Extreme Poverty Rising, Existing Government Programs Could Do More Sheila R

An Urban Institute New Federalism Program to Assess National Survey of America’s Families Changing Social Policies

THE URBAN INSTITUTE Series B, No. B-45, April 2002

Extreme Poverty Rising, Existing Government Programs Could Do More Sheila R. Zedlewski, Linda Giannarelli, Joyce Morton, and Laura Wheaton

By raising Many have heralded welfare reform as an getting the assistance they need and may unqualified success. More single parents be facing greater hardship. participation in are working, and welfare caseloads have This brief sets out to understand how plummeted. The official poverty rate decreased program participation has affect- programs already in decreased from 13.7 percent in 1996 to 11.8 ed poverty since welfare reform. It reviews percent in 1999. While many commentators changes in poverty from 1996 to 1998 and place, policymakers debate the causes of these improvements— compares families’ current economic status citing the strong economy, federal and with a scenario that assumes full participa- can take an important state policies designed to “make work tion in government support programs. pay,” and welfare reform—few dispute the The findings show that the govern- step toward improving strong positive outcomes.1 ment offers a stronger safety net than is Less well known, however, is the delivered. In 1998, if all families with chil- the economic well- unfortunate story buried within these posi- dren participated in the post-reform gov- tive outcomes. When all types of income ernment safety net programs for which being of children. are taken into account, extreme poverty they qualified, poverty would have (income below 50 percent of the federal declined by more than 20 percent and poverty level) has increased. Despite phe- extreme poverty by 70 percent. This nomenal growth in the U.S. economy in decline provides a strong rationale for the late 1990s, more children lived in changing existing programs to provide single-parent, extremely poor families in “family-friendly” delivery systems and 1998 than in 1996. This surprisingly nega- more standardized eligibility requirements. tive outcome largely reflects an increase in By increasing participation in programs the number of low-income families that already in place, policymakers can take an have either left, or chosen not to enroll in, important step toward improving the eco- government support programs (including nomic well-being of children. cash welfare and food stamps). The reasons for the decline in program Data and Methods participation are difficult to pinpoint. The analysis uses the 1997 and 1999 Changing attitudes about seeking help National Surveys of America’s Families from the government, the increasing com- (NSAF) and the Urban Institute’s Transfer plexity of program rules, and states’ new Income Model, version 3 (TRIM3). The welfare programs have no doubt all played NSAF is a large, nationally representative a role in reducing participation. The survey of the civilian, noninstitutionalized decline suggests that many families are not population under age 65 that overrepre- ASSESSING THE NEW FEDERALISM An Urban Institute Program to Assess Changing Social Policies

