The Child Support Demonstration Evaluation Research Summary

Maria Cancian and Daniel R. Meyer, Principal Investigators

Collaborating Researchers: Judi Bartfeld, Tonya Brito, Emma Caspar, Judith Cassetty, Jane Collins, Steven T. Cook, Thomas Corbett, Robert Haveman, Thomas Kaplan, Sandra Magana, Katherine Magnuson, David Pate, Arthur Reynolds, Jennifer Roff, Gary Sandefur, Nora Cate Schaeffer, Judith Seltzer, Geoffrey Wallace, Barbara Wolfe, and James Ziliak

CSDE Contributors | List of Reports | Summary available in PDF format [103 KB]

February 2007

Note: Bracketed numbers in text link to full reports, and correspond with the List of Reports below.

Under the Temporary Assistance for Needy Families (TANF) program, states were given greater flexibility in determining how to distribute child support paid on behalf of children whose mothers were receiving welfare payments. The Wisconsin approach was unique. Wisconsin alone allowed all child support paid by noncustodial parents to pass through to the family and disregarded such payments in calculating welfare benefits; most other states decided to retain all the child support paid to offset welfare payments. This policy was put in place in 1997, as part of the original Wisconsin Works (W-2) program.

The Wisconsin policy was the subject of a full evaluation from its inception. The Child Support Demonstration Evaluation (CSDE), conducted by the Institute for Research on Poverty, included several primary components: a statewide random assignment experimental evaluation; quantitative nonexperimental evaluations using Wisconsin and national data on child support policies; analyses of policy implementation and monitoring; and qualitative explorations of family dynamics and responses to the new state policies. The results of all these analyses are summarized here.

The Policy Context

In 1996, the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) ended Aid to Families with Dependent Children (AFDC), and gave states substantial flexibility in designing replacement programs. Under the old program rules, most states passed through to recipient families the first $50 of child support collected each month, and disregarded this amount in calculating AFDC benefits. Any child support above $50 was, however, retained by the state and federal governments. The new program, TANF, allowed states to set their own policies regarding the state’s share of child support payments (the federal government continued to retain its share). Under TANF rules, most states chose to retain the entire amount of child support collected, passing none of the money on to the custodial parent and children.

Under a federal waiver received in 1997, Wisconsin was allowed to implement a different policy as part of W-2. Families were able to receive the full amount of monthly child support collected (“full pass-through”), and all of that child support income was ignored in the determination of W-2 eligibility (“full disregard”).

The CSDE Experimental Evaluation

The evaluation of the Wisconsin policy began shortly after the program itself went into operation. The central component of the CSDE was a random-assignment evaluation; most W-2 families received a full pass-through and disregard of monthly child support, but some child support was withheld from a randomly selected control group. This approach allowed evaluators to attribute any observed differences in outcomes between the two groups to the difference in the treatment of child support. The experimental evaluation and related research drew on a large, longitudinal database incorporating administrative data from several sources, and three waves of data from a longitudinal survey, the Survey of Wisconsin Works Families. The CSDE was completed in several phases, largely corresponding to experiment and policy changes.

Phase 1: The Primary Experimental Results

These results appear in the Phase 1 Final Report, completed after the end of the initial random assignment. Volume One [1] of this report includes the most comprehensive analyses of outcomes found in the experimental evaluation. Using data from the first two waves of the survey of W-2 families, it covers both direct effects on child support paid and received, and such potential secondary effects as mothers’ and fathers’ employment and earnings, parents’ interactions, and child well-being.

For both the experimental and control groups, the amount of support increased as time passed. The experimental group received more child support in each of three years, with a difference of $138 in the first year. The impact was larger among some subgroups, including those who had not received AFDC in the past two years (a difference of $156) and those who had a child support order at entry (a difference of $190).

This impact largely reflects the mechanical effects of the policy change, in which experimental-group mothers receive the full amount paid and control-group mothers do not receive the full amount if they receive W-2 benefits. We also find small, but statistically significant differences in the percentage of fathers who paid support, and thus the percentage of mothers who receive support. The proportion of mothers receiving support increases over time. Those in the full pass-through and disregard group are more likely to receive support in each year, though by the third year the difference is no longer statistically significant at conventional levels. Among those mothers who had not received AFDC in the prior two years, the impact was particularly large, with 35 percent of experimental-group mothers receiving support in the first year, compared to 30 percent of those in the control group. In subsequent work, we have found that not only was there an effect on whether fathers paid support, there was also an effect on how much fathers paid [28].

We also find significantly higher rates of paternity establishment in the first year for the experimental group. Among the nearly 15,000 nonmarital children who did not have a legal father when they entered W-2, 17.2 percent had paternity established by the end of the first year, compared to 15.7 percent of those in the control group. The difference disappeared by the end of the second year. Similar to the other outcomes, the difference is particularly large for those who had not recently received AFDC. In summary, a full pass-through and disregard is associated with faster paternity establishment, but not with a long-term increase.

