Health Affairs

This Article Abstract Submit a response to this article Services E-mail this article to a friend Alert me to new issues of the journal

H E A L T H  T R A C K I N G :
M A R K E T W A T C H
 
30 November 2006  
  

The Uninsured And The Affordability 
Of Health Insurance Coverage

Examining subgroups of uninsured Americans uncovers
certain patterns of coverage gaps,
but affordability remains a key concern.


by Lisa Dubay, John Holahan, and Allison Cook


ABSTRACT:

The 2005 Current Population Survey (CPS) is used to estimate what share of uninsured Americans are eligible for coverage through Medicaid or the State Children's Health Insurance Program (SCHIP), need financial assistance to purchase health insurance, and are likely able to afford insurance. Twenty-five percent are eligible for public coverage, 56 percent need assistance, and 20 percent can afford coverage. This varies across uninsured populations: 74 percent of children are eligible for public programs, and 57 percent and 69 percent of parents and childless adults, respectively, need assistance. A central conclusion is that a large percentage of uninsured adults need help purchasing health insurance. [Health Affairs 26, no. 1 (2007): w22-w30 (published online 30 November 2006; 10.1377/hlthaff.26.1.w22)]

The predominant vehicle for health insurance coverage in the United States is employers, which cover 161 million nonelderly people. Another large source of coverage, particularly for low-income people, is Medicaid, which covers children, parents, the disabled, and in some states, other adults. Children and some adults, mostly parents, also receive coverage through the State Children's Health Insurance Program (SCHIP). Together, Medicaid and SCHIP provide insurance to almost thirty million people.⁴ Another fourteen million people obtain coverage through the direct purchase market, and others are covered through Medicare and military health insurance programs. Together these various types of insurance extended coverage to 209.5 million Americans in 2004. The remaining 45.5 million nonelderly Americans were uninsured.⁵

People who lack health insurance fall into three main groups. First, some adults and children are eligible for Medicaid and SCHIP but do not participate. This could be because of administrative barriers, limited outreach efforts, or lack of knowledge about eligibility for public health insurance coverage, or because families themselves do not make the necessary efforts to obtain coverage.⁶

Another group has incomes above Medicaid and SCHIP eligibility levels but finds coverage too costly given their incomes. Health insurance premiums have risen dramatically in recent years: Premiums for private-sector employees of all firm sizes averaged about $3,700 for individuals and $10,000 for families in 2004.⁷ Given high and rising premiums, some firms--particularly smaller firms with low-wage workers--do not offer coverage. Other workers have an offer of coverage, but employers pay only a relatively small share of the cost, particularly for dependent coverage, leaving a large share for the employee.⁸ Moreover, some firms that offer coverage do not make all workers eligible--for example, new employees or seasonal and temporary workers. Self-employed workers and others without an offer of employer-sponsored coverage often find coverage unaffordable. Although most of these people can obtain coverage in the nongroup market, it can be quite expensive, especially relative to premiums in the group market. In addition, people in the nongroup market can be denied coverage altogether, be offered coverage that excludes certain services for health reasons, or face high premiums because of health status.

A third group appears to have sufficient income to afford coverage but are nonetheless uninsured. These people are probably not offered employer coverage, since take-up rates of employer offers for higher-income people are very high.⁹ Some people in this group may be self-employed and consequently face the purchase of insurance in the nongroup market. Some may be denied coverage or face extremely high premiums because of their poor health status. Others may simply choose to go without coverage.

In this paper we provide a profile of the uninsured that divides them into these three groups. We conducted the analysis separately for children, parents, and nonparents because policy options are different for each group. We found that 24.7 percent of the uninsured are eligible for public health insurance programs, 55.7 percent are in the "need assistance" category, and 19.6 percent are likely to be able to afford coverage on their own. There is much variation in this distribution across population groups, with 74 percent of uninsured children being eligible for existing public programs and 57 percent and 69 percent of uninsured parents and childless adults, respectively, being in the "need assistance" category. Consequently, absent a universal coverage solution, a range of policies will be needed to address the problem of uninsurance.

