through a spouse) were offered ESI themselves through their own workplace. This issue, noted in Appendix J, User Note #6 of the Technical Documentation, prevented us from using the that data in our offer imputation analysis of 2018 ACS data. We therefore relied on the 2018 CPS-ASEC microdata to inform an imputation of offers of ESI onto both the 2017 and 2018 years of ACS microdata. Although the methods discussed below describe our application of the 2018 CPS-ASEC imputation model onto the 2017 ACS, we applied each of these steps onto the 2018 ACS as well. This repetition relied on the assumption that the characteristics that predict workplace offers of ESI (age, earnings, full-time/part-time status, and current health coverage sources within the tax filing unit) did not change between 2017 and 2018. While an imputation based on lagged data may introduce error due to annual trends between the years, we have little evidence indicating a change in ESI offer rates across the period.1 Thus, our use of lagged ESI offer data should not be interpreted as a new source of bias nor a meaningful loss of accuracy.

Since the health insurance coverage variables available in the CPS-ASEC 2018 capture sources of coverage at any point during calendar year 2017 (versus at the time of survey, as with the offer rate variable), a subset of sampled individuals had a change in their employer-based coverage status across the two distinct time periods.2 Therefore, among workers who potentially experienced a shift in offer status across the two time periods, we recoded or imputed offer rates in 2017 using the offer status in 2018. After constructing this revised offer variable for workers in CPS, we aggregated the results at the tax filing unit level to create a prediction model to apply to the ACS. Below we describe these recodes and imputation. The programming code, written using the statistical computing package R v.3.6.1, is available upon request for people interested in replicating this approach for their own analysis.

Recoding and Imputing Offer Rate Data in the CPS

As a first step in our analysis, we divided CPS-ASEC survey respondents into five distinct groups:

  1. All individuals who did not work during 2017 and also did not hold an offer of ESI in 2018 were assumed not to have an offer in 2017.
  2. All individuals who reported being an ESI policyholder (that is, anyone reporting having taken-up their offer of ESI) during 2017 and also reported holding an offer of ESI during early 2018 were assumed to have an offer in 2017.
  3. All workers in 2017 who held their own ESI policies during 2017 but then reported not holding an offer during 2018 were re-coded as holding an offer of ESI in 2017.
  4. All non-workers during 2017 who reported holding an offer during 2018 were re-coded as not holding an offer of ESI in 2017.
  5. Some workers during 2017 who did not report being ESI policyholders but did report holding an offer of ESI during early 2018 were imputed to not have an offer of ESI during 2017.

For many groups, including those in groups (1) and (2) listed above, the offer status did not change across the two time periods. In contrast, we recoded offer status for people in groups (3) and (4): every non-offered worker in group (3), which includes people who held ESI policies in their own name in 2017, were considered to have an offer of ESI in 2017, and offered workers in group (4), which includes people who did not work themselves in 2017, were considered to not have their offer of ESI in 2017. Last, we implemented a probability-based random sample imputation of offers of ESI for people in group (5), described in more detail below. Only a subset of the group was re-coded from holding an offer in 2018 to not holding an offer in 2017. The number of workers selected from this population was equal to the population size of (3) subtracted by the population size of (4), thereby assuming an unchanging offer rate for the total worker population across the period.

Imputing Offer Rates for CPS Respondents with Ambiguous Offer Rate Status

The CPS-ASEC worker-level regression model was designed to be applied to a single dataset where ESI offer status is known at one point in time but not another. The code mentioned above includes programming to apply the model to the Current Population Survey (CPS-ASEC) (for years 2014 on). For the analysis underlying KFF’s current estimates of ACA eligibility, we apply the regression model to workers in the 2018 CPS-ASEC.

We use the 2018 point-in-time worker offer variable provided by the US Census Bureau3 to create a binomial, dependent variable that identifies a respondent as a recipient of an offer of employer-sponsored insurance at his or her workplace in early 2018. The dependent variable was constructed at the worker-level based on individuals not holding their own ESI policy at time of interview and also reporting an ESI offer or eligibility to be covered that was then voluntarily declined.

We use the following independent variables to predict offer status in 2017 among workers not covered by their own ESI during both 2017 and early 2018 but potentially holding an offer of ESI in 2017:

  • Any public coverage,
  • Any nongroup coverage,
  • Worker earnings among all jobs,
  • Full-time versus part-time status,
  • Age of worker,
  • Work within the construction industry.

