This Article Abstract Submit a response to this article Services E-mail this article to a friend Alert me to new issues of the journal

H E A L T H  A F F A I R S
W E B  E X C L U S I V E 

3 April 2002

John Holahan and Mary Beth Pohl

ABSTRACT:

The number of uninsured Americans fell in 2000 for the second consecutive year. The reduction has been attributed to the continued expansion of employer-sponsored insurance. However, the increase in employer coverage among adults was offset by declines of other types of coverage. For children, increases in public coverage plus the growth in employer-sponsored insurance led to the reduction in the number of uninsured children. Over the longer period (1994–2000), one of great economic growth, the uninsurance rate was essentially the same at the end as at the beginning. The rate of employer-sponsored insurance increased sharply, so that more people had employer coverage. However, these increases were offset by reductions in other forms of coverage, particularly Medicaid and state-sponsored insurance and private nongroup coverage, so the overall rate of uninsurance did not change.

For the second consecutive year the Current Population Survey (CPS) revealed that the number of uninsured nonelderly Americans fell in 2000. The number of uninsured declined by 570,000; uninsured children actually declined by 700,000, while the number of uninsured adults increased by 130,000. The primary reason for the reduction in the number of uninsured at first glance appears to be the increase in employer-sponsored coverage.¹ However, expansions in employer coverage occurred for both children and adults. It was the additional expansions in Medicaid and state program coverage (particularly the State Children’s Health Insurance Program, or SCHIP) for children that led to the reduction in the uninsured rate for children. Similar reductions in uninsurance were not observed for adults where the increases in employer coverage were offset by declines in other types of coverage.

During 1994–2000, a period of great economic prosperity, the rate of uninsurance remained essentially unchanged at around 17 percent. The period’s prosperity brought extraordinary changes in income distribution. The number of low-income Americans fell by 8.2 million, while the numbers of middle- and high-income Americans rose by 2.0 million and 19.2 million, respectively. Among each of the three income groups, however, uninsurance rates actually increased. It was the movement of so many people up the income distribution, where uninsurance rates were lower, that kept the uninsurance rate essentially level over the period.

The rate of uninsurance stayed constant despite large increases in employer-sponsored insurance—from 64.3 percent in 1994 to 67.3 percent in 2000. This increase, together with population growth, resulted in a 15.9 million increase in the number of Americans with employer coverage. Although more low-income Americans (those with family incomes below 200 percent of the federal poverty level) gained coverage, the primary reason for the overall increase was, again, the large movement up the income distribution. The period’s tight labor markets allowed people to take jobs with higher earnings and a higher likelihood of employer coverage.

Why didn’t employer coverage have more of an impact on the uninsurance rate between 1994 and 2000? The primary reason is a decline in coverage in public programs such as Medicaid and other state-subsidized insurance. As the economy improved, many low-income persons left Medicaid, obtained higher-paying jobs, and enrolled in employer plans. The enactment of welfare reform in 1996 also contributed to reduced public coverage among low-income Americans. (In the late 1990s this was partly offset by the introduction of SCHIP.) Private nongroup coverage also fell; this could be attributable to problems in the individual insurance market, but it also might reflect the fact that as labor markets tightened, workers were more likely to obtain insurance through their jobs. Finally, there was also a large reduction in coverage through federal programs such as military and Medicare coverage.² Declines in all of these types of coverage may have been independent of the growth in employer coverage. Alternatively, the growth in employer coverage may have displaced other forms of coverage. In either case, Americans were no less likely to be uninsured even amid tremendous economic growth.

The 1994–2000 period actually consisted of two subperiods. Between 1994 and 1998 the increase in employer coverage was more than offset by declines in Medicaid or state coverage, other public coverage (principally military), and private nongroup coverage; the result was a larger number of uninsured persons.³ After 1998, Medicaid and state coverage stabilized, and the increase in employer coverage reduced the number of uninsured persons.⁴

In this paper we first examine changes in insurance coverage between 1999 and 2000 because of the recent attention given to the smaller number of uninsured persons in 2000. We then put the same data in the context of a longer period, the economic boom of the 1990s. We examine CPS data from 1994 through 2000, showing changes in both private and public coverage and in the uninsurance rates for adults and children.

