Public And Private Health Insurance: Stacking Up The Costs
- Leighton Ku (Leighton.Ku{at}gwumc.edu)
Abstract
Some proposals to expand health insurance coverage for people with low incomes are based on expansions of public programs, such as Medicaid or the State Children’s Health Insurance Program (SCHIP), while others rely on the use of tax subsidies for individuals to purchase private insurance. Analyses of data from the 2005 Medical Expenditure Panel Survey indicate that total medical spending is much lower when coverage is provided by Medicaid or SCHIP than it is when coverage is provided by private insurance. Public insurance is particularly advantageous from the consumer’s perspective because associated out-of-pocket spending is far lower.
Out-of-pocket health spending, not provider payment rates, primarily accounts for the large difference in costs between those covered by Medicaid and those covered by private health insurance.
A perennial issue in health reform proposals is whether to expand coverage for low-income people, who are the majority of the uninsured, through public programs such as Medicaid or the State Children’s Health Insurance Program (SCHIP) or through the use of tax subsidies for individuals to purchase private health insurance. For example, in explaining President George W. Bush’s veto of legislation to reauthorize and expand SCHIP, the White House objected because “our goal should be to move children who have no insurance to private coverage,” not to public coverage.1 Many of the objections raised by the president and other critics of the SCHIP expansion reflect concerns that public insurance coverage might “crowd out” private coverage. The Bush administration favors using tax deductions to subsidize private insurance purchases and letting states divert funds from Medicaid to subsidize “basic, private coverage.”2 Republican presidential candidate John McCain has similarly supported tax subsidies to purchase private insurance and not the expansion of public programs. In contrast, during the primary season, Democratic presidential hopeful Barack Obama supported expanding public programs for low-income people, as well as providing subsidies to help them purchase private insurance.3
In light of concerns about escalating national health spending, the comparative cost of public and private health insurance is an important issue. Both public and private health spending is rising faster than the rest of the economy.4 Some intuitively believe that a public insurance expansion would be more costly than expanding private insurance because Medicaid covers a wider range of benefits than those typically covered by private health insurance and requires less patient cost sharing. On the other hand, Medicaid provider payment rates are typically lower than those offered by private insurers. Analyses by Jack Hadley and John Holahan indicate that total per person medical spending was lower under Medicaid than under private insurance, after controlling for differences in insured people’s health and other risk characteristics.5
In addition to considering total spending, it is also important to consider costs from the consumer’s perspective. An insurance plan that offers sparse benefits or that has very high cost-sharing burdens could be inexpensive for the payer but quite burdensome for the beneficiary. For example, a lower-premium plan that includes a $1,000–$2,000 deductible might be affordable from the payer’s perspective, but a low- or moderate-income consumer might have great difficulty meeting the deductible and forgo effective primary care services or medications.
This study updates and extends the analyses conducted by Hadley and Holahan.6 We update the data used to 2005 (from 1996–1999), we examine differences in out-of-pocket spending under Medicaid and private insurance, and we estimate the effects of providing either Medicaid or private health insurance to low-income uninsured adults and children.
Study Data And Methods
Data source.
We analyzed the 2005 full-year consolidated file of the Medical Expenditure Panel Survey (MEPS), a nationally representative survey of noninstitutionalized U.S. households. In five interviews conducted over two and a half years, MEPS collects detailed information about individuals’ demographic and health characteristics and medical care use. The Agency for Healthcare Research and Quality (AHRQ) also conducts a survey of medical providers, to validate these self-reported utilization data and add medical spending estimates.7
Study sample.
We examined nonelderly adults (ages 19–64) and children (ages 0–18) with incomes below 200 percent of the federal poverty level (in the continental United States, about $32,180 for a family of three in 2005). We focused on those who were (1) insured by Medicaid or SCHIP for twelve full months during 2005; (2) insured by private health insurance for twelve months; or (3) uninsured for the full year. Those with part-year health insurance coverage were excluded from the analysis. (To exclude those with severe disabilities, we excluded Medicaid beneficiaries receiving Supplemental Security Income [SSI] benefits.)
Definitions.
Total medical expenditures include payments to health care providers from any source, including Medicaid; SCHIP; employer-sponsored insurance; individual insurance; Medicare; out-of-pocket payments; Department of Veterans Affairs (VA); workers’ compensation; other federal, state, or local insurers; other charitable sources; and free care or bad debt provided by public facilities. Medical expenditures are defined as payments to providers, not premiums or capitation payments paid to insurers. They do not include nursing home expenditures or Medicaid disproportionate-share hospital (DSH) payments.
