Tax Policy: Health Insurance Tax Credit Participation Rate Was Low
(Letter Report, 05/02/94, GAO/GGD-94-99).
This report provides information on the health insurance tax credit,
which was established to encourage low-income workers to buy private
health insurance for their families. GAO discusses (1) the estimated
participation rate, in part to determine whether the potentially
eligible population was aware of the health insurance credit, and (2)
the health insurance tax credit's influence on low-wage workers'
purchase of health insurance.
--------------------------- Indexing Terms -----------------------------
REPORTNUM: GGD-94-99
TITLE: Tax Policy: Health Insurance Tax Credit Participation Rate
Was Low
DATE: 05/02/94
SUBJECT: Health insurance
Disadvantaged persons
Insurance premiums
Health insurance cost control
Employee medical benefits
Subsidies
Tax credit
Federal aid programs
Tax refunds
Taxpayers
IDENTIFIER: Census Bureau Current Population Survey
Earned Income Tax Credit
Robert Wood Johnson Foundation Health Care for the
Uninsured Program
Census Bureau Survey of Income and Program Participation
SIPP
**************************************************************************
* This file contains an ASCII representation of the text of a GAO *
* report. Delineations within the text indicating chapter titles, *
* headings, and bullets are preserved. Major divisions and subdivisions *
* of the text, such as Chapters, Sections, and Appendixes, are *
* identified by double and single lines. The numbers on the right end *
* of these lines indicate the position of each of the subsections in the *
* document outline. These numbers do NOT correspond with the page *
* numbers of the printed product. *
* *
* No attempt has been made to display graphic images, although figure *
* captions are reproduced. Tables are included, but may not resemble *
* those in the printed version. *
* *
* A printed copy of this report may be obtained from the GAO Document *
* Distribution Facility by calling (202) 512-6000, by faxing your *
* request to (301) 258-4066, or by writing to P.O. Box 6015, *
* Gaithersburg, MD 20884-6015. We are unable to accept electronic orders *
* for printed documents at this time. *
**************************************************************************
Cover
================================================================ COVER
Report to the Chairman and the Ranking Minority Member, Subcommittee
on Health, Committee on Ways and Means, House of Representatives
May 1994
TAX POLICY - HEALTH INSURANCE TAX
CREDIT PARTICIPATION RATE WAS LOW
GAO/GGD-94-99
Health Insurance Tax Credit
Abbreviations
=============================================================== ABBREV
AGI - Adjusted Gross Income
CPS - Current Population Survey
CBPP - Center on Budget and Policy Priorities
EBRI - Employee Benefit Research Institute
EITC - Earned Income Tax Credit
IRS - Internal Revenue Service
Letter
=============================================================== LETTER
B-255852
May 2, 1994
The Honorable Fortney H. (Pete) Stark, Chairman
The Honorable William Thomas, Ranking Minority Member
Subcommittee on Health
Committee on Ways and Means
House of Representatives
This report completes our work on your request for information on the
health insurance tax credit, which was established to encourage
low-income workers to purchase private health insurance coverage for
their families. You asked us to describe the administration of the
health insurance tax credit and to study the effect of the tax credit
on the purchase of health insurance. Our September 1991 fact sheet
on the administrative aspects of the health insurance tax credit
responded to the first part of your request.\1 This report responds
to the second part on the effect of the credit on the purchase of
health insurance by low-income families.
This report discusses (1) the estimated participation rate, in part
to determine whether the potentially eligible population was aware of
the health insurance credit; and (2) the health insurance tax
credit's influence on low-wage workers' purchase of health insurance.
The observations in this report are based on data from two random
samples of tax year 1991 tax returns claiming the earned income tax
credit, responses from a postcard survey mailed to taxpayers in the
sample, interviews with 48 potentially eligible earned income credit
taxpayers, and reviews of several independent studies. Only the data
from our samples of tax returns may be generalized to the population
of taxpayers who claimed the earned income tax credit and the health
insurance tax credit.
During our work on this request, the health insurance tax credit was
repealed by the Omnibus Budget Reconciliation Act of 1993. We
discussed our preliminary findings with your staff and they requested
that we complete this work, as much of the information would be
helpful in discussions on health care reform.
--------------------
\1 See Tax Administration: Administrative Aspects of the Health
Insurance Tax Credit (GAO/GGD-91-110FS, Sept. 12, 1991).
RESULTS IN BRIEF
------------------------------------------------------------ Letter :1
There is no clear measure of the size of the population eligible for
the health insurance credit. However, on the basis of comparisons
between the Census Bureau's Current Population Survey (CPS) and
Internal Revenue Service (IRS) data, we estimated that about
one-quarter of those who potentially were eligible actually claimed
the health insurance credit in 1991. We identified two reasons for
this: (1) those eligible may not have been aware of the credit, and
(2) the credit may not have been sufficient to encourage them to
purchase insurance coverage.
Our study indicated that a lack of awareness may have prevented more
eligible taxpayers from taking the health insurance credit. IRS
conducted an extensive outreach campaign in 1991 and 1992 regarding
the basic Earned Income Tax Credit (EITC), but it did not emphasize
the health insurance credit per se in its awareness-raising efforts.
The Center on Budget and Policy Priorities (CBPP), a low-income
advocacy group that has also promoted the basic EITC, likewise did
not give much attention to the health insurance credit in its
outreach efforts. Our interviews with a small sample of low-income
taxpayers suggested that they were unaware of the health insurance
credit, and they may have learned about it only after purchasing
insurance coverage.
Income was the major factor determining whether a family purchased
health insurance. Our analysis of randomly selected tax year 1991
tax returns showed that EITC recipient families with annual incomes
above $10,000 were 3.4 times more likely to pay for health insurance
than those with incomes below $10,000. This analysis also showed
that those with only one employer, rather than several, were more
likely to receive the health insurance credit.
We found that the health insurance credit reimbursed only a small
percentage of taxpayers' reported costs of health coverage. Also,
other studies have found a low response to health insurance
subsidies. Therefore, the health credit, as it existed, probably did
not provide much incentive to encourage families to purchase
insurance if they otherwise would not have done so. The maximum
health insurance credit available in 1991 was $428. Our analysis of
tax year 1991 tax returns showed that the health insurance credit
paid, on average, $233, or 23 percent, of the average reported health
insurance premium of $1,029 for credit recipients. This $1,029
average annual premium for recipients represented only a fraction of
the total cost of employer-provided health insurance, because
employers generally pay a significant part of the cost. A limited
number of studies by other researchers that have addressed this issue
have similarly found that income was the primary determinant of
whether individuals purchased health insurance, and neither
individuals nor employers were very responsive to health insurance
subsidies.
BACKGROUND
------------------------------------------------------------ Letter :2
The Omnibus Budget Reconciliation Act of 1990 provided an additional
credit to taxpayers who qualified for the EITC and contributed toward
the purchase of health insurance for a qualifying child.\2 The health
insurance credit was intended to offer an incentive to low-income
(EITC-eligible) taxpayers to purchase health insurance coverage for
their children. Even taxpayers with no tax liability could receive
the credit in the form of a tax refund. The law took effect on
January 1, 1991.
For tax year 1991 over 13.3 million taxpayers received approximately
$9.9 billion in basic EITC payments, while about 2.3 million also
received health insurance credits totalling $496 million. The
Omnibus Budget Reconciliation Act of 1993 revised the provisions of
the EITC and repealed the health insurance credit effective December
31, 1993.
--------------------
\2 In order to be a qualifying child, an individual must satisfy a
relationship test, a residency test, and an age test.
ELIGIBILITY FOR THE CREDIT
---------------------------------------------------------- Letter :2.1
Qualified health insurance expenses for which the credit was
available were premiums paid during the tax year for health insurance
coverage that included one or more qualifying children. These
expenses included only those relating to the cost of coverage. Thus,
medical expenses such as co-payments or other deductibles, as well as
other out-of-pocket medical expenses, were not qualified as health
insurance expenses.
In addition, qualified insurance expenses did not include amounts
paid by an employee who contributed to his or her employer-sponsored
health plan on a pretax basis. For example, employees who made
contributions toward health insurance through flexible benefit plans
paid taxes on gross wages less those contributions. Such
contributions did not qualify for the health insurance tax credit.
However, if the employee contributed to an employer-sponsored plan an
amount that was included in taxable income, that amount was eligible
for the credit.
CALCULATING THE CREDIT
---------------------------------------------------------- Letter :2.2
The health insurance credit was calculated on the basis of a
taxpayer's earned income. For tax year 1991, over the phase-in range
(incomes between $1 and $7,100), the credit was 6 percent of earned
income. In the second range, for earned incomes between $7,100 and
$11,250, the credit remained at its maximum ($428). In the last
range, for higher earned incomes ($11,250 through $21,250), the
credit phased out at a rate of 4.285 percent of earned income and
disappeared for earned incomes of more than $21,250.
The allowable credit was limited to no more than the actual cost to
the taxpayer of premiums paid for family coverage. Thus, the credit
was limited to the lesser of the amount calculated by the taxpayer on
the basis of earned income or the actual qualified health insurance
expenses. For example, a person with earned income of $6,500 and one
qualifying child for whom the taxpayer was paying $30 a month ($360
per year) in health insurance premiums could have received a health
insurance credit of no more than $360. In this example, the maximum
health insurance credit allowed was $392, but the credit was limited
to the actual amount paid ($360).
