urban institute nonprofit social and economic policy research

Variations among States in Health Insurance Coverage and Medical Expenditures

How Much Is Too Much?

Read complete document: PDF

PrintPrint this page
Share on Facebook Share on Twitter Share on LinkedIn Share on Digg Share on Reddit
| Email this pageE-mail
Document date: June 01, 2002
Released online: June 01, 2002

Assessing the New Federalism Discussion Paper No. 02-07

Assessing the New Federalism is a multiyear Urban Institute project designed to analyze the devolution of responsibility for social programs from the federal government to the states. It focuses primarily on health care, income security, employment and training programs, and social services. Researchers monitor program changes and fiscal developments. Alan Weil is the project director. In collaboration with Child Trends, the project studies changes in family well-being. The project provides timely, nonpartisan information to inform public debate and to help state and local decisionmakers carry out their new responsibilities more effectively.

Key components of the project include a household survey, studies of policies in 13 states, and a database with information on all states and the District of Columbia. Publications and database are available free of charge on the Urban Institute's web site: http://www.urban.org. This paper is one in a series of discussion papers analyzing information from these and other sources.

This paper received special funding from The Robert Wood Johnson Foundation as part of the Urban Institute's Assessing the New Federalism project. The project received additional funding from The Annie E. Casey Foundation, the W.K. Kellogg Foundation, The Henry J. Kaiser Family Foundation, The Ford Foundation, The John D. and Catherine T. MacArthur Foundation, the Charles Stewart Mott Foundation, The David and Lucile Packard Foundation, The McKnight Foundation, The Commonwealth Fund, the Stuart Foundation, the Weingart Foundation, The Fund for New Jersey, The Lynde and Harry Bradley Foundation, the Joyce Foundation, and The Rockefeller Foundation.

The nonpartisan Urban Institute publishes studies, reports, and books on timely topics worthy of public consideration. The views expressed are those of the authors and should not be attributed to the Urban Institute, its trustees, or its funders.



The Assessing the New Federalism States

Variations in Uninsurance Rates

The Insurance Gap

Medicaid and Other State Coverage

Insurance Gaps and Coverage Ratios

Medicaid Acute Care Spending

Other State Spending on Health Services

Long-Term Care Expenditures

Variations in Access and Health Status

Summary and Conclusions


Health insurance coverage in the United States for children and nonelderly adults varies considerably among states, both in the ways people are covered and the extent to which they are covered. State spending on services for these populations also varies greatly. This paper explores this variation in considerable detail. We provide information on private and public insurance coverage, uninsurance rates, and on expenditures on Medicaid and other programs states and localities use to support health services. The fundamental question is how much variation is acceptable in a system which is intended to cover low-income Americans, but leaves so many decisions on the extent of their coverage and the resources devoted to their care to states? Fundamentally, how well is this decentralized system working in providing coverage and access to Americans, particularly low-income Americans, regardless of where they live?

The American "system" is a mix of both private and public insurance. Many factors determine the availability of both types of coverage. By far, the most important source of coverage for the nonelderly population is through employers. Employers provide or contribute to the coverage of 69 percent of nonelderly Americans. The likelihood of employers offering coverage varies with a number of factors, including wage levels, firm size, industry, and characteristics of the working population. Since these characteristics vary among states, employer-sponsored insurance also varies. In addition, about 4 percent of nonelderly Americans obtain coverage privately in the individual insurance market, through Medicare because of disability, and through federal programs for civilians associated with the military.

Many of those not covered through private insurance or federal programs are covered through state-run programs, primarily Medicaid, but more recently the State Children's Health Insurance Program (SCHIP). A few states also provide state-subsidized insurance for populations not covered through Medicaid or SCHIP, primarily childless adults or legal immigrants.1

Decisions on the extent of coverage in public programs are essentially a state responsibility, though more than half the costs are borne by the federal government. Medicaid programs will spend about $248 billion in 2002 for health care services. The federal government pays from 50 percent of the cost in 11 relatively high-income states up to 76 percent in Mississippi. About $142 billion of total Medicaid spending is federal and $106 billion is state. Medicaid has traditionally provided coverage to low-income elderly and disabled persons, to nondisabled parents (mostly women) and children receiving cash assistance, and to low-income pregnant women and children. Many of these groups must be covered as part of state participation in Medicaid. Others such as those who meet the categorical eligibility requirements but have higher medical expenses, i.e., the medically needy, may be covered at the option of the state. States can cover "other groups" including higher income parents and children. States can also cover higher income people, including childless adults, through Section 1115 Medicaid research and demonstration waivers or at their own expense. In the Medicaid program there is a set of mandatory benefits states must provide, with the option of covering a wide range of other benefits. There are limits on the use of premiums and cost sharing.

The SCHIP, enacted in 1997, allowed states to cover children with income levels above those eligible for Medicaid. States have more flexibility in benefit design and use of premiums and cost sharing in SCHIP than in Medicaid—up to 200 percent of the federal poverty level or higher. Federal matching rates for Medicaid and SCHIP vary inversely with state per capita income, but are 30 percent higher under SCHIP than under Medicaid. More recently, states have been allowed to use SCHIP funds to cover parents through demonstration waivers, and a limited number of states have done so.

Under this structure states are free to extend coverage as far beyond minimum standards as they choose. Under Medicaid (post welfare reform) and SCHIP, states can use income disregards liberally to extend coverage to eligible groups virtually without limit. Considerable variation may therefore be expected and perhaps intended. But there also seems to be a national interest in extending coverage and in addressing the problems of the uninsured. And if there is a national interest, then wide variations in outcomes are problematic.

The national interest is evident from the fact that the U.S. enacted the Medicare program to cover all those over age 65 and those with serious disabilities. While ultimately failing, the health reform debate of 1993-1994 resulted in competing proposals to expand coverage from across the political spectrum. The large federal contribution to the Medicaid and SCHIP programs also reflects the recognition that there is a national interest in extending insurance coverage to low-income populations regardless of where they live. The fact that federal contributions vary inversely with state per capita income is in recognition that the poor may be adversely affected by limits on resources in low-income states and that low-income Americans should not be adversely affected as a result. Recent interest in extending subsidies through federal tax credits also suggests that even those who oppose expanding government insurance programs accept that there is a national interest in extending coverage.

But it could also be argued that there is not a consensus with regard to coverage and spending for all populations. While the nation has clearly expressed itself with regard to coverage of those over 65 and those with serious disabilities, and to a great extent for coverage of children, it is not clear that any consensus exists for parents and even less so for non-parents. States are now required to cover some parents—those whose incomes would have qualified them for AFDC in 1996—but coverage beyond this level is completely at state discretion. States can only cover non-parents with Section 1115 waivers, which are often difficult to obtain, or by using their own funds. Even for children the consensus may be limited to the poor and near-poor.

If there is not a consensus on a national interest, substantial variation may be acceptable and is clearly to be expected. Coverage and spending are a state responsibility and states are expected to make choices consistent with their political preferences. Nonetheless, it could still be true that the current system has produced variations that are beyond what even those who believe coverage and spending is properly the province of states would find acceptable. As noted above, both Republicans and Democrats made proposals to substantially extend coverage in 1993 and 1994. Recent proposals to extend coverage through federal tax credits would suggest a strong willingness to extend federal support. In other words, it is possible to both believe that there is not a national consensus, that coverage of low-income populations is a state responsibility and also be concerned that state flexibility will yield outcomes that are unacceptable.

