The Importance of Family-Based Insurance Expansions:
New Research Findings about State Health Reforms

by Leighton Ku and Matthew Broaddus

This report in PDF
Press Release:  HTML or PDF

To access the PDF files, right-click on the underlined text, click "Save Link As," download to your directory, and open document in Adobe Acrobat Reader.

A national consensus has emerged in recent years on the importance of extending publicly-funded health insurance coverage to low-income children under the State Children's Health Insurance Program (SCHIP) and Medicaid. Yet substantial numbers of children eligible for these programs remain uninsured.

This analysis presents the result of new research on whether extending insurance coverage to low-income parents affects enrollment among children. This analysis also reviews recent research that examines other effects of state initiatives to extend eligibility for health insurance to low-income parents. The key findings are:

This research is timely because the federal government and a number of states are considering whether to build upon recent insurance expansions for children by adding coverage expansions for their parents. As a result of recent federal policy changes, states have several options available under which they can institute family-based coverage initiatives that include low-income parents. (See box below.)

In the past two years, 10 states — California, Connecticut, the District of Columbia, Maine, Missouri, New Jersey, New York, Ohio, Rhode Island and Wisconsin — have approved or implemented Medicaid eligibility expansions that cover all members of families with children, including the parents, with incomes up at least 100 percent of the poverty line (and in many cases, up to 185 percent or 200 percent of the poverty line), using a new option for family coverage that the 1996 federal welfare law created. These ten states join five that already had expanded Medicaid eligibility for families by using Medicaid waivers — Delaware, Hawaii, Oregon, Tennessee and Vermont — and two other states with state-funded adult expansions that include parents. (These states are Minnesota and Washington.)

In addition, on July 31, 2000, the Health Care Financing Administration of the U.S. Department of Health and Human Services announced it would begin to approve waivers under the SCHIP program under which states may use SCHIP funds in certain circumstances to extend coverage to the parents of children being uninsured. In states that meet the conditions for these waivers and elect to apply, these waivers open a new avenue for parent coverage expansions.

Furthermore, Congress may consider new initiatives in this area. In July, a major new legislative option, the FamilyCare Act of 2000 (H.R. 4927 and S. 2923), was introduced in the Senate by Senator Edward Kennedy and a bipartisan group of sponsors and by Rep. John Dingell and others in the House. The Administration's budget contains a similar proposal. In a recent vote on the Senate floor, a version of this bill drew support from a majority of senators.(1) This legislation would allow states to expand their SCHIP programs to extend coverage to the parents of children covered under Medicaid and SCHIP and would provide $50 billion in additional federal funding for this purpose between 2002 and 2010. The FamilyCare Act goes substantially beyond the current options by increasing SCHIP funding and permitting the use of the enhanced SCHIP matching rate to extend coverage to parents under either Medicaid or separate state programs.(2)


Insurance Coverage and Uninsurance Rates

Most uninsured low-income children in the nation are now eligible for public insurance coverage. A recent analysis has found that more than 90 percent of uninsured children with incomes below 200 percent of the poverty line are already income-eligible for Medicaid or SCHIP.(3) However, many of the eligible children are not participating and 25 percent of the low-income children (i.e., children below 200 percent of the poverty line) were uninsured in 1998. The major challenge facing policy officials is how to increase the rate of enrollment for children who are already eligible for Medicaid and SCHIP.

Most states are much less generous in offering insurance coverage to the parents of these children, however. In a median state, parental eligibility for Medicaid ends at about 60 percent of the poverty line (about $10,000 for a family of four), about two-thirds lower than the eligibility level for children.(4) The share of low-income parents who are uninsured (34 percent) is substantially higher than the uninsurance rate for children.


Research Findings on the Effect of Parent Expansions on Child Enrollment Rates

In addition to decreasing the proportion of parents who are uninsured, initiatives that expand public insurance coverage of parents may also help stimulate children's enrollment. For example, program administrators in Wisconsin have stated, "The single most important goal of BadgerCare [Wisconsin's program that extends coverage to families, including parents, with incomes up to 185 percent of the poverty line] is to provide health care to uninsured children. We believe that family-based coverage will be more effective than child-only coverage in achieving this goal."(5)

The economic theory behind such a belief is straightforward. Decisions to apply for a health insurance program are made at a family level, generally by the parent(s). In deciding whether to participate, a family must weigh the costs (such as out-of-pocket expenses and time taken off from work to apply) versus the benefits (such as reduced medical care expenses, improved health, and a feeling of security that a family member has insurance). This cost-benefit assessment becomes more appealing if more people in a family can gain coverage through a single application. Covering parents thus ought to increase the probability that the family applies, thereby leading also to greater child enrollment.