LIMITED PARTICIPATION IN THE
HEALTH INSURANCE CREDIT
------------------------------------------------------------ Letter :3
IRS data show that 2.3 million low-income taxpayers received the
health insurance credit in tax year 1991. An unpublished Employee
Benefit Research Institute (EBRI) analysis of the Census Bureau's
1992 CPS estimated that almost 9 million low-income families could
have qualified for the health insurance credit in 1991, based on
their eligibility for the EITC and having had private health
coverage. We used the IRS data on the number of taxpayers who
received the credit and the EBRI estimate of the eligible population
to estimate the health insurance credit participation rate of about
26 percent in 1991. This was significantly lower than the
participation rate for the basic EITC, which was estimated to be
between 80 and 86 percent in 1990.\3
We obtained from EBRI a tabulation of the 1992 CPS, which estimated
that about 8.8 million families appeared to meet the eligibility
requirements for the health insurance credit in 1991 (see table 1).\4
Because CPS does not track the number of people who qualify
specifically for the health insurance credit, we used a tabulation of
CPS data to estimate the number as accurately as possible. This
figure actually reflects, based on CPS definitions, the number of
families that had a worker with an adjusted gross income under
$21,250; had children under age 18; and had private health insurance
at some time in 1991. These parameters do not match precisely with
the Internal Revenue Code's eligibility criteria for the health
insurance credit. For example, CPS families are not the same as
taxpaying units, children shown as part of a family in CPS may not be
qualifying children according to the eligibility criteria, CPS does
not track whether families paid for health insurance or whether
insurance was extended to children in the household, and CPS measures
income somewhat differently from the Code. Due to inadequacies in
the data, we were unable to determine the extent to which each of
these differences may over- or undercount the population potentially
eligible for the health insurance credit. (See app. I.)
Table 1
Potentially Eligible Health Insurance
Credit Population in 1991
Millions of
families Percent of total
-------------------- ------------------ ------------------
Private coverage
------------------------------------------------------------
Employer coverage\a 7.8 63.4
Other private 1.0 8.1
coverage
============================================================
Total private 8.8 71.5
coverage
No health insurance 3.5 28.5
============================================================
Total 12.3 100.0
------------------------------------------------------------
Note: Sampling errors associated with EBRI tabulations of the 1992
CPS are less than plus or minus 6 percent at the 95 percent
confidence level.
\a Employer coverage is that provided in some way, either directly or
indirectly, through one's employer. The data do not indicate to what
degree health care premiums are subsidized by employers.
Source: Employee Benefit Research Institute Unpublished Tabulation
of the 1992 CPS Information.
According to IRS data, 2.3 million taxpayers received the health
insurance credit in 1991. If, according to EBRI tabulations of CPS
data, 8.8 million families were eligible for the health insurance
credit, the participation rate for tax year 1991 was only 26 percent,
which is significantly lower than the participation rate for the
basic EITC, which was estimated to be between 80 and 86 percent in
1990.\5
An alternative to estimating the participation rate in the program is
to look at the credit's success in terms of the target population
(families who qualified for the credit as well as those without
insurance who theoretically would be encouraged to purchase coverage
as a result of the credit's availability). We defined the target
population as all low-income (EITC-eligible) taxpayers with children;
the credit was intended to offer an incentive to encourage taxpayers
to purchase health insurance coverage for their children. According
to EBRI tabulations of CPS data, 3.5 million potentially eligible
families were uninsured in 1991 (see table 1). If this total were
added to the 8.8 million families who were potentially eligible and
had privately provided health coverage in that year, the target
population would have been 12.3 million. With this 12.3 million as
the denominator, the health insurance credit would have had a
participation rate of only 19 percent. Given either scenario, the
health insurance credit participation rate was significantly lower
than the participation rate for the EITC, which was estimated to be
between 80 and 86 percent of the eligible population in 1990.\6
--------------------
\3 See Scholz, John Karl. "The Earned Income Tax Credit:
Participation, Compliance and Antipoverty Effectiveness." Institute
for Research on Poverty, Discussion Paper No. 1020-93. University
of Wisconsin, Madison (August 1993).
\4 The CPS is conducted monthly by the Bureau of Census and provides
data on employment status, income, and other characteristics of
families across the United States. In the March 1991 supplement to
the survey, interviewers asked questions regarding families' health
insurance status. It is from the CPS that GAO and research
institutes collect most of their demographic and statistical
information of this nature on the U.S. population.
\5 Scholz, John Karl, op. cit.
\6 Scholz, John Karl, op.cit.
LIMITED OUTREACH MAY HAVE
AFFECTED AWARENESS AND
PARTICIPATION
------------------------------------------------------------ Letter :4
Lack of outreach specifically pertaining to the health insurance
credit may have resulted in low taxpayer awareness of the credit.
Because the health insurance credit was first offered in 1991,
extensive outreach efforts would have been required to inform
low-income taxpayers of the credit's availability. IRS' outreach
efforts regarding the EITC were extensive in 1991 and 1992, as were
those conducted by the CBPP low-income advocacy group, but neither
outreach program emphasized the health insurance credit.
IRS officials in various divisions said that while IRS has promoted
the EITC widely, it was not active in promoting the health insurance
credit. IRS' annual business plans for 1991 and 1992 dictated that
the agency conduct a major outreach program regarding the EITC in
those years. IRS promoted the EITC nationwide through various
channels and under the coordination of several IRS divisions. IRS'
EITC outreach efforts included radio, television, and newspaper
advertisements; posters and mail stuffers; coverage in Publication
596 and on various federal tax forms; question and answer columns and
tax information supplements in newspapers; and a folder of
information distributed to Volunteer Income Tax Assistance sites and
other interest groups for use in their outreach efforts or for
distribution to potentially eligible taxpayers themselves. The
outreach materials providing more detailed information on the EITC
(such as the folder of information distributed to volunteers and IRS
Publication 596) also described the two supplemental credits (the
health insurance and young child credits), or they at least mentioned
that these credits were available to qualifying taxpayers. IRS did
not promote the health insurance credit independently from the EITC.
According to one IRS official, the health insurance credit was not
promoted independently because the first criterion in qualifying for
the health insurance credit is EITC eligibility.
CBPP also conducted an extensive nationwide EITC awareness-raising
campaign in 1991 and 1992 but, like IRS, promoted the health credit
only secondarily. The co-coordinator of the EITC outreach campaign
at CBPP told us that although the EITC is widely supported by
low-income advocacy groups and appears to have a high participation
rate, the health insurance credit did not receive such support.
Because CBPP was not a strong proponent of the health insurance
credit, it did not actively promote the credit.
CBPP did not support the health insurance credit for several reasons,
according to the EITC campaign co-coordinator. First, he said that
the health insurance credit made it more difficult to teach people
about the EITC and made the EITC more complicated for low-income
taxpayers. In addition, he believes the credit did not subsidize
health insurance costs sufficiently to encourage families to purchase
health insurance. Finally, IRS did not monitor the quality of health
insurance coverage being reimbursed by the credit. The CBPP was also
concerned that health insurance providers would take advantage of the
availability of the health insurance credit to offer "flimsy" health
insurance policies to low-income families specifically to qualify
them for the credit.\7
--------------------
\7 The Oversight Subcommittee of the House Committee on Ways and
Means held hearings on March 4, 1993, to discuss the sale of inferior
policies to low-income families and potential abuses of the health
insurance credit.
LOW LEVEL OF AWARENESS OF
THE HEALTH INSURANCE CREDIT
---------------------------------------------------------- Letter :4.1
A lack of awareness may have been a contributing factor for the low
participation in the health insurance credit in 1991. In March and
April 1993, we interviewed taxpayers at IRS taxpayer service sites in
six cities regarding their awareness of the health insurance
credit.\8 Of the 233 taxpayers screened, 48 said they qualified for
the EITC. Of those, 35 said they were unaware of the health
insurance credit. Those taxpayers who were aware of the credit
typically learned about it from the Schedule EIC while they completed
their tax returns.\9 At that point, only those who had already paid
for health coverage could claim the credit. If taxpayers learned of
the credit's availability only when completing their tax returns, it
was too late for the credit to have any impact on their decisions to
purchase health insurance for the prior year.
Responses to our nationwide postcard survey sent in May and June of
1993 (including follow-up) to 1,839 randomly selected EITC recipients
from our basic EITC only and health insurance credit samples\10 also
indicated a lack of awareness. About one-third of basic EITC only
recipients who responded to the survey (128 out of 366 responding)
indicated that they had family health insurance at some time in 1991
but did not claim the health insurance credit.\11 (See app. I for a
detailed discussion of the methodology.) We believe possible
explanations for this are that (1) they were unaware of the credit,
(2) they had insurance for a short period of time and did not include
it when the tax return was prepared, and (3) they may not have
actually qualified for it because the employer paid the full cost of
family coverage. However, on average, 75 percent of workers employed
by companies that offered health benefits paid some portion of the
cost for family coverage in 1990.\12 Thus, it was unlikely that many
of the 128 respondents who indicated they had family insurance
coverage some time during the year worked for employers that paid 100
percent of the cost of their family coverage. Therefore, many of
these 128 respondents probably paid some part of their family health
insurance coverage and did not claim the health insurance tax credit
for which they were eligible. Either they were unaware of the
credit, or they had health insurance for only a short period of time
and did not include it when they prepared their returns.
