The Impact of certificate of need laws on infant and post-neonatal mortality

Abstract

Certificate of Need (CON) laws require healthcare providers to obtain prior government approval before expanding facilities or services. First mandated by federal legislation in 1974 and adopted by nearly every state, these laws were designed to contain healthcare costs by restricting supply. Their effects on infant health outcomes have never been rigorously estimated. This paper provides the first causal estimates of CON laws' effects on infant and post-neonatal mortality, exploiting the staggered adoption and repeal of CON programs across states from 1968 to 2019. Using a county-year panel constructed from CDC Restricted Use Vital Statistics data, I estimate effects separately across two policy margins - adoption and repeal - and two outcome families, with race-stratified analyses motivated by the persistent Black-White infant mortality gap. Identification relies on the staggered difference-in-differences estimator of Callaway and Sant'Anna (2021), which recovers cohort-specific average treatment effects while avoiding the biases of two-way fixed effects regression under heterogeneous treatment timing. On the adoption margin, CON laws are associated with a statistically significant increase in aggregate infant mortality of 0.791 deaths per 1,000 births (SE = 0.325), approximately 4 percent of the adoption-era baseline rate of 19.2 deaths per 1,000 births. Pre-treatment event-study coefficients are tightly clustered near zero, supporting the parallel trends identifying assumption. Race-stratified adoption-era models are directionally consistent - point estimates are positive for White infant mortality (0.348), Black infant mortality (0.974), and Black post-neonatal mortality (0.676) - but none reach statistical significance once the panel is stratified by race. On the repeal margin, the overall infant mortality estimate is a precise null (ATT = 0.151, SE = 0.378), with confidence intervals ruling out effects larger than roughly 10 percent of the repeal-era baseline. Post-neonatal and race-stratified repeal estimates are indistinguishable from zero. These findings are consistent with a mechanism in which CON entry restrictions reduced access to obstetric, neonatal, and outpatient pediatric services during the formation period of the U.S. hospital CON regime, modestly raising infant mortality in the adoption era.