Ex-ante Moral Hazard & Health Insurance: An RD Analysis of Changes in Coverage & Non-fatal Unintentional Injury Rates at Age 19

This paper is motivated by the finding that, in a regression discontinuity (RD) framework using CDC national estimates and year of age among 13–24-year-olds, the non-fatal unintentional injury rate has abruptly dropped at age 19 for most of the period since the early 2000s, but not in recent years. Specifically, the discontinuity was a significant 7–8% during 2002–2007 and 5% over 2008–2012, but has been small and insignificant since 2013. This pattern broadly matches the increasing eligibility for parental health insurance coverage over time among non-students ages 19 and above. Mandates extending parental coverage past age 18 were implemented by many states starting in 2007; ultimately the Affordable Care Act (ACA), as one of its better-received provisions, expanded allowable coverage until age 26 starting in late 2010. A decreasing, and eventually disappearing, discontinuous fall in injuries at age 19 is therefore consistent with ex-ante moral hazard behavior induced by parental health coverage: as dependents become less likely to lose eligibility upon turning 19-years-old, they have less incentive to become more careful to avoid injuries requiring costly medical treatment.

To further investigate this hypothesis, I estimate RD models using age in months on 15–22-year-old respondents to the National Survey of Family Growth (NSFG), most recently administered in 2002–2003 and then nearly continuously from 2006 through 2015. Consistent with the injury findings, current private insurance coverage among males dropped precipitously, by a significant 10 points (16%), through 2012, but by a much smaller, insignificant amount thereafter. However, other aspects of the age 19 change in health coverage, and how it varies temporally, are at odds with the injury results. The probability that males lose private coverage rises slightly, rather than declines, during the 2008–2012 state mandate and early ACA period. Meanwhile, among females, the loss in private coverage is negligible throughout the period.

More broadly, for each gender and distinct NSFG cycle (2002–2003, 2006–2010, 2011–2015), the likelihood of current coverage falls significantly and substantially at age 19, by an average of 10 points, or over 100% of the uninsurance rate among 18-year-olds. For females, this is almost exclusively attributable to losing Medicaid/SCHIP coverage. For males, this is largely caused by losing private coverage, although the overall effect persists into the later post-ACA period in which the main source of male coverage loss becomes Medicaid/SCHIP. The size of the age 19 increase in the uninsurance rate declines only slightly in magnitude, and not at all relative to the age 18 rate, from before to after the ACA mandate.

An implication of these results is that the comparative roles of private and public coverage in becoming uninsured upon turning age 19, and their differential effects by gender, bear further investigation. Other age 19 effects warranting additional examination include significant reductions in the probabilities that females report excellent health – but only post-ACA, and in contradiction to the injury results – and that males, but not females, have a usual place to obtain care.