To test the effect of broader family coverage on child participation rates, we conducted an analysis to answer a basic question: Have states that undertook broad Medicaid eligibility expansions that included parents increased participation among children to a greater degree than states without such parent expansions? We examined Medicaid participation rates among children under six years of age with family incomes below 133 percent of the poverty line, a group that has been eligible for Medicaid in all states since 1990.(6) If states that instituted broad eligibility expansions that include parents experienced great increases in participation rates among these children, that would suggest that covering parents boosts participation among already-eligible children.

We compared three groups of states:

Children under six with family incomes below 133 percent of the poverty line have been eligible for Medicaid in all states since 1990. As a result, any changes in the participation rate of these children should not be due to changes in their own eligibility but might have been influenced by changes in the eligibility of other family members. The methodology and other technical aspects of this analysis are discussed in the appendix to this paper.

In 1990, before the three states in the first state group had implemented their broad Medicaid expansions, they had a 51 percent participation rate among young low-income children (Figure 1). In other words, 51 percent of the children under six with family incomes below 133 percent of the poverty line were enrolled in Medicaid in these states. This was about the same participation rate as the rate in 1990 in states that did not subsequently adopt a broad expansion.

Figure 1

In 1998, after the broad expansions were in effect in these three expansion states, the child participation rate in these states stood at 67 percent. In the no-expansion states, by contrast, the child participation rate edged up only to 54 percent. In other words, the states with broad, early expansions experienced a 16 percentage point increase in their young child participation rate, while the other states experienced a much-smaller three percentage point increase.(8)

Medicaid participation rates among young children thus grew 13 percentage points more in the early expansion states than in the states without a parent expansion. This difference was statistically significant at a 95 percent confidence level.(9)

Using a similar approach, we also examined changes in Medicaid participation by parents with incomes under 133 percent of the poverty line. We did this to verify that changes in parent participation actually occurred in the early-expansion states. As Figure 2 shows, the percentage of low-income parents enrolled in Medicaid increased by four percentage points in the early-expansion states between 1990 and 1998, while declining by eight percentage points in the states with no expansions.


The decline in Medicaid participation among parents in the states without expansions should not be surprising; it likely is a result of the substantial welfare caseload reductions of recent years. What is striking is that in the early-expansion states, parents' Medicaid enrollment rose despite welfare caseload declines.

The net difference in growth rates for parent participation between the early-expansion states and the no-expansion states is 12 percentage points. This difference is also statistically significant. The 12 percentage-point differential in changes in parent participation closely parallels the 13 percentage-point differential among young children, as described above.

Interpreting These Results

Is it possible that these findings are just a coincidence, caused by factors unrelated to broad Medicaid eligibility expansions that include parents? Perhaps, but the design of this analysis rules out most such possibilities.

First, the early-expansion states and the no-expansion states started out in 1990 with essentially the same child participation rates in Medicaid, as well as with similar parent participation rates. This suggests the states initially were similar in these respects. We measured the 1990-1998 change in participation rates to help control for even the small initial differences. We found that children's participation grew faster in the early-expansion states.

Second, the differences in changes in child participation do not appear to be due to variations in the performance of states' economies. The proportion of the population below the poverty line was similar in the early-expansion and no-expansion states in both 1990 and 1998. (The early-expansion states had an average 15 percent poverty rate in both 1990 and 1998, while the no-expansion states had a 15 percent poverty rate in 1990 and 14 percent in 1998). There were no major differences in the trajectories of these states' economies.

Other analyses have shown that Medicaid participation shrank as states' welfare caseloads fell.(10) This raises the question of whether there were different patterns of welfare caseload declines in the groups of states we compared. Analyses of data from the Census Bureau's Current Population Survey indicate that the number of young low-income children in AFDC or TANF declined about the same amount in the early-expansion states (a 42-percent reduction from 1990 to 1998) and the no-expansion states (a 44-percent reduction). Differences in welfare caseload declines consequently do not explain the variation in the changes in Medicaid participation among young children in these groups of states. Both groups of states experienced large reductions in welfare caseloads.

Still another possibility is that the increase in Medicaid participation among young children might be due to additional publicity surrounding the state expansions or other procedural changes, such as simplified applications that may make it easier to enroll, rather than to the family-based eligibility expansions themselves. It is difficult to disentangle these effects, since state reforms that expand Medicaid eligibility often are accompanied by publicity and new procedures. We believe the best evidence that the broad eligibility expansions themselves led to increases in participation among already-eligible children lies in the fact that the net increase in parents' Medicaid participation in the early-expansion states as compared to the no-expansion states (12 percentage points) mirrors the net increase in young children's participation in these states (13 percentage points). This strongly suggests the linkage of parent and child participation. While children were equally eligible in expansion and no-expansion states, there were sharp differences in eligibility criteria for parents across the states, with the expansion states having much higher income eligibility criteria for parents than the states without expansions.