--------------------
\8 The taxpayers interviewed were selected at random from those who
happened to walk into the selected IRS service sites while we were
there.
\9 Not all EITC recipients filed the Schedule EIC (see table II.1).
Prior to tax year 1991, IRS awarded the basic EITC to taxpayers who
appeared to qualify for the credit regardless of whether they filed a
Schedule EIC or not. In tax year 1991, IRS began sending a form
letter requesting additional information from taxpayers not filing
the Schedule EIC who appeared to qualify for the credit.
\10 Our two samples of returns were randomly selected from the
universe of 13.3 million taxpayers who received the EITC for tax year
1991. Throughout this report, we refer to those who received the
EITC but not the health insurance credit as the "EITC only"
population (10,982,192 taxpayers), even though they may have received
other supplemental credits, such as the young child credit. We refer
to taxpayers who received the EITC as well as the health insurance
credit as the "health insurance credit population" (2,247,032
taxpayers).
\11 These results are based on postcards mailed to 928 basic EITC
only recipients, 366 of whom responded.
\12 This figure is based on the Health Insurance Institute of
America's 1990 Employer Survey. The survey results are given at the
95-percent confidence interval with a sampling error of plus or minus
2 percent.
SUBSIDIES MAY NOT INCREASE
HEALTH INSURANCE COVERAGE AMONG
LOW-INCOME POPULATION
------------------------------------------------------------ Letter :5
Taxpayers' income, not subsidies, was the major difference between
those taxpayers who claimed the credit and reported having health
insurance and taxpayers not claiming the health insurance credit.
Analysis of our two samples of 1991 tax returns revealed significant
differences between taxpayers who claimed the basic EITC only in 1991
and those who claimed the health insurance credit in that year. Both
income and number of employers were sources of differences as to
whether or not taxpayers claimed the health insurance credit and, we
assume, had health insurance coverage. The most important difference
was income.
INCOME IS MOST SIGNIFICANT
DIFFERENCE BETWEEN EITC AND
HEALTH CREDIT RECIPIENTS
---------------------------------------------------------- Letter :5.1
Analysis of our samples of tax year 1991 tax returns showed that
adjusted gross income (AGI) was the most significant factor
determining whether or not a taxpayer would have claimed the health
insurance credit. EITC taxpayers with AGIs greater than $10,000 were
3.4 times more likely to have claimed and received the health
insurance credit in 1991 than those with incomes below $10,000.
The health insurance credit recipient population had an average AGI
29.8 percent higher than that of the EITC only population ($14,019
compared with $10,799, respectively, as shown in table II.2).\13 AGI
varied by the IRS region in which taxpayers lived. However, despite
regional variation, the difference between the AGIs reported by the
EITC only and health insurance credit populations was statistically
significant at the national level as well as in each individual
region. AGI is the only variable we studied that was statistically
significantly different between the two populations in every region.
Although taxpayers in IRS' North-Atlantic region paid the highest
average premiums ($1,254) because they had the highest average AGI,
their average health insurance credit reimbursement was low ($209, or
17 percent). Conversely, although taxpayers in the Southeast region
had the lowest average health care premiums ($924) because they had
the lowest average AGI, their average reimbursement rate was higher
($250, or 27 percent). (Tables II.3 and II.7 highlight the
differences in demographics between the EITC and health insurance
credit populations by IRS region.)
The higher average income among the health insurance credit recipient
population also accounts for a greater distribution of this group in
the "phase-out" portion of the EITC. (As discussed earlier, with
AGIs greater than $11,250 the EITC began to "phase out" and became $0
for AGIs of more than $21,250.) Our analysis of samples of tax year
1991 tax returns shows that both the EITC only and health insurance
credit recipient populations fell predominantly in the phase-out
portion of the EITC cycle (AGIs greater than $11,250). However, the
proportion was much greater for the health insurance credit
population: 47 percent of the EITC only population fell in the
phase-out portion, compared with 73 percent of the health insurance
credit population (see figs. 1 and 2).
Figure 1: Distribution of EITC
Recipients by Phase of EITC
Cycle (1991)
(See figure in printed
edition.)
Source: GAO random sample of 957 EITC recipients in 1991.
Figure 2: Distribution of
Health Insurance Credit
Recipients by Phase of EITC
Cycle (1991)
(See figure in printed
edition.)
Source: GAO random sample of 942 health insurance credit recipients
in 1991.
Another factor differentiating between the two populations that helps
to explain why one group claimed and received the credit and
presumably had health insurance while the other did not was the
number of employers reported on tax returns. The health insurance
credit recipients were more likely to have one employer (60 percent)
than EITC only recipients (54 percent). (See tables II.1 and II.2.)
In our samples of tax year 1991 tax returns, taxpayers who were
self-employed were 1.5 times and taxpayers with one employer were 1.3
times more likely to have received the health insurance credit than
those with multiple employers. Thus, the number of employers
taxpayers worked for over the course of a year appeared to have an
impact, although less significant than AGI, on the likelihood that
they would have health insurance.
Taxpayers with more than one employer may be more likely to be
working part-time or on a temporary basis for several employers.
Such workers are less likely to be offered health coverage by their
employers than those working full-time for one employer.
We looked at other demographic characteristics, such as number of
dependents and taxpayer filing status, to see if they differed
between the two populations and may have had an impact on whether
taxpayers purchased health insurance or not. (See table I.1.)
However, these differences did not help to explain why one group
claimed the health insurance credit while the other did not.
--------------------
\13 All estimates derived from our samples of tax returns are given
at the 95 percent confidence interval with associated sampling errors
of less than plus or minus 10 percent, unless otherwise stated.
HEALTH INSURANCE CREDIT
PROBABLY DOES NOT ENCOURAGE
LOW-INCOME FAMILIES TO
PURCHASE HEALTH COVERAGE
---------------------------------------------------------- Letter :5.2
Health insurance coverage for individuals is not generally available
in small, incremental packages, and individuals therefore must incur
large premiums to obtain coverage. However, employer-subsidized
coverage usually significantly reduces the cost to the covered
employee. According to the Health Insurance Association of America,
the average cost of individually purchased family health insurance
under an employer-sponsored plan was $4,260 in 1991.\14 According to
this source, the average employer subsidy of family insurance
coverage, if any, was about 72 percent of the insurance premium,
depending on the type of plan offered. The Association also reported
that in 1991, 77 percent of all employees worked in firms offering
health insurance (although about 11 percent of full-time employees
and 86 percent of part-time employees were ineligible for their
employers' plans). Forty-two percent of firms offered some type of
health insurance to employees in 1990.\15
Because the health insurance credit reimbursed recipients for a small
percentage of their reported health insurance costs, we believe the
credit did not likely encourage many families to purchase coverage.
Although we could not determine from our analysis whether the health
insurance credit actually increased coverage among low-income
families, several studies we reviewed indicated that subsidies of
health insurance premiums, offered to either individuals or
employers, were not likely to increase coverage greatly.
The average health insurance premium reported by health insurance
credit recipients in our 1991 sample was $1,029, and the average
health insurance credit received by our sample population was $233.
(See table II.2.) Four taxpayers out of 942 in our sample reported
insurance premiums in 1991 below $25, and 1 reported a premium of
$6,060. However, almost all (88 percent) of those in our sample of
health credit recipients reported premiums under $2,000: 61 percent
reported premiums under $1,000, and 80 percent reported premiums
under $1,500. Thirty-six taxpayers in our sample received $20 or
less in credit, based on their AGIs and reported insurance premiums,
while 113 taxpayers received the maximum credit of $428. The health
insurance credit reimbursed an average of 23 percent of taxpayers'
reported costs of coverage.
--------------------
\14 Health Insurance Association of America figures presented in this
report are based on the Association's 1990 and 1991 nationwide
Employer Surveys. At the 95 percent confidence level, the sampling
errors for the 1990 and 1991 surveys are plus or minus 2 percent.
\15 This percentage varied by type, location, and size of industry.
For example, only 32 percent of retail trade firms offered health
coverage, compared with 55 percent of manufacturing firms.
Thirty-five percent of firms in the South offered health coverage to
employees, compared with 51 percent in the Northeast. Only 27
percent of firms with fewer than 10 employees offered health benefits
to workers.
OTHER STUDIES INDICATED
INCOME WAS KEY FACTOR
INFLUENCING THE PURCHASE OF
INSURANCE
---------------------------------------------------------- Letter :5.3
The studies we reviewed also indicated income was the key factor
influencing an individual to purchase insurance. In addition, some
studies investigated whether subsidizing health insurance costs would
increase demand for the insurance. The subsidies studied appeared to
have little effect, whether they were in the form of reductions in
premium prices for individual policies or for employers who offered
coverage to their employees.
Two studies by a researcher then at the Urban Institute,\16 which
address the demand for health insurance, indicated that income was
the prime factor affecting demand for health insurance coverage: the
higher a family's income, the greater the demand for health
insurance. One study based on a model of the demand for individual,
nongroup, health insurance found that income was the primary
determinant of whether individuals purchased nongroup health
insurance. A hypothetical 28-year-old, married, white male living in
Maryland was found to have a 25-percent chance of purchasing
self-pay, nongroup insurance if his income was below $5,000; a
30-percent chance with income between $5,000 and $10,000; a
42-percent chance with income between $10,000 to $20,000; and a
57-percent chance with income above $20,000.
