Changes in Medicaid Enrollment Patterns for Children and Their Parents Following Welfare Reform

Transcription

1 Contract No.: (Task Order 2) MPR Reference No.: Changes in Medicaid Enrollment Patterns for Children and Their Parents Following Welfare Reform April 2005 Marilyn Ellwood Angela Merrill Submitted to: Centers for Medicare & Medicaid Services Office of Strategic Planning 7500 Security Boulevard Baltimore, MD Submitted by: Mathematica Policy Research, Inc. 955 Massachusetts Ave., Suite 801 Cambridge, MA Telephone: (617) Project Officer: William Clark Project Director: Marilyn Ellwood

2

3 A CKNOWLEDGEMENTS T he authors would like to thank our current project officer at CMS, William Clark, as well as the former project officer Penelope Pine, who is now retired, for their guidance and support throughout the study. We would also like to thank Karyn Anderson at CMS for thoughtful comments on an earlier draft of this report.. In addition, the paper benefited from the comments of Embry Howell at the Urban Institute and Leighton Ku at the Center for Budget and Policy Priorities. Many people at Mathematica Policy Research, Inc. (MPR) contributed to this report, including Mei-Ling Mason, Robert Schmitz, and Margaret Hallisey.

4 THIS PAGE LEFT INTENTIONALLY BLANK FOR DOUBLE SIDED COPYING

5 C ONTENTS EXECUTIVE SUMMARY... XI STUDY STATES... 2 DATA SOURCES... 3 STUDY LIMITATIONS... 3 KEY ELIGIBILITY GROUPS... 4 TREATMENT OF SCHIP ELIGIBILITY... 6 BACKGROUND... 6 RESULTS... 8 Shifts in Enrollment by Eligibility Group... 8 Changes in Monthly Entry and Exit Rates...11 Changes in Enrollment Duration and Continuity...13 Changes in the Demographic and Other Characteristics of Medicaid Children and Adults...14 Changes in Utilization and Expenditure Patterns...16 SUMMARY AND DISCUSSION...18 FUTURE RESEARCH...22 REFERENCES...25 TABLES FIGURES APPENDIX

6 THIS PAGE LEFT INTENTIONALLY BLANK FOR DOUBLE SIDED COPYING

7 T ABLES Table 1 CHILD MEDICAID ENROLLMENT BY ELIGIBILITY GROUP IN 10 STATES, 1994 AND PERCENT DISTRIBUTION OF CHILD MEDICAID ENROLLMENT BY ELIGIBILITY GROUP IN 10 STATES, 1994 AND ADULT MEDICAID ENROLLMENT BY ELIGIBILITY GROUP IN 10 STATES, 1994 AND PERCENT DISTRIBUTION OF ADULT MEDICAID ENROLLMENT BY ELIGIBILITY GROUP IN 10 STATES, 1994 AND AVERAGE MONTHLY ENTRY AND EXIT RATES FOR MEDICAID IN 10 STATES, 1994 AND MEDICAID ENROLLMENT DURATION AND CONTINUITY AMONG CHILDREN IN 10 STATES, 1994 AND MEDICAID ENROLLMENT AND CONTINUITY AMONG ADULTS IN 10 STATES, 1994 AND DEMOGRAPHIC DISTRIBUTION OF CHILDREN IN MEDICAID IN 10 STATES IN 1999, AND PERCENT DIFFERENCE FROM DEMOGRAPHIC DISTRIBUTION OF ADULTS IN MEDICAID IN 10 STATES IN 1999, AND PERCENT DIFFERENCE FROM CHILD MEDICAID EXPENDITURES AND UTILIZATION IN THREE STATES, 1994 AND ADULT MEDICAID EXPENDITURES AND UTILIZATION IN THREE STATES, 1994 AND 1999

8 THIS PAGE LEFT INTENTIONALLY BLANK FOR DOUBLE SIDED COPYING

9 F IGURES Figure 1 NATIONAL CHILD AND ADULT MEDICAID ENROLLMENT, FFY (IN MILLIONS) 2 RATIO OF 1999 CHILD PUBLIC INSURANCE ENROLLMENT TO 1994 MEDICAID CHILD ENROLLMENT

10 THIS PAGE LEFT INTENTIONALLY BLANK FOR DOUBLE SIDED COPYING

11 E XECUTIVE S UMMARY B etween 1995 and 1999, welfare caseloads of children and their parents fell by almost 50 percent. Medicaid enrollment also declined (although much less so than welfare), even though eligibility criteria for adults remained generally unchanged and, with the inclusion of State Children s Health Insurance Programs (SCHIP), were actually expanded for children. Drops in Medicaid coverage were unexpected and cause for concern, since uninsurance rates for low-income families kept rising during this period. Other research has shown that numerous factors probably contributed to the decline in Medicaid, including federal and state welfare reform efforts, a strong economy and shortcomings in how Medicaid eligibility was administered in the years immediately following passage of federal welfare reform legislation in This study used Medicaid administrative data for children and their parents in 10 states in 1994 and 1999, years that represent the time before and the time after welfare reform, to see whether declines in Medicaid enrollment were accompanied by other eligibility-related changes as well. Key questions included: Did the eligibility groups that families used to qualify for Medicaid shift following welfare reform? How did the dynamics of enrollment compare for both time periods? To what extent were people entering and leaving Medicaid each month? Did enrollment duration and continuity remain the same? Did the demographic characteristics of enrollees change? What happened to service utilization patterns and expenditure levels? The 10 study states were Arkansas, California, Florida, Kentucky, Maine, Michigan, New Jersey, Pennsylvania, Washington, and Wisconsin. State Medicaid Research Files (SMRF) for 1994 and Medicaid Analytic Extract (MAX) files for 1999 were the primary data sources. Supplemental data on SCHIP enrollment was used as well.