Moreover, efforts to boost children's enrollment in Medicaid were relatively commonplace across states by 1998 and were not peculiar to the expansion states. For example, in 1998, some 40 states had a mail-in application for children in Medicaid, 40 had eliminated assets tests for children and 41 had simplified their applications for children.(11) It seems unlikely that much of the difference in changes in participation rates among young children can be explained by differences between early-expansion and no-expansion states in practices aimed at boosting enrollment among children. (Nevertheless, it stands to reason that state and local agencies should conduct effective outreach and simplify their enrollment procedures; sound policy requires effective implementation.)

While the findings we present here are not as rigorous as those that might be obtained from a randomized experiment, they offer relatively clear evidence that states can increase the rate of enrollment among children by adopting broad expansions that include parents. Simply stated, covering parents helps expand insurance coverage for children.


Insurance Expansions Can Reduce Uninsurance Levels with Minimal Crowd Out

A different policy issue relating to Medicaid expansions is whether such expansions lead to a reduction in the proportion of adults who are uninsured. If increases in Medicaid coverage are achieved by people dropping private coverage and switching to Medicaid, there will be no net decline in the proportion of people who are insured.

Several recent studies have looked at whether states that have broader Medicaid coverage (or similar state-funded insurance programs) have lower uninsurance rates. These studies have found that, on average, states with broader adult eligibility have lower proportions of uninsured adults than states without such policies.(12) In other words, Medicaid expansions help shrink the ranks of uninsured adults. One of these studies, conducted by Schoen and her colleagues, also found that uninsured adults have more unmet medical needs and lower health care access than adults with Medicaid coverage.

A related area of research involves investigating the "crowd out" problem, or the extent to which Medicaid or SCHIP expansions displace private employer-sponsored insurance. Expanding public insurance would be problematic if all, or a large fraction, of those gaining public coverage simply dropped private, employer-sponsored insurance. It is beyond the scope of this paper to review all of the research concerning crowd out, most of which involves national analyses of the effects of Medicaid child eligibility expansions during the late 1980s and first half of the 1990s. In a recent, comprehensive review of this research, most of the studies indicated that about 10 percent to 20 percent of the gain in Medicaid coverage is offset by a reduction in private coverage.(13) That is, there was an 80 percent to 90 percent net increase in insurance coverage, because most of those who joined the programs were previously uninsured. The number gaining coverage far exceeded the number switching from private insurance, resulting in a substantial net gain in insurance coverage.

Moreover, most studies of this issue examined Medicaid expansions that did not contain anti-crowd-out provisions (such as provisions requiring that people be uninsured before they can enroll). The legislation establishing SCHIP requires states to develop procedures to limit crowd out, and states typically require that children be uninsured prior to enrolling them in separate SCHIP-funded programs. Similarly, the federal government has required states developing demonstration programs like TennCare to include anti-crowd out procedures. Such anti-crowd out policies, which are largely unstudied, ought to reduce further the level of displacement.

Of particular interest here is recent research regarding state programs that expanded family-based coverage. One recent study analyzed TennCare, the largest state insurance expansion of recent years. The study found, using data from the Current Population Survey, that between 1992/93 and 1997/98, the percentage of Tennesseans below 200 percent of the poverty line who had Medicaid coverage climbed from 30 percent to 38 percent, while the percentage of people in that income category who lacked insurance fell from 28 percent to 21 percent. These figures indicate that the increase in Medicaid enrollment was paralleled by a shrinkage in the ranks of the uninsured and suggest that little of the increase in Medicaid enrollment resulted from people switching from private coverage to Medicaid. Although it is a moderately poor state, Tennessee now has one of the lowest percentages of uninsured people of any state in the nation.(14)

Further buttressing this conclusion, the research found that TennCare led to substantial increases in Medicaid coverage without any statistically significant change in private insurance coverage for low-income Tennesseans, indicating there was no significant crowd out.(15) One possible reason for these favorable results is that TennCare had anti-crowd-out rules, requiring that applicants be uninsured before they could join and that families with incomes above the poverty line pay a portion of the TennCare premiums on a sliding-scale basis.

These new findings are consistent with an earlier study of MinnesotaCare, which found that only seven percent of enrollees said they gave up private insurance to join the program: three percent dropped employer-sponsored coverage while four percent dropped nongroup insurance policies.(16) On the other hand, there have been anecdotal reports of crowd-out in Rhode Island's family-based expansion to its RIteCare program.(17) To address these concerns, the state plans to modify its eligibility policies to bar adults who are offered employer-sponsored coverage from RIteCare and to subsidize the purchase of employer-sponsored coverage instead.(18)


Broad State Expansions Can Improve Health Care Access for Adults and Children

The most important question is: Do eligibility expansions that include parents help uninsured people gain better access to health care and improve health care utilization? This question is more complex than it might seem since Medicaid (or SCHIP) eligibility expansions are typically accomplished through the use of managed care plans. While most would expect that insurance would increase health care use, the type of insurance offered might affect access or utilization.

