Until 1984, the Office of Child Support Enforcement (OCSE) provided financial incentives to state OCSE offices for one objective: cost-effectiveness.  States were rewarded for child support collected on behalf of families receiving Aid to Families with Dependent Children (AFDC) divided by administrative costs.

In 1980, the Federal Government began offering states 66% matching funds to work the nonAFDC caseload, mirroring matching funds to AFDC cases. In 1984, incentive payments were also provided for the nonAFDC caseload although they were capped at 115% of the AFDC incentive payments.  This rewarded states for their collection efforts on behalf of the nonAFDC population and additionally maintained a balance of effort between the AFDC and nonAFDC caseload. 

How the nonAFDC caseload was handled by states once collections were incentivized is the focus of this paper.  Did states work the entire nonAFDC caseloads or did the incentive cap result in states cream-skimming the more affluent cases?  We hypothesize that such behavior might create a “donut-hole” in service provision for low income, yet nonAFDC, cases. 

With the passage of the Child Support Performance and Incentive Act (CSPIA) in 1998, the one-factor formula (cost-effectiveness) became a five-factor formula, adding: paternity establishment, establishment of child support orders, collection of current support orders, and collection of arrearage payments.

Methodology, Hypotheses, and Data

We create a forcing variable, AFDC welfare eligibility, to identify the impact of custodial parents being just above the AFDC/TANF eligibility cutoff on state OCSE performance on the four incentive metrics established by CSPIA.  We use a regression discontinuity design (RDD) which is well-known to produce consistent estimates of local average treatment effects and is arguably the best quasi-experimental approach to replicating experimental impact estimates, if properly implemented.  This analysis is conducted for the periods 1991-1995 (pre-CSPIA) and for 2000-2005 (post-CSPIA).  This analysis shows if and how the new incentive formula caused states to change their service provision to families just above the AFDC (or TANF) eligibility lines.

Our units of analysis are women who have children, in state i, year t, and welfare eligibility bandwidth k% of eligibility.  These state-year-cutoff cells will provide a large number of observations as well as a running variable, the AFDC eligibility.  Because AFDC eligibility is multidimensional (it is adjusted by number of children and has income and asset limits), some work has to be done to conform these factors into RDD, but early indications in our data suggest this problem can be ameliorated. 

The primary sources of data for this paper include the March CPS and SIPP as well as the TRIM3 system and the Welfare Rules Database available through the Urban Institute.

Conclusion

This research seeks to contribute to our understanding of CSE and federal/state relationships by examining how changes in federal incentive payment formulas altered the composition of families receiving OCSE services.  Specifically, did the 1984 formula lead to states cream-skimming nonAFDC cases and a failure to handle hard-to-serve cases?  If so, did the new CSPIA incentive formula lead to a more uniform provision of services?