12 xii FINDINGS Study findings showed that the declines in Medicaid enrollment following welfare reform were accompanied by changes in other eligibility-related patterns by Key results included: Following welfare reform, the main pathway to Medicaid eligibility for children became the poverty-related groups. Before welfare reform, children (as well as adults) were most likely to qualify for Medicaid because they received cash assistance through the Aid to Families with Dependent Children (AFDC) program. However, following the decline of welfare caseloads, the study states shifted to using various mandatory and optional poverty-related coverage groups, including Medicaid SCHIP (M-SCHIP) and 1115 programs, as the primary route to Medicaid eligibility for children. This occurred probably because the poverty-related income thresholds were higher in 1999, and in most states, the income thresholds were uniform for children ages 1 to 18, in contrast to differing income thresholds by age group for children in The uniform thresholds made the poverty-related coverage simpler for states to administer and no doubt easier for parents to understand. States may also have shifted into poverty-related coverage some children who would have qualified for Medicaid under the AFDC rules in The SCHIP program was instrumental in helping states expand poverty-related child coverage under Medicaid. The decline in enrollment for Medicaid children from 1994 to 1999 would have been much worse were it not for the SCHIP program. SCHIP s higher federal matching rate helped 8 of the 10 study states in 1999 expand coverage for children through M-SCHIP programs that increased Medicaid eligibility income thresholds, particularly for older children. M-SCHIP children accounted for about 2 percent of Medicaid children across the study states in SCHIP helped Medicaid in other ways as well. States report that SCHIP programs steered many low-income children to Medicaid as result of SCHIP outreach and the Medicaid screening requirement. In addition, SCHIP efforts to simplify eligibility requirements and the application process also encouraged many states to reconsider these aspects of their traditional Medicaid programs. In 1999, the AFDC/Section 1931 group continued to be the most common route to Medicaid eligibility for adults, but utilization of the other eligibility group (which included transitional Medicaid coverage) increased in many states as well. All states showed a decrease from 1994 to 1999 in the number and proportion of adults enrolled in Medicaid through the AFDC/Section 1931 eligibility group. (Section 1931 of the Social Security Act provides the rules for determining AFDC-related Medicaid eligibility after welfare reform.) Nevertheless, the AFDC/1931 eligibility group remained the most common route to Medicaid eligibility for adults. Half the study states also had a noticeable rise in the use of the other eligibility group by 1999, probably because of increasing use of Executive Summary

13 transitional Medicaid coverage, for families whose increased earnings made them ineligible under the Section 1931 rules. Even though states could have modified their Section 1931 provisions to expand Medicaid coverage for adults beyond the old AFDC rules, few had taken advantage of this flexibility in Turnover in monthly Medicaid enrollment remained about the same for both children and adults in 1994 and 1999, but the proportion of Medicaid children and adults with gaps in their Medicaid enrollment increased from 1994 to Little information has been available on the level of turnover in state Medicaid programs, yet turnover is a major administrative challenge to states, with many persons newly entering the program each month, and many leaving as well. In 1999, most of the study states reported that on average 5 percent or more of adult enrollees each month were not enrolled in Medicaid the previous month, and 5 percent or more of adult enrollees were leaving Medicaid each month. For children, entries were 3 percent or more of enrollees each month in all but one of the states, and leavers were 2.5 percent or more of all enrollees in all the study states. These rates were similar to those reported for For both years, these levels of turnover for both adults and children on Medicaid required that state Medicaid programs spend considerable resources processing applications and terminations. Exits from Medicaid are appropriate when children or parents no longer satisfy program eligibility requirements. However, some of the turnover was likely caused by families losing coverage even though they remained eligible, and then regaining it within the same year, a phenomenon sometimes referred to as churning. During 1999, more than 5 percent of child enrollees and more than 7 percent of adults had one or more gaps in their Medicaid enrollment in most study states, higher than the rates reported in Enrollment gaps are a concern, since they can interrupt established relationships with providers as well as continuity in health care services. For both children and adults, there were some shifts from 1994 to 1999 in age, race/ethnicity, and the rate of private insurance coverage. In 1999, Medicaid was reaching a slightly different population of children than in 1994, with increases in the proportion of older children (ages 6 to 18), minority children (especially Hispanics), and children with private health insurance in addition to Medicaid. Changes in race/ethnicity and private insurance coverage were similar for adults. These changes seem plausible, given the higher income thresholds for children (especially those who were older) and the general growth in minority populations. In addition, there were small increases from 1994 to 1999 in the proportion of adults in the under-20 age group and the 45-to-64 group. Adults in the group would have been the most likely to obtain jobs in a good economy. Based on data in three states, both children and adults on Medicaid in 1999 were more likely to be users of Medicaid services than in 1994; hospital user rates also increased for adults, but not for children. xiii Executive Summary