sents low-income families. Like simi- gibility because of immigrant status. also apply for in-kind assistance lar government surveys, each round Thus, the estimates of poverty reduc- through the Food Stamp program of the NSAF asked families about tion achieved by full program partici- and for child care and housing subsi- income received during the prior pation are somewhat overstated.6 dies. Each of these programs, howev- year (1996 and 1998). The TRIM3 er, has very different minimum model corrects the NSAF’s underre- Changes in the Government income requirements, rules about porting of government benefits (a Safety Net work, and asset tests (see Zedlewski condition that plagues all surveys in Federal and state welfare reforms 2002). In addition, unlike many bene- this area),2 adds the value of food enacted since 1996 have affected fam- fits, child care and housing assistance stamps and the earned income tax ilies’ benefit eligibility and participa- are not guaranteed and are awarded credit (EITC), imputes federal payroll tion in a number of ways. Some of on a first-come, first-served basis. and income taxes, and estimates the changes positively affected partic- Medical assistance may be available annual out-of-pocket child care ipation. For example, states that to low-income families, but most expenses. raised the amount that individuals states limit coverage for parents to The TRIM3 model corrects for can earn and still receive TANF bene- those that have cash assistance or the underreporting of Supplemental fits increased the number of eligible have recently left TANF. Coverage Security Income (SSI), Temporary families and encouraged participa- for their children, however, is more Assistance for Needy Families tion. In contrast, state diversion broadly available. Working families (TANF), and food stamp benefits by strategies, such as requiring substan- receive assistance through the applying individual state rules to tial proof of job search prior to enroll- refundable EITC.7 Legal immigrant determine eligibility for each individ- ment, may have discouraged pro- families may or may not be eligible ual and family in the survey file.3 The gram participation. for some of these safety net benefits model then compares benefit eligibil- The 1996 federal reforms to other depending on the program, their ity with income sources reported by parts of the safety net also affected state of residence, and (sometimes) NSAF families and adds benefits for benefit eligibility and participation. their work history. As the next sec- families that failed to report them by In addition to restricting SSI eligibili- tion shows, the drop in participation matching detailed control totals ty for children, the legislation elimi- following the reforms led to higher developed from state and federal nated SSI and food stamp benefit eli- extreme poverty among families with administrative data. gibility for legal, noncitizen immi- children. The augmented NSAF data pro- grants. (Subsequent legislation enact- vide estimates of families’ total dis- ed in 1997 and 1998 restored eligibili- The Poverty Story: 1996–1998 posable income—defined as all ty for some noncitizen groups.) The Our examination of participation’s sources of cash income plus the value reforms also cut food stamp benefits effect on family well-being looks at of food stamps and the EITC, minus by lowering the payment standard, two types of poverty measures. The federal taxes and out-of-pocket child freezing the standard deduction, first is the official U.S. measure of care expenses.4 We report poverty increasing the amount of income poverty, which compares families’ estimates using cash income alone counted against benefits, and limiting cash income to the poverty threshold (following the official Census Bureau food stamp eligibility to three but excludes two of the most impor- estimate) and using an alternative months for able-bodied adults with- tant government benefits—food measure based on disposable out dependents working less than 20 stamps and the EITC. Because of the income.5 hours per week. Finally, the legisla- importance of food stamps (a “near The TRIM3 program eligibility tion gave states more funding and cash” benefit) in reducing poverty, estimates allow us to calculate the consolidated programs for child care we also extend the cash-income mea- change in income that would occur if assistance. sure to include these benefits. all families took the benefits available The changes made the already The second, more comprehensive to them. The “full-participation sce- complex safety net system even more measure includes all government nario” holds earnings and nonwel- difficult to navigate. Low-income citi- assistance, but deducts federal pay- fare sources of income constant, but zen families can apply for cash assis- roll and income taxes as well as out- recalculates eligible families’ SSI, tance through SSI (if there is a severe- of-pocket child care expenses from TANF, and food stamp benefits. ly disabled family member) or TANF income to adjust for the different Owing to data limitations, the esti- (if they have not reached a benefit needs of working and nonworking mates do not take into account ineli- time limit). Low-income families can families.8 The official U.S. “cash-