This study found that such factors as the individual's age, race,
marital status, frequency of employment turnover, and level of risk
aversion, as well as the presence of children, each affected demand,
but to a lesser extent than income.\17 This study also indicated that
older individuals were more likely to buy health insurance, as were
those who were caucasian, married, had one full-time job, were highly
risk-averse, and had children. However, none of these factors showed
as great an impact on demand for health insurance as did income. The
other study on workers' characteristics corroborated these findings.
--------------------
\16 Katherine Swartz, "The Demand for Self-Pay Health Insurance: An
Empirical Investigation." Unpublished, June 1988; Katherine Swartz,
"Characteristics of Workers Without Employer-Group Health Insurance."
Trends in Health Benefits, July 1989, pp. 101-114.
\17 We were not able to test for most of these variables in our
analysis because such information as age, race, and level of risk
aversion are not included in tax returns.
STUDIES ALSO INDICATED
THAT SUBSIDIES MAY HAVE
HAD LITTLE EFFECT ON
COVERAGE
-------------------------------------------------------- Letter :5.3.1
The Urban Institute researcher's study on demand for health insurance
showed that significant reductions in the price of premiums for
self-pay, nongroup health insurance did not greatly increase the
low-income individual's demand for health insurance coverage.\18
According to the study, about 40 percent of adults would purchase
nongroup health insurance for themselves or their families, assuming
actual health insurance premiums. A subsidy of 10 percent of the
premium amount would increase coverage by 1.0 percent, and a subsidy
equal to 75 percent of the cost of coverage probably would increase
coverage by about 13 percent. Thus, subsidies in this range would
appear to have little effect on the purchase by individuals of
nongroup health insurance policies.\19
A study by RAND, a nonprofit policy analysis research institution,
and funded by the U.S. Department of Labor in large part supported
these findings.\20 This study estimated that about 50 percent of all
families not offered health insurance coverage through their
employers would purchase coverage on their own. According to this
study, the figure is lower for low-income families: only 20 percent
of families with incomes below $5,000 in 1988 would have purchased
coverage on their own. Overall, families' demand for health
insurance appeared to increase as subsidies reduced the price of
coverage, although low-income families were far less responsive to
subsidies than those at higher income levels. The study concluded
that large subsidies would be required to increase participation
rates significantly, and many workers may still not purchase coverage
due to low income levels.
Another study by two senior economists at RAND investigated
individuals' purchases of group insurance, i.e., the supply and
demand for health insurance among workers.\21 This study found that
most workers (82 percent) were employed by firms that offered health
insurance to employees. Of those, 2 percent did not take the
employer-provided coverage and did not have insurance coverage from
other sources. Those who did not take the employer-provided coverage
were largely low-wage, young workers who worked predominantly
part-time in small firms or on a seasonal or temporary basis. Thus,
they were less able to afford the cost of coverage and refused offers
for coverage even if subsidized by employers. These economic and
demographic characteristics matched those of the 18 percent of
workers who were not offered insurance by employers, suggesting that
they, too, were unlikely to purchase health insurance even if
subsidized. The authors concluded that although tax credits or
subsidies "may stimulate some additional purchase of insurance," such
mechanisms are unlikely to do very much.
Subsidies of up to 50 percent of insurance costs offered to employers
similarly may not increase their provision of health insurance to
low-income workers. A study published in the Journal of the American
Medical Association evaluated whether small employers (with under 20
employees) who did not provide health insurance to employees would do
so if given a 50-percent reduction (subsidy) in the price of
coverage.\22 The researchers found that in the first year the subsidy
was offered (1988), it resulted in a 3.5 percentage point increase in
the number of small employers surveyed that began offering insurance
and had not offered insurance in the preceding year.\23 Over time,
the researchers estimated that the subsidy may increase the number of
firms offering insurance by as much as 16.5 percent. However, the
authors believed this to be an upper limit of the subsidy's impact,
and they noted that even under ideal conditions, these results
highlighted the limitations of voluntary programs to increase
employer coverage.
Researchers directing The Robert Wood Johnson Foundation's Health
Care for the Uninsured Program found that most small employers
(generally with 25 or fewer employees) were not willing to offer
health insurance to their employees even if premium reductions of 25
to 50 percent of prevailing rates were offered.\24 The primary
reasons cited by employers for not offering coverage were the high
cost of doing so, the high cost to employees (because small groups
are seen as higher risk by insurers), and employees (often low-wage
or temporary) not demanding coverage.
A study performed by RAND and funded by the U.S. Department of Labor
suggested that small employers may be more responsive to subsidies
than indicated by the Robert Wood Johnson researchers.\25 This study
reported that 41 percent of small employers currently offered
insurance to employees. They estimated that the percent of small
firms offering coverage would increase by about 6 percentage points
with a 5-percent subsidy and an additional 6 percentage points for
each added 5-percent subsidy, up to 20 percent. However, they
concluded that moderate subsidies of 15 percent, for example, would
make "only modest progress in reducing the number of firms who do not
insure their workers."
--------------------
\18 In the recent study, Tax Incentives and the Decision to Purchase
Health Insurance: Evidence from the Self-Employed, Jonathan Gruber
and James Poterba found that subsidies may have a significant effect
on the demand for health insurance coverage among higher income
workers. The study supports our principal finding that income is the
primary factor affecting the demand for coverage. (Working Paper No.
4435, National Bureau of Economic Research, Cambridge, MA, August
1993.)
\19 For the same hypothetical man mentioned above, with all other
factors held constant, the probability of his purchasing insurance
went up only slightly as his premium dropped in price. This man had
a probability of purchasing insurance of 0.40 if his premium was $120
per month, compared to 0.43 if the premium was halved to $60 per
month.
\20 M. Susan Marquis and John L. Buchanan, "Subsidies and National
Health Care Reform: The Effect on Workers Demand for Health
Insurance Coverage," in Health Benefits and The Workforce. U.S.
Department of Labor, Washington, D.C., 1992. pp. 85-92.
\21 Long and Marquis, "Gaps in Employer Coverage: Lack of Supply or
Lack of Demand?" Health Affairs, Supplement 1993, pp. 282-293. This
study was based on the 1988 CPS and addressed the supply and demand
for health insurance among workers.
\22 Thorpe, et.al. "Reducing the Number of Uninsured by Subsidizing
Employment-Based Health Insurance: Results From a Pilot Study."
Journal of the American Medical Association, February 19, 1992, pp.
23-30.
\23 These findings are based on a 1989 survey of 1,006 small
employers in New York State, 485 of which did not offer insurance in
the preceding year.
\24 W.D. Helms, et. al. "Mending the Flaws in the Small-Group
Market." Health Affairs, Summer 1992, pp. 7-27.
\25 Arleen Leibowitz and Michael Chernew, "The Firm's Demand for
Health Insurance," in Health Benefits and The Workforce. U.S.
Department of Labor, Washington, D.C., 1992. pp. 77-83.
OBJECTIVES, SCOPE, AND
METHODOLOGY
------------------------------------------------------------ Letter :6
Our objectives were to (1) estimate the participation rate, in part
to determine whether the population of low-income taxpayers was aware
of the health insurance credit; and (2) assess whether the health
insurance credit influenced low-income families to purchase health
insurance.
To estimate the participation rate, we used the 1992 CPS, which
contained 1991 data, for information on the uninsured population and
the number of families who were potentially eligible for the health
insurance credit in 1991. We reviewed other data sources, the Survey
of Income and Program Participation and Current Population Survey,
for example, and we used data tabulations from the EBRI to estimate
the size of the population that might have qualified for the health
insurance credit in 1991.
To explore the awareness of the health insurance credit among the
potentially eligible population, we interviewed staff at the Center
on Budget and Policy Priorities (CBPP) and IRS officials in various
divisions regarding their health insurance credit outreach efforts.
We also screened 233 taxpayers who were visiting IRS taxpayer service
sites in San Francisco, Atlanta, Philadelphia, Albany, Cincinnati,
and Chicago over a period of 1 day; and we interviewed 48 of these
taxpayers who said they were eligible for the EITC.
To examine the influence of the health insurance credit on low-income
taxpayers' purchase of health coverage, we analyzed data from two
statistically random samples of tax year 1991 tax returns. (See app.
I for a detailed discussion of the methodology used.) To identify
differences between individuals who claimed the credit and those who
did not, we obtained information on the demographic characteristics
of taxpayers who received the basic EITC only (957 taxpayers in our
sample) and those who received the EITC and the supplemental health
insurance credit (942 taxpayers in our sample).\26 We did not verify
the accuracy of information contained in the tax returns, nor did we
verify the adequacy of the health insurance policies purchased by
health insurance credit recipients.
We also sent a nationwide postcard survey to 928 basic EITC only and
911 health insurance credit recipients from our samples of 1991
returns in order to determine (1) if respondents had health insurance
and whether it was purchased through their employers or individually;
and (2) if respondents were employed and, if so, whether employment
was full- or part-time.\27 Finally, we gathered information from
studies on the price of health insurance nationwide and the effect of
prices on the demand for health insurance to determine if subsidies
lowering the price of health insurance were likely to increase
coverage rates. (See app. II for the detailed results of our
analysis.)
