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  T R A C K I N G : 
T R E N D S
W E B  E X C L U S I V E   

28 January 2004  
  

Changes In Health Insurance Coverage 
During The Economic Downturn: 2000–2002

Low-income Americans, particularly males and nonparents, fared the worst,
as gains in public programs failed to offset lost employer-sponsored coverage.

By John Holahan and Marie Wang

ABSTRACT:

Using Current Population Survey data from 2000–2002, this paper documents the changes that led the uninsured population to grow by 3.8 million during that time period. All of the increase in the uninsured occurred among adults, and two-thirds was among low-income adults. The extent to which the loss of employer coverage resulted in people becoming uninsured depended on their access to public programs: Children were more likely than adults to gain public coverage; women more likely than men; and parents more likely than nonparents. Middle- and higher-income Americans were also affected because many lost income and because rates of employer coverage were lower.

On 30 September 2003 the U.S. Census Bureau released its annual report on insurance coverage, showing that the number of nonelderly Americans without health insurance had increased by 2.4 million in 2002—the largest single increase in more than a decade.¹ All of the increase in the uninsured occurred among adults; the number of uninsured children did not change. The Census Bureau report indicated that the primary culprit was the drop in employer-sponsored insurance (ESI). This was the second consecutive year in which the number of people with such coverage declined. The decline reversed several years of rising ESI rates. In fact, between 1998 and 2000 a sizable increase in ESI rates (6.4 million) had contributed to a decline of 2.1 million in the number of uninsured Americans.

In this paper we explore the new Census Bureau data in greater detail. We focus on how changes in coverage rates for people at different income levels interact with changes in income distribution and population growth to affect the number of uninsured people. We present data on 2000–2002, a period in which there was both an economic downturn and a sharp increase in health care costs, including health insurance premiums.² The period also saw a dramatic increase in public coverage, which offset the decline in ESI for some groups but not for others.³

Data And Methods

In this paper we use data from the 2001, 2002, and 2003 March Supplements to the Current Population Survey (now called the Annual Social and Economic Supplement). Unlike the Census report, we use the income of the health insurance unit rather than household income. A health insurance unit includes members of a nuclear family who can be covered under one health insurance policy (that is, policyholder, spouse, children under age nineteen, and full-time students under age twenty-three). Counting the income of all members of the household, as the Census Bureau does, can overstate income because it can include incomes of all relatives and unrelated people living together.⁴ The income of a health insurance unit more accurately reflects a person’s available income when purchasing private insurance or determining eligibility for public programs.⁵

We present data for three income groups using poverty thresholds: less than 200 percent of the federal poverty level, 200–399 percent, and 400 percent and higher.⁶ An advantage of using poverty thresholds is that they adjust for family size and inflation. The household income categories used by the Census Bureau do not adjust for either.⁷

Changes In Coverage

An overview. For the nonelderly population as a whole, the number of uninsured people increased by 1.4 million in 2001, and another 2.4 million in 2002 (Exhibit 1). Declining ESI rates in both years reduced the number of people with ESI by 2.8 million. At the same time, the number of people in Medicaid or other state programs (including the State Children’s Health Insurance Program, or SCHIP) rose each year, adding 3.7 million to public coverage. There were also small increases in Medicare and private nongroup coverage.⁸ Because the coverage expansions were not sufficient to offset the ESI decline and growth in the population, the number of uninsured people rose by 3.8 million between 2000 and 2002.

The changes in coverage are different for children and adults. The rate of ESI for children declined in each year, with a cumulative reduction of 2.5 percentage points and an overall decline of 1.3 million children covered by employers. Medicaid and SCHIP expanded dramatically (2.9 percentage points), resulting in an increase of 2.4 million covered children between 2000 and 2002. The increase in Medicaid and SCHIP was sufficient to offset the drop in ESI as well as the small increase in the number of children (growth of only 1.0 million during the two years). As a result, there was essentially no change in the number of uninsured children.