Study Data And Methods

Data. We used data from the most recent Annual Social and Economic Supplement to the Current Population Survey (CPS), conducted by the U.S. Census Bureau in 2005, which we adjusted for the "Medicaid undercount." We also used a detailed Medicaid and SCHIP eligibility model to identify those who are eligible but not enrolled. The analysis is based on health insurance unit (HIU) income to better identify those who are likely or unlikely to be able to afford health insurance coverage.

The CPS is the most frequently cited national survey on health insurance for Americans. It has both strengths and weaknesses.¹⁰ Strengths include that it is fielded annually, has samples that are fairly large, is conducted in person, has a high response rate, and has excellent income data. A major weakness is that there is little agreement as to whether it is measuring the uninsured for an entire year, as intended, or whether responses more closely reflect the uninsured at a point in time. There is also some evidence that Medicaid is underreported.

The way in which the CPS asks people to report their insurance coverage would seem to lead to an estimate of the number of uninsured people for the entire previous year. But comparisons to other surveys suggest that the number of people without coverage is much closer to point-in-time estimates and well above full-year estimates.¹¹ In its most recent release, the Census Bureau stated that its estimates were more closely in line with point-in-time estimates of the uninsured.¹² We accept this assessment.

A second issue is that there is evidence that the CPS understates Medicaid enrollment and possibly overstates the number of uninsured people. Recent evidence suggests that the Medicaid undercount might be primarily overstating private coverage: People who respond seem to accurately indicate whether or not they have health insurance but often make mistakes about what kind.¹³ This evidence comes from a study in which the researchers took a list sample of people known to be enrolled in Medicaid and conducted a telephone survey asking them to report their coverage. About 85 percent correctly reported having Medicaid. Another 10 percent reported some other form of coverage such as employer or private nongroup coverage, and 5 percent reported being uninsured.

A related question is that to the extent that there is an undercount, can Medicaid administrative data be regarded as a gold standard for comparison? There are a number of reasons to believe that the administrative data overstate the number of people in the program.¹⁴ In particular, there is evidence of double-counting and the possibility that some people remain in the administrative data after they have obtained another type of coverage. This could be especially problematic for children because of the use of continuous eligibility and the opportunity to be counted in both Medicaid and SCHIP in a given year.

To adjust for the Medicaid undercount but also to reflect the fact that administrative data could be overstating the number of enrollees, we established targets of 50 percent of the difference between the CPS-reported number of Medicaid and SCHIP enrollees and administrative totals. Because of the results cited above, we assumed that two-thirds of those to whom we are assigning Medicaid coverage were reporting private coverage and that one-third were reporting being uninsured. In summary, we increased the number of people reporting Medicaid coverage but not to the full extent of the administrative totals, and we reduced the number of uninsured people below what is reported in the CPS data.

Defining "affordability.""Affordability" of coverage is a subjective concept, and its definition depends on what is thought to be the appropriate share of income that one should be expected to pay to have health insurance. The definition varies considerably in both the literature and in policy and political debates. The share that is appropriate is undoubtedly much lower at very low income levels because a certain minimum amount of income is needed to cover basic necessities. In fact, SCHIP sets a maximum level for premiums and other cost sharing at 5 percent of family income. But at moderate income levels, people should probably be expected to pay somewhat more. Massachusetts has set subsidized premiums for people with incomes of 100 percent and 300 percent of the federal poverty level to be 1.8 percent and 4.7 percent of income for individuals, respectively, and somewhat higher for couples.¹⁵

With employer-sponsored coverage, employees explicitly contribute 16 percent and 26 percent of the cost of health insurance coverage for individual and family premiums on average, respectively, and some additional portion of the cost through forgone wages. At moderate levels of income--such as 300 percent of poverty--average employee contributions to employer-sponsored coverage for individuals and families constitute less than 5 percent of income.¹⁶

Setting aside the question of what share of income should be spent on health insurance for it to be considered "affordable," it is difficult to determine the price of insurance faced by those who are uninsured. In 2005, 81 percent of uninsured workers were not offered or eligible for coverage; thus, only 19 percent had an offer of employer-sponsored coverage that they did not take up.¹⁷ As a result, we assumed that most of the uninsured would be buying insurance in the private nongroup market.

