Chapter
3
Factors Affecting Insurance Premiums
The effect of proposals to increase coverage would depend in part on the premiums charged and the value of the coverage provided. In particular, the costs of a subsidy that covers a specified percentage of policy premiums would be affected by the amount of those premiums, whereas the impact of a fixed-dollar subsidy on coverage rates would depend on the share of the premiums it covers. Thus, the factors that determine premiums also affect the impact that a proposal has on insurance coverage and the federal budget.
In general, the premium charged for a private health insurance policy is equal to the sum of two components: the average amount that an insurer expects to pay for services covered under the plan; and a loading factor that reflects the insurer’s costs of operating the plan (including administrative expenses and a return on investment). An insurer’s costs for covered services in turn reflect the scope of benefits that are included, the plan’s cost-sharing requirements, and the health status of the plan’s enrollees.
The aggregate effects of those factors are illustrated by examining current premium levels. Reflecting the choices that individuals and families currently make, premiums for employment-based plans are expected to average about $5,000 per year for single coverage and about $13,000 per year for family coverage in 2009. Premiums for policies purchased in the individual insurance market are much lower—about one-third lower for single coverage and half that level for family policies.1 In large part, those differences reflect the fact that policies purchased in the individual market cover a lower share of enrollees’ health care costs, on average, which also encourages enrollees to use somewhat fewer services. At the same time, average administrative costs are higher for individually purchased policies. The remainder of the difference in premiums probably arises because people who purchase individual coverage have lower expected costs for health care to begin with. In other words, if both employment-based and individually purchased policies covered the same enrollees, the difference in their premiums would be smaller.
Those premiums could change under proposals that would modify the health insurance market or extend coverage to individuals who are currently uninsured. Some proposals could affect premiums by requiring that individuals enroll in plans that meet certain design specifications in order to qualify for subsidies or comply with a mandate. For example, proposals could require that plans cover certain services, limit the amount of cost sharing that would be required of enrollees, or be "actuarially equivalent" to an existing plan.2 The more comprehensive the insurance coverage, the higher the premium would be. Because of the resulting increase in the use of health care services, total spending also would be greater under proposals that reduced cost sharing. CBO has concluded that a 10 percent decrease in enrollees’ out-of-pocket costs would typically cause average spending on health care to increase by 1 percent to 2 percent.
In addition, premiums could be affected by proposals that changed insurers’ management of covered benefits. Most people who have private health insurance are enrolled in some form of managed care plan. Those plans use various techniques to contain health care spending, including negotiating lower fees with a network of providers, requiring that certain services be authorized in advance by the plan or by the patient’s primary care physician, monitoring the care of hospitalized patients, and varying cost-sharing requirements to encourage the use of less expensive prescription drugs. Proposals that restricted plans’ use of such management tools would tend to yield higher premiums and health care spending.
Another factor affecting the level of premiums is the cost of administering a health plan. Some administrative costs (such as those for customer service) vary with the number of enrollees in a plan, but others (such as those for sales and marketing efforts) are more fixed—that is, those costs are similar whether a policy covers 100 enrollees or 100,000. As a result of those economies of scale, the average share of the policy premium that covers administrative costs varies from about 7 percent for employment-based plans with 1,000 or more enrollees to nearly 30 percent for policies purchased by very small firms and by individuals. Some administrative costs are unavoidable, but proposals that shift enrollment away from the small-group and individual markets have the potential to avoid the added administrative costs per enrollee that are observed in those markets. In other cases, however, trade-offs may arise between reducing administrative costs and limiting overall health costs and insurance policy premiums because some administrative costs are incurred when using management tools designed to limit health care spending.
Proposals could change the types of coverage in which many people are enrolled as well as expand coverage to include people who would have otherwise been uninsured. The greatest effects on health care spending are likely to be for the latter group because their use of health care services could increase substantially once they became insured. After accounting for differences in the demographic characteristics and health status of the two populations, CBO estimates that the uninsured use about 60 percent as much care as similar people who are insured.
On the basis of a review of the research literature and original data analysis, CBO concludes that if all people who are currently uninsured were enrolled in health insurance coverage that is similar in design to a typical employment-based plan, they would use between 75 percent and 95 percent as much care as the previously insured. The remaining gap reflects CBO’s assessment that, on average, people without insurance have a somewhat lower propensity to use health care services—a tendency that would persist if they became covered under a new program. Providing all uninsured people with such coverage would thus cause total demand for health care services to increase by 2 percent to 5 percent.
Those estimates are sensitive to the effects of proposals on the extent of coverage that people receive, however; that is, more extensive coverage would have a larger effect. In addition, how proposals that do not achieve universal or near-universal coverage would affect people’s health care spending depends on the extent to which the uninsured would be covered under a plan and on assumptions about the underlying demand for health care among people who would become insured. For more incremental increases in insurance coverage rates, CBO would assume that people who enrolled under a new program would have a greater propensity to use medical care than those who did not enroll. Depending on the design of such a proposal, those newly covered individuals might use health care services at a rate comparable with—or even greater than—that of people with similar demographic characteristics and health status who are currently insured.
In addition, studies indicate that about one-third of the services the uninsured population uses either are provided for free or yield lower total payments to providers than if the same services were provided to privately insured individuals. To the extent that uncompensated care became compensated, spending for the currently uninsured population would rise even if they did not use more services.
Design of Benefits and Cost Sharing
A health insurance plan is essentially a contract between an insurer and an enrollee. In exchange for premium payments, the insurer agrees to cover certain medical services that are specified in the plan. The plan also details the share of costs that both the insurer and the enrollee will bear for each of those services. Thus, two key design elements of a health insurance plan are its scope of covered benefits and its cost-sharing requirements.
Nearly all health insurance policies cover hospitalization, physicians’ services, and prescription drugs—the three largest categories of spending on health care—but greater variation in coverage exists for dental care and more specialized medical services (such as infertility treatments). Legislative proposals to increase the number of insured people could require that health insurance plans cover certain types of medical services. Under such proposals, individuals (or their employers) might not qualify for subsidies or fulfill a mandate unless they were covered by plans that included those benefits.
Benefit mandates ensure that enrollees who may need those services will have coverage for them, but they also tend to raise insurance premiums in order to cover the added costs of the services. The extent of the premium increase resulting from a mandate would depend not only on the costs of the services involved and the likelihood they would be used by enrollees but also on whether health insurance policies would have covered those services in the absence of a mandate. Moreover, because many states already require coverage of various benefits, the impact of any federal mandates would depend on their scope relative to those existing state requirements and their applicability to plans that fall outside the purview of state regulation.
