RESEARCH FINDINGS ON THE ECONOMY
AND HEALTH INSURANCE COVERAGE
Summaries of Studies by the Urban Institute and
The Center for Studying Health System Change
April 23, 2004
RESEARCH FINDINGS ON THE ECONOMY AND
HEALTH INSURANCE COVERAGE
The Robert Wood Johnson Foundation commissioned studies from the Urban Institute
and the Center for Studying Health Systems Change (CSHSC) to explore the impact of
the downturn in the economy on insurance coverage. The Urban Institute’s analyses use
data from the 1999 and 2002 National Survey of America’s Families. The CSHSC
analysis uses the 2001 and 2003 Community Tracking Study Household Survey. Despite
differences in the two surveys—in both the questions they ask and the time periods they
cover—certain key findings are consistent across the studies.
Rates of employer-sponsored insurance (ESI) declined in both time periods.
Factors contributing to the decline included reductions in employment, shifts from
full-time to part-time work or to jobs less likely to offer coverage, and declines in
take-up of ESI when it was offered.
The low-income population—those with income less than 200 percent of the
federal poverty level (FPL)—experienced the largest declines in ESI. (In 2003,
the federal poverty level was $18,400 for a family of four.)
Uninsurance rates would have risen much more than they did, if public
programs—primarily Medicaid and SCHIP—had not picked up the slack.
CHANGES IN LABOR FORCE STATUS AND HEALTH INSURANCE
Jack Hadley of the Urban Institute explores the relationships between changes in labor
force status and changes the health insurance coverage of nonelderly adults and children
over the 1999 to 2002 period. After declining steadily from an annual rate of 7.5 percent
in 1992 to 4.0 percent in 2000, the national unemployment rate began increasing in early
February 2001. It peaked at 6.5 percent in January 2003 and was still at 6.0 percent in
February 2004 when 8.8 million people were unemployed.1 Labor force participation
also fell over this period, from 67.3 percent in February 2000 to 65.9 percent in February
2004. This means that an additional 1.4 percent of the population was no longer looking
for work or in the labor force, on top of the 6.0 percent who were unemployed.
Coinciding with these changes in labor force status, the share of the population without
health insurance coverage all year grew from 14.2 percent in 2000 to 15.2 percent in
Bureau of Labor Statistics, http://data.bls.gov, accessed March 5, 2004.
2002, an increase of almost four million people.2 This change reversed the decline in
uninsurance that began in 1997 when 16.1 percent of the population was uninsured.
Although the unemployment rate in the mid-1990s was similar to the rates the economy
has experienced in the last two years,3 the underlying structures of the labor and
insurance markets may be quite different now. The growth of the internet, for example,
has enabled businesses to reduce their full-time work forces and rely much more on
“outsourcing,” both to individual contractors and consultants and, more recently, to
foreign workers. These changes may have weakened the employed population’s access
to group insurance coverage offered by their employers. (Some people have access to
group insurance through unions and associations, or through former employers under the
At the same time, the cost of private insurance has been increasing relative to both
workers’ earnings and general inflation, compared to the mid-1990s. In the last three
years, premiums for employer-sponsored insurance have increased between 10.9 and 13.9
percent per year, while workers’ wages have grown between 2-3 percent per year and
average household income fell by 3 percent.4 Another change relative to the mid-1990s
is that workers’ premium contributions for family coverage have been growing more
rapidly than for single coverage, and the overall share of the premium paid by employers
has been shrinking.5
The interactive effects of these multiple factors have produced significant changes in
employer-sponsored coverage. The paper describes these changes by income class and
labor force status.
The number of unemployed adults increased by almost 3.5 million between 1999
and 2002. Almost 90 percent of these adults were from low- and lower-middle-
income families. Low- (below 200 percent of the FPL) and lower-middle-income
(200-400 percent of the FPL) adults also had the largest increases in the
percentage unemployed, 3.3 percent and 2.1 percent respectively.
Mills, Robert and Shailesh Bhandari. “Health Insurance Coverage in the United States: 2002.” U.S.
Census Bureau, September 2003.
Annual unemployment rates were 6.1 percent and 5.6 percent in 1994 and 1995, compared to 5.8 percent
and 6.0 percent in 2002 and 2003.
Holtz-Eakin, Douglas. “The Uninsured and Rising Health Insurance Premiums.” Testimony before the
Subcommittee on Health, Committee on Ways and Means, U.S. House of Representatives, March 9, 2004
(available from http://www.cbo.gov); DeNavas-Walt, Carmen, Robert Cleveland and Bruce Webster.
“Income in the United States: 2002.” U.S. Census Bureau, September 2003.