This paper addresses how well this federal-state system has done in meeting what to some degree is a national goal of extending insurance coverage to low-income Americans. How much does insurance coverage and the resources devoted to health care vary among states? How much of this variation is due to differences in employer- sponsored insurance that leave states facing problems of very different dimensions? Can states be expected to overcome low levels of private coverage? How much is due to variations in state efforts? Would variations in uninsurance rates still be great even if there was less variation in employer coverage?

The paper begins by providing data on uninsurance rates for the nonelderly population and for children and adults. Variations in employer-sponsored insurance and in other forms of coverage are shown. We use the concept of the insurance gap, the percentage of the population not covered through employer-sponsored insurance or other private coverage, Medicare, or CHAMPUS to show the size of the population that would be uninsured if not for state efforts.2 Next, data on the share of the states' population covered through Medicaid and SCHIP are presented, along with estimates of the share of the gap covered by those two programs. It is the size of the gap together with the share of the gap closed by Medicaid and SCHIP (which we term the coverage ratio) that determines uninsurance rates across states.

State Medicaid spending is then examined, showing Medicaid spending per capita, per low-income person, and relative to the insurance gap. Finally, it is occasionally argued that some states (and their cities and counties) spend more money on direct service provision, such as public hospitals, rather than extend public insurance programs, largely because of an aversion to participation in federal entitlement programs. We compare differences in state spending on these other direct service programs to variations in spending on Medicaid and SCHIP to see if in fact there is an inverse relationship.

We conclude that the variations in uninsurance rates among states are extremely large. Much of this is attributable to variations in employer-sponsored insurance, particularly when we look at the entire nonelderly population regardless of income. States face problems of a very different magnitude when beginning to address the needs of their populations. For low-income populations, i.e., those below 200 percent of the federal poverty level (FPL), variations in employer-sponsored insurance are also important. But public efforts to close the insurance gap also matter greatly in determining uninsurance rates of low-income populations. Finally we show that levels of spending per low-income person vary by approximately a factor of four and that state and local spending on direct services is positively, not negatively, correlated with Medicaid spending.

In this paper we use data on insurance coverage from the Current Population Survey (CPS). We merge three years of the CPS to increase the size of the state samples. We use the CPS primarily because it provides data on all 50 states, which we use for regression analyses and to provide results on selected variables in the Appendix. For information on variations in health status and access to care, we rely on the National Survey of America's Families. For many of the tables, we rely on data for the 13 states from the Assessing the New Federalism (ANF) project. This is done for two reasons. First, over the past several years, the ANF project has developed an extensive amount of information on health policy in these 13 states. Second, limiting the tables to 13 states makes the information easier to present than if data were provided on all 50 states. For comparisons of spending across states, we use Medicaid expenditure data from HCFA 2082 and HCFA 64 databases. For data on "other" state spending we used data from the National Association of State Budget Officers and from the American Hospital Association.

The Assessing the New Federalism States

The thirteen states in the Assessing New Federalism project vary considerably in terms of the features of their publicly sponsored health insurance programs. The key characteristics of coverage in these states in 1999 (we use 1999 to be consistent with the data which follow) were as follows.

Massachusetts - with a Medicaid Section 1115 waiver and SCHIP funds, the state developed its MassHealth program, which extended coverage to children to 200 percent of FPL and the long-term unemployed up to 133 percent of FPL. There were provisions to cover parents and non-parents between 133 percent and 200 percent of the federal poverty level, either through their employers or a special program for the uninsured. Massachusetts also covered the short-term unemployed who were receiving unemployment compensation with incomes up to 400 percent of the federal poverty level.

Minnesota - provided coverage to children and pregnant women with incomes up to 275 percent of FPL under its Section 1115 Medicaid research and demonstration waiver. Minnesota also had a state-subsidized health insurance program, MinnesotaCare, which provides coverage to children and parents with incomes below 275 percent of FPL, and single adults with incomes below 175 percent of FPL.

New York - has had high AFDC income levels and extended Medicaid eligibility for pregnant women, infants, and children well beyond federally mandated levels. New York also had an SCHIP program, an extension of its state-subsidized Child Health Plus program, which provided coverage up to 230 percent of FPL. The state has also provided broad health insurance coverage to its general assistance or home relief population.

Washington - had generous Medicaid eligibility covering all children under age 19 with family incomes below 200 percent of FPL. Individuals and families with incomes up to 200 percent of the federal poverty level could buy into the Basic Health Plan (BHP) on a subsidized basis. Those with higher incomes could join the BHP by paying the full premium.

California - had broad eligibility for Medicaid primarily because the AFDC income level was well above the national average and the state had a medically needy program. The state extended coverage under SCHIP to children up to 250 percent of FPL. In 1999 California had not yet extended coverage to parents to 100 percent of FPL and did not have coverage for non-parents.

Michigan - had relatively broad Medicaid eligibility, including a medically needy program. Children are covered up to 200 percent of the federal poverty level under SCHIP. The state had a small subsidized insurance programs in Wayne County for the general assistance population.

New Jersey - had AFDC income levels above the national average and had a medically needy program. The state expanded eligibility for children under SCHIP to 200 percent of the federal poverty level. In 1999 the state had not yet adopted its current extensive coverage for parents and non-parents.

Wisconsin - extended Medicaid eligibility to children and parents with family incomes up to 185 percent of FPL under its new BadgerCare program. BadgerCare was funded under both as Medicaid Section 1115 and an SCHIP waiver. BadgerCare does not cover non-parents.

Florida - extended coverage to 185 percent of FPL for pregnant women and infants but otherwise provided little coverage beyond federally mandated levels. AFDC income levels were well below the national average. Florida covered children up to 200 percent of FPL under SCHIP, building upon its state-funded Healthy Kids program. Florida did not provide coverage to non-parents.

Mississippi and Alabama - had limited Medicaid coverage because of low AFDC income standards and no medically needy program. They both covered children under SCHIP up to 200 percent of FPL. There was no subsidized insurance coverage for non-parents.

Texas and Colorado - had limited Medicaid coverage because of low AFDC income standards; Colorado had no medically needy program and Texas one with very low income thresholds. Colorado covered children under SCHIP to 185 percent of the poverty level and Texas to 100 percent (it has subsequently increased coverage to 200 percent). There was no subsidized insurance coverage for non-parents.

Table 1. Variation in Uninsurance Rates among States
  Total Nonelderly Children Adults
  All <200% All <200% All <200%
Texas 26.7% 45.7% 25.3% 37.7% 27.4% 51.7%
California 23.7% 40.3% 19.4% 28.9% 25.8% 47.9%
Florida 22.9% 38.5% 18.8% 29.2% 24.5% 43.9%
Mississippi 21.8% 37.6% 18.2% 28.5% 23.7% 43.9%
New York 19.5% 33.9% 14.0% 21.6% 22.1% 41.9%
Alabama 18.1% 33.5% 14.9% 24.0% 19.5% 39.7%
Colorado 17.3% 38.1% 14.4% 30.6% 18.6% 42.6%
New Jersey 17.3% 37.3% 13.3% 25.4% 19.0% 44.1%
Washington 14.6% 28.9% 10.3% 15.5% 16.5% 36.7%
Michigan 13.5% 28.1% 9.8% 18.1% 15.4% 35.3%
Massachusetts 12.7% 24.4% 8.7% 15.0% 14.3% 30.3%
Wisconsin 11.6% 26.2% 8.8% 16.3% 12.9% 32.5%
Minnesota 9.8% 25.2% 8.3% 19.2% 10.6% 29.7%
Balance of Total 16.2% 32.0% 13.5% 23.8% 17.4% 37.6%
Grand Total 18.2% 34.8% 15.1% 25.5% 19.7% 41.0%
Source: Urban Institute tabulations of the March 1998-2000 CPS.