Empirical evidence on the effect of benefit mandates on premiums and coverage is limited. A recent study sponsored by the state of Maryland found that the total cost of services covered by the state’s benefit mandates equaled 15 percent of all covered claims.3 That figure overstates the extent to which benefit mandates raise health insurance premiums nationally, for two reasons: first, because Maryland mandates more benefits than most other states; and second, because some insurers would have covered the mandated benefits even if they had not been required to do so (a factor noted in the study). On the basis of data on mandated benefits in other states and evidence on the extent to which insurers cover such benefits in the absence of mandates, CBO assumes that, averaged across the country as a whole, existing state benefit mandates increase premiums in the individual and small-group markets by approximately 2 percent to 3 percent.4
Cost-sharing requirements—the amount that consumers are required to pay out of pocket when they use health care services—can take the form of deductibles, co-insurance, or copayments. Deductibles are the amount of spending an enrollee must incur before coverage begins; coinsurance and copayments are a portion of spending an enrollee pays at the time of service. Most private insurance plans also limit enrollees’ financial exposure through an annual cap on out-of-pocket spending. (See Chapter 1 for additional discussion of cost-sharing requirements.)
A proposal to increase health care coverage could specify either minimum or maximum levels of cost sharing that would be allowed in order for an insurance policy to qualify for a subsidy or fulfill a mandate. For example, in order to contribute to a health savings account (which allows enrollees to pay many of their out-of-pocket costs using tax-preferred funds), an individual must be enrolled in a health insurance policy that in 2009 has an annual deductible of at least $1,150 for single coverage or $2,300 for family coverage and has an annual limit on out-of-pocket spending that does not exceed $5,800 or $11,600, respectively.
Some proposals might also include additional subsidies to reduce or eliminate cost sharing for lower-income individuals and families—to reflect their more limited ability to pay for services. The Medicaid program fills that role for low-income Medicare enrollees by offering to cover their cost-sharing requirements under Part A and Part B of that program. About 12 million Medicare enrollees with low income and few assets are entitled to subsidies that reduce or eliminate the deductible or other cost-sharing requirements under the Medicare drug benefit. Such subsidies would also raise many of the same issues about identifying and enrolling eligible individuals that arise in designing and implementing premium subsidies. (See Chapter 2 for a discussion of the issues that arise in targeting such assistance toward lower-income individuals.)
Changes in cost-sharing requirements primarily affect premiums by shifting the share of spending that is covered by the policy between the insurer and the enrollee. Those changes can also affect premiums, however, by causing total health care spending to increase or decrease. The best available evidence about the effects of cost sharing on spending for health care comes from the RAND Health Insurance Experiment, a large-scale study that was conducted between 1974 and 1982.5 The RAND study measured the effects of cost sharing on the use of services, expenditures for health care, and health outcomes by randomly assigning nonelderly people to several different types of health insurance plans and tracking their experience over time. A major advantage of using random assignment is that differences in outcomes across plans can be attributed to the design features of each of the plans rather than to the characteristics of the people who were enrolled in them.
The RAND study found that, compared with a plan in which all care was free, a plan with no deductible and a coinsurance rate of 25 percent reduced spending on covered services by about 20 percent; with a coinsurance rate of 95 percent, spending fell by about 30 percent.6 (The differences in health care costs that would be covered by the plan were even larger; compared with the free-care plan, covered costs were about 40 percent lower with 25 percent coinsurance and about two-thirds lower with 95 percent coinsurance.) The RAND study also found that the effect of cost-sharing requirements varied with the type of services provided. Overall, though, cost-sharing requirements resulted in less use of health care services. Compared with study participants who received free care, those with cost-sharing requirements made, on average, one to two fewer visits to their doctors and had 20 percent fewer hospitalizations during a year. The reduction in the use of health care services that resulted from cost-sharing requirements did not have a substantial impact on health outcomes for the general population, although some adverse effects were observed for low-income people in poor health.
Even though the RAND study was conducted more than 25 years ago, its findings continue to be widely used by analysts because of the strength of its experimental design and the limited availability of more recent evidence. Although the provision of health care and the design of insurance plans have changed considerably since the study was conducted, the implications of those changes for the impact of cost sharing are not obvious. Most of the plans examined in the RAND experiment were indemnity plans, which essentially reimbursed enrollees for the health care costs that they incurred; since that time, the spread of managed care has provided insurers with tools to contain health care spending. To try to capture those effects in its analysis, CBO supplements the results from the RAND study with more recent findings and advice from experts in the health care industry.
Based on its review of the literature and discussions with outside experts, CBO assumes that a 10 percent decrease in enrollees’ out-of-pocket costs will generally cause the total spending on their health care to increase by 1 percent to 2 percent. CBO uses that assumption for all types of plans and applies an additional factor to account for the effect of benefit management on spending. The effect of cost sharing on spending varies by type of service. For example, the use of hospital care is less sensitive to cost-sharing requirements than is the use of prescription drugs. A similar analysis would be applied to proposals that provided subsidies to reduce cost sharing for lower-income enrollees.
One useful way to compare health insurance plans with different design features is by examining their actuarial value. That summary statistic measures the share of health care spending for a given population that would be covered by each plan and thus reflects both covered services and cost-sharing requirements (see Box 3-1). Although actuarial value provides one measure of the comprehensiveness of benefits offered by an insurance plan, it does not capture all features of a plan—such as the utilization controls and size of the provider network—that affect the benefits that are delivered. Because of differences in those and other features, such as plans’ administrative costs and the populations they cover, plans with the same actuarial value may charge different premiums.
The actuarial values of current insurance plans vary across employers and between the group market and the individual market. For employment-based plans, actuarial values—expressed as the share of a given population’s medical claims that would be covered by the plan—are typically between 65 percent and 95 percent, with an average value that is between 80 percent and 85 percent. Deductibles and other cost-sharing requirements are typically larger for policies purchased in the individual insurance market, where actuarial values generally range from 40 percent to 80 percent, with an average value that is between 55 percent and 60 percent.
Actuarial-value calculations could be incorporated into legislative proposals in various ways. Proposals that specified a particular benefit design could allow plans to deviate from that design so long as they provided actuarially equivalent benefits. For example, the Medicare drug benefit specifies a standard benefit with a specific deductible, coinsurance rate, and catastrophic threshold (above which enrollees pay about 5 percent of their drug costs).7 But drug plans are allowed to reduce the deductible, vary the coinsurance rate, or use tiered copayments for different types of drugs so long as the plan’s overall actuarial value remains the same and certain other actuarial tests are met. Drug plans are not, however, allowed to increase the deductible or change the catastrophic threshold.
In a similar manner, proposals could require that a qualified plan be actuarially equivalent to an existing plan. For example, the standard Blue Cross and Blue Shield plan available in the Federal Employees Health Benefits program could be used as a model. The impact of that approach would depend partly on the extent to which the model plan differed from the insurance plans that individuals currently have. According to CBO’s analysis, the standard Blue Cross and Blue Shield plan has an actuarial value that is slightly above the national average for employment-based plans. In evaluating a proposal that established the actuarial value of that plan as the minimum requirement for all qualified plans, CBO would account for the increased spending that would result as lower-value plans were enhanced to meet that higher standard.
