SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. LITERATURE REVIEW AND INSTITUTIONAL DETAILS
  5. DATA AND EMPIRICAL APPROACH
  6. RESULTS
  7. CONCLUSION
  8. ACKNOWLEDGMENTS
  9. REFERENCES
  10. Biographies

Health disparities related to sexual orientation are well documented and may be due to unequal access to a partner's employer-sponsored insurance (ESI). We provide the literature's first evaluation of legislation enacted by California in 2005 that required private employers within the state to treat employees in committed same-sex relationships in the same way as employees in different-sex marriages with respect to ESI. Our analysis uses data on sexual orientation, partnership, and health insurance from the 2001 to 2007 California Health Interview Surveys (CHIS). Prior to the reform, partnered gay men and lesbians were significantly less likely to have ESI in someone else's name than partnered heterosexuals. Pooling data from 2001 to 2007, we find that the reform had no effects on differences in insurance outcomes between gay and straight men. We find some evidence that the reform increased partnership, reduced full-time employment, and increased health insurance coverage among lesbians relative to heterosexual women. The increases in insurance coverage for lesbians are consistent with a role for expanded dependent ESI, suggesting that such policies may reduce sexual orientation-based insurance disparities among women.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. LITERATURE REVIEW AND INSTITUTIONAL DETAILS
  5. DATA AND EMPIRICAL APPROACH
  6. RESULTS
  7. CONCLUSION
  8. ACKNOWLEDGMENTS
  9. REFERENCES
  10. Biographies

Disparities in access to health care as well as in health outcomes related to sexual orientation are well documented (see e.g., Cochran & Mays, 2000; Institute of Medicine, 2011; Solarz, 1999). Whereas the disproportionate impact of the HIV/AIDS epidemic on the gay male community is well known, public health research has also shown that sexual minorities are at increased risk for cancer, body weight problems, depression and other mental health disorders, substance use, and smoking (see e.g., Carpenter, 2003; Cochran & Mays, 2000; Stall & Wiley, 1988; Daling et al., 1987; Stall et al., 1999). Other work finds that lesbians are significantly less likely than other women to get routine preventive health care such as pap smears and breast cancer screenings and that gay adults are significantly more likely than heterosexuals to report unmet medical needs and difficulty obtaining care (see e.g., Denenberg, 1995; Diamant, Schuster, & Lever, 2000).1 However, the causes of these disparities are not well understood.

A low rate of insurance coverage among gay adults2 is commonly cited as a primary reason for their reduced access, and indeed several recent studies document lower rates of insurance coverage for gay men and lesbians relative to heterosexual men and women (Ash & Badgett, 2006; Buchmueller & Carpenter 2010; Harris Interactive/Witeck-Combs, 2002; Diamant et al., 2000; Cochran et al., 2001; Heck, Sell & Gorin, 2006). One reason that sexual minorities might have lower rates of coverage is that employers have historically treated same-sex partners of employees differently than heterosexual spouses with respect to fringe benefits such as health insurance. Although same-sex domestic partner benefits have increased over time, many employers do not offer them, and in most states, employers are not legally required to do so.

Government intervention with respect to health insurance benefits for same-sex partners has taken a variety of forms. Many state governments have extended a variety of fringe benefits—including health insurance—to the same-sex partners of government employees for several decades.3 More recently, the issue of same-sex partner benefits has been on the policy agenda in numerous states that have addressed this issue by legalizing gay marriage, civil unions, or some other domestic partner registration status within the state (e.g., New Jersey, Vermont, Massachusetts, California). There has been no evidence, however, on whether and to what extent these laws have affected the disparity in health insurance between sexual minorities and heterosexual individuals.

In this paper, we address this gap by examining differences in health insurance coverage related to sexual orientation in California using data from the California Health Interview Survey (CHIS) for the years 2001, 2003, 2005, and 2007. These data provide direct measures of sexual orientation and partnership for large, representative samples as well as detailed information on sources of health insurance coverage. From a policy perspective, a key feature of our CHIS sample is that it provides data before and after the implementation of legislation in California that effectively required private employers within the state to extend health benefit eligibility to their employees in committed same-sex relationships in the same way as employees in different-sex marriages.