While there are differences in some components of government costs, we find no difference in overall government costs between the experimental and control groups. Although more child support is passed through to those in the experimental group, not all of this is at the expense of government, since some consists of additional support that would not have been paid in the absence of the full pass-through and disregard. More importantly, the reform also generated cost savings in other areas, especially in W-2 cash payments. Finally, our estimates of cost differences do not consider the administrative savings that may have resulted from a simpler child support system.

Phase 2: Longer-term Outcomes for Families

The Phase 2 Final Report [7] presents longer-term outcomes for two groups: (1) the first cohort of families (those included in the Phase 1 report), and (2) a second cohort of randomly assigned families that entered W-2 six months or more after the first group. The report covers three to four years of experience with the new policy. Evaluation continued to demonstrate that the pass-through and disregard policy increased child support receipts. Because the later cohort entered after some improvements in W-2 and child support pass-through policy implementation, we had hypothesized that there would be a stronger experimental effect among this group. For the most part, we did not see that effect.

The End of Random Assignment

Beginning July 1999, all new families entering W-2 received the full pass-through and disregard. Among families already receiving W-2, those cases previously assigned to the control group continued to receive a partial pass-through and disregard until July 2002. At that point all cases, new and old, began to receive the full pass-through and disregard. These changes defined two additional cohorts of families receiving the full pass-through and disregard—those entering during a time of transition, when some older cases still received a partial pass-through and disregard, and those entering during a time when all cases received the full pass-through and disregard.

Comparisons of Outcomes in the Original and Extended Evaluation

The final annual report [25], completed as part of the extended evaluation, describes findings from six years of observations for the original two randomly assigned cohorts of families, and two to five years of observations for the two later-entering cohorts that received the full pass-through and disregard. The report corroborates results from the earlier years of the experiment that showed faster paternity establishment and child support payments. It also shows that most of the experimental effects did not persist, likely because families moved quickly off W-2 cash assistance and were only briefly exposed to differences in the experimental treatments.

The finding of limited longer-range effects does not necessarily mean that the full pass-through and disregard policy has not been beneficial. The narrow differences separating the partial pass-through and disregard as defined in the evaluation and the full pass-through and disregard received by most families mean that all W-2 cases in Wisconsin have benefited from a fairly generous disregard policy, compared to the no-disregard policy used in the majority of states.

The Nonexperimental Evaluations: Contexts for the Experimental Evaluation

In the CSDE, nonexperimental approaches were used to further explore potential effects of pass-through and disregard policy. The nonexperimental evaluations in the CSDE included four main components: three quantitative analyses that used both national and Wisconsin-specific data sources [6] and an ethnographic study of fathers of children in W-2 families [5]. Overall, the results of these analyses support the conclusion that increasing the pass-through and disregard will increase paternity establishment and payment and receipt of child support. They are discussed in the Report on Nonexperimental Analyses and compared to the original experimental findings [4].

Experimental and nonexperimental evaluations have different strengths. A key limitation of experimental designs is that they provide information only on the comparison of the policy regimes actually tested; they tell us little about the effects of other potential policies. Currently, most states retain all child support, and in those that pass-through and disregard some amount, the typical amount is $50 per month. Thus the results of the Wisconsin experimental design, which rely on a comparison of a full pass-through and disregard with a pass-through/disregard of the greater of $50 or 41 percent, cannot provide direct information on the effect of a full pass-through and disregard compared to full retention or compared to a straight $50 per month pass-through and disregard. In related research we compared outcomes as pass-through and disregard policies varied across states and over time [4, 22]. In theory, this approach should enable us to estimate the effects of the full range of pass-through and disregard policies observed. The key limitation to this analysis is the possibility that even after controlling for other observable factors, the groups facing different regimes differ in ways other than the pass-through and disregard policy.

An ethnographic study [5] provided a detailed description of the experience of African American fathers of children in W-2 Families. The study showed that the men had little understanding of child support policy. It also showed that many fathers were very involved in their children’s lives, and were willing to participate in the child support system but wanted their child support payments to provide a more direct benefit to their children.

The nonexperiemental results were largely consistent with the experimental results. The quantitative analysis found that a larger pass-through and disregard is associated with a statistically significant increase in the rate of paternity establishment and a small statistically significant increase in the proportion of cases with collections. Taken as a whole, the results support the conclusion that increasing the pass-through and disregard will not only increase the receipt of child support (a mechanical effect), but will also increase the probability of payment of child support (a behavioral effect) and may affect the amount paid.

Broadening the Scope of CSDE Research

In addition to the experimental and nonexperimental evaluations, the CSDE included a range of related research drawing from the core administrative and survey data, from ethnographic and field research, and from specialized data collection.

Understanding the Effects of a Full Pass-Through and Disregard

The most recent CSDE research efforts have increased the scope of the research. For example, the original evaluation included a process and implementation analysis [1, 3] which documented the challenges of explaining the new child support policy, especially in the context of a dramatic welfare reform. Building on this work, we completed a series of studies of participants’ understanding of the policy change.