Among adults, the ESI rate dropped sharply in both years. Between 2000 and 2002 it fell 2.8 percentage points, resulting in a decline of 1.5 million adults with employer coverage. Medicaid and state programs increased by only 0.6 percentage points over the two-year period. There were small increases in other forms of coverage, but these were not enough to offset the decline in ESI. As a result, the uninsurance rate rose in both years (a cumulative increase of 1.7 percentage points). The total number of adults grew much faster (an increase of 4.8 million) than the total number of children (1.0 million). Because of both the increase in the uninsurance rate and growth in the population, the number of uninsured adults increased by 1.5 million in 2001 and another 2.3 million in 2002 (Exhibit 1).

Changes by income level. Exhibit 2 shows changes in the population and in the distribution of income between 2000 and 2002. Change in the size of the nonelderly population is the net effect of births and net immigration less deaths and the number turning age sixty-five; change in income distribution reflects not only the impact of these changes but also movements within the existing population. Most of the population changes in these two years resulted in an increase in the number of low-income people, which reflects shifts in the income distribution because of the economic downturn.

Of the 5.8 million increase in the nonelderly population, 5.0 million or 86 percent of the increase was among those whose incomes were below 200 percent of poverty. Because low-income people are much more likely than people with higher incomes to be uninsured, this shift in income distribution had a strong impact on the likelihood of coverage.

Children. Exhibit 3 examines changes in coverage by income level among children and adults separately. The largest drop in ESI for children was for those in families with incomes below 200 percent of poverty. For low-income children, the rate of ESI fell by 3.5 percentage points, a sizable drop given the low level of employer coverage for low-income children in 2000 (36.1 percent). This was more than offset by an increase of 4.6 percentage points in Medicaid and SCHIP coverage rates. The result was actually a slight decline in the uninsurance rate among low-income children.

For children in families with incomes at 200–399 percent of poverty, the drop in ESI was somewhat smaller (1.6 percentage points). This was also offset by an increase (2.0 percentage points) in Medicaid and SCHIP coverage rates.⁹ Children in families above 400 percent of poverty also lost ESI, but increases in private nongroup and public coverage were sufficient to avoid an increase in the uninsurance rate.

Adults. Reductions in ESI for adults were only partially offset by increases in enrollment in public programs; as a result, the number of uninsured adults increased. Rates of ESI among low-income adults dropped sharply (2.6 percentage points). Medicaid and state coverage increased but by only 0.7 percentage points, much less of an increase than was observed for low-income children. The result was an increase of 1.8 percentage points in the uninsurance rate for low-income adults. As noted earlier, the number of low-income adults had risen sharply. More low-income adults combined with a higher uninsurance rate meant that 2.5 million more adults joined the ranks of the uninsured. About two-thirds of this increase occurred among adults with incomes below 200 percent of poverty.

Adults with incomes at 200–399 percent of poverty also fared poorly. Their rate of ESI fell 2.5 percentage points. There was a small increase in Medicaid and state coverage, as well as in Tricare and Medicare. These increases were not sufficient to offset the drop in ESI rates, and the uninsurance rate increased by 1.7 percentage points (900,000 adults).

Even adults with incomes above 400 percent of poverty were affected by a drop in ESI rates. There was a small increase in private nongroup coverage, but the uninsurance rate increased, which left 500,000 more adults in this income group lacking coverage.

Changes in coverage for adults, by age. Most of the increase in the uninsured was among young (ages 19–34) and middle-aged (ages 35–54) adults. The near-elderly (ages 55– 64) fared relatively well, as most retained insurance coverage (Exhibit 4). The number of young and middle-aged adults increased by 1.1 million and 1.0 million, respectively. In contrast, the number of near-elderly people grew much faster (2.7 million); the rate of growth in this group was much faster than in the other two age groups, reflecting the aging of the baby-boom generation (Exhibit 5).

There were also important differences in the changes in the income distribution within each of these age categories, which have implications for changes in coverage. The number of low-income young adults grew by two million, while the number in the two higher income groups declined. Similarly, the number of middle-aged adults in the lowest income group grew by 1.8 million, again much faster than the rate of growth in the overall population, while the number in the higher income groups fell. In contrast, there were large increases in near-elderly people in the highest income group: Two-thirds of the increase in the near-elderly occurred among those with incomes of at least 400 percent of poverty.