The regression model was sub-populated to remove respondents already covered by their own ESI and also to remove non-workers. Since this imputation does not account for the affordability of the offer or whether it meets the minimum value test, we included an assumption that workers in tax filing units with a MAGI below 250% FPL do not hold affordable offers of ESI and therefore might be eligible to purchase subsidized coverage on the Exchanges.4

As mentioned above, we assume an unchanging offer rate for the total worker population across the two time periods. We determined the needed size of the population to impute by subtracting the population of (4) from the population of (3) to ensure an equivalent number of offers were gained and lost. This left only workers who reported holding an offer of ESI during early 2018, since (3) represented a larger count of workers than (4). We then calculated the probability that each worker in the dataset was offered ESI during calendar year 2017 based on our 2018 CPS-ASEC regression model. Next, we selected workers within the potential population (5) using the sampling probabilities resultant from our model.

Construction and Application of ACS Regression Model

For the analysis underlying KFF estimates of ACA eligibility, we construct a prediction model of having an offer of ESI using the 2018 CPS-ASEC and then apply this regression to tax filing units in the 2017 ACS to estimate who has an ESI offer in ACS.

We aggregate the worker offer variables constructed the 2018 CPS-ASEC as described above to create a binomial, dependent variable that identifies each tax filing unit as either holding or not holding an affordable offer of employer-sponsored insurance.

We use the following independent variables to predict offer status among tax filing units:

  • Any senior citizen in the household,
  • Oldest member of the tax-filing unit,
  • Any member of the tax-filing unit has employer-sponsored insurance coverage,
  • Any member of the tax-filing unit has nongroup coverage,
  • Any uninsured individuals in the tax filing unit,
  • Share of adults working full-time and part-time, and
  • Highest worker earnings.

Since the imputation of documentation status (discussed in Technical Appendix B) required a multiply-imputed approach, this secondary imputation and subsequent worker sampling was only conducted once per implicate, keeping the number of ACS implicates to five.

Endnotes

Issue Brief

  1. U.S. Department of Health and Human Services, Office of The Assistant Secretary for Planning and Evaluation, 2019 Poverty Guidelines. Available at: https://aspe.hhs.gov/poverty-guidelines.

    ← Return to text

  2. Kaiser Family Foundation State Health Facts, “Status of State Action on the Medicaid Expansion Decision,” accessed January 2020, http://kff.org/health-reform/state-indicator/state-activity-around-expanding-medicaid-under-the-affordable-care-act/.

    ← Return to text

  3. Of the states not moving forward with the expansion, only Wisconsin provides full Medicaid coverage to adults without dependent children. For state-by-state information on Medicaid eligibility, see The Kaiser Family Foundation State Health Facts. “Medicaid Income Eligibility Limits for Adults as a Percent of the Federal Poverty Level.” Data Source: Based on state-reported eligibility levels as of January 1, 2019, collected through a national survey conducted by the Kaiser Commission on Medicaid and the Uninsured with the Georgetown University Center for Children and Families. Available at: http://kff.org/health-reform/state-indicator/medicaid-income-eligibility-limits-for-adults-as-a-percent-of-the-federal-poverty-level/.

    ← Return to text

  4. National and state-by-state estimates of the number of people in the coverage gap may change from year to year due to several factors, including differences in the underlying data, small changes in state Medicaid eligibility, and declines in the number of uninsured people by state as economic conditions change.

    ← Return to text

  5. Kaiser Family Foundation analysis of 2018 American Community Survey (ACS), 1-Year Estimates.

    ← Return to text

  6. Kaiser Family Foundation analysis of the 2018 American Community Survey (ACS), 1-Year Estimates.

    ← Return to text

  7. The "100%-138% FPL" category presented here uses a Marketplace eligibility determination for the lower bound (100% FPL) and a Medicaid eligibility determination for the upper bound (138% FPL) in order to appropriately isolate individuals within the range of potential Medicaid expansions but also with sufficient resources to avoid the coverage gap.

    ← Return to text

  8. The vast majority of these people are eligible for tax credits to subsidize the cost of coverage in the Marketplace, though some (e.g., people with an offer of employer coverage) may not qualify for tax credits.

    ← Return to text

  9. Ken Cannon, Jenna Burton, and MaryBeth Musumeci, Adult Behavioral Health Benefits in Medicaid and the Marketplace (Washington, D.C.: Kaiser Family Foundation, June 11, 2015), https://www.kff.org/medicaid/report/adult-behavioral-health-benefits-in-medicaid-and-the-marketplace/.

    ← Return to text

  10. MaryBeth Musumeci, Julia Paradise, Erica L. Reaves, and Henry Claypool, Benefits and Cost-Sharing for Working People with Disabilities in Medicaid and the Marketplace (Washington, D.C.: Kaiser Family Foundation, October 15, 2014), https://www.kff.org/medicaid/issue-brief/benefits-and-cost-sharing-for-working-people-with-disabilities-in-medicaid-and-the-marketplace/.