California's reform consisted of two companion pieces of legislation. The first, California's Domestic Partner Rights and Responsibilities Act—commonly known as AB205—which was passed in 2003 and took effect on January 1, 2005, amended the state family code to give same-sex domestic partners many of the same rights and responsibilities already afforded heterosexual married couples, including adoption rights, hospital visitation privileges, and equal state income tax treatment.4 AB205 was widely interpreted as requiring private employers to treat domestic partners of employees the same as spouses. The second bill, AB2208 was adopted in 2004 and took effect on January 2, 2005 (one day after AB205). AB2208 amended the state insurance code to make it consistent with the provisions of AB205. Both laws were motivated by concerns about equality in treatment of gay and lesbian couples compared to heterosexual couples; this is explicit in the language of both bills. For employers, the reforms mean that any fringe benefits such as health insurance commonly extended to heterosexual spouses of employees must also be extended to same-sex partners of employees. In so doing, this policy had the potential to increase health insurance coverage among partnered gay men and lesbians. Ours is the first study to directly evaluate the effects of these increasingly common policies.5

Of course, this description raises several important considerations that may complicate empirical evaluations of California's reform. First, because AB205 changed many other benefits besides just access to health insurance for partnered compared to nonpartnered gay men and lesbians, it is plausible that partnership itself changed as a result of the law. A key feature of our analysis is that we test whether partnership rates changed appreciably for gays and lesbians relative to heterosexuals coincident with AB205 implementation. Second, the reform may have changed the sources of health insurance coverage without changing overall rates of health insurance coverage by sexual orientation. For example, a partnered gay person who had previously purchased individual insurance may drop that coverage in favor of enrolling as a dependent on his or her partner's employer-sponsored plan. Or, some people who were working full-time to gain access to employer-sponsored insurance (ESI) may reduce their hours in response to being able to access a partner's health insurance. In such cases, we would not observe a change in the overall likelihood of having health insurance coverage, despite the fact that important behavioral changes occurred. We address these possibilities by examining the effects of the reform on the probability an individual has any insurance coverage and the probability an individual works full-time, in addition to examining effects on the main outcomes of interest: the probability of having health insurance from various sources, especially ESI in someone else's name.

LITERATURE REVIEW AND INSTITUTIONAL DETAILS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. LITERATURE REVIEW AND INSTITUTIONAL DETAILS
  5. DATA AND EMPIRICAL APPROACH
  6. RESULTS
  7. CONCLUSION
  8. ACKNOWLEDGMENTS
  9. REFERENCES
  10. Biographies

Research on health insurance gaps by sexual orientation face formidable data challenges because most representative survey data sets do not include direct measures of sexual orientation at the individual level.6 To overcome this limitation, researchers have analyzed large data sets that allow identification of minority sexual orientation status indirectly through information on household sex composition and intrahousehold relationships (since two men, or women, living together in a married-like relationship are very likely to be gay, or lesbian; Carpenter, 2004). These studies find that individuals in same-sex relationships are significantly less likely to have health insurance than individuals in different-sex relationships (Ash & Badgett, 2006; Buchmueller & Carpenter 2010; Heck, Sell, & Gorin, 2006).

Several patterns from the literature on sexual orientation and health insurance emerge that are relevant for considering the potential effects of policies such as AB205 and AB2208. First, the evidence base for a disparity in insurance coverage is stronger for lesbians than it is for gay men. For example, studies that examine both men and women consistently find that the gap between lesbians and straight women is larger than the associated gap between gay men and straight men (Buchmueller & Carpenter, 2010; Heck, Sell, & Gorin, 2006). Second, partnership rates among lesbians are close to the partnership rates among heterosexual women, whereas gay men are much less likely to be partnered than straight men (Carpenter & Gates, 2008). Finally, previous research on gay marriage or domestic partner registries in other settings indicates that lesbians are more likely to take advantage of these registration benefits than are gay men (Badgett, Gates & Maisel, 2008; Gates, Badgett, & Ho, 2008). These factors all suggest that policies such as those we study here have greater latitude for affecting sexual orientation-based disparities in health-insurance-related outcomes for females than males.