Young adults experienced the sharpest drop in the rate of ESI (Exhibit 4). Increased coverage from Medicaid and increases in private nongroup coverage were not enough to offset the decline in ESI. As a result, the uninsurance rate among young adults rose from 24.7 percent to 27.5 percent, and two million more young adults were uninsured. Most of this increase occurred among low-income young adults (data not shown). The share of low-income young adults with ESI fell dramatically, from 34.0 percent in 2000 to 30.5 percent in 2002. Some of this was offset by an increase in Medicaid/state coverage; nonetheless, the number of uninsured, low-income young adults rose 2.2 percentage points, or 1.4 million people. Young adults in the two higher income groups also saw significant decreases in ESI. As a result, the uninsurance rate rose for higher-income young adults (adding 500,000 more uninsured young adults).

Middle-aged adults also fared poorly (Exhibit 4). Again, there were significant declines in ESI, which were partially offset by increases in Medicaid/state coverage. The uninsurance rate rose from 14.1 percent to 15.9 percent, and 1.7 million more middle-aged adults were without insurance.

Most of the decline in coverage in this group occurred among those below 200 percent of poverty, as among young adults. Because of the decline in the ESI rate (from 34.7 percent to 32.5 percent) and the increased number of people in this group, the number of uninsured people in this age and income group rose 1.1 million. Middle- and higher-income adults in this age range also saw declines in ESI. For the group at 200–399 percent of poverty, this resulted in a statistically significant increase in the uninsurance rate and another 500,000 uninsured people.

In contrast, the near-elderly fared well, experiencing increases in income and no change in the uninsurance rate. There were no statistically significant declines in ESI for any of the three income groups and no change in the uninsurance rate.

In sum, of the 3.9 million more adults who lacked insurance in this two-year period, 2.4 million had incomes below 200 percent of poverty. Another 800,000 had incomes at 200– 399 percent of poverty. More than half of the 3.9 million newly uninsured people were young adults; only 200,000 were near-elderly.

Changes in coverage by sex and parental status. Parents are somewhat more likely than nonparents to be covered by public programs both because of the historic link to public assistance programs and because several states have used new authority to extend coverage to parents.¹⁰ A smaller number of states have extended coverage to nonparents. Because public assistance programs have been directed at single parents, who are predominantly female, women are more likely to be covered than men.

Both parents and nonparents experienced declines in ESI (Exhibit 6).¹¹ Parents were more likely than nonparents to have increased enrollment in public programs. As a result, the increase in the uninsurance rate was somewhat less for parents than for nonparents.¹² Among nonparents, 2.8 million more were uninsured, compared with 1.0 million parents. There was a much greater increase between 2000 and 2002 in the number of nonparents (4.1 million) than parents (0.7 million), and there are many more nonparents (107.2 million) than parents (66.4 million); as a result, almost three-quarters of the increase in the uninsured occurred among nonparents.

We also found that men had a somewhat greater likelihood of losing ESI and less of an increase in coverage through Medicaid/state programs. As a result, the increase in uninsurance among men was much greater than among women. Because of these factors, 2.5 million more men were without health insurance, compared with 1.3 million more women.

Changes in coverage by family work status. The decline in ESI came about, at least in part, because the population shifted toward families with fewer workers. The CPS data show that full-time employment declined and part-time employment increased over this period. Between 2000 and 2002 the number of adults working full time for the full year fell by 900,000, while the number working full time for part of the year increased by 300,000 and the number working part time increased by 1.1 million. The number of people in families with two full-time workers fell by 2.7 million, while the number of people in families with one full-time worker increased by 3.5 million (Exhibit 7). The number of people living in families with only part-time workers or with no worker also rose. In general, the likelihood of having health insurance is highest in households with two full-time workers and lowest in households with no worker.

Within each type of family (except those with only part-time workers), the likelihood of having ESI also declined. As shown earlier, the overall rate of ESI fell by 2.7 percentage points. But in none of the four family-work-status categories did the rate of ESI decline by that much. Thus, while the likelihood of having ESI declined within each category, the large overall decline must be attributable in part to the lower likelihood of full-time employment. The decline in ESI rates for full-time workers suggests that workers were less likely to be offered coverage or to take it up if it was offered.

