|
D A T A W A T C H
E X P A N S I O N P R O P O S A L S
7 June 2005
Variations In The Impact Of
Health Coverage Expansion Proposals
Across States
Statewide variation cancels out federal uniformity of benefits,
with varying effects on the uninsured across the country.
By Sherry Glied and Douglas Gould
ABSTRACT:
Most estimates of the consequences of alternative health insurance proposals focus on national impact, but the extent of cross-state diversity in uninsurance rates, economic and labor-market characteristics, and health care markets suggests that the impact of strategies will also vary. We illustrate this variation by comparing the effects of standard tax credit and Medicaid expansion proposals across states. Some states do well (or poorly) under all policies; others benefit under some but not others. Across policies, state effects on uninsurance rates vary by at least a factor of 2.5. Uniform national strategies that target the uninsured do not generate uniform national outcomes.
The fifty states differ in their demographic and economic circumstances and in their health insurance markets. This variability has animated discussion about how health policy responsibilities should be divided between the federal and state governments.1 Yet estimates of the consequences of alternative federal health insurance expansion strategies generally ignore this diversity and focus exclusively on national impact. Because of the extent and nature of cross-state variation, different national expansion strategies, applied uniformly across states, are likely to have quite different impacts in different states.
Consider the variation in current insurance coverage rates. In Rhode Island the three-year average uninsurance rate (2000–2002) among the nonelderly was 9.2 percent, while in Texas it was 26.3 percent (Exhibit 1). The effect of any new federal policies—however designed—on the total number of uninsured people is likely to be more limited in Rhode Island, where a larger share of people already have coverage, than in Texas.
Most expansion proposals target recipients by family income or poverty status. In Louisiana, 27 percent of all families have incomes below the federal poverty level. In New Hampshire, by contrast, only 10.5 percent do. In high-income states such as New Hampshire, a smaller share of the population will be eligible under any given income or poverty threshold than in low-income states such as Louisiana. Similarly, some policies target the uninsured in small firms, but the degree to which uninsured people are concentrated in such firms varies by state.
State history also matters. In some states, state expansion policies have made many people eligible for public coverage.2 Federal policies that expand eligibility for public programs are likely to have little effect on participation among those already eligible. Finally, states vary in their health insurance markets and in the cost of private coverage. A fixed tax credit will have a larger impact on participation in states where premiums are low than in those where premiums are high.
In this paper we examine the impacts of five types of insurance expansion policies on coverage by state. First, we consider refundable tax credits for the non-group market, a proposal of the type that has been advanced by the Bush administration. Next, we analyze tax credits for small-firm workers and their dependents. Finally, we consider three types of policies that increase the eligibility limits for Medicaid or the State Children’s Health Insurance Program (SCHIP): an expansion of Medicaid eligibility to include all low-income adults, an expansion to low-income uninsured children not now eligible for SCHIP, and an extension of Medicaid eligibility to all parents of SCHIP-eligible children.
Policy Types And Existing National Estimates
We began by examining state characteristics that are likely to be important in generating variation across states. We then performed estimates of the effect of a set of standard strategies that have been evaluated in prior research, focusing on the diversity in their effects among states. In each case, we calibrated our baseline estimates so that the national effects of each strategy corresponded to those in existing widely accepted and published estimates. We then classified states into quintiles according to how a policy affects that state’s uninsurance rate relative to the national average. Details of our modeling strategy and comparisons of our estimates to previously published figures are available in an online technical appendix.3
Tax credit proposals. Under the Bush administration’s tax credit proposal, nonelderly people who buy nongroup insurance (not public or employer-sponsored insurance plans) would receive tax credits that vary with the recipients’ adjusted gross incomes.4 The maximum tax credit would be 90 percent of the premium or $1,000 per adult and $500 per child, or $3,000 for families with two adults and two or more children. The poorest Americans (single adults with less than $15,000 in adjusted gross income or families with less than $25,000) would receive the maximum tax credits, which would be phased down for those with higher incomes. Eligibility would phase out at $30,000 for single adults and $60,000 for families. Jonathan Gruber concluded in 2002 that 3.3 million people would become newly insured through the tax credit proposal, while the Lewin Group estimated in 2000 that the net increase in insured people would be 3.8 million.5
We also considered a second tax credit proposal: the small-firm tax credit. This is a credit available to all employees of firms with twenty-five workers or fewer and their dependents. The tax credits cover 25 percent of recipients’ employer-sponsored insurance premiums. The estimated number of newly insured people under this proposal ranges from 0.5 million (James Reschovsky and Jack Hadley) to 2.6 million (Lewin Group).6
Public insurance expansion proposals. The first proposal we considered would expand the federal limit on Medicaid coverage to 133 percent of the federal poverty level. In most states today, even the poorest U.S. adults must also be pregnant, disabled, or otherwise classified as “medically needy” to participate in Medicaid, which leaves many poor people ineligible.7 The expansion we examined would be available to all nonelderly adults (ages 19–65) with annual household incomes below 133 percent of poverty, regardless of family structure or medical condition.