In order to learn about the health insurance credit and studies
pertaining to the credit, we reviewed relevant literature and
interviewed U.S. government officials and experts in the fields of
health policy and health insurance. We discussed the information in
this report related to IRS with an Assistant to the Commissioner who
agreed with the basic message as it affected IRS. We conducted our
field work between January 1993 and September 1993, in accordance
with generally accepted government auditing standards.
We plan no further distribution of this report until 30 days after
its issuance date, unless you publicly release its contents earlier.
After 30 days, we will send copies to various congressional
committees, Members of Congress, the Secretary of the Treasury, and
other interested parties.
Major contributors to this report are listed in appendix III. Please
contact me on (202) 512-5407 if you have any questions about the
report.
Jennie S. Stathis
Director, Tax Policy and
Administration Issues
--------------------
\26 The estimates presented in this report, which we developed from
our two random samples of taxpayers, are given as point estimates and
have associated sampling errors of plus or minus 10 percent of the
given estimate at the 95 percent confidence level, unless otherwise
indicated.
\27 We did not send postcards to 60 of the 1,899 of the sample
taxpayers in our sample because they were under IRS review or audit.
SAMPLING AND DATA ANALYSIS
METHODOLOGY
=========================================================== Appendix I
This appendix describes the methodology we used to determine the
participation rate in the health insurance tax credit and the factors
affecting demand for health insurance. Information is included
concerning (1) selection of our tax return sample, (2) analysis of
tax return data, (3) the use of CPS and IRS data to estimate the
health insurance credit participation rate, and (4) administration of
the postcard survey.
SELECTION OF TAX RETURN SAMPLE
--------------------------------------------------------- Appendix I:1
We obtained from IRS a computer tape containing the names and Social
Security Numbers of 13.3 million taxpayers who received the EITC for
tax year 1991, including about 2.3 million who also received the
health insurance credit in that year. We stratified the 13.3 million
EITC recipients into 2 groups: group A, which received the basic
EITC and the health insurance credit (2.3 million taxpayers); and
group B, which received the basic EITC only (11 million taxpayers).
We selected 2 statistically random samples of tax returns--1,000 from
group A and 1,000 from group B. We requested from IRS the original
tax returns selected for our sample. IRS was not able to locate 59
of the returns, which reduced our sample size to 1,941.
We eliminated 42 of the selected tax returns from our samples because
of incomplete or conflicting data on the tax returns. Thus, we
conducted our data analysis on the remaining 1,899 tax returns: 957
from the universe of basic EITC only recipients and 942 from the
universe of those who received both the EITC and the health insurance
credits.
Estimates listed in this report show point estimates developed on the
basis of random sampling. Because we used random samples of returns
from the universe of taxpayers who received the EITC in 1991, we also
computed sampling errors to assess the reliability of results. Point
estimates by themselves are not adequate representations of
statistical results because the population value being estimated can
fall within a range around the point estimate. Accordingly, we
calculate range estimates, also known as confidence intervals, to
indicate the precision of the estimate. The range estimate,
designated as the confidence interval, is computed by adding and
subtracting the sampling error from the point estimate.
For the point estimates displayed in this report, the sampling error
amounts to less than plus or minus 10 percent of the estimate at the
95 percent confidence level, unless otherwise stated. For example, a
point estimate of 26 percent has an associated sampling error of less
than plus or minus 2.6 percentage points. Thus, we can say that we
are 95-percent confident that the population value being estimated is
somewhere in the range between 23.4 percent and 28.6 percent. There
is a 5-percent chance that the value being estimated is outside the
stated range for the populations of basic EITC only or health
insurance credit recipients.
ANALYSIS OF TAX RETURN DATA
--------------------------------------------------------- Appendix I:2
We used logit analysis, a form of loglinear modeling, to test
associations between the independent variables (taxpayer demographic
characteristics) and the outcome of whether or not the taxpayer
received the health insurance credit. We performed statistical
significance tests to see whether the variables differed between the
samples of basic EITC only and health insurance credit recipients and
computed expected frequencies for each taxpayer characteristic to
determine which factors appeared to be influential in determining
whether a family received the health insurance credit in 1991. We
then computed odds ratios to determine the likelihood of a taxpayer
receiving the health credit, given the various demographic
characteristics (such as number of dependents, taxpayer filing
status, etc.).
The objective of the multivariate analysis was to determine which
variables, after we controlled for the effects of other variables,
had statistically significant relationships with the outcome. To
accomplish this, we compared a series of logit models that allowed
for associations among the independent variables but varied in terms
of the effects of the independent variables on the outcome. We began
with a base model that postulated no association between the
variables and the outcome, and we built a series of hierarchical
models varying one variable at a time. For each model tested, we
chose the one that fit most closely with the data. In the final
model, the only two variables that showed a significant effect were
adjusted gross income and number of employers. The expected
frequencies obtained from the preferred model were used to estimate
the odds on receiving the health credit and the odds ratios to
indicate how greatly those odds varied by income and number of
employers. Using the odds ratios, we were able to determine to what
extent one outcome was more likely than another given a set of
demographic characteristics (e.g., how much more likely one taxpayer
was to receive the health credit than another, given their AGIs).
Table I.1 shows the expected case frequencies, odds, and odds ratios.
Table I.1
Expected Case Frequencies, Odds, and
Odds Ratios Based on the AGI, Number of
Employers Model, and Filing Status
EITC & Odds on Odds
Taxpayer health receiving ratios: Odds
Adjusted gross Number of filing EITC only credit health employer ratios:
income employers status cases cases credit effect AGI effect
-------------- ------------ ---------- ---------- ---------- ---------- ---------- ----------
<$10,000 0\a HH 15.516 8.484 .5468 1.52 1.00
MFJ 16.163 8.837 .5468 1.52 1.00
1 HH 157.472 75.528 .4796 1.33 1.00
MFJ 58.799 28.201 .4796 1.33 1.00
2 HH 77.359 26.641 .3444 .96 1.00
MFJ 20.084 6.916 .3444 .96 1.00
3 HH 46.336 16.664 .3596 1.00 1.00
MFJ 24.271 8.729 .3596 1.00 1.00
>=$10,000 0\a HH 10.076 18.924 1.8782 1.52 3.43
MFJ 14.245 26.755 1.8782 1.52 3.43
1 HH 192.255 316.745 1.6475 1.33 3.44
MFJ 83.474 137.526 1.6475 1.33 3.44
2 HH 66.424 78.576 1.1830 .96 3.43
MFJ 64.133 75.867 1.1830 .96 3.43
3 HH 24.157 29.843 1.2354 1.00 3.44
MFJ 53.235 65.765 1.2354 1.00 3.44
----------------------------------------------------------------------------------------------------
Legend
HH=Head of household
MFJ=Married filing jointly
Note 1: The numbers displayed in this table are expected frequencies
based on the preferred model from the categorical data analysis.
Note 2: The observed data that gave rise to these expected
frequencies excluded 45 (out of 1,899 total) cases involving persons
whose filing status was `single,' as their number was too small to
include in these multivariate analyses.
\a Majority with 0 employers had a Schedule C or F attached to the
return indicating the taxpayer was self-employed. Other cases in
this group did not indicate the source of their income.
Source: GAO samples of 1,899 EITC recipients in tax year 1991.
According to this model, an EITC recipient was 3.4 times more likely
to receive the health insurance credit if the taxpayer's AGI was
$10,000 or more than if the AGI was less than $10,000. EITC
recipients with a single employer were also 1.3 times more likely to
claim the health insurance credit than taxpayers showing multiple
employers (regardless of the taxpayer's AGI).
USE OF CPS AND IRS DATA TO
ESTIMATE PARTICIPATION RATE
--------------------------------------------------------- Appendix I:3
We used the 1992 CPS, containing 1991 data, for information on the
uninsured population and the number of families who appeared to be
eligible for the health insurance credit in 1991. As mentioned
previously the CPS is conducted monthly by the Census Bureau and
collects primarily labor force data on the civilian noninstitutional
population. CPS interviewers poll about 60,000 households across the
country regarding their income, employment, and other issues
throughout each year. In the March supplement to the annual survey,
interviewers ask questions regarding families' health insurance
status. The CPS provides a cross-sectional ("snapshot") view of
families' insurance status; the survey does not provide information
on changes in health insurance status over the previous year or on
the duration of coverage. A positive response to the CPS March
supplement questions regarding health insurance coverage denotes at
least some coverage during the previous year.\1
We obtained a tabulation of the 1992 CPS from the Employee Benefit
Research Institute (EBRI) showing that 8.8 million families appeared
to meet the eligibility requirements for the health insurance credit
in 1991. This figure reflected the number of families that had a
worker; reported an AGI under $21,250 (the eligibility ceiling for
1991); had at least one child under age 18; and reported having
private health insurance at some time in 1991. These characteristics
do not match exactly with the IRS eligibility criteria because CPS
and IRS do not use the same definitions of key variables (such as
taxpayers and families), and they track demographic characteristics
somewhat differently.