An important change has been made in the CPS over the past two years: Namely, respondents were asked to confirm the lack of insurance for household members. Prior to the March 2000 CPS, individuals were asked if anyone in their household had any of several forms of insurance. Those who did not indicate having some form of insurance were regarded as being uninsured. In the March 2000 (and 2001) CPS, respondents were asked to verify that the lack of a positive indication of some form of coverage really meant that they were uninsured. The verification question in 2001 reduced the percentage of nonelderly uninsured persons from 17.2 percent to 15.8 percent (41.9 million to 38.4 million).⁵

The first section of this paper analyzes the verified CPS data. Thus, the rates of uninsurance presented here are lower than they have been in previous analyses. In the second section, where we explore changes in coverage that occurred between 1994 and 2000, we turn to the unverified data because they were the only data available before the March 2000 CPS. The following discussion presents CPS data categorized by insurance type, income, and other characteristics. Although CPS survey respondents were able to choose multiple types of health insurance on the surveys, their responses were classified in a hierarchy, with each respondent assigned only one type of coverage.⁶ To analyze income, respondents were grouped by their total health insurance unit income in relation to census poverty thresholds.⁷

Changes In Insurance Coverage, 1999 And 2000

Coverage by income group. Looking only at the aggregate changes in the insurance distribution from 1999 to 2000, one sees a small decline in the uninsurance rate and 570,000 fewer persons lacking insurance (Exhibit 1). For all income groups, employer coverage increased by about one percentage point; the increase in the coverage rate coupled with population growth (2.07 million) added 3.33 million to the ranks of Americans insured through an employer. The rate of coverage by Medicaid, SCHIP, and other state programs did not change, but population growth alone added 230,000 persons to Medicaid or other state coverage. Declines in private nongroup coverage somewhat offset the increase in employer coverage rates.

This general trend masks some interesting and important changes that occurred within the overall aggregate. Changes in coverage rates affected income groups differently; in addition, there were large income gains. In particular, the number of persons below 200 percent of poverty fell by 2.27 million; the number at 200–400 percent of poverty increased by 2.6 million; and the number above 400 percent of poverty increased by 1.74 million.

Insurance coverage improved among persons below 200 percent of poverty, and there were fewer low-income individuals in 2000 than in 1999 (Exhibit 1)—the strong economy seemed to move many into higher-paying jobs. Also, the rate of employer coverage and of Medicaid/state coverage also rose (together these increases were significant). Because the low-income population declined, there was no change in the number of new enrollees in Medicaid and other state programs despite the increase in the rate of coverage. As a consequence of the increased likelihood of both employer and Medicaid/state coverage, the uninsurance rate for low-income persons fell. This decline, coupled with the decline in the low-income population, meant that the number of low-income Americans without health insurance fell by 1.67 million.

The middle-income population did not fare as well. The rate of employer coverage fell, probably because, as wages in many low-skill jobs rose, many families moved above 200 percent of poverty without obtaining health insurance at the same rates as those enjoyed by persons already in this income group. Because the number of middle-income Americans increased, the number of middle-income Americans with employer coverage increased by 1.27 million. The result was an increase in the uninsurance rate and 1.18 million more uninsured persons. Thus, the increase in the number of uninsured middle-income Americans offset much of the decline in the number of uninsured low-income Americans.

The picture was strikingly different for high-income Americans. Because employer coverage increased and because there was an overall increase in the number of Americans above 400 percent of poverty, the number with employer coverage increased by 2.17 million. The uninsurance rate was unchanged, but because of population growth, there was a net decrease of 80,000 in the number of uninsured high-income Americans.

