Children and Welfare Reform:

What Policy Theories are Being Implemented in States

Where Most Poor Children Live?

By

Cathy Marie Johnson

Associate Professor

Williams College

Thomas Lewis Gais

Director, Federalism Research Group

Rockefeller Institute of Government

Catherine Lawrence

Research Associate, Federalism Research Group

Rockefeller Institute of Government

When Aid to Families with Dependent Children was replaced by Temporary Assistance for Needy Families (TANF), a main argument for adopting a block grant rather than an entitlement was that states should have leeway to decide how to structure their welfare programs. Advocates of devolution argued that states could accomplish more with less money if they were not constrained by federal requirements. States could identify their most important problems and figure out how to mitigate them by devising solutions pertinent to their own communities and economic conditions. Devolution would thus allow states to achieve responsive government and policy innovation.

Opponents were not sanguine, however, about the possibility that states would meet these expectations. Concerned about incentives for supplantation, they feared states might use federal funds to replace rather than buttress state funding for social welfare policies. Moreover, many state governments had a reputation of poorly representing disadvantaged populations, suggesting little interest in searching for new ways to reduce poverty. Also, even if states wanted to pursue creative policies, few had the resources or the administrative capacity necessary for policy innovation.

In this article, we look at the welfare programs of states that faced a particular problem—high rates of child poverty. Most political actors admit that child poverty is a problem because it limits child development and prospects for a full adult life, even though they differ on the nature of the problem and its causes. Do states that share this problem have similar welfare programs? Do these states use the flexibility granted to them under devolution to develop innovative policies concerning children? Do they differ in important ways from other states, such as states with low child poverty rates? We investigate how sixteen states with high rates of child poverty responded to welfare reform.[*] In 1999 these states had poverty rates among children above 20 percent, ranging from 21 percent in Kentucky to 27 percent in New Mexico. To help us see whether these states are implementing distinct approaches to welfare reform, we compare their policies to states with low rates of child poverty.[†]

While AFDC was and TANF is a children’s program, political actors do not all agree on what constitutes child wellbeing or what government should do to advance it. In previous research, we argued that there were three different policy theories about child wellbeing contained within the 1996 Personal Responsibility and Work Opportunity Reconciliation Act (Johnson and Gais 2001). The environment theory maintains that children reap psychological and sociological benefits from being part of a family in which the head of the household works. The resource theory holds that children benefit from increased resources as caregivers enter and progress in the workforce and not simply from having a working parent. The family structure theory contends that children benefit from being born and raised in a particular kind of family—specifically married, two-parent families—and suffer when raised in a single parent family or by unmarried parents. Drawing on field research conducted in 1997-98—that is, the first years of state implementation of TANF—we found that states were most likely to implement the environment theory and were less likely to create policies, programs, structures, and processes that put the other two theories into effect.

In this article, we revisit those findings in light of a wider base of data, including two additional rounds of field research, and with a specific focus on sixteen states. We find that states with high rates of child poverty continue to have TANF policies consistent with the environment theory; they emphasize going to work and staying off or getting off public assistance. These states have not expanded the resources available to poor families, although many other states have expanded work supports and other resources for working families since the first years of TANF implementation. Finally, all states have done relatively little with respect to family structure theory, although states with high child poverty devote more money to such policies.

The Environment Theory

The first theory involves changing the child’s home environment by moving parents into the workforce. According to this line of thought, a child should be raised in an environment strongly connected to the world of work. A working parent would provide a role model for children by demonstrating that adults are expected to work for their livelihood and not rely on public assistance. Having a working parent would also structure the home environment by creating daily routines and schedules, which some proponents viewed as important for child development. Whether employment increased household resources so that children were no longer poor is not central to this theory. Rather, exposing children to the world of work and removing them from the world of welfare are essential elements for improved child wellbeing.