14 xiv The proportion of children using Medicaid services during 1999 ranged from 87 to 98 percent across the three study states with available data. These rates were 11 to 27 percent higher than those reported for For adults, the Medicaid user rates ranged from 79 to 99 percent. These rates were 9 to 14 percent higher than those reported for The hospital user rates for adults ranged from 20 to 29 percent, compared to 19 to 26 percent in Study data did not include diagnosis codes, so information was not available to determine whether the higher user rates for Medicaid children and adults in 1999, meant that enrollees were sicker than in Perhaps the increased service use among children meant simply that more were using preventive services. In addition, the higher overall user rates and hospital user rates among adults could have occurred because pregnant women constituted a higher proportion of overall adult enrollment in 1999 than in However, it might also be that children and adults with health problems were more likely than those in better health to have newly enrolled or remained on Medicaid following welfare reform. This would be consistent with the findings of other researchers: that sicker people remained on welfare and enrolled in Medicaid immediately following welfare reform. Study data were not conclusive on any changes in expenditure patterns for children and adults from 1994 to STUDY LIMITATIONS The study had some limitations. First, the data cover the years 1994 and Enough time has passed that care should be taken in using study results to draw conclusions about current enrollment patterns. Nevertheless, the study findings document an important time of transition, as state Medicaid programs disengaged from their long-term reliance on welfare programs as the main source of enrollment, particularly for children. A second limitation is that the study sample was limited to 10 states. Although these states accounted for 34 percent of all child and adult Medicaid enrollees in 1996 (when the federal welfare reform legislation was passed), they may not be representative of all states. Third, utilization and expenditure information was available for only 3 of the 10 study states, largely because by 1999 many of the children and parents in the other states had enrolled in managed care programs. FUTURE RESEARCH This research has documented several shifts in Medicaid enrollment patterns from 1994 to 1999, as Medicaid caseloads for both children and adults declined nationwide. However, over five years have passed since the end of the study time period. In the intervening years, Medicaid enrollment for children has increased dramatically in most states, and many states have also reported increases in adult enrollment. In addition, states have continued to expand Medicaid eligibility through 1115 waiver and SCHIP programs, as well as changes to Section 1931 provisions. In 2003, half the state Medicaid programs were covering children with family income to 150 percent of the poverty line or higher. Most states now provide continuous coverage guarantees to children for 6 to 12 months. Most states have also opened up Medicaid to two-parent families on the same terms as for single-parent families and have increased the earned income disregards in their Section 1931 programs (beyond the disregards previously used for AFDC), so that the effective income limit is higher. States Executive Summary

15 have also implemented other strategies to promote Medicaid enrollment, including greater client outreach, worker training, improvements to computer systems, and eligibility practices that make it easier to qualify and retain coverage. As a result, enrollment patterns have probably continued to change. Beginning in 1999, CMS required that all states routinely submit Medicaid administrative data of the type used for this study. Therefore, information should be available to continue monitoring aspects of enrollment that were covered in this report. Ongoing research on Medicaid enrollment patterns will help monitor Medicaid s progress in providing low-income families with a stable source of health care coverage. xv Executive Summary

16 THIS PAGE LEFT INTENTIONALLY BLANK FOR DOUBLE SIDED COPYING

17 C HANGES IN M EDICAID E NROLLMENT P ATTERNS FOR C HILDREN AND T HEIR P ARENTS F OLLOWING W ELFARE R EFORM B etween 1995 and 1999, state and federal welfare reform, as well as a strong economy, caused welfare caseloads covering children and their parents to plummet (Bell 2001). The number of welfare recipients dropped almost 50 percent in just four years, from a monthly average of 14.2 million children and parents in 1995 to 7.2 million in Welfare recipients, who accounted for over half of child and adult Medicaid enrollment in 1995, were not expected to lose Medicaid coverage as a result of welfare reform. The federal welfare reform law, the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) of 1996, tried to minimize any adverse effects welfare reform might have on Medicaid. The new rules delinked eligibility for Medicaid and cash assistance and required states to determine Medicaid eligibility according to the Aid to Families with Dependent Children (AFDC) rules in place when PRWORA was passed. In addition, during this period states continued to phase in a federally mandated eligibility expansion that extended Medicaid coverage to all children born after September 30, 1983, with families with income below the federal poverty level (FPL). Also important, Congress passed the State Child Health Insurance Program (SCHIP) as part of the Balanced Budget Act of SCHIP provided states with more funds and greater flexibility to address insurance coverage for children. States could use SCHIP funds to expand their Medicaid coverage of children (called M-SCHIP programs). They could develop separate child health insurance programs (known as S-SCHIP programs), or they could use both approaches. Therefore, for children in particular, there was reason to expect that Medicaid coverage would continue to grow after Nevertheless, after many years of steady growth, Medicaid enrollment in Federal Fiscal Year (FFY) 1996 through FFY 1999 fell below the FFY 1995 levels for both children and adults. As shown in Figure 1, Medicaid enrollment for children peaked at 21.6 million ever enrolled in FFY 1995, before dropping to a low of 20.7 million in FFY Adult enrollment reached 9.6 million ever enrolled in FFY 1995 but fell to 8.6 million in FFY The declines in Medicaid enrollment, though much smaller than the declines in welfare, caused concern, since uninsurance rates for low-income families continued to rise during this period. As discussed in greater detail below, research has shown that many families leaving