Coverage of children. Much of this general picture holds true for children, with some important differences (Exhibit 2). Children benefited from an increase in employer coverage, but again this seemed to be largely due to the income gains experienced by families with children coupled with an increased rate of employer coverage among children in high-income families. Children also benefited from increases in Medicaid/state coverage. Together these increases offset a decline in private nongroup coverage, resulting in 700,000 fewer uninsured children.

Among children with family incomes below 200 percent of poverty, Medicaid/state coverage increased by almost two percentage points or by 260,000 children, probably reflecting the growth of SCHIP, which also is believed to have raised the number of children on Medicaid.⁸ This increase would have been greater had it not been for the 820,000 decline in the number of children below 200 percent of poverty, which removed many of them from program eligibility. The number of low-income children with private nongroup coverage also fell. The net effect of these partially offsetting changes was 760,000 fewer uninsured low-income children.

A somewhat different picture emerges for middle-income children. Because of the growing economy, 480,000 more children were in the middle income bracket in 2000. While children in families that experienced income gains were no doubt more likely to have employer coverage than they were before, the rate of employer coverage for middle-income children in 2000 nonetheless fell relative to 1999. At the same time, 180,000 middle-income children gained coverage through public programs, presumably SCHIP. The increased public coverage was not enough to offset the increase in population and the decline in employer coverage; thus, the number of uninsured middle-income children rose by 180,000.

Employer coverage rates increased among children with family incomes above 400 percent of poverty—920,000 children gained access to employer coverage. This increase offset a decline in private nongroup coverage; as a result, there were 120,000 fewer uninsured high-income children.

Coverage of adults. Adults also benefited from increased employer coverage, but for them there was no expansion of public coverage comparable to SCHIP (Exhibit 2). The increase in employer coverage was offset by small declines in each of the other types of coverage (significant as a group); as a result, the uninsurance rate did not change. Population growth caused the number of uninsured adults to rise by 130,000.

The number of low-income adults without insurance fell by 910,000, primarily because of a 1.45 million decline in the number of low-income adults. None of the coverage changes was significant. As with children, middle-income adults were hurt by a lower rate of employer coverage, for reasons explained above. The uninsurance rate increased, and the number of uninsured middle-income adults rose by 1.01 million, which offset the decline in uninsurance among low-income adults.

The number of high-income adults with employer coverage rose by 1.25 million, slightly more than the increase in the number of adults above 400 percent of poverty. The increase in the number with employer coverage was offset by a small decline in the number with military and Medicare coverage; the number of uninsured high-income persons did not change.

Coverage of low-income children and adults. Exhibits 3 and 4 show insurance coverage changes for children and adults below 300 percent of poverty in more detail. The number of uninsured children below poverty fell by 320,000. Tighter labor markets seem to mean that even poor children gained employer coverage, although these rates are still very low (increasing from 20.4 percent to 21.6 percent, not significant). For these poor children, Medicaid/state coverage was stable. The number of uninsured children below poverty fell primarily because of the overall decline in the number of those children.

The picture is different for children between 100 and 199 percent of poverty. The rate of employer coverage for them was much higher than it was for those below poverty but did not change. However, Medicaid/state coverage rose significantly. There was an increase of 510,000 in the number of children with Medicaid/state coverage despite the fact that the number of children between 100 and 199 percent of poverty fell by 130,000 (not shown). This seems likely to be attributable to SCHIP, which targeted much of this income group. The large increase in Medicaid/state coverage is also associated with a drop in private non-group coverage but no significant change in employer coverage.

Among children between 200 and 299 percent of poverty, employer coverage rates were much higher than for low-income children, but they actually fell two percentage points. This decline meant that 290,000 fewer children had access to employer coverage (not shown). Some of this was offset by increased coverage of children in Medicaid/state programs, presumably SCHIP, but overall the number of uninsured middle-income children increased by 100,000.