Partly due to data limitations, we know very little about the sources of health insurance coverage for gay men and lesbians, particularly when compared to the large body of research on sources of insurance for heterosexuals. A Harris Interactive (2002) report based on an online survey indicates that while only 3 percent of sexual minority individuals have health insurance through a partner, 14 percent of heterosexual adults report dependent ESI. Ponce et al. (2010) find that gay men and lesbians are 42 percent and 28 percent as likely as straight men and women, respectively, to have dependent ESI. Dependent coverage is an important source of health insurance for heterosexual married women and to a lesser extent married men (Buchmueller, 1996/1997; Farber & Levy, 2000). Several studies suggest that access to ESI coverage through a spouse has important effects on behavior. Married workers, especially women, are significantly more likely to decline coverage from their own employer when their spouse is offered ESI (Buchmueller, 1996/1997; Zimmer, 2009). Other studies find that access to spousal health insurance coverage is negatively related to labor supply for married women (Abraham & Royalty, 2006; Buchmueller & Valletta, 1999; Kapinos, 2009; Olson, 1998; Wellington & Cobb-Clark, 2000).

The effect of laws such as AB205 and AB2208 depend in part on how common it is for employers to allow gay or lesbian employees to cover their partners as dependents on their health insurance in the absence of legislation. We know very little about the prevalence of such policies, though available evidence indicates that the practice of extending same-sex domestic partner health benefits by private employers is far from universal. The 2011 National Compensation Survey (NCS) indicated that just 29 percent of all workers in private industry had access to health benefits for same-sex partners. Firm characteristics—especially firm size—are strongly correlated with the likelihood of offering same-sex domestic partner health benefits. For example, although a slight majority (57 percent) of Fortune 500 firms now offer health insurance benefits to same-sex partners of their sexual minority employees (Human Rights Campaign, 2009), a 2009 employer survey sponsored by the Kaiser Family Foundation indicated that just 20 percent of firms with fewer than 200 employees offer these benefits to same-sex domestic partners of their employees (Kaiser Family Foundation, 2009).7

In summary, there is a fair amount of evidence that gays and lesbians are less likely to have health insurance than heterosexuals, and some evidence suggests that a source of this disparity is reduced access to a partner's ESI. Same-sex domestic partner health insurance benefits were far from universal in the period prior to California's reform, suggesting a potential role for the law to increase access to a partner's ESI. In addition to changes in insurance, previous research on dependent ESI among heterosexuals suggests that labor supply effects are also possible. Finally, several considerations, including the much higher prevalence of partnership among lesbians, suggest that any effects are likely to be larger for lesbians than for gay men.

DATA AND EMPIRICAL APPROACH

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. LITERATURE REVIEW AND INSTITUTIONAL DETAILS
  5. DATA AND EMPIRICAL APPROACH
  6. RESULTS
  7. CONCLUSION
  8. ACKNOWLEDGMENTS
  9. REFERENCES
  10. Biographies

Our data for this study are the 2001, 2003, 2005, and 2007 waves of the CHIS. These surveys are administered by telephone to over 40,000 households using random digit dialing (RDD) methods in each year. The CHIS data are unique in that they provide information on self-reported sexual orientation for all adults. For privacy reasons, the question on sexual orientation is not available in the public use CHIS data file, though we have obtained access to these data through a confidential data agreement with the UCLA Data Access Center at the Center for Health Policy Research.