Variations in Uninsurance Rates

Table 1 shows the uninsurance rates for our 13 ANF states and the balance of the nation for 1997-1999. The data show that during this three-year period an average of 18.2 percent of the nonelderly population was without insurance, and there were great differences among states. The highest uninsurance rates were in Texas (26.7), California (23.7 percent), Florida (22.9 percent), and Mississippi (21.8 percent). The lowest rates were in Minnesota (9.8 percent), Wisconsin (11.6 percent), Massachusetts (12.7 percent), Michigan (13.5 percent), and Washington (14.6 percent).

For those below 200 percent of the poverty level, 34.8 percent lacked insurance nationally. Again, the states with the highest rates were Texas (45.7 percent) and California (40.3 percent). The states with the lowest rates were again Minnesota (25.2 percent), Wisconsin (26.2 percent), Massachusetts (24.4 percent), Michigan (28.1 percent), and Washington (28.9 percent). The data also show that our sample of 13 states had somewhat higher uninsurance rates than did the rest of the nation; the balance of states had uninsurance rates for the low-income population almost 3 percent less than the U.S. as a whole.

For all children the uninsurance rates averaged 15.1 percent for the nation; uninsurance rates ranged from 8.3 percent in Minnesota to 25.3 percent in Texas. The same four states, Texas, California, Florida, and Mississippi, had the highest uninsurance rates for all children and the lowest rates again were in Minnesota, Wisconsin, Massachusetts, Michigan, and Washington. For children under 200 percent of FPL, the uninsurance rate in the U.S. was 25.5 percent. Texas had by far the highest uninsurance rate (37.7 percent); the lowest rate was in Massachusetts (15.0 percent).

Nonelderly adults were much more likely to be uninsured than children, 19.7 percent vs. 15.1 percent. The highest uninsurance rates again were in Texas (27.4 percent) and California (25.8 percent); the lowest rates were in Minnesota (10.6 percent) and Wisconsin (12.9 percent). The uninsurance rate for adults under 200 percent of the poverty level nationally was 41.0 percent. Texas and California had the highest rates (51.7 percent and 47.9 percent, respectively); the lowest rates were in Minnesota and Massachusetts (29.7 percent and 30.3 percent, respectively).

The Insurance Gap

Table 2 provides data on variations in employer-sponsored insurance as well as other insurance not provided by the states. Employer-sponsored insurance together with individual insurance, Medicare and CHAMPUS determine the number of people who will be uninsured if they are not covered by Medicaid and other state-subsidized insurance. In other words, the complement to private and federal coverage insurance rate is the "insurance gap." For example, table 2 shows that across all states 69.1 percent of the nonelderly are covered through employer-sponsored insurance, and a total of 73.0 percent are covered through ESI, individual insurance, Medicare, and CHAMPUS. The remainder, 27.0 percent, are in the insurance gap.

Table 2. Employer Coverage and the "Insurance Gap"
  All Incomes Low Income (<200% FPL)
  ESI Rate Private and
Fed Rate
Gap Rate ESI Rate Private and
Fed Rate
Gap Rate
California 60.8% 64.9% 35.1% 29.7% 33.8% 66.2%
Texas 61.7% 65.2% 34.8% 31.8% 35.6% 64.4%
Mississippi 63.6% 68.2% 31.8% 35.7% 40.4% 59.6%
New York 64.3% 67.1% 32.9% 29.2% 32.4% 67.6%
Florida 64.4% 69.4% 30.6% 36.5% 42.8% 57.2%
Alabama 69.1% 73.1% 26.9% 38.2% 43.5% 56.5%
Washington 70.6% 76.1% 23.9% 34.9% 42.0% 58.0%
Massachusetts 73.0% 76.0% 24.0% 36.6% 40.6% 59.4%
Colorado 73.2% 78.5% 21.5% 39.5% 47.2% 52.8%
Michigan 73.6% 76.3% 23.7% 37.1% 41.2% 58.8%
New Jersey 74.2% 76.9% 23.1% 39.9% 42.9% 57.1%
Minnesota 78.1% 82.2% 17.8% 39.7% 46.4% 53.6%
Wisconsin 78.9% 82.3% 17.7% 45.2% 51.7% 48.3%
Balance of Total 71.8% 75.9% 24.1% 40.6% 46.1% 53.9%
Grand Total 69.1% 73.0% 27.0% 36.8% 41.7% 58.3%
Source: Urban Institute tabulations of the March 1998-2000 CPS.

Employer-sponsored insurance rates vary from 60.8 percent in California and 61.7 percent in Texas to 78.1 percent in Minnesota and 78.9 percent in Wisconsin. The private/federal coverage rate averages roughly 4 percentage points above the ESI rate. From the private and federal coverage, the size of each state's insurance gap can be determined. The gaps are largest in California and Texas (35.1 percent and 34.8 percent, respectively) and lowest in Minnesota and Wisconsin (17.8 percent and 17.7 percent, respectively). This means that California and Texas begin with much larger uninsurance problems than do Minnesota and Wisconsin. Stated differently, the variations in employer-sponsored insurance largely determine the problems states face in thinking about the uninsured and in considering the scope of public programs.

The right hand side of table 2 shows the ESI rates and insurance gaps for the those below 200 percent of FPL by state. For the nation, 36.8 percent of those below 200 percent of FPL are covered by employer-sponsored insurance; 41.7 percent are covered by ESI, individual insurance, Medicare, and CHAMPUS. The remainder, or the gap, is 58.3 percent. ESI rates vary for those below 200 percent of FPL from 29.2 percent in New York and 29.7 percent in California to highs of 45.2 percent in Wisconsin. These result in gaps as high as 67.6 percent in New York and 66.2 percent in California and as low as 48.3 percent in Wisconsin. The table shows that states like New York, California and Texas begin with very large insurance gaps for the low-income population. At the other extreme, states like Wisconsin, Colorado, and Minnesota face substantially smaller problems.

Why is there such variation in employer-sponsored insurance among states? The answer depends on what factors affect whether individuals have ESI and how much these factors vary among states. Several factors affect the likelihood of ESI.3 First, ESI rates vary by industry. Manufacturing and government employees are much more likely to have health insurance coverage than those in agriculture, construction, and retail trades. Second, workers in large firms are far more likely to have coverage than those in small firms. Third, workers in unions are far more likely to be covered. Fourth, insurance is less likely at lower wage and income levels. Those with low wages are typically less skilled and have lower education levels. Firms may not offer insurance because the cost of health insurance would be a very high share of compensation. Moreover workers with low skills and limited ability to command high compensation may prefer cash to health insurance. Fifth, holding all these factors constant, blacks, Hispanics and noncitizens are much less likely to have health insurance.

A recent study by Shen and Zuckerman analyzed how these and other factors varied among states and attempted to explain why some states had high rates of ESI and others low rates.4 They found that while industry and firm size were important in explaining whether individuals had ESI, they did not vary greatly among states in most cases. However other variables such as income, education, race-ethnicity, and citizenship both contributed to explaining the likelihood of having ESI and varied significantly among states. They concluded that states with low ESI rates have populations with fewer skills and less human capital. Because of these workforce characteristics, these states get fewer of the high human capital, high paying jobs that also have ESI and as a result have high uninsurance rates.