Out-of-pocket payments include self-paid payments for deductibles, copayments, or coinsurance or payments for noncovered services; they do not include what consumers pay for health insurance premiums. Out-of-pocket payments were largely self-reported by respondents, although they were also imputed or edited based on other information, including the survey of providers.8
Spending models.
We developed two models of medical spending: (1) for whether a person had any medical spending (or out-of-pocket spending); and (2) for the level of annual medical spending among those who had any such spending. The models controlled for type of insurance, a battery of health measures, and demographic characteristics of adults and children.9
Study Findings
Differences in health conditions and other characteristics, 2005.
Health status differences.
We found important variations in the characteristics of low-income people whose insurance status differed in 2005, variations that affected their medical care use and spending. Adults with Medicaid coverage for a full year were more likely than low-income adults who had private insurance for the full year to be in fair or poor health, to have fair or poor mental health, to have activity limitations, to have chronic health conditions, or to have experienced a pregnancy (Exhibit 1⇓). By every measure, adult Medicaid recipients had much higher medical needs, even after those with severe disabilities were excluded. Uninsured adults appeared to be somewhat healthier than Medicaid recipients but not as healthy as the privately insured, although they were the least likely of the three groups to have chronic health problems or to have had a baby. There were fewer health differences among the three groups of children. Overall, children had many fewer health problems than adults.
Selected Health And Demographic Characteristics Of Low-Income Adults And Children, By Full-Year Insurance Status In 2005
Demographic differences.
There were other important differences in the low-income groups shown in Exhibit 1⇑. About three-quarters of Medicaid adults were women, compared to about half of the privately insured and uninsured. Medicaid adults were more likely than privately insured adults to be African American or Hispanic, although half of the Medicaid adults were white or “other.” Uninsured adults were more likely than insured adults to be Hispanic. Compared to privately insured low-income adults, adults with Medicaid coverage and the uninsured were less educated, less likely to be employed, and poorer. Children with Medicaid or SCHIP were more likely than the privately insured to be from a minority ethnic group, and many uninsured children were Hispanic. Also, Medicaid/SCHIP and uninsured children tended to be poorer than privately insured children.
Differences in unadjusted medical spending, 2005.
Unadjusted total annual per person medical spending for low-income adults covered by Medicaid for the full year was higher than spending for full-year, privately insured adults (Exhibit 2⇓). Uninsured adults had the lowest total mean spending. The main reason that adults with Medicaid coverage had higher spending than the privately insured is that they were more likely to have serious health conditions. Not surprisingly, uninsured adults had the lowest spending because they obtained less care, even though their health status appeared to be slightly worse than that of low-income privately insured adults. Unadjusted average out-of-pocket spending for low-income adults was lowest for those with Medicaid coverage; the uninsured and those with private insurance spent close to the same amount (Exhibit 2⇓).
Unadjusted Annual Per Person Medical Spending On Low-Income Adults And Children, By Full-Year Insurance Status In 2005
Unadjusted total annual medical spending for low-income children was similar for children covered by Medicaid or SCHIP and those covered by private health insurance and was lower for uninsured children (Exhibit 2⇑). Out-of-pocket spending was similar for uninsured children and those with private insurance but was lower for those with Medicaid or SCHIP.
Although the predominant payer for Medicaid recipients was Medicaid, some payments were also made by Medicare, private insurance, and other sources. Some people with Medicaid coverage for a full year may be covered by other insurers for a portion of the year. Similarly, low-income adults covered by private insurance for the full year could have payments from Medicaid, Medicare, or other sources, too, although private insurance was their dominant payer in 2005. The role of other payment sources was particularly important for the uninsured. Total medical spending for the uninsured exceeded their out-of-pocket payments by two- to threefold, which indicates the contributions of uncompensated care.
Comparing Medicaid and private insurance spending in 2005.
To make Medicaid and privately insured individuals more comparable, we used multivariate analysis and simulation methods. We pooled the full-year Medicaid and privately insured groups and conducted multivariate analyses of the effects of private versus public insurance coverage on annual per person total and out-of-pocket medical spending, controlling for factors such as health conditions, age, sex, race/ethnicity, education, and region of the country. The detailed results of these analyses are contained in an appendix to this paper.10 Briefly, the analyses found that after health and demographic characteristics were controlled for, Medicaid/SCHIP coverage was associated with significantly lower per person medical spending.