Gabel, John et al. “Health Benefits in 2003: Premiums Reach Thirteen-Year High as Employers Adopt
New Forms of Cost Sharing.” Health Affairs 22 (5) September/October 2003: pp. 117-126.
Upper-middle-income (400-600 percent of the FPL) adults also experienced a
relatively large drop in the employment rate of 1.9 percentage points, mostly due
to a high percentage dropping out of the labor force.
The drop in employment among low- and lower-middle-income adults was
accompanied by a decrease in full-time employment and an increase in part-time
employment—in which access to ESI is less likely. By contrast, upper-middle
and high-income workers shifted away from self-employment to employment by
firms that potentially offer employer-sponsored insurance coverage.
Overall, the number of uninsured nonelderly adults increased by 1.9 million from
1999-2002, with the uninsurance rate rising by 0.5 percentage point to 17.0
percent. The largest increase occurred among the unemployed, whose uninsurance
rate went from 33.7 percent in 1999 to 34.6 percent in 2002. The unemployed
accounted for almost 60 percent of the increase in the number of uninsured adults,
and their uninsurance rate was more than twice that of employed adults.
The increase in uninsurance was strongly related to family income, rising by 1.9
percentage points for low-income adults and 0.6 percentage point for lower-
middle-income adults, with essentially all of the newly uninsured adults coming
from families in those income brackets. The rate of uninsurance was also
strongly related to family income, with 36 percent of all low-income adults being
uninsured in 2002. The uninsurance rate was similar for the employed (38.4
percent) and the unemployed (40.2 percent) and lowest for those not in the labor
force (29.3 percent), who had the highest rate of public coverage.
By contrast, 14.7 percent of lower-middle-income adults were uninsured in 2002.
The unemployed had the highest rate of uninsurance, 23.2 percent, while the
employed and those not in the labor force had similar uninsurance rates.
Uninsurance rates among adults in higher income groups were much lower.
Among workers, those who were either self-employed or worked part time had
the largest increases in uninsurance rates. The uninsurance rate also increased
among low- and lower-middle-income workers who worked full time.
Despite declines in ESI, the uninsurance rate among children fell by 2.5
percentage points—representing 1.7 million children—between 1999 and 2002.
The decrease was similar for children in families with and without a working
parent, although the uninsurance rate was about 6 percentage points higher for
children without a working parent (15.0 percent versus 9.1 percent). The decline
in uninsurance reflects an expansion of public coverage, which offset the decrease
in private coverage.
FACTORS CONTRIBUTING TO THE DECLINE IN EMPLOYER-SPONSORED
In a separate study, Linda Blumberg and John Holahan of the Urban Institute analyze the
decline in ESI that occurred during the 1999-2002 period. These years were largely
dominated by the downturn in the U.S. economy, although the first year represented a
continuation of the broad economic expansion of the 1990s. The researchers estimate
changes in the probability of working, of receiving an offer of ESI, and of taking up that
offer, among unmarried adults, married adults, and children. They decompose the
changes in ESI into these components and assess the share of the change attributable to
each. They also determine how the changes over this time period differed for low-
income and higher-income populations, as well as the variations that occurred by
The share of the non-elderly population with ESI coverage fell from 69.2 percent
in 1999 to 66.9 percent in 2002, a change of 2.2 percentage points. Had the rate
of ESI coverage stayed at 1999 levels, 5.9 million more individuals would have
had such coverage in 2002 than was actually the case.
All three groups -- dependent children, unmarried adults and married adults --
experienced declines in ESI. Dependent children had the largest declines (3.1
percentage points), followed by unmarried adults (2.3 percentage points) and
married adults (1.7 percentage points).
Declines in ESI coverage were particularly sharp for the low-income (below 200
percent of the FPL) population, which was affected by a reduced likelihood of
working, a lower probability of being offered insurance by their employer, and a
decline in the rate of take-up ESI, when it was available. ESI coverage for the
low-income non-elderly population fell by just under 5 percentage points. This
amounts to a decline of about 12 percent—substantially larger than the drop (less
than 2 percent) experienced by the higher-income non-elderly population. The
researchers estimate that almost half of that decline resulted from lower take-up
rates, with most of the remainder attributable to changes in work and to lower
offer rates by employers in almost equal proportions.
There was a shift in work from large to small employers, where the likelihood of
ESI coverage is lower. The probability of working for a small employer increased
and the probability of working for a large employer declined for unmarried adults.
There was also a modest increase in the probability of working in government for
this group. At the same time, offer rates declined in small establishments. Thus,
while employment of unmarried adults was shifting to small establishments and to
government, the offer rate for those workers was falling. For those working in
small establishments, there were also declines in take-up rates for unmarried
adults and children.