Estimates of nongroup premiums for the uninsured population are limited except for industry surveys on the price of purchased policies, which are biased downward by underwriting and selection processes. To assess the potential affordability of nongroup coverage, we used national data on the cost of insurance in the small-group market and calculated the shares of family income at 300 percent and 400 percent of poverty that premiums at this level would require.

As a proxy for premiums in the nongroup market, we used premiums for individual and family policies available to firms with fewer than ten workers using the 2004 Medical Expenditure Panel Survey Insurance Component (MEPS-IC).¹⁸ These premiums are $3,998 for an individual policy and $9,961 for a family policy. For a single person in 2004, three times the poverty level was $28,935; four times the poverty level was $38,580 ($57,921 and $77,228, respectively, for a family of four). Thus, people would pay 13.8 percent of income at 300 percent of poverty and 10.4 percent of income at 400 percent of poverty. Family policies would cost 17.2 percent of family income at 300 percent of poverty and 12.9 percent of family income at 400 percent of poverty.

For this analysis, we set the threshold of affordability at 300 percent of poverty. Particularly for families, this seems to be at the outer reach of affordability. Therefore, we conducted sensitivity analyses that examined the change in the population in the "need assistance" group if we used alternative methods for defining affordability.

This approach has two limitations. First, our estimates might understate the share of income that premiums would account for among those living in high-cost states or with poor health status and certain health conditions.¹⁹ They might overstate the share of income among those living in states with low health care costs or in good health. Second, we do not know what the benefit packages or cost sharing are for the policies for which we have premium data. The percentage of income that families should be expected to pay cannot be divorced from their out-of-pocket liability.

Eligibility simulation. To identify people who are eligible for Medicaid and SCHIP, we used a detailed eligibility simulation model designed to mimic the eligibility determination process in each state. We modeled eligibility through pathways specific to adults and children. Because health insurance data in the CPS are considered current to the calendar year prior to its release, income, resource, and categorical eligibility rules reflect eligibility for 2004.

To simulate eligibility, we used HIUs developed from the CPS data. HIUs include all people who would be eligible for coverage under one insurance policy (that is, policyholder, spouse, children under age nineteen, full-time students under age twenty-three). We used HIUs instead of CPS family or subfamily units because they more closely approximate the groupings used in determining eligibility for public coverage. The family composition, work status of the adults, age of the children, earned and unearned income, assets, and child care expenses were compared with detailed state-specific Medicaid and SCHIP eligibility rules to determine eligibility.

We simulated eligibility for children using rules for welfare-related eligibility under Medicaid and poverty-related eligibility under Medicaid and SCHIP. We simulated eligibility for adults using rules for Section 1931, Section 1115 waivers, Transitional Medical Assistance (TMA), Ribicoff, and Medically Needy. We imputed Medicaid eligibility to individuals reporting receipt of Supplemental Security Income (SSI) on the CPS. Finally, those reporting Medicaid/SCHIP coverage but for whom we could not determine eligibility were deemed eligible for Medicaid/SCHIP. In addition to determining categorical, income, and resource eligibility, the model took into account immigration status based on whether states provide coverage for legal immigrants who arrived after 1996 and who have been in the country for less than five years or for five years or more.²⁰

Study Results

In 2004, 44.6 million Americans were uninsured, accounting for 17.5 percent of the non-elderly noninstitutionalized U.S. population after adjustment for the Medicaid undercount. The adjustment reduced the number of uninsured people by 900,000. With an uninsurance rate of 10 percent, children were underrepresented in the uninsured group relative to adults (Exhibits 1 and 2). This is in large part because of Medicaid and SCHIP, which extend eligibility for coverage to 53 percent of all U.S. children (41.4 million of 78.0 million).

Parents had higher rates of uninsurance than children: 16 percent of all parents were uninsured. The 11.1 million uninsured parents represent about 25 percent of both the uninsured and the total nonelderly populations. Finally, childless adults had the highest rate of uninsurance (23 percent) and accounted for the majority of the uninsured (57 percent), despite constituting only 43 percent of the non-elderly population (Exhibits 1 and 2).