2 An Urban Institute Program to Assess Changing Social Policies ASSESSING THE NEW FEDERALISM

income poverty” measure yields high- 20 million persons were poor in 1996 stamps reveals a decline of 1.7 million er poverty levels than the more according to the disposable-income persons. A reduction in the use of comprehensive “disposable-income measure, compared with 23.7 million food stamps mitigated the improve- poverty” measure because it omits persons by the official poverty mea- ment in cash-income poverty for per- significant sources of income. The dis- sure. Food stamps accounted for a sons in single-parent families. posable income poverty measure bet- large share of this difference (2.3 mil- ter reflects families’ economic status lion fewer poor persons). The net Extreme Poverty because it takes into account their effect of the refundable EITC decreas- Factoring in government assistance buying power. ed poverty by another 1.4 million per- has a dramatic effect on extreme Changes in both the cash and dis- sons, despite deductions for other tax poverty levels (figure 2). The number posable income measures of poverty, payments and out-of-pocket child care of persons in families with children reported below, show a disturbing expenses. counted as extremely poor is more trend. While overall family poverty Both cash-income and disposable- than halved when food stamps are declined significantly, extreme pover- income poverty declined between considered income (3.2 million com- ty for persons in all families with chil- 1996 and 1998 for individuals in fami- pared with 7.5 million), and even dren and those in single-parent families with children, including those liv- fewer persons are classified as lies shrunk much less, and by the ing in single-parent families. Based on extremely poor based on disposable disposable-income measure it rose. cash income, about 3.8 million fewer income. Food stamp benefits have an The difference between having income persons in families with children were especially large effect on extreme below one-half of the poverty level and poor in 1998 than in 1996. The decline poverty figures because benefits are 50–100 percent of the poverty level in poverty for all persons is slightly significant at this income level and a makes a big difference in economic less when food stamps are included in large share of these families hardship. A family of three is poor income (3.5 million persons) and participate. when its income falls below $14,269; it when total disposable income is used According to the cash-income is extremely poor when its income to measure poverty (3.3 million). measure, extreme poverty declined falls below $7,135 (in 2001 dollars). The change in poverty for persons substantially between 1996 and 1998 in single-parent families follows a among persons living in families with Poverty similar pattern. The cash measure of children. About 6.3 million persons Poverty is lower when more sources poverty shows the largest decline lived in extreme poverty in 1998, 1.2 of government assistance are taken between 1996 and 1998 (2 million per- million fewer than in 1996. By this into account (figure 1). For example, sons). The measure that includes food measure, fewer persons in single- parent families also lived in extreme FIGURE 1. Poverty among Families with Children, by Income Definition, poverty in 1998 than in 1996, although 1996–1998 the reduction in extreme poverty was relatively small. These declines in extreme poverty

25 23.7 All Families shrink as more sources of income are Single-Parent Families 21.4 taken into account. For example, 19.9* 20.0 20 when food stamps are included as 17.9* 16.7* income, extreme poverty among families with children was unchanged 15 between 1996 and 1998. The disposable income poverty measure shows a 10 slight increase in extreme poverty for Persons (millions) Persons 14.0 all families with children. Fewer fami- 12.0* 12.4 11.7 5 10.7* 9.9* lies in extreme poverty participated in food stamps in 1998 than in 1996, and 0 increased work-related expenses 1996 1998 1996 1998 1996 1998 Cash Income Cash Income+Food Stamps Disposable Incomea (taxes and child care) further reduced Poor disposable incomes. Disposable-income measures Source: The 1997 and 1999 National Surveys of America’s Families, augmented by the Urban Institute’s TRIM3. a. Cash plus food stamps and the EITC, less federal taxes and out-of-pocket child care. show extreme poverty among persons * Significantly different from the 1996 figure at the 90 percent confidence level or higher.

3 ASSESSING THE NEW FEDERALISM An Urban Institute Program to Assess Changing Social Policies

FIGURE 2. Extreme Poverty among Families with Children, by Income parent families had incomes below Definition, 1996–1998 poverty under the full-participation scenario than in the actual 1998 data. All Families 8 7.5 Single-Parent Families This difference could have reduced 7 6.3* the number of poor individuals in 6 single-parent families from 9.9 mil-