Individuals in the 2001 CHIS were asked “Are you gay, (lesbian) or bisexual?” Since 2003, the CHIS has asked adults “Do you think of yourself as straight–heterosexual, gay (lesbian), or bisexual?” If the individual did not immediately volunteer an answer, the interviewer was directed to read the following statement: “Straight or heterosexual people are attracted to, or have sex with, people of the opposite sex; gays/lesbians are attracted to, or have sex with, people of the same sex; and bisexuals are attracted to, or have sex with, people of both sexes.” We drop a very small share of individuals who report they do not know their sexual orientation or who refused a response to the sexual orientation question. Throughout this analysis, we also drop individuals who report being bisexual. Although these groups are independently interesting, bisexual is not a well-defined concept when studying policies related to same-sex partner benefits since we do not observe the sex of any respondent's partner. This is because CHIS does not include a complete household sex roster.

The CHIS includes standard sociodemographic variables, including age, race and ethnicity, education, rural and urban residency, and employment. We identify the partnership status of each CHIS respondent using responses to a question about marital status. Specifically, individuals are asked, “Are you now married, living with a partner in a marriage-like relationship, widowed, divorced, separated, or never married?” We define an individual as “partnered” if she reports being either married or living with a partner in a marriage-like relationship.8 Unfortunately, because the CHIS is an individual-level (not a household-level) survey, we do not observe detailed information on the employment or insurance coverage of partners or other household members for all adults in the sample.9

We restrict attention to 25- to 64-year olds to focus on a period after most people have completed their education. Table 1 presents the means of these variables by gender and sexual orientation. The patterns in these variables are similar to those that have been reported previously using earlier waves of these data (Carpenter, 2005; Carpenter & Gates, 2008) and are also similar to patterns from other large, representative data sets (Black et al., 2000; Black, Sanders, & Taylor, 2007).

Table 1. Descriptive characteristics, California Health Interview Survey (CHIS) 2001 to 2007
 (1) Gay Males(2) Straight Males(3) Lesbians(4) Straight Females
Notes
  1. Author calculations, 2001 to 2007 CHIS, adults age 25 to 64, weighted means (standard errors in parentheses).

Age41.2 (0.034)42.9 (0.075)42.9 (0.422)43.5 (0.064)
White0.525 (0.019)0.451 (0.003)0.523 (0.024)0.454 (0.003)
Black0.051 (0.008)0.050 (0.001)0.046 (0.010)0.060 (0.001)
Latino0.082 (0.011)0.121 (0.002)0.072 (0.012)0.106 (0.002)
Asian/Pacific Islander0.057 (0.009)0.085 (0.002)0.022 (0.007)0.089 (0.002)
Less than high school degree0.030 (0.007)0.083 (0.002)0.031 (0.007)0.071 (0.002)
High school degree0.151 (0.016)0.230 (0.003)0.137 (0.015)0.222 (0.002)
Some college0.264 (0.017)0.260 (0.003)0.263 (0.021)0.298 (0.003)
College degree or more0.555 (0.019)0.426 (0.003)0.570 (0.023)0.408 (0.003)
Partnered (married or living with a partner)0.405 (0.018)0.726 (0.003)0.584 (0.024)0.695 (0.003)
Any health insurance0.850 (0.016)0.867 (0.003)0.887 (0.016)0.900 (0.002)
Works full time0.691 (0.017)0.778 (0.003)0.690 (0.021)0.522 (0.003)
N1,78347,7331,14067,460

The CHIS data also provide detail on both the type (e.g., public, ESI, private nongroup) and source (e.g., own name, as a dependent) of health insurance coverage for each respondent. We consider several outcomes related to insurance coverage and its sources. Individuals in the CHIS are asked a series of detailed questions regarding whether they are covered by a variety of specific sources, such as Medicare, Medicaid (MediCal in California), employer plans, or other individually purchased plans. We use the CHIS-recoded responses to create several indicator variables, including insurance from any source, Medicaid, own-name ESI, dependent ESI, and individually purchased insurance. We do not separately examine the very small share of individuals who report having CHIP (the public insurance program targeted to children), Medicare, or “other public” insurance, though we do not exclude them from the analysis.

We begin by examining descriptive evidence on differences in insurance coverage (overall and by source), paying particular attention to the period prior to California's reform (i.e., the 2001 and 2003 CHIS waves). This is an independently interesting and valuable exercise, because in many ways our large samples of population-based data with detailed information on source of insurance coverage improve on previous research that has used much smaller and geographically isolated samples, convenience samples, or couples-based samples and that has focused mainly on overall differences in insurance coverage.