18 2 Medicaid during this period became uninsured, which indicates that parents were not moving to jobs with affordable health insurance. Researchers have also concluded that numerous factors probably contributed to the unexpected decline in Medicaid, including shortcomings in how Medicaid eligibility was administered in the years after welfare reform. The present study used Medicaid administrative data to see whether the declines in Medicaid enrollment following welfare reform were accompanied by other eligibility-related changes. Cross-sectional data from 10 states in 1994 and 1999 were analyzed to represent the time before and the time after welfare reform. Key questions include:! Did the eligibility groups that families used to qualify for Medicaid shift following welfare reform?! How did the dynamics of enrollment compare for both time periods? To what extent were persons entering and leaving Medicaid each month? Did enrollment duration and continuity remain the same?! Did the demographic characteristics of child and parent Medicaid enrollees change?! What happened to expenditure levels and service utilization patterns? This study was one of five funded by the Centers for Medicare & Medicaid Services (CMS) in 1998 to examine the impacts of welfare reform on Medicaid. Other studies looked at the reasons for Medicaid enrollment declines following welfare reform; longitudinal Medicaid enrollment patterns for families exiting welfare, with particular attention to the use of transitional Medicaid coverage as a result of increased earnings; and Medicaid enrollment for noncitizen immigrants enrolled in the SSI program. 1 STUDY STATES This analysis focuses on the Medicaid experiences of children and adults in 10 states Arkansas, California, Florida, Kentucky, Maine, Michigan, New Jersey, Pennsylvania, Washington, and Wisconsin. These states represented 42 percent of AFDC enrollees and 34 percent of all adult and child Medicaid enrollees nationally in The four other papers resulting from this project include How Economic Factors and Policy Changes Affected States Medicaid Enrollment and Expenditure per Enrollee: , by Bowen Garrett and Alshadey Yemane, Urban Institute, July 2002; Medicaid Eligibility, Takeup, Insurance Coverage, and Health Care Access and Use Before and After Welfare Reform: National Changes from 1994 to 1997, by Amy Davidoff, Bowen Garrett and Alshadye Yemane, Urban Institute, June 2001; Medicaid Patterns Before and After Welfare Reform for Noncitizen Immigrants on SSI, by Marilyn Ellwood, Mathematica Policy Research, February 2005; and Medicaid Enrollment After Welfare Exit: Changes Since Welfare Reform, by Angela Merrill and Marilyn Ellwood, Mathematica Policy Research, April Changes in Medicaid Enrollment Following Welfare Reform

19 The study states were chosen based on a number of factors, including the availability of administrative data, the level of Medicaid enrollment, low managed care penetration, regional diversity, availability of detailed Medicaid eligibility codes, and contrasting AFDC and Medicaid eligibility policies. The availability of administrative data was a consideration since only 31 state Medicaid programs submitted electronic eligibility and claims data to CMS prior to Selecting states with low managed care penetration was necessary so that the study analysis could yield information on the types of services being used and the level of expenditures. For the most part, managed care organizations have not submitted detailed claims information to state Medicaid programs. The Appendix presents a summary of welfare and Medicaid eligibility income and asset limits, other eligibility rules, and coverage groups before and after PRWORA for the 10 chosen states. 3 DATA SOURCES This study uses Medicaid administrative data maintained by CMS. Data for 1994 are from the State Medicaid Research File (SMRF) Person Summary File. Data for 1999 are from the Medicaid Analytic Extract (MAX) file, which replaced SMRF in that year. SMRF and MAX are research-ready calendar year files produced from the Medicaid Statistical Information System (MSIS) by CMS. They contain individual-level records with month-by-month eligibility information. These Medicaid administrative data files also include summary variables of health care expenditures and utilization, compiled from the four claims files inpatient, long-term care, prescription drugs, and other. STUDY LIMITATIONS Three limitations to the study should be noted. First, study data focus on state Medicaid enrollment patterns in 1994 and Enough time has passed so that care should be taken in using study results to draw any conclusions about current enrollment. Nevertheless, the study findings document an important time of transition, as state Medicaid programs disengaged from their long-term reliance on welfare programs as the main source of enrollment, particularly for children. A second limitation is that the study sample is limited to 10 states. Although these states accounted for 34 percent of all child and adult Medicaid enrollees in 1996, they might not be representative of all states. Third, the study was able to examine summary measures of health care utilization and expenditures in only 3 of the 10 study states: Arkansas, Florida and Kentucky. Five states California, Michigan, New Jersey, Pennsylvania, and Wisconsin were excluded from the utilization and expenditure analysis because of large proportions (greater than 50 percent) of enrollees in managed care in 1999, for whom utilization and detailed expenditure information were not available. Maine and Washington were excluded because some of their utilization data contained anomalous patterns. Changes in Medicaid Enrollment Following Welfare Reform