5 lion (shown in figure 1) to 7.1 mil-

4 lion. And 1.4 million fewer persons 3.2 3.2 2.9* would have been in extreme 3 2.8 Persons (millions) Persons 5.0 4.6* poverty—more than enough to offset 2 the 300,000-person increase in 2.1* 1 1.7 1.6 1.9* extreme poverty actually experienced 0 1996 1998 1996 1998 1996 1998 by single-parent families between Cash Income Cash Income+Food Stamps Disposable Incomea 1996 and 1998. Extremely Poor The potential poverty decreases are particularly notable for those in Source: The 1997 and 1999 National Surveys of America’s Families, augmented by the Urban Institute’s TRIM3. Note: Extremely poor are those with income below 50 percent of the poverty level. extreme poverty (figure 4). The full- a. Cash plus food stamps and the EITC, less federal taxes and out-of-pocket child care expenses. participation scenario shows extreme * Significantly different from the 1996 figure at the 90 percent confidence level or higher. poverty declining about 70 percent living in single-parent families— 1999. While the uninsured rate of and poverty dropping 20 percent in those most at risk of needing wel- these parents did not change signifi- 1998. Government transfer programs fare—increasing more substantially. cantly (because of a slight gain in pri- are much more likely to move fami- About 300,000 more persons in vate health coverage), the decline is lies above 50 percent of poverty than single-parent families lived in notable because Medicaid provides to lift families out of poverty. TANF, extreme poverty in 1998 than in 1996. the fullest package of coverage with for example, pays a maximum cash The rise reflects shifts in benefits and limited, or no, out-of-pocket costs. benefit equal to about 36 percent of expenses. Many single parents who poverty for a family of three, on aver- moved into the labor market during Full Participation Would age. Families that combine substan- this period left welfare and the Brighten the Poverty Story tial earnings with food stamps and Food Stamp program. Their earnings, The full-participation analysis shows the EITC, however, can move above however, were not sufficient to offset that poverty and extreme poverty poverty, as long as their child care the income loss resulting from the would have been substantially lower costs do not eat away a substantial drop in food stamp benefits. In addi- if families took full advantage of the portion of their income. tion, while many of these parents government safety net (figure 3). For Of course, full participation in became eligible for the EITC, they example, based on disposable government programs would also began to pay payroll taxes and income, if all families with children increase government program costs. child care expenses. In sum, dispos- participated fully in the government The TRIM3 model estimates that able income declined sufficiently to programs they qualified for, 3.8 mil- food stamps, TANF, and SSI would move more parents and their chil- lion fewer persons would have been have cost $23 billion more in 1998 (in dren into extreme poverty. poor and 2.0 million fewer persons 2001 dollars) if all low-income fami- One other change in program would have been extremely poor lies with children took advantage of policy may have also affected the compared with the actual 1998 data. these programs’ benefits. Estimates well-being of extremely poor fami- The potential decline in poverty is based on full participation (even for lies. While our analysis does not substantially smaller when using the families eligible for low benefits who deduct out-of-pocket health care less-comprehensive cash-income might find the “cost” of participating costs, some evidence suggests that measure (1.6 million). particularly high) may somewhat these costs (or at least the risk of For persons living in single- overstate these costs. In addition, higher costs) increased. In particular, parent families, the decline in potential increases in cash TANF ben- the NSAF showed that Medicaid cov- disposable-income poverty resulting efits ($10 billion) arising from the erage declined significantly among from full participation would have greater demand may have restricted extremely poor single parents, from been even more substantial. About states from using TANF dollars to 48 percent in 1997 to 41 percent in 1.8 million fewer persons in single- provide some new benefits for work-

4 An Urban Institute Program to Assess Changing Social Policies ASSESSING THE NEW FEDERALISM

FIGURE 3. Estimated Change in Poverty from Full Participation in tions in poverty that could have been Government Support Programs, 1996–1998 achieved by combining increased

Single-Parent Families earnings with government assistance All Families were diminished. Poverty Extreme Poverty As noted, it is impossible to iso- Cash Income Disposable Income Cash Income Disposable Income late the reasons for low and declining 1996 1998 1996 1998 1996 1998 1996 1998 0 program participation. Some of the -0.577^ -0.974^* -0.738^ -0.860^ drop likely reflects changing family -1.269^ -1.434^* -1.400^* -1 -1.777^* attitudes about using government -1.001^ -1.250^ assistance. Studies also show that the -1.617^* -1.657^ -2

Persons (millions) Persons -1.892^* -2.029^* complexity of program rules and fear of stigma have played a role in dis- -3 couraging participation (see, for example, McConnell, Ponza, and -4 -3.832^* -3.996^ Cohen 1999). A recent report by the U.S. General Accounting Office (2001),