We then use the full-pooled 2001–2007 CHIS sample to test whether California's reform differentially affected outcomes for gay men and lesbians relative to heterosexual individuals by examining how outcomes changed for these groups relative to heterosexuals after the enactment of the reforms in 2005. This difference-in-differences (DD) model takes the following form:

  • display math(1)

where GAY/LESBIAN is an indicator variable equal to 1 for individuals reporting a gay or lesbian sexual orientation, and X is a vector of demographic variables that includes age and its square, four education dummies (less than high school, some college, bachelor's degree, and master's and Ph.D. degree, with high school degree as the excluded category), four race and ethnicity dummies (black, Hispanic, Asian and Pacific Islander, and other multiple race, with white as the excluded category), a dummy for being partnered, and five dummies for urban location (second city, suburban, small town, rural, and urban status not ascertained, with urban location as the excluded category). POST REFORM is an indicator variable equal to 1 for observations after the implementation date of the AB205 and AB2208 laws (January 1 and 2, 2005). The error term ε in (1) is assumed to be iid, and we estimate models separately for males and females. The coefficient of interest is β4 and represents the relative effect of the reform on outcomes for gays and lesbians compared to heterosexuals. The key identifying assumption in this simple DD model is that there were no other shocks to relative outcomes over this period for gays and lesbians relative to heterosexuals.

The distribution of interview dates in the CHIS is far from uniform in a way that is fortuitous for our research design. AB205 and AB2208 were passed in 2003 and 2004, respectively, and took effect in January 2005. Since insurance policies are generally revised in the fall to take effect the following calendar year, we expect that individuals would have seen the new benefits beginning in calendar year 2005. The 2005 CHIS interviews did not begin until July 2005 and actually continued into March and April of 2006. This means that there should have been ample time for individuals to have responded to the new benefits in the 2005 data. In other words, the 2005 CHIS wave is legitimately after the reform. All 2007 CHIS interviews were completed from June 2007 to March 2008, just before the California Supreme Court unexpectedly legalized same-sex marriage.

Finally, we note that an alternative strategy for evaluating California's reform would be to use partnership status as an additional treatment status in a triple-differences framework since AB205 and AB2208 changed access to ESI for partnered gays and lesbians only. A limitation of this approach is that if partnership itself was affected by the reform, a triple-difference model is inappropriate because of composition changes in the treatment and control groups coincident with the policy of interest. Indeed, we show below that there is some evidence that partnership increased among lesbians relative to heterosexual women coincident with AB205, which is plausible given that the policy increased many benefits to same-sex partnership. Given these changes in partnership, we do not estimate triple difference models.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. LITERATURE REVIEW AND INSTITUTIONAL DETAILS
  5. DATA AND EMPIRICAL APPROACH
  6. RESULTS
  7. CONCLUSION
  8. ACKNOWLEDGMENTS
  9. REFERENCES
  10. Biographies

Sexual Orientation and Insurance Coverage Prior to the Reform

To provide a sense of the potential impact of California's reform, we begin the analysis by documenting differences in health insurance coverage prior to the law. Table 2 presents means for the various insurance outcomes for men (top panel) and women (bottom panel); the first four columns present descriptive statistics for the 2001 and 2003 CHIS data prior to the reform. For completeness, we also present in the last four columns the associated descriptive statistics for the 2005 and 2007 CHIS data. We present means separately for nonpartnered gays and lesbians (columns 1 and 5), nonpartnered heterosexuals (columns 2 and 6), partnered gays and lesbians (columns 3 and 7), and partnered heterosexuals (columns 4 and 8) (recall that the ESI provision of the reform pertains only to partnered gays and lesbians). The patterns in Table 2 show that although there are not large differences in the percentage of gay and straight men and women who have any health insurance (see bottom of Table 1), there are interesting differences by type of coverage and partnership status. For men in the top panel of Table 2, we find that prior to the reform, gay men are slightly more likely to have Medicaid and individually purchased insurance, regardless of partnership status. The higher rate of individually purchased insurance among gay men compared to straight men (regardless of partnership status) is interesting in light of common perception and past research suggesting discriminatory practices related to a minority sexual orientation in the individual insurance market (Zellers, McLaughlin, & Frick, 1992). Nonpartnered gay men are also more likely to have own-name ESI compared with nonpartnered straight men. In contrast, in the partnered subsample straight men are more likely to have ESI, either in their own name or as a dependent, and this more than offsets their lower rate of Medicaid and individually purchased insurance compared to partnered gay men.