Table 1: Effects of TennCare on Adults' Health Care Use
Selected Health Measures for Adults Newly Covered Adults Uninsured Adults
Percent of women with Pap smear in past year 73.4% 51.4%
Percent who had blood pressure measured in past year 92.2% 74.1%
Percent who needed to see a doctor but did not 33.6% 63.8%
Percent who needed to see a specialist but did not 9.9% 30.5%
Percent who took prescription at lower level than recommended 11.3% 21.9%
Percent with usual source of health care 92.3% 71.0%
Percent who always visit the same provider 69.1% 55.4%
Percent who paid more than $100 out-of-pocket for care in last year 11.9% 23.2%
Source: Moreno and Hoag 2000

A important new study of TennCare by Lorenzo Moreno and Sheila Hoag of Mathematica Policy Research highlights the value of coverage expansions for adults and children alike.(19) The researchers compared adults and children who are covered by the TennCare expansions but would be ineligible under traditional Medicaid eligibility rules with similar low-income uninsured Tennesseans, using rigorous analysis of survey data. Since TennCare was open to both parents and childless adults, the report does not distinguish between parents and childless adults.

The table below recaps a number of the most important findings for adults. All results shown are significant at the 95 percent confidence level and include statistical adjustments for other differences between the newly covered and the uninsured. The differences shown here consequently are attributable to the effects of insurance, not to other underlying differences between these groups.(20)

The study indicates that Medicaid expansions for low- and moderate-income adults can:

The study shows that children benefit from insurance expansions in a similar fashion. Table 2 summarizes the findings relating to children.

Table 2: Effects of TennCare on Children's Health Care Use
Selected Health Measures for Children Newly Covered Children Uninsured Children
Percent with well-child visits on schedule 82.8% 51.3%
Percent of children three or older with a dental check-up in past year 71.2% 54.8%
Percent who needed to see a doctor but did not 5.5% 31.9%
Percent who needed dental care but did not get it 14.9% 29.8%
Percent with usual place of health care 98.3% 73.7%
Percent who always see the same provider 57.3% 39.4%
Percent who paid more than $100 out-of-pocket for care in last year 4.9% 11.6%
Source: Moreno and Hoag 2000

In short, both adults and children gain when provided insurance coverage. These data show that insurance expansions can be particularly helpful for adults, since they tend to have greater medical needs than children and to experience greater difficulty in securing care when uninsured. For example, 65 percent of the uninsured adults in the study did not get medical care when they thought they needed it, as compared to 32 percent of the uninsured children. Similarly, about twice as many uninsured adults as uninsured children had out-of-pocket medical expenses exceeding $100 in the preceding year. Both adults and children benefit when offered insurance, but the burdens of being insured often are more serious for adults.

Findings from Oregon

New research on the effects of Oregon's broad Medicaid expansion — the Oregon Health Plan, which, like TennCare, was launched in 1994 — also is significant. The Oregon initiative extended Medicaid eligibility to uninsured adults and children up to 100 percent of the poverty line. Like TennCare, OHP also involved a shift to mandatory managed care. A distinctive element of Oregon's program was the development of a prioritized list of medical conditions and treatments, which were used to define the benefit package, although in practice there have been very few cases where care was denied because someone needed a low-priority service.

Researchers from Health Economics Research, Inc. have completed preliminary studies comparing OHP recipients with uninsured food stamp recipients.(21) Compared to the uninsured food stamp recipients, adult OHP recipients were:

The researchers found similar positive results for children from the insurance expansions in OHP.

In another part of this study, the researchers compared OHP recipients to privately-insured food stamp recipients. They found no noteworthy differences in health care access or utilization between these two groups. Although OHP recipients were enrolled in managed care plans and subject to rationing under OHP, their health care utilization was similar to that of the group with private insurance. The form of insurance did not matter as much as having any insurance at all.



The research summarized in this report points to three key findings:

These results are based on the experiences of a handful of pioneering states that implemented family-based expansions earlier in the decade and of the hundreds of thousands of people who gained coverage as a consequence. By contrast, some health reform proposals, such as health care tax credits, are based largely on theoretical analyses, with little real-world experience to provide guidance about how to design such programs, administer them or how many people might gain coverage. The Medicaid-based family coverage expansions stand out as road-tested examples of state policy innovations. We are not claiming that programs like TennCare or OHP were ideal; the states had difficulties implementing the new policies, and the programs were sometimes controversial. New programs often have initial problems, but their experiences are instructive and help teach other states how to avoid predictable pitfalls through careful planning and implementation.