Looking at the uninsured by whether they are eligible for public coverage, not eligible for public coverage but in families that would need assistance to make coverage affordable, and not eligible but for whom coverage is likely affordable provides critical information regarding the types of strategies that would be needed to address the problem of uninsurance (Exhibit 3). A quarter of the uninsured are eligible for public coverage but not enrolled. Another 56 percent would need financial assistance to make the purchase of private non-group insurance affordable. The remaining 20 percent are not currently eligible for public coverage and live in families whose incomes deem the purchase of private nongroup coverage affordable. Thus, the majority of the uninsured have sufficiently low incomes that at least partial financial assistance would be necessary to make coverage affordable. This overall picture masks the variation in the distribution of the eligible, "need assistance," and affordable groups for children, parents, and childless adults (Exhibit 2).

Children. Fifty-three percent of all children are eligible for public coverage through Medicaid or SCHIP. Their high rate of uninsurance (14 percent) relative to noneligible children and the higher income eligibility thresholds relative to their parents and other adults mean that children who are eligible for Medicaid and SCHIP account for 74 percent of all uninsured children. Consequently, increasing participation among those already eligible is a critical mechanism for eradicating uninsurance among children. Another 11 percent of uninsured children live in families with incomes that make the purchase of nongroup coverage unaffordable. Children who are ineligible for public programs and who have family incomes above 300 percent of poverty account for only 15 percent of uninsured children (Exhibit 2).

Parents. Parents are generally much less protected by eligibility for public coverage programs than their children. Consequently, parents who are eligible for Medicaid and SCHIP represent a much lower-income population than children who are eligible, and eligible parents account for a smaller share of uninsured parents than do children. Twenty-seven percent of parents who are eligible for Medicaid or SCHIP are uninsured. Although uninsured eligible parents constitute 5 percent of all parents, they account for 28 percent of uninsured parents. Noneligible parents with incomes below 300 percent of poverty constitute the bulk of uninsured parents (57 percent). In contrast, only 5 percent of non-eligible parents with incomes above 300 percent of poverty are uninsured, but because they are 51 percent of all parents (34.4 million of 67.4 million), they account for 15 percent of uninsured parents (Exhibit 2).

Childless adults. Eligibility for public insurance coverage among childless adults is limited to the few states that cover such people with state-only funds or through 1115 waiver authority, certain disabled populations, and pregnant women with no other children. Consequently, childless adults eligible for public programs constitute only 8 percent of all uninsured childless adults. Childless adults eligible for public coverage are uninsured at a rate of 18 percent. Those who are ineligible for public coverage and with family incomes below 300 percent of poverty are uninsured at a rate of 43 percent and constitute 69 percent of all uninsured childless adults. Childless adults who are ineligible for public programs and who have incomes above 300 percent of poverty are uninsured at a much lower rate than other childless adults, 10 percent, and represent 23 percent of uninsured childless adults (Exhibit 2).

Sensitivity analysis. Because it is possible to define affordability in many different ways, we conducted a sensitivity analysis to see the effect of setting the affordability threshold at different levels. Specifically, how much did the "need assistance" group change if we changed the income level above which coverage was deemed affordable? Because the eligible-but-uninsured group is defined by Medicaid eligibility standards, the number of uninsured people in this group is unaffected. All of the effect is on the division between the other two groups. Because premiums as a percentage of income are relatively high at 300 percent of poverty, we first examined the effect of setting the threshold at 400 percent of poverty. As an alternative, we set the threshold at 10 percent of income; that is, we regarded anyone who faced a premium cost above 10 percent of income as needing financial assistance to purchase coverage. Next, we set the threshold at 15 percent of income. Finally, because premiums could be lower because of higher deductibles or reduced benefits, we examined the impact of setting the threshold at 10 percent of income but reducing premiums by 20 percent. Reduced premiums could mean that people would face higher out-of-pocket costs if they had greater cost sharing or fewer benefits and thus face a different affordability problem.