20 4 KEY ELIGIBILITY GROUPS This study focuses on Medicaid enrollment among children, as well as their parents and other caretaker relatives (referred to as adults in study tables). 2 Part of the analysis looks at the eligibility groups that children and their parents used to qualify for coverage. Since there are many potential pathways to Medicaid eligibility, and because distinguishing among these pathways can be very complicated, the analysis relies on the broad eligibility groups used for federal Medicaid reporting. Federal Medicaid administrative data during the study period used four primary eligibility groups to classify children and adults: 1. AFDC/1931 Group. Before PRWORA, this group (sometimes referred to as cash-related Medicaid) included those children and adults who qualified for Medicaid because they received AFDC welfare benefits. After PRWORA, this group encompassed those who qualified for Medicaid under the Section 1931 rules. (Section 1931 of the Social Security Act provides the rules for Medicaid AFDC-related eligibility after welfare reform.) PRWORA replaced AFDC with the Temporary Assistance for Needy Families (TANF) program, and it was assumed that most TANF enrollees would continue to qualify for Medicaid under the Section 1931 rules. Therefore, the AFDC/1931 group in 1999 included most of those who qualified for TANF benefits in each state, even though receipt of such benefits was not the reason they qualified for Medicaid. It was assumed that states would eventually have people who qualified for Medicaid under the Section 1931 rules but did not qualify for TANF benefits. PRWORA gave states broad flexibility to increase income and asset disregards as part of their Section 1931 rules, thus allowing more children and parents to qualify for Medicaid. However, it was not until 1999 that many states began to change their Section 1931 rules so that persons other than TANF recipients were reported to the Section 1931 group in Medicaid administrative data. 2. Medically Needy Group. Providing coverage for the medically needy is optional, but in 1994 and 1999 more than two-thirds of the states had medically needy programs covering children and their parents, including all the study states except Pennsylvania. States that cover medically needy groups use a higher income threshold than the AFDC cash assistance level to determine eligibility. In addition, applicants with income above the medically needy thresholds must be allowed to qualify for Medicaid by spending down, a provision that allows applicants to deduct incurred medical expenses from their income to determine financial eligibility for Medicaid. 2 People who qualified for Medicaid under the aged and disabled eligibility groups were excluded from the study. Also excluded were women in the months when they were enrolled in family planning waiver programs in three states Arkansas, California, and Florida since these women were receiving very limited services during these months. Changes in Medicaid Enrollment Following Welfare Reform

21 5 3. Poverty-Related Group. This group generally encompasses children and pregnant women who qualify for Medicaid through any of the mandatory poverty-related requirements or optional poverty-related expansions. States must extend full Medicaid benefits to all children under age 6 and to all pregnant women with family income below 133 percent of FPL. In addition, states are required to cover all children born after September 30, 1983, with family income below 100 percent of FPL. At their option, most states elected to use considerably higher income thresholds for their poverty-related child and pregnant women coverage by In particular, many states used the enhanced federal matching available through SCHIP to establish higher povertyrelated income thresholds in Medicaid for children. For example, in 1999 several of the study states had M-SCHIP programs that covered children to 150 percent of FPL. Children qualifying through these programs were reported as being in the poverty-related group, as were children and adults covered by eligibility expansions through state 1115 waiver programs. Finally, the povertyrelated group could include children and pregnant women who qualified for Medicaid as a result of the 1902(r)(2) option. 3 In 1999, the only study state that covered adults other than pregnant women under the poverty-related group was Wisconsin, as a result of its 1115 program, which extended coverage to parents. 4. Other Group. Children and adults who qualify for Medicaid through a mixture of mandatory and optional coverage not reported under the previous eligibility groups are reported under the other eligibility group, which includes for example, children and adults qualifying for up to 12 months of transitional medical assistance (TMA) because family earnings caused them to lose AFDC/1931 eligibility. The other group also includes presumptive eligibility coverage, emergency coverage for immigrants not otherwise eligible for Medicaid, institutional coverage, coverage for foster care and adoptive children, and coverage for children qualifying through the so-called Ribicoff rules. 4 When PRWORA was passed, little change was expected with regard to the numbers and distribution of children and adults across these four key eligibility groups, and the Congressional Budget Office did not anticipate any drops in Medicaid enrollment for children and their parents (Congressional Budget Office 1996). However, when state welfare caseloads began their rapid decline and then SCHIP was passed, it became clear that shifts were occurring in how children and their parents qualified for Medicaid. Therefore, a study objective was to analyze how the pathways to Medicaid eligibility were changing. A final point about eligibility groups is that it is possible for individuals to qualify for Medicaid under more than one group. So, states have to decide the order in which they will 3 As with Section 1931, the 1902(r)(2) provision allows states to use more liberal income and asset disregards than those used for AFDC. 4 The Ribicoff option allows states to open up Medicaid to children through age 20, regardless of family structure (neither SCHIP nor the federal poverty-related expansions cover 19- or 20-year-old children). Changes in Medicaid Enrollment Following Welfare Reform