The broad structure of the federal act contains the policy elements critical to the environment theory. Work requirements and time limits on assistance advance the theory’s central goals of getting adults into the workforce and off the rolls. Thus, if states develop welfare programs consistent with these federal contours and do little or nothing else regarding children, their programs fit with the environment theory. Time limits on assistance, stricter sanctions, work requirements, and other policies and signals stressing work would by themselves be expected to have beneficial effects on children.

Much research has found that states moved quickly to develop and implement welfare programs that strongly signaled the importance of employment and getting off the rolls (Nathan and Gais, 1999; Gais, Nathan, Lurie, and Kaplan, 2001). States with high rates of child poverty are no exception, and overall, their welfare programs are highly consistent with the environment theory. They have adopted policies intended to discourage individuals from coming on the rolls, strong work requirements that apply to women with children, and work programs emphasizing “work first” or immediate placement in entry level employment. Indeed, with respect to these policies, their TANF programs are similar to those states with low rates of child poverty.

All but two of the high poverty states have at least one policy designed to encourage applicants to work and discourage them from coming on the rolls. These include a formal diversion program and mandatory job search requirements for applicants. States with low child poverty are only slightly less likely to have either of these policies. Four of those 17 states have neither a diversion program or mandatory job search. But 13 have at least one, and like the high poverty states, they also prefer a diversion program to mandatory job search.

For individuals who receive TANF cash assistance in the high poverty states, participation in state work programs is required shortly after going on the rolls. Similar to other states in the nation, most of these states require that adults comply with work requirements immediately upon or shortly after receipt of aid. The low poverty states were again slightly more generous with five states using the federal government’s 24 month time frame. But 10 of these 17 states also imposed work requirements immediately and two gave adults just a few months before requiring their participation in the work program.

States have backed up these expectations about work by making it clear they apply to women with children. Under federal law, states can exempt single parents caring for a child under one year of age from work requirements. Among the high poverty states, seven states have opted not to do this. Only one state, Texas, uses a higher age, granting an exemption to single mothers with toddlers. This is not out of line with the policies of other states although there is more variation among the states with low child poverty. More of these states have a higher age limit for their exemption but more also make this decision on a case-by-case basis.

States with high child poverty are less likely to recognize child care problems as a good cause for noncompliance with these work requirements. Federal law specifies that states cannot sanction individuals for not complying with work requirements if child care is unavailable for children up to age five. Some states, though, have expanded this good cause reason to include school-age children. States with high child poverty rates were less likely to do this than states with low child poverty, a move consistent with the environment theory because it weighs so heavily the importance of work.

Accompanying these strong work requirements are tough sanctions and lifetime limits on receipt of welfare. Sanctions enable states to enforce their work requirements, and high poverty states resemble low poverty states in the structure of their penalties. A majority of both groups partially reduce grants in the first instance of noncompliance and then end cash assistance entirely for the ultimate sanction. Only five of the high poverty states and four of the low poverty states end the ultimate sanction when the adult comes into compliance with the program requirement; a majority of states continue the sanction for some set period, although this ranges considerably from a low of one month in Arizona to a lifetime sanction in Mississippi and Georgia.

Time limits underscore the need for adults to enter the labor market quickly in hopes of gaining the work experience needed to acquire steady employment before time limits apply. Few states have decided to forego time limits on the receipt of aid, and not surprisingly, none of the states with high child poverty rates have decided to do this, although Arizona has an intermittent time limit (24 out of 60 months) with no lifetime ceiling. Among the low poverty states, only Vermont has no time limit at all, and Massachusetts has a policy like Arizona’s. Most of the high and low poverty states have adopted the federal government’s five-year limit, with a few in each group opting for lower limits. Most states in both groups have also decided not to extend benefits to children if their families hit the lifetime limits.