Census data indicate that in 1998, there were 7.1 million uninsured parents in families with incomes below 200 percent of the poverty line, and that one-third of all low-income parents — 34 percent — were uninsured. To a large extent, this is because there are substantial gaps in both employer-based health insurance and Medicaid coverage for low-income working parents. Low-wage workers often are in jobs that do not offer insurance. In 1996, only 43 percent of workers earning $7 or less per hour (at that time, slightly more than 100 percent of the poverty line for a family of three with one full-time worker) were offered employer-based insurance, and many workers could not afford to purchase insurance even when offered it.(22) Indeed, working parents with incomes below the poverty line are twice as likely to be uninsured as poor, nonworking parents.(23)

Although most states now provide Medicaid or SCHIP coverage to children with incomes up to 200 percent of the poverty line, eligibility is far less generous for the parents of these children. One-third of the states now cover parents with incomes at or above the poverty line (including those where legislation has been passed but the program has not yet been implemented); the other two-thirds of the states are well below that level. Indeed, the median state Medicaid income eligibility limit for parents is about 60 percent of the poverty line.

The proportion of Americans who lack health insurance has been rising in recent years. Analyses indicate that a major factor in the increasing proportion of people without insurance has been welfare caseload declines, which have lowered Medicaid participation.(24) One of the major functions of family-based Medicaid expansions in states that have instituted them is to help address the loss of insurance coverage that has been an unintended consequence of welfare reform. Helping poor families attain self-sufficiency entails enabling low-wage working parents to secure health insurance rather than going without insurance or having to go on welfare or sink deeper into poverty to obtain it.

Seventeen states have launched initiatives for family-based health insurance expansions that include parents. In most cases, these expansions were financed as Medicaid expansions, with federal matching funds used to augment state funding. Through guidance issued on July 31, 2000, the Department of Health and Human Services has further expanded the range of options for states by indicating that under certain circumstances, it will approve waivers under which SCHIP funds can be applied for parent coverage. The new FamilyCare legislative proposal would enable states to provide family coverage to a much-greater degree, as it would allocate a substantial amount of new federal funding that would be available at an enhanced matching rate for this purpose. It also would permit substantial state flexibility in the use of these funds.

Given the strength of the economy and the presence of sizeable federal and state budget surpluses, there is a window of opportunity for states and the federal government to invest in family health through family-based insurance expansions. The research discussed here indicates that there are practical and tested ways to extend coverage to parents that shrink the ranks of the uninsured and lead to increases in enrollment among children, with the result that access to health care is improved for large numbers of low-income working families.

How States Can Undertake Family Coverage Expansions

  • Medicaid options created by the 1996 welfare law. The 1996 Personal Responsibility and Work Opportunity Reconciliation Act permits states to adopt "less restrictive" methods of computing income or assets in determining the eligibility of families with children. For, example, a state can opt to disregard the amount of income between its previous Medicaid income eligibility limit for families and 100 percent of the poverty line, thus effectively moving the income limit up to the poverty line. These expansions can cover families with children, but not single adults nor childless couples. This new authority has opened the door for a new wave of family-based expansions because these changes do not require special HCFA approval nor budget neutrality.
  • Demonstration project waivers. Until this new Medicaid option was created, the main way states could cover parents was by using special demonstration project waivers, under which HCFA may permit major changes in Medicaid programs. Three important conditions apply: (a) the project must be budget-neutral - that is, it must cost no more to the federal government than the program would otherwise cost; (b) the waiver typically has a five-year time limit, although it may be renewed; and (c) it must be approved by HCFA, which may impose additional requirements that it judges appropriate given the agency's oversight function. These waivers may be combined with parent expansions permitted under the welfare law or under SCHIP expansions, to further customize their programs, such as adding anti-crowd out rules or sliding-scale premiums.
  • SCHIP. On July 31, 2000, HCFA issued guidance about how states may apply for demonstration project waivers to modify their SCHIP programs.(25) The guidance explains how states may use SCHIP funds (at the higher federal matching rate that accompanies such funds) for parent expansions under certain circumstances. The key requirement is that the state must show that it has already made substantial efforts to cover low-income children: it must cover children under age 19 up to 200 percent of the poverty line and must have implemented a number of procedures that make it easier for children to enroll, such as mail-in joint Medicaid/SCHIP applications, 12-month continuous eligibility, elimination of asset tests, or presumptive eligibility for children. States may use regular Medicaid funding to cover parents up to 100 percent of the poverty line and use SCHIP funds to finance eligibility for those at higher levels.
  • State-funded expansions. In the early 1990s, the states of Washington and Minnesota expanded eligibility to parents or childless adults using state funds, without federal matching. Although the states now use Medicaid or SCHIP funds to provide coverage for children and pregnant women, they continue to use state funds to assist parents and childless adults. Earlier this year, New Jersey enacted legislation to expand Medicaid for parents with incomes up to 133 percent of the poverty line. The legislation also provides for use of state funds (including tobacco settlement funds) to cover childless adults up to 100 percent of the poverty line and parents between 133 percent and 200 percent of the poverty line.



