Statement of
Douglas Holtz-Eakin
Director

Health Care Spending and the Uninsured

before the
Committee on Health, Education,
Labor, and Pensions
United States Senate

January 28, 2004

Mr. Chairman and Members of the Committee, I appreciate the opportunity to be here today to discuss the characteristics of people without health insurance and the factors that contribute to the growth of health care expenditures. While more than 240 million people in the United States have health insurance today through a variety of private and public sources, millions of others do not have such coverage; and the percentage of Americans who are uninsured has risen in each of the last two years for which information is available. At the same time, health care spending has continued to rise.

In my testimony today, I will discuss some important characteristics of the uninsured population that have received relatively little attention but that have important implications for federal policies to expand insurance coverage. I will also discuss factors contributing to increases in health care spending and will describe the relationship between health care costs and insurance coverage.
 

Characteristics of the Uninsured Population

In recent years, it has been frequently stated that about 40 million Americans lack health insurance coverage. That estimate, by itself, presents an incomplete and potentially misleading picture of the uninsured population. The uninsured population is constantly changing as people gain coverage and lose coverage. Furthermore, people vary greatly in the length of time that they remain uninsured. Some people are uninsured for long periods of time, but more are uninsured for shorter periods.

Policies aimed at increasing insurance coverage will be more effective if designed in light of the dynamic nature of the uninsured population as well as the distinction between the short-term and long-term uninsured. For people with short spells of being uninsured, policies might have the goal of filling the temporary gap in coverage or of preventing such a gap from occurring. For people with longer periods without insurance, policies might seek to provide or facilitate an ongoing source of coverage.

There are several alternative measures of the number of people who lack insurance coverage. One describes those people who do not have coverage for a sustained period (say, one year)--the long-term uninsured. Alternatively, another identifies how many individuals have experienced any episode of uninsurance during a particular period. Finally, the most commonly used measure (a mixture of those two others) counts the number of individuals without insurance on any particular day or week. Those different approaches yield different numbers because of the continual movement of people into and out of the uninsured population. The Congressional Budget Office's (CBO's) recent analysis⁽¹⁾ found that in 1998:

• Between 21 million and 31 million people were uninsured all year;
  
  
• At any point in time during the year, about 40 million people were uninsured; and
  
  
• Nearly 60 million people were uninsured at some point during the year (see Figure 1).

Figure 1.

Estimated Number of Nonelderly People Without Health Insurance in 1998

Source: Source: Congressional Budget Office.

Note: Note: The Survey of Income and Program Participation is conducted by the Census Bureau. The Medical Expenditure Panel Survey is conducted by the Agency for Healthcare Research and Quality. The National Health Interview Survey is sponsored by the Centers for Disease Control and Prevention, which reports only the point-in-time estimate.

CBO conducted the analysis for 1998 because that was the most recent year for which suitable data were available to construct all three measures. More recent analyses by researchers at the Agency for Healthcare Research and Quality indicate that those three measures of the uninsured remained fairly stable in the subsequent period from 1998 to 2001.⁽²⁾

About 30 percent of Americans under age 65 who become uninsured in a given year remain so for more than 12 months, while nearly half obtain coverage within four months (see Figure 2).⁽³⁾ Those estimates were obtained by CBO using data from the Census Bureau's Survey of Income and Program Participation for 1996 through 1999. They are very similar to the findings of previous studies that have examined earlier time periods.
 

Figure 2.

Distribution of Uninsured Spells Among Nonelderly People in a Given Year, by Duration

Source: Congressional Budget Office based on data from the 1996 panel of the Survey of Income and Program Participation, which followed respondents over a period of 41 months (from March 1996 through July 1999).

Note: These estimates are based on uninsured spells that began between July 1996 and June 1997.