Source: The 1997 and 1999 National Surveys of America’s Families, augmented by the Urban Institute’s TRIM3 to correct for underreport- which documents the wide array of ing of government benefits and to adjust for the value of food stamps and taxes. Note: The full-participation scenario assumes that all eligible families would participate in TANF, SSI, and food stamps, holding constant government assistance programs and earnings and other sources of income. *Significantly different from 1996 level at the 90 percent confidence level or higher. income eligibility policies, concludes ^Significantly different from historical results (figures 1 and 2). that the federal government should ing poor families (such as child care). and extremely poor families. The support state and local demonstration Participation in TANF would also number of hardships reported by the programs that test alternative benefit count against benefit time limits, less poor, however, was significantly delivery systems. potentially deterring participation by lower (47 percent of families with The recent weakening in the econ- working families that could combine a incomes between 50 and 100 percent omy will likely put more families at small welfare check with earnings. of the poverty level reported two or risk of poverty, making program Still, the rise in government costs more hardships, compared with 61 changes that welcome, rather than would be significant and must be percent of extremely poor families). In discourage, families from enrolling in weighed against the high costs of addition, near-poor families (income appropriate assistance programs even poverty to individuals and to society. of 100–150 percent of the poverty more essential. While the demonstra- level) reported significantly fewer tion programs recommended by the Hardships Facing Families in hardships than those living in poverty. GAO may ultimately identify the Poverty These findings suggest that full partic- most effective benefit-delivery strate- Families with children living in ipation in government assistance pro- gies, incremental improvements could extreme poverty find it difficult to grams, which would lift many fami- be implemented immediately. provide basic food and shelter for lies out of poverty into the near-poor Standardizing income eligibility, work their families. Indeed, about 4 in 10 category, would ease the economic rules, asset tests, and treatment of dif- extremely poor families report two or hardships currently facing poor fami- ferent family members’ incomes for more problems providing food for lies with children. TANF families as well as for working their families, and 3 in 10 report diffi- poor families could raise participation culties paying rent or utilities in the Implications levels. In addition, information shar- prior 12 months (table 1). Almost a The 1996 federal reforms shifted ing across the agencies that administer quarter report living in crowded hous- responsibility for cash assistance to these programs—including human ing (more than 2 persons per bed- the states, cut food stamps and SSI services offices, housing authorities, room), and 1 in 10 have lived without benefits, and expanded child care the public health service, and the a telephone for one month or longer. assistance. At least in part because of social security office—would likely Changes in government programs the changes in the safety net, partici- streamline the eligibility maze con- that increase program participation pation in government assistance pro- fronting low-income families. States could substantially ease economic grams has declined. At the same time, must ensure that TANF rules hardship. The types of hardships mea- extreme poverty based on disposable designed to encourage work do not sured by the NSAF were about the income increased, especially among discourage families from using inter- same for individuals living in poor single-parent families. Potential reduc- im government assistance. Public and