Table 3. Variations in Medicaid Coverage
  Total Enrolled in Medicaid or State-Funded Health Programs/Population
  Nonelderly Children Adults
  All Incomes <200% FPL All Incomes <200% FPL All Incomes <200% FPL
Colorado 4.2% 14.7% 7.3% 21.8% 2.8% 10.4%
New Jersey 5.8% 19.8% 11.1% 31.0% 3.6% 13.4%
Wisconsin 6.1% 22.1% 10.2% 31.2% 4.1% 16.3%
Florida 7.7% 18.7% 15.6% 30.5% 4.4% 11.8%
Minnesota 8.0% 28.4% 13.4% 37.6% 5.2% 21.5%
Texas 8.1% 18.7% 15.5% 28.3% 4.4% 11.5%
Alabama 8.8% 23.0% 18.0% 36.1% 4.8% 14.4%
Washington 9.3% 29.1% 17.8% 46.2% 5.6% 19.1%
Mississippi 10.0% 21.9% 17.7% 32.9% 6.0% 14.4%
Michigan 10.1% 30.7% 17.7% 43.9% 6.2% 21.3%
Massachusetts 11.3% 35.0% 21.1% 50.7% 7.2% 25.1%
California 11.4% 25.9% 20.4% 38.4% 6.9% 17.4%
New York 13.4% 33.8% 23.6% 47.6% 8.6% 24.7%
Balance of Total 7.9% 21.9% 14.3% 31.6% 4.9% 15.3%
Grand Total 8.8% 23.5% 16.0% 34.5% 5.4% 16.2%
Source: Urban Institute tabulations of the March 1998-2000 CPS.

Medicaid and Other State Coverage

Table 3 provides data on coverage among states through Medicaid and other state programs. The first column of table 3 shows that Mississippi, California, New York, Michigan, and Massachusetts cover at least 10 percent of their nonelderly populations on Medicaid and other state insurance programs. Colorado covers the smallest share of their population (4.2 percent) among our 13 states. The second column shows that Texas, Florida, and New Jersey cover less than 20 percent, and Colorado less than 15 percent, of the population below 200 percent of FPL. In contrast, New York, Michigan, and Massachusetts cover over 30 percent.

Generous coverage of the low-income population does not always translate into broad coverage of the overall population; much depends on income distribution. Alabama and Mississippi cover a larger proportion of their populations with Medicaid or state-funded health programs than does Minnesota despite covering a small share of their low-income population. This occurs because Alabama and Mississippi have such large low-income populations (below 200 percent of the poverty level) that despite their lower coverage rates of these populations they end up extending coverage to a higher share of their overall population than do some states with more generous coverage policies.

The middle two columns of table 3 show that California, New York, and Massachusetts cover over 20 percent of their children through state programs; in contrast Colorado covers 7.3 percent. Similarly Massachusetts, New York, and Washington cover over 45 percent of their low-income children; Colorado and Texas less than 30 percent. Substantially fewer adults are covered by states in public programs. Coverage for the adult population in public programs ranges from 2.8 percent in Colorado to 7.2 percent in Massachusetts and 8.6 percent in New York. Colorado, Texas, and Florida cover less than 12 percent of their low-income adults while New York and Massachusetts cover over 25 percent.

Insurance Gaps and Coverage Ratios

Tables 4 and 5 provide data by state on insurance gaps, coverage ratios, and uninsurance rates. In table 4 we present coverage at all income levels with the 13 states ordered by the size of their insurance gap; table 5 presents the same data for the low-income population. Again, the size of the gap primarily reflects the importance of employer-sponsored insurance. The coverage ratios—the extent of Medicaid or state health program divided by the gap—show the role of public policy. The uninsurance rate depends on both the size of the insurance gap and state efforts to reduce the size of the gap.

Table 4. Insurance Gaps and Uninsured Rates, All Incomes
  Nonelderly Children Adults
  Gap Rate Mcaid and State/Gap Uninsurance Rate Gap Rate Mcaid and State/Gap Uninsurance Rate Gap Rate Mcaid and State/Gap Uninsurance Rate
California 35.1% 32.5% 23.7% 39.9% 51.3% 19.4% 32.7% 21.2% 25.8%
Texas 34.8% 23.4% 26.7% 40.7% 38.0% 25.3% 31.8% 13.8% 27.4%
New York 32.9% 40.7% 19.5% 37.6% 62.7% 14.0% 30.7% 28.1% 22.1%
Mississippi 31.8% 31.4% 21.8% 35.9% 49.2% 18.2% 29.7% 20.3% 23.7%
Florida 30.6% 25.2% 22.9% 34.5% 45.4% 18.8% 29.0% 15.3% 24.5%
Alabama 26.9% 32.8% 18.1% 32.9% 54.7% 14.9% 24.3% 19.9% 19.5%
Massachusetts 24.0% 47.1% 12.7% 29.8% 70.7% 8.7% 21.5% 33.4% 14.3%
Washington 23.9% 38.9% 14.6% 28.0% 63.4% 10.3% 22.0% 25.2% 16.5%
Michigan 23.7% 42.8% 13.5% 27.5% 64.4% 9.8% 21.7% 28.8% 15.4%
New Jersey 23.1% 25.3% 17.3% 24.4% 45.5% 13.3% 22.6% 15.9% 19.0%
Colorado 21.5% 19.4% 17.3% 21.7% 33.5% 14.4% 21.4% 13.1% 18.6%
Minnesota 17.8% 44.7% 9.8% 21.7% 61.6% 8.3% 15.8% 32.7% 10.6%
Wisconsin 17.7% 34.4% 11.6% 19.0% 53.7% 8.8% 17.0% 24.3% 12.9%
Balance of Total 24.1% 32.8% 16.2% 27.8% 51.5% 13.5% 22.3% 21.9% 17.4%
Grand Total 27.0% 32.6% 18.2% 31.1% 51.6% 15.1% 25.0% 21.5% 19.7%
Source: Urban Institute tabulations of the March 1998-2000 CPS.

Clearly, the size of the gap may be affected by the size of public programs, i.e., generous public coverage may reduce the take-up of employer coverage offers, increasing the size of the gap. That crowding out of private coverage occurs has been well documented, though there is some dispute as to the extent of it.5 While research has shown that ESI rates (and thus the gaps) are largely driven by factors associated with human capital such as income and education, states like Minnesota, Massachusetts, and Washington that have extended coverage well above the poverty level may be having some effect on their ESI rates, and as a result the gaps we report may be biased downward. This is less likely to be an issue in low-coverage states such as Texas or Mississippi because eligibility is not extended far enough up the income distribution to much affect the working population. In these states ESI rates are low (and gaps high) because of human capital factors, as discussed above.

The first column of table 4 shows once again that the gap for the nonelderly at all income levels ranges from 35.1 percent in California and 34.8 percent in Texas to 17.8 percent in Minnesota and 17.7 percent in Wisconsin. A comparison of columns 1 and 3 of table 4 shows that the size of the gap clearly affects the level of the uninsurance rate. California and Texas have the largest gaps and the highest uninsurance rates, similarly Minnesota and Wisconsin have the smallest gaps and the lowest uninsurance rates.