The Lewin Group estimated that if Medicaid eligibility were increased to 133 percent of poverty for childless adults, in the context of other expansions targeting parents, 4.7 million previously uninsured, childless adults would become insured. Bill Custer and Tom Wildsmith, and Judith Feder and colleagues, estimated in 1999 that six to seven million adults would gain coverage if eligibility were expanded to 100 percent of poverty.8
The second proposal we evaluated expands the federal limit on SCHIP coverage to 300 percent of poverty. As of March 2005, six states have raised SCHIP eligibility to 300 percent of poverty or higher.9 We assumed that all states would raise their eligibility levels to the federal limit. This overstates the likely impact of this policy in states that might not choose to participate at this level. Ken Thorpe and Curtis Florence estimated in 1999 that an expansion of SCHIP eligibility to 300 percent of poverty would raise the population of insured children by 2.4 million.10 Feder and colleagues estimated this number at 3.7 million.11
The third proposal we evaluated would allow all parents of SCHIP-eligible children to become eligible for Medicaid. Several published reports concluded that the likelihood of a child’s being insured increases as his or her parent becomes insured.12 Thorpe and Florence, and Feder and colleagues, estimated that 2–3.8 million parents would gain coverage under such an expansion.13
Study Data And Methods
We combined three years of the March Current Population Survey (CPS)—2001, 2002, and 2003—for our analyses and restricted our sample to those under age sixty-five. We matched data from the March 2001 CPS with information on health insurance offers (not available in the March files) from the February 2001 CPS, the most recent available offer data. Exhibit 1 lists the uninsurance rate for each state, computed from these data.
The CPS does not include information on job-based or nongroup health insurance premiums. We used tabulations of the Medical Expenditure Panel Survey Insurance Component (MEPS-IC) data as our source for group premium data.14 We computed nongroup insurance premiums using data from eHealthInsurance and from Jon Gabel and colleagues, modified using methods described in the technical appendix and adjusted for age, sex, health status, and rate regulation practices of each state so that they conform to observed premiums in the group market.15
We calibrated our national estimates to published estimates that did not incorporate the effects of expansion proposals on the behavior of employers that now offer coverage. There is much disagreement about how to model employers’ behavior. In practice, modelers generally use methods in which the magnitude of behavior changes is proportional to the number of people gaining eligibility for new coverage. This means that although national estimates of the effects of proposals would certainly change with the inclusion of employers’ behavior, the relative ranking of states, our focus here, would not be greatly affected.
Variations Among States
Exhibit 2 contains the characteristics of each state that are most important in assessing the varying impact of different expansion proposals.
Percentage of uninsured people living in poverty. All of the expansion policies we considered, except the small-firm tax credit, target uninsured people and determine eligibility based on income. Under these policies, the cutoffs for eligibility are set nationally. Targeting subsidies to lower-income people makes sense. In general, people with higher incomes are better able to afford unsubsidized coverage, and more people are covered by employer-sponsored insurance in higher-income states.16 But national income cutoffs do not account for state-level differences in the cost of health care or cost of living more generally. Although absolute incomes are much higher in Alaska than in Oklahoma, for example, the much higher cost of living in Alaska leaves the two states with similarly high uninsurance rates but very different eligibility for income-based expansion proposals.