Whether a required level of coverage was defined using the actuarial value of a benchmark plan or with reference to specific cost-sharing requirements, the way in which those values were updated over time would have important implications for a proposal’s effects on the federal budget and on coverage rates. If a requirement regarding the actuarial value of plans was fixed in nominal dollars (that is, not adjusted for inflation), plans would cover a declining share of health care costs as those costs rose. Alternatively, if plans were required to cover a specified percentage of health care costs, their actuarial value in dollar terms would rise along with those costs. Similar issues would arise if requirements were imposed on cost sharing. If deductibles or other cost-sharing requirements were fixed in nominal dollars, the share of costs covered by an insurance plan would increase over time as health care costs rose—making the coverage more valuable but also increasing its premium. (An example is the deductible under Part B of Medicare, which remained at $100 from 1991 to 2004.)
Those issues can be addressed by indexing a plan’s parameters, but the choice of index can significantly affect the cost of a new program and the scope of coverage provided. If the required actuarial value of plans was specified in dollar terms and updated using a general inflation index (such as the consumer price index) rather than a health-specific index (such as growth in per capita health expenditures), that value would probably decline in future years relative to the cost of health care because health care spending is expected to grow more rapidly than general price levels.
Over the past 30 years, private insurance coverage in the United States has largely shifted away from indemnity policies and toward managed care plans. Such plans vary considerably, but all use management techniques to try to contain health care spending. Provisions that would restrict plans’ use of such techniques would probably increase premiums and health care spending, although the amount of the increase would depend on the details of the provisions.
Most people who have private health insurance are enrolled in some type of managed care plan. Among workers covered by employment-based insurance, about 58 percent are enrolled in preferred provider organizations, about 20 percent are enrolled in health maintenance organizations, and about 12 percent are covered by point-of-service plans.8 Those plans vary in their use of provider networks, authorization requirements for more expensive treatments, and other features. About 8 percent of workers are enrolled in policies known as consumer-directed health plans, which combine a high deductible with an account that enrollees can use to help finance their out-of-pocket costs. Those plans generally feature provider networks and other requirements that are similar to PPO plans but are not included in the count of PPO enrollees.
Average premiums for those plans also vary (see Figure 3-1); that variation reflects not only differences in benefit design and management but also differences in administrative costs, the populations covered by the plans, and the geographic areas that those plans serve. Consequently, those premiums do not measure the average costs of different types of plans in delivering a particular benefit package to a particular enrolled population.
Average Annual Premiums for Covered Workers for Single and Family Coverage, by Plan Type, 2008
Source: Congressional Budget Office based on data from Kaiser Family Foundation and Health Research and Educational Trust, Employer Health Benefits: 2008 Annual Survey (Washington, D.C.: Kaiser/HRET, September 2008).
Note: CDHP = consumer-directed health plan; POS = point-of-service plan; PPO = preferred provider organization; HMO = health maintenance organization.
One important way that managed care plans seek to control costs is by developing networks of providers—including hospitals, physicians, and laboratories—and negotiating discounts within those networks. Many providers are willing to charge lower fees to a plan’s enrollees in the expectation that the plan will direct patients to them (and to avoid the loss of patients that could occur if the provider was excluded from a plan’s network). Despite those efforts, the rates that health plans pay hospitals have risen substantially in many market areas in recent years, as hospitals have achieved stronger bargaining positions in their dealings with plans. The enhanced bargaining position of hospitals has been attributed in part to consolidation among hospitals and also to strong consumer preferences that hospitals not be excluded from a plan’s network.9
Managed care plans differ in their coverage for services that enrollees receive from providers who are not in a plan’s network. That coverage affects plans’ costs directly as well as their leverage in negotiations with providers. PPOs cover services received from any licensed provider, but they encourage enrollees to receive care from providers in their network by charging them less for such care. HMOs do not provide any coverage for services received outside their provider network (except in emergencies). POS plans offer a middle ground: They cover services received outside a plan’s network (usually for a higher charge), but they typically apply some of the same management tools that HMOs use to limit costs within the network.
The differences between PPOs and HMOs have generally narrowed in recent years. In the past, HMOs typically had smaller provider networks than PPOs. In response to consumer demand for broader networks, however, HMOs have generally increased the size of their networks, and many are now similar to PPOs in that respect; indeed, some insurance carriers use the same provider network for their HMO and PPO products.
Use of Other Cost-Containment Practices
Health plans use a variety of other practices to contain health care costs. One approach is to manage access to expensive medical benefits by requiring prior authorization before the services will be covered. A second approach is to use price signals—that is, variations in cost-sharing requirements—to encourage enrollees to use less expensive medical care. Managed care plans also may use evaluations of providers on both price and quality terms to give feedback to those providers or to structure the information and incentives given to enrollees.
Managing Access and Use. Controlling enrollees’ access to more specialized (and expensive) medical services can help health plans manage their costs. Many plans require prior authorization for nonemergency hospital admissions and other selected services. To limit access to expensive drugs, many plans use "step therapy"—a process in which patients are required to begin treatment with less expensive alternatives (such as generic drugs) and then switch to a more expensive drug only if necessary. Enrollees in some HMO and POS plans must select a primary care physician who is responsible for approving referrals to specialists, but that approach is less common than in the past. More generally, differences between HMOs and PPOs have diminished as many managed care plans eliminated or relaxed some of their cost-control procedures in response to widespread complaints in the late 1990s from consumers and providers. However, some of those cost-containment procedures were subsequently reinstated or replaced by new procedures to limit spending.10
Some managed care plans also seek to monitor the performance of providers and the use of services more actively. To limit the length of hospital stays, plans may conduct "concurrent reviews" of the care and medical condition of hospitalized patients; that is, the insurer may review the medical necessity of a treatment upon or shortly after a patient’s hospital admission and continue to monitor the services provided during the course of his or her treatment. Some insurers also apply such reviews to patients receiving other types of care, such as rehabilitation and skilled nursing care.11 Insurers sometimes use information from medical claims to obtain detailed information about the care furnished by individual providers (a procedure known as provider profiling). That information is then used to give feedback to providers on how their practice patterns compare with those of their peers and to identify providers who are furnishing inappropriate or excessive care (and who might be removed from the plan’s network as a result).
Varying Cost Sharing. Plans encourage enrollees to use providers within their network by requiring lower cost sharing for in-network care. In some cases, plans also use differences in cost-sharing requirements or other techniques to influence consumers’ choices within their approved networks or range of covered treatments and services. For example, plans generally establish a drug formulary or list of drugs that the plan covers (which is akin to a provider network). In addition, plans typically try to limit spending on prescription drugs by negotiating price discounts or rebates from drug manufacturers in return for giving their drugs preferred status on the formulary—and with it, lower copayments for enrollees. Plans also encourage enrollees to use lower-cost generic versions of drugs when they are available, by setting the lowest copayment amounts for those drugs.
More recently, some plans have begun using the information collected from provider profiling to designate a preferred "tier" of providers based on quality and cost standards. For example, an insurer might rank providers on the basis of the total cost per medical "episode," which accounts for complications, readmissions to the hospital, and other factors that go beyond a consideration of the provider’s fees. In addition, enrollees may be given financial incentives—such as lower cost-sharing requirements—to receive their care from higher-tier providers.