But not all of the variation in uninsurance rates is explained by the insurance gap. The four states that have high coverage ratios make significant inroads into the size of their uninsured populations. New York (40.7 percent), Massachusetts (47.1 percent), Michigan (42.8 percent), and Minnesota (44.7 percent) substantially reduce their uninsurance rates by providing generous public coverage. The impact of public policy can be seen by simple comparisons of states with similar gaps. For example, Texas and New York have gap rates that are similar in magnitude, 32.9 to 34.8 percent. But New York has a coverage ratio of 40.7 percent while Texas has a coverage ratio of 23.4 percent. The result is an uninsurance rate of 19.5 percent in New York and 26.7 percent in Texas. Similarly, Colorado and Minnesota have gaps of 21.5 percent and 17.8 percent respectively, but Minnesota has a coverage ratio of 44.7 percent and Colorado of 19.4 percent. The result is that the uninsurance rate of the nonelderly in Colorado is 17.3 percent versus 9.8 percent in Minnesota.

Similarly, the middle three rows show that the size of the gap clearly affects the uninsured rates for children. Texas and California have the largest gaps and the highest uninsurance rates; Minnesota and Wisconsin the smallest gaps and the lowest uninsurance rates. The middle row shows the differences in state efforts. New York, Washington, Michigan, and Minnesota cover over 60 percent of the children in the gap through Medicaid and other state programs, and Massachusetts over 70 percent. The result is that their uninsurance rates are lower than one would expect given the size of their gaps. Comparisons of states with similar gaps are again instructive. California has a gap of 39.9 percent and New York 37.6 percent, but New York covers 62.7 percent of the population of the children in the gap and California 51.3 percent. The result is that New York has a substantially lower uninsurance rate.

Table 5. Insurance Gaps and Uninsured Rates, Low-Income Populations (<200% FPL)
  Nonelderly Children Adults
  Gap Rate Mcaid and State/Gap Uninsurance Rate Gap Rate Mcaid and State/Gap Uninsurance Rate Gap Rate Mcaid and State/Gap Uninsurance Rate
New York 67.6% 49.9% 33.9% 69.2% 68.7% 21.6% 66.6% 37.0% 41.9%
California 66.2% 39.2% 40.3% 67.3% 57.1% 28.9% 65.4% 26.7% 47.9%
Texas 64.4% 29.0% 45.7% 66.0% 42.9% 37.7% 63.2% 18.2% 51.7%
Mississippi 59.6% 36.8% 37.6% 61.4% 53.6% 28.5% 58.3% 24.8% 43.9%
Massachusetts 59.4% 59.0% 24.4% 65.8% 77.1% 15.0% 55.4% 45.3% 30.3%
Michigan 58.8% 52.2% 28.1% 62.0% 70.8% 18.1% 56.6% 37.6% 35.3%
Washington 58.0% 50.2% 28.9% 61.7% 74.9% 15.5% 55.8% 34.2% 36.7%
Florida 57.2% 32.7% 38.5% 59.7% 51.1% 29.2% 55.7% 21.1% 43.9%
New Jersey 57.1% 34.7% 37.3% 56.4% 54.9% 25.4% 57.6% 23.3% 44.1%
Alabama 56.5% 40.7% 33.5% 60.1% 60.0% 24.0% 54.1% 26.6% 39.7%
Minnesota 53.6% 53.0% 25.2% 56.8% 66.1% 19.2% 51.2% 41.9% 29.7%
Colorado 52.8% 27.9% 38.1% 52.4% 41.6% 30.6% 53.0% 19.7% 42.6%
Wisconsin 48.3% 45.8% 26.2% 47.6% 65.7% 16.3% 48.8% 33.3% 32.5%
Balance of Total 53.9% 40.6% 32.0% 55.4% 57.0% 23.8% 52.9% 29.0% 37.6%
Grand Total 58.3% 40.3% 34.8% 60.0% 57.4% 25.5% 57.1% 28.3% 41.0%
Source: Urban Institute tabulations of the March 1998-2000 CPS.

A similar picture emerges for adults; states with the highest gap rates have the highest uninsurance rates and vice versa, states with the smallest gaps have the lowest uninsurance rates. And again states with extensive public programs—Minnesota, Massachusetts, and New York—have a significant effect on their adult uninsurance rates.

The story changes somewhat when we look at table 5, which shows similar data for the low-income population. Here again there is a rough correspondence between the gap rate and the uninsurance rate, but the role of public programs is much more significant. Massachusetts, Michigan, Washington, and Minnesota all cover over 50 percent of the low-income population in the insurance gap and New York just under 50 percent. As a result, all of these states reduce their uninsurance rates below where they would be based solely on these gaps. Comparing again New York and California, the gap rates are similar, but New York has a much higher coverage ratio. As a result, the uninsurance rate in New York is much lower than in California. Similarly, Minnesota and Colorado have similar size gaps but Minnesota has a much higher coverage ratio for its low-income population. The result is that the uninsurance rate in Minnesota is 25.2 percent versus 38.1 percent in Colorado.

The rest of table 5 shows the effect of state programs to cover children and adults. In both cases there is a rough correspondence between the size of the gap, Medicaid and state health programs, and the uninsurance rate for low-income children. For example, states such as Massachusetts, Michigan, and Washington all cover more than 70 percent of the children in the insurance gap and Wisconsin, Minnesota, and New York over 65 percent. New York, Michigan, Washington, and Wisconsin all cover more than 30 percent of low-income adults in the insurance gap, while Minnesota, and Massachusetts cover over 40 percent. Each of these states' public programs reduce their uninsurance rate below those of other states with similar insurance gaps. In contrast, the low coverage rates in Texas and Colorado (about 40 percent for children and 20 percent for adults) result in higher uninsurance rates for both low-income children and adults than would be expected if only gaps mattered.

In summary, the size of the gap is clearly important in explaining the uninsurance rate for the nonelderly population at all income levels. For low-income people, the likelihood of working in firms that offer employer-sponsored coverage is lower and public programs play a greater role. Public efforts are particularly important for children. But state efforts to cover low-income adults also affect their uninsurance rates.

To test for the relative importance of the size of the insurance gaps and the effectiveness of state policies to close the insurance gaps, we use data from all 50 states to estimate two simple regression models that have the uninsurance rate as the dependent variable. The gap rate is the independent variable in the first model and the ratio of Medicaid and state coverage to population is the independent variable in the second. (Including both in the same regression comes close to creating an identity with the result that there are very low standard errors and unreliable results.) The two models were estimated separately for the full populations, by income, and for children and adults.

Table 6. Insurance Gaps and Medicaid Coverage: Which Matters?
  Coefficient T-Statistic R2
Nonelderly, all incomes
    Gap Rate 1.06 11.83 0.74
    Medicaid and State / Population 0.07 0.62 0.01
Children, all incomes
    Gap Rate 0.9 6.32 0.45
    Medicaid and State / Population 0.02 0.31 0.00
Adults, all incomes
    Gap Rate 1.06 15.34 0.83
    Medicaid and State / Population 0.02 0.26 0.00
Nonelderly, low income
    Gap Rate 0.82 3.73 0.22
    Medicaid and State / Population -0.36 -3.97 0.25
Children, low income
    Gap Rate 0.53 1.59 0.05
    Medicaid and State / Population -0.67 -5.02 0.34
Adults, low income
    Gap Rate 0.95 5.42 0.38
  Medicaid and State / Population -0.26 -3.76 0.23
Source: Urban Institute tabulations of the March 1998-2000 CPS with dollar figures from HCFA 2082 and NASBO data. Note: Regressions were estimated in logs, allowing the coefficients to be interpreted as elasticities.