22 6 test for eligibility across groups. Often, eligibility for multiple groups is attempted before an individual is found to qualify for coverage. Before welfare reform, the automated systems in most states used a hierarchical approach to eligibility determination, testing for the groups using the lowest income thresholds first (Ellwood 1999). In the years following welfare reform, some states moved to testing first for the eligibility groups with higher income limits and less complicated rules (particularly for children and pregnant women). TREATMENT OF SCHIP ELIGIBILITY When the study was conceived, the SCHIP program was still in its planning stages, and the study design did not anticipate how the SCHIP program would affect state Medicaid programs. Nevertheless, by the time the study data were ready, it was apparent that state SCHIP programs were having an important impact in expanding public health care coverage for low-income children. To follow this, some study tables break out M-SCHIP enrollment numbers as a subset of the Medicaid poverty-related child group. However, the Medicaid administrative data used in the study do not include information on children enrolled in S- SCHIP programs. To help fill this gap in the data aggregate S-SCHIP enrollment figures for each state from the SCHIP Enrollment Data System (SEDS), maintained by CMS, are included in some tables. This may introduce double-counting among children who were enrolled in both Medicaid/M-SCHIP and S-SCHIP in a calendar year, but it gives a fuller picture of children s public health insurance enrollment at the end of the study period. Any double-counting is likely to be minor in All the study states but Washington had implemented SCHIP programs by Eight of the nine states had an M-SCHIP program (all but Pennsylvania), and six of the nine had S-SCHIP programs (California, Florida, Maine, Michigan, New Jersey, and Pennsylvania). 5 BACKGROUND In 1996, PRWORA ended the federal entitlement for cash welfare benefits by replacing the AFDC program with TANF, a block grant program that instituted more stringent work requirements, a time limit on cash benefits, and sanctions for noncompliance, as well as new supports such as child care. These policy changes were expected to reduce welfare participation, although not to the extent that actually occurred. Most of the research on welfare reform concludes that the strong economy during the mid-1990s was just as important in reducing welfare caseloads as the policy changes (need reference). What about Medicaid? Before PRWORA, families eligible for AFDC were automatically eligible for Medicaid. In enacting PRWORA, Congress did not intend to undercut Medicaid eligibility; instead, the legislation severed, or delinked the automatic eligibility tie between cash assistance and Medicaid. PRWORA required that states determine Medicaid eligibility under the AFDC rules in place in 1996 (when PRWORA was passed). As mentioned earlier, the Congressional Budget Office predicted no reduction in 5 S-SCHIP programs in Kentucky and Washington were implemented after the study period. Changes in Medicaid Enrollment Following Welfare Reform

23 Medicaid participation or spending as a result of PRWORA even though welfare caseloads were expected to fall. The assumption seemed to be that former AFDC recipients would continue to qualify for Medicaid under the old AFDC rules, even if they were not enrolled in the TANF program. Section 1931 refers to the section of the Social Security Act that stipulated the rules for determining Medicaid AFDC-related eligibility after welfare reform. Section 1931 required states to continue using income and resource standards, methodologies, and deprivation requirements that were in effect under their AFDC plans on July 1, However, it also gave states broad flexibility to modify the AFDC eligibility rules, including how income and assets were counted. For example, states could elect to increase earned income disregards and eliminate asset tests in determining coverage under the Section 1931 provisions. Nevertheless, states were slow to use this new flexibility (Maloy et al. 2002), and in 1999 they were just beginning to take advantage of using the Section 1931 provisions to expand Medicaid eligibility. Despite the delinking of Medicaid eligibility from TANF and the Section 1931 flexibility, Medicaid caseloads began to decline in 1996, following a decade of increases. This might have resulted because welfare recipients leaving Medicaid were getting jobs with health insurance. However, studies of welfare leavers suggest that many children and adults lost Medicaid coverage after leaving welfare even though most of them remained eligible, and that few families entering employment reported gaining employer-based health insurance (Guyer 2000; and Greenberg 1998). One study found that a year after leaving welfare, almost half of women and one-third of children were uninsured (Garret and Holahan 2000b). Families who did not have health insurance after leaving welfare are more likely to return to welfare (Loprest 2002). While an improving economy and rising incomes did play a role in causing Medicaid enrollment to decline, changes in behavior and administrative practices played a larger role (Ku and Garret 2000; and Garrett and Yemane 2002). The implementation of welfare reform and delinking, accompanied by lack of knowledge among both families and eligibility workers about eligibility options, probably had a number of unintended effects on Medicaid enrollment and retention (Chavkin et al. 2000; Ellwood 1999; Pavetti et al. 2002; and Ragan 2003). These effects included decreased applications as a result of TANF diversion programs 6 and job search requirements; delay or inappropriate denial of Medicaid applications due to TANF application requirements; problems with delinking computer systems and procedures, which caused families to be terminated from Medicaid at the same time as from cash assistance; inappropriate termination of Medicaid for TANF sanctions or failure to meet reporting requirements; lack of knowledge of potential Medicaid eligibility when leaving TANF; and welfare stigma. 7 6 Diversion programs offer incentives (such as one-time emergency payments) to dissuade families from applying for ongoing TANF cash assistance. Changes in Medicaid Enrollment Following Welfare Reform