With respect to their work programs, states with high child poverty operate programs that emphasize work first activities. While these states are slightly more likely to disallow activities related to education and training than activities related to immediate employment, they accept most of the work activities allowable under federal law and compare favorably to states with low child poverty. But in most states, individuals participating in the work programs are involved in activities related to immediate entry into the labor market, such as unsubsidized employment, job search, or work experience. Among high poverty states, on average, 87 percent of participating adults are engaged in work activities related to immediate employment, similar to the low poverty states at 88 percent. Some of the high poverty states do have higher proportions of adults participating in education and job training activities, such as Arizona at 37 percent and Georgia at 27 percent. We know from Institute field reports that this training is not necessarily for higher-level, better-paying jobs. In Georgia, for example, “clients are told to ‘get a job, any job,’” and training programs are short term and generally used for individuals who cannot find entry-level positions through job search.

Resource Theory

Proponents of the resource theory focus on increasing family income and other resources in order to improve children’s wellbeing. Maternal employment is an important part of this endeavor, not because it changes the environment in which children are raised, but because it enables families to meet their material needs. Employment alone, however, generally does not raise family incomes sufficiently, leading proponents of the resource theory to emphasize “make work pay” initiatives, including income subsidies and work supports such as child care assistance.

Our initial review of state implementation in 1997-98 found that states did little to advance the resource theory, as they devoted their early efforts to signal expectations about work (Johnson and Gais 2001). Since then, however, states experienced fiscal surpluses under TANF, and many responded by increasing spending on services and benefits for working families, such as child care subsidies, enhanced earnings disregards, and transportation assistance (Gais, Nathan, Lurie, and Kaplan 2001).

But this response was conditioned by the structure of the block grant, which allocated different levels of resources per child in needy families across the states. The formula used for the TANF block grants froze into place the sharp differences in federal funding across states under AFDC. Such differences were reinforced by the maintenance of effort (MOE) requirements under TANF, which were keyed to state expenditures in the same time period. A supplemental grant reduced some of these differences in total federal and state spending, but they remained substantial. As Table 1 indicates, states varied greatly in the fiscal size of their federal and state TANF programs relative to the number of low-income children living in their boundaries. These differences are strongly correlated with the proportion of children who are poor in each state. Most of the states with high child poverty have few resources per low-income child under the block grant, while just the opposite pattern exists for states with low concentrations of poor children. TANF thus provides and mandates fewer resources per poor child in states where such children are most likely to live.

Table 1

Number of states by fiscal resources per low-income child under TANF, stratified by proportion of children under the federal poverty level

	Under $401 per child	$401- $800 per child	$801- 1,200 per child	$1,201-1,600 per child	Over $1,600 per child
States with highest proportions of low-income children (N = 16)	4	6	4	1	1
States with relatively low proportions of low-income children (N = 17)	1	4	4	1	7
All states (N = 50)	7	16	14	3	9

Spending ratios per child are calculated as total federal TANF, state MOE, and federal supplemental dollars, divided by the number of children in the state under 125 percent of the federal poverty level (averaged across 1997, 1998, and 1999); rows represent states with different proportions of children under the federal poverty level (for example, states in top row with highest proportion are those with 20 percent or more of the children in the state under the federal poverty level)

Fiscal resource differences across states are related to state policy choices, even after political and other state-to-state differences are taken into account (Gais and Weaver 2002). We found that states with high rates of child poverty were less likely to adopt policies that made it easier for families to combine work and cash assistance than their low-poverty counterparts. Most states increased their earnings disregards after the enactment of TANF in 1996, allowing families to keep more of their income as they increased their wages and hours worked. States with high concentrations of child poverty were less likely to adopt generous earnings disregards when compared to other states, particularly those with low child poverty, and the disregards tended to decline over time, especially after the first four months of work. The message of these states was that combining earnings and assistance was only a short-term option. Even if such states did adopt significant disregards, their benefit levels remained low (with two exceptions, New York and California) and thereby limited the income supplements for working families. Thus, among states with high child poverty, the potential gains for families from combining cash assistance and earnings were smaller on average than for other states.