People with less education, those with low income, and Hispanics are more likely than others to be uninsured (see Table 1). They are also somewhat more likely to remain uninsured for long periods. For example, people in families in which no one attended college account for 64 percent of uninsured spells of more than 12 months but only 49 percent of uninsured spells that end within four months (see Table 2). That difference probably reflects, at least in part, the fact that people who did not attend college are less likely than others to have access to employment-based insurance.

Table 1.
Nonelderly People Without Health Insurance in 1998, by Selected Characteristics
(In percent)
		Nonelderly People
Characteristic		Uninsured at Any Time During the Year		Uninsured All Year		Distribution of the Population Uninsured All Year
Age
	Less than 19	26.8		7.3		24.9
	19-24	41.9		14.4		13.7
	25-34	31.1		12.3		21.9
	35-44	20.2		9.3		19.7
	45-54	15.1		7.6		12.6
	55-64	14.0		6.7		7.2
Race/Ethnicity
	White, Non-Hispanic	18.4		6.3		48.4
	Black, Non-Hispanic	33.4		10.7		15.3
	Hispanic	47.4		22.5		30.8
	Other	31.1		10.9		5.5
Family Income Relative to the Poverty Levela
	Less than 200 percent	47.9		19.5		74.9
	200 percent to 399 percent	17.4		5.3		19.8
	400 percent or more	6.0		1.6		5.3
Educationa, b
	No high school diploma	50.4		24.6		28.4
	High school graduate	33.1		12.7		36.4
	Some college coursework	22.1		7.3		26.6
	Bachelor's degree or higher	9.9		2.6		8.7
Family Employment Statusa
	At least one full-time worker all year	15.0		5.9		42.9
	Part-time or part-year work only	46.1		16.1		46.6
	No work	32.8		13.1		10.6
Health Statusc
	Excellent	23.7		8.9		28.8
	Very good	25.1		9.3		32.8
	Good	24.6		9.1		24.5
	Fair	25.1		8.7		8.9
	Poor	25.3		10.3		5.1
Memorandum:
Total Nonelderly Population		24.5		9.1		100.0
Source: Congressional Budget Office based on an analysis of data from the 1996 panel of the Survey of Income and Program Participation.
a. For family-level variables, families are defined as health insurance eligibility units, which are composed of individuals who could be covered as a family under most private health insurance plans.
b. Education measures the highest education level among the adults in the family.
c. Information on health status was collected only for survey respondents who were at least 15 years of age.

Table 2.
Comparison of the Characteristics of Nonelderly People with Short Uninsured Spells and Long Uninsured Spells
(In percent)
		Duration of Uninsured Spell
Characteristic		Four Months or Less		More Than 12 Months
Total		100.0		100.0
Agea
	Children	47.3		37.5
	Adults	52.7		62.5
Race/Ethnicity
	White, Non-Hispanic	56.7		48.8
	Black, Non-Hispanic	19.7		18.2
	Hispanic	18.4		27.6
	Other	5.2		5.4
Family Income Relative to the Poverty Levelb,c
	Less than 200 percent	61.6		77.0
	200 percent to 399 percent	26.7		21.0
	400 percent or more	11.7		7.0
Educationa,c
	No high school diploma	17.8		26.6
	High school graduate only	31.0		37.6
	Some college	35.5		26.8
	Bachelor's degree or higher	15.6		9.0
Source: Congressional Budget Office based on an analysis of data from the 1996 panel of the Survey of Income and Program Participation.
Note: Estimates in this table are based on uninsured spells that began between July 1996 and June 1997. People with uninsured spells lasting more than 24 months had very similar characteristics to those of people with uninsured spells lasting more than 12 months.
a. Age and education were measured as of the first month of the uninsured spell. Education measures the highest education level among the adults in the family.
b. Family income relative to the poverty level was computed as the mean over the four-month period before the beginning of the uninsured spell.
c. For family-level variables, families are defined as health insurance eligibility units, which are composed of individuals who could be covered as a family under most private health insurance plans.

