Introduction and Background

The health care provided in the United States can confer tremendous benefits, extending and improving lives. But the high and rising costs for health care in this country impose an increasing burden on individuals, businesses, states, and the federal government, and a substantial number of people may have trouble paying for that care because they do not have health insurance. Those issues are related: People seek insurance to protect themselves against the risk of experiencing financial hardship when they need expensive medical care, and yet rising costs for health care also make health insurance policies more expensive and thus more difficult to afford.

Lack of insurance can limit access to care, but having insurance can increase spending by encouraging the use of services that have relatively low clinical value. More generally, despite spending more per capita than other countries, the United States lags behind lower-spending countries on several metrics, including life expectancy and infant mortality. Indeed, evidence suggests that a substantial share of spending on health care contributes little if anything to the overall health of the nation, but finding ways to reduce such spending without also affecting services that improve health will be difficult.

Spending on health care services and related activities accounts for about 17 percent of gross domestic product (GDP)—an expected total of about $2.6 trillion in 2009—and under current law that share is projected to reach nearly 20 percent by 2017 (according to the Centers for Medicare and Medicaid Services). Correspondingly, annual health expenditures per capita are projected to rise from about $8,300 to about $13,000 in nominal (current-dollar) terms over that period. Federal spending accounts for about one-third of those totals, and federal outlays for the Medicare and Medicaid programs are projected to reach about $720 billion in 2009.

Over the longer term, rising health care costs represent the single greatest challenge to balancing the federal budget. The Congressional Budget Office (CBO) projects that under current law, federal spending on Medicare and Medicaid alone will grow from about 4 percent of GDP in 2009 to about 12 percent of GDP by 2050.

The number of people who lack health insurance is also on the rise. Recent estimates indicate that about one in four nonelderly residents (those age 64 or younger) is uninsured at some point during the year, and at any given point in 2009 at least 45 million residents are expected to lack insurance—a figure that has grown rather steadily over time and that CBO projects to reach about 54 million by 2019.¹ The nonelderly population will also increase, so the share of that population projected to be uninsured at a given point during the year will grow more slowly, from about 17 percent in 2009 to about 19 percent in 2019. (Those estimates for 2009 do not reflect the recent deterioration in economic conditions, which could result in a larger uninsured population.) The number of uninsured individuals is expected to increase because health insurance premiums are likely to continue rising considerably faster than income, which will make insurance more difficult to afford.

The challenge of seeking to increase the number of people who have health insurance while attempting to constrain the high and rising costs of health care has led to proposals that would substantially modify the health insurance system in this country. Because the Medicare program already provides nearly universal coverage to the elderly, those proposals generally focus on options for providing coverage to and reducing costs for the nonelderly population; even so, options that would reduce Medicare spending could be used to offset the government’s costs for insurance expansions and could have broad effects on spending for health care.

This report describes the assumptions that CBO would use in estimating the effects of key elements of such proposals on federal costs, insurance coverage, and other outcomes; the evidence on which those assumptions are based; and, if the evidence points to a range of possible effects rather than a precise prediction, the factors that would influence where a proposal falls within those ranges. (A companion report—Budget Options, Volume 1: Health Care—describes CBO’s estimates of the effects on the federal budget of numerous specific options related to health care and health insurance.)

This document does not provide a comprehensive analysis of any specific proposal; rather, it identifies and discusses many of the critical factors that would affect estimates of various proposals. In particular, it considers the types of issues that would arise in estimating the budgetary effects of proposals to:

Wherever possible, the analysis presented here describes in quantitative terms how CBO would estimate the budgetary and other effects of such proposals. In other cases, it describes the components that a proposal would have to specify in order to permit estimation of its effects on federal spending or other outcomes. This report reflects the current state of CBO’s analysis of and judgments about the likely response of individuals, employers, insurers, and providers to changes in the health insurance and health care systems. The details of particular policy specifications and the way in which they are combined, as well as new evidence or analysis related to the issues discussed here, could affect future CBO estimates of the effects of large-scale health insurance proposals.

Because such proposals could incorporate a number of the elements that are discussed in this report, they could have interactions that are difficult to predict. Those proposals could also affect both tax revenues and outlays. Estimates of the impact on revenues of proposals to change the federal tax code are prepared by the staff of the Joint Committee on Taxation (JCT) and would be incorporated into any formal CBO estimate of a proposal’s effects on the federal budget. In preparing this report, CBO consulted with the JCT staff on the behavioral considerations that are incorporated into both agencies’ estimates.

CBO’s analysis of whether people obtain insurance focuses on the scope of the coverage that the available policies provide and the cost to them. All else being equal, more comprehensive coverage will be more expensive. The cost of an insurance policy also depends on who enrolls in the plan; how much health care they seek; and how much doctors, hospitals, and other providers of care are paid for their services. Although those considerations are closely related, this report analyzes the following questions:

Proposals to modify the health insurance system that include subsidies would probably have the most immediate and direct impact on the federal budget. Their costs would depend primarily on the nature and extent of those subsidies, the number of people who take advantage of them, and the scope of insurance coverage that is purchased or provided as a result. This report also considers other effects, including any federal administrative costs and challenges that might be involved in implementing a proposal; the effects on eligibility for and spending under other federal programs; the impact of provisions that seek to reduce spending on health care by encouraging consumers to make healthier choices and providers to change some of the ways in which they practice medicine; and other macroeconomic effects or budgetary implications that a proposal might have.

The question of whether and how any net increases in federal spending for health care and health insurance would be financed by policy changes outside the health sector is beyond the scope of this report. Whether a proposal makes health insurance more affordable for a given individual or family would depend not only on its impact on the health insurance premiums that they face but also on the effect that its financing mechanisms have on the household’s budget. To the extent that such proposals are financed by provisions that fall outside the health sector—through increases in tax revenues or reductions in spending for other federal programs—those effects are not addressed in this report.

As background for the discussion of the broad policy options presented in subsequent chapters of this report, the remainder of this chapter describes the primary sources of health insurance coverage, the reasons that people lack coverage, the extent and nature of the coverage that is currently purchased, and the main components and drivers of health care spending.

The primary purpose of health insurance is to protect individuals against the risk of financial hardship when they need expensive medical care. In principle, most people would be willing to pay an insurance premium that was somewhat higher than their own expected costs for health care in order to avoid that risk, but in practice many people with low income or high expected costs might consider the premiums they would face to be unaffordable.

Over the years, various policies have been adopted that subsidize insurance coverage for certain groups. Medicare provides highly subsidized coverage to the elderly and also insures several million people under the age of 65 who are disabled—two groups that have relatively high costs for health care. The Medicaid program and related initiatives offer free or low-priced coverage to many children and (to a more limited degree) their parents; Medicaid also covers many elderly and disabled individuals who have low income and few assets (and thus would have difficulty paying for insurance). Most employers offer health insurance to their workers and most workers enroll in a plan, motivated in part by a tax subsidy for employment-based insurance. People may also be able to purchase coverage in the individual insurance market, but that coverage is not generally subsidized. Those sources of coverage also vary in the ease of enrollment, which affects their attractiveness.

Because health insurance provides more benefits to people who incur relatively high costs for health care, health insurance coverage generally—or specific health insurance plans—may attract enrollees with above-average costs, a phenomenon known as "adverse selection." Conversely, people with low expected costs for health care may be reluctant to pay an insurance premium that reflects the average costs of all enrollees, or they might prefer to wait until they develop a health problem to sign up for coverage. To the extent that such adverse selection occurs, average insurance premiums (or the costs of government subsidies for insurance) would tend to rise to reflect the higher spending per enrollee. The potential for adverse selection exists with almost any health insurance plan, but the manner in which it arises and the mechanisms used to address it differ across insurance markets.

The availability of health insurance affects not only who enrolls but also how much health care people consume. People who are insured are likely to use more health care than they would if they had to pay the full costs of those services—a phenomenon economists call "moral hazard." To offset that tendency toward increased use, health insurance policies typically feature some degree of cost sharing by enrollees. Health plans may also seek to control their costs and premiums by using various methods of managing care and by varying the range of benefits offered. Of course, those features also affect the premiums for health insurance policies and the attractiveness of the coverage to enrollees.

In the United States, most people obtain health insurance coverage from either public or private sources, but about 17 percent of the nonelderly population will be uninsured in 2009 (see Table 1-1).³ Insurance obtained through an individual’s employment is the primary source of coverage for the nonelderly.

Note: The nonelderly population excludes people in institutions and residents of U.S. territories.

Employment-Based Insurance. In 2009, roughly 160 million people under the age of 65—or about three out of every five nonelderly Americans—are expected to have health insurance that is provided through an employer or other job-related arrangement, such as a plan offered through a labor union. That figure includes active workers, spouses and dependents who are covered by family policies, and nonelderly retirees.

One prominent feature of employment-based insurance is that employers generally contribute a large share of the total premium; that is, the amount that is directly and visibly deducted from workers’ paychecks for health insurance (called the employees’ contribution) usually represents a relatively small share of the average cost per enrollee. According to a survey of firms conducted in 2008, employers contribute 73 percent of the cost of a family policy for their workers and 84 percent of the cost of single coverage, on average.4 One reason employers make those contributions is to encourage broad participation by their employees, so as to limit the potential for adverse selection.