Effects of Managed Care on Premiums and Spending
Determining the effects of the various cost-containment tools can be difficult because health plans use different combinations of them, and plans vary along a number of other dimensions. Consequently, much of the published research has focused on comparing HMOs (which have traditionally used more stringent cost-containment methods) with other types of plans. The evidence suggests that HMOs deliver a given package of benefits at a lower cost than PPOs and other plans. In particular, studies have found that HMOs reduce the use of hospital services and other expensive services.12 Because those studies rely largely on data that are more than a decade old, however, they probably overstate the differences that exist today between HMOs and other types of plans. On the basis of the available evidence, CBO estimates that plans making more extensive use of benefit-management techniques would have premiums that are 5 percent to 10 percent lower than plans using minimal management techniques.
The slowdown in health care spending that occurred in the 1990s—as private insurance coverage shifted away from indemnity policies and toward various forms of managed care—provides additional evidence on the potential effects of managed care on spending. Before 1993, health care spending generally grew at a faster rate than gross domestic product. From 1993 to 2000, the share of workers with private health insurance who were enrolled in some kind of managed care plan rose from 54 percent to 92 percent. During that period, total spending for health care remained nearly constant as a share of the economy, at about 13.8 percent of GDP. Many analysts believe that the growth in managed care plans contributed significantly to the slowdown in the growth of health care spending during that period.
By the end of the 1990s, opposition to the restrictions imposed by managed care plans was growing among consumers and providers. The plans responded by relaxing those restrictions, and enrollment shifted to more loosely managed PPO plans. Health care spending also began to increase at a faster rate, rising to 16.0 percent of GDP in 2006. Other factors, however, have undoubtedly contributed to the growth in health care spending relative to the size of the economy since 2000; hospital mergers became more pervasive, for example, enhancing hospitals’ leverage in negotiating with health plans.
Regulating the Operations of Health Plans
Proposals to change the health insurance market or to subsidize insurance purchases might include provisions affecting the management of health plans. During the past decade, for example, the Congress has considered several versions of legislative proposals—commonly referred to as a "Patients’ Bill of Rights"—that would have restricted insurers’ management of health benefits. Although lawmakers did not enact those proposals, some states adopted similar provisions restricting health insurers that operate in their jurisdiction. (As discussed in Chapter 1, plans purchased in the individual insurance market and most plans purchased by smaller employers are subject to state regulations, whereas the majority of plans offered by larger employers are exempt.) In modeling the effects of such proposals, CBO considers the nature of any provisions governing the plan’s structure, utilization management, and provider networks and their interaction with existing state requirements.
Types of Provisions. Proposals like the Patients’ Bill of Rights could change how health insurers interact with enrollees, in several ways. Under some proposals, insurers would be required to cover certain types of care, such as visits to specialists, without a referral from an enrollee’s primary care physician. Past proposals also would have granted enrollees rights of redress, allowing those who had been denied coverage for a particular service to appeal the decision or pursue other remedies in civil courts. Transactions among insurers, providers, and enrollees are another area of concern, with legislative proposals addressing how information about a plan is presented to enrollees or specifying rules for the prompt payment of claims to providers.
Other provisions could also regulate insurers’ networks of providers. Any-willing-provider laws require that health plans include in their network any provider who agrees to abide by the terms and conditions of the plan’s contract. Many states enacted such laws in the 1990s, but those laws do not apply to employment-based plans that are exempt from state regulation. Network-adequacy requirements would establish rules about the number of different types of providers that plans must have in their network, and restrictions on provider profiling would limit plans’ ability to use their analysis of medical claims and other factors to exclude providers from their networks or to develop tiered networks.
Effects on Health Insurance Premiums. In its previous analyses of proposals to create a Patients’ Bill of Rights in 1999 and 2001, CBO generally determined that many of their provisions—which are similar to those described above—would increase spending on health care.13 Since then, however, many health plans have dropped certain cost-containment procedures or replaced them with other techniques; to the extent that such changes were not anticipated, the magnitude of CBO’s estimates of the effects of new proposals that affect plans’ management techniques may differ from its previous findings.
For certain provisions that CBO analyzed, the effects today would most likely differ from what the agency previously estimated. For example, CBO estimated that a federal any-willing-provider law or federal network-adequacy requirements and proposals requiring plans to cover certain types of care—including visits to specialists without prior authorization, visits to an emergency room if a "prudent layperson" would have regarded the patient’s condition as an emergency, and the routine costs of enrollment in approved clinical trials—would, in combination, have increased private health insurance premiums by amounts ranging from 1.2 percent to 1.7 percent. If reintroduced today, however, similar provisions would probably have a smaller impact on premiums; to an extent not anticipated in CBO’s original estimates, many health plans have acceded to consumers’ preferences for broader access to care by expanding the size of their provider networks and eliminating or reducing some of their restrictions on the use of covered services.
For other provisions, CBO’s estimates of the effects would be similar to the agency’s previous estimates. For instance, the effects of proposals to expand enrollees’ access to the courts for pursuing civil remedies to settle disputes with insurers would probably be similar to the effects that were estimated in 2001 because the expectation in the original estimates that the legal environment would not change substantially has, so far, proved to be accurate. At that time, CBO estimated that the combined effect of various provisions creating new civil remedies and establishing grievance processes would have been to increase premiums by 1.1 percent to 1.7 percent.
Administrative Costs of Health Plans
Proposals to change the regulation of insurance markets—as well as many other types of proposals—could affect the costs of health insurance by changing the administrative costs of health plans (sometimes referred to as "administrative load"). In this discussion, administrative costs refer to any expenses insurers incur that are not payments for health care services, including the profits retained by private insurers and the taxes paid on those profits.
Administrative costs can be divided into three categories:
- Marketing costs include expenses for advertising, sales, enrollment processing, customer service, billing, and actuarial and underwriting activities. (Underwriting involves an assessment of an applicant’s health and expected use of health care in order to determine what premium to charge.)
- Costs associated with medical activities include expenses for claims review and processing, medical management (such as utilization review, case management, quality assurance, and regulatory compliance), and provider and network management (contracting with doctors and hospitals and maintaining relations with providers).
- General administrative costs are difficult to allocate to a specific function; they include expenses for information technology, general management overhead, profits, and taxes.
According to a recent analysis, administrative costs for private health insurance totaled $90 billion in 2006 (see Table 3-1), of which about $24 billion was for marketing and related costs, roughly $14 billion was for medical activities, and about $52 billion was for general expenses (including $9 billion in tax payments and $21 billion in after-tax profits).14 Overall, those costs accounted for about 12 percent of private insurance premiums.
Administrative Costs for Private Health Plans, by Category, 2006
Source: Congressional Budget Office based on Diana Farrell and others, Accounting for the Cost of U.S. Health Care, 2008: A New Look at Why Americans Spend More (San Francisco: McKinsey Global Institute, December 2008).