Among adults (Exhibit 4), changes in employer coverage were similar to those for children, but adults did not benefit from public program expansions to the extent that children did. Among adults below poverty, there were no significant changes in any kind of coverage. Similarly, among adults between 100 and 199 percent of poverty, there were no significant changes in employer or private non-group coverage. There also was no change in Medicaid/state coverage. As a result, the uninsurance rate did not change, but population changes resulted in a decline of 240,000 uninsured adults at this income level. In contrast, the number of uninsured adults between 200 and 299 percent of poverty rose by 630,000. Employer coverage declined for this group, and again, adults did not benefit from public program expansions. However, it was primarily the growth in the number of adults in this income bracket that led to an increase in the number of uninsured persons.

The Broader Picture: What Happened During The Boom Of The 1990s?

Exhibits 5 and 6 show what has happened to various kinds of insurance coverage between 1994 and 2000, the entire period of the economic boom of the 1990s. This was also the period in which state and federal welfare reform reduced the number of persons receiving cash welfare benefits. Here we rely on the unverified CPS data; thus, the uninsurance rates for 2000 are higher than shown in earlier exhibits.

Surprisingly, between 1994 and 2000 the uninsurance rate for the nonelderly was nearly unchanged. As noted above, this number stayed constant despite the fact that the aggregate rate of employer coverage rose three percentage points. As employer coverage was growing throughout the entire period, Medicaid coverage fell between 1994 and 1998 and then leveled off. There were also reductions in military/Medicare and private non-group coverage. These declines offset the increase in employer coverage.

The subperiods of 1994–1998 and 1998–2000 tell different stories. In the earlier period, declines in Medicaid/state and private non-group coverage more than offset the increase in employer coverage; thus, the overall uninsurance rate rose. After 1998 the increase in employer coverage more than offset the declines in other kinds of coverage, and the uninsurance rate fell.

Between 1994 and 2000 a major change took place in the income distribution, resulting in fewer low-income Americans and many more with incomes above 400 percent of poverty (Exhibit 5). The rate of employer coverage rose for persons below 200 percent of poverty. However, it fell for middle-income Americans and stayed roughly constant for those above 400 percent of poverty. Thus, there were fewer low-income Americans, but those whose incomes stayed low were more likely to have employer coverage at the end of the period than at the beginning. As many moved up the income distribution, they were far more likely to obtain employer coverage than they had been before, but middle-income Americans as a whole were less likely to have employer coverage in 2000 than in 1994.

For those below 200 percent of poverty, the rate of Medicaid/state coverage fell two percentage points, with the entire decline occurring prior to 1998. Private nongroup coverage also fell; the result was a one-percentage-point increase in the uninsurance rate from 1994 to 2000. Thus, the rise in the rate of employer coverage was not sufficient to offset the declines in Medicaid/state and private nongroup coverage for persons below 200 percent of poverty.

For middle-income Americans, the decline in employer coverage together with reductions in military/Medicare and private nongroup coverage raised the uninsurance rate over the entire period. The uninsurance rate among persons above 400 percent of poverty also rose. This reflected declines in both employer and private nongroup coverage.

It is interesting that the uninsurance rate rose for each of the three income groups. It only remained constant overall because so many persons moved up the income distribution, where the likelihood of being uninsured was lower. A related point is that all of the growth in the uninsured over the period occurred among the nonpoor. The number of uninsured persons below 200 percent of poverty actually fell by 1.84 million, primarily because there were fewer low-income Americans.

How Many Uninsured People Will There Be In A Recession?

This analysis examines a period of great prosperity. As of this writing, the nation was in a recession, and there is concern that the number of uninsured persons will rise sharply. One study has estimated that 2.2 million more persons became uninsured in 2001.⁹ No one knows how high the unemployment rate will go; thus, it is difficult to predict how much the number of uninsured Americans will increase.

This paper has shown the important role played by the growth in employer coverage during the late 1990s. Much of this growth has been attributable to increased rates of coverage among low-income Americans coupled with income expansion. These could be reversed in an economic slowdown. Firms facing reduced demand for their products might drop health insurance. Others might increase employee contributions, causing some employees to drop their coverage. Premiums have risen sharply in the past few years, which could exacerbate the effect of an economic slowdown.