The percentage of a state’s uninsured population with incomes below 100 percent of poverty is an important indicator of the extent to which income-based coverage policies would expand eligibility in that state (Exhibit 2). There is a nearly twofold difference among states in the percentage of the uninsured earning less than 100 percent of poverty. Identical income-targeted policies will expand eligibility much more in Hawaii, where 43 percent of the uninsured earn less than 100 percent of poverty, than in New Hampshire, where only 24 percent do.
Percentage of uninsured people tied to small firms. In Vermont, Idaho, and Montana, about 60 percent of uninsured people have a connection to a small firm. By contrast, in several states, including Virginia; Washington, D.C.; and Louisiana, fewer than 40 percent of uninsured people are linked to small firms (Exhibit 2).
The firm-size distribution of employment is a function of local industrial patterns and population density. Small firms are more common in rural than in urban areas.17 Policy proposals that target the small-group market will be more effective in states with a small firm–oriented distribution of employment than in states with more large firms.
Cost of health insurance in the nongroup market. The cost of health care, and of health insurance, varies among states. Controlling for income, states with higher health care costs tend to have higher uninsurance rates.18 In public program expansions, public payers absorb cross-state variation in costs. In Medicare, for example, costs per beneficiary are more than twice as high in Miami as in Minnesota.19 The higher costs in Miami are spread among taxpayers nationally. In Medicaid, the federal match means that additional expenses associated with higher health care costs are divided between state and federal taxpayers.
In the tax credit proposal, by contrast, the amount of the individual subsidy is fixed nationally. Individuals, not the federal or state governments, must bear the burden of higher-than-average health care costs. A tax credit will cover a larger share of the cost of insurance—and presumably lead more people to take up coverage—in areas where existing nongroup premiums are relatively low. In Utah and Kansas, we estimate that nongroup premiums average less than $2,000 per person. In Maine, New Hampshire, New Jersey, and New York, they average more than $4,000 per person (Exhibit 2). An equal-size tax credit would have much smaller effects in the latter states than in the former.
The problem of high health care costs is compounded in the many states with high health insurance costs that also have high costs of living and a distribution of income that is above the national average. A person’s ability to afford health insurance depends on the relationship between income and the costs of living and of health care. Moderate-income residents of states with high health care costs might find insurance just as unaffordable as do lower-income residents of states with low health care costs. Under an income-targeted tax credit program, moderate-income residents of states with high health costs will be eligible only for partial tax credits that will cover only a small share of the cost of insurance.
In unregulated health insurance markets, nongroup premiums also vary with age, sex, and health status. Older, less healthy people in unregulated states might find coverage very costly, even if the average nongroup policy price is low. On the other hand, many of the states with the highest average nongroup premiums employ some form of community rating.20 The presence of community rating drives up average premiums because the insured risk pool in community-rating states tends to be older and less healthy. High average premiums in these states mean that coverage may be particularly inaccessible to young, healthy people.
Eligibility levels of public insurance programs. States already differ greatly in the extent to which they have expanded their Medicaid and SCHIP programs.21 Seven states have already expanded coverage for single adults to 100 percent of poverty, and twenty have expanded coverage for parents to this level or higher (Exhibit 2). A federal expansion to 100 percent of poverty would be useful for state finances (and might permit these states to expand coverage higher up the income scale) but would not directly affect uninsurance rates in these states. In six states, SCHIP/ Medicaid eligibility for children is at 300 percent of poverty or higher. Again, a federal expansion to this level would alter the financing of these programs in these states but would not likely lead to reductions in uninsurance rates.
Study Results
We assigned states to quintiles according to the effect of a proposal on their uninsurance rate, relative to the national average effect (Exhibit 3). We prefer to report our results in quintiles because small variations in rates are likely meaningless, given the uncertainties of the estimation process.
Tax credits. Most uninsured people are eligible for at least a partial tax credit (Exhibit 3). We estimated that 11 percent of the currently uninsured population (about 4.6 million people) would gain health coverage under the proposal, reducing the overall U.S. uninsurance rate by 1.7 percentage points. Declines in the uninsurance rate by state vary by a factor of nearly 5: from 4.4 percent in New Hampshire to 20.5 percent in Utah. In the median state (Connecticut), 10.6 percent of the uninsured population would gain coverage under the tax credit. The states that would experience the greatest percentage declines in their uninsured populations under this proposal, such as Utah, Kansas, and Oregon, share two features: a low average nongroup premium and a large proportion of the uninsured population earning less than 100 percent of poverty and hence eligible for the full credit.