Although employers may appear to pay most of the costs of their workers’ health insurance, economists generally agree that workers ultimately bear those costs. Employers’ contributions are simply a form of compensation, and if labor markets are competitive (which is generally the case), an employee’s total compensation should equal his or her contribution to the revenue of the firm. Thus, when an employer offers to pay for health insurance, it pays less in wages and other forms of compensation than it otherwise would, keeping total compensation about the same.⁵

That relationship can be difficult to observe and may not hold perfectly for every worker at every instant. In particular, workers who turned down an employer’s offer of subsidized health insurance generally would not see an immediate or corresponding increase in their wages. Moreover, firms offering health insurance actually tend to pay higher wages than firms that do not do so, but those differences in total compensation reflect disparities in the skill and productivity of the workers, not a failure to pass on the costs of providing insurance. For their part, many employers behave as though they do bear the costs of the insurance plans they offer (as reflected in their efforts to control those costs). Nevertheless, the available evidence indicates that employees as a group ultimately bear the costs of any payments an employer makes for health insurance.⁶

How the costs of employers’ contributions are allocated among different types of workers and how quickly wages would adjust to changes in those contributions is less clear. In principle, workers who would obtain more benefits from health insurance coverage—such as older workers, who have higher average costs for health care—would be willing to accept a greater reduction in their wages than other workers would accept in return for that coverage. The extent to which that phenomenon occurs in practice, however, is uncertain.⁷ Similarly, it could take labor markets several years to adjust to unexpected changes in employers’ costs for health care. For purposes of estimating the impact of proposed legislation, however, CBO makes the simplifying assumption that total compensation is fixed and that changes in the costs of health insurance translate immediately into offsetting changes in wages and other forms of compensation; the JCT staff makes the same assumption when estimating the effects of proposals on revenue collections.

Compared with the individual insurance market, employment-based coverage offers several advantages, particularly for employees of larger firms. Unlike wages, the employer’s costs for providing that coverage are excluded from the enrollee’s taxable income. As a result, that portion of employees’ compensation is not subject to individual income and payroll taxes. In addition, most employees are also able to exclude the portion of the premium that they pay. For a typical worker, that favorable tax treatment provides a subsidy from the government that reduces the net cost of employment-based health insurance by about 30 percent.

That tax subsidy provides an incentive for workers to obtain insurance through their employer and for their employer to provide it. Because out-of-pocket costs for health care do not generally receive a tax subsidy, workers also have an incentive to secure more extensive coverage, thereby increasing the share of spending for health care that is covered and decreasing the share that they pay out of pocket. The value of the exclusion from taxation is generally somewhat larger for workers with higher income because they face higher income tax rates (although they may also face lower rates of payroll taxation).

Employment-based insurance offers a number of other advantages. For example, because sales and marketing costs for insurers are relatively fixed, as the number of enrollees covered by an employer’s policy increases, those fixed costs can be spread over a larger number of enrollees. As a result, the average premium needed to purchase a given amount of coverage is lower for employees of larger firms. Some analysts have suggested that employers also act as employees’ agents, using their power to bargain for lower premiums, sorting out the employees’ options, and making it easier for them to choose an insurance plan.⁸ In particular, employers may take steps that substantially simplify the process of enrolling in a health insurance plan, and the use of automatic payroll deduction to pay for employees’ premiums may also encourage participation.

Another important feature of employment-based insurance is that policies offered by firms of all sizes are subject to certain federal requirements, but most policies offered by larger firms are exempt from state insurance laws and regulations. That distinction stems from the provisions of the Employee Retirement Income Security Act, which are described in Box 1-1. As a result, policies offered by smaller employers generally must comply with requirements that vary by state regarding the benefits they cover, the premiums that insurers may charge, and other terms of purchase. (Those regulations are discussed further in Chapter 4.) Policies provided in the large-group market, by contrast, generally face few legal constraints regarding their benefits and premiums. One exception is that, among workers who are similarly situated (that is, workers who are in the same class of employment and work in the same geographic location), employers may not vary employees’ contributions to premiums on the basis of their health.

Whether employers offer coverage largely reflects the aggregate preferences of their workers, but for several reasons smaller firms are less likely to offer insurance than larger firms. Overall, about half of the workers at very small firms (those that have fewer than 25 employees) are offered coverage and are eligible for it, compared with 77 percent of the workers at firms with 100 to 999 employees and 86 percent of the workers at firms with 1,000 or more employees (see Table 1-2).⁹ One reason is that households with lower income find it more difficult to accept lower wages in return for health insurance, and smaller firms are more likely to employ low-wage workers. Another reason is that policies purchased by smaller firms incur higher administrative costs per enrollee, so the share of the policy premium that covers medical costs is lower, reducing the attractiveness of such policies. Because employees of larger firms constitute most of the total workforce, the percentage of all workers who are offered coverage—about three out of four—is closer to the proportion for larger firms.

The share of workers who are enrolled in employment-based coverage has varied somewhat over time, partly reflecting changes in the mix of employment and partly tracking fluctuations in the business cycle. According to recent surveys of employers, that share rose from 62 percent in 1999 to 65 percent in 2001 but has fallen since then and stands at 60 percent in 2008.¹⁰ The coverage rate has been somewhat more volatile for smaller firms (those with fewer than 200 workers); that rate was 52 percent in 1996, rose to 58 percent in 2001, and fell back to 52 percent in 2008. Studies have attributed the recent decline in enrollment to a combination of modest reductions in the number of employers offering insurance, shifts in employment toward firms and industries that are less likely to offer health insurance coverage, and a reduction in enrollment rates among workers who are offered coverage. The estimated impact of each of those factors varies, however, depending on the specific years examined, the data used, and the methodology employed.

One source of employment-based health insurance that has received considerable attention is the Federal Employees Health Benefits (FEHB) program, which provides coverage to about 8 million active and retired federal employees in 2008. Under that program, several private health insurance plans are available nationwide, and in most regions employees have a range of local plans available to them as well. The federal government covers 75 percent of the cost of each participating plan up to a limit set at 72 percent of the national average premium; to purchase a policy more expensive than that, the enrollee has to pay the added costs (although those payments may also be excluded from taxable income).¹¹ Like employees of private firms that offer a choice of insurance plans, federal workers may generally sign up for coverage or change plans only during an annual open-enrollment season—a rule that limits their opportunities to wait until they develop a health problem to enroll or to switch plans for health reasons and thus limits the degree of adverse selection that can occur.

Although employment-based insurance has certain advantages, the central role of employers in sponsoring coverage also has disadvantages. Unlike federal workers, many employees are not offered a choice of insurance plans, and others may have only a few plans from which to select, so the plan in which they enroll might not fit their preferences. Furthermore, employees and their dependents typically have to change plans when changing jobs and could become uninsured if their new employer does not offer coverage—potentially making them reluctant to switch jobs in the first place (a phenomenon known as "job lock").¹² In addition, employees who become disabled or too sick to keep their job may eventually lose their employment-based coverage.

Individually Purchased Insurance. Overall, CBO estimates that about 10 million nonelderly individuals will be covered by a policy purchased in the individual insurance market in 2009. In principle, anyone may purchase coverage in that market—to cover only themselves or their family as well—but in practice that option may be more attractive to some people than to others. (Such coverage is sometimes called "nongroup" insurance to distinguish it from group coverage, which is primarily employment based.)

The potential for adverse selection may be stronger in the individual market than in the employment-based market, partly because people can apply for individual insurance at any time and may therefore wait until a health problem arises before seeking coverage and partly because applicants do not have to be healthy enough to work. To address those possibilities, insurers usually "underwrite" the policy—a process by which they assess the health risk of applicants. Although most applicants end up being quoted a standard premium rate (which usually varies by age), underwriting can result in adjustments to premiums, adjustments to benefits (for example, to exclude coverage of known health conditions), or denials of coverage. As a result, individuals who have more health problems may face higher premiums when they apply for coverage. Some states, however, prohibit or limit those practices—which generally has the effect of reducing premiums charged to older or less healthy applicants and raising premiums for younger and healthier applicants (as discussed further in Chapter 4).

Individual insurance products have some other advantages and disadvantages compared with employment-based coverage. Some applicants may be able to obtain basic insurance protection (such as "catastrophic coverage" plans) in the individual market at a relatively low cost. That market generally offers consumers a greater choice of plans, and the coverage may be portable from one job to another. Insurers incur greater administrative costs for policies sold in the individual market, however, and those costs are built into the policy premiums. Compared with the enrollment process for an employment-based plan, the effort required of applicants to search for a policy and sign up for coverage in the individual market can be considerably greater. In general, individually purchased coverage does not receive favorable tax treatment, which also makes its effective price higher.¹³

Reflecting those disadvantages, participation in the individual insurance market is relatively low. Only about 1 percent of nonelderly adults who are offered employment-based coverage (either by their own employer or through a spouse) elect to purchase individual coverage. Even among people who lack other coverage options, only about 20 percent elect to purchase a policy in the individual market; the rest are uninsured. In many cases, individually purchased policies are held for relatively short periods of time—serving to cover individuals between jobs, for a short period following college (a point at which children may become ineligible for coverage under their parents’ plan), or between retirement and age 65 (the age of eligibility for Medicare).

Medicare. Medicare provides coverage for about 37 million people who are age 65 or older, and it also covers about 7 million nonelderly people who are disabled (and generally become eligible after a two-year waiting period) or have severe kidney disease.¹⁴ In 2008, about 80 percent of Medicare’s beneficiaries are insured through the traditional fee-for-service program, which pays providers for services directly using prices set administratively; the rest have chosen to receive coverage through private insurers that contract with Medicare to provide program benefits in return for a fixed monthly payment per enrollee (known as the Medicare Advantage option). About 3 percent of people under age 65 are covered by Medicare (see Table 1-1), but their average costs to the program are substantial—more than $35,000 per person in 2007 for those with kidney failure and roughly $8,000 per person for other disabled enrollees.