Admittedly, however, the role of cash assistance under TANF is small and generally shrinking. Like other states, states with high child poverty rates have shifted their spending away from cash assistance and toward services and in-kind benefits important for working families, such as child care, transportation assistance, job services, post-employment training, and case management. Still, they have increased their spending on these nonassistance programs much less than states with low child poverty.

To understand whether states are really making major changes in their human service priorities—or whether they are simply shifting around program expenditures to take advantage of TANF’s flexibility—the Institute conducted a study of overall changes in spending on all nonhealth human services before and after the implementation of welfare reform. Data from 17 states in fiscal years 1995 and 1999 are available now. These data included all nonhealth expenditures in state budgets—not just TANF money—including spending from federal as well as state sources. Of these 17 states, five have high rates of child poverty and six have low rates of child poverty. We examine broad spending in these states to see overall shifts in state priorities, not just idiosyncratic differences among states in decisions about what programs to move under the broad TANF or MOE umbrellas.

In 1995, before TANF, important differences in median expenditure levels already existed between states that had high child poverty rates and those with low rates in all spending categories, demonstrated in Table 2. Regarding cash assistance and other basic needs, states with many poor children spent less than half of what the low-child-poverty states spent per poor child. The difference was greater for work supports. Finally, since most states spent little on child care, the difference between states with high and low child poverty rates was not great.

In 1999, these differences persisted, but their magnitudes changed. States with high child poverty rates reduced their spending on cash assistance per poor child by more than half, while increasing their spending in all other categories. However, these increases were not nearly as large as the spending increases in the states with low child-poverty rates. As a consequence, 1999 showed larger differences between the median spending levels of states with high and low poverty levels—most notably in the case of child care.

Table 2

Median levels of spending on social programs per poor child,

Fiscal Years 1995 and 1999

	Cash assistance	Other Basic Needs	Child Care	Work Supports	Other Welfare Related
	1995
States with highest proportions of low-income children (N = 5)	395	351	128	86	112
States with low proportions of low-income children (N = 6)	813	834	166	247	256
	1999
States with highest proportions of low-income children (N = 5)	163	469	190	206	128
States with low proportions of low-income children (N = 6)	294	1,045	367	513	309

Why have states with high child poverty rates failed to increase their spending on services to the same degree as other states?[‡] Many of these are southern and western states that historically offered lower levels of benefits to low-income families, and their political cultures, economic conditions, and other factors affecting such policies probably did not change enormously in the 1990s. However, because of the relatively small block grants per poor child in most of these states, the states did not have the resources under the TANF block grant to greatly increase their spending on services and in-kind benefits—even if their political and economic circumstances had changed substantially. Also, despite the fact that the southern and mountain states typically showed larger cash caseload declines than other states, these declines did not produce the same fiscal savings under their cash assistance programs as did states with more generous benefit levels.

That is not the complete story for these high poverty states, however. New York and California had relatively large cash assistance programs and benefited from large TANF grants. Yet neither of these states underwent the same degree of transformation—from reliance on cash assistance to a greater focus on services and in-kind supports—as was common in many other states with similarly large TANF grants. In the fiscal effects dataset, California and New York are among the few states that still rely much more heavily on cash assistance than most other states in fiscal year 1999. Although we cannot yet be sure, one factor seems to be the heavy reliance of these states on their counties to implement welfare reforms. Although some counties may be quite eager and able to shift toward service strategies, many are not. The weaker increase in spending on services among high child poverty states may thus be attributable in part to the small TANF grants, political traditions, and other variables that have limited resources for poor families in the southern and mountain states; and on the fact that child poverty is also concentrated in two large and highly decentralized states that have faced important institutional barriers in shifting toward effective service strategies.

Family Structure Theory

We have called the third theory, highlighted in the preamble of the 1996 federal legislation, the family structure model. Children are assumed to do best when they are born to and raised by married couples living together--and to do poorly under virtually any other arrangement. Because they do not try to reduce the number of single mother families, policies that require work and reduce welfare receipt among single mothers are considered to be imprudent by the most vocal proponents of this theory. Welfare policy instead, they argue, should focus on encouraging and sustaining marriage, and discouraging the birth of children out of wedlock.