Small-firm tax credits. The small-firm tax credit proposal would have a much smaller aggregate impact than the individual tax credit (Exhibit 3). It is estimated to increase the number of newly insured Americans by 1.4 million people, about 3.3 percent of the uninsured. The effect across states varies by a factor of 2.4: from 2 percent of the uninsured in Washington, D.C., to 4.7 percent in Montana. All of the states that would experience greater-than-average increases in their insured populations, such as Montana, Idaho, and Maine, have relatively large proportions of their nonelderly, uninsured populations employed in (or dependents of employees in) small firms.
Low-income expansion. The proposal to expand public coverage to adults with incomes below 133 percent of poverty would decrease the U.S. uninsurance rate by 1.9 percentage points and increase the number of insured Americans by 4.7 million, approximately 11.5 percent of the uninsured population. The range of effects across states, however, is much wider. The effects of this policy vary so greatly because several states have already enacted expansions for this population.22 Thus, the decrease in the uninsured population ranges from no effect in states with existing adult Medicaid eligibility limits above 133 percent of poverty, such as Vermont, Utah, and Massachusetts, to a high of 18.3 percent in Alabama and West Virginia. States that would be affected most by this proposal have not yet expanded coverage and have large shares of their uninsured populations living in poverty and lacking an existing offer of employer-sponsored coverage. Although federal eligibility expansions would have little impact on coverage in states that have already expanded eligibility, new federal funding might relieve a share of the state burden there.
SCHIP eligibility levels. In aggregate, expanding SCHIP to children under 300 percent of poverty would make 1.0 million of the 9.3 million uninsured children eligible for SCHIP. Approximately 420,000 of the newly eligible children, however, are expected to take up public insurance. Historically, take-up rates for SCHIP have been inversely proportional to income, so that expansions to higher-income children tend to have smaller effects. The program is thus estimated to insure about 4.5 percent of uninsured children. The range of effects on the entire uninsured population is expected to vary from zero in states, such as New Jersey, Missouri, and Maryland, where eligibility already exceeds this level, to 4.7 percent in states, such as South Carolina, with very low existing SCHIP eligibility. In the median state (Iowa) the overall effect would be 1.1 percent.
Medicaid for parents of SCHIP-eligible children. Although the immediate aim of the proposed Medicaid expansion policy to parents of SCHIP-eligible children is to provide uninsured adults with health coverage, it also would affect uninsured children because children are more likely to be enrolled in coverage if their parents are also eligible. This proposal is estimated to increase the number of insured Americans by approximately two million, about 5.2 percent of the U.S. uninsured population. The effects by state range from 0.7 percent of the uninsured in Tennessee to 10.3 percent in Arkansas, with a median of 4.7 percent (South Dakota). The effects are greatest in states with the largest populations of low-income families.
Effects on range of uninsurance rates. Each of the proposals considered here would narrow the range of uninsurance rates across the states (Exhibit 3). All would have their greatest effects in the states with the highest uninsurance rates. But the extent of narrowing is modest, even for the three policies with the largest national effects. Prior to reform, the nonelderly uninsurance rates among the states had a range of 17.1 percentage points, varying from 9.2 percent to 26.3 percent (see Exhibit 1). Under the expansion to parents of SCHIP children, the range would narrow by 2.4 percentage points (Exhibit 3). The tax credit reform would have more than twice as large an aggregate impact but slightly less effect on the range, narrowing it by just 1.4 percentage points. The expansion to low-income adults would have the largest effect on the number of uninsured people and would narrow the range of uninsurance rates by 2.3 percentage points.