When it was created, Medicare had two primary components: Part A, which generally covers hospital care and other services provided by institutions; and Part B, which generally covers physicians’ services and various forms of outpatient care. Enrollment in Part A is free of charge and essentially automatic for individuals (and their spouses) who have sufficient earnings subject to payroll taxes to qualify for Social Security benefits; certain others may enroll but must pay a monthly premium. To participate in Part B, enrollees must pay a monthly premium that covers about 25 percent of the program’s average costs. Although participation is voluntary, seniors who choose not to participate in Part B when they are first eligible are subject to penalties if they decide to enroll at a later date—penalties that are intended to discourage eligible individuals from waiting to develop a health problem before they enroll. As a result of those provisions, nearly 95 percent of individuals who are eligible to enroll in Part B do so. Many of those who do not enroll have retiree coverage from a former employer that limits the benefits they would receive from enrolling in Part B (and may also exempt them from the late-enrollment penalty).

A voluntary outpatient prescription drug benefit—known as Part D—was added to Medicare in 2006; its premium subsidy and penalty for late enrollment are similar to Part B’s. About 70 percent of the people who are eligible to participate in Part D have chosen to do so.¹⁵ Analysis by the Centers for Medicare and Medicaid Services (CMS) indicates that a majority of those nonenrollees have drug coverage from another source that is at least as comprehensive as the Medicare benefit, but about 10 percent of the Medicare population appears to lack substantial drug coverage.

Medicaid and the State Children’s Health Insurance Program. Medicaid is the main source of health insurance coverage for Americans who have very low income, and the smaller State Children’s Health Insurance Program (SCHIP) provides coverage for children in families that have somewhat higher income. Unlike the Medicare program, which does not take into account income or assets when determining eligibility and is federally financed, Medicaid and SCHIP are needs-based assistance programs that are jointly financed by the federal government and state governments.

CBO estimates that at any given point in 2009, roughly 64 million nonelderly individuals will be eligible for Medicaid or SCHIP coverage and that about 43 million will be enrolled.¹⁶ Eligibility for Medicaid was originally limited to very low income families with dependent children and to poor elderly or disabled individuals. Over the past two decades, coverage has been extended to children in families with somewhat higher income and to pregnant women. Nonelderly, nondisabled adults who have no children are generally ineligible for the program. Able-bodied parents and children represent about three-fourths of all Medicaid enrollees, but about 70 percent of the program’s spending is for the remaining enrollees who are either elderly or disabled and have low income and few assets.

Subject to broad federal requirements governing eligibility and benefits, the Medicaid program is largely administered by the states, and thus its specific features may vary considerably from state to state. On average, the federal government covers about 57 percent of the costs of the health care services received by enrollees (the share varies among states and is higher for states with relatively low per capita income). State Medicaid programs cover a comprehensive set of services, including hospital care (both inpatient and outpatient), physicians’ services, nursing home care, home health care, and certain additional services for children. States have the authority to cover other services and populations and have used that authority extensively.¹⁷ They may also apply to the federal government for waivers from various federal Medicaid rules.

SCHIP was established in 1997 to provide coverage to children whose family income is above the eligibility levels for Medicaid. States generally cover children in families that have income up to 200 percent of the federal poverty level (or about $44,000 for a family of four in 2009), but some states have higher income limits and some cover parents as well as their children. Like Medicaid, SCHIP is jointly funded by the federal government and the states, but the federal share of costs is higher for SCHIP—covering 70 percent of health care claims, on average. States have a fair amount of discretion in designing and implementing their programs: They may expand Medicaid, create a new state system specifically for SCHIP, or use some combination of the two approaches.¹⁸

SCHIP is currently authorized in law through March 2009. Consistent with statutory guidelines, CBO assumes in its baseline spending projections that federal funding for the program in later years will continue at $5.0 billion, the base amount provided for the first half of fiscal year 2009. In fiscal year 2008, the program’s budget authority was $6 billion and its outlays were about $7 billion. Because average costs per enrollee are expected to rise, CBO projects that average enrollment would decline from a peak of about 5.3 million in 2008 to about 2 million in 2018 under that assumption about future funding. (References to Medicaid in the remainder of this chapter also include SCHIP.)

Other Sources of Coverage. A significant number of people obtain insurance coverage from various other sources including the military, universities (for students), and other organizations. CBO estimates that roughly 12 million people will be covered under such arrangements in 2009. Although military coverage could be considered a form of employment-based insurance, it is typically counted separately. The Department of Veterans Affairs provides some health care to military veterans, but its programs are not considered a comprehensive health insurance plan; similarly, the Indian Health Service provides some care to Native Americans and Alaska natives but is not counted as a source of health insurance (such programs are discussed more extensively in Chapter 6).

About 45 million people, or about 15 percent of the total U.S. population, will be uninsured at any given point in 2009, by CBO’s most recent estimates. Because the elderly have near-universal coverage from Medicare, many analyses of the uninsured focus on the nonelderly population, about 17 percent of which is expected to lack coverage in 2009. Those estimates for 2009 do not reflect the recent deterioration in economic conditions, which could result in a larger uninsured population.

In many cases, people’s insurance status varies over the course of a year. For example, CBO’s analysis of survey data showed that between 57 million and 59 million people—or roughly one-fourth of the nonelderly n population—were uninsured at some point during 1998. The average number of people who were uninsured at a give point in 1998 was smaller—between 39 million and 44 million, of which 21 million to 31 million were uninsured for all of that year.¹⁹ CBO also found that for those who became uninsured at some point between July 1996 and June 1997, nearly half had spells of uninsurance lasting four months or less and about one in six had spells lasting two years or more.

According to CBO’s projections, the average number of people who are uninsured at any one time will rise to about 54 million, or about 19 percent of the nonelderly population, by 2019. The number of uninsured individuals is expected to increase because health insurance premiums are likely to rise considerably faster than income, which will make insurance more difficult to afford.

Characteristics of the Uninsured. The purchase of health insurance in the United States is voluntary, so the main reason that people are uninsured is that they are unwilling or unable to purchase coverage. Several characteristics are associated with insurance status—including income, age, being offered insurance at work, or being eligible for public coverage—but whether they are a causal factor or are merely correlated with coverage rates is not always clear.

Because the costs of health insurance can represent a substantial share of income for lower-income individuals and families who are not eligible for subsidized public coverage, it is not surprising that coverage patterns are strongly correlated with income. In particular, as income rises, the share of nonelderly people who are uninsured or have public coverage declines and the share with private coverage rises (see Figure 1-1). In 2009, the highest rates of uninsurance—about 30 percent—will be found among people whose family income is below 200 percent of the federal poverty level. For people in that group that have insurance, those with family income below the poverty line will be much more likely to have public coverage, whereas those with income above the poverty line will be more likely to have private insurance. Only about 12 percent of people below the poverty line will have private coverage; that rate rises to 40 percent for those between 100 percent and 200 percent of the poverty level. For people whose income is between 200 percent and 400 percent of the poverty level, by contrast, 74 percent have private coverage and 16 percent are uninsured. For people with income above 400 percent of the poverty level, 90 percent have private coverage and 4 percent are uninsured.

Patterns of Health Insurance Coverage for Nonelderly People, by Family Income Relative to the Federal Poverty Level, 2009

Another characteristic that is associated with the lack of health insurance, at least among adults, is age. Younger adults are particularly likely to be uninsured—about 27 percent of those ages 18 to 34 lacked coverage, compared with about 14 percent of those ages 45 to 64 in 2007—possibly reflecting a lower perceived need for using health care services (younger people are generally healthier) as well as lower average income and assets.²⁰ Those younger adults make up about one-fourth of the nonelderly population but represent about 40 percent of the uninsured. Children under the age of 18 account for about the same share of that population but are much less likely to be uninsured.

Not surprisingly, rates of coverage are also associated with whether an individual (or a close family member) is offered insurance at work. In part that correlation probably reflects differences in income—firms with more low-wage workers are less likely to offer coverage—but even within a given income range, workers in relatively small firms (which are less likely to offer coverage) are much more likely to be uninsured than workers in larger firms (see Figure 1-2). For example, among full-time workers whose income is between 100 percent and 200 percent of the federal poverty level, CBO projects that 56 percent of those employed by very small firms (fewer than 25 employees) will be uninsured in 2009, compared with 30 percent for those employed by larger firms (those with 100 or more workers). Determining cause and effect is difficult, however, because workers with less of a desire for insurance or who consider coverage unaffordable would be more likely to join firms that do not offer coverage and pay those workers higher wages instead.

Looking at income levels and insurance options simultaneously may provide additional insights about the uninsured population. For example, CBO projects that among the uninsured in 2009, 17 percent will have family income above 300 percent of the poverty level (about $65,000 for a family of four); 18 percent will be eligible for but not enrolled in Medicaid; and 30 percent will be offered, but will decline, coverage from an employer (see Figure 1-3). Some people will be in more than one of those categories at the same time—so overall, about half of the uninsured will meet at least one of those three criteria. Conversely, the rest of the uninsured are projected to have relatively low income and to lack both an offer of employment-based coverage and eligibility for public coverage.

Note: This analysis categorizes uninsured nonelderly people according to whether they will meet any of the following criteria in 2009: Their family income will be above 300 percent of the federal poverty level; they will have an offer of employment-based insurance (EBI); or they will be eligible for Medicaid or the State Children’s Health Insurance Program (SCHIP). The Congressional Budget Office estimates that a very small number of people will have family income above 300 percent of the federal poverty level and will be eligible for Medicaid or SCHIP.