Several caveats must be considered in using CPS data to estimate the
size of the population eligible for the health insurance credit and
in comparing CPS figures to IRS data on the population that actually
received the credit. First, sampling errors must be considered in
using the CPS (or any other sample survey). Interviewing techniques,
the way in which various answers are interpreted, respondents'
inability to recall information, errors in imputing missing
responses, and errors in tabulating and processing responses account
for nonsampling variability, the extent of which is not known. Some
researchers believe that these errors have resulted in what appears
to be underreporting of income in the CPS, particularly in relation
to IRS estimates of income. Unreported income in CPS data would
result in overestimation of the size of the population eligible for
the health insurance tax credit, which would cause the participation
rate estimate to be too low. Second, CPS counts families in such a
way that its figure may significantly underrepresent the number of
families or taxpayers eligible for the health insurance credit. CPS
counts all related members of a household as one family, although
there may be more than one taxpaying unit for EITC purposes in each
CPS family.\2 This would cause a participation rate estimate to be
too high. Third, CPS defines children (either "own" or "related") as
those under age 18 and uses no residency test.\3 In determining
eligibility for the health insurance credit, the Code defines
qualifying children as those under age 19 (or under age 24 if
full-time students) who lived in the household for more than 6 months
of the tax year (12 months for eligible foster children) that bear a
certain relationship to the taxpayer. The impact of these
differences on the estimated size of the eligible population is
unknown. Fourth, CPS does not track whether insurance coverage is
extended to children, nor whether the householder paid a premium for
coverage. As a result, the CPS figure may overestimate the eligible
population because only those otherwise-eligible taxpayers who
actually paid a premium for health insurance and extended coverage to
one or more qualifying children would be eligible to claim the health
insurance credit.\4 To the extent this overestimates the eligible
population the estimated participation rate would be too low.
Finally, what CPS considers to be household income differs slightly
from the definition of earned income used in determining EITC and
health
insurance credit eligibility.\5 However, the CPS definition of income
fits closely with the definition of AGI. Thus, we believe the
different definitions of income are immaterial in the comparison of
CPS and IRS data for our purposes. Although we know there are biases
in the various data sources, we were unable to determine their net
effect.
--------------------
\1 However, some researchers believe that those surveyed may respond
to the question with information concerning their current health
insurance status rather than their status during the survey period
(the previous year).
\2 CPS defines families as two or more individuals living together in
a household who are related by birth, marriage, or adoption. One
family member is the householder. Two or more people living in the
same household who are related to one another, but not related to the
householder, form an unrelated subfamily and are not included in the
count of families by CPS. If a subfamily is related to the
householder (e.g., a young married couple living with the husband's
or wife's parents, or the daughter and grandchild of the
householder), the members of the related subfamily are counted as
members of the householder's family. Households are defined as all
the persons who occupy a housing unit. A household includes the
related family members and all unrelated persons living in the
household.
\3 CPS defines "own" children in a family as sons and daughters,
including stepchildren and adopted children, of the householder.
"Related" children include "own" children and all other children
living in a household who are related to the householder by blood,
marriage, or adoption. CPS does not require that a child live in a
household for a certain period of time to be considered a child of
that householder.
\4 Statistics from the Health Insurance Association of America, "The
Health Insurance Tax Credit and Medicaid Expansion: Eligible
Populations" (Washington, D.C., 1991) indicate that in 1990
approximately 25 percent of those working for employers who offered
health benefits paid nothing for their family coverage. If this
holds true for the low-income population, our EBRI estimate of the
eligible population would have to be reduced by 25 percent (because a
taxpayer who paid nothing for coverage is ineligible for the health
credit). HIAA data also show that the low-income population tends to
work for employers (especially small firms) and in industries (retail
and service sectors, for example) that are less likely to offer
health benefits to employees.
\5 For the purposes of EITC and Health Credit eligibility, earned
income includes the following: wages, salaries and tips; union
strike benefits; long-term disability benefits; self employment
earnings; voluntary salary deferrals; U.S. combat pay and military
subsistence allowance; meals and lodging provided by employers;
anything of value received for services performed. CPS data include
all of the above as income, as well as accrued interest on retirement
accounts.
ADMINISTRATION OF POSTCARD
SURVEY
--------------------------------------------------------- Appendix I:4
In May and June 1993 (including follow-up), we sent a postcard survey
to 1,839 taxpayers (928 basic EITC only recipients and 911 health
insurance credit recipients) from our samples of tax year 1991 tax
returns.\6 The postcard survey asked two two-part questions of
taxpayers: (1) At any time during 1991 or 1992, did any children
living in your home (age 18 or younger) have health insurance? and
(2) In 1991 and in 1992, were you employed? The possible answers to
question 1 were as follows: health coverage bought through my/my
spouse's employer; health coverage bought by me/my spouse on our own;
or no health insurance. For question 2, possible answers were as
follows: employed mostly full-time; employed mostly part-time; or
mostly unemployed. We sent out 2 mailings of the postcard survey,
the first to all 1,839 taxpayers and the second to those who did not
respond to the first mailing. We received responses from 810
taxpayers: 366 (39.4 percent) from our sample of EITC recipients and
444 (48.7 percent) from our sample of EITC only and health insurance
credit recipients. (See tables I.2 through I.5.)
The overall response rate of the postcard survey was 44 percent,
which was not sufficient to allow us to project the survey results
over the entire population of EITC only and health insurance credit
recipients. Therefore, we did not weight the responses in order to
estimate what the results would indicate for the sample as a whole.
However, we found that the characteristics of the population that
responded to the postcard survey were for the most part similar to
the characteristics of the 1,839 taxpayers in our sample population.
One exception pertains to AGI: We found that the postcard respondent
population of the basic EITC only recipients had a higher average AGI
than the sample population overall. The postcard respondent
population showed an average AGI of $11,806, compared to $10,799 for
the entire basic EITC only sample population.
Table I.2
Postcard Survey Responses for 1991, for
Taxpayers Who Received the Basic EITC
Only
Source of Employed Employed
family mostly mostly Mostly
coverage full-time part-time unemployed Total
------------ ---------- ---------- ---------- ==========
Employer 80 16 8 104
sponsored
Purchase 20 1 3 24
own
coverage
No health 148 43 36 227
insurance
============================================================
Total 248 60 47 355\a
------------------------------------------------------------
\a Eleven EITC only respondents did not answer all questions and are
not captured in this matrix.
Source: GAO postcard survey.
Table I.3
Postcard Survey Responses for 1992, for
Taxpayers Who Received the Basic EITC
Only
Source of Employed Employed
family mostly mostly Mostly
coverage full-time part-time unemployed Total
------------ ---------- ---------- ---------- ==========
Employer 83 14 7 104
sponsored
Purchase 18 5 3 26
own
coverage
No health 137 40 50 227
insurance
============================================================
Total 238 59 60 357\a
------------------------------------------------------------
Nine EITC only respondents did not answer all questions and are not
captured in this matrix.
Source: GAO postcard survey.
Table I.4
Postcard Survey Responses for 1991, for
Taxpayers Who Received the EITC and
Health Insurance Credit
Source of Employed Employed
family mostly mostly Mostly
coverage full-time part-time unemployed Total
------------ ---------- ---------- ---------- ==========
Employer 219 10 22 251
sponsored
Purchase 96 10 10 116
own
coverage
No health 34 15 23 72
insurance
============================================================
Total 349 35 55 439\a
------------------------------------------------------------
\a Five health insurance credit respondents did not answer all
questions and are not captured in this matrix.
Source: GAO postcard survey.
Table I.5
Postcard Survey Responses for 1992, for
Taxpayers Who Received the EITC and
Health Insurance Credit
Source of Employed Employed
family mostly mostly Mostly
coverage full-time part-time unemployed Total
------------ ---------- ---------- ---------- ==========
Employer 210 10 15 235
sponsored
Purchase 90 9 8 107
own
coverage
No health 46 16 35 97
insurance
============================================================
Total 346 35 58 439\a
------------------------------------------------------------
\a Five health insurance credit respondents did not answer all
questions and are not captured in this matrix.
Source: GAO postcard survey.
--------------------
\6 We did not send postcard surveys to 60 of the 1,899 sample
taxpayers used in our analysis because they were under IRS audit at
the time of our review.
SUPPLEMENTAL TABLES
========================================================== Appendix II
The tables in this appendix provide additional details on GAO's
analysis of the EITC and health insurance credit recipient
populations. We provide this data based on their potential
usefulness in the ongoing health care reform debate. These tables
are based on our two nationwide statistical random samples of tax
year 1991 tax returns. The EITC only population sample consisted of
957 tax returns, and the health insurance credit sample consisted of
942 tax returns. Our sample data is projectable over the 1991 EITC
only population of about 11 million families, or the 1991 health
insurance credit population of about 2.2 million families. Data in
these tables can be projected to the appropriate population at a 95
percent confidence level with associated sampling errors of less than
plus or minus 10 percent, unless otherwise stated.
Table II.1 presents information pertaining to the two samples. The
demographic characteristics are presented for each sample population
separately and include data on adjusted gross income stratified in
$5,000 increments; type of return filed; preparer of sample tax
returns; filing status of sample taxpayers; number of employers shown
on sample tax returns; number of sample taxpayers who filed
supplemental IRS schedules or forms; and the number of sample returns
analyzed from each IRS region.