Table 2. Health insurance outcomes by sexual orientation and partnership status before and after California's reform, CHIS 2001 to 2007
 (1) Non-Partnered Gay/Lesbian(2) Non-Partnered Straight(3) Partnered Gay/Lesbian(4) Partnered Straight(5) Non-Partnered Gay/Lesbian(6) Non-Partnered Straight(7) Partnered Gay/Lesbian(8) Partnered Straight
 Before Reform (2001 and 2003 CHIS)After Reform (2005 and 2007 CHIS)
Notes
  1. Author calculations, 2001 to 2003 CHIS, adults age 25 to 64, weighted means (standard errors in parentheses).

Men        
 Has any insurance0.857 (0.021)0.759 (0.007)0.864 (0.026)0.906 (0.003)0.814 (0.040)0.743 (0.010)0.881 (0.031)0.915 (0.004)
 Has individually purchased insurance0.108 (0.016)0.065 (0.003)0.081 (0.018)0.059 (0.002)0.110 (0.027)0.071 (0.005)0.099 (0.022)0.057 (0.003)
 Has Medicaid0.108 (0.017)0.102 (0.005)0.051 (0.016)0.047 (0.002)0.114 (0.020)0.110 (0.007)0.041 (0.014)0.042 (0.003)
 Has employer-sponsored insurance (ESI) in own name0.619 (0.026)0.543 (0.008)0.611 (0.036)0.646 (0.005)0.552 (0.041)0.491 (0.011)0.610 (0.042)0.649 (0.006)
 Has ESI in someone else's name0.012 (0.007)0.018 (0.002)0.093 (0.021)0.150 (0.004)0.012 (0.006)0.021 (0.004)0.064 (0.016)0.151 (0.004)
 Works full time0.672 (0.026)0.670 (0.008)0.721 (0.034)0.817 (0.004)0.661 (0.037)0.646 (0.011)0.732 (0.035)0.830 (0.004)
 Works part time0.109 (0.018)0.085 (0.005)0.082 (0.021)0.048 (0.002)0.071 (0.013)0.099 (0.007)0.115 (0.028)0.052 (0.003)
 N6448,37229216,4285627,06928515,512
Women        
 Has any insurance0.799 (0.034)0.855 (0.005)0.890 (0.027)0.922 (0.003)0.855 (0.050)0.839 (0.007)0.955 (0.017)0.926 (0.003)
 Has individually purchased insurance0.073 (0.023)0.078 (0.003)0.079 (0.018)0.079 (0.002)0.075 (0.022)0.070 (0.004)0.052 (0.014)0.071 (0.002)
 Has Medicaid0.069 (0.019)0.175 (0.005)0.040 (0.015)0.057 (0.002)0.097 (0.024)0.178 (0.006)0.083 (0.025)0.051 (0.003)
 Has ESI in own name0.599 (0.041)0.571 (0.006)0.655 (0.036)0.381 (0.004)0.610 (0.061)0.546 (0.008)0.676 (0.036)0.401 (0.005)
 Has ESI in someone else's name0.016 (0.007)0.027 (0.002)0.110 (0.023)0.395 (0.004)0.034 (0.020)0.031 (0.003)0.104 (0.021)0.384 (0.005)
 Works full time0.710 (0.036)0.604 (0.006)0.728 (0.033)0.470 (0.004)0.637 (0.059)0.606 (0.008)0.687 (0.036)0.500 (0.005)
 Works part time0.068 (0.016)0.112 (0.004)0.079 (0.019)0.146 (0.003)0.130 (0.048)0.107 (0.005)0.126 (0.025)0.147 (0.003)
N29213,49328021,01225112,24231720,713