24 8 RESULTS Shifts in Enrollment by Eligibility Group Consistent with the national pattern, most of the study states had declines in child Medicaid enrollment from 1994 to 1999, and all the states had declines in adult Medicaid enrollment. One objective of the study was to investigate whether these declines occurred across all eligibility groups, or whether the use of some pathways to eligibility changed following welfare reform. Children. Among the study states, California, Florida, and Wisconsin showed the greatest drops in child Medicaid enrollment, with 11 to 14 percent fewer children enrolled in 1999 than in 1994 (Table 1). Pennsylvania, Michigan, and Kentucky experienced smaller declines. The four states with increases in child Medicaid enrollment were Arkansas (40 percent), Washington (21 percent), New Jersey (3 percent), and Maine (1 percent). In each of the study states in 1994, the main route to Medicaid eligibility for children was through the AFDC program. However, by 1999, all 10 study states had experienced a major reduction in Medicaid enrollment for children who qualified through the AFDC/1931 provisions. The extent of this reduction ranged from 22 percent in California to 57 percent in Wisconsin; three states had declines of 50 percent or more. This occurred even though by 1999 five of the study states had increased the income limits they used for Section 1931 coverage, five had increased their asset limits, and one had eliminated asset testing altogether (see the Appendix table). Therefore, even though AFDC-related rules for Medicaid (as implemented through the Section 1931 provisions) were no more restrictive in 1999 than in 1994, and in over half the states Section 1931 rules expanded eligibility, many fewer children qualified for Medicaid through this pathway. However, as indicated below, all the states had other eligibility group options for extending child coverage, which could explain in part the decline in the use of the AFDC/1931 coverage group for children. Most of the study states also experienced declines from 1994 to 1999 in children enrolling in Medicaid through the medically needy group, although this group accounted for only a small proportion of Medicaid children in both 1994 and Since federal requirements tie medically needy income thresholds to the income standards used for AFDC-related eligibility, the income thresholds are quite low in most states (see the Appendix for medically needy income limits). Therefore, it is not surprising that few children qualify for Medicaid under the medically needy group, and that this number would fall over time as states moved to higher income thresholds for poverty-related coverage. In 1999, enrollment increases in the poverty-related group offset declines in child AFDC/1931 and medically needy enrollment. Child enrollment in the poverty-related group grew in every state, with increases in 1999 ranging from 50 to 274 percent above 1994 levels. In all but two of the states (Pennsylvania and Washington), the poverty-related eligibility group included a new M-SCHIP program, and in most of the study states, the income standards for children were more liberal than the federal mandates. For infants in 1999, the standard was 185 percent or more of FPL in all the states (compared with the federal requirement of 133 percent). For children ages 1 to 5, six states had income thresholds Changes in Medicaid Enrollment Following Welfare Reform

25 above the federal requirement of 133 percent of FPL. Finally, seven states used income thresholds higher than 100 percent of FPL (the federal requirement) for children over age 5. In some states, growth in enrollment was aided by optional expansions to the povertyrelated group. For example, in 1997 Arkansas expanded its poverty-related group through an 1115 program that extended Medicaid to all children with family income less than 200 percent of FPL. Wisconsin expanded child coverage in 1999 through an M-SCHIP expansion and an 1115 program, so that an income threshold of 185 percent of FPL applied to children of all ages. In 1995, Washington expanded its poverty-related eligibility to 200 percent of FPL using the 1902(r)(2) provisions. Seven of the study states also showed growth in the number of children qualifying for Medicaid through the other eligibility group from 1994 to As a reminder, this group includes children whose families qualified for up to 12 months of transitional medical assistance because family earnings caused them to lose AFDC/1931 eligibility. More families moving from welfare to work could have caused this group to grow. Table 2 shows how the distribution of children on Medicaid, by eligibility group, shifted between 1994 and AFDC/1931 Medicaid was the largest eligibility category for children in all study states in 1994, accounting for more than half of child enrollment in all but two of the study states. By 1999, however, declines in AFDC/1931 coverage meant that the three other broad Medicaid eligibility groups accounted for the majority of children in all the states but California. Poverty-related coverage (including M-SCHIP and 1115 programs) took up most of the decline in AFDC/1931 coverage. In 1999, California was the only study state in which enrollment in the poverty-related group was not substantial. The distribution of coverage there was notably different from the distribution in the other study states, with more children continuing to be eligible through the AFDC/1931 provisions, a much smaller proportion of children in the poverty-related group, and a higher proportion in the medically needy group. By 1999, California had moved to higher income eligibility limits under its Section 1931 program (a result of higher disregards for earnings) and its medically needy program, and California had increased its allowable asset level. In addition, the way eligibility was assessed meant that children often qualified for medically needy coverage before they qualified for poverty-related coverage. Note that six of the study states opted to expand public health insurance coverage for children through S-SCHIP programs, in addition to their Medicaid coverage. Table 1 also includes enrollment levels for these S-SCHIP programs in If S-SCHIP coverage is considered as well as Medicaid, the decrease in child public insurance coverage from 1994 to 1999 was reduced. For example, in Florida, if S-SCHIP and Medicaid coverage are counted, 9 7 To avoid double-counting, enrollees are classified by their principal eligibility group (the group in which they spent the most time during the year). In the case of equal periods in multiple coverage groups, people are assigned to their most recent eligibility group. Changes in Medicaid Enrollment Following Welfare Reform