The results shown in table 6 demonstrate that among populations at all income levels, the size of the gap is clearly the more important in explaining the uninsurance rate. The gap ratio in the regressions for all nonelderly, children, and adults is clearly positive and highly significant. This single variable explains a high share of the variation in the uninsurance rate. The first row shows that a 1.00 percent increase in the gap would result in a 1.06 percent increase in the uninsurance rate for the nonelderly and similar increases for both children and adults. At the same time the results show that the Medicaid/state coverage variable is not statistically significant in any of these regressions and contributes nothing to explaining the uninsurance rate.

Among the low-income population, however the gap is less important. The coefficients are smaller in magnitude, thus increases in the insurance gap among low-income populations result in less of an increase in the uninsurance rate. The gap ratio is statistically significant in the nonelderly and adult equations but not in the child equation. A lesser effect is not surprising given that many low-income people are not in the labor force and, therefore, would not be affected by the provision of employer-sponsored insurance. In contrast, the Medicaid/state coverage variable is negative and the variable is significant (at the .05 level) in each of the three equations. Among all the low-income nonelderly, a one percent increase in Medicaid/state coverage results in a 0.36 percent reduction in the uninsurance rate. For children a one percent increase in Medicaid/state coverage results in a 0.67 percent reduction in the uninsured rate, and for adults a 0.26 percent reduction. Since Medicaid and state coverage is directed at the low-income population it is not surprising that it has a greater effect here than on the total population.

Thus it is clear that the size of the gap is the key determinant of the uninsurance rates across states for the population as a whole. But for the low-income population both are important. The bigger the gap, the higher the uninsurance rate, but the greater the extent of public coverage the lower the uninsurance rate.

Medicaid Acute Care Spending

Beyond coverage, expenditures are another important indicator of the relative contributions of states to low-income populations health care use. Table 7 shows acute care Medicaid spending (for children, adults, and the nonelderly disabled) per capita, Medicaid spending per person under 200 percent of the poverty level, and Medicaid spending per person in the gap. All three are clearly related. Expenditures reflect not only the extent of coverage but also generosity of benefits, levels of provider payments, and so on.

Table 7 shows that Colorado and Wisconsin spend the least on a per capita basis, $197 and $219, respectively. Massachusetts and New York spend the most, $435 and $692, respectively. Per capita spending reflects not only the generosity of benefits but also the share of the population covered. Medicaid spending per low-income person (column 2) shows slightly different results. The states with the lowest levels of spending are Texas ($608) and Mississippi ($720); Massachusetts ($1,507) and New York ($1,866) remain the highest spending states. It is noteworthy that there is a roughly threefold difference among states in the level of spending per low-income person even when both federal and state spending are combined. Column 3 shows the ratio of Medicaid spending to the number of people in the gap—here Texas and California spend the least while again, Massachusetts and New York spend the most. The difference between highest and lowest is roughly threefold.

Table 7. Medicaid Spending On Acute Care for the Nonelderly, 1998
  Federal and State Spending State Spending
  Medicaid $/Population Medicaid $/<200% FPL Medicaid $/Gap Medicaid $/Population Medicaid $/<200% FPL Medicaid $/Gap
Colorado $197 $761 $918 $95 $366 $441
Wisconsin $219 $868 $1,241 $90 $357 $511
Minnesota $235 $946 $1,318 $112 $453 $631
Texas $251 $608 $721 $95 $229 $272
Alabama $272 $734 $1,013 $84 $225 $311
New Jersey $276 $1,013 $1,193 $138 $506 $596
Florida $291 $779 $953 $129 $345 $423
California $310 $756 $883 $151 $369 $431
Washington $314 $1,126 $1,313 $150 $539 $628
Mississippi $318 $720 $998 $73 $165 $229
Michigan $336 $1,114 $1,419 $156 $517 $659
Massachusetts $435 $1,507 $1,814 $217 $754 $907
New York $692 $1,866 $2,104 $346 $933 $1,052
Balance of United States $303 $915 $1,261 $116 $349 $481
United States $325 $940 $1,203 $139 $402 $515
Source: Urban Institute tabulations of the March 1998-2000 CPS with acute care Medicaid spending for children, adults, and the nonelderly disabled from 1998 HCFA 2082 data.

The federal Medicaid program matches each state's expenditures with a formula that allows federal spending to be inversely related to state income, i.e., the lowest income states have the highest federal matching rates. The intent is to even out levels of spending. Clearly these results show that this is far from reality. The reason for this is that there are even greater variations in state effort. The last three columns show state expenditures from their own resources. On a per capita basis the states with the lowest expenditures are Mississippi and Alabama ($73 and $84 respectively), the two highest states again are Massachusetts and New York ($217 and $346). Here differences between the highest and lowest states are at least threefold even when New York is excluded. The differences are even larger when we look at state Medicaid spending per person below 200 percent of the poverty level. Here, the lowest spending states were Mississippi ($165), Alabama ($225), and Texas ($229). In contrast, Massachusetts spent $754 and New York $933. Thus, there is at least a fourfold difference in state expenditures for low-income persons. In part this reflects differences in coverage per low-income person, but the variations in spending are greater than variations in coverage.

A similar picture emerges when we look at the number of people in the gap. Mississippi and Texas cover fewer individuals in the insurance gap than do states like Massachusetts and New York. Thus, the differences in spending are large: Mississippi spends $229 and Texas $272, but at the other extreme Massachusetts spends $907 and New York $1,052. Again differences in spending per person in the gap are greater than differences in coverage.

Variations among states in Medicaid/SCHIP spending tend to mirror variations in incomes. Higher income states not only have greater ability to pay for services but typically have fewer low-income people to serve. To test the relationship between spending and income, we again use data for 50 states to estimate two simple regressions. The first was the log of federal and state Medicaid and SCHIP spending per capita against the log of state income per capita and second, the log of state spending per capita against the log of state income per capita. We then repeated these regressions using the log of spending per low-income person (below 200 percent of FPL). The results in table 8 show that a one percent increase in income per capita is not related to federal and state spending per capita but is associated with a one percent increase in state spending per capita. Federal matching funds which are greater in low-income states offset the effects of higher state spending in high-income states. The bottom line is that federal and state spending per capita does not vary systematically with income.

Table 8. How Does State Medicaid/SCHIP Spending Change as State per Capita Income Increases?
  Coefficient T-Statistic
Medicaid/SCHIP Spending per Capita
    State and Federal -0.01 -0.04
    State Only 1.03 3.02
Medicaid/SCHIP Spending per Low-Income Person
    State and Federal 1.09 4.03
    State Only 2.13 6.64
Source: The Urban Institute 2001. Expenditures from HCFA 2082 and NASBO data. Population estimates from tabulations of the March 1998-2000 CPS.

However, the results in the spending per low-income person regressions are somewhat different. A one percent increase in income per capita is associated with a 1.09 percent increase in federal and state spending per low-income person. Thus, as state income increases so does state Medicaid spending per low-income person, including the federal matching funds. Regressing state expenditures (from states' own resources) per low-income person against income per capita reveals that a one percent increase in income per capita is associated with a 2.13 percent increase in state spending. Because Medicaid/SCHIP spending per low-income person from states' own resources increases so much as state incomes increase, the higher federal matching contributions for low- income states are not sufficient to offset the greater propensity to spend in higher income states.

