Association for Public Policy Analysis & Management

The early childhood period is marked by a sensitivity to investments and insults that have a lasting effect on development and subsequently, adulthood well-being. It is unsurprising that maltreatment in early childhood predicts a litany of disadvantages, yet children age five and younger disproportionately account for 40 percent of substantiated abuse and neglect cases and nearly 90 percent of child maltreatment fatalities. Myopic funding structures incentivize removal over prevention, causing families with largely tractable hardships to fall in a gap where the risk severity is too low for preventative measures yet high enough to result in a maltreatment report or removal. Lack of access to affordable healthcare is one such hardship. Medicaid is a protective resource without close or suitable substitutes that without, can heighten the risk of child maltreatment through impoverishment and diminished mental and physical health. Accordingly, the aim of this paper is to explore the effect of access to Medicaid on child abuse and neglect rates using a restricted, administrative data of child maltreatment reports from 2009 – 2015 (NCANDS) capturing the full census of N=10,225,138 investigations aggregated to the county-month level. I identify the effect of access to Medicaid using the exogenous temporal and geographic variation induced by the county-level expansion in California from 2011 – 2012. I compare the pre- and post- expansion abuse and neglect rates of children living in expansion counties to those in non-expansion counties with a difference-in-difference framework (DD) estimated by OLS models weighted by the county population of children younger than five. As 12 states also adopted a Medicaid policy that covers maternal depression screening, I employ triple difference (DDD) models to test depression as a possible mechanism. To ensure exact counterfactuals and to account for the possible presence of pre-trends and omitted variable bias, I also employ synthetic control models (SC). I test for effect heterogeneity by stratifying models by multiple markers of disadvantage.

The pattern of results drawn from my preferred estimates suggests that Medicaid significantly reduces the number of maltreatment reports by up to 14.31 reports per month (5.9 percent of SD) and increases the substantiation rate by 1.02 percentage points (9.6 percent of SD). Reports made by medical personnel fall by nearly 17 reports per county-month, suggesting that newly-enrolled families are taking advantage of the preventative aspects of Medicaid services. These findings imply that expanding Medicaid reduces the probability that lower-risk families encounter CPS, allowing caseworkers to investigate and service higher-risk cases. Given that the primary population served has a higher prevalence of the major risk factors for maltreatment, expanding Medicaid to the remaining 18 states could be an efficacious solution to preventing maltreatment among young children, especially the low-to-moderate risk families, for whom the stress associated with hardship can be plausibly removed or reduced under a more generous healthcare system. This paper provides practitioners and policymakers new evidence to inform decisions about funding and expanding Medicaid and child protection strategies, and the possible negative consequences of a retraction of benefits or generosity through work requirements.