    ← Return to text

  11. Larisa Antonisse, Rachel Garfield, Robin Rudowitz, and Madeline Guth, The Effects of Medicaid Expansion under the ACA: Updated Findings From a Literature Review (Washington, D.C.: Kaiser Family Foundation, August 2019), https://www.kff.org/medicaid/issue-brief/the-effects-of-medicaid-expansion-under-the-aca-updated-findings-from-a-literature-review-august-2019/.

    ← Return to text

  12. Ibid.

    ← Return to text

  13. Jennifer Tolbert, Kendal Orgera, Natalie Singer, and Anthony Damico, Key Facts about the Uninsured Population (Washington, D.C.: Kaiser Family Foundation, December 2019), https://www.kff.org/uninsured/issue-brief/key-facts-about-the-uninsured-population/.

    ← Return to text

  14. The methods for arriving at this estimate can be found on the Kaiser Family Foundation Subsidy Calculator and the Kaiser Family Foundation “Change in Average Marketplace Premiums by Metal Tier”, available here: http://www.kff.org/interactive/subsidy-calculator/ and https://www.kff.org/health-reform/state-indicator/change-in-average-marketplace-premiums-by-metal-tier/.

    ← Return to text

  15. Samantha Artiga, Kendal Orgera, and Anthony Damico, Changes in Health Coverage by Race and Ethnicity since Implementation of the ACA, 2013-2017 (Washington, D.C.: Kaiser Family Foundation, February 2019), https://www.kff.org/disparities-policy/issue-brief/changes-in-health-coverage-by-race-and-ethnicity-since-implementation-of-the-aca-2013-2017/.

    ← Return to text

  16. Kaiser Family Foundation, Status of State Medicaid Expansion Decisions: Interactive Map, January 2020, https://www.kff.org/medicaid/issue-brief/status-of-state-medicaid-expansion-decisions-interactive-map/.

    ← Return to text

  17. Kaiser Family Foundation, “Medicaid Waiver Tracker: Approved and Pending Section 1115 Waivers by State,” (Washington, DC, Kaiser Family Foundation, December 2019), https://www.kff.org/medicaid/issue-brief/medicaid-waiver-tracker-approved-and-pending-section-1115-waivers-by-state/.    

    ← Return to text

Data and Methods
  1. State Health Access Data Assistance Center. 2013. “State Estimates of the Low-income Uninsured Not Eligible for the ACA Medicaid Expansion.” Issue Brief #35. Minneapolis, MN: University of Minnesota. Available at: http://www.rwjf.org/content/dam/farm/reports/issue_briefs/2013/rwjf404825.

    ← Return to text

  2. Van Hook, J., Bachmeier, J., Coffman, D., and Harel, O.  2015. “Can We Spin Straw into Gold? An Evaluation of Immigrant Legal Status Imputation Approaches” Demography. 52(1):329-54.

    ← Return to text

  3. Based on state-reported eligibility levels as of January 1, 2019. Eligibility levels are updated to reflect state implementation of the Medicaid expansion as of January 2020 and 2018 Federal Poverty Levels but may not reflect other eligibility policy changes since January 2019. The Kaiser Family Foundation State Health Facts. Data Source: Kaiser Family Foundation with the Georgetown University Center for Children and Families. Medicaid and CHIP Eligibility, Enrollment, Renewal, and Cost Sharing Policies as of January 2019: Findings from a 50-State Survey, (Washington, DC: Kaiser Family Foundation, March 27, 2019), Available at: https://www.kff.org/medicaid/report/medicaid-and-chip-eligibility-enrollment-and-cost-sharing-policies-as-of-january-2019-findings-from-a-50-state-survey/.