Other State Spending on Health Services

It is sometimes argued that states that spend less on Medicaid do so because of their aversion to federal entitlement programs, but they make up for this by spending more on public hospitals and clinics and on other programs that directly provide services to low-income individuals. To examine this we use data from the National Association of State Budget Officers (NASBO) on several kinds of state spending on health care. NASBO reports expenditures on Medicaid and SCHIP; for consistency we chose to use our own Medicaid expenditure data and used federal reports on SCHIP expenditures in 1999.

NASBO provides data on state employee health benefits and on corrections; these are not related to the issue we wished to address and were excluded. NASBO also had data on state expenditures on higher education, insurance and access expansion initiatives, public health-related expenditures, state facility-based services, and community-based services. All of these can contribute to care for the poor. Higher education includes expenditures on state university-based teaching hospitals, and these hospitals usually provide significant amounts of care to the uninsured. State insurance and access expansions include state coverage initiatives as well as funding for high-risk pools and other insurance subsidies. Public health-related expenditures include spending on local health clinics, the Ryan White AIDS program, nonfederal Indian health care, pharmaceutical assistance for the elderly, childhood immunization programs, women, infants, and children (WIC) programs, AIDS testing, and breast and cervical cancer screening. State facility-based services include spending on state hospitals, state operated long-term care facilities, and a variety of other facilities. Community-based services include rehabilitation services, alcohol and drug abuse treatment, mental health community services, and vocational rehabilitation services.

Table 9. Federal and State Spending on Medicaid, SCHIP, and Other Services for the Low-Income Nonelderly
  Federal and State Spending State Spending
  Medicaid and SCHIP $ / Low-Income Pop Other State Health Programs $ / Low-Income Pop Medicaid and SCHIP $ / Low-Income Pop Other State Health Programs $ / Low-Income Pop
Texas $615 $510 $231 $372
Mississippi $729 $325 $166 $220
Alabama $756 $304 $230 $244
California $764 $394 $372 $260
Colorado $776 $457 $371 $218
Florida $795 $458 $350 $329
Wisconsin $870 $613 $358 $319
Minnesota $946 $631 $453 $573
New Jersey $1,029 $665 $512 $588
Michigan $1,122 $473 $520 $317
Washington $1,126 $553 $539 $438
Massachusetts $1,543 $1,243 $766 $1,028
New York $1,928 $762 $955 $584
Balance of Total $929 $550 $368 $404
Grand Total $956 $540 $412 $396
Source: The Urban Institute 2001. Expenditures from HCFA 2082, NASBO, and AHA data, with denominators from the March 1998-2000 CPS.

Some spending on local hospitals is financed with local revenues. We use American Hospital Association data on local revenues to estimate local government contributions by states. Unfortunately the problem with this data is that not all hospitals report. However, data from the hospitals that do report reveal that local government revenues amount to only 1.1 percent of hospital revenues. Thus, this underreporting could not have a significant effect on the results shown below.

Table 9 provides data on these various functions. We compare this with state spending on Medicaid, using the same data shown in table 7, together with the addition of SCHIP expenditures. We show this both for federal and state spending and for state-only expenditures; the table shows spending relative to the number of people below 200 percent of FPL.

The results show that state expenditures on these "other services" tend to track Medicaid expenditures. Expenditures are lowest in Alabama and Mississippi for these "direct services" and spending in these states is close to the bottom for Medicaid and SCHIP. Similarly, New York and Massachusetts spend the most for these services, just as they do for Medicaid and SCHIP. The picture remains the same when we look at state spending from their own resources. States with the lowest level of spending on those services (Mississippi, Alabama, and Colorado) are among the states with the lowest expenditures on Medicaid and SCHIP. And again, New York and Massachusetts are among the highest spending states for both Medicaid and non-Medicaid spending.

To test whether there was an inverse relationship between state and federal direct spending and Medicaid and SCHIP spending we used data from all 50 states to estimate simple regressions of direct spending against Medicaid and SCHIP expenditures. The results are shown in table 10. In each case there is a positive relationship; the higher spending on Medicaid and SCHIP the higher direct service spending. This is true whether we look at spending per capita or per low-income person and whether we look at
state expenditures from their own resources or federal and state spending.

Table 10. Is State Spending on Direct Health Services Related to Spending on Medicaid and SCHIP?
  Coefficient T-Statistic
State and Federal Spending
    All Incomes 0.18 1.92
    Low Income 0.28 2.74
State Spending
    All Incomes 0.34 2.04
    Low Income 0.53 3.40
Source: The Urban Institute 2001. Expenditures from HCFA 2082 and NASBO data. Estimates of the low income population from tabulations of the March 1998-2000 CPS.

Long-Term Care Expenditures

The previous tables in this paper have looked at Medicaid acute care spending. Medicaid also pays for a substantial amount of long-term care expenditures in the United States. Table 11 provides data on Medicaid long-term care spending per aged and disabled enrollee. Not all elderly and disabled enrollees use long-term care services, but this is as good an estimate of the eligible population as is available. Most long-term care spending is for nursing home care, while the fastest growing component is home- and community-based services. Table 11 shows that Medicaid expenditures for long-term care services per aged and disabled enrollee varies from $11,968 in New York and $11,682 in Minnesota to $3,167 in Alabama and $2,408 in Mississippi. The federal government pays more in low-income states than in high-income states. Thus, the discrepancies are greater if we look simply at state effort. Here the variations range from $5,984 in New York and $5,591 in Minnesota to $971 in Alabama and $552 in Mississippi. Clearly, long-term services available to aged and disabled individuals also vary enormously among states.

Table 11. Total and State Long-Term Care Expenditures per Enrollee for Aged, Blind, and Disabled Enrollees, by State, FFY 1998
  Expenditures per Enrollee
  Federal and State State
New York $11,968 $5,984
Minnesota $11,682 $5,591
Massachusetts $8,051 $4,026
Wisconsin $7,648 $3,148
New Jersey $7,504 $3,752
Washington $6,300 $3,014
Colorado $5,909 $2,838
Michigan $5,554 $2,578
Texas $4,468 $1,685
Florida $3,285 $1,457
California $3,208 $1,565
Alabama $3,167 $971
Mississippi $2,408 $552
Average for Rest of Nation $5,285 $2,151
United States $5,639 $2,484
Source: Urban Institute estimates based on data from the Centers for Medicare & Medicaid Services (CMS). Long-term care includes nursing facilities, intermediate care facilities for the mentally retarded, home- and community-based care for the functionally disabled elderly, home- and community-based waiver services, home care, personal care services, targeted case management, and hospice services. The denominator is the number of aged, blind, and disabled enrollees who participated in Medicaid for any length of time during the fiscal year. The number of people actually using these services cannot be accurately counted using publicly available data sources.

Variations in Access and Health Status

Table 12 shows the variations among the same 13 states in the likelihood of low-income children and adults having a usual source of care (as a measure of access) and the percent of the population in fair or poor health (as a measure of health). Among low-income children, states with the highest likelihood of lacking a usual source of care are California, Mississippi, and Texas, states where low-income children are least likely to lack a usual source of care are Massachusetts, Wisconsin, and Minnesota. Among low-income adults, those in Texas and California are more likely to lack a usual source of care and those in Massachusetts, Michigan, Minnesota, and Wisconsin are least likely. The results tend to track variations in uninsurance rates.