Reduced employment also had a noticeable effect on the number of uninsured people. There was an increase of 1.6 million uninsured Americans in families with one full-time worker. But the remainder (2.2 million of the increase in uninsured Americans) was among people living in families with only part-time workers or with no worker.

Changes in coverage by firm size. The CPS data show that the number of workers increased slightly (0.5 million) between 2000 and 2002. This seems to conflict with Bureau of Labor Statistics (BLS) data showing a decline of about 1.45 million workers.¹³ BLS data show changes in employment in establishments, while the CPS data include those who report self-employment. As a result, the CPS data show a small increase in the number of workers, even though at the same time they also indicate a reduction in the employment rate (the share of the population age sixteen and older who are employed) and an increase in the unemployment rate (the share of those in the labor force who are not able to find work).

Consistent with earlier data, workers experienced a decline in ESI and a small increase in Medicaid/state coverage (Exhibit 8). The uninsurance rate among workers rose 1.5 percentage points (2.2 million more workers without coverage).

The number of people working in large firms (1,000 or more workers) declined by 1.8 million, while the number of people working in small firms (up to twenty-four workers) rose 2.1 million. Employment in midsize firms (25–999) stayed roughly constant (an 0.2 million increase). The increase in small-firm employment reflects in part an increase in self-employment (94 percent of those who report self-employment reported being in small firms). Many workers who lost jobs in large firms might have either started their own firms or reported being self-employed. Since the likelihood of having health insurance declines with firm size, a shift of workers from large firms to small firms or self-employment is likely to result in a decline in the ESI rate. That, in fact, occurred.

There were small declines in the ESI rate in both large and midsize firms (Exhibit 8). In large firms the rate fell from 83.7 percent to 82.6 percent; in midsize firms, from 78.2 percent to 77.3 percent. The decline in the rate of ESI was much higher in small firms, from 56.9 percent to 53.6 percent. We conclude from these data that the ESI rate declined for two reasons: (1) Workers moved to employment arrangements that were less likely to provide health insurance, and (2) the likelihood of having health coverage fell in each firm-size category.

Most of the increase in the number of uninsured workers occurred among those working in small firms (Exhibit 8). Not only were more people working in small firms, but the likelihood of being uninsured increased by 2.1 percentage points. Thus, 1.5 million of the 2.2 million increase in the number of uninsured workers were those working in small firms; there was an increase of 0.3 million among workers in midsize firms, and the number of uninsured workers in large firms grew by 0.4 million.

Exhibit 8 also shows coverage of workers by industry. We divided industries into those with high and low ESI rates, with rates of 80 percent or more considered high. Industries with high ESI rates included finance and manufacturing; those with low rates included construction, most service industries, and wholesale and retail trade. The results tended to be similar to those for firm size, perhaps because high-ESI industries tend to be larger firms. The results showed a decline of 2.7 million workers in high-ESI industries (mostly manufacturing) and an increase of 3.2 million in low-ESI industries. Both high- and low-ESI industries had reductions in ESI rates and increases in uninsurance rates. Because the drop in ESI was larger within low-ESI industries and because these industries added workers, most of the increase in the uninsured (2.0 million of the 2.2 million) occurred among workers in low-ESI industries.

Changes in coverage by region. We also examined the changes in coverage that occurred during 2000–2002 across regions (data not shown). All regions experienced a decline in ESI, and all regions, except the West, had significant increases in Medicaid/state coverage. The largest effects were in the South. The ESI rate in the South fell by 3.4 percentage points, and the uninsurance rate increased from 18.3 percent to 19.9 percent. Because the South also experienced the largest increase in population growth, the number of uninsured people in this region increased by 1.8 million—almost half of the overall increase. About two-thirds of the increase in uninsurance in the South occurred among those below 200 percent of poverty.