Table II.1
Demographic Characteristics of EITC and
Health Insurance Credit Recipients in
GAO Samples
(Tax year 1991)
Demographic
characteristic Number Percent Number Percent
-------------------- -------- -------- -------- --------
Adjusted gross income:
------------------------------------------------------------
Less than $5,000 169 17.7 36 3.8
$5,000 to $9,999 274 28.6 145 15.4
$10,000 to $14,999 283 29.6 330 35.0
$15,000 or more 231 24.1 431 45.8
============================================================
Total 957 100.0 942 100.0
Type of return filed:
------------------------------------------------------------
1040 (paper return) 250 26.1 285 30.3
1040A 367 38.4 294 31.2
1040 (electronic) 340 35.5 363 38.5
============================================================
Total 957 100.0 942 100.0
Preparer of return:
------------------------------------------------------------
Taxpayer/spouse 446 46.6 472 50.1
Paid preparer 505 52.8 465 49.4
Other 6 0.6 5 0.5
============================================================
Total 957 100.0 942 100.0
Filing status:
------------------------------------------------------------
Married 325 34.0 368 39.1
Head of household 599 62.6 562 59.7
Single 33 3.5 12 1.3
============================================================
Total 957 100.0 942 100.0
Number of employers:
------------------------------------------------------------
None shown 57 6.0 63 6.7
One 513 53.6 567 60.2
Two 236 24.7 191 20.3
Three 72 7.5 68 7.2
Four 56 5.9 32 3.4
Five 9 0.9 32 3.4
More than five 14 1.5 7 0.7
============================================================
Total 957 100.0 942 100.0
IRS Region:
------------------------------------------------------------
North-Atlantic 77 8.0 85 9.0
Mid-Atlantic 97 10.1 98 10.4
Southeast 237 24.8 314 33.3
Central 90 9.4 98 10.4
Southwest 160 16.7 135 14.3
Midwest 93 9.7 121 12.9
Western 203 32.3 91 9.7
============================================================
Total 957 100.0 942 100.0
Schedules filed:\a
------------------------------------------------------------
Schedule EIC 903 94.4 930 98.7
Schedule A 50 5.2 66 7.0
Schedule C or F 107 11.2 164 17.4
Form 2441 78 8.2 172 18.3
------------------------------------------------------------
Note: Percentages may not add to 100 due to rounding.
\a Taxpayers may file any or all of these schedules. Therefore, the
figures do not add to 100 percent.
Source: GAO samples of 1,899 EITC recipients in tax year 1991.
Table II.2 shows the averages of each of the key variables in the two
samples. The most significant difference between the two groups was
AGI; however, the number of employers was also statistically
significant.
Table II.2
Averages of Key Variables
(Tax year 1991)
EITC only Health credit
Demographic characteristic population population
------------------------------ ------------- -------------
Adjusted gross income $10,799 $14,019
Number of children claimed 1.4 1.5
Number of dependents claimed 1.7 1.7
Number of employers 1.6 1.5
Number of W-2 forms submitted 1.7 1.5
Credits received:
------------------------------------------------------------
Basic EITC $784 $719
Health insurance credit \a $233
Young child credit $228 $234
Child care credit $500\b $443
Cost of health coverage \c $1,029
claimed
------------------------------------------------------------
Note: Averages based on sample data.
\a EITC only population did not receive the health insurance credit.
\b The sampling error for the average child care credit received by
EITC recipients is plus or minus 14 percent.
\c EITC only population did not report health coverage costs.
Source: GAO samples of 1,899 EITC recipients in tax year 1991.
Table II.3 shows AGI in $5,000 increments and the proportion of each
sample that fell into the AGI categories, by IRS region. The highest
and lowest average AGIs for both the EITC only and health insurance
credit recipient populations in tax year 1991 were found in the IRS
North-Atlantic and Southeast regions, respectively. Despite regional
variation, the difference between the AGIs reported by the EITC only
and health insurance credit populations was statistically significant
at the national level as well as in each individual region. Sampling
errors associated with regional estimates are all less than plus or
minus 11.5 percent for the EITC population and 9 percent for the
health insurance credit population.
Table II.3
Adjusted Gross Income Distribution for
Basic EITC Only and Health Insurance
Credit Recipients in GAO Samples, by IRS
Region
(Tax year 1991)
Region
and Percen Percen Percen Percen
credit Number t Number t Number t Number t Number
-------- ------ ------ ------ ------ ------ ------ ------ ------ ------
Sample total
--------------------------------------------------------------------------------
EITC 169 17.7 274 28.6 283 29.6 231 24.1 957
Health 36 3.8 145 15.4 330 35.0 431 45.8 942
credit
North-
Atlantic
EITC 13 16.9 17 22.1 23 29.9 24 31.2 77
Health 2 2.4 12 14.1 23 27.1 48 56.5 85
credit
Mid-
Atlantic
EITC 14 14.4 32 33.0 23 23.7 28 28.9 97
Health 2 2.0 11 11.2 45 45.9 40 40.8 98
credit
Southeas
t
EITC 45 19.0 72 30.4 73 30.8 47 19.8 237
Health 12 3.8 59 18.8 115 36.6 128 40.8 314
credit
Central
EITC 17 18.9 24 26.7 26 28.9 23 25.6 90
Health 5 5.1 13 13.3 34 34.7 46 46.9 98
credit
Southwes
t
EITC 29 18.1 41 25.6 54 33.8 36 22.5 160
Health 5 3.7 25 18.5 41 30.4 64 47.4 135
credit
Midwest
EITC 16 17.2 22 23.7 26 28.0 29 31.2 93
Health 7 5.8 16 13.2 37 30.6 61 50.4 121
credit
Western
EITC 35 17.2 66 32.5 58 28.6 44 21.7 203
Health 3 3.3 9 9.9 35 38.5 44 48.4 91
credit
--------------------------------------------------------------------------------
Source: GAO samples of 1,899 EITC recipients in tax year 1991.
The difference between the incomes of taxpayers who received the EITC
only and those who received the health insurance credit is
illustrated by looking at the two groups in terms of federal poverty
levels (see table II.4). Fifty-four percent of the EITC only
population fell below the federal poverty level in 1991, compared
with 29 percent of the health insurance credit recipient population.
Eighty-three percent of the EITC only population lived below 150
percent of the poverty level in 1991, compared with 69 percent of the
health insurance credit population (see table II.6).
Because of the difference in magnitudes of the potential credit, the
EITC (maximum $1,235) had a much greater impact on taxpayers' after
tax AGIs than the health insurance credit (maximum $428). Unlike the
EITC, which benefits many recipients by raising their after tax AGIs
above the poverty level, the health insurance credit was not large
enough to have such an impact. When we included the EITC amount in
taxpayers' AGIs, the percentage of those living below the poverty
level declined (see table II.4). The EITC brought 7 percent of both
EITC only and health insurance credit recipients (or nearly 1 million
taxpayers) above the federal poverty level in 1991. The health
insurance credit only decreased slightly the percentage of health
insurance credit recipients living below poverty. One percent of
health insurance credit recipients (or about 22,000 taxpayers) were
brought above the poverty level as a result of the health insurance
credit. Adding the supplemental young child and child and dependent
care credits to taxpayers' AGIs, however, did not increase AGIs
enough to lift the status of any of those in our sample above the
poverty level.
Table II.4
Impact of EITC and Supplemental Credits
on Recipients' Poverty Status
(Tax year 1991)
Percent of EITC Percent of health
only population insurance credit
below poverty population below
level poverty level
-------------------- ------------------ ------------------
AGI 54 29
AGI+EITC 47 22
AGI+EITC+HIC\a 47 21
AGI+EITC+HIC\a+YCC 47 21
AGI+EITC+HIC\a+YCC+
CARE 47 21
------------------------------------------------------------
Legend
HIC=health insurance credit
YCC=young child credit
CARE=child and dependent care credit
\a EITC only population did not receive the health insurance credit.
Source: GAO samples of 1,899 EITC recipients in tax year 1991.
Table II.5 shows the national weighted average poverty thresholds by
family size at the poverty level and at 150 percent of the poverty
level. Table II.6 shows the percentage of EITC and health insurance
credit recipients that fall below these poverty levels by IRS region.
The incomes are based on taxpayers' reported AGIs and do not include
any benefits received from the various EITC credits.
Table II.5
1991 National Weighted Average Poverty
Thresholds
150 percent of the
Weighted poverty weighted poverty
Family size threshold threshold
-------------------- ------------------ ------------------
One $6,932 $10,398
Two 8,865 13,298
Three 10,860 16,290
Four 13,924 20,886
Five 16,456 24,684
Six 18,587 27,881
Seven 21,058 31,587
Eight 23,605 35,408
Nine or more 27,942 41,913
------------------------------------------------------------
Source: Bureau of the Census, 1992 Current Population Survey (CPS).
Table II.6
Percentage of EITC Only and Health
Credit Recipient Populations Below the
Federal Poverty Thresholds
(Tax year 1991)
Percent Percent Percent Percent
of EITC of health of EITC of health
only credit only credit
populatio populatio populatio populatio
Region n n n n
---------------- --------- --------- --------- ---------
North-Atlantic 42 25 69 60
Mid-Atlantic 50 22 78 64
Central 54 35 79 70
Southeast 54 29 85 71
Southwest 54 35 87 74
Midwest 46 30 77 72
Western 64 21 89 60
Sample average 54 29 83 69
------------------------------------------------------------
Note: Sampling errors for regional figures given in the above table
are all less than plus or minus 11.5 percent.