Increased state flexibility under TANF is particularly large in this program area. States are free to deny aid to unwed teen mothers and to children born to parents already receiving assistance. Other elements in the federal act, such as restrictions on teen mothers, performance bonuses for states that reduce out-of-wedlock births, and special funds for teaching sexual abstinence, were designed to encourage states to create and implement programs that try to increase the proportion of children born in married, intact households. States with high child poverty rates may find family structure theory particularly attractive as multiple parents may increase the economic and emotional resources available to a child and thus reduce the pressure on other institutions to provide those resources.

During the congressional debate about the adoption of welfare reform in 1995-1996, much of the discussion about family structure centered on proposals to deny aid to unwed teen mothers and to children born to women already receiving assistance. The focus was on manipulating eligibility requirements in hopes of reducing the number of children born out of wedlock. None of the states have adopted a policy of denying aid to unwed teen mothers, and half of the states with high child poverty rates have adopted family caps, compared to 8 of the 17 states with low child poverty rates. Because family caps apply to children born to married as well as unmarried parents, it is unclear to what extent they are expected to alter family structure. Overall, states have not moved aggressively to try to alter family structure by denying aid to single mothers and their children.

They have relaxed, however, eligibility rules that made it more difficult for two-parents to receive aid. Under AFDC, three eligibility restrictions could be applied to two-parent families; 15 of the 16 states with high child poverty adopted all three, compared to 12 of the 17 states with low child poverty. Among the states with high child poverty, 10 now have no restrictions for two-parent families, compared to 12 of the 17 low poverty states, and only one (Mississippi) maintains three restrictions. Relaxing all three of these rules for two-parent families has been the modal response across the states; thirty-three states have eliminated all restrictions.

Other than these changes regarding eligibility, spending patterns among the states indicate how limited state attention to family structure strategies has been. In federal fiscal year 2000, the fifty states spent minimal portions of their TANF grants (including MOE) on efforts to promote or maintain two-parent families and efforts to prevent unplanned pregnancies. State spending on family structure does vary, however, by the concentration of child poverty in the state, if only slightly. States with low child poverty rates spent on average 0.30% of their TANF dollars on two parent family formation and pregnancy prevention. States with the highest levels of child poverty spent the most, on average 1.96% on the combined efforts of two-parent formation and pregnancy prevention.[§]

State initiatives in this area are quite varied. Some states, such as Arizona and Oklahoma, focus on promoting marriage through exhortation and information. Arizona has adopted a marriage commission, a marriage skills course, and free marriage manuals to all applicants of marriage licenses. Oklahoma got national attention for its “marriage ambassadors,” two academicians who give seminars on relationship skills. West Virginia has adopted a financial incentive by offering a cash bonus to single women who marry while on welfare, a practice the state was planning to discontinue until it received national attention during reauthorization discussions. Some programs are service oriented, such as West Virginia’s youth development programs, while others are educational, such as Florida’s new requirement that all high school students take a class on marriage and the family. Other states target family planning and pregnancy prevention. In Georgia, family planning is deeply integrated into the welfare system as counseling is mandatory for all TANF recipients. In New York and Texas, however, family planning efforts are generally independent of TANF, funded by other sources, and administered outside of the welfare agency.

The variation in these state programs may eventually lead to a few innovative policies widely adopted by other states. Certainly, the Bush administration hopes that it can spur creativity through encouragement and grant money. But the initiatives adopted to date suggest few cohesive ideas about what government can do to induce adults to bear and raise children in married two-parent families. While many argue that children should be raised in two-parent families, there are more claims about the desired outcome than agreement on ways to achieve it. Despite the flexibility in the federal law, states have not embraced family structure theory as a key strategy in current welfare programs.