5 ASSESSING THE NEW FEDERALISM An Urban Institute Program to Assess Changing Social Policies

FIGURE 4. Estimated Percentage Change in Disposable-Income Poverty from training—should not be forgotten. In Full Participation in Government Programs, 1996–1998 these tough economic times, however, these strategies seem less saleable Poverty Extreme Poverty 1996 1998 1996 1998 1996 1998 1996 1998 than steps that improve the safety net 0 already offered by the government. -10 -11^ Endnotes -20 -18^* -20^ -23^* 1. Recent positive outcomes since welfare -30 Single-Parent Families reform have led some to argue that studies All Families -40 predicting higher levels of child poverty as a result of welfare reform were wrong. However, -50 these studies all assumed that federal benefit Persons (millions) Persons -53^ -60 time limits were fully implemented, that the -58^ economy performed as it did in 1995, and that -70 -70^* proposed benefit cuts were permanent. We -73^* -80 now know that assumptions about the economy proved to be too pessimistic (at least through 2000) and that some of the immigrant benefit cuts have been restored. Further, federal benefit time limits will only gradually roll Source: The 1997 and 1999 National Surveys of America’s Families, augmented by the Urban Institute’s TRIM3 to correct for underreporting of government benefits and to adjust for the value of food stamps and taxes. out across the states between 2001 and 2003. Note: The full-participation scenario assumes that all eligible families would participate in TANF, SSI, and food stamps, holding constant earnings and other sources of income. 2. Wheaton and Giannarelli (2000) show a *Significantly different from 1996 level at the 90 percent confidence level or higher. ^Significantly different from historical results (figures 1 and 2). deterioration in reports of government assistance in the March Current Population Survey, Table 1. Type and Number of Economic Hardships Reported by Families with the basis for the federal government’s poverty estimates. The use of corrected NSAF data for Children, by Disposable Income Level (percent), 1999 this analysis eliminates the chance that a change in income is due to higher levels of Extremely Poor Poor Near Poor underreporting of government assistance in Economic Hardship (Below 50% of FPL) (50–100% of FPL) (101–150% of FPL) one year relative to the other. Type: 3. The model uses the same procedures as a Two or more food problems 42 41 32* caseworker to determine eligibility. These pro- Crowded housing 24 24 18 Housing costs exceed 50% of income 46 17* 7* cedures are explained fully in Giannarelli, Unable to pay rent or utilities 30 32 29 Morton, and Wheaton (forthcoming). Moved in with others 5 6 3 4. The NSAF asked families how much out-of- Without phone for month or more 10 6 6 Without car 40 41 20 pocket expenses they incurred for child care in the past month. We use these reports to extrap- Number of hardships: olate annual child care spending for employed One 25 35* 33* families. Note that the use of out-of-pocket Two 31 23* 20* Three or more 30 24 13* child care spending indirectly accounts for increased government spending on child care Source: The 1999 National Survey of America’s Families. subsidies because these subsidies reduce fami- Note: Measures of poverty are based on total disposable income, defined as all cash plus food stamps and the EITC, less taxes and child care lies’ out-of-pocket costs. expenses. FPL = federal poverty level 5. Note that our estimates omit the value of *Significantly different from results for extremely poor at the 90 percent confidence level. some of the nutrition programs for low-income families (school lunch and school breakfast) private sector groups also have a role Our results demonstrate that pri- and the value of housing assistance (because of to play—effective outreach that edu- vate and public sector efforts to incorrect reports of assistance in round one of the NSAF). cates families about public assistance increase program participation could 6. For example, based on Current Population and facilitates their participation reduce poverty and economic hard- Survey estimates of the number of legal immi- would help. At the agency level, ship. grants ineligible for food stamps in 1998, efforts to develop family-friendly Alternative ways of increasing immigrant benefits are overstated by 7 percent. offices that can accommodate the the disposable incomes of families— 7. Congress recently increased the child tax schedules of working parents are such as minimum wage increases credit and made it refundable. This credit will needed. and investments in education and provide additional assistance to families with