Discussion And Policy Implications
There is no such thing as an “average” state. Thus, the national effects of policies can be misleading. A policy that serves one state well may be relatively ineffective in a neighboring state. Some states would do relatively well—and others would do relatively poorly—under all of the federal reform proposals we have considered here. States with low nongroup premiums, low average incomes, and few prior expansion efforts would tend to do well under all proposals. Others, such as Maryland, Massachusetts, New York, and Wisconsin, which have already undertaken expansions and have high nongroup premiums, would do relatively worse than average under all proposals. States with relatively low health care costs and moderate incomes (such as California, Oregon, and Washington) would do better with tax credits than with public expansions; lower-income states with moderately high health costs (such as Alabama, Kentucky, and West Virginia) would do relatively better under public expansions.
The debate about whether the proper locus of health policy is state or federal often contrasts interstate variability generated by state experimentation with the uniformity of federal health policy. Some analysts argue that federal health policymakers have greater health care expertise, better revenue collection abilities, and the ability to avoid the race to the bottom inherent in intrastate competition. Others assert that state policymakers can use their understanding of local conditions to craft policies that best reflect states’ values and priorities.
Our results suggest that in the context of expansions to low-income uninsured populations, this theoretical dichotomy does not apply. The factors that generate variation in uninsurance rates among states will themselves affect the application of federal policy. Uniform federal income eligibility limits will leave more people uninsured in high-income states where the cost of living is high. Uniform expansions in eligibility will have less effect on uninsured people in states that generously subsidize safety-net providers and have expanded public program eligibility. Uniform tax credits will provide less benefit to people in states where the cost of medical care is high. In each of these situations, layering uniform national policies over the existing differences among states will not necessarily narrow these differences. Instead, a policy goal of interstate uniformity in outcomes may require interstate variability in policies.
The authors thank the Commonwealth Fund Task Force on the Future of Health Insurance for research support.
NOTES
1. J. Holahan, A. Weil, and J.M. Wiener, “Which Way for Federalism and Health Policy?” Health Affairs,16 July 2003, content.healthaffairs.org/cgi/content/abstract/hlthaff.w3.317 (18 March 2005); and M. Sparer, Medicaid and the Limits of State Health Reform (Philadelphia: Temple University Press, 1996).
2. J. Holahan and M. Pohl, “Leaders and Laggards in State Coverage Expansions,” in Federalism and Health Policy, ed. J. Holahan, A. Weil, and J.M. Wiener (Washington: Urban Institute Press, 2003), 179–214.
3. The online appendix is available at content.healthaffairs.org/cgi/content/full/hlthaff.w5.259/DC2.
4. U.S. Department of the Treasury, General Explanations of the Administration’s Fiscal Year 2005 Revenue Proposals, February 2005, www.treas.gov/offices/tax-policy/library/bluebk05.pdf (10 May 2005).
5. House Committee on Ways and Means, “Health Care Tax Credits to Decrease the Number of Uninsured, Statement of Jonathan Gruber, Ph.D.,” Serial no. 107-58 (Washington: U.S. Government Printing Office, 2002), 123–128; and Lewin Group, Health Coverage 2000: Cost and Coverage Analysis of Eight Proposals to Expand Health Insurance Coverage (Princeton, N.J.: Robert Wood Johnson Foundation, 2000).
6. J.D. Reschovsky and J. Hadley, “Employer Health Insurance Premium Subsidies Unlikely to Enhance Coverage Significantly,” Issue Brief no. 46, December 2001, www.hschange.org/CONTENT/392/392.pdf (10 May 2005); and Lewin Group, Covering America: Cost and Coverage Analysis of Ten Proposals to Expand Health Insurance Coverage (Princeton, N.J.: RWJF, 2003).
7. In only eight of the fifty-one states (including Washington, D.C.) are all adults with incomes below 100 percent of the federal poverty level eligible for Medicaid.
8. Lewin Group, Health Coverage 2000; W.S. Custer and T.F. Wildsmith, “Estimated Cost and Coverage Impact of the HIAA Proposal to Cover the Uninsured,” 17 May 1999, www.insureusa.org/plan/costreport.htm (8 September 2004); and J. Feder, C. Uccello, and E. O’Brien, The Difference Different Approaches Make: Comparing Proposals to Expand Health Insurance, Pub. no. 1532 (Menlo Park, Calif.: Henry J. Kaiser Family Foundation, 1999).