The reasons people remain uninsured even though they are offered employment-based coverage or are eligible for Medicaid are not always clear. In the case of employment-based coverage, the share of the premium that the employee must pay may be relatively high, or the employee may simply place a low value on having insurance. As for Medicaid, studies indicate a mixture of reasons for failing to enroll. Some people may not be aware that they are eligible; others may be deterred by the application process or see some stigma associated with a program for low-income families. An additional factor is that people who are eligible for Medicaid may be enrolled when they are hospitalized and then may gain retroactive coverage for recent medical expenses; thus, eligibility—even without enrollment—gives them some degree of protection against high medical costs and may reduce the incentive to enroll sooner.

Use of Health Care by the Uninsured. How the uninsured obtain health care affects both their incentives to seek insurance coverage and the impact that policies designed to reduce the number of uninsured have on spending and health. Many of the uninsured receive care from free clinics and other community health centers, which are funded by a combination of federal and state sources and private donations. Others may use traditional health care providers—hospitals as well as physicians in private practice—and pay all charges for the services they receive.

In many cases, however, people who are uninsured receive treatments from traditional providers for which they either do not pay or pay very little, which is known as "uncompensated care." Hospitals that participate in Medicare and offer emergency services are required by law to stabilize any patient who arrives, regardless of whether he or she has insurance or is able to pay for that care. In addition, most hospitals are nonprofit organizations and thus have some obligation to provide care for free or for a minimal charge to members of their community who could not afford it otherwise. For-profit hospitals also provide such charity or reduced-price care.²¹

Estimates of how much uncompensated care the uninsured receive vary depending on the data sources and methods used and the categories of spending that are included in the analysis. Some measures of uncompensated care compare the amount that providers are actually paid for their services with their list prices or posted charges for those services. A more useful comparison, however, is with the total payments that providers would receive for the same service when treating a privately insured patient, because that amount (which is generally much lower than the list price) more closely resembles their costs.

A recent study by Hadley and others, which used that analytic approach, examined a sample of medical claims for uninsured individuals and projected that they would receive about $28 billion in uncompensated care in 2008.²² That study also examined reports by doctors and hospitals and derived a higher estimate: Their gross costs of providing uncompensated care would be about $43 billion in 2008, of which $8 billion would come from doctors and $35 billion would come from hospitals. But as the study noted, at least a portion of those costs could be offset by added payments under Medicare and Medicaid to hospitals that treat a disproportionate share of low-income patients (and by similar dedicated payments made under other federal and state programs). Another recent study found that, as a group, office-based physicians roughly "broke even" when treating uninsured patients because some of those patients paid more than the doctors would have received for treating a privately insured patient.²³ (The issue of whether and to what extent the net costs of providing uncompensated care are shifted to other payers in the health sector is discussed in Chapter 5.)

The uninsured generally use fewer health care services than people who have insurance, although estimates regarding the magnitude of the difference also vary. The study by Hadley and others estimated that an individual who is uninsured for all of 2008 will use about $1,700 worth of care—including about $540 in uncompensated care—or less than half as much as someone who is privately insured all year would use (see Table 1-3). The disparity in the amount spent for care is even larger; subtracting uncompensated care yields an estimate that spending incurred by and on behalf of people who are uninsured for the entire year (about $1,160) is about 30 percent of the amount spent for people who are privately insured all year (about $3,900). Spending by and for those who are insured for part of the year (about $3,000) falls between those two points. According to those estimates, average out-of-pocket payments are similar for each group, although those payments cover a higher share of total spending for the uninsured.

Insurance Status	Out-of-Pocket Spending	Third-Party Payments		Uncompensated Care	Total
Insurance Status	Out-of-Pocket Spending	Insurance	Other^a	Uncompensated Care	Total



	Dollars of Spending

Uninsured for Full Year	583	0	567	536	1,686
Insured for Part of the Year	550	2,030	260	145	2,983
Privately Insured for Full Year	681	3,018	215	0	3,915
Insured for Full Year	654	3,563	246	0	4,463


	Shares of Spending (Percent)
Uninsured for Full Year	35	0	34	32	100
Insured for Part of the Year	18	68	9	5	100
Privately Insured for Full Year	17	77	5	0	100
Insured for Full Year	15	80	6	0	100

Source: Congressional Budget Office based on data from Jack Hadley and others, "Covering the Uninsured in 2008: Current Costs, Sources of Payment, and Incremental Costs," Health Affairs, Web Exclusive (August 25, 2008), pp. W399–W415. The authors used data from the Medical Expenditure Panel Survey, 2002–2004, and adjusted the data to 2008.

Reflecting a range of other findings on that topic, CBO estimates a somewhat smaller disparity in the use of health care services than the study by Hadley and others would indicate.²⁴ According to several other studies and CBO’s own analysis of data for the nonelderly population, the uninsured do use fewer health care services than the insured, but the difference is generally in the range of 30 percent to 50 percent. (See Chapter 3 for a more extensive discussion of those estimates.) Studies comparing the insured and uninsured populations usually account for any differences that are observed in the demographic characteristics and health status of those populations, which would affect their use of health care. Thus, CBO would expect an uninsured person to use 30 percent to 50 percent fewer health care services, on average, than a person who is similar in other respects but has typical private insurance coverage. Among people who have similar demographic characteristics and health status, there are two possible reasons why those who are uninsured would use fewer services than those who are insured: First, some of the uninsured may simply be less inclined to seek health care, resulting in less use of services; and second, the prospect of having to pay the full cost of the services they receive gives them an incentive to use less medical care or less expensive services.

A related consideration is whether the lack of insurance has adverse effects on health. Some studies examining the treatment of serious health conditions have found relatively clear links between insurance coverage and health outcomes.²⁵ For example, uninsured individuals who develop cancer generally have poorer outcomes and die more quickly than cancer patients who have private health insurance. That difference is attributed partly to later diagnosis for the uninsured; broader analyses of the uninsured population have found that they are less likely to receive screening tests, such as mammograms. Similarly, uninsured individuals who have heart disease are less likely to receive expensive treatments for it and also have higher rates of mortality than those who have heart disease but are privately insured.

For more routine care, however, disentangling the effects on health of being uninsured from the impact of other factors that are associated with lack of insurance is more difficult. One recent and comprehensive review of the literature noted that most studies of such effects on health simply compare insured and uninsured individuals and thus do not account for underlying differences between those populations.²⁶ Some studies with a better design have examined the effects of expanding eligibility for public insurance programs and have found specific health benefits for the targeted populations, but broad health improvements stemming from insurance coverage have been difficult to identify. For example, one recent study found that the creation of Medicare had no discernible effect on the mortality rates of the elderly during the first 10 years of the program’s operation.²⁷ Of course, reduced mortality is a relatively crude measure of the benefits conferred by medical care, but the ability to analyze other outcomes, such as quality of life, is constrained because those effects are more difficult to measure.

In addition to differences in the sources of and financing for health insurance and health care, coverage varies by the type of health plan providing it, the scope of services that are covered, and the cost-sharing requirements and limits that apply. That variation largely reflects different approaches to controlling costs for insured individuals and can have substantial effects on the premiums charged for an insurance policy (as discussed in Chapter 3).

Types of Plans. Through the 1980s, private health insurance coverage in the United States typically took the form of an "indemnity" policy, which reimbursed enrollees for their incurred costs, left it to them and their doctors to determine what care to provide, and largely allowed doctors and hospitals to set the prices for those services. As health care costs grew rapidly in the 1980s, however, private insurance coverage began to shift from indemnity policies toward other types of health plans, involving various degrees of managed care (as described below) and negotiated pricing.

One form of managed care plan that emerged was a preferred provider organization (PPO). PPOs establish lists or networks of preferred doctors and hospitals and—to give enrollees an incentive to use those providers—charge more for care received outside the plan’s network. The preferred providers thus gain a higher volume of patients and, in return, usually accept lower negotiated payment rates for each service from the health plan. According to a major survey of employers conducted by the Kaiser Family Foundation, PPOs are the most common type of managed care plan, accounting for about 58 percent of enrollees in employment-based plans in 2008.²⁸ (That survey is the primary source of statistics about coverage and benefits cited in this subsection.)

At the same time, more stringent forms of managed care, such as health maintenance organizations (HMOs), also grew in prominence. Like PPOs, those plans establish networks of providers; unlike PPOs, they offer no coverage for services received outside their networks (except for emergencies). HMOs have also instituted various measures to limit the use of certain services, such as requiring patients to get a referral from a primary care physician in order to see a specialist or to obtain prior authorization from the plan before using some types of specialty care. Some HMOs are fully integrated; the plan owns the hospitals, and doctors work on salary. A more common arrangement, however, is to have a network of independent hospitals and physicians’ practices in which providers either receive a fixed payment per patient (in the case of some primary care physicians) or are paid negotiated rates on a fee-for-service basis. As a share of enrollment in employment-based plans, HMOs peaked at roughly 30 percent in the mid-1990s and then fell, reaching about 20 percent in 2008.

Point-of-service (POS) plans have emerged as a kind of middle ground between PPOs and HMOs. Like PPOs they allow enrollees to go outside a plan’s network for care (albeit at a higher charge), but like HMOs they typically require enrollees to secure referrals for specialty care from a primary care physician within the plan’s network. More common among small firms, they accounted for 12 percent of enrollment in employment-based plans in 2008.

