Child support, welfare dependency, and women's labor supply.(includes appendices)

I. Introduction

Policy efforts to alleviate the high rate of poverty among single-parent families have taken place along two lines: (i) defining and enforcing private child support obligations(1) and (ii) providing income support through an array of government-funded welfare programs including Temporary Assistance for Needy Families (TANF), Food Stamps, and the Earned Income Tax Credit. The comprehensive welfare reform of 1996 was aimed at reducing the public fiscal burden of fighting poverty by reducing the availability of welfare benefits, with the intention that some poor families would find adequate private sources of income support. Thus, the well-being of single mothers increasingly depends on their own capacity to earn income and on the extent to which child support income is paid.

Interest in child support reform has been driven by the fact that the failure of private individuals to provide income support increases the burden on public income transfer programs. Because poor single mothers will continue to be eligible for welfare benefits after welfare reform, this fiscal burden could be reduced further if private child support payments were increased. An added attraction of child support reform, even after welfare reform, is that keeping women with children off of welfare improves work incentives. The high benefit reduction rate (or the implicit tax rate) of the TANF program means that a woman on welfare earning an extra dollar of income typically sees her total post-welfare income rise by a small fraction of the extra dollar. If child support payments, combined with earnings and other sources of non-welfare income, are sufficiently high to keep a woman off of the welfare rolls, her earnings are not implicitly taxed at such high rates. Thus, child support reform is desirable because it may both reduce public welfare expenditures and avoid the distortionary effects embedded in the welfare benefit formula.

A few authors have provided insight into the connection between child support outcomes and welfare program participation (See Robins 1986, Bergmann and Roberts 1987, Graham and Beller 1989, Oellerich et al. 1989, Garfinkel et al. 1990, and Nixon 1994). Others have estimated the impact of proposed reforms of the child support system on female labor supply (See Beller and Graham 1985, Graham and Beller 1989, Garfinkel et al. 1990, and Graham 1990). With the exception of Graham and Beller (1989), Graham (1990), and Nixon (1995), all of these previous studies have disregarded the possibility that child support outcomes, labor supply, and welfare program participation are jointly determined. Unobserved heterogeneity across families introduces simultaneous-equations bias in conventional estimates of the effect of child support income on women's economic decisions.

In this study, I use data on women from the Panel Study of Income Dynamics (PSID) who experience marital breakup to determine the potential effects of child support reforms on female labor supply and welfare recipiency. My empirical framework addresses the endogeneity of child support payments in women's labor market decisions. This study also corrects for the selectivity bias in analyzing welfare participation and labor supply decisions over time due to the fact that remarriage rates among divorced/separated women may be related to factors affecting labor market choices. The statistical analysis recognizes the simultaneity of women's decisions regarding remarriage, welfare participation, labor force participation, and hours of work. In this way, I demonstrate that child support policies that increase support payments would (i) decrease welfare participation and (ii) increase the proportion of divorced/separated mothers who work. Despite some reductions in work among non-welfare mothers, the net effect on aggregate hours of work is positive - increases in child support income imply increases in average hours of work.

Finally, I assess the likely effects on female welfare participation, work behavior, and income of alternative child support policies determining the level of support obligations. Results show that significant impacts on the fraction of women who work and on welfare participation are achievable only through improvements in child support collections for mothers who currently receive nothing or through government-assured support payments. Nevertheless, significant savings in welfare benefit payments may be achieved even without dramatic improvements in child support collections. This study's results can thus help to guide the continuing evolution of welfare reform.

II. Child Support and Welfare Policy

The welfare reform legislation of 1996 gave states substantial flexibility to change their welfare programs for single mothers, formerly called Aid to Families with Dependent Children (AFDC) and renamed TANF. The variety and novelty of these state-level changes means that policymakers wishing to evaluate the likely effects of child support reform in the new welfare regime face an extremely difficult task. Given the lack of precedent on which to base predictions about the effects of child support reform under TANF, I turn to the next best alternative. Analyzing the pre-reform welfare system yields a useful benchmark for individual state policymakers, because TANF in most states will still tax away increases in either earnings or nonlabor income. The benefit structure is thus essentially unchanged, with the following exceptions described below.

First, states were required to impose work requirements on aid recipients under TANK The actual requirements put in place so far vary substantially from state to state, in terms of (i) whether the work requirements are in force earlier than the federally mandated 24 months following the start of a welfare spell, (ii) the number of hours per week, and (iii) whether activities such as job search or job training satisfy the definition of "work."(2) An important aspect of the federal mandate is that it applies only to the use of federal funds - a state can independently fund welfare benefits for subpopulations of single mothers upon whom it does not wish to impose work requirements. Thus, it is not clear what fraction of the TANF population will actually face binding work requirements. Because work requirements will mostly affect women in welfare spells lasting longer than two years (a minority of welfare spells), I will conduct an additional policy analysis at the end of the paper only on the subgroup of women in their first two years following marital breakup.

Second, states were allowed to impose limits on the amount of life-time participation for given individuals, with each state's limit not to exceed five years. Each state was allowed to exempt 20 percent of the caseload from this cutoff. (Again, the federal mandate does not exclude states from paying welfare benefits, funded from state sources, that are not time-limited.) As is the case for work requirements, some states have adopted time limits shorter than the federal mandate. Predicting the effects of this reform is well beyond the scope of this study. It is probably safe to say that time limits will have small effects on most divorced or separated women, however, because their expected time on welfare is well less than five years.

Third, states were allowed to end the one-for-one displacement of welfare benefits by child support payments. The policy simulations in Section VI will explicitly consider the effects of alternative ways of taxing child support income. This study thus serves as a guideline for individual states so that they can make informed policy choices.

III. The Effects of Child Support on Welfare Participation and Labor Supply

This section describes the effects of child support on welfare participation and labor supply suggested by economic theory and discusses their empirical identification. Theory suggests that increases in child support income will (i) decrease welfare participation, (ii) increase the likelihood of working, and (iii) have an ambiguous effect on total hours of work. This discussion serves to motivate the statistical model presented in the following section of the paper. The usual consumption-leisure diagram is the most convenient way to demonstrate women's decisions regarding welfare and work. While utility maximizing behavior is assumed, one does not need to postulate a specific utility function in order to illustrate the effects of interest.

A. Graphical Analysis

Figure 1 shows the effects of an increase in child support income on a divorced mother's budget constraint. For simplicity, the woman is assumed to have no nonlabor income initially and the AFDC(3) earnings disregards for work expenses and child care are ignored. In the absence of AFDC, her opportunities are represented by the line OBX, the slope of which is equal to (minus) the wage rate. The AFDC budget constraint is represented by OGBX. The length of the segment OG is the guarantee level and point …