OPTIONS FOR LOW- AND MIDDLE-INCOME WORKERS WHO LOSE
ACCESS TO EMPLOYER-SPONSORED COVERAGE
Based on the experience of the recent recession, Urban Institute researchers Sharon Long
and Yu-Chu Shen consider what is likely to happen to low- and middle-income workers
who lose access to ESI. Without ESI, an individual’s coverage options narrow to
obtaining coverage through a family member’s job (if available), purchasing non-group
insurance, or enrolling in public programs. Non-group insurance, however, may be
unaffordable for such workers and their families, and their income may not be low
enough for them to qualify for public coverage. For purposes of their analysis, the
researchers define non-group coverage as unaffordable if the premium exceeds 10
percent of family income.
While low-income (below 200 percent of the FPL) workers are more likely to be
uninsured and are most vulnerable to the loss of ESI coverage, many lower-
middle-income (200-400 percent of the FPL) workers are also in a precarious
position. Uninsurance for such workers increased between 1999 and 2002,
reflecting a loss in ESI coverage and a decline in the share of workers with
coverage options outside of ESI.
Factors driving the changes in ESI coverage for low- and lower-middle-income
workers between 1999 and 2002 included changes in the nature of their jobs and
shifts in the structure of the insurance market.
Many low- and lower-middle-income workers have few coverage options in the
absence of ESI. This is particularly problematic for low-income workers, as
nearly half of them have no viable alternatives outside of the ESI market.
Moreover, the affordability of non-group insurance declined sharply between
1999 and 2002: the shares of low- and middle-income workers who would face a
non-group premium that is less than 10 percent of their income dropped 10
percentage points over this period.
When workers do lose ESI many are likely to become uninsured. Although
dependent ESI coverage, public coverage, and non-group coverage would all
increase substantially if ESI were no longer an option, the overall levels of each
would remain relatively low. That finding is true for both low- and lower-middle-
income workers: Long and Shen estimate that 48 percent of low-income workers
and 31 percent of lower-middle-income workers would become uninsured if ESI
were no longer available to them.
CHANGES IN INSURANCE COVERAGE AND ACCESS TO CARE FOR
MIDDLE-INCOME AMERICANS, 1999-2002
Linda Blumberg and John Holahan of the Urban Institute also examine how insurance
coverage and access to health care of middle-income Americans changed between 1999
and 2002. They contrast the experience of the middle-income population with that of
lower- and higher-income Americans over the period.
How middle-income Americans fared in the last few years is of interest for several
reasons. First, through most of this period (2000-2002), there was a downturn in the U.S.
economy which affected their economic well being. Unemployment increased and many
Americans experienced reductions in income. Second, there was a continuing expansion
in public insurance, particularly for children through the State Children’s Health
Insurance Program (SCHIP). While this generally affected families with income below
200 percent of the FPL, there was some expansion of coverage above 200 percent of the
FPL. Third, these years encompassed a period in which insurance premiums increased
considerably faster than wages, making it more difficult for employers to offer coverage.
To the extent that employee contribution requirements increased, middle-income
Americans may also have found it more difficult to accept employers’ offers of
There is also concern that along with the increasing number of uninsured Americans, an
even larger number of middle-and higher-income individuals and families feel vulnerable
to the potential loss of coverage. The 2002 report by the Census showed that households
with income of more than $75,000 accounted for more than half of the increase in the
uninsured.7 This figure was somewhat misleading because it reflected increases in the
number of people living in households with more than $75,000 of income.8 Some of the
increase in the “high-income” uninsured reflected people with low incomes moving into
households with other people whose earnings lifted the household’s income above
$75,000. Nonetheless, the report fueled the belief that the large number of uninsured in
the U.S. was no longer just a low income problem, but was spreading to the middle class.
Another reason for concern is that public efforts to expand coverage have focused on
low-income populations (particularly children) and more recently their parents. Thus,
many of the poor have access to public coverage, although many unfortunately do not
participate in programs for which they are eligible and uninsurance rates remain high. At
the same time, those with higher incomes are likely to have offers of employer-sponsored
insurance and low uninsured rates as a result. Those in the middle are more vulnerable
John Holahan “Changes in Employer Sponsored Health Insurance coverage”, Snapshots of America’s
Families, The Urban Institute, September 2003.
U.S. Census Bureau. “Health Insurance Coverage: 2001”. September 2002.
John Holahan, Catherine Hoffman and Marie Wang, “The New Uninsured of Middle Class Americans: Is
it Real”, Kaiser Commission on Medicaid and the Uninsured, March 2003.
because they are more likely to work in firms that do not offer coverage, yet their
incomes are too high to be eligible for public coverage.