Another design option that has arisen in recent years is a consumer-directed health plan, which combines a high-deductible insurance policy with an account that enrollees can use to finance their out-of-pocket payments on a tax-preferred basis. (In other respects, those plans are usually similar to PPOs.) As of 2008, those plans account for about 8 percent of enrollment in employment-based coverage; one form of consumer-directed plan (known as a health savings account) can also be purchased in the individual insurance market.²⁹

Scope of Covered Services. Both public and private health insurance plans generally cover hospitalizations, visits to doctors and other outpatient care, tests and imaging services (such as X-rays), and prescription drugs. Coverage varies to a greater extent for dental care and vision-related services, particularly when care is discretionary (for example, laser surgery to correct vision problems is typically not covered). According to a 2004 survey of employers, about 20 percent offered vision benefits and two-thirds offered dental benefits (although nearly all firms with more than 500 employees offered dental benefits and about half of those firms offered vision benefits).³⁰ Another source of variation is government requirements to cover certain types of benefits (such as infertility treatments) or the services of specific providers (such as chiropractors), which some states impose and others do not. Those mandates generally affect policies offered in the individual market and by small employers.

Cost-Sharing Requirements. A more significant way in which health insurance plans vary, even among the broad categories of plans noted above, is their cost-sharing structure. Most plans include one or more of the following provisions:

Cost-sharing requirements typically differ by type of plan. According to the 2008 Kaiser/HRET survey of employment-based health insurance plans, almost 20 percent of HMO enrollees face a deductible in 2008, compared with about 68 percent of PPO enrollees. Among PPO enrollees, deductibles for care received within the plan’s provider network average about $560 for single coverage and about $1,300 for family coverage in 2008. For hospital care, some enrollees face separate deductibles, and most (about 69 percent) are subject to coinsurance or copayments.

Most HMO and PPO plans that have a deductible exempt visits to a physician’s office for care received within the network. Enrollees typically have a fixed copayment of around $20 for seeing a primary care physician and around $25 for seeing a specialist physician within their network. For visits outside the network, PPO enrollees who have met the deductible typically pay coinsurance in the range of 30 percent to 35 percent (thus encouraging enrollees to use network providers and also limiting the plan’s liability for those costs). Most people who have employment-based insurance must also pay a portion of the costs for advanced diagnostic tests and outpatient surgery (coinsurance is more common) and for emergency room and urgent care visits (copayments are more common).

Most plans also limit total out-of-pocket spending that enrollees might incur in a given year. For PPO plans, median levels of the out-of-pocket maximum are roughly $2,000 for single coverage and $4,000 for family cover-age in 2008, although those limits vary considerably across plans. Nearly half of HMOs do not have an out-of-pocket limit, but those plans typically have no deductible and relatively low cost sharing for individual services, so enrollees would be unlikely to incur very high out-of-pocket costs in the aggregate.

Many plans vary the amount of coinsurance by the type of service or exempt some services from the general deductible in an attempt to create differing incentives for enrollees to use certain types of care. For example, preventive services may have little or no cost sharing, either because insurers want to encourage their use or because those benefits are attractive to enrollees. Similarly, plans typically exempt prescription drugs from their general deductible and require relatively low copayments for less expensive generic drugs. Conversely, plans that cover dental and vision services may charge a separate deductible for them, require higher rates of cost sharing, or limit the maximum annual benefits that enrollees can receive.

Cost-sharing requirements tend to be higher in the individual insurance market, reflecting not only insurers’ efforts to control the health care spending of their enrollees but also enrollees’ desire for lower premiums (because those policies are generally not subsidized through the tax code). One survey of policies purchased in the individual market in late 2006 and early 2007 found that about 70 percent of single policies had deductibles of more than $1,000 and about two-thirds of family policies had deductibles of more than $2,000.³¹ Largely because they cover a smaller share of enrollees’ health care costs, the premiums for those policies are generally lower than the average premiums observed for employment-based insurance (even though the premiums for individually purchased policies include higher administrative costs per policy).

Cost-sharing requirements in the Medicaid program tend to be much lower than those in employment-based or individually purchased plans—typically $1 to $3 for a doctor’s visit or $2 to $3 for a brand-name drug prescription—reflecting the limited income of Medicaid recipients. Cost-sharing requirements may be more substantial under SCHIP but are generally limited to about 5 percent of enrollees’ family income.

Cost sharing under the Medicare program varies widely by service. In 2009, enrollees will face a deductible of about $135 for physicians’ services and will be charged 20 percent coinsurance beyond that point. Some services, such as lab tests and home health care, are free to the enrollee. Most hospital admissions require a deductible of about $1,070, however, and the effective coinsurance rates for some skilled nursing care and outpatient hospital services may exceed 30 percent. In addition, the program does not cap annual out-of-pocket costs. To limit their financial exposure, most Medicare enrollees have some form of supplemental insurance that covers most or all of their cost-sharing obligations. That supplemental coverage typically comes from a former employer, the Medicaid program, a Medicare Advantage plan, or an individually purchased medigap policy.

Both the amount and rate of growth of spending for health care have important implications for proposals that would seek to expand insurance coverage, reduce that spending, or do both simultaneously. The budgetary impact of subsidizing insurance coverage depends in part on the health care costs that would be covered, and the effects of efforts to control costs depend on how those efforts influence the factors that cause cost growth. Other key aspects of health care spending include its concentration (a relatively small share of individuals account for the bulk of expenditures in any given year), how much an individual’s health care costs vary from year to year, and the substantial variation in average spending that is observed from one region of the country to another.

For 2009, the Centers for Medicare and Medicaid Services projects that national health expenditures will total $2.6 trillion. The bulk of that spending—about five out of every six dollars spent in the health sector—is for personal health expenditures. That category includes such services and supplies as hospital care, physicians’ and clinical services, and prescription drugs, among others. The remaining expenditures are for broad categories of spending that support but do not provide health care, including the administrative costs of private and public insurers; the outlays of public health departments and related activities; and investments in medical research, equipment, and structures.

Data on national health expenditures can be broken down in two basic ways: by the sources of payment and by the types of services provided (see Table 1-4). Private spending will account for about 54 percent of the total in 2009, and public outlays—primarily for the Medicare and Medicaid programs—will account for the remaining 46 percent.³² About 65 percent of private health care costs are covered by insurance, and the rest are paid out of pocket or from other sources (including philanthropy). As for the types of services provided, hospital care and physicians’ services combined will account for about half of all health care expenditures, outpatient prescription drugs for 10 percent, and the administrative costs of public and private insurers for about 7 percent. (Administrative costs borne by doctors, hospitals, and other providers are financed through the payments they receive for their services.) The remainder of personal health care expenditures is primarily for dental and other professional care, home health and nursing home care, and medical equipment.

				Private				Public								Total, NHE
				Insured	Out-of- Pocket	Other^a	Sub- total	Medi- care^b		Medicaid^b				Other^c	Sub- total	Billions of Dollars	Per- centage
				Insured	Out-of- Pocket	Other^a	Sub- total	Medi- care^b		Federal		State		Other^c	Sub- total	Billions of Dollars	Per- centage



				Personal Health Care Expenditures

Hospital Care				291	27	36	355	230		79		61		75	445	800	31

Physicians' and
Clinical Services				264	55	34	353	108		22		16		34	180	533	21

Dental and Other
Professional Care				79	65	3	148	15		6		4		4	29	177	7

Prescription Drugs				113	56	0	170	52		14		10		19	95	264	10

Home Health and Nursing
Home Care				17	43	6	67	51		48		39		6	144	210	8

Medical Equipment and
Other Personal Care				3	51	7	61	10		34		27		14	85	146	6
				___	___	__	____	___		___		___		___	___	____	__

	Subtotal, Personal
	Health Care			767	298	87	1,152	467		203		156		152	978	2,131	83


				Other Expenditures
Administration and Net Cost of Private Insurance				112	0	2	113	28	^d	16	^d	13	^d	14	71	184	7

Public Health Activity				0	0	0	0	0		0		0		72	72	72	3

Research, Equipment, and Structures				0	0	104	104	0		0		0		65	65	169	7
				___	__	___	___	__		__		__		___	___	___	__

	Subtotal, Other			112	0	105	217	28		16		13		151	207	424	17

		Total, NHE
			Billions of dollars	879	298	193	1,369	495		219		169		303	1,186	2,555	100
			Percentage	34	12	8	54	19		9		7		12	46	100

Compared with other developed countries, the United States devotes a substantially larger share of its economy to health care and related expenditures. That share was about 16 percent of gross domestic product in 2006—up from about 8 percent in 1975. Under current law, that share is projected to reach nearly 20 percent by 2017 (the last year of the current CMS projections). By contrast, spending for health care among the other countries that belong to the Organisation for Economic Co-operation and Development (OECD) averaged about 9 percent of GDP in 2006.³³

Comparisons of growth rates for health care spending across countries can be sensitive to the time period used and to other factors included in the analysis (such as age, average income, or overall rates of economic growth). Some comparisons indicate that real (inflation-adjusted) growth rates have been similar across developed countries, which might suggest that common forces are causing spending to rise despite substantial differences in their health care systems; other studies conclude, however, that the United States has experienced faster growth in the share of GDP spent on health care than have other, comparable nations.³⁴

Within the United States, growth rates in health care spending have varied over time but have generally outpaced those in the overall economy. An exception was the period between 1993 and 2000, when the share of GDP spent on health care held nearly constant at about 14 percent, but spending growth has accelerated since then. Over extended periods, the annual growth rate of health care spending per capita in the United States has typically exceeded the growth rate of GDP per capita by 2 percentage points or more, accounting for the substantially larger share of the economy that spending for health care now represents. CBO projects that as the share of family and state budgets devoted to health care grows even larger, growth in health care expenditures will eventually moderate even in the absence of changes in federal law. By CBO’s estimates, spending per capita will nevertheless continue to grow more quickly than the economy as a whole—about 1.7 percentage points faster, on average—and total health care spending will reach nearly 40 percent of GDP by 2050.³⁵

Over the past 30 years, federal spending on Medicare and Medicaid has nearly tripled as a share of GDP, rising from about 1.5 percent in 1975 to about 4.0 percent in 2007. According to CBO’s projections, such spending will reach about 12 percent of GDP by 2050 under current policies, but substantial uncertainty surrounds that estimate. If spending per enrollee continued growing over the next four decades as quickly as it has over the past four—about 2.5 percentage points faster than per capita GDP—then federal spending on those programs would reach about 17 percent of the economy. If, instead, spending per enrollee grew at the same rate as GDP per capita, demographic changes alone would push those federal expenditures to about 6 percent of GDP in 2050.