Note: * = between zero and $500 million.
A common metric that is used to assess an insurer’s administrative costs is the ratio of claims payments to the total premium, referred to as the "medical loss ratio." (The difference between the medical loss ratio and 100 percent is the share of the premium devoted to administrative expenses.) When comparing two plans that are equivalent on other dimensions, such as the total premium and quality of service, a low loss ratio could indicate a plan that is run less efficiently. But a loss ratio is not always indicative of a plan’s efficiency or value.15 For example, a health plan that devotes more resources to managing the use of health care services might have a relatively low loss ratio but also a lower overall premium. In contrast, a more lightly managed plan might have a high loss ratio but a correspondingly higher overall premium and might be covering more services that provide limited health benefits. The former plan, despite its low loss ratio, may well be preferable because of its lower overall premium for the package of services that it provides. Thus, a loss ratio provides just one way of evaluating a health plan’s administrative expenses.
Variation of Administrative Costs
Administrative costs typically vary not only by the type of insurance plan but also by the size and nature of the group being insured. Among employment-based plans, the share of the premium that pays for administrative costs varies significantly by the size of firms, from about 7 percent for firms with at least 1,000 employees to 26 percent for firms with 25 or fewer employees.16 The latter loading factor is comparable with the one seen in the individual insurance market, where administrative costs account for nearly 30 percent of premiums.
To a large extent, the variation in administrative costs among private plans reflects economies of scale. Some types of administrative costs, such as sales and marketing expenses, are relatively fixed for the group being insured; thus, the larger the group, the smaller the cost per enrollee. In particular, plans that are sold to individuals and small groups are more likely to incur fees for insurance agents and brokers to handle the responsibilities that larger firms generally delegate to their human resources departments—such as finding plans and negotiating premiums, providing information about the selected plans, and processing enrollees. Because large firms can spread those costs over a greater number of enrollees, their average administrative costs per enrollee are lower.
Other factors appear to play a lesser role in the variation of average administrative costs across markets. One commonly cited difference is that underwriting is used in the individual and small-group markets, but those efforts appear to account for a relatively small share of insurers’ administrative costs and thus seem unlikely to explain the higher administrative costs per enrollee that are observed in those markets. Plans sold in the individual and small-group markets are also generally subject to state taxes on the premiums they collect, whereas the plans offered by large employers are generally exempt from such requirements. Other expenses—such as the costs of responding to telephone calls from enrollees and providers with questions regarding coverage and payments—are roughly proportional to the number of enrollees (at least for broadly similar populations) and thus would probably constitute a similar share of the premiums for groups of different sizes.
Potential Effects of Proposals on Administrative Costs
Depending on their design, proposals could have a significant impact on the administrative costs involved in providing health insurance—which, in turn, could have a substantial effect on policy premiums. Administrative costs would probably be affected indirectly by proposals that altered the number of insurers, the size of purchasing pools, and insurers’ responsibilities. Some proposals might seek to limit the amount spent on administrative costs by specifying a minimum loss ratio, but the net effect of such proposals on insurance premiums or health care spending is uncertain.
Trade-offs are likely to arise between the number of insurance plans that are offered to consumers and the total administrative costs incurred by all insurers. Because some administrative costs are largely fixed, duplication of functions would arise in proportion to the number of insurers participating in the market. Greater competition among insurers, however, would also tend to provide stronger incentives to control costs and thus could yield lower total premiums despite causing aggregate administrative costs to increase.
Proposals that would organize insurance purchasers into larger groups could avoid some of the high administrative costs observed in the individual and small-group markets. In the extreme, if a proposal established a purchasing system under which all insurers incurred administrative costs that were comparable with the costs of large employment-based plans, average policy premiums would be about 3 percent lower than they would be if administrative costs for individual and small-group purchasers remained at their current levels. Administrative savings, however, might be smaller if plans still had to rely on insurance agents and brokers to enroll workers who were not employed by large firms or if other entities had to perform similar functions.
Some proposals would try to directly limit administrative costs by mandating minimum loss ratios—that is, by specifying that the amounts spent on benefits should be at least some specified percentage of the premium. That strategy could be problematic, however, because a high loss ratio may not imply greater efficiency on the part of an insurer. Moreover, whether insurers serving the individual and small-group markets could increase their loss ratios simply because they were required to do so is not clear, so the effects of such requirements on those markets are hard to predict. If the requirement was set too high, insurers would probably exit the market.
Effects of Gaining Insurance Coverage on Health Care Use and Spending
Proposals that expand coverage to people who currently lack insurance would lead to an increase in their use of medical services, which in turn would affect the costs of those proposals and their impact on spending for health care. The extent to which the demand for care would increase depends partly on the number and characteristics of the newly enrolled individuals—including their health status and their preferences for medical care—and partly on the scope of the coverage that they obtain. Estimating that likely impact presents a number of challenges.
Based on a review of the research literature and original analysis, CBO concludes that if all people who are currently uninsured were enrolled in insurance coverage equivalent to a typical employment-based plan, they would use about 75 percent to 95 percent as much medical care as people who are currently insured (and also have the same demographic characteristics and health status). Those figures provide a benchmark for analyzing the impact of various coverage expansions. Depending on their design, proposals for more incremental coverage expansions could provide coverage to a group of people who would use at least as much health care as similar people who are currently insured.
Estimates of Demand for Health Care by the Uninsured
How much more care the uninsured would seek and the impact that such an increase would have on premiums and spending depend in part on how much care they now receive. According to several studies and CBO’s own analysis of the nonelderly population, the uninsured use about 50 percent to 70 percent as many health care services as the insured.17 A key challenge in estimating the impact of a coverage expansion is sorting out the extent to which that disparity stems from the uninsured’s lack of coverage, how much reflects other observable differences between the insured and the uninsured, and what role is played by differences that researchers cannot easily observe.
Although there are substantial demographic differences between the insured and the uninsured, some of those differences have offsetting effects on their relative use of services. For example, younger adults are represented disproportionately in the uninsured population, whereas the insured population is more likely to contain children (who tend to use fewer health care services than average) and older adults (who have above-average use). As a result, differences in age do not appear to explain much of the overall disparity in use of services between the insured and the uninsured. Differences in health status may play a larger role. CBO’s analysis of survey data indicates that the share of the nonelderly population reporting their health as fair or poor is higher among the uninsured (10 percent) than among the privately insured (5 percent).
A more difficult factor to assess is whether the uninsured differ from those with insurance in other less observable ways that affect their demand for health care services. Understanding the reasons that the uninsured currently do not have insurance could also provide some insight into how they would respond to an increase in coverage.
The uninsured are not a monolithic group, however, and there are many reasons that they lack coverage. Some uninsured individuals may have a strong preference for health insurance but lack coverage because of limited financial resources. If those financial constraints were relaxed, their use of health services might become comparable with that of otherwise similar people who have insurance. Other people may not purchase insurance because they place a relatively low value on health care or think they will not need to use it. Still others may be willing to accept more risk than those who enroll in health insurance plans or may believe that they will be able to obtain the care they need without insurance. Such individuals may not substantially increase their use of health care services even if they become insured.