    ← Return to text

  4. Non-MAGI pathways for nonelderly adults include disability-related pathways, such as SSI beneficiary; Qualified Severely Impaired Individuals; Working Disabled; and Medically Needy. We are unable to assess disability status in the ACS sufficiently to model eligibility under these pathways. However, previous research indicates high current participation rates among individuals with disabilities (largely due to the automatic link between SSI and Medicaid in most states, see Kenney GM, V Lynch, J Haley, and M Huntress. “Variation in Medicaid Eligibility and Participation among Adults: Implications for the Affordable Care Act.” Inquiry. 49:231-53 (Fall 2012)), indicating that there may be a small number of eligible uninsured individuals in this group. Further, many of these pathways (with the exception of SSI, which automatically links an individual to Medicaid in most states) are optional for states, and eligibility in states not implementing the ACA expansion is limited. For example, the median income eligibility level for coverage through the Medically Needy pathway is 38% of poverty in states that are not expanding Medicaid and that use this eligibility pathway. (See: MACPAC, Medicaid Income Eligibility Levels as a Percentage of the FPL for Individuals Age 65 and Older and Persons with Disabilities by State, 2019. Available at: https://www.macpac.gov/wp-content/uploads/2015/01/EXHIBIT-37.-Medicaid-Income-Eligibility-Levels-as-a-Percentage-of-the-Federal-Poverty-Level-for-Individuals-65-and-Older-and-Persons-with-Disabilities-by-State-2019.pdf.

    ← Return to text

Technical Appendix A: Household Construction
  1. Steven Ruggles, Sarah Flood, Ronald Goeken, Josiah Grover, Erin Meyer, Jose Pacas, and Matthew Sobek. IPUMS USA: Version 8.0 [dataset]. Minneapolis, MN: IPUMS, 2018. https://doi.org/10.18128/D010.V8.0

    For a detailed description of how IPUMS constructs family interrelationships variables, see https://usa.ipums.org/usa/chapter5/chapter5.shtml

    ← Return to text

  2. According to the Public Use Microdata Sample (PUMS) documentation, "Estimates generated with PUMS microdata will be slightly different from the pretabulated estimates for the same characteristics published on data.census.gov.  These differences are due to the fact that the PUMS files include only about two-thirds of the cases that were used to produce estimates on data.census.gov, as well as additional PUMS edits."

    ← Return to text

  3. Medicaid eligibility in 2018 is based on 2018 poverty guidelines, available at: U.S. Department of Health and Human Services, Office of The Assistant Secretary for Planning and Evaluation, Poverty Guidelines. https://aspe.hhs.gov/2018-poverty-guidelines.  Tax credit eligibility in 2018 is based on 2017 poverty guidelines, available at:  U.S. Department of Health and Human Services, Office of The Assistant Secretary for Planning and Evaluation, 2017 Poverty Guidelines. https://aspe.hhs.gov/2017-poverty-guidelines.

    ← Return to text

  4. See Internal Revenue Service, Publication 501, Table 1.2018: Filing Requirements Chart for Most Taxpayers. Available at: https://www.irs.gov/publications/p501#en_US_2018_publink1000270109.

    ← Return to text

  5. See Internal Revenue Service, Publication 501, Qualifying Relative. Available at: https://www.irs.gov/publications/p501#en_US_2018_publink1000196863.

    ← Return to text

  6. A detailed explanation of Medicaid and Marketplace income counting rules can be found in Center on Budget and Policy Priorities webinar available at: http://www.healthreformbeyondthebasics.org/wp-content/uploads/2013/08/Income-Definitions-Webinar-Aug-28.pdf

    ← Return to text

  7. A detailed explanation of Medicaid and Marketplace HIU size calculations can be found in the Center on Budget and Policy Priorities webinar available at http://www.healthreformbeyondthebasics.org/wp-content/uploads/2013/08/Household-Definitions-Webinar-7Aug13.pdf

    ← Return to text

  8. This is the same underlying data as the 2020 Health Insurance Marketplace Calculator.  Available at: http://kff.org/interactive/subsidy-calculator/  For a more detailed examination of plans available in the Health Insurance Marketplaces in 2020, see Kaiser Family Foundation, How ACA Marketplace Premiums Are Changing by County in 2020.  Available at: https://www.kff.org/health-costs/issue-brief/how-aca-marketplace-premiums-are-changing-by-county-in-2020/.

    ← Return to text

  9. See Congressional Budget Office, Economic Projections.  Available at: https://www.cbo.gov/system/files/2019-08/51135-2019-08-economicprojections_1.xlsx.

    ← Return to text

  10. See Internal Revenue Service, Publication 501, Table 1.2018: Filing Requirements Chart for Most Taxpayers. Available at: https://www.irs.gov/publications/p501#en_US_2018_publink1000270109.

    ← Return to text

  11. See Internal Revenue Service, Publication 501, Qualifying Relative. Available at: https://www.irs.gov/publications/p501#en_US_2018_publink1000196863.

    ← Return to text

Technical Appendix B: Immigration Status Imputation
  1. State Health Access Data Assistance Center. 2013. “State Estimates of the Low-income Uninsured Not Eligible for the ACA Medicaid Expansion.” Issue Brief #35. Minneapolis, MN: University of Minnesota. Available at: http://www.rwjf.org/content/dam/farm/reports/issue_briefs/2013/rwjf404825

    ← Return to text

  2. Van Hook, J., Bachmeier, J., Coffman, D., and Harel, O.  2015. “Can We Spin Straw into Gold? An Evaluation of Immigrant Legal Status Imputation Approaches”  Demography. 52(1):329-54.

