Same-Sex Marriage and Gains in Employer-Sponsored Insurance for US Adults, 2008–2017

Abstract

Objectives. To estimate the effects of same-sex marriage recognition on health insurance coverage.

Methods. We used 2008–2017 data from the American Community Survey that represent 18 416 674 adult respondents in the United States. We estimated changes to health insurance outcomes using state–year variation in marriage equality recognition in a difference-in-differences framework.

Results. Marriage equality led to a 0.61 percentage point (P = .03) increase in employer-sponsored health insurance coverage, with similar results for men and women.

Conclusions. US adults gained employer-sponsored coverage as a result of marriage equality recognition over the study period, likely because of an increase in dependent coverage for newly recognized same-sex married partners.

Same-sex marriage had been gaining state recognition over a period of years when, in 2015, its nationwide equality to different-sex marriage was confirmed through the Supreme Court decision Obergefell v. Hodges.¹ Although not an explicit aim of same-sex marriage recognition, one benefit for those in same-sex relationships is equal access to spousal employer-sponsored health insurance (ESI) through marriage.

Marriage has long offered a means of accessing ESI coverage for members of different-sex couples. Moreover, it is unsurprising that prior to marriage equality, ESI coverage was 6 percentage points lower for both women and men in same-sex relationships than for those in heterosexual relationships.² Another pre-equality study confirmed these findings, showing that rates of dependent ESI coverage were 15 percentage points lower among lesbian partnered women and 8 percentage points lower among gay partnered men compared with their heterosexual partnered peers.³ Marriage equality has the potential to reduce ESI disparities and possibly decrease the inequities in uninsurance observed in several studies of the period predating marriage equality.^2–5 Findings from early in the nationwide postequality period show that the recognition of same-sex marriage is associated with having health insurance of any type among gay men (although not among lesbian women).⁶

Insurance coverage helps facilitate access to health services, which is important for sexual minorities, a population characterized by health disadvantages such as a higher risk of early mortality.⁷ Sexual minorities are at elevated risk of delaying care because of costs,⁸ and improved access to health insurance has the potential to play a role in reducing these delays. In light of the current policy discussion on universal health care coverage and the contemporary nature of marriage equality, it is critical to understand what the legal recognition of same-sex marriage means in terms of numbers of individuals newly covered by ESI. We fill this need by estimating the effects of same-sex marriage recognition on ESI and other health insurance outcomes.

METHODS

This study used 2008–2017 public-use data from the American Community Survey (ACS).⁹ The ACS is an ongoing annual household survey of approximately 3.5 million housing units collected by the US Census Bureau. We created data on the timing of states’ recognition of same-sex marriage from publicly available online sources.¹⁰ (For dates, see Appendix, available as a supplement to the online version of this article at http://www.ajph.org). The study sample included all nonelderly respondents aged 18 to 64 years, irrespective of their sexual orientation or relationship status.

We estimated the effect of marriage equality on ESI as our primary analysis. We also studied the effect on any type of private coverage (including ESI, direct purchase plans, and TRICARE or other military health care), public coverage (Medicare, Medicaid, Department of Veterans Affairs insurance, as well as state or local public insurance programs), and uninsurance (no health insurance or only Indian Health Services coverage).

We used a difference-in-differences approach that leveraged state–time variation in recognition of marriage equality to estimate changes to health insurance. The model accounted for gender, race, ethnicity, education, age, age squared, family size, and income, and included an indicator for young adults aged 26 years or older in 2010 or after, since this group gained eligibility for private dependent coverage in 2010. We included state-by-year unemployment and state and year fixed effects. Standard errors are heteroscedasticity-robust and clustered by state (see online Appendix for model details).

We expected that our intent-to-treat approach of including all nonelderly, nondisabled adults would produce small effect sizes since the affected individuals were a minority population. Our approach allowed us to avoid using potentially weak proxies of sexual orientation. We adjusted models by using analytic weights for the full US population. The analyses were linear probability models. Coefficients could be interpreted as percentage point changes, or multiplied by population sizes to estimate numbers of people affected. We conducted all analyses with Stata 14 MP statistical software (StataCorp LP, College Station, TX).

RESULTS

Marriage equality produced a 0.61 percentage point (P = .03) increase in ESI, which accounted for most of the 0.74 percentage point (P = .07) increase in having any private insurance (Figure 1). This translates to an estimated 11.6 million person-years of ESI coverage over the period 2008 to 2017. Despite these gains in ESI coverage, there was no measurable decline in uninsurance and no change in public insurance (see online Appendix for detailed results). Results were similar for men and women (not shown).

FIGURE 1—

Estimated Changes to Person-Years (in Millions) of Insurance Coverage After Marriage Equality: United States, 2008–2017

Note. ESI = employer-sponsored health insurance. Shaded bars represent estimates statistically significant at P < .05. Estimates shown were calculated by multiplying the regression-estimated percentage point changes by the frequency weighted number of observations over the study period. Data are from the American Community Survey from 2008 to 2017. For more detail, see online Appendix (available as a supplement to the online version of this article at http://www.ajph.org).

DISCUSSION

Millions of US adults gained employer-sponsored coverage after marriage equality recognition over the period 2008 to 2017, likely because of an increase in dependent coverage for same-sex partners following the policy change. Our results are consistent with prior work that found that people in same-sex relationships were less likely to have ESI when they lived in a state that did not recognize same-sex marriage.^2,3

Contrary to previous research that found gains in health insurance only among men,⁶ we found similar results for men and women. This is in part because of differences in data, study population, and approach: our analysis included more years of postequality data, and we did not use proxies for sexual orientation to estimate a treatment-on-the-treated effect. The lack of change to uninsurance in our study, despite gains in ESI, is not surprising. Those with new ESI may have switched from nongroup private plans or from public health insurance, rather than exclusively from being uninsured.

This study has limitations. Although we suspect that the changes in insurance coverage occurred among sexual minority couples who married or had existing unions recognized, we were unable to identify the specific mechanisms by which the change occurred or the population most likely affected by the policy (i.e., sexual minorities). Restricting our analysis to people in same-sex relationships would require the assumption that marriage equality did not influence respondents’ willingness to identify as part of a same-sex couple. In addition, our study included states that did not legalize same-sex marriage until 2015, resulting in some short postequality periods. If any effects on health insurance grew or changed over time, we were limited in our ability to capture this in the analysis. Future work should distinguish the plausible short- and long-run impact of marriage equality on health insurance.

We suggest that future research on same-sex marriage should go beyond the net effects to explore context and dynamics, as it is likely that other status characteristics such as geographic location, age, income, or family structure can intersect with sexual orientation in ways that are meaningful for health insurance outcomes.¹¹

PUBLIC HEALTH IMPLICATIONS

Marriage equality played a critical role in expanding access to employer-sponsored insurance. This has far-reaching health benefits for those newly insured, and may offer out-of-pocket savings or access to more comprehensive benefits for those switching from a nongroup or public plan.