Adults are somewhat more likely than children to remain uninsured for long periods. The availability of Medicaid coverage may explain some of that discrepancy: coverage is available to many children in low-income families, but the majority of low-income adults are not eligible for the program. In addition, evidence suggests that single adults without children may be less inclined to seek insurance, on average, than other adults are, which may lead them to experience long spells without insurance.

The vast majority of the uninsured are in working families. Some 43 percent of the people who were uninsured all year in 1998 were in families in which at least one person worked full time all year, and 47 percent were in families in which at least one person worked part time or for a portion of the year (see Table 1, column 3). Studies have found that over three-quarters of uninsured workers are not offered insurance by their employer. Low-wage workers are less likely to be offered insurance by their employer and less likely to accept it if it is offered.

According to the Census Bureau's Current Population Survey, the number and percentage of Americans who are uninsured increased in 2001 and 2002, after falling the previous two years.⁽⁴⁾ From 2000 to 2002, the number of nonelderly people who were uninsured increased from 39.4 million to 43.3 million, or from 16.1 percent of the nonelderly population to 17.3 percent.⁽⁵⁾ That rise in uninsurance rates was associated with a drop in the percentage of nonelderly people covered by employment-based insurance (from 67.1 percent to 64.2 percent), which was partially offset by an increase in the percentage covered by Medicaid and the State Children's Health Insurance Program (from 10.4 percent to 11.9 percent). Those changes in coverage rates, while significant, are smaller than those that occurred in the early 1990s, when the share of the nonelderly population covered by employment-based insurance fell by more than 5 percentage points.⁽⁶⁾

Just how much of the recent increase in uninsurance rates results from rising costs for premiums, rising unemployment, or other factors is unknown. Clearly, though, losing a job may put a worker's employment-based health insurance at risk. In preliminary results from a recent analysis, CBO found that health insurance coverage rates declined significantly among people who received unemployment insurance (UI) benefits for at least four consecutive months in 2001 or early 2002. Some 82 percent of such workers had health insurance coverage (from any source) before they began receiving UI benefits, but only 58 percent had coverage by the final month of those benefits. Federal legislation (the Consolidated Omnibus Budget Reconciliation Act of 1985, known as COBRA) requires firms with 20 or more employees to continue offering health coverage to workers who separate from their firm. However, firms may charge former employees up to 102 percent of the full (group) premiums for that coverage. Therefore, unemployed workers may face a large increase in their out-of-pocket premiums under COBRA. The reduction in coverage estimated for recipients of unemployment insurance probably stems, in part, from many of those people opting not to purchase coverage under that law.

Rising health insurance premiums resulting from the recent large increases in health care spending overall offer a plausible explanation for at least some of the reduction in coverage. To explore that issue, I will now discuss the magnitude and causes of the spending increases and then turn to what is known about the relationship between health care spending and insurance coverage.
 

Historical and Recent Trends in Health Care Spending

Health care is a large and growing sector of the economy. The United States spent $1.6 trillion on health care in 2002, an amount more than five times as great in real (inflation-adjusted) terms as that spent in 1970. Per capita spending increased from about $1,300 in 1970 (in 2002 dollars) to about $5,450 in 2002, for an average rate of real growth of 4.5 percent per year (see Table 3). The economy as a whole has grown over that period as well, but not as quickly, with the result that health spending as a percentage of gross domestic product (GDP) has more than doubled, from 7.0 percent in 1970 to 14.9 percent in 2002. The mid-1990s saw a brief slowdown in real spending growth per capita (the rate was 2.6 percent per year from 1992 to 1997), but higher rates of growth have since returned: from 1997 to 2002, real per capita health care spending grew at an average annual rate of 4.6 percent--which is similar to its approximate long-term rate of increase.

Table 3.
Increases in Health Care Spending Between 1970 and 2002
	1970		2002
Total Health Care Spending per Capita (2002 dollars)	1,313		5,449
Total Health Care Spending as a Percentage of GDP	7.0		14.9
Source: Centers for Medicare and Medicaid Services, Office of the Actuary, National Health Statistics Group.
Note: Figures were adjusted for inflation using the gross domestic product (GDP) implicit price deflator.