Discussion

The number of uninsured Americans increased by 3.8 million between 2000 and 2002. The primary factor was the widespread decline in employer-sponsored insurance. The effects were felt most acutely by low-income Americans, but even high-income Americans had lower rates of ESI in 2002 than in 2000. Reductions in ESI were attributable to declines in employment; shifts of employment from large to small firms (or self-employment) and from high- to low-ESI industries; and the rising cost of health care, which was likely to have affected employer offer rates, take-up rates, or both. The predominant way in which Americans have health insurance coverage is through employers, but in the past two years (2000–2002) we have seen that this coverage is quite vulnerable to economic fluctuations.

The increase in uninsurance that resulted from the decline in ESI disproportionately affected low-income Americans, who experienced the sharpest drop in ESI rates. The increases in uninsurance were also disproportionately high among younger adults, men, and people living in the South. Whether the loss in ESI resulted in losing coverage depended on access to public programs. Thus, children fared far better than adults, women fared better than men, and parents fared better than nonparents. The significant increase in public coverage that occurred in 2000–2002 contributed to sharp increases in both federal and state Medicaid spending. But without the expansion of public coverage, the increases in the number of uninsured people would have been much greater than those we have reported here.

Middle- and higher-income Americans were not immune from feeling the effects. People with incomes at or above 200 percent of poverty saw reductions in the rate of ESI and increased uninsurance. About one-third of the increase in the uninsured occurred among people at these income levels. They also were affected in another way: Job losses and reduced incomes caused many to move below 200 percent of poverty. But despite the increase in the number of uninsured Americans in higher income groups, 64 percent of the uninsured in 2002 still had incomes below 200 percent of poverty.

Whether the number of uninsured people continues to increase will depend on the same factors that affected coverage in 2000–2002. The key questions are, first, whether ESI will continue to decline in the face of slow economic growth and rising health care premiums; and, second, whether Medicaid and SCHIP can continue to increase enrollment in the face of state budget problems.

If the ESI rate stabilizes or rises because job growth results in expanding the offering of health benefits and if coverage through Medicaid and state programs increases or even remains constant, then the number of uninsured people will fall as it did between 1998 and 2000. But if the ESI rate continues to fall because of slow job growth, a continued shift from high- to low-ESI industries and from large firms to small firms or self-employment, or premium growth exceeding wage growth, the number of uninsured people is likely to continue to rise.

The recent growth in Medicaid has contributed to severe pressures on state budgets. States have closed budget gaps through a variety of measures while still accommodating annual growth rates in Medicaid of about 12 percent.¹⁴ To the extent that states are finding these growth rates unsustainable, eligibility standards may be affected. If coverage through Medicaid and state programs declines because of tightened eligibility standards, reduced outreach, or efforts to make enrollment more difficult, increases in public coverage will not offset the decline in ESI, and the number of uninsured people could rise even more sharply.

The authors gratefully acknowledge the valuable comments of Jack Hadley, Catherine Hoffman, Genevieve Kenney, and Stephen Zuckerman.