26 10 child health insurance coverage in 1999 declined by only 2 percent relative to Without S-SCHIP coverage, the decline was 12 percent. For Pennsylvania, the number of children with public insurance coverage declined by 8 percent if only Medicaid is considered, while it increased by 1 percent if S-SCHIP is considered as well. If the M-SCHIP and S-SCHIP programs had not existed in 1999, results would have been much different. Figure 2 is a bar chart comparing child enrollment with 1994 levels, showing the proportion of 1999 public insurance enrollment in each state attributable to the M-SCHIP and/or S-SCHIP programs, as well as to traditional Medicaid. 8 Only Arkansas and Washington would have shown child enrollment gains in 1999 without SCHIP. In addition, the level of decline in many states would have been much greater without SCHIP. For example, without the M-SCHIP and S-SCHIP programs, 1999 child enrollment in public insurance in Florida would have dropped by 15 percent, instead of 2 percent, from Adults. Every state experienced a decline in overall adult Medicaid enrollment from 1994 to 1999, ranging from 9 percent in Arkansas to 37 percent in Michigan (Table 3). 9 In the six states where both child and adult enrollment decreased, the declines for adults greatly exceeded those for children. As with children, the main route to Medicaid eligibility for adults in 1994 was the AFDC group, and this group experienced the greatest reduction in enrollment. Between 1994 and 1999, adult enrollment in the AFDC/1931 group fell from 24 percent (in Florida) to 70 percent (in Michigan). Four states had declines of 50 percent or more, even though by 1999 six of the study states had moved to take advantage of the broad flexibility allowed by Section 1931, by increasing earned income or asset disregards, or both. Unlike the pattern with children, the declines in AFDC/1931 adult enrollment were not offset by increases in the three other Medicaid coverage groups. This is not surprising, since the eligibility provisions for adults outside the AFDC/1931 group were fairly restrictive. Federal requirements dictate that the income thresholds for state medically needy programs be no more than one-third higher than the AFDC-related income standard (adjusting for family size), and states sometimes opt to set them lower than this maximum. The povertyrelated provisions for adults extended only to pregnant women in all the states but Wisconsin. As mentioned earlier, Wisconsin used an 1115 waiver to expand adult coverage. None of the SCHIP programs were covering adults in Medicaid adult enrollment showed a net decline even though many states experienced modest increases for adults in the medically needy, poverty-related, or other eligibility groups. From 1994 to 1999, four states had increases in adult medically needy enrollment. Five had more adults qualifying under the other eligibility group in 1999 than in 1994, 8 Traditional Medicaid refers to Medicaid eligibility groups other than M-SCHIP. 9 In 2003, Pennsylvania officials informed CMS that their Medicaid administrative data for FFY 2000 overcounted poverty-related adult enrollment by about 40,000, and that this problem existed in earlier years as well. It was not possible to identify the people who should be removed from the state s 1994 and 1999 data. Therefore, the Pennsylvania numbers on poverty-related adult enrollment and total adult enrollment are overstated, and calculations that include adults in the poverty-related group may not be reliable. Changes in Medicaid Enrollment Following Welfare Reform

27 which probably reflects increased enrollment under the transitional medical assistance provisions (for those no longer qualifying for AFDC/1931 as a result of increased earnings). Seven states (excluding Pennsylvania) had more adults qualifying under their povertyrelated provisions. Wisconsin showed the largest gain in poverty-related adult enrollment (227 percent), as a result of its 1115 BadgerCare program, which in July 1999 extended coverage to parents to 185 percent of FPL. For the other six states, the increases in the adult poverty-related group ranged from 7 to 82 percent. In all these states, only pregnant women could qualify under the poverty-related adult rules. Even though some of these states had large gains between 1994 and 1999 in enrollment under the poverty-related rules, it seems unlikely that this reflects an increase in the number of pregnant women enrolled in Medicaid. Birth rates of women ages 15 to 44 fell from 65 per 1,000 in 1994 to 61 per 1,000 in 1998 (Bachu 1995; and Bachu and O Connell 2000). One possibility for the increase is that it was simply more straightforward for states to determine eligibility for pregnant women through the poverty-related rules in 1999 than in 1994, with the delinking of welfare and Medicaid eligibility. Adults had less of a shift than children in the distribution of enrollment across eligibility groups from 1994 to 1999 (Table 4). The AFDC/1931 eligibility group remained the largest adult eligibility group in 1999 for all the study states except Arkansas and Wisconsin. Nevertheless, there was a shift. In 1994, the AFDC/1931 group accounted for half or more of adult Medicaid enrollment in all the study states, but in 1999 this was true only in Florida and New Jersey. Further, the other eligibility group accounted for 29 percent or more of adult enrollment in five of the study states in 1999, but in only one state in Changes in Monthly Entry and Exit Rates Little information has been available on the level of turnover in the Medicaid program, yet turnover is a major administrative challenge to states. Unlike the situation with private insurance and Medicare, Medicaid eligibility in most states is very dynamic, with many persons newly entering the program each month, and many leaving as well. Table 5 shows the average monthly entry and exit rates for Medicaid for both children and adults in 1994 and Entry rates are defined as the proportion of people who enrolled in a given month and were not enrolled in the previous month, and exit rates are defined as the proportion of persons enrolled in a month who were not enrolled the following month. Generally, state enrollment levels increase when monthly entry rates exceed exit rates. It is impossible to come up with a standard of what might be considered a reasonable entry rate for Medicaid each month. Medicaid entry rates might increase for several reasons, including rising unemployment, expansions in eligibility criteria or special outreach activities. Similarly, entry rates could fall with increased employment opportunities or new eligibility restrictions. Standards for exit rates are also difficult to determine. Some exits are always expected each month, since eligibility is redetermined for Medicaid enrollees at least once a year, and inevitably some persons are found to no longer qualify for coverage. Clearly, Medicaid exit is appropriate when circumstances have changed, so that individuals are no longer eligible. However, concern has always existed that some people leave Medicaid even 11 Changes in Medicaid Enrollment Following Welfare Reform