Methodology and Technical Discussion of Effects of Broad Expansions
on Young Children's Medicaid Enrollment

Methods. To measure the effects of broad Medicaid expansions on the enrollment of children under six in Medicaid, we used data from the Current Population Survey (CPS), the nationally representative Census survey used most often to track insurance trends. For each of the three group of states described in the main body of this paper, we estimated the percentage of children under six with family incomes below 133 percent of the poverty line for whom Medicaid coverage was reported in 1990 and 1998.(26) We also measured the percentage of parents with incomes below 133 percent of the poverty line who reported Medicaid coverage in these years.

We then measured the change in child and parent participation rates between 1990 and 1998 for each of the three groups of states. We also compared the net difference in the changes in participation rates across these groups of states, comparing the changes in states with early expansions to the changes in states with no expansions and also comparing these states to states with late or non-Medicaid expansions. This assessment of the net difference in trends is sometimes called a "difference in difference" or "pre/post comparison group" design and is considered a relatively rigorous evaluation approach.

Since the CPS is a sample survey, we used statistical methods the Census Bureau recommends to compute standard errors.(27) The standard errors are higher for the early- and late- expansion states than for the no-expansion states, since the population size is smaller in the two groups of expansion states.

There are some pitfalls to the use of the CPS, but the design of this analysis compensates for most of them. First, the CPS undercounts the number of people participating in Medicaid, as compared to administrative data.(28) Second, there was a change in CPS questions in 1994 that slightly altered the reporting of children's health insurance, primarily affecting reporting about children whose health insurance was provided by a nonresident. Since this is a "difference in difference" analysis, however, these problems should cancel each other out, because this change should affect early-expansion states and no-expansion states equally. Even if Medicaid enrollment is undercounted, the comparison across states should still be valid .

Results for children. Table A-1 presents the findings for the early-expansion states (Hawaii, Oregon and Tennessee), the later and non-Medicaid-expansion states, and the states with no broad expansions. In 1990, before there were expansions of eligibility for parents, both the early-expansion and the no-expansion states had a 51 percent participation rate for young low-income children. States with later or non-Medicaid expansions started out with a higher participation rate, 63 percent, indicating they differed at the outset. By 1998, the percentage of young children on Medicaid was 67 percent in the early-expansion states, 65 percent in the late-expansion states, and 54 percent in the no-expansion states. The changes in child participation rates between 1990 and 1998 were 16, 2 and 3 percentage points, respectively.

As seen in Table A-2, the key finding is that young child participation rates grew 13 percentage points more in the early-expansion states than in states with no expansions (16 percentage point growth in early-expansion states minus 3 percentage point growth in no-expansion states = 13 percentage point net difference). The net difference was statistically significant at the 95 percent confidence level. (There was no significant difference in the change in participation for young children in the no-expansion states as compared to the late-expansion states.)

Results for parents. Table A-1 presents similar data about participation by low-income parents. The key finding with regard to parents is that while early-expansion states appeared to have an increase in parents' participation rates from 40 percent to 44 percent, the no-expansion states experienced a decline of 8 percentage points, and the late-expansion states saw a 5-percentage point drop in parent participation rates.(29)

As shown in Table A-2, the net difference in the change in participation rates for the early-expansion as compared to the no-expansion states was 12 percentage points, which was statistically significant. Although most states in the country experienced a reduction in Medicaid participation among low-income parents between 1990 and 1998, the early-expansion states exhibited a quite different pattern, one of modest growth in the proportion of low-income parents enrolled. It seems reasonable to interpret the general reduction in parent participation in other states as being related to reductions in welfare caseloads, which broadly affected Medicaid participation in the late 1990s. The different pattern in the early-expansion states suggests that the broadened parent eligibility criteria that these states adopted both helped to offset some of the effects on Medicaid coverage for parents of the welfare caseload reductions and resulted in more working poor parents being reached and enrolled.