Conclusion

Compared to the other policy theories concerning child wellbeing, adoption of the environment theory is relatively widespread among the states, including those with high rates of poverty. Policies consistent with the environment theory are easier to design and implement. The policy ideas are not complicated, and the assumption that child wellbeing can be achieved by having a working parent reduces considerably the obligation to provide additional programs for children. This does not mean that policies consistent with the environment theory hold out no pitfalls for these states. Successful implementation of the environment theory relies heavily on the private sector for jobs, and states with high rates of child poverty may find it difficult for their local economies to generate sufficient entry-level positions.

Although states are doing more than they were in 1997-98 to implement the family formation goals of TANF, we still see a rather scattershot, incremental approach by all the states. Although states with high child poverty rates are spending more on average than other states, even in these states efforts are marginal and unfocused.

Perhaps most difficult to implement—and certainly more costly—is the resource theory. This approach to welfare reform has spread significantly since the first years of TANF implementation as many states built greater work incentives into their cash assistance programs and began to transform their welfare systems toward the provision of work supports and services. However, these developments did not spread evenly among all states. Most important, the expansion of services and the provision of the most generous disregards are much less evident among states with high concentrations of child poverty. To the extent that the resource theory is viewed as critical to child wellbeing (see Duncan and Lansdale 2001), the problem of extending elements of this model to these states ought to be a priority. Important policy changes to consider include revising the fiscal formula in a way that provides greater resources to states with high rates of child poverty, while at the same time ensuring that states spend the money on low-income families, and helping to build the capacity of and incentives for local governments to deliver work supports and other services to poor families.

References

Boyd, Donald J., Patricia L. Billen, and Phil Dearborn. 2002. The Fiscal Effects of Welfare Reform: Final Report. Albany, NY: Rockefeller Institute of Government.

Duncan, Greg J., and P. Lindsay Chase-Lansdale. 2001. “Welfare Reform and Children’s Well-Being.” In The New World of Welfare. Edited by Rebecca Blank and Ron Haskins. Washington, DC: Brookings Institution Press.

Gais, Thomas L., Richard P. Nathan, Irene Lurie, and Thomas Kaplan. 2001. “Implementation of the Personal Responsibility Act of 1996. In The New World of Welfare. Edited by Rebecca Blank and Ron Haskins. Washington, DC: Brookings Institution Press.

Gais, Thomas, and R. Kent Weaver. 2002. “State Policy Choices Under Welfare Reform.” In Welfare Reform and Beyond: The Future of the Safety Net. Edited by Isabel V. Sawhill, R. Kent Weaver, Ron Haskins, and Andrea Kane. Washington, DC: Brookings Institution Press.

Johnson, Cathy M., and Thomas L. Gais. 2001. “Welfare Reform, Management Systems, and Policy Theories of Child Well-Being.” In For Better and For Worse: Welfare Reform and the Well-Being of Children and Families. Edited by Greg J. Duncan and P. Lindsay Chase-Lansdale. New York: Russell Sage Foundation.

Nathan, Richard P., and Thomas L. Gais. 1999. Implementing the Personal Responsibility Act of 1996: A First Look. Albany, NY: Rockefeller Institute Press.

---

[*] The states are Alabama, Arizona, Arkansas, California, Florida, Georgia, Kentucky, Louisiana, Mississippi, Montana, New Mexico, New York, Oklahoma, South Carolina, Texas and West Virginia. Although these constitute less than one-third of the states, 58 percent of all poor children in the U.S. live in these states

[†] The states with low rates of child poverty are Colorado, Connecticut, Indiana, Iowa, Kansas, Maine, Maryland, Massachusetts, Minnesota, Nebraska, New Hampshire, New Jersey, Utah, Vermont, Virginia, Washington, and Wisconsin.

[‡] Although states with moderate rates of child poverty are not displayed in the tables, in fact they too show higher median levels of spending on social programs per poor child than the states with the highest rates of child poverty.

[§] States with moderate levels of child poverty spent 0.60%.