The percentage of the population needing assistance did change somewhat, depending on the assumptions, but the change did not alter our fundamental conclusions (Exhibit 4). The number of children who need financial assistance would increase to 16.3 percent if we used 400 percent of poverty and 20.0 percent if the threshold were set at 10 percent of income. Similarly, the number of childless adults who would need financial assistance would increase to 78.5 percent if the threshold were set at 400 percent of poverty and 82.6 percent if the threshold were set at 10 percent of income. The number needing financial assistance would fall to 71.0 percent if a 15 percent income threshold were used and 77.1 percent if a 10 percent income threshold were used but premiums were reduced. Comparable changes in the percentage needing financial assistance were seen for parents. Regardless of the affordability threshold used, the bulk of uninsured parents and childless adults are not eligible for public coverage and likely not able to afford private coverage.

Discussion

Reducing the number of uninsured people among the 25 percent who are eligible for public programs would require extensive outreach efforts and simplified enrollment and redetermination procedures, including easing requirements for documentation of income, assets, and citizenship and possibly various types of incentives or penalties. Unfortunately, provisions in the Deficit Reduction Act of 2005 requiring proof of citizenship as part of the eligibility determination process could dampen participation among those who are truly eligible for the program.

The largest group of uninsured people--the 56 percent who need financial assistance to afford coverage--present a more difficult public policy challenge. Some form of partial assistance such as sliding-scale subsidies or income-related tax credits would be necessary to reduce the cost burden for these families enough to encourage them to purchase in the nongroup market. Expansions in public programs are another option for covering these populations, although it is far easier to do so for children and parents than for childless adults, given current Medicaid and SCHIP eligibility rules. In the absence of an expansion in public coverage, health insurance market reforms, purchasing pools, or high-risk pools, or some combination, might be needed to bring the cost of coverage available to this group into the affordable range. Such reforms could also reduce the income-related subsidies that would be required to make coverage affordable. It might also be necessary to mandate that people obtain coverage once income-related financial support is available, as has been recently enacted in Massachusetts.

The 20 percent of the uninsured in the "affordable" group might be less of a policy concern than those with lower incomes. But many people whom we have assigned to this group could lack health insurance because of health conditions; that is, poor health status might mean that the purchase of insurance for some people in this group is either unaffordable or unavailable for purchase, despite our having deemed them in the affordable range. Insurance reforms or expansions and subsidization of high-risk pools might be necessary. A mandate might also be necessary to assure that this group does not become a "free rider" in the case of a catastrophic health episode. However, such mandates should be coupled with an assurance of affordability for all, but particularly for those who are older, with chronic conditions, or in poor health.

This research was funded by the Robert Wood Johnson Foundation. The authors are grateful for the very helpful suggestions of the anonymous reviewers.