Table A-1:
Changes in Medicaid Participation Rates of Children Under 6 and Parents with Family Incomes Below 133 Percent of the Poverty Line, 1990 to 1998
1990 1998 Percentage-Point   Change
States with Early, Broad Expansions
(Tennessee, Hawaii, Oregon)
% Young Children on Medicaid 50.9% 67.2% 16.3%
     Standard error 4.1% 3.9% 5.7%
% Parents on Medicaid 39.6% 43.8% 4.1%
     Standard error 3.3% 3.3% 4.6%
States with Later or Non-Medicaid Expansion
(Minnesota, Washington, Delaware, Massachusetts, Vermont, New York)
% Young Children on Medicaid 62.9% 65.1% 2.2%
      Standard error 2.3% 2.3% 3.3%
% Parents on Medicaid 50.0% 44.7% -5.3%
      Standard error 1.9% 1.8% 2.6%
States with No Broad Expansions by 1998
(All other states)
% Young Children on Medicaid 50.8% 53.9% 3.2%
      Standard error 0.9% 0.9% 1.3%
% Parents on Medicaid 34.9% 26.8% -8.1%
      Standard error 0.7% 0.6% 0.9%

Table A-2: Comparison of Net Differences in Changes in Participation Rates
for the Three State Groups from 1990 to 1998

Percentage-Point Difference Standard Error (in percentage points) Significance
Early vs. No-Expansion States
Difference in Growth Rate for Children 13.1% 5.9% *
Difference in Growth Rate for Parents 12.3% 4.7% *

Late vs. No-Expansion States

Difference in Growth Rate for Children -1.0% 3.5% n.s.
Difference in Growth Rate for Parents 2.8% 2.8% n.s.
Source: March 1991 and 1999 Current Population Surveys, as analyzed by the Center on Budget and Policy Priorities.

* Difference in growth rates is significant with 95 percent confidence.

Parents' and young children's participation rates in Medicaid grew significantly more between 1990 and 1998 in states with early, broad family-based expansions than in states without such expansions. Since children under six with incomes below 133 percent of the poverty line were eligible for Medicaid in all states throughout this period, this indicates that factors unrelated to children's eligibility Medicaid — and in particular, the expanded coverage for parents — are responsible for the increased Medicaid participation among young children in these states.

End Notes:

1. A version of the FamilyCare Act was offered as an amendment to the marriage penalty tax bill on July 14, 2000. Despite the fact that there was no advance discussion, the amendment received a favorable vote of 51-47. For procedural reasons, however, the amendment required 60 votes and thus did not pass.

2. Under a new policy announced by HCFA, states may seek waivers to use SCHIP funds and enhanced matching rate to extend coverage to parents. These waivers are limited in their impact because both children and parents must be accommodated within existing state SCHIP allocations. The FamilyCare Act would provide substantially more funds to help cover parents and would eliminate the need for states to apply for a time-limited waiver.

3. Matthew Broaddus and Leighton Ku, "More Than 9 Out of 10 Low-Income Uninsured Children Are Now Income-Eligible for Child Health Coverage," Center on Budget and Policy Priorities, forthcoming.

4. Higher parental income limits may apply for 6 to 12 months after leaving welfare due to employment

5. Peggy Bartels and Pris Boroneic, "BadgerCare: A Case Study of the Elusive New Federalism," Health Affairs, 17(6):165-69, Nov./Dec. 1998.

6. The Omnibus Budget Reconciliation Act of 1989 required states to implement this expansion by April 1, 1990. Many states exercised options to expand eligibility to children even before then. National Governors Association Center for Policy Research, MCH Update, State Coverage of Pregnant Women and Children, Jan. 1990 and Jan. 1991. By 1998, many states had increased income eligibility limits for young children beyond 133 percent of the poverty line, but our analysis is confined to children under that level since they were eligible in all states throughout the period this study covers.

7. Although Tennessee froze enrollment of new uninsured applicants during some periods and Hawaii eventually scaled back its eligibility standards, both programs still represent major program expansions, and caseload levels in both states were substantially higher than they had been before these programs began.

8. There was no significant change in the young child participation rate between 1990 and 1998 in the states that had late or non-Medicaid expansions. Some increase in the child participation rate in these states might occur later, but more recent data are not yet available.

9. Since these analyses are based on survey samples, the estimates have a margin of error. The 90-percent confidence interval for the difference between these two groups of states in the increase in participation rates for children under six is from 3 percentage points to 23 percentage points.

10. Leighton Ku and Brian Bruen, "The Continuing Decline in Medicaid Coverage," The Urban Institute, Dec. 1999. Families USA, "Go Directly to Work, Do Not Collect Health Insurance: Low Income Parents Lose Medicaid," June 2000.

11. Donna Cohen Ross and Wendy Jacobson, Free and Low-Cost Health Insurance: Children You Know are Missing Out, Center on Budget and Policy Priorities, 1999, pp. 146-7.

