The spillover effects of health insurance benefit mandates on public insurance coverage: Evidence from veterans

Highlights

• We consider the spillover effect of private insurance regulation on public coverage.

• We examine the impact of mental health parity on ESI and public insurance coverage.

• We focus on veterans who have better access to public insurance compared to others.

• Mental health parity reduces coverage through ESI by 2.1% points.

• The drop in ESI is largely offset by Veterans Affairs (VA) and Medicaid/Medicare.

Abstract

This study examines how regulations in private health insurance markets affect coverage of public insurance. We focus on mental health parity laws, which mandate private health insurance to provide equal coverage for mental and physical health services. The implementation of mental health parity laws may improve a quality dimension of private health insurance but at increased costs. We graphically develop a conceptual framework and then empirically examine whether the regulations shift individuals from private to public insurance. We exploit state-by-year variation in policy implementation in 1999–2008 and focus on a sample of veterans, who have better access to public insurance than non-veterans. Using data from the Current Population Survey, we find that the parity laws reduce employer-sponsored insurance (ESI) coverage by 2.1% points. The drop in ESI is largely offset by enrollment gains in public insurance, namely through the Veterans Affairs (VA) benefit and Medicaid/Medicare programs.

Introduction

In recent decades, a commonly used policy tool to regulate the quality of health insurance markets is the introduction of benefit mandates, which require private health insurance plans to cover certain health benefits (e.g., maternity benefits). Despite their role in enhancing access to quality health care, benefit mandate laws increase the costs of health insurance (Bailey, 2014, Bailey and Blascak., 2016),¹ and thus may decrease private insurance coverage rate. Standard economic theory suggests that individuals who optimize at a “minimum” health insurance plan may become uninsured if the “minimum” plan is prohibited by the benefit mandate laws. Nevertheless, such a “quality vs. quantity” tradeoff is suggested to be of minimal concern. Gruber (1994a) and Kaestner and Simon (2002) show that the major high-cost benefit mandates in the 1980s and 1990s had little impact on private employer-sponsored insurance (ESI) coverage. These estimates of a null effect on ESI coverage may have contributed to the popularity of mandate laws. As of 2009, more than 2100 benefits and providers are mandated through a variation of state laws (The Council for Affordable Health Insurance, 2009).

The objective of this study is to provide new evidence on the effects of benefit mandates. In addition to the effect on ESI coverage, the primary focus of prior studies, we investigate the spillover effect onto public health insurance programs. We hypothesize that an individual who does not value the improvement of insurance quality as much as the increase in costs may find ESI coverage less attractive, and switch to previously disfavored public insurance (Costa and Garcia, 2003). Such switching, if significant, would alter cost and benefit estimates, and possibly, how we evaluate mandate policies including how they are targeted or applied.

We examine the proposed spillover effect of benefit mandates among a sample of veterans. Veterans fit our research setting well because they usually have better access to publicly provided insurance compared to non-veterans. Veterans are often eligible for the Veterans Affairs (VA) health care system, which provides health care at a minimal cost. Veterans are also more likely to have work-preventing health conditions, which would qualify them for Medicaid/Medicare through the SSI/SSDI programs. These public programs provide veterans alternative avenues to health care services. Veterans may thus be more likely to opt out of ESI coverage when benefit mandates increase costs, while non-veterans lack similar flexibility.

For health insurance benefit mandates, we focus on state mental health parity laws −regulations that require equal coverage for mental health care services and physical health care services for fully insured plans.² The self-insured group health plans are exempt from state parity laws because of the Employee Retirement Income Security Act (ERISA). Before the implementation of the parity laws, mental health coverage was more limited than coverage for physical ailments. With the implementation of mental health parity laws, insurers are prohibited from discriminating between coverage for mental and physical health care. These parity laws are expected to significantly increase the costs of health insurance. Kirschstein (2000) summarizes simulation model results from multiple sources and reports that the implementation of mental health parity would increase the typical plan premiums by 3–5%, and the cost increase could be reduced by managed care. The Council for Affordable Health Insurance (2009) analyzes observational data from insurance companies, and finds that adding equal mental health benefits to an insurance plan that does not have such components would increase the cost by 5–10%.

The empirical strategy explores how state mental health parity laws affect insurance coverage (ESI and public insurance programs) of veterans. To identify causal effects, we exploit variation in the timing of enactment of different state mental health parity laws. The enactment of parity laws is linked to information on veteran’s insurance coverage, which comes from the March supplements to the Current Population Survey. We implement a generalized difference-in-differences (DD) model using a sample of male veterans of ages 22–64. We further supplement the DD model using a difference-in-difference-in-differences (DDD) model, which includes male non-veterans of ages 22–64 as a within-state comparison group.

Econometric results from both models indicate that mental health parity laws affect the health insurance coverage of veterans. Results from the base model show that state mental health parity laws reduce ESI coverage by 2.1% points. The reduction in ESI coverage is largely offset by increases in coverage under public health insurance programs. Importantly, for veterans of ages 22–54, mental health parity laws increase coverage through the VA by 1.2% points. For veterans of ages 55–64, mental health parity laws increase coverage through Medicaid/Medicare by 1.6% points. On net, the overall health insurance coverage rate does not change significantly. Subgroup analysis suggests that the effects are more pronounced among veterans without college education, veterans who are unmarried, and veterans who are employed at small firms.

One particular contribution of this study is to examine a previously understudied aspect of health insurance benefit mandates: the effects of worker selection. While economic theory suggests that workers would self-sort into employment and ESI coverage based on preferences for mandated benefits, empirical work does not separately consider the possible sorting behavior. Instead, prior studies on ESI coverage restrict the analysis sample to employed workers. One explicit assumption made by the sample restriction is that employment status is not affected by insurance mandates. This assumption, however, is not consistent with economic theory (Summers, 1989, Feldman, 1993) or empirical evidence (Gruber, 1994b, Lahey, 2012).

This analysis differs from existing studies on insurance benefit mandates by relaxing this assumption. First, we provide a simple conceptual framework to demonstrate worker selection graphically. The framework characterizes how an individual maximizes utility by choosing labor supply, monetary income, and health insurance benefits. Then, we incorporate the theoretical prediction in conducting an econometric analysis by allowing employment status to be endogenous and allowing public insurance coverage to be a possible alternate. Findings confirm that workers selecting into employment and workers choosing between ESI and public insurance are significant behavior adjustments. This study is closest to Andersen (2015), which also focuses on the determinants of worker selection into employment and ESI. The focus of Andersen (2015) is how medical conditions affect worker selection. In contract, our focus is on eligibility for public insurance.

This analysis also contributes to a broader literature on the interaction of public insurance and private insurance. Prior studies show that public insurance affects private insurance coverage. Notably, many papers provide evidence of a “crowd-out” effect – public health insurance expansions decrease private insurance coverage (Cutler and Gruber, 1996, Gruber and Simon, 2008, Brown and Finkelstein, 2008; and Hamersma and Kim, 2013) and labor supply (Boyle and Lahey, 2010, Garthwaite et al., 2014). A few studies find a “crowd-in” effect – public health insurance expansions increase private insurance coverage in particular markets. For example, Clemens (2015) shows Medicaid expansions increase private insurance coverage in community-rated markets. However, the reverse relationship – how regulations on private insurance market affect public insurance coverage – is rarely studied. The only paper on the “reverse crowd-out” that we are aware of is Dillender (2015), which shows that the dependent coverage mandate of the Affordable Care Act (ACA) reduces applications to Worker’s Compensation. Our study instead looks at the coverage through public health insurance programs.