Table 12. Variations in Access and Health Status, Low Income Children And Adults
  Percent with No Usual Source of Care Percent in Fair or Poor Health
  Children Adults Children Adults
Texas 14.4% 27.5% 11.7% 27.0%
California 13.7% 26.8% 11.0% 26.6%
Mississippi 11.0% 15.2% 12.9% 28.8%
New Jersey 9.1% 20.4% 11.0% 27.3%
Florida 9.1% 22.6% 7.6% 26.8%
Alabama 9.0% 17.3% 8.1% 27.9%
Colorado 9.0% 19.7% 9.3% 16.1%
Washington 8.1% 22.0% 8.0% 20.3%
New York 7.8% 16.5% 9.7% 27.4%
Michigan 7.7% 14.8% 7.9% 23.5%
Wisconsin 6.4% 16.0% 6.0% 15.9%
Minnesota 3.8% 12.0% 4.9% 12.8%
Massachusetts 3.6% 14.6% 7.0% 21.3%
Balance of States 6.9% 16.6% 5.7% 21.8%
Total 8.9% 19.5% 7.9% 23.8%
Source: Urban Institute tabulations of the National Survey of America's Families (NSAF), 1999.

Similarly, low-income children are more likely to be reported to be in fair or poor health in California, Mississippi, New Jersey, and Texas. With the exception of New Jersey, all of these are states with high uninsurance rates. At the other extreme, Minnesota and Wisconsin have the lowest percentages of children in fair or poor health. Similarly, low-income adults in Alabama, Florida, Mississippi, New Jersey, and Texas are most likely to be in fair or poor health. Those in Minnesota, Wisconsin, and Colorado are least likely to be in fair or poor health.

These results are not intended to necessarily imply a causal relationship between insurance coverage, access to care, and health status, although these relationships have been established convincingly elsewhere.6 But the data clearly suggest that the states that have low levels of coverage and high uninsurance rates have not done so because they do not have access problems or that their low-income populations are in good health. The results show that with few exceptions the states that have higher uninsurance rates among low-income populations tend to be the same states with significant access problems and higher percentages of the population in fair or poor health.

Summary and Conclusions

There are widespread variations in uninsurance rates among states. Uninsurance rates vary by a factor of roughly two for low-income populations and about three for nonelderly populations as a whole—states with higher uninsurance rates tend to have larger low-income populations. These variations hold for both children and adults. There are substantial variations across the country in the extent of employer-sponsored insurance coverage primarily related to the skills and human capital embodied in the workforce. Variations in employer-sponsored insurance are major determinants of variations in uninsurance rates of children and adults when we look at state populations regardless of income. As a result, states face problems of very different dimensions, with some states having a very large gap to fill and with others a relatively small gap to address. Public programs vary depending on the extent of the initiatives that individual states have taken to extend coverage. When we look at the low-income populations both the extent of employer-sponsored coverage and the generosity of public programs affect uninsurance rates.

Spending on Medicaid as well as other direct service programs varies considerably among states. Medicaid spending varies by roughly a factor of 3 among our 13 states and by even more when we look at spending from states' own resources. State spending per low-income person on Medicaid and SCHIP increases sharply as state incomes increase such that federal matching contributions, which vary inversely with incomes, are not sufficient to reduce spending differentials; the result is that Medicaid spending (federal and state) increases in proportion to increases in state incomes. Finally, states with lower rates of public coverage and high uninsurance rates are also the states with more access problems, and more of their low-income populations are more likely to be in fair or poor health.

One can only conclude from this evidence that the existing federal-state structure has left us with serious equity problems. The redistribution of resources, as intended by the Medicaid matching formula, which varies federal contributions inversely with state financial capacity, has simply not resulted in anything close to equity in coverage and in the resources that are expended on low-income populations.

States are not wholly to blame for this outcome. The problems they face vary greatly because of differences in their employment base. Low rates of employer coverage raise the fiscal stakes for states and have proven very difficult to overcome. But there are also major differences in state efforts. Some of these variations reflect differences in ability to pay but much reflects differences in the willingness of states to devote resources to health care for low-income populations. This is a matter of political philosophy and political will and is perhaps to be expected in a system with prominent decisionmaking roles devolved to states. Whether the variation in outcomes is acceptable is another matter. Unfortunately, the resulting differences in lack of coverage affect access and health, and ultimately work effort, productivity, and contributions to national well-being. In this sense, decisions states make with respect to their own citizens ultimately affect the citizens of other states.

The results suggest that states need to be given stronger incentives to go further in providing health care for the low-income population. Possibly with higher matching rates and more flexibility over program design as in the current SCHIP program, states may be induced to make greater efforts to expand coverage, expand more resources, and improve outcomes. With strong enough incentives, even the most reluctant states may be unable to ignore the availability of federal funds. But it is also likely that this would not be sufficient. If it is not, achieving a reasonable degree of equity in coverage and access may require shifting of the full responsibility for financing health care for low-income populations to the federal government.


1 For a more thorough summary of these state programs, see John Holahan and Mary Beth Pohl, "States as Innovators in Coverage," Urban Institute Discussion Paper 02-08, forthcoming.

2 Brenda C. Spillman, "Adults without Health Insurance: Do State Policies Matter?" Health Affairs 19(4):178-87.

3 Cantor, Joel C., Stephen H. Long, and M. Susan Marquis. 1995. "Private Employment-Based Health Insurance in Ten States." Health Affairs Summer 1995: 199-211; Cunningham, Peter J., and Paul B. Ginsburg. 2001. "What Accounts for Differences in Uninsurance Rates Across Communities?" Inquiry 38:6-21; Farber, Henry S., and Helen Levy. 1998. "Recent Trends in Employer-Sponsored Health Insurance Coverage: Are Bad Jobs Getting Worse?" National Bureau of Economic Research Working Paper 6709; Hadley, Jack, Randall Bovbjerg, and Marc Rockmore, "Factors Associated with Variations Across States and Communities in Low-Income Workers Insurance Coverage," unpublished paper; Long, Stephen H., and M. Susan Marquis. 1993. "Gaps in Employer Coverage: Lack of Supply or Lack of Demand?" Health Affairs Supplement 1993: 282-293; Trenholm, Christopher, and Susanna Kung. 2000. "Disparities in State Health Insurance Coverage: A Matter of Policy or Fortune?" Academy for Health Services Research and Health Policy.

4 Yu-Chu Shen and Stephen Zuckerman, Why is There State Variation in Employer-Sponsored Insurance?, Urban Institute working paper, forthcoming.

5 Lisa Dubay "Expanding Public Insurance Coverage and Crowd-Out: A Review of the Evidence." Opinions for Expanding Health Insurance Coverage: What Difference Do Different Approaches Make? Judith Feder and Sheila Burke, editors (Washington D.C.: Henry J. Kaiser Foundation, 1999).

6 Jack Hadley "Sicker and Poorer: The Consequences of Being Uninsured," Kaiser Family Foundation (Washington, D.C.: May 2002) www.kaiser.org.

See the pdf version for Appendix Tables

Topics/Tags: | Governing | Health/Healthcare

Usage and reprints: Most publications may be downloaded free of charge from the web site and may be used and copies made for research, academic, policy or other non-commercial purposes. Proper attribution is required. Posting UI research papers on other websites is permitted subject to prior approval from the Urban Institute—contact publicaffairs@urban.org.

If you are unable to access or print the PDF document please contact us or call the Publications Office at (202) 261-5687.

Disclaimer: The nonpartisan Urban Institute publishes studies, reports, and books on timely topics worthy of public consideration. The views expressed are those of the authors and should not be attributed to the Urban Institute, its trustees, or its funders. Copyright of the written materials contained within the Urban Institute website is owned or controlled by the Urban Institute.

Email this Page