NOTES

1. C. DeNavas-Walt, B.D. Proctor, and C.H. Lee, Income, Poverty, and Health Insurance Coverage in the United States: 2005, Pub. no. P60-231, August 2006, http://www.census.gov/prod/2006pubs/p60-231.pdf (accessed 30 October 2006).
2. J. Hadley and J. Holahan, "How Much Medical Care Do the Uninsured Use, and Who Pays for It?" Health Affairs 22 (2003): w66-w81 (published online 12 February 2003; 10.1377/hlthaff.w3.66).
3. D.U. Himmelstein et al., "Illness and Injury as Contributors to Bankruptcy," Health Affairs 24 (2005): w63-w73 (published online 2 February 2005; 10.1377/hlthaff.w5.63).
4. Because people can report multiple types of coverage on the CPS, we created a coverage hierarchy so that each person was counted only once in the coverage distribution. Employer-sponsored insurance is at the top of the hierarchy, followed by Medicaid/SCHIP, other federal coverage, private nongroup coverage, and uninsured. The number of nonelderly people reporting Medicaid/SCHIP in 2004, regardless of whether they reported another type of coverage, is thirty-four million.
5. These numbers are based on data from the 2005 Annual Social and Economic Supplement to the CPS, released in August 2005. In this paper we adjusted these data to adjust for a Medicaid undercount. Thus, the data presented will differ slightly from these estimates. C. DeNavas-Walt, B.D. Proctor, and C.H. Lee, Income, Poverty, and Health Insurance Coverage in the United States: 2004, Pub. no. P60-229, August 2005, http://www2.census.gov/prod2/popscan/p60-229.pdf (accessed 30 October 2006).
6. G.M. Kenney, J.M. Hadley and A. Tebay, "Familiarity with Medicaid and SCHIP Programs Grows and Interest in Enrolling Children Is High," Snapshots of America's Families 3, no. 2, July 2003, http://www.urban.org/UploadedPDF/310817_snapshots3_no2.pdf (accessed 17 November 2006).
7. Agency for Healthcare Research and Quality, MEPS HC-089: 2004 Full Year Consolidated Data File, October 2006,
http://www.meps.ahrq.gov/mepsweb/data_stats/download_data/pufs/h89/h89doc.pdf (accessed 17 October 2006).
8. Ibid.
9. L. Clemans-Cope, B. Garrett, and C. Hoffman, "Changes in Employees' Health Insurance Coverage, 2001-2005," Issue Paper, October 2006, http://www.kff.org/uninsured/7570.cfm (accessed 30 October 2006).
10. G.M. Kenney, J. Holahan, and L. Nichols, "Toward a More Reliable Federal Survey for Tracking Health Insurance Coverage and Access," Health Services Research 41, no. 3 (2006): 918-945.
11. K. Swartz, "Interpreting the Estimates from Four National Surveys of the Number of People without Health Insurance," Journal of Economic and Social Measurement 14, no. 3 (1986): 233-242; and P.F. Short, Counting and Characterizing the Uninsured, Paper no. 2, December 2001, http://www.umich.edu/~eriu/pdf/wp2.pdf (accessed 14 November 2006).
12. DeNavas-Walt et al., Income, Poverty, and Health Insurance Coverage in the United States: 2005.
13. K.T. Call et al., "Uncovering the Missing Medicaid Cases and Assessing Their Bias for Estimates of the Uninsured," Inquiry 38, no. 4 (2001/02): 396-408.
14. Ibid.
15. Commonwealth of Massachusetts, Commonwealth Health Insurance Connector Authority Board, "Eligibility and Hearing Process for Commonwealth Care," Draft Regulations, 12 October 2006, http://www.hcfama.org/act/Documents/Comm%20Care%20Draft%20Regs%2020061010.pdf (accessed 20 October 2006).
16. Calculations based on G. Claxton et al., Employer Health Benefits: 2005 Annual Survey, September 2005, http://www.kff.org/insurance/7315.cfm (accessed 15 November 2006); K. Swartz and D.W. Garnick, "Adverse Selection and Price Sensitivity When Low-Income People Have Subsidies to Purchase Health Insurance in the Private Market," Inquiry 37, no. 1 (2000): 45-60; and S.K. Long and J.A. Graves, "What Happens When Public Coverage Is No Longer Available?" January 2006, http://www.kff.org/medicaid/7449.cfm (accessed 30 October 2006).
17. Urban Institute tabulations based on merged data from the 2005 Contingent Workers and Alternative Employment Supplement to the CPS and the 2005 Annual Social and Demographic Supplement to the CPS prepared by Lisa Clemans-Cope and Bowen Garrett.
18. AHRQ, MEPS HC-089.
19. J. Gabel et al., "Individual Insurance: How Much Financial Protection Does It Provide?" Health Affairs 21 (2002): w172-w181 (published online 17 April 2002; 10.1377/hlthaff.w2.172); and J. Hadley and J.D. Reschovsky, "Health and the Cost of Nongroup Insurance," Inquiry 40, no. 3 (2003): 235-253.
20. Although the CPS does not contain data on immigrants' documentation status, data are available to determine how many years an immigrant has resided in the country. Since we could not determine immigrants' documentation status, our simulation overstated the eligibility of immigrants for public programs, because some people whom we determined to be eligible might not be because of their documentation status. We estimated that approximately 10 percent of children, 19 percent of parents, and 1 percent of childless adults whom we simulated to be eligible are likely ineligible because of documentation status.

Lisa Dubay is a research scientist at the Johns Hopkins Bloomberg School of Public Health, in Baltimore, Maryland. John Holahan (jholahan{at}ui.urban.org) is director of the Health Policy Center, Urban Institute, in Washington, D.C.; Allison Cook is a research assistant there.

DOI: 10.1377/hlthaff.25.1.w22
©2006 Project HOPE–The People-to-People Health Foundation, Inc.