9. Kaiser Family Foundation, “Income Eligibility Levels for Children under SCHIP, as a Percent of Federal Poverty Level (FPL), 2004,” July 2004, www.statehealthfacts.org/cgi-bin/healthfacts.cgi?action=compare&category=Medicaid+%26+SCHIP&subcategory=Children%27s+Medicaid+and+SCHIP+Eligibility&topic=Income+Eligibility%2d%2d+Separate+SCHIP (18 March 2005).
10. K.E. Thorpe and C.S. Florence, “Covering Uninsured Children and Their Parents: Estimated Costs and Number of Newly Insured,” Medical Care Research and Review 56, no. 2 (1999): 197–214.
11. Feder et al., The Difference Different Approaches Make.
12. See, for example, L. Ku and M. Broaddus, The Importance of Family-based Health Insurance Expansions: New Research Findings about State Health Reforms (Washington: Center on Budget and Policy Priorities, 2000); J.M. Lambrew, Health Insurance: A Family Affair—A National Profile and State-by-State Analysis of Uninsured Parents and Their Children (New York: Commonwealth Fund, 2001); and L. Dubay and G. Kenney, Covering Parents through Medicaid and SCHIP: Potential Benefits to Low-Income Parents and Children, Pub. no. 4022 (Menlo Park, Calif.: Henry J. Kaiser Family Foundation, 2001).
13. Thorpe and Florence, “Covering Uninsured Children”; and Feder et al., The Difference Different Approaches Make.
14. Agency for Healthcare Research and Quality, “Table I.C.1: Average Total Single Premium (in Dollars) Per Enrolled Employee at Private-Sector Establishments That Offer Health Insurance by Firm Size and Selected Characteristics: United States, 2001,” 2001, www.meps.ahrq.gov/MEPSDATA/ic/2001/Tables_I/TIC1.htm (10 May 2005).
15. J. Gabel et al., “Individual Insurance: How Much Financial Protection Does It Provide?” Health Affairs, 17 April 2002, content.healthaffairs.org/cgi/content/abstract/hlthaff.w2.172 (18 March 2005); ehealthInsurance, “The Costs and Benefits of Individual Health Insurance Plans,” UITxCreditFactSheet_9.25.02, September 2002, available from the authors upon request (send e-mail to Sag1{at}columbia.edu); and Georgetown University Health Policy Institute, Health Insurance Info, 2004, www.healthinsuranceinfo.net (11 March 2005). Rate Regulation Practices are available under “Read the Consumer Guides On-line” in lower left area of site.
16. J. Holahan, “Variation in Health Insurance Coverage and Medical Expenditures: How Much Is Too Much?” in Federalism and Health Policy, 111–144.
17. B. Headd, “The Characteristics of Small-Business Employees,” Monthly Labor Review (April 2000): 13–18, available online at stats.bls.gov/opub/mlr/2000/04/art3full.pdf (10 May 2005).
18. S. Glied and K. Jack, “Macroeconomic Conditions, Health Care Costs, and the Distribution of Health Insurance,” NBER Working Paper no. w10029 (Cambridge, Mass.: National Bureau of Economic Research, 2003); and M. Chernew, D. Cutler, and P.S. Keenan, “Increasing Health Insurance Costs and the Decline in Insurance Coverage,” Health Services Research (forthcoming).
19. J.E. Wennberg, E.S. Fisher, and J.S. Skinner, “Geography and the Debate over Medicare Reform,” Health Affairs, 13 February 2002, content.healthaffairs.org/cgi/content/abstract/hlthaff.w2.96 (21 March 2005).
20. Kaiser Family Foundation, “Individual Insurance Market Reforms.”
21. Holahan and Pohl, “Leaders and Laggards.”
22. State Coverage Initiatives, “State Coverage Matrix: Medicaid Section 1115 Waivers,” statecoverage.net/medicaid-1115.htm (10 May 2005).
Sherry Glied (Sag1{at}columbia.edu) is a professor of health policy and management and department chair, Mailman School of Public Health, Columbia University, in New York City. Douglas Gould is a senior programmer there.
DOI:
10.1377/hlthaff.w5.259
©2005 Project HOPE–The People-to-People Health
Foundation, Inc.
|