Following the approach of Hadley and Holahan, we illustrate the effect of Medicaid versus private coverage by simulating two scenarios: (1) what total and out-of-pocket medical spending would be if those covered by Medicaid (or SCHIP) were instead covered by private insurance; and (2) what spending would be if those covered by private insurance were instead covered by Medicaid (or SCHIP). To do this, we used the characteristics of the Medicaid/SCHIP group—their health conditions, age, sex, income, education, race/ethnicity, region of the country, and so on—and applied the coefficients derived for those with private insurance. We also undertook the same type of exercise for those who had private insurance, applying the coefficients found for people covered by Medicaid or SCHIP. Because of the estimation models, findings from simulation models should be compared only to each other, not to the unadjusted values shown in Exhibit 2⇑.
From Medicaid to private insurance.
The results are summarized in Exhibit 3⇓. The models estimate that if an average low-income Medicaid adult were instead covered by private health insurance for a full year, total spending would climb from $5,671 per person per year to $7,126, an increase of 26 percent. The level of out-of-pocket payments would rise from $197 to $1,293, an increase of 556 percent. (Out-of-pocket payments include not only deductibles, copayments, and coinsurance, but also payments for medical services or goods not covered by insurance.)
Predicted Spending For Low-income Adults And Children If They Were Covered By Medicaid Or Private Health Insurance For A Full Year, 2005
If an average low-income child covered by Medicaid or SCHIP were instead covered by private health insurance, total spending would rise from $909 per person per year to $1,247, an increase of 37 percent. The out-of-pocket payments made for this child’s medical care would escalate from $42 to $314, which is 648 percent higher.
From private insurance to Medicaid.
The reverse is also true. If an average low-income adult with full-year private insurance switched to Medicaid, total annual spending would fall from $4,527 to $3,586, and out-of-pocket expenses would decline from $859 to $124. If a privately insured child were covered by Medicaid/SCHIP, annual spending of $1,348 would fall to $995, and out-of-pocket spending would decline from $323 to $40.
Insurance reimbursement versus out-of-pocket spending.
We can view the difference in the estimated total spending and out-of-pocket spending as the amounts paid by insurance. This indicates that the amounts paid by private insurance are 6–8 percent higher than those paid by Medicaid (Exhibit 3⇑). However, most of the difference in the total medical spending under Medicaid or private health insurance is attributable not to the amounts paid by insurers but, rather, to insured people’s out-of-pocket spending.
From the perspective of a family, the differences in out-of-pocket spending can be substantial. Consider a hypothetical family of four, with income at the poverty level ($19,350 in 2005), covered by Medicaid. Assuming spending for two average adults and two average children, aggregate out-of-pocket spending under Medicaid for the family would be less than $500 per year but would be about $3,200 if they had typical private insurance. Shifting this family from Medicaid to private insurance could increase average out-of-pocket medical spending by more than $2,700, or about 13 percent of the family’s income.
From uninsured to Medicaid or private insurance.
Finally, we simulated what would happen if an average low-income uninsured adult or child received coverage under Medicaid/SCHIP or private insurance. We applied the coefficients from the models described earlier to the characteristics of the sample of low-income adults and children who were uninsured for the full year (Exhibit 4⇓).
Predicted Per Person Medical Spending For Uninsured Adults And Children If They Received Medicaid Or Private Coverage Instead Of Being Uninsured, 2005
There are three important results. First, total medical spending would be about 26–30 percent higher if an uninsured person were covered by private insurance than by Medicaid or SCHIP. (For adults, $3,084 for Medicaid versus $3,899 for private insurance; for children, $918 for Medicaid/SCHIP versus $1,194 for private insurance.) The consumer’s out-of-pocket spending would be 600–750 percent higher under private insurance than under public insurance. (For adults, $109 for Medicaid versus $771 for private insurance; for children, $36 for Medicaid/SCHIP versus $305 for private insurance.) Second, because uninsured people have fewer health risks than those currently with Medicaid coverage, they should be less costly to cover than current Medicaid recipients. The predicted total spending for an adult Medicaid recipient was $5,671 per year in 2005 (as seen in Exhibit 3⇑) but would be only $3,084 if an uninsured adult received Medicaid coverage. Third, the uninsured would be able to secure more medical care if they received public coverage, while their out-of-pocket costs would be reduced.
