Panel Paper: Ex Ante and Ex Post Moral Hazard: Lessons from the ACA and Risky Sex

Friday, November 3, 2017
Hong Kong (Hyatt Regency Chicago)

*Names in bold indicate Presenter

Barton Jay Willage, Cornell University

Health insurance has played a prominent role in the rapid growth in medical spending. However, the effect of insurance on risky health behaviors (ex ante moral hazard) is rarely discussed, let alone disentangled from the price effect of insurance (ex post moral hazard). Many studies find little or inconsistent evidence that insurance changes health behaviors. A major obstacle to observing ex ante moral hazard is that health behaviors often take years or decades to result in adverse health shocks, for example smoking and cancer. Future, rather than current, insurance status covers and theoretically affects many current health behaviors.

This examines how insurance affects risky sex and the resulting health consequences. There are two advantages of focusing on risky sex, defined here as sex between a man and a woman without a condom. First, many of the health consequences of risky sex occur quickly and should be affected by current insurance status. For instance, the effects of an unplanned pregnancy and the contraction of a sexually transmitted infection occur within one year. Second, recent policy changes allow me to partially isolate the behavior effect from the overall effect of insurance. I leverage two components of the Patient Protection and Affordable Care Act (ACA) to isolate the effect on risky behaviors from the overall effect of insurance. (1) The adult dependent coverage mandate of 2010 causes an exogenous shock to the extensive margin – how many people are insured. (2) The zero cost-sharing for prescription contraception mandate of 2012 – which essentially made prescription birth control free for women with insurance – changes the intensive margin of health insurance in terms of the degree to which each person is insured. By using two policies, I can disentangle some of the ex ante moral hazard effect caused by the zero cost-sharing mandate from the overall effect of health insurance caused by the dependent coverage mandate. Additionally, since I am using changes in federal law instead of state-level policies, there is less concern about policy timing endogeneity.

I use state-year level data and an extension of the dose-response difference-in-differences approach to determine the effect of each policy on a number of outcomes related to risky sex, such as condom sales, sexually transmitted disease, and fertility. The treatment for each policy in my analysis is the percent of each state that is potentially exposed; uninsured young adults are exposed to the dependent coverage mandate, and people with insurance are affected by the requirement that insurance cover contraception with zero cost-sharing.

The hypothesized effects of zero cost-sharing for contraceptives are clear-cut: lowering the cost of contraceptives decreases the cost of having sex without a condom and leads to substitution away from condoms. This, in turn, increases STDs and decreases fertility. The predictions of the overall effect of health insurance are mostly ambiguous. There is strong empirical evidence of ex ante moral hazard in risky sex in the form of increased STDs, and this result is obscured in the overall effect of health insurance.