As those figures suggest, the rate at which health care spending grows relative to the economy is the most important determinant of the country’s long-term fiscal balance; it exerts a significantly larger influence on the budget over the long term than other commonly cited factors, such as the coming retirement of the baby-boom generation.³⁶ Rising health care spending represents a challenge not only for the federal government but also for private payers. Indeed, trends in both sectors reflect many of the same underlying forces, so controlling federal outlays over the long term will be difficult without addressing the forces that are also causing private spending for health care to rise.

The effects of proposals to reduce spending on health care depend in part on how they affect the factors that are driving the growth of that spending. The factor with the greatest impact on spending growth is probably the development and diffusion of medical technology (broadly defined). Other influences include the aging of the population; reductions in the share of costs paid out of pocket; growth in the relative prices of health care services; and the growing prevalence of chronic health problems. (A recent CBO report analyzed several of those factors and provides additional information about the studies used to estimate their effects.³⁷) In addition, the manner in which health care services are financed probably has an effect on the amount of spending and could also affect its growth rate.

Advances in Medical Technology. Many analysts attribute the bulk of the growth in health care spending to the development and diffusion of new medical "technology"—a term that is defined broadly to include new procedures and treatments as well as new medicines and devices. Some breakthrough developments permit the treatment of previously untreatable conditions; such innovations can confer substantial benefits, but they also add new sources of spending. Other advances may simply improve medical outcomes (compared with those provided by older treatments) but at added costs. In some cases, however, new procedures and treatments—or broader application of existing ones to new types of patients—could add to spending without yielding better outcomes. Whatever the magnitude of the health benefits may be, studies indicate that about half of the growth in health care spending over the past several decades reflects changes in medical care made possible by the development of new treatments and procedures.

Improvements in medical technology do not have to increase costs; technological innovation could reduce the unit cost of treating a given health problem and could also reduce total spending. Under current arrangements, however, the nature of technological advances in medicine and the changes in clinical practice that have ensued in the United States have tended to raise total spending—because the treatments themselves are expensive, because the number of people receiving them grows rapidly, or for both reasons. In the aggregate, that tendency may well reflect the willingness of individuals to pay for the added spending through higher insurance premiums; as some observers have noted, the demand for health plans offering "1960s medicine at 1960s prices" appears to be low.³⁸ Decisions about whether to cover new technologies, however, are ultimately made by public and private insurers, and the benefits and costs of those technologies may not be carefully evaluated in each case, in part because the information needed to do so is lacking in many situations.³⁹

In assessing the role of medical technology, analysts also considered other sources of past spending growth (including increases in income and rising administrative costs for insurers, as well as those listed above). Yet each of those factors individually has accounted for a relatively small share of that growth, and collectively they can account for about half of total spending growth—even using estimated effects toward the upper end of the range for each factor. Analysts have thus attributed the large residual effect to technology because it is the one remaining force that could be responsible for cost growth (and because the effects of technology on spending are hard to measure directly).⁴⁰ Even if that conclusion is correct, it still leaves open the question of what underlying forces are causing technological changes to be adopted or why those changes tend to yield net increases in spending.

Aging of the Population. One noteworthy finding from studies that have analyzed past spending growth is that the impact of aging has been relatively small. The elderly do use more health care than the nonelderly, and the share of the population that is elderly increased by about 30 percent between 1965 and 2005—from 9.5 percent to 12.4 percent. By itself, however, that change would cause total spending on health care to rise by about 16 percent and thus accounts for only about 3 percent of the total cost growth over that period. (After adjusting for general price inflation using the GDP implicit price deflator, per capita spending on health care grew by more than 500 percent between 1965 and 2005.)

Aging has had a larger effect on federal spending for health care, however, primarily because nearly all residents become eligible for Medicare once they turn 65. In particular, the impending eligibility of the baby-boom generation will have a substantial effect on the share of GDP devoted to Medicare as a result of the increase in enrollment, but that effect pales in comparison with the likely impact of continued increases in health care spending per enrollee. According to CBO’s analysis, future demographic changes will account for somewhere between one-fifth and one-third of the increase in federal spending on Medicare and Medicaid over the next 25 to 75 years, and rising outlays per enrollee (over and above demographic effects) will account for the remainder.⁴¹

Reductions in the Share of Costs Paid for Out of Pocket. Another important factor that both reflects and has contributed to rising health care expenditures is the declining proportion of those costs that are paid out of pocket—and the corresponding increase in the share covered by insurance. According to the estimates of national health expenditures produced by CMS, out-of-pocket payments accounted for 33 percent of all personal health care expenditures in 1975; by 2000, that share had fallen to 17 percent, and it declined to 15 percent in 2006.

Reducing the share of costs that patients have to pay generally increases their demand for care, and studies have concluded that more extensive insurance coverage is responsible for about 10 percent of historical spending growth. But that estimate does not account for the effect that expanded insurance coverage has on the diffusion of medical technology. By contrast, a recent study that examined the effects of Medicare’s introduction found that a broad expansion of insurance coverage had much larger effects on spending. It attributed part of the impact to more rapid and widespread adoption of existing treatment methods (such as those provided by cardiac intensive care units), although some questions remain about the precise magnitude of those effects.⁴²

Out-of-pocket payments have continued to decline slightly as a share of health care spending in recent years, despite recent increases in cost-sharing levels. For example, the average deductible for single coverage in an employment-based PPO plan tripled between 2000 and 2008 (rising from $187 to $560).⁴³ However, total out-of-pocket payments have not increased as quickly, and spending covered by insurance has also risen substantially, roughly keeping pace with the increases in out-of-pocket costs. Indeed, the overall rise in health care spending and the decline in the share paid out of pocket have had roughly offsetting effects on the share of GDP accounted for by out-of-pocket costs, which has held steady over the past three decades at about 2 percent. Even so, such increases in cost-sharing requirements have raised concerns that some people who have insurance coverage may be underinsured. For example, a recent study estimated that about 25 million insured adults faced relatively high out-of-pocket costs (as a share of their income) in 2007, up from about 16 million in 2003.⁴⁴

Financing of Health Care Services. The way in which health care services are financed also affects the amount of health care spending and could affect its growth rate as well. With the exception of some HMOs, most health care provided by doctors in the United States is currently paid for on a fee-for-service basis. In some cases (such as hospital stays under Medicare), a fixed payment is made for a bundle of related services. Such payments encourage doctors and hospitals to deliver a given service or bundle efficiently, but they can also create an incentive to provide additional services or more expensive bundles if the payments exceed the costs of providing care.

Fee-for-service payments may yield a higher quantity or a greater intensity of services at any given time, but whether that type of payment contributes to the rate of spending growth is less clear. Because that method of financing has been in place for many years, it could have affected the amount of spending in a constant way without changing the growth rate of spending. Consistent with that view, an older study found that growth rates of spending in HMOs and fee-for-service plans did not differ substantially.⁴⁵ Compared with other payment systems, fee-for-service payment could encourage or at least facilitate the adoption of newer, more costly services, but whether that happens depends on how quickly fees are established for new treatments and on the level at which those fees are set. (See Chapter 5 for additional discussion of fee-setting mechanisms.)

Growth in the Relative Prices of Health Care Services. Growth in payment rates that exceeds general price inflation has probably contributed to the increase in the share of GDP devoted to health care. Between 1975 and 2005, the increase in the medical component of the consumer price index was nearly twice as large as the increase in prices overall—which might suggest that price increases for health care have played a large role in cost growth.

Measuring price inflation in the health sector can be difficult, however, both because it is hard to control for changes over time in the quality or type of the products being compared (which can make historical price comparisons misleading) and because discounts negotiated by private insurers are typically confidential. Such problems can arise with any price index but may be particularly acute for health care because of the relatively large role played by technological advances and because the prevalence of insurance obscures the price of many transactions. Despite those challenges, some observers have suggested that prices for health care, when properly measured, have actually grown at rates comparable with or lower than general inflation and that prices have not played a substantial role in the growth of U.S. health care spending over time. But other analyses (which are also cited in CBO’s January 2008 paper on the growth of health care spending) suggest that rising relative prices for medical care may have accounted for as much as 10 percent to 20 percent of past spending growth.

Whether or not they have contributed to the growth in spending, price levels affect total spending, so the methods used to set those levels can also play an important role. In some cases, private insurers may have difficulty negotiating low prices for health care items and services, whereas public purchasers have sometimes intervened to obtain relatively low prices. Limited competition among doctors and hospitals in some parts of the country hampers the ability of private insurers to negotiate lower payment rates for their services. In the case of prescription drugs, public policy (through patents) gives manufacturers monopoly power, which leads to higher prices when drugs are introduced but also encourages those drugs to be developed in the first place. Federal and state purchasers have established payment systems that yield lower prices for drugs (under Medicaid and the health program for military veterans) and for doctors and hospitals (under Medicare and Medicaid), although many doctors are unwilling to accept Medicaid’s payment rates. (See Chapter 5 for additional discussion.)