Part-year coverage.
We limited our analyses to those who had Medicaid or private insurance or were uninsured for a full year. Many are uninsured for part of the year, and health insurance coverage expansions could affect them also. We expect that similar patterns exist for those uninsured for part of the year. However, earlier analyses have found that those with gaps in coverage often have higher medical expenses when they first get insurance as they obtain services they lacked while uninsured but moderate their use later.11
Policy Implications
Limitations.
Although our analyses controlled for an extensive battery of health and demographic characteristics, our findings are limited because the data are nonexperimental in nature, and we cannot control for possible unobserved factors that might explain why people “self-select” into Medicaid, private insurance, or uninsured status that may also affect their health spending. In addition, although MEPS is the best national data set for this type of analysis, the complex nature of data collection and the use of self-reported and imputed data may limit its accuracy.
Impact of changed conditions.
These analyses are based on data about low-income people in 2005. Policy or economic changes since then might affect the results. Health spending of every type has grown since then and is expected to continue to climb.12 The Deficit Reduction Act, enacted in early 2006, gave states options to increase Medicaid cost-sharing levels, particularly for recipients with incomes above the poverty level, although only a portion of states are expected to adopt these options, and cost-sharing levels are expected to rise very gradually.13 Data from the Henry J. Kaiser Family Foundation/Health Research and Educational Trust (HRET) Employer Health Benefits Survey indicate that deductibles and co-payments in private health plans are also rising.14
But there is no reason to believe that the basic cost relationships are changing. Our findings are similar to those reported by Hadley and Holahan using 1996–1999 data: Medicaid is a less expensive method of providing health insurance to low-income populations than private insurance is.
Differences from Hadley and Holahan.
Our findings differ from those of Hadley and Holahan in one important respect. They interpreted the large difference in Medicaid and private health insurance spending as being primarily attributable to differences in provider payment rates. We found that the main difference is attributable to higher out-of-pocket spending under private health insurance. Hadley and Holahan did not investigate such differences. We agree that provider payments in Medicaid tend to be lower than in private insurance and that differences in medical care use are modest, after accounting for health status. Modest increases in Medicaid provider payment rates could improve patients’ access to and quality of care. Even if Medicaid payments were increased somewhat, medical spending under Medicaid would likely remain lower than under private coverage.
Impact of large out-of-pocket payments.
Large out-of-pocket payments can have serious repercussions. A number of recent studies have found that problems in the affordability of health care are mounting because of the rising out-of-pocket obligations faced by low- and moderate-income families.15 Even if they can obtain private health insurance, higher cost sharing—especially higher deductibles—makes it harder for many low-income families to afford medical care. A substantial body of research documents that higher cost sharing can make it harder for lower-income people to obtain the medical care or medications they need and can result in poorer health status.16
Limitations of tax subsidies.
Proposals to expand health coverage using tax subsidies for individuals to purchase private health insurance typically focus on using the subsidies to help pay health insurance premiums. This study suggests that even if tax policies could help low-income families pay premiums for basic private insurance, such families might then face substantial out-of-pocket spending for deductibles, copayments, coinsurance, or benefit exclusions that create serious barriers to care. (Some tax policies could affect out-of-pocket payments. For example, tax-deductible health savings accounts, or HSAs, can pay out-of-pocket medical expenses and reduce their effective cost by the user’s marginal income tax rate. However, since low-income people have no or low marginal tax rates, the tax deduction offers little financial benefit to them.)
Our results indicate that efforts to expand health insurance coverage for low-income populations, whether conducted at the national or state level, would be less costly to society and much less costly to financially strapped beneficiaries if the expansions were based on public insurance like Medicaid and SCHIP, which provide coverage for services by private providers and often through private health plans, instead of being based on subsidies for private health insurance.
Footnotes
-
Leighton Ku (Leighton.Ku{at}gwumc.edu) is a professor in the Department of Health Policy, School of Public Health and Health Services, at the George Washington University in Washington, D.C. Matthew Broaddus is a research analyst at the Center on Budget and Policy Priorities, also in Washington.
-
This work was conducted at both the George Washington University and the Center on Budget and Policy Priorities. At the center, the work was supported in part through grants from the David and Lucile Packard Foundation and the Nathan Cummings Foundation. The authors appreciate helpful comments from Bob Greenstein, Sara Rosenbaum, Edwin Park, Tom Selden, Jack Hadley, and anonymous reviewers.