The uninsurance rate for low-income (below 200 percent of the FPL) adults
increased by 1.9 percentage points because of a decline in ESI that was only
partially offset by an increase in Medicaid and state program coverage. For low-
income children, there were sharper declines in ESI and larger increases in public
coverage because of the introduction of the SCHIP program and the related
increase in Medicaid enrollment. As a result, the uninsured rate for low-income
children fell by 5.6 percentage points.
Lower-middle-income (200 – 400 percent of the FPL) adults fared better than
low-income adults. Increases in Medicaid and private non-group coverage offset
declines in ESI and there were no significant declines in the uninsurance rate.
Similarly, for lower-middle-income children, the decline in employer sponsored
insurance of 3.7 percentage points was offset by a slightly larger expansion of
Medicaid and again, there was no increase in the uninsurance rate.
Upper-middle-income (400 percent – 600 percent of the FPL) experienced
increases in employer sponsored insurance, public coverage and private non-group
coverage and the uninsurance rate fell by 1.5 percentage points. Further analysis
showed that this reduction in uninsurance occurred only among childless adults.
In contrast, upper-middle-income children actually saw a reduction in coverage
due to the decline in both employer-sponsored insurance and private non-group
coverage. It appeared that premium increases for private dependent coverage
(group and non-group) may have led to a decline in coverage for children in this
Several changes in access mirrored the changes in coverage. Low-income adults
who experienced a decline in coverage also reported significant increases in the
likelihood of unmet need for care, particularly prescription drugs and dental care.
In contrast, low-income children who experienced an increase in coverage reported
a decline in unmet need for medical care or surgery and in dental care.
Among lower-middle-income adults there was a large increase in the number
reporting unmet need for care, primarily because of unmet need for prescription
drugs. It may be that the shift from ESI to Medicaid and non-group coverage led
to reduced access to care. Among lower-middle-income children, there was an
increase in the share of parents reporting confidence in their ability to obtain care
for their family, possibly reflecting the substitution of Medicaid or SCHIP for
employer sponsored insurance.
Individuals in virtually all income groups reported an increase in unmet need for
prescription drugs. This may reflect the rising cost as well as reduced coverage for
CONTINUING EFFECTS OF THE RECESSION ON EMPLOYER-SPONSORED
COVERAGE THROUGH 2003
Peter Cunningham and Len Nichols of the Center for Studying Health System Change
have developed findings from the most recent health insurance coverage data available,
which allow them to focus on the ongoing effects of the employment recession on
coverage, by comparing insurance rates between 2001 and 2003.
Rates of ESI fell between 2001 and 2003 by about 3.6 percentage points for the
non-elderly population and 2.5 percentage points for workers. This represents a
reduction in ESI of 2.3 million people by 2003. But the overall uninsured rate did
not increase by a statistically significant amount because public programs—
mostly Medicaid and SCHIP—picked up the slack.
Recent weak labor markets and increases in health insurance premiums had the
greatest impacts on the low-income (below 200 percent of the FPL) population.
They and lower-middle-income people (between 200 and 400 percent of the FPL)
were more likely than higher-income people (above 400 percent of the FPL) to
experience decreases in employment and in offer rates of ESI among those who
were employed. Some with access to ESI were also more likely to decline the
offer. But the researchers conclude that reduced access to coverage—in the form
of working at all and being offered coverage if one worked—was the primary
cause of lower ESI coverage during this period. Although there is some evidence
of middle-income families experiencing declines in ESI coverage, the impacts on
low-income families were far greater.
The decreased likelihood of being offered ESI was concentrated among low-
income workers in small firms—those with less than 50 employees. The share of
such workers with ESI offers fell by 6.3 percentage points over the 2001-2003
Comparing 2003 with 2001, a higher percentage of uninsured people who
declined ESI—both workers and others with access to ESI—came from families
with income above 400 percent of FPL. But that increased share primarily
reflects the fact that access to ESI declined for lower-income populations, so
fewer low-income people had opportunities to decline coverage.
The researchers estimate that 20 percent of the non-elderly uninsured in 2003 had
access to ESI through their own work or that of a family member, and declined it.
o These uninsured decliners included 9.2 percent of the low-income (below
200 percent of the FPL) population, 2.8 percent of the lower-middle-
income (200 – 400 percent of the FPL) population, and 1.3 percent of the
higher-income (above 400% of the FPL) population.
o More than half of the uninsured decliners were in low-income families,
and more than half were in families with children. In addition, almost half
of them were in families with children and both parents present, and one-
fifth of them were children under age 18.