|
| OCSE Home . Program Information . News . Publications . Policy . State Links . OCSE Search . Help |
Final Report
Department of Health and Human Services
Office of Child Support Enforcement
August 2003
OVERVIEW
The study underlying this report was conducted for HHS, Office of Child Support Enforcement (OCSE) under contract with the Lewin Group. The study considered how such external factors as demographic, economic, and programmatic variables affect state child support performance on the five child support incentive measures-paternity establishment, order establishment, current support, arrears support, and cost-effectiveness. The report found several external factors affected state performance and suggested a potential arithmetic adjustment that would alleviate the effect of these factors for four of the five incentive measures (all but paternity establishment). However, due to study limitations, an insufficient adjustment period to the new system, and related concerns detailed further below, the Secretary does not recommend changes to the system at this time.
BACKGROUND
In 1998, Congress enacted the Child Support Performance and Incentive Act (CSPIA), to revise the incentive structure to reward states for performance on a larger number of their establishment and enforcement practices. Specifically, Congress linked incentive payments to a state's performance in five areas:
In addition to other studies of the new incentive system, CSPIA required a report on how various demographic and economic variables affect state performance. OCSE contracted with the Lewin Group to conduct this study.
THE LEWIN STUDY ON STATE DEMOGRAPHIC, ECONOMIC, AND PROGRAMMATIC VARIABLES
The study by the Lewin Group utilized performance results reported by states in fiscal years 1999 and 2000. To isolate the relationships between the performance indicators and their determinants, a regression model was developed for each performance measure. In addition, the Lewin report provides an arithmetic model that could be used to adjust raw performance data reported from each state with the aim of holding states harmless from economic, programmatic, and demographic characteristics that affect performance but are out of the state child support enforcement agencies' control.
State-level data on over 50 demographic, economic, and programmatic variables that have potential relationships with child support performance were considered. The final six models relied on different combinations of 12 explanatory variables, with an additional variable for data reliability/audit passage or failure.
Economic factors:
Demographic factors:
Programmatic factors:
Data Reliability variable:
Study Results from the Lewin Report
The ability to explain variation in performance through demographic, economic, and programmatic variables varied widely between the five performance measures. No factors were found that relate clearly to all five measures. The factors determined to possibly affect performance are listed below, followed by HHS interpretation of the results, and our recommendation to not alter the performance based incentive system.
Demographic factors associated with state performance variation: Stability of the local population, percent living in urban areas, and the age of TANF household heads were all found to be related to state performance. A higher share of urban dwellers was associated with weaker performance on all the indicators except paternity establishment. The study found that the more stable a state's population-as evidenced by the percent who remain in the same house from year to year-the better the state performance. Finally, the study found that states with TANF caseloads with younger heads of households exhibited weaker performance for IV-D paternity establishment, current collections, and cost-effectiveness.
Economic factors associated with state performance variation: A robust economy -- as evidenced either by high personal incomes, low poverty rates, strong job growth, or high rates of employment among males -- was found to be associated with better performance in general. This, of course, was expected in a program that collects money. A higher poverty rate was associated with weaker performance on the cases with orders and arrearage measures. Personal incomes did a better job in predicting rates of current collections, with higher personal incomes linked to better performance. A higher rate of males not working were shown to depress performance on current collections and cost- effectiveness. Finally, a higher rate of job growth was associated with better performance on the arrearage measure.
Programmatic factors associated with state performance variation: Program universality, cases per staff member, expenditure per case, and the process for establishing orders were found to be related to performance. The study found that states with a higher share of IV-D cases receiving TANF exhibit weaker performance on the statewide paternity, cases with orders, current collections, arrearages, and cost- effectiveness measures. In addition, the study found that resources devoted to enforcement-expressed in total dollars or staff-also show relationships with performance. Specifically, a lower ratio of cases to total program staff is related to better performance in cases with orders and current collections. An increase in total spending per case appears to strengthen performance for statewide paternity but weakens the cost-effectiveness ratio. Finally, the process by which states establish child support orders appears related to their performance on cases with orders. The adoption of administrative processes appears to be related to better performance in order establishment.
Study Limitations
The analyses had several limitations. First, data quality continues to be problematic for some states, particularly with the paternity establishment percentages, cases paying toward arrears, and current collections. To adjust for the most serious problems of missing or miscalculated data, certain states were dropped from the analyses by Lewin. This resulted in limited observations in some of the analyses.
Second, there was also concern about having omitted potential determinants of performance for which no measure is available. For example, in addition to the number of staff in each state, the quality of staff and management may also affect performance. Likewise, the functionality of a state's computer system may affect performance. In short, there are a number of factors that may affect performance that were omitted from the analysis. To the extent those omitted variables are important in explaining performance, the findings may be misleading, oversimplified, or focused on the easiest-to-measure variable rather than the true causal factors. Similarly, the system of testing variables utilized in the report may overemphasize the relative importance of certain explanatory variables.
Finally, the statistical models rely on a relatively limited number of observations. In the future, researchers will benefit from additional years of data, which will allow time-series analyses and pooling over larger numbers of years.
RECOMMENDATIONS OF THE SECRETARY OF HEALTH AND HUMAN SERVICES
At this time, we recommend that no alterations be made to the child support incentive system with the intent of adjusting for demographic, economic, and programmatic differences among the states.
We acknowledge that there are circumstances and situations other than state efforts that may affect how well or how poorly a state may perform on incentive measures. However, in addition to the study limitations listed above that severely restrict the informative value of the results, we think that adjustments that respond to the study's concerns may cause more harm than good for the following reasons:
In our view, any effort to alter the system so that incentives are paid on the basis of variables other than performance risks the credibility of the system as a whole and the intended benefits for families. In short, we find the regression models presented in this report too speculative to merit alteration of the current incentive system.