We examined participants’ knowledge of child support policy rules and found that many parents do not fully understand the policy. We found evidence that child support agency staff provided useful information, and that those mothers who reported having heard media information were also better informed, which suggests that there are ways to directly improve policy knowledge [11, 16]. Our findings also suggest that the effects of the experiment would have been larger had the policy been better understood [11].
A report which examined workers’ knowledge of policy rules [17] found that staff in Milwaukee County, which had the largest concentration of families affected by the policy, believed that adoption of the full pass-through and disregard has increased custodial parents’ cooperation in the establishment of child support. Staff in several counties also noted that the pass-through and disregard policy facilitates the efforts of W-2 case managers to build constructive relationships with program applicants.
Two studies featuring focus groups and extensive interviews with white, black, and Latino families in Wisconsin [20, 21] found that both custodial and noncustodial parents were familiar with the pass-through and disregard policy in 2005, in contrast to the lack of knowledge found in a similar study of fathers in 1999 [5]. It was also clearer to participants in 2005 than in the earlier study that the policy benefited families.
A report examining the experiences of the American Indian population served by Wisconsin's W-2 program [24] looked at this subgroup of the population within its unique context of demographics, socioeconomic status, and different regulatory jurisdictions (e.g., tribal courts), and found that it differed substantially from the non-Indian W-2 population.

Other CSDE reports focusing on the effects of the full pass-through and disregard include a cost-benefit analysis [19], and a difference-in-difference analysis [18]. The cost-benefit analysis found that the majority of the net cost of the policy to the federal government of around $9.8 million over nearly six years is due simply to the loss of the federal share of child support that is passed through, and that the policy results in a net savings to the state of approximately $7.7 million over the same period largely because of lower average child care subsidies for those in the full pass-through and disregard group. The difference-in-difference analysis made use of the opportunity provided by the end of the child support pass-through experiment to assess the changes in outcomes for custodial and noncustodial parents associated with the full pass-through and disregard policy. We found consistently larger changes for those in the group formerly receiving the partial pass-through, but only the difference arising from the mechanical effect of the change to full pass-through on child support received was statistically significant.

Family Well-Being and Family Dynamics

Volume 2 of the Phase 1 report [2] documented welfare, child support, employment and a range of other indicators of family well-being for W-2 participants over the first two years of the program. Researchers found, for example, that despite the difficulties they faced, custodial parents had high levels of employment and substantial growth in employment and earnings over the short period considered. Average family income increased, and receipt of cash assistance payments declined dramatically over the first 24 months after families entered W-2. Subsequent analysis built on this work examining longer-term outcomes and special populations.

An analysis of the effects of the full pass-through and disregard on marriage, cohabitation, and living arrangements [23] found evidence that mothers receiving the full pass-through and disregard were less likely to be cohabiting with men who were not the father of their children, and were more likely to be single.
Several projects used ethnographic research to provide a more detailed understanding of the interactions of welfare and child support policies and the implications for family well-being. An examination of mothers’ family networks and livelihood [14] documented the complex family care responsibilities, health and other issues that shape participation in work.
An analysis of W-2 cases that transitioned to SSI caretaker cases [13], included a quantitative analysis which found that recipients of the SSI caretaker supplement were better off than before they entered the program, as well as a qualitative analysis which found that the population remains disadvantaged even while receiving the supplements, and continues to face many challenges.

Understanding Complicated Families

The original CSDE evaluation found that many W-2 families had a very complex family and household structure, including multiple children and also multiple fathers. Later CSDE projects focused on understanding complicated families and related policy.

An analysis of the implications of complicated families for child support policy [8] found that family complexity resulting from multiple-partner fertility is quite common, and suggests that states’ child support policies need to reflect this complexity.
An exploration of legal issues arising for complicated families [12] identified and analyzed cross-state variation in how guidelines treat additional dependents resulting from multiple-partner fertility.
The inevitable challenges inherent in designing a coherent policy for complex families were explored in related research [26, 27].
Finally, an ethnographic study of the experiences of families in other states [15] suggested that low levels of child support payments by low-income fathers are more likely attributable to low resources than to multiple-partner fertility. In the absence of support from their child’s father, most mothers worked in low-wage jobs, and many had a new partner contributing key economic resources.

Child Support Enforcement Projects

With the data collected for the CSDE, we have also been able to explore several topics that help us understand the lives of TANF families and the potential role played by child support. Two projects provide important background information on child support enforcement. The results of an analysis of the effect of child support arrears on subsequent payments [9] suggest that lying-in orders in particular, but not high arrears in general, contribute to reduced child support compliance among fathers of W-2 children. Another study found little evidence that child support enforcement efforts had any effect on nonmarital fertility or marriage [10].

What the Future Holds

Wisconsin’s child support pass-through and disregard policy changed again in January 2006; the Federal waiver that permitted the CSDE has expired, and the full-pass through is being phased out. This most recent change falls outside the scope of the CSDE. At the same time, Federal policy is changing in ways to encourage other states to follow Wisconsin’s original innovative approach. The 2006 TANF reauthorization allows states to pass-through and disregard the first $100 per month of child support for one child families, and $200 per month for larger families without reimbursing the Federal government for its share of the support. The results of the CSDE demonstrate the potential advantages of this policy.