Recent growth in real spending on prescription drugs has been especially rapid--at more than 14 percent per year on average from 1997 to 2002, making it the fastest growing category of health spending during the period. Despite the recent rapid increase in spending for prescription drugs, they currently account for only about 10 percent of all national health expenditures. That relatively small (but growing) share of expenditures should be kept in mind when evaluating whether drugs are a major driver of increasing costs.

Federal spending on health care, principally Medicare and Medicaid, is subject to the same cost pressures facing the system as a whole. Total federal health spending as a percentage of GDP was 1.7 in 1970 and 4.8 in 2002. If the recent rate of growth in spending persists, federal outlays on health care will continue to increase as a proportion of GDP. Projections of future spending on Medicare and Medicaid depend critically on the assumed rate of "excess cost growth."⁽⁷⁾ Under an assumed rate of excess cost growth of 2.5 percent (a rate that is slightly lower than the long-term historical average), federal spending on Medicare and Medicaid would climb to 21.3 percent of GDP in 2050 (see Figure 3).⁽⁸⁾ Under a more optimistic assumption of 1 percent excess cost growth, federal Medicare and Medicaid spending would reach 11.5 percent of GDP in 2050. To put those estimates in perspective, the entire federal budget currently consumes 20 percent of GDP.
 

Figure 3.

Total Federal Spending for Medicare and Medicaid Under Different Assumptions About Excess Cost Growth

Source: Congressional Budget Office.

Note: Excess cost growth is the extent to which the growth of spending per enrollee in those programs exceeds the growth of per capita gross domestic product (after accounting for changes in the age and sex composition of the population).

 

Factors Behind the Continuing Growth in Health Care Spending

Most analysts agree that the perennial increases in health care spending that have occurred over recent decades are associated with the diffusion of new medical technologies, or as one analyst has described it, "the enhanced capabilities of medicine."⁽⁹⁾ Recent advances, including pharmaceutical innovations, have made available to patients and physicians a wealth of new medical therapies, many unheard of in even the relatively recent past. The economic incentives for innovation and the development, deployment, and utilization of new technologies in the U.S. health care system has led generally to higher levels of spending. Some medical advances permit the treatment of previously untreatable conditions, introducing new categories of spending. Others, relative to older modes of treatment, improve medical outcomes at added cost, expanding existing spending.

It is occasionally suggested that advances in technology can lead to reduced spending, and that may be the case in some instances. Vaccinations, for example, may sometimes offer the potential for savings, and certain types of preventive medical care may help some patients avoid costly acute care hospitalizations. But, overall, examples of new therapies for which long-term savings have been clearly demonstrated are few. Improvements in medical care that decrease mortality by helping patients avoid or survive acute health problems paradoxically increase overall spending on health care, as those (surviving) patients live to utilize health services through old age.

Even when a particular service becomes cheaper to provide over time, higher aggregate spending can still result as practice patterns emerge and the service is used with greater frequency. Comparing increased expenditures on computers and information technology with those on health care is instructive. As technological innovations permitted profitable computer processing at a fraction of the previous cost, total spending on computers did not decrease--it skyrocketed, as more consumers made more intensive use of what became available. Why do few people regard increasing spending on information technology as a problem requiring a remedy? Let me suggest that the reason is that the market for information technology works the way a market is intended to function: businesses and consumers weigh alternatives and face the full costs of what they use. In health care, two factors combine to produce a different result: payments made by third parties typically buffer patients from the full cost of the medical services they use, and the inherent complexity of medical practice forces patients to rely on the judgment of providers who, depending on the reimbursement system, may have an incentive to provide more care (under a fee-for-service arrangement) or less care (under capitation).