6 An Urban Institute Program to Assess Changing Social Policies ASSESSING THE NEW FEDERALISM

children, but is outside the time period ana- Sections on Government Statistics and Social add a child care component to TRIM3, lyzed here. Statistics: 236–241. which was used to estimate the number of 8. See Citro and Michel (1995) for a full descrip- Zedlewski, Sheila. 2002. “Family Incomes: children and families eligible for federally tion of the alternative poverty measure recom- Rising, Falling, or Holding Steady?” In funded child care subsidies. mended by the National Academy of Sciences. Welfare Reform: The Next Act, edited by Alan They also recommended deducting Weil and Kenneth Finegold. Washington, Joyce Morton is a senior pro- out-of-pocket health expenses from income. D.C.: Urban Institute Press. grammer/analyst and one of the developers of the current References About the Authors version of the TRIM3 microsimulation model. Ms. Citro, Constance, and Robert Michel. 1995. Morton has extensive knowledge of the Sheila R. Zedlewski is direc- Measuring Poverty, A New Approach. core programs modeled by TRIM3. She tor of the Urban Institute’s Washington, D.C.: National Academy Press. also has in-depth knowledge of large data Income and Benefits Policy Giannarelli, Linda, Joyce Morton, and Laura sets including the Census Bureau’s Center. Her research deals Wheaton. Forthcoming. “Using the NSAF in Current Population Survey (CPS), the with welfare reform, low- the TRIM3 Model.” Washington, D.C.: The Survey of Income and Program income program participation, and pover- Urban Institute. Assessing the New Federalism Participation (SIPP), and the National ty. Her recent articles deal with the linkage Methodology Report. Survey of America’s Families (NSAF). between welfare reform and food stamp McConnell, Sheena, Michael Ponza, and Rhoda participation, and the changing character- Laura Wheaton is a research Cohen. 1999. “Report on the Pretest of the istics of families participating in cash Reaching the Working Poor and Poor associate in the Income and assistance programs. Elderly Survey,” vol. 1. Washington, D.C.: Benefits Policy Center at the Mathematica Policy Research. Urban Institute. Linda Giannarelli is a senior U.S. General Accounting Office. 2001. Means- research associate in the Tested Programs: Determining Financial Urban Institute’s Income and Eligibility Is Cumbersome and Can Be Benefits Policy Center. Her Simplified. Washington, D.C.: General research focuses on govern- Accounting Office Report 02-58. ment programs for low-income families. Wheaton, Laura, and Linda Giannarelli. 2000. She directs the Urban Institute’s Welfare “Underreporting of Means-Tested Transfer Rules Database project and is the principal Programs in the March CPS.” American investigator for the development and Statistical Association 2000 Proceedings of the maintenance of the TRIM3 microsimulation model. She recently led the effort to

7 THE URBAN INSTITUTE Nonprofit Org. 2100 M Street, N.W. U.S. Postage Washington, D.C. 20037 PAID Permit No. 8098 Mt. Airy, MD

Address Service Requested

For more information, call Public Affairs: This series presents findings from the 1997 and 1999 rounds of the National Survey of America’s Families (NSAF). Information on more than 100,000 people was gathered in each round from more 202-261-5709 than 42,000 households with and without telephones that are representative of the nation as a or visit our Web site, whole and of 13 selected states (Alabama, California, Colorado, Florida, Massachusetts, Michigan, http://www.urban.org. Minnesota, Mississippi, New Jersey, New York, Texas, Washington, and Wisconsin). As in all sur- To order additional copies veys, the data are subject to sampling variability and other sources of error. Additional information of this publication, call on the NSAF can be obtained at http://newfederalism.urban.org. 202-261-5687 or visit our online bookstore, The NSAF is part of Assessing the New Federalism, a multiyear project to monitor and assess http://www.uipress.org. the devolution of social programs from the federal to the state and local levels. Alan Weil is the project director. The project analyzes changes in income support, social services, and health programs. In collaboration with Child Trends, the project studies child and family well-being. This analysis and paper were funded by The David and Lucile Packard Foundation. The ANF project has also received funding from The Annie E. Casey Foundation, the W.K. Kellogg Foundation, The Robert Wood Johnson Foundation, The Henry J. Kaiser Family Foundation, The Ford Foundation, The John D. and Catherine T. MacArthur Foundation, the Charles Stewart Mott Foundation, The McKnight Foundation, The Commonwealth Fund, the Stuart Foundation, the Weingart Foundation, The Fund for New Jersey, The Lynde and Harry Bradley Foundation, the Joyce Foundation, and The Rockefeller Foundation.

THE URBAN INSTITUTE The views expressed are those of the authors and do not necessarily reflect those of 2100 M Street, N.W. the Urban Institute, its board, its sponsors, or other authors in the series. Washington, DC 20037 Permission is granted for reproduction of this document, with attribution to the Urban Copyright © 2002 Institute. Phone: 202-833-7200 Fax: 202-293-1918 E-mail: [email protected]