As we have seen, employer coverage has been substituted for other forms of coverage over the past several years. This will likely be reversed, although how much so will bear heavily on how much the uninsurance rate will grow. Medicaid coverage is likely to increase as more become eligible because of falling incomes. Medicaid enrollment has risen in prior recessions.¹⁰ Whether Medicaid and other state programs will respond as they have in the past is unknown. State budget shortfalls are large and may keep states from allowing enrollment to expand. In the past several years employer coverage also seems to have substituted for private nongroup coverage. Whether persons losing employer coverage will be able to buy increasingly costly individual coverage is also unknown. How these scenarios play out will eventually determine the actual increase in the number of uninsured Americans in the coming years.

The authors are grateful for the valuable comments received from Genevieve Kenney, Steve Zuckerman, and several anonymous reviewers.

NOTES

1. U.S. Census Bureau, Health Insurance Coverage: 2000 (Washington: US Government Printing Office, September 2001); and R. Pear, “Number of Uninsured Drops for Second Year,” New York Times, 28 September 2001.
2. We removed the elderly (age 65 and older) and active-duty military from the CPS analysis. Military health includes military retirees and dependents receiving health care from the Uniformed Services, TRICARE, the Civilian Health and Medical Program of the Uniformed Services (CHAMPUS), VA Health Care, or CHAMPVA.
3. J. Holahan and J. Kim, “Why Does the Number of Uninsured Americans Continue to Grow?” Health Affairs (July/Aug 2000): 188–196.
4. J. Holahan, Why Did the Number of Uninsured Fall in 1999? (Washington: Kaiser Commission on Medicaid and the Uninsured, January 2001).
5. S. Zuckerman and M.B. Pohl, “Verifying Health Insurance Coverage in the CPS: Impact on Estimates of the Uninsured over Time,” Health Policy Brief (Washington: Urban Institute, forthcoming).
6. The hierarchy is as follows: insurance through an individual’s own employer or another’s group plan, including coverage outside the household; Medicaid, SCHIP, or other state-funded health coverage; federally funded health coverage through military and veterans’ coverage and Medicare; coverage through private insurance that is not employer or group insurance; and uninsured. For verified CPS data, those classified as uninsured responded to the verification questions as being uninsured; for unverified CPS data, this is a residual category for those who did not report having any of the other types of insurance coverage over the course of the year.
7. Changes in income distribution that we report are not affected by the use of health insurance units.
8. V. Smith and E. Ellis, Medicaid Budgets in Stress: Survey Findings for State Fiscal Year 2000, 2001, and 2002 (Washington: Kaiser Commission on Medicaid and the Uninsured, October 2001).
9. Covering the Uninsured, “Two Million Americans Lost Their Health Insurance in 2001; Largest One-Year Increase in Nearly a Decade” (Press release prepared by Families USA for a Robert Wood Johnson Foundation–sponsored partnership, Covering the Uninsured, 12 February 2002). The figure of 2.2 million more uninsured persons in 2001 is calculated from Bureau of Labor Statistics data using US Census Bureau methodology (R.L. Bennefield, Who Loses Coverage and for How Long? Dynamics of Economic Well-Being: Health Insurance, 1993–1995 [Suitland, Md.: US Census Bureau, Economics and Statistics Administration, August 1998]) and is based on work from researchers at the Henry J. Kaiser Family Foundation and Massachusetts Institute of Technology.
10. Bowen Garrett, Urban Institute, unpublished estimates, 2001; and J. Holahan and B. Garrett, “Rising Unemployment and Medicaid,” Health Policy Online, no. 1 (Washington: Urban Institute, 16 October 2001).

John Holahan is director of the Health Policy Center at the Urban Institute in Washington, D.C. Mary Beth Pohl is a research assistant there.

©2002 Project HOPE–The People-to-People Health Foundation, Inc.