12. Cathy Schoen, Barbara Lyons, Diane Rowland, Karen Davis and Elaine Puleo, "Insurance Matters for Low-Income Adults: Results from a Five-State Survey," Health Affairs, 16(5): 163-71, September/October 1997; Brenda Spillman, "Adults without Health Insurance: Do State Policies Matter?" Health Affairs, 19(4):178-187, July/August 2000.

13. Lisa Dubay, "Expansions in Public Health Insurance and Crowd-Out: What the Evidence Says," Kaiser Project in Incremental Health Reform, Menlo Park: Kaiser Family Foundation, October 1999.

14. Leighton Ku, Marilyn Ellwood, Sheila Hoag, Barbara Ormond and Judith Wooldridge, "The Evolution of Medicaid Managed Care Systems and Eligibility Expansions in Section 1115 Projects," Report to the Health Care Financing Administration from the Urban Institute and Mathematica Policy Research, May 2000, forthcoming in Health Care Financing Review. Also see Christopher Conover and Hester Davies, The Role of TennCare in Health Policy for Low-Income People in Tennessee, The Urban Institute, February 2000.

15. For Tennesseans with incomes between 200 and 400 percent of the poverty line, the rate of nongroup insurance coverage fell. However, the reduction in nongroup coverage was not significantly different from the broader trend of falling nongroup coverage for the nation as a whole. This may have been part of a broader national trend, caused by the general increase in the cost of nongroup insurance policies, rather than a result of the Medicaid expansion.

16. Kathleen Call, et al., "Who Is Still Uninsured in Minnesota? Lessons from State Reform Efforts," Journal of the American Medical Association, 278(14): 1191-95, October 8, 1997.

17. One HMO indicated that many of the people it gained under RIte Care had previously been covered under its commercial policies. Christopher Rowland, "House Passes Bill to Stem RIte Care's Huge Deficit" Providence Journal, June 28, 2000.

18. These changes have been approved by the state legislature but require a waiver that must be approved by HCFA.

19. Lorenzo Moreno and Sheila Hoag, "Covering the Uninsured Through TennCare: Does It Make a Difference?," Report to the Health Care Financing Administration from Mathematica Policy Research, Inc., March 24, 2000. A similar study of Hawaii's QUEST program has been conducted, but results are not yet available for dissemination.

20. The researchers used multivariate statistical models to control for differences in income, employment, health status, education, and other factors that might affect health care use. The authors tested for selection bias (i.e., they tested for the possibility that TennCare recipients had other, unmeasured baseline differences from the uninsured individuals to whom they were compared) and found no evidence this was a problem.

21. Janet Mitchell, Susan Haber, Galina Khatusky and Suzanne Donoghue, "Impact of the Oregon Health Plan on Access and Satisfaction of Low-Income Adults," Health Economics Research, Inc. Draft manuscript, January 2000. Janet Mitchell, Susan Haber, Galina Khatusky and Suzanne Donoghue, "Children in the Oregon Health Plan: How Have They Fared?" Health Economics Research, Inc. Results presented at Association of Public Policy and Management Conference in Washington, DC, November 1999.

22. Philip Cooper and Barbara Schone, "More Offers, Fewer Takers for Employment-Based Health Insurance: 1987 and 1996," Health Affairs, 16(6): 142-49, November/December 1997.

23. Jocelyn Guyer and Cindy Mann, "Employed But Not Insured: A State-by-State Analysis of the Number of Low-Income Working Parents Who Lack Health Insurance," Center on Budget and Policy Priorities, March 1, 1999

24. John Holahan and Johnny Kim, "Why Does the Number of Uninsured Americans Continue to Grow," Health Affairs, 19(4):188-94, July/August 2000.

25. Dear State Health Official letter, signed by Timothy Westmoreland, HCFA, dated July 31, 2000. Available at

26. We used gross income in determining low-income status and did not account for factors such as income disregards or assets tests used in computing Medicaid eligibility; there are no data on states' use of disregards and assets test in 1990. Because we did not have such data for 1990, we did not make such adjustments in either year. Discrepancies in income or assets rules ought to affect the number of children in both the numerator and denominator in roughly equal amounts and probably would have minimal effect on changes in participation rates.

27. Census Bureau and Bureau of Labor Statistics, "Source and Accuracy of the Data for the March 1999 Current Population Survey Microdata File," ( See the authors for more technical detail, if desired.

28. See Ku and Bruen, op cit.

29. The change from 40 percent to 44 percent participation for parents in the early-expansion states was not statistically significant, due to the small sample sizes. However, the key finding — the 12 percentage point difference in changes for early-expansion vs. no-expansion states — was significant at a 95-percent confidence level.