NOTES

1. U.S. Census Bureau, Health Insurance Coverage in the United States: 2002 (Washington: U.S. Government Printing Office, September 2003).
2. J. Bernstein, “The Jobless Recovery,” Issue Brief no. 186 (Washington: Economic Policy Institute, January 2003); Bureau of Labor Statistics, “Establishment Data: Historical Employment, Table B-1, Employers on Nonfarm Payrolls by Major Industry Sector, 1954 to Date,” ftp.bls.gov/pub/ suppl/empsit.ceseeb1.txt (20 January 2004); S. Heffler et al., “Health Spending Projections for 2002–2012,” Health Affairs, 7 February 2003, content.healthaffairs.org/cgi/content/abstract/hlthaff.w3.54 (29 October 2003); and B.C. Strunk and P.B. Ginsburg, “Tracking Health Care Costs: Trends Stabilize but Remain High in 2002,” Health Affairs, 11 June 2003, content.healthaffairs.org/cgi/content/abstract/hlthaff.w3.266 (29 October 2003).
3. We conducted similar analyses using the National Survey of America’s Families (NSAF) data from 1999 and 2002. The results showed similar patterns, although there were some differences, such as an increase in coverage for children and no change for adults. Much of the difference was attributable to the use of 1999 data, which meant that the period included one year of strong economic growth and two years of the economic downturn. See the following policy briefs from the Urban Institute’s series, Shapshots of America’s Families III (September 2003): J. Holahan, “Changes in Employer-Sponsored Health Insurance Coverage,” No. 9, www.urban.org/UploadedPDF/310849_snapshots3_no9.pdf; S. Zuckerman, “Gains in Public Health Insurance Offset Reductions in Employer Coverage among Adults,” No. 8, www.urban.org/UploadedPDF/310850_snapshots3_no8.pdf; and G. Kenney, J. Haley, and A. Tebay, “Children’s Insurance Coverage and Service Use Improve,” No. 1,
www.urban.org/UploadedPDF/310816_snapshots3_no1.pdf (all 20 January 2004).
4. An alternative is to limit the analysis to the family, which includes all related members who are living together. This classification is broader than the health insurance unit but more restrictive than the household. However, because neither private nor public insurance is typically available to all people within the family, the health insurance unit is a more appropriate measure. For example, most adults who live with their parents would not be part of the family’s private health insurance policy and would be eligible for public programs based on their own income, not the family’s income.
5. We repeated the analysis using the family unit rather than the health insurance unit and found virtually no differences in the changes between 2000 and 2002. Using the family income measure, there are fewer people below 200 percent of poverty and more in higher income groups. As a result, there are fewer uninsured people below 200 percent of poverty and more in higher income groups. For example, using the family unit, 74.3 million people are below 200 percent of poverty, compared with 85.7 million using health insurance units. In contrast, 98.2 million people are above 400 percent of poverty using the family unit, compared with 91.0 million using the health insurance unit. This results in 27.8 million uninsured people below 200 percent of poverty using the health insurance unit and 23.3 million using the family unit. Above 400 percent of poverty, there are 5.1 million uninsured people using the health insurance unit and 7.3 million using the family unit.
6. The federal poverty threshold was $18,100 for a family of four in the forty-eight contiguous states and the District of Columbia in 2002. See U.S. Department of Health and Human Services, “The 2002 HHS Poverty Guidelines,” 11 September 2003, aspe.hhs.gov/poverty/02poverty.htm (5 January 2004).
7. Since respondents can report more than one type of coverage, we use the following hierarchy to classify people by insurance coverage: employer coverage; Medicaid and state programs (including SCHIP); Tricare, Medicare, or other military coverage; and individual nongroup insurance. Our results did not change when we repeated the analysis in Exhibit 1 without the hierarchy.
8. For ease of exposition, we group Tricare and Medicare because they cover only small numbers of people in our study population (people under age sixty-five). Over the two-year period, Medicare added 300,000 adults. There was no net change in the number of people. Other data not shown indicate that this increase in Medicaid/SCHIP was predominantly among children covered by Tricare, although there was an increase in the coverage of adults and a decline in the coverage of children.
9. Other data not shown indicate that this increase in Medicaid/SCHIP was predominantly among children in families with incomes below 300 percent of poverty, as expected.
10. J. Holahan and M. Pohl, “Leaders and Laggards in State Coverage Expansions,” in Federalism and Health Policy, ed. J. Holahan, A. Weil, and J. Wiener (Washington: Urban Institute Press, 2003), 179–214.
11. “Parents” are defined as adults age nineteen and above with dependent children age eighteen or younger living in the same household.
12. The difference between parents and nonparents in the change in public coverage was statistically significant (p < .10), but the difference in the change in uninsurance rates was not.
13. BLS, “Establishment Data: Historical Employment, Table B-1.”
14. J. Holahan and B. Bruen, “Medicaid Spending: What Factors Contributed to the Growth between 2000 and 2002?” (Washington: Kaiser Commission on Medicaid and the Uninsured, September 2003).

John Holahan (jholahan{at}ui.urban.org) directs the Health Policy Center at the Urban Institute in Washington, D.C.; Marie Wang is a research assistant there.


DOI: 10.1377/hlthaff.W4.31
©2004 Project HOPE–The People-to-People Health Foundation, Inc.