Source: GAO samples of 1,899 EITC recipients in tax year 1991.
The reported cost of insurance and the health insurance credit
reimbursement rate both varied by region (see tables II.7 and II.9).
Taxpayers in the IRS North-Atlantic region paid the highest average
premiums: $1,254, representing 9 percent of taxpayers' AGIs in that
region.\1 However, because these taxpayers also had the highest
average AGI in the nation, their average health insurance credit was
low (only $209), yielding less than a 17 percent reimbursement rate.
Taxpayers in the IRS Southeast region showed the lowest average
health care premiums: $924, representing 7 percent of taxpayers'
AGIs in that region. Conversely, because taxpayers in the Southeast
region showed the lowest average AGI, their average health insurance
credit was high ($250), yielding the highest reimbursement rate of 27
percent.
Table II.7 shows how the average reported cost of health insurance,
health insurance credit amount, and the health insurance credit
reimbursement rate varied by IRS region.
Table II.7
Average Adjusted Gross Income, Reported
Health Coverage Cost, Health Insurance
Credit Amount, and Reimbursement Rates
for Health Insurance Credit Recipients,
by IRS Region
(Tax year 1991)
Percentag
Average Average e of cost
Average coverage health reimburse
Region AGI cost credit d
---------------- --------- --------- --------- ---------
North-Atlantic $14,707 $1,254 $209 16.6
Mid-Atlantic 14,359 1,112 249 22.4
Southeast 13,602 924 250 27.1
Central 14,055 963 219 22.7
Southwest 13,922 938 227 24.2
Midwest 14,028 1,215 220 18.1
Western 14,544 1,049 223 21.3
------------------------------------------------------------
Note: The sampling errors associated with the average cost of
coverage and average health insurance credit amount in each region
are all less than plus or minus 22 percent.
Source: GAO samples of 1,899 EITC recipients in tax year 1991.
Table II.8 shows sample taxpayers' reported health insurance costs by
AGI (in $500 and $5,000 increments, respectively). Over 60 percent
of health insurance credit recipients in our sample reported costs of
less than $1,000, and 88 percent reported costs less than $2,000 in
1991.
Table II.8
Reported Health Insurance Costs for
Health Insurance Credit Recipients in
GAO Sample, by AGI
(Tax year 1991)
Less $500 $1,000 $1,500 More
than to to to than Total\
AGI $500 $999 $1,499 $1,999 $2,000 a
------------ ------ ------ ------ ------ ------ ======
Less than 24 5 0 3 4 36
$5,000
66.7 13.9 0.0 8.3 11.1 100.0
Percent
$5,000 to 62 38 17 12 16 145
$9,999
42.8 26.2 11.7 8.2 11.0 100.0
Percent
$10,000 to 96 92 71 30 41 330
$14,999
29.1 27.9 21.5 9.1 12.4 100.0
Percent
$15,000 to 95 115 82 27 43 362
$19,999
26.2 31.8 22.7 7.5 11.9 100.0
Percent
$20,000 or 26 19 10 4 10 69
more
37.7 27.5 14.5 5.8 14.5 100.0
Percent
============================================================
Total 303 269 180 76 114 942
number\a
32.2 28.6 19.1 8.1 12.1 100.0
Total
percent
------------------------------------------------------------
\a Percentages may not add to 100 due to rounding.
Source: GAO sample of 942 health insurance credit recipients in tax
year 1991.
Table II.9 shows sample taxpayers' reported health insurance costs in
$500 increments, by IRS region.
Table II.9
Reported Health Insurance Costs for
Health Insurance Credit Recipients in
GAO Sample, by IRS Region
(Tax year 1991)
Less $500 $1,000 $1,500 More
than to to to than Total\
IRS Region $500 $999 $1,499 $1,999 $2,000 a
------------ ------ ------ ------ ------ ------ ======
North- 24 27 11 2 21 85
Atlantic
28.2 31.8 12.9 2.4 24.7 100.0
Percent
Mid- 28 26 23 9 12 98
Atlantic
28.6 26.5 23.5 9.2 12.2 100.0
Percent
Central 42 23 15 7 11 98
Percent 42.9 23.5 15.3 7.1 11.2 100.0
Southeast 97 101 67 26 23 314
Percent 30.9 32.2 21.3 8.3 7.7 100.0
Southwest 42 43 26 11 13 135
Percent 31.1 31.9 19.3 8.2 9.6 100.0
Midwest 34 29 23 11 24 121
Percent 28.1 24.0 19.0 9.1 19.8 100.0
Western 36 20 15 10 10 91
Percent 39.6 22.0 16.5 11.0 11.0 100.0
============================================================
Total\a 303 269 180 76 114 942
Percent 32.2 28.6 19.1 8.1 12.1 100.0
------------------------------------------------------------
\a Percentages may not add to 100 due to rounding.
Source: GAO sample of 942 health insurance credit recipients in tax
year 1991.
Table II.10
Distribution of the Universe of EITC Tax
Returns, by IRS Region
(Tax year 1991)
Basic Health
EITC insuranc
only e credit
Universe returns Percent returns Percent Total
---------- -------- -------- -------- -------- ========
Southeast 2,776,82 25.3 714,172 31.8 3,491,00
9 1
Western 2,090,40 19.0 231,537 10.3 2,321,94
3 0
Southwest 1,796,31 16.4 339,913 15.1 2,136,22
3 6
Mid- 1,135,59 10.3 235,977 10.5 1,371,57
Atlantic 8 5
Central 1,116,98 10.2 252,200 11.2 1,369,18
4 4
Midwest 1,087,53 9.9 291,460 13.0 1,378,99
4 4
North- 978,531 8.9 181,773 8.1 1,160,30
Atlantic 4
============================================================
Total 10,982,1 100.0 2,247,03 100.0 13,229,2
92 2 24
------------------------------------------------------------
Source: IRS supplied data and IRS Annual Report for 1991.
Table II.11
Distribution of GAO's Sample of EITC Tax
Returns, by IRS Region
(Tax year 1991)
Basic Health
EITC insuranc
only e credit
Sample returns Percent returns Percent Total
---------- -------- -------- -------- -------- ========
Southeast 237 24.8 314 33.3 551
Western 203 21.2 91 9.7 294
Southwest 160 16.7 135 14.3 295
Mid- 97 10.1 98 10.4 195
Atlantic
Central 90 9.4 98 10.4 188
Midwest 93 9.7 121 12.9 214
North- 77 8.0 85 9.0 162
Atlantic
============================================================
Total 957 100.0\a 942 100.0 1,899
------------------------------------------------------------
\a Detail does not add to total due to rounding.
Source: GAO samples of 1,899 EITC recipients in tax year 1991.
Table II.12
Individual Tax Returns Filed, by IRS
Region
(Tax year 1991)
EITC and
health
insurance Basic EITC Health
Individual credit only credit
IRS region returns returns returns returns
------------ ---------- ---------- ---------- ----------
Southeast 21,417,538 3,491,001 2,776,829 714,172
Western 19,321,596 2,321,940 2,090,403 231,537
Southwest 14,800,011 2,136,226 1,796,313 339,913
Mid- 15,154,014 1,371,575 1,135,598 235,977
Atlantic
Central 14,038,934 1,369,184 1,116,984 252,200
Midwest 14,932,423 1,378,994 1,087,534 291,460
North- 14,301,480 1,160,304 978,531 181,773
Atlantic
============================================================
Total 113,965,99 13,229,224 10,982,192 2,247,032
6
------------------------------------------------------------
Source: IRS supplied data and IRS Annual Report for 1991.
Table II.13
Percentage of Individual Tax Returns
Filed, by IRS Region
(Tax year 1991)
EITC and
health
insurance Basic EITC Health
credit only credit
Individual percent of percent of percent of
IRS region returns region region region
------------ ---------- ---------- ---------- ----------
Southeast 18.8 16.3 13.0 3.3
Western 17.0 12.0 10.8 1.2
Southwest 13.0 14.4 12.1 2.3
Mid- 13.3 9.1 7.5 1.6
Atlantic
Central 12.3 9.8 8.0 1.8
Midwest 13.1 9.2 7.3 2.0
North- 12.5 8.1 6.8 1.3
Atlantic
============================================================
Total 100.0 11.6 9.6 2.0
------------------------------------------------------------
Source: GAO calculations based on IRS supplied data and IRS Annual
Report for 1991.
--------------------
\1 Estimates showing regional averages of the cost of coverage and
the health insurance credit have associated sampling errors of less
than plus or minus 22 percent in all cases.
MAJOR CONTRIBUTORS TO THIS REPORT
========================================================= Appendix III
GENERAL GOVERNMENT DIVISION,
WASHINGTON, D.C.
Thomas J. McCool, Associate Director, Financial Institutions
and Markets Issues
Michael Brostek, Assistant Director, Tax Policy and Administration
Issues
Mary G. Phillips, Assignment Manager
Thomas M. Beall, Technical Advisor
SAN FRANCISCO REGIONAL OFFICE
Ralph T. Block, Regional Management Representative
George A. Zika, Evaluator-in-Charge
Susan J. Kramer, Site Senior
Hans R. Bredfeldt, Technical Advisor
Samuel H. Scrutchins, Technical Advisor