Rising Prevalence of Health Problems. Spending on health care would also be expected to grow if Americans were developing more health problems or were becoming more likely to contract diseases, but the evidence on those points is mixed. Perhaps the most alarming trend has been the growth in obesity over the past several decades. According to one set of surveys, the share of the adult population that is obese grew from about 23 percent in 1988 to about 34 percent in 2004, and the share that is either obese or overweight increased from 56 percent to 67 percent over that period.⁴⁶ CBO’s analysis indicates that the share of spending growth attributable to rising weight over a similar period is between 4 percent and 12 percent, depending on the methodology used.⁴⁷

More generally, determining whether spending on health care is rising because Americans are getting sicker is difficult. Trends in the incidence and prevalence of chronic and acute health problems have varied—some rates have increased, some have decreased, and some have held steady. For example, cancer is a leading cause of death and a major source of health care spending, but the incidence of cancers has declined slightly since 1990. In other cases, the analysis is complicated by the fact that reported rates of disease prevalence may rise when new treatments for the disease become available. Moreover, increases in the intensity of treatment may also increase the likelihood of diagnosing a disease (even if the true prevalence of the disease has not changed).

For example, obesity is associated with many serious medical conditions, including diabetes, heart disease, and high blood pressure. According to one government survey, the share of adults with diabetes grew by about 2 percentage points between 1988 and 2004, from about 8 percent to about 10 percent.⁴⁸ The share of the population being treated for diabetes grew even faster—by more than 50 percent among those with private insurance, according to one study—partly because the probability that someone with diabetes would be diagnosed also increased by about 10 percent.⁴⁹ Over that same period, however, the fraction of adults who have high blood pressure held constant at about 18 percent (in part because an increasing share of patients were taking medications to lower their blood pressure). Meanwhile, the percentage of adults with high cholesterol has fallen steadily and is now about half what it was in the early 1960s, partly because of the development and use of cholesterol-lowering drugs. Overall, it is not clear what role changes in the prevalence of disease—as opposed to increases in the rate at which existing diseases are diagnosed and in the intensity of their treatment—are playing in the growth of health care spending.

individuals’ spending and the substantial geographic variation in average spending levels across the United States. In any given year, the vast majority of spending on health care is generated by the relatively small share of individuals who use extensive or expensive services. Furthermore, people with high health care costs in one year tend to have above-average costs the next year; below-average costs for health care are also likely to persist.

In addition to individual variation, average health care spending varies sharply from one region of the country to another in ways that are not explained by regional differences in age or measures of sickness and that do not appear to yield better health overall in the high-spending regions. Reducing those differences in spending without harming health would improve the efficiency of the health sector, but steps to achieve that goal would undoubtedly prove quite challenging and complex to implement.

Concentration and Persistence of Individuals’ Health Care Spending. The concentration of annual health care spending among a relatively small share of the population has been well documented, both among the nonelderly and in the Medicare program. For example, CBO analyzed spending by nonelderly individuals who had health insurance and found that 13 percent of them used more than $5,000 worth of care in 2004. That high-spending subgroup (with average costs of about $15,000) accounted for about 68 percent of the health care costs for that population. If the threshold is lowered to $2,000 worth of care, the share of nonelderly insured people with higher spending increases to 30 percent, and the share of health care spending attributable to those individuals rises to 86 percent. By contrast, about 55 percent of that population used less than $1,000 worth of care in 2004, and their collective spending amounted to only 6 percent of the total (with average spending of about $300). Among the Medicare population, similar degrees of concentration are observed. In 2001, the most expensive 5 percent of enrollees accounted for about 43 percent of program spending in one year, and the top 25 percent accounted for 85 percent of spending.⁵⁰

Protecting themselves against the relatively low risk of incurring substantial costs is the main reason most people seek health insurance, and the uncertainty about those costs is large enough that most people who can afford to purchase insurance do so. Even so, a predictable element of health care spending also affects the type and extent of insurance coverage that people seek. CBO’s analysis found that nonelderly insured individuals who used less than $1,000 worth of care in 2003 had a 78 percent chance of using less than $1,000 worth of care in 2004 but only a 5 percent chance of using more than $5,000; their average costs in 2004 were about $1,200 (see Table 1-5). By contrast, individuals who used more than $5,000 worth of care in 2003 had a 37 percent chance of using more than $5,000 worth of care in 2004 and only a 24 percent chance of using less than $1,000 worth; their average costs in 2004 were about $7,800. In some cases,

those correlations are an artifact of the calendar year and simply reflect treatments begun in late 2003 that continued into early 2004. But in other cases, those raw year-to-year correlations may understate the extent to which individuals can anticipate their likely needs for health care in the near future; even if they have similar spending initially, people who have had health problems that are unlikely to recur and those who have conditions that are more chronically costly can use that information in choosing a health insurance plan.

		Percentage of 2003 Subgroup with Total 2004 Spending in Range				Spending in 2004 (Dollars)
Total 2003 Spending Range (Dollars)	Percentage of Population in Range	0 to $1,000	$1,000 to $5,000	$5,000 or More	All Ranges	Mean	Median



Zero to 1,000	50	78	17	5	100	1,214	279

1,000 to 5,000	35	41	47	13	100	2,597	1,313

5,000 or More	15	24	39	37	100	7,765	3,316

Source: Congressional Budget Office based on merged data for 2003 and 2004 from the Medical Expenditure Panel Survey conducted by the Department of Health and Human Services, Agency for Healthcare Research and Quality.

Notes: The figures include only individuals who were under the age of 65 in 2003 and were privately insured for all of that year.

CBO increased total health care spending for 2003 to 2004 dollars by using the growth in health care spending per capita as estimated from the national health expenditures compiled by the Department of Health and Human Services.

As more time passes, the difference in average spending between those who initially had high expenditures and those who initially had low expenditures would tend to decline further, reflecting a common statistical phenomenon known as "regression to the mean." People with very high initial spending were probably hospitalized, which is unlikely to happen year after year (even though they may have an above-average chance of being hospitalized again). Conversely, some of those who had low initial spending may develop a chronic or acute health problem that generates higher costs. As a result, when examined over longer periods of time, health spending appears to be less concentrated. For example, looking at Medicare spending over a five-year period (from 1997 to 2001), CBO found that the most expensive 5 percent of

Medicare beneficiaries accounted for 27 percent of total Medicare spending (compared with 43 percent in 2001 alone) and that the top 25 percent of beneficiaries accounted for 68 percent of total five-year spending (compared with 85 percent in 2001). Analysis of younger populations has been limited by lack of data, but one study simulated expenditure patterns for workers from age 25 to age 60 and suggested that the most expensive 25 percent of employees would account for roughly half of expenditures over that 35-year period.⁵¹

Geographic Variation in Health Care Spending. Spending on health care varies not only from person to person (because of differences in their health and in the treatments they receive) but also from region to region in the United States. In particular, per capita health care spending varies widely within the Medicare program, and yet that variation is not correlated with measures of the quality of care that beneficiaries receive or with available metrics of overall health outcomes. In 2004, for example, Medicare spending per beneficiary ranged from about $5,600 in South Dakota to about $8,700 in Louisiana. Yet a comparison of composite quality scores for medical centers and average Medicare spending per beneficiary shows that facilities in states with high average spending are no more likely to provide recommended care for some common health problems than are facilities in states with lower spending (see Figure 1-4). Health care spending per capita also varies widely when examined for the entire population—ranging from roughly $4,000 in Utah to $6,700 in Massachusetts in 2004—but the connection between that variation and health outcomes has not been examined as closely. Medicaid spending per enrollee also varies considerably among states (partly reflecting differences in the population covered and benefits provided).

Source: Congressional Budget Office based on data from the Centers for Medicare and Medicaid Services and from the Department of Health and Human Services, Agency for Healthcare Research and Quality, National Healthcare Quality Report, 2005 (December 2005), Data Tables Appendix, www.ahrq.gov/qual/nhqr05/index.html.

Notes: The composite measure of the quality of care, based on Medicare beneficiaries in the fee-for-service program who were hospitalized in 2004, conveys the percentage who received recommended care for myocardial infarction, heart failure, or pneumonia.

The observed variations in Medicare spending per enrollee are even greater when examined by the area in which enrollees generally receive their hospital care, but a link between higher spending and better health in that population is still hard to discern. In 2005, average costs ranged from about $5,200 in the regions with the lowest spending to nearly $14,000 in those with the highest spending (those averages were adjusted to account for differences in the age, sex, and race of Medicare beneficiaries in the various regions). According to one study, higher-spending regions did not have lower mortality rates than lower-spending regions, even after adjustments were made to control for different rates of illness among patients and across regions.⁵² That study also found that higher spending did not slow the rate at which the elderly developed functional limitations (a measure of their difficulties in taking care of themselves).

Other studies of spending variation reach somewhat different conclusions, even though they also suggest opportunities to improve the efficiency of the health sector. Some research suggests that health overall might not suffer if medical practice in higher-spending regions changed to match that of lower-spending regions. Patients who would benefit most from more expensive treatments, however, might be made worse off as a result, whereas patients who would do better with less expensive treatments would gain. Other, older studies of geographic variation indicate that there may be room to reduce spending without harming health in both high-use and low-use regions of the country, because a large share of certain surgeries were performed in both types of regions even though they were found to be clinically inappropriate or of equivocal value.