    ← Return to text

  3. This data source is a change from previous KFF analyses, which used estimates from the Department of Homeland Security.

    ← Return to text

  4. Pew updates these estimates periodically. We use the most recent estimates available at the time of our analysis. We draw on Pew directly for all published data and interpolate years missing from their trend. Our analysis uses the year applicable to the year for the data sets to which we apply the regression model. The most recent estimates as of the time of our analysis were: J Passel, D Cohn. Mexicans decline to less than half the U.S. unauthorized immigrant population for the first time. (Pew Research Center), June 2019. Available at: https://www.pewresearch.org/fact-tank/2019/06/12/us-unauthorized-immigrant-population-2017/.

    ← Return to text

  5. Indicators that imply legal status include: (i) respondent entered the US prior to 1980, or (ii) respondent is enrolled in any of the following public programs: Medicare, military health insurance, public assistance, Supplemental Security Income, or Social Security Income.

    ← Return to text

  6. For more information, see SHADAC 2013, footnote 6. The table created for this function contains estimates of the undocumented across 2013-2018.

    ← Return to text

  7. For more detail, see documentation available at: National Health Interview Survey. 2018 Imputed Income Files. August, 2019. Available at: https://www.cdc.gov/nchs/nhis/nhis_2018_data_release.htm.

    ← Return to text

  8. As an example of this calculation, we found that approximately 59% of undocumented uninsured individuals did not have health coverage in 2013.  We allow the undocumented rate to drop slightly after 2013. We base the percent drop in the uninsured rate that we see in the Kaiser Family Foundation's Survey of the Low-Income Population and the ACA (which has a direct measure of citizenship) for 2013 to 2014, which is an 11% decline, to estimate an uninsured rate in 2014 for the undocumented (52%). We use the ratio of that drop relative to the drop for citizens (less than half the scale of the drop for citizens) to estimate a 7% drop from 2014 to 2015, getting us to a 49% uninsured rate in 2015 and repeat this until 2018, resulting in the final undocumented uninsured rate of 46% in calendar year 2018.  Prior to implementing this new sampling dimension, we found unrealistic drops in the uninsured rate of the undocumented population that we largely attributed to our prediction model's inability to discern this group from legally-present non-citizens, many of whom are eligible for assistance under the ACA's coverage expansions.  Although a few states have implemented programs that allow for coverage of the undocumented population, these programs are state-funded and relatively small in scale compared to the nationwide coverage expansions accompanying the ACA.

    ← Return to text

Technical Appendix C: Imputation of Offer of Employer-Sponsored Insurance
  1. Our analysis of the National Health Interview Survey finds that 67.2% of nonelderly workers had an offer of ESI in 2017 compared to 67.8% in 2018.

    ← Return to text

  2. For example, anyone who did not work during 2017 who then held an offer of ESI in early 2018 would appear incongruous in our CPS-based eligibility model.  In the other direction, workers covered by health insurance through their own employer in 2017 who lost their offer of ESI during the early months of 2018 (perhaps due to a job change) would also appear incongruous due to the discrepancy across the two time periods.

    ← Return to text

  3. Available at: https://www.census.gov/data/datasets/time-series/demo/health-insurance/cps-asec-research-files.html  For more detail about these microdata, see: J. Abramowitz, B. O'Hara.  New Estimates of Offer and Take-up of Employer-Sponsored Insurance (US Census Bureau), 2016.  Available at: https://www.census.gov/library/working-papers/2016/demo/Abramowitz-2016.html.

    ← Return to text

  4. For an explanation of affordability, see: Kaiser Family Foundation. Employer Responsibility Under the Affordable Care Act. July 2019. Available at: https://www.kff.org/infographic/employer-responsibility-under-the-affordable-care-act/.

    ← Return to text

The Henry J. Kaiser Family Foundation Headquarters: 185 Berry St., Suite 2000, San Francisco, CA 94107 | Phone 650-854-9400
Washington Offices and Barbara Jordan Conference Center: 1330 G Street, NW, Washington, DC 20005 | Phone 202-347-5270

www.kff.org | Email Alerts: kff.org/email | facebook.com/KaiserFamilyFoundation | twitter.com/kff

Filling the need for trusted information on national health issues, the Kaiser Family Foundation is a nonprofit organization based in San Francisco, California.