Other factors have also contributed to increases in health spending. One obvious example is the aging of the population. Among adults, medical spending generally increases with age. As the number of elderly people rises with the aging of the very large baby-boom generation, health spending will naturally grow. However, over the past half century, aging has played a relatively minor role in the very large increases in spending that have occurred.

Other contributing factors include the growth in personal income over time and the spread of health plans over recent decades. Because medical care is a desired service, people naturally purchase more of it as their income increases. And health insurance, as economists are fond of pointing out, effectively drives down the cost of care from the consumer's perspective, resulting in a higher quantity demanded than would otherwise be the case. But the best estimates of the effects of those two factors suggest that they, too, fail to explain much of the surge in spending in recent decades.

Claims are often heard about unwarranted expenditures. One example is so-called defensive medicine, which refers to medical tests or procedures of little or no clinical value that are ordered by physicians solely in the interest of avoiding lawsuits. Another example is what some people term physician-induced demand, which refers to spending that is brought about at least in part by providers' desire to augment their own income. While the magnitude of spending associated with such practices has been the subject of considerable debate, those factors do not appear to explain much of the growth in spending.⁽¹⁰⁾

What I have presented here is a simple discussion of a complex issue, and I do not intend for it to represent an exhaustive or definitive review of the subject. The association between technological change and rising medical expenditures is the manifestation of a complex system of economic incentives that need to be examined in more detail. A greater understanding of the possible role of the third-party payment system in creating incentives for innovation and the diffusion of technologies, for example, could inform public policy aimed at addressing the continuing increases in spending. At the same time, policymakers could choose to spend more in light of the quality enhancements resulting from the remarkable medical advances that have been made in recent years. The point to emphasize (and about which there is general consensus) is that the way new medical technologies have been adopted and utilized has generally led to more health spending over time; that factor lies at the heart of increasing expenditures for health care. In the absence of a change in overall incentives, those pressures can be expected to continue.
 

Rising Spending and Health Insurance Premiums

Health insurance premiums, like total spending, have undergone increased growth in recent years following a brief slowdown. Estimates from one survey of private firms show that growth in employer-based premiums has risen every year since 1996, exceeding 10 percent in real terms from 2002 to 2003.⁽¹¹⁾ According to that survey, the average annual premium for an employer-based policy for an individual is now about $3,400, and for a family, more than $9,000. Recent declines in employer-based coverage may be related to increases in premiums, though the relationship between premiums and the prevalence of coverage is not necessarily a simple one. While it is natural to believe that the purchases of any good tend to decline when its price goes up, the case of health insurance is complicated by the fact that a general upward trend in the cost of medical services can make insurance more appealing, because in the absence of insurance, covering potentially costly medical needs is more difficult. Furthermore, changes in premiums can present an incomplete picture if plan characteristics change simultaneously (and there is evidence that average deductibles have recently increased in certain types of plans).

Employee contributions for employer-based health insurance have remained a constant proportion of total premium costs in recent years; because total premiums are rising, however, the dollar value of employee contributions is rising as well. Perhaps related to the growth in employee contributions is a reported decline in the rates at which employees take up the offer of employer-based coverage. Three factors may drive changes in the number of covered workers: the number of firms offering health benefits, the proportion of workers eligible for health benefits among firms offering coverage (part-time workers generally are not eligible), and the rate at which workers accept coverage (if it is offered by their employer). Recent analysis shows that as much as three-fifths of the recent decrease in employer-based coverage is attributable to workers' declining to enroll.⁽¹²⁾
 

Conclusions

In sum, I have stressed that the commonly cited estimates of the number of people who lack health insurance present an incomplete picture of the uninsured population. That population is dynamic, and individuals vary greatly in the length of time that they remain uninsured. The varying characteristics of the uninsured should be kept in mind when developing policies to expand insurance coverage. I have also discussed the relationship between advancing medical technologies and increasing health care expenditures and noted that the correlation between technological change and expenditure growth needs to be better understood. Both federal spending and private spending on health care are likely to continue to increase in the immediate future and to constitute a greater share of GDP.