Association for Public Policy Analysis & Management

About half of the children in the U.S. will live with a single parent before age 18. The poverty gap between one- and two-parent families has contributed to calls to strengthen child support policy as a way to reduce poverty and increase the income of single-parent families. In 2013, child support accounted for 70% of the mean annual personal income of custodial parents below poverty. However, less than half of custodial parents with a child support order received full payments, limiting the contribution of child support to reducing single parent poverty. Understanding the factors accounting for limited payments is a central concern of policymakers and researchers. Substantial prior research has relied on administrative data, but these data include limited information on families. Other research, with a richer set of covariates, relies on survey data. But how well do survey data reflect the actual payment of child support? Data from surveys might differ from administrative data for several reasons, which can limit the potential to identify the effects of the interventions as well as to compare results across studies using different data sources. This paper exploits integrated survey and administrative data on a large sample of disadvantaged noncustodial parents to examine the extent to which survey and administrative reports on child support are consistent and what is correlated with any discrepancies.

This paper compares the self-reported amounts of child support paid and orders of over 10,000 noncustodial parents with administrative records. We use new data from the Child Support Noncustodial Parent Employment Demonstration program, a federally-funded eight-state intervention for people behind in their child support payments and with limited employment, collected between October 2013 and September 2016. Participants in the baseline survey were asked to self-report the amount of child support they owed in the last month and the amount they paid in the last month. We compare the distribution of survey and administrative reports, discrepancies, and describe sample characteristics. We then use OLS regression to estimate the relationship between key characteristics (including family structure, income, age, education, race, and criminal history) and the size of the discrepancy. We explore the extent to which discrepancies between survey reports and administrative records are associated with the number of cases for which the noncustodial parent owes support (related to multiple partner fertility). We also evaluate hypotheses related to informal child support payments accounting for apparent over reporting of formal child support in surveys, relative to administrative data.