Both because individuals’ preferences for health care vary and those preferences are not easy to observe, estimating the impact of gaining health insurance coverage on the use of medical services is difficult. If individuals who are more likely to use health care are also more likely to have insurance, simple comparisons of the insured and uninsured populations would overstate the impact of becoming insured. An ideal research strategy would randomly assign individuals to an insured or uninsured group and see how much care they use—but people would be understandably reluctant to participate in such an experiment. Short of that, researchers have used three broad methodological approaches to examine the extent to which people who are uninsured would increase their use of services if they were provided with coverage:
|
■ |
Simulations based on findings from the RAND Health Insurance Experiment, which randomly assigned individuals to different insurance plans; |
|
■ |
Analysis of so-called natural experiments, in which coverage under Medicaid or Medicare has been extended to individuals who previously lacked insurance; and |
|
■ |
Studies that have compared the use of services by people who are insured with that of people who are uninsured, taking into account various differences among the two populations. |
None of those approaches resolves all of the methodological issues that arise when trying to estimate an uninsured individual’s likely use of health care if provided with insurance. Reflecting the different strengths and weaknesses of the approaches—as well as their differing sources of data and analytic techniques—studies based on those approaches have yielded a wide range of estimates of those effects.
The studies also differ in what they examine. Some studies focus on people’s use of services (primarily doctors’ visits and hospitalizations), and others analyze spending. The impact that covering uninsured individuals has on spending for health care and health insurance premiums depends both on the quantity of services that they use and on the amount per service that is paid to the providers of their care. Analyzing those elements separately can be useful, however, because differences in payment rates can complicate comparisons between insured and uninsured individuals. In particular, a substantial minority of the care that the uninsured receive is uncompensated or undercompensated—that is, they either pay nothing for it or pay less than the amount that a provider would receive for treating an insured patient. To the extent that such care became compensated under a proposal to expand coverage, health care spending for the uninsured would increase, regardless of whether their use of care also rose. (Other factors that might affect the impact of an insurance expansion on spending, including any constraints on the supply of health care services and any effects on other payment rates from reductions in uncompensated care, are discussed in Chapter 5.)
Simulations Based on Experimental Evidence. One study used the results of the RAND Health Insurance Experiment to simulate spending for individuals enrolled in hypothetical insurance plans as well as for individuals with no insurance.18 Although that study used older data, its results can be adjusted so that they reflect current levels of health care spending. The simulations indicate that, on average, people enrolled in a plan with a $400 deductible, 25 percent cost sharing, and a $4,000 maximum on out-of-pocket spending (a plan that is roughly equivalent to typical employment-based plans) would incur about 13 percent more in health care expenditures than similar people who are uninsured. Because the prices paid for services were standardized in that analysis, that difference in spending also reflects the projected differences in use of services.
Basing simulations on the findings from the RAND experiment builds on both the strengths and weaknesses of that study. Because random assignment ensured that enrollees in different plans were comparable, its design allowed researchers to isolate responses to variations in the extent of insurance coverage. Yet the RAND study did not include any individuals who were uninsured; the closest design resembled a high-deductible plan, which covered more than half of its enrollees’ health care costs. The researchers therefore had to extrapolate from the RAND results to simulate spending among the uninsured. One concern with that approach is that uninsured individuals may be more reluctant to seek treatment than a comparison of enrollees in low-deductible and high-deductible plans would indicate. Extrapolating from those results may also fail to reflect certain constraints on the medical care available to the uninsured. For example, some physicians do not accept new patients who are uninsured. As a result, simulations based on the RAND study’s results may underestimate the increased use of services that would occur if insurance coverage was extended to people who are uninsured.
Studies of Expansions in Medicaid and Medicare. Some studies have examined the change in service use that occurs when uninsured people become eligible for Medicaid or Medicare. The creation or expansion of such programs can provide useful insights, but only to the extent that people gain insurance coverage for reasons that are unrelated to their health or their preferences about health care. One study examined the impact of Medicaid expansions between 1984 and 1992, finding that children who became eligible for the program increased their likelihood of visiting a physician at least once during the year by about 10 percent and roughly doubled their probability of being hospitalized.19 A more recent study on expanding eligibility for Medicaid found similar effects on visits to physicians.20
Two other studies examined health care use by previously uninsured individuals shortly before and after they became eligible for Medicare at age 65—a natural experiment that is similar in many respects to a proposal that yields near-universal coverage. One study, which focused on a subset of clinical services, found that the use of preventive care by the previously uninsured rose substantially once they were eligible for Medicare but remained below the levels seen for individuals who had been insured before age 65.21 In addition, visits for arthritis treatments not only increased substantially among those who lacked insurance before becoming eligible for Medicare but also reached a higher level than was seen for the continuously insured. The other study compared overall numbers of physicians’ visits and hospital admissions and found that, among the near elderly, use of care was about 15 percent lower for those who lacked insurance coverage compared with those who were insured.22 Once they enrolled in Medicare, previously uninsured individuals increased their health care use by 30 percent to 40 percent and ended up with higher levels of use than were observed for Medicare enrollees who had been insured before becoming eligible for Medicare at age 65 (although the differences were not always statistically significant).
One advantage of the studies of Medicaid and Medicare is that they clearly isolate the effects of gaining insurance coverage; one limitation is that their results may not be applicable to the entire uninsured population. To some extent, those studies may reflect the responsiveness of people who are uninsured primarily because of financial constraints—and who are thus more likely to increase their use of health services once they receive coverage. More specifically, the findings related to Medicaid expansions may apply only to similar proposals. On the one hand, Medicaid coverage has relatively low cost sharing, so it could have stimulated demand for care to a greater extent than a typical insurance policy; on the other hand, the observed impact may have been dampened because some of the people who gained eligibility for Medicaid would have otherwise had private coverage and because some doctors do not accept Medicaid’s relatively low payment rates.
Comparisons of Insured and Uninsured Populations. Other studies have used statistical methods to try to isolate the effects of insurance when comparing the use of medical services by people who are insured and those who are uninsured. Those studies attempt to identify and adjust for other differences between the two populations (such as income and health status) that would be expected to influence their use of services. To the extent that the insured and the uninsured differ in ways that are not observed in the data, however, those studies may not have isolated the effect of insurance on the use of medical services. The studies themselves also vary along several dimensions. Some studies examine the use of services, and others analyze spending; some look at a cross-section of insured and uninsured people, and others focus on changes in coverage over time. In addition, the studies may analyze different subsets of the nonelderly population. (Because a large number of studies have made such comparisons, this discussion highlights only a few of them.)