What factors contribute to geographic variation? Some of the differences in spending reflect varying rates of illness as well as differences in the prices that Medicare pays for the same service (those prices are adjusted on the basis of local costs for labor and equipment in the health sector). But according to researchers at Dartmouth, differences in illness rates account for less than 30 percent of the variation in spending among areas, and differences in prices can explain another 10 percent—indicating that more than 60 percent of the variation is due to other factors.⁵³ Differences in income or the preferences of individuals for specific types of care appear to explain little of the variation in spending. Unmeasured differences in the demand for care could be important, but some of the variation in medical practice probably is attributable to regional differences in the supply of medical resources (specialist physicians or health care facilities, for example) and the propensity to take advantage of the financial incentives provided by Medicare or other payers in developing and using those resources. Overall, patterns of treatment in high-spending areas tend to be more intensive than those in low-spending areas; that is, in high-spending areas, a broader array of patients will receive more costly treatments.⁵⁴

In sum, the evidence about variation in spending suggests that efficiency gains in the health care system are possible: Expenditures in high-spending regions could probably be lowered without producing worse outcomes, on average, or reducing the overall quality of care. But if policies that reduced expenditures in high-spending areas did not successfully target ineffective or harmful treatments—a challenging task—they might not lead to increased efficiency and could result in worse health outcomes.

Those figures are CBO estimates and differ somewhat from the estimates of the uninsured population released annually by the Census Bureau (the most recent of which indicated that about 46 million people were uninsured at any given point in 2007). Like other estimates, CBO’s figures represent the civilian, noninstitutionalized population; they therefore include unauthorized immigrants but exclude residents of nursing homes, people who are incarcerated, and U.S. citizens living abroad. See Chapter 6 for further discussion of such populations.

Even if a given labor market was not competitive, firms operating in that market would still be expected to hold total compensation fixed, so that other forms of compensation would be reduced to offset the costs of providing health insurance. The allocation of compensation among wages, health insurance, and other fringe benefits would reflect the preferences of workers and the firms’ efforts to attract employees.

One study examined the impact of a state mandate to cover maternity benefits and found that reductions in the wages of women of child-bearing age and their spouses roughly offset the average costs of providing those benefits. See Jonathan Gruber, "The Incidence of Mandated Maternity Benefits," American Economic Review, vol. 84, no. 3 (June 1994), pp. 622–641.

Jeff Liebman and Richard Zeckhauser, Simple Humans, Complex Insurance, Subtle Subsidies, Working Paper No. 14330 (Cambridge, Mass.: National Bureau of Economic Research, September 2008).

Among firms that have similar numbers of workers, the share of firms reporting that they offer coverage to their employees is generally larger than the share of employees reporting that they have an offer, but that discrepancy simply reflects the fact that some workers at firms that offer coverage are not eligible to enroll in it. For example, many part-time workers are ineligible.

Kaiser/HRET, Employer Health Benefits: 2008 Annual Survey; and Employer Health Benefits: 1999 Annual Survey (October 1999).

Workers who previously held employment-based insurance may seek coverage in the individual insurance market, and insurers must generally offer them a policy if they apply, but some workers may find the terms of that coverage unattractive. See Chapter 4 for additional discussion.

That figure represents average enrollment and excludes nonelderly individuals living in institutions (such as nursing homes) and people living in U.S. territories. CBO has also projected that the total number of individuals enrolled in Medicaid at any point during 2009 (including elderly and institutionalized enrollees and residents of territories) will be 65 million, of which about 59 million will be nonelderly. Many of those individuals will be enrolled in the program for only part of the year.

According to one estimate, total spending on optional populations and benefits accounted for about 60 percent of the program’s expenditures in 2001. Of that total, 30 percent was spent to provide optional benefits to mandatory groups; 50 percent, to provide mandatory benefits to optional groups; and 20 percent, to provide optional benefits to optional groups. See Kaiser Commission on Medicaid and the Uninsured, Medicaid Enrollment and Spending by "Mandatory" and "Optional" Eligibility and Benefit Categories (Washington, D.C.: Henry J. Kaiser Family Foundation, June 2005), p. 11.

U.S. Census Bureau, Income, Poverty, and Health Insurance Coverage in the United States: 2007, P60-235 (August 2008).

Jack Hadley and others, "Covering the Uninsured in 2008: Current Costs, Sources of Payment, and Incremental Costs," Health Affairs, Web Exclusive (August 25, 2008), pp. W399–W415. That study also reported that uncompensated care would total about $56 billion in 2008 if all costs not paid out of pocket by the uninsured were included in the tally. But that amount would seem to be an overestimate because the study found that, even though no payments were made by insurers, about half of those costs were directly compensated by various third parties (such as workers’ compensation programs).

Jonathan Gruber and David Rodriguez, How Much Uncompensated Care Do Doctors Provide? Working Paper No. 13585 (Cambridge, Mass.: National Bureau of Economic Research, November 2007).

If the study by Hadley and others underestimated the number of services used by uninsured individuals, its estimate of uncompensated care could also be correspondingly low. (That factor could account for the higher estimate of uncompensated care that study derived using reports by doctors and hospitals.) If, instead, the study overestimated the number of services used by insured individuals, that would not necessarily affect the estimate of uncompensated care.

Helen Levy and David Meltzer, "The Impact of Health Insurance on Health," Annual Review of Public Health, vol. 29 (April 2008), pp. 399–409. One study that sheds some light on the impact of health insurance on health is the RAND Health Insurance Experiment, which randomly assigned large groups of nonelderly individuals to different health insurance plans and tracked their experience over several years. In general, the study found that participants who faced cost sharing did not have worse health than those who got all of their care for free; one exception was lower-income participants with prior health problems, who did not control their blood pressure as effectively when they faced cost sharing. An important limitation of the study, however, is that no participants lacked insurance. For additional discussion of those findings, see Congressional Budget Office, Consumer-Directed Health Plans: Potential Effects on Health Care Spending and Outcomes (December 2006), pp. 54–55.

AHIP Center for Policy Research, Individual Health Insurance 2006–2007: A Comprehensive Survey of Premiums, Availability, and Benefits (Washington, D.C.: America’s Health Insurance Plans, December 2007).

If the cost of the tax expenditure from excluding premiums for employment-based insurance (estimated to be roughly $250 billion at the federal level in 2007) was included in public spending rather than private spending, then the public share of spending would be about 57 percent. See Chapter 2 for additional discussion.

Organisation for Economic Co-operation and Development, OECD Health Data 2008, www.oecd.org. Adjusting for differences in income would reduce that disparity somewhat because income in the United States is higher than the OECD average and higher income is correlated with higher spending on health care.

Congressional Budget Office, The Long-Term Outlook for Health Care Spending (November 2007). The figure for per capita cost growth reflects the projected rate of growth after accounting for the aging of the population, referred to as "excess" cost growth. Those projections assume that no changes are made in federal policies.

A more precise description would label the residual as the effect of changes in medical technology that are not attributed to other observed forces. For example, increases in income can account for some growth in health care spending, and the mechanism through which that growth occurs might also be the greater use of medical technology.

Kaiser/HRET, Employer Health Benefits: 2008 Annual Survey; and Employer Health Benefits: 2000 Annual Survey (September 2000).

See Cathy Schoen and others, "How Many Are Underinsured? Trends Among U.S. Adults, 2003 and 2007," Health Affairs, Web Exclusive (June 10, 2008), pp. W298–W309. The study relied on self-reported income and out-of-pocket health care costs of survey respondents; it defined individuals as underinsured if their health plan’s deductible exceeded 5 percent of their income or if their out-of-pocket costs exceeded either 10 percent of their income (for those with family income above 200 percent of the poverty level) or 5 percent of their income (for those with family income below 200 percent of the poverty level).

National Center for Health Statistics, Health, United States, 2007, DHHS Publication No. 2007-1232 (November 2007). Recent data on obesity rates suggest that those rates may have leveled off, but it is probably too early to tell whether that development is temporary or is likely to endure.

National Center for Health Statistics, Health, United States, 2007. About 3 percent of the population was estimated to have undiagnosed diabetes in both years (which was determined by conducting medical tests on survey participants).

Matthew J. Eichner, Mark B. McClellan, and David A. Wise, Insurance or Self-Insurance? Variation, Persistence, and Individual Health Accounts, Working Paper No. 5640 (Cambridge, Mass.: National Bureau of Economic Research, June 1996).

See John E. Wennberg, Elliott S. Fisher, and Jonathan S. Skinner, "Geography and the Debate Over Medicare Reform," Health Affairs, Web Exclusive (February 13, 2002), pp. W96–W114; and Center for the Evaluative Clinical Sciences, Dartmouth Medical School, The Dartmouth Atlas of Health Care 1999 (Lebanon, N.H.: Health Forum, Inc., 1999), pp. 22–23.

	Total Employees		Employees Offered Health Insurance
Size of Firm (Number of employees)	Number (Millions)	Percent	Number (Millions)	Percent


Fewer than 25	31.0	22	14.9	48
25 to 99	17.6	13	12.7	72
100 to 999	27.2	19	21.0	77
1,000 or More	63.9	46	54.9	86

All	139.7	100	103.5	74

	Number (Millions)	Percent



	Source of Coverage

Employment-Based^a	160	61
Individually Purchased	10	4
Medicare	7	3
Medicaid^b	43	17
Other^c	12	4


	Insurance Status

Insured, Any Source^d	216	83
Uninsured	45	17