One recent example of a cross-sectional study compared spending on health care for the insured and uninsured populations and then sought to project what expenditures for the uninsured would be if they gained a typical level of coverage.23 The study used data on the nonelderly population from the Medical Expenditure Panel Survey (MEPS), a large-scale survey that collects information on individuals’ insurance coverage and use of health care services. The analysis controlled for differences in demographic, health, and socioeconomic characteristics between the insured and uninsured populations. To account for the fact that a large share of services received by the uninsured are uncompensated, the study also sought to adjust their spending figures upward so that they would reflect payment rates for the privately insured. The study estimated that spending on health care for the uninsured—and, by implication, their use of services—would increase by about 70 percent if they became continuously insured; the estimated impact was larger for individuals who had been uninsured all year and smaller for those who spent only part of the year uninsured. The resulting amount of spending per person was similar to that observed for privately insured individuals.
Other studies have tracked individuals from one year to the next to see what happens when they gain or lose insurance coverage. One recent study used MEPS data to analyze the health care expenditures of insured nonelderly adults.24 It found that individuals who were also insured in the previous year had expenditures that were similar to those who had been previously uninsured—suggesting that spending for the uninsured will rise to the level seen for the insured once they gain coverage. That approach has the advantage of avoiding the need to estimate how much uncompensated care the uninsured receive. An important limitation of that approach, however, is that individuals who became insured in the second year may not be representative of the entire uninsured population; that is, they may have obtained coverage partly because of a greater preference for medical care (or a greater need for care) than those who remained uninsured in the second year. In particular, some people may have been enrolled in Medicaid when they were hospitalized or treated in an emergency room—in which case it was their higher health care spending that caused them to become insured rather than their insurance causing them to use more health care.
More generally, the results of such comparisons are sensitive to the methodologies used and to the types of people included in the analysis. To examine that sensitivity, CBO conducted its own analysis of MEPS data to track changes in insurance coverage from year to year. The analysis measured use of services rather than expenditures, included children and nonelderly adults, and excluded people who had public coverage.25 CBO conducted the analysis separately for two groups of people—those who were insured for only part of the first year and those who were uninsured throughout the entire first year—and then compared the use of services in the following year by people in each group who were insured or uninsured during all of that year. Grouping people on the basis of their coverage in the first year was designed to limit the extent to which people gaining or losing coverage in the second year differed with respect to their attitudes toward health insurance and medical care. Comparing the use of services rather than expenditures holds aside any differences in the prices paid for the same services—and the challenges of measuring uncompensated care—that could affect the results.26 One downside of that approach, however, is that it could miss differences in the types of services provided to insured and uninsured individuals during a given visit to a physician or a hospital stay.
Among people who were insured for only part of the first year, those who were insured for all of the second year used 26 percent more health care services during that year than those who were uninsured for all of the second year. More specifically, people in that group who were insured for all of the second year used 88 percent as much care in that year as those who were continuously insured in both years, whereas those who were uninsured for all of the second year used 70 percent as much care as the continuously insured. The possibility that some of those who became insured in the second year obtained coverage partly because of a greater preference for medical care (or a greater perceived need for care) than those who were uninsured in the second year also remains an issue with that analytic approach. To the extent that such sorting occurred, however, it would mean that the true effect of gaining insurance coverage was smaller than the 26 percent estimate.
Among people who were uninsured throughout the entire first year, CBO found that those who were insured during all of the second year used 29 percent more services during that year than people who were uninsured throughout the year. That is, those who gained coverage in the second year used 67 percent as much care as people who were continuously insured in both years, whereas those who remained uninsured in the second year used 52 percent as much care as the continuously insured. Thus, in both analyses, people who gained coverage in the second year increased their use of services, but they did not use the same amount of care as people who had been continuously insured in both years.
Although the wide range of estimates generated by different studies makes it difficult to be certain about the effects of gaining health insurance coverage on health care use and spending, some central tendencies can be observed. A 2005 review of the research literature in that area analyzed studies using the strongest methodologies and concluded that extending insurance coverage to the uninsured would increase the number of physicians’ visits by 30 percent to 50 percent for children and by 60 percent to 100 percent for adults.27 The studies that were reviewed obtained a broader range of estimates for the effects on children’s use of inpatient hospital care; among adults, the estimated increases in hospital use ranged from about 40 percent to 80 percent. That review encompassed studies evaluating Medicaid expansions and research comparing insured and uninsured populations; it did not include the findings based on the RAND study or some of the more recent research on Medicare.
Examining the full range of studies also highlights the potential biases in each type of analysis. The findings based on the RAND experiment may have understated the dampening effect of being uninsured on expenditures and thus underestimated the increase that would result from gaining insurance coverage. Conversely, studies relying on comparisons of currently insured and uninsured people, or on changes in their insurance coverage from one year to the next, may overstate the true effect of gaining coverage because those studies may not fully account for differences between the two groups in their attitudes toward health care and health insurance. Studies using natural experiments resulting from policy changes often yield intermediate results, but those findings generally reflect how the target populations would respond and thus may not apply to proposals that would achieve near-universal coverage.
Based on its review of the literature and analysis of health care data, CBO has adopted an intermediate range of estimates. Specifically, the agency expects that providing all of the uninsured with health insurance coverage equivalent to a typical employment-based plan would increase their demand for medical services to a level that is between 75 percent and 95 percent of the level of similar people who are currently insured. (To the extent that the insured and uninsured populations differ in age and health status, CBO would make additional adjustments to account for the effects of those differences on health care use and spending.) Relative to current amounts of health care use by the uninsured—for which estimates average around 60 percent of the amount seen for insured individuals—those assumptions reflect an increase of between 25 percent and 60 percent that would result from gaining insurance coverage. Compared with currently projected levels of health care use for the population as a whole, that rise in the demand for services would constitute an increase that is between 2 percent and 5 percent.28
Those figures provide a framework for analysis, but the effects of specific proposals will depend in part on the extent of coverage that the uninsured receive; that is, more extensive coverage would have a larger effect on health care use and spending. In addition, the impact of proposals that did not achieve universal or near-universal coverage would depend on the extent to which the uninsured would be covered under a plan and on assumptions about the underlying demand for health care among those who would become insured. For more incremental increases in insurance coverage rates, CBO would assume that people who enrolled under a new program would have a greater propensity to use medical services than those who did not enroll. Depending on the design of such a proposal, those newly covered individuals might use health care services at a rate comparable with—or even greater than—that of people who have similar demographic characteristics and health status and are currently insured.
All else being equal, the increase in use of services by previously uninsured individuals would also yield a corresponding increase in health care spending. In addition, measured spending for the uninsured population would rise as uncompensated care that they had received was compensated by their insurance plan. Even so, the assessment that the entire pool of uninsured individuals would have somewhat lower use of services under a typical employment-based policy means that they would have somewhat lower total spending per person than those who are currently insured. The expected effect on spending also depends on several other factors, however; in particular, the rates used to pay providers are an important consideration, and limits on the ability and willingness of health care providers to meet an increase in demand could affect both utilization rates and payment rates (see Chapter 5 for more details).
See Didem Bernard and Jessica Banthin, Premiums in the Individual Insurance Market for Policyholders Under Age 65: 2002 and 2005, AHRQ Statistical Brief No. 2002 (Agency for Healthcare Research and Quality, April 2008); 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); and the Kaiser Family Foundation and Health Research and Educational Trust, Employer Health Benefits: 2005 Annual Survey (Washington, D.C.: Kaiser/HRET, September 2005) and Employer Health Benefits: 2007 Annual Survey (September 2007).
Actuarially equivalent plans cover the same share of health care spending for a given population; see Box 3-1 for a more detailed explanation.
Maryland Health Care Commission, Study of Mandated Health Insurance Services: A Comparative Evaluation (January 1, 2008), http://mhcc.maryland.gov/health_insurance/mandated_1207. pdf.
For evidence on the extent to which insurers would cover mandated benefits in the absence of the mandates, see Maryland Health Care Commission, Study of Mandated Health Insurance Services; and Jonathan Gruber, "State-Mandated Benefits and Employer-Provided Health Insurance," Journal of Public Economics, vol. 55 (1994), pp. 433–464.
For a description of the RAND experiment, see Joseph P. Newhouse and the Insurance Experiment Group, Free for All?: Lessons from the RAND Health Insurance Experiment (Cambridge, Mass.: Harvard University Press, 1993).
The plans that required coinsurance also featured an annual limit on enrollees’ out-of-pocket costs that was equal to the lesser of $1,000 (in nominal dollars) or a percentage of family income.
The standard drug benefit also specifies an initial coverage limit; enrollees who do not have additional coverage are responsible for all of their drug costs between that point and the catastrophic threshold (in the so-called doughnut hole).
The Kaiser Family Foundation and Health Research and Educational Trust, Employer Health Benefits: 2008 Annual Survey. Only about 2 percent of workers are enrolled in indemnity plans that do not use managed care tools.
See Justin S. White, Robert E. Hurley, and Bradley C. Strunk, Getting Along or Going Along? Health Plan-Provider Contract Showdowns Subside, Issue Brief No. 74 (Washington, D.C.: Center for Studying Health System Change, January 2004).
Glen P. Mays, Gary Claxton, and Justin White, "Managed Care Rebound? Recent Changes in Health Plans’ Cost Containment Strategies," Health Affairs, Web Exclusive (August 11, 2004), pp. W4-427 to W4-436.
Plans use various methods to pay providers, some of which are designed to give providers incentives to limit spending. The payment methods that plans use and their effects are examined in Chapter 5.
See Robert H. Miller and Harold S. Luft, "HMO Plan Performance Update: An Analysis of the Literature, 1997–2001," Health Affairs, vol. 21, no. 4 (July/August 2002), pp. 63–86.
See the Congressional Budget Office’s cost estimates for S. 6, Patients’ Bill of Rights Act of 1999 (June 16, 1999); for H.R. 2315, Patients’ Bill of Rights Act of 2001 (July 20, 2001); for S. 1052, Bipartisan Patients’ Bill of Rights Act (July 20, 2001); and for H.R. 2563, Bipartisan Patient Protection Act (July 26, 2001). All of CBO’s estimates reflect the marginal effect of the federal legislation on costs. That is, they compare the projected costs under the new law with the estimated costs that would have prevailed in the absence of the new law (recognizing that some health plans may have made the proposed changes even without the new law).
See Diana Farrell and others, Accounting for the Cost of U.S. Health Care, 2008: A New Look at Why Americans Spend More (San Francisco: McKinsey Global Institute, December 2008). National health expenditure data show a comparable estimate of about $89 billion for 2006.
For a discussion, see James C. Robinson, "Use and Abuse of the Medical Loss Ratio to Measure Health Plan Performance," Health Affairs, vol. 16, no. 4 (July/August 1997), pp. 176–187.
Lower estimates of administrative costs for large-employer groups may reflect only the fees paid to insurers who act as third-party administrators but who do not assume financial risk for operating an employer’s plan; when the employer is acting as the insurer, some administrative costs are borne by the employer but may not be included in the estimates.
For example, see M. Susan Marquis and Stephen H. Long, "The Uninsured Access Gap and the Cost of Universal Coverage," Health Affairs (Spring 1994), pp. 211–220; and Brenda C. Spillman, Stephen Zuckerman, and Bowen Garrett, Does the Health Care Safety Net Narrow the Access Gap? Discussion Paper 03-02 (Washington, D.C.: Urban Institute, April 2003).
Joan L. Buchanan and others, "Simulating Health Expenditures Under Alternative Insurance Plans," Management Science, vol. 37, no. 9 (September 1991), pp. 1067–1090.
Janet Currie and Jonathan Gruber, "Health Insurance Eligibility, Utilization of Medical Care, and Child Health," Quarterly Journal of Economics, vol. 111, no. 2 (May 1996), pp. 431–466. Because actual enrollment in Medicaid reflects preferences about insurance and may be triggered by a health problem, the study compared groups who gained eligibility for the program (regardless of whether they actually enrolled) to groups whose eligibility did not change.
Jessica S. Banthin and Thomas M. Selden, "The ABCs of Children’s Health Care: How the Medicaid Expansions Affected Access, Burdens, and Coverage Between 1987 and 1996," Inquiry, vol. 40, no. 2 (Summer 2003), pp. 133–145.
J. Michael McWilliams and others, "Impact of Medicare Coverage on Basic Clinical Services for Previously Uninsured Adults," Journal of the American Medical Association, vol. 290, no. 6 (August 13, 2003), pp. 757–764.
J. Michael McWilliams and others, "Use of Health Services by Previously Uninsured Medicare Beneficiaries," New England Journal of Medicine, vol. 357, no. 2 (July 12, 2007), pp. 143–153.
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.
Lisa Ward and Peter Franks, "Changes in Health Care Expenditure Associated with Gaining or Losing Health Insurance," Annals of Internal Medicine, vol. 146, no. 11 (June 2007), pp. 768–774. That study also found that individuals who lost insurance coverage had expenditures while uninsured that were comparable with those of continuously uninsured individuals.
CBO used MEPS data for 1997 to 2005 rather than data for a single two-year period to expand the size of the samples and thus increase the precision of the estimates. The data were adjusted for age, sex, health status, education, and income so that the analysis compared individuals who appear similar in observed characteristics other than insurance status.
CBO measured the use of health care services using a single index that reflected the use of physicians’ and hospital services, weighted by average expenditures for each kind of use. Thus, reported payments for individual services did not affect the comparisons between insured and uninsured individuals.
Thomas C. Buchmueller and others, "The Effect of Health Insurance on Medical Care Utilization and Implications for Insurance Expansion: A Review of the Literature," Medical Care Research and Review, vol. 62, no. 1 (February 2005), pp. 3–30.
The uninsured currently account for about 8 percent of health care use nationwide; a 25 percent increase in their use of services would thus translate into a 2 percent increase for the country as a whole, and a 60 percent increase would translate into a rise of nearly 5 percent in total use.