The means for women by sexual orientation and partnership status in the pre-reform period are reported in the first four columns of the bottom panel of Table 2. In the years just prior to reform, lesbians were less likely to have any health insurance than heterosexual women. As with men, the pattern with respect to sources of coverage varies with partnership status. Among nonpartnered women, differences in ESI coverage (either own-name or in someone else's name) are very small. For this subsample, the difference in overall coverage rates is driven by a much higher rate of Medicaid coverage for heterosexual women: Fully 17.5 percent of nonpartnered heterosexual women have Medicaid, compared to only 6.9 percent of nonpartnered lesbians. In contrast, there are large differences in ESI coverage by sexual orientation in the partnered subsample. Partnered lesbians are 27.4 percentage points more likely than partnered heterosexual women to have own-name ESI. This difference in own-name ESI coverage, however, is more than offset by the fact that partnered heterosexual women are much more likely to have ESI coverage as a dependent than partnered lesbians (39.5 percent vs. 11 percent).

Overall, the results in Table 2 suggest a greater potential effect of California's reform for lesbians than for gay men. Among men, differences in any health insurance coverage were modest, and it appears that gay men are able to compensate for low levels of ESI coverage with higher rates of individually purchased insurance and Medicaid. Moreover, partnership prevalence among gay men was low (about 40 percent), meaning that the potential role for AB205 to affect relative outcomes for men is limited. Among women, we find much larger differences in insurance coverage associated with sexual orientation, and for partnered women the gap is largely due to a much lower likelihood for lesbians to have dependent ESI. This large difference in the source of ESI is potentially a target of the reform if partnered women were constrained due to lack of legal rights to access a same-sex partner's ESI. Moreover, the high rates of partnership among lesbians (nearing 60 percent) also increase the latitude for the reform to have meaningful effects on outcomes.

Estimating the Effect of California's Reform

We now turn to a direct evaluation of AB205 and AB2208. As described in the previous section, we estimate DD models (i.e., equation (1)) that compare changes in outcomes for gay men and lesbians (the treatment groups) before and after the reform to the associated changes in outcomes for heterosexual men and women (the control groups). Before turning to health-insurance-related outcomes, however, we first examine in Table 3 a range of other outcomes that may speak to the appropriateness of our research design and to other behavioral changes that may have been due to the policy. The format of Table 3 is as follows: The top panel presents results for males, and the bottom panel presents results for females. Each column within each panel is from a separate regression that includes the standard demographic controls, and we report the relevant coefficients on the GAY/LESBIAN indicator, the POST REFORM indicator, and the interaction of the two. The coefficient of primary interest in the DD models is the one on the interaction term.

Table 3. Changes in partnership and employment surrounding California's reform, 2001 to 2007 CHIS
Outcome is [RIGHTWARDS ARROW](1) Partnered(2) Employed at All(3) Employed Full-Time
Notes
  1. Statistical robust standard errors below in parentheses. Each column within each panel is a separate regression. Models also include controls for age and its square, race, education, and urban location.

  2. *Significant at 10 percent; **significant at 5 percent; ***significant at 1 percent.

Males   
 Gay−0.302***−0.067***−0.116***
 (0.022)(0.018)(0.021)
 Post reform0.007−0.0000.008
 (0.006)(0.005)(0.006)
 Gay × post reform.016−0.0350.004
 (0.035)(0.030)(0.032)
 R-squared0.070.080.09
 N49,63149,63149,631
Females   
 Lesbian−0.153***0.100**0.169***
 (0.027)(0.020)(0.024)
 Post reform0.003−0.0030.025***
 (0.005)(0.005)(0.006)
 Lesbian × post reform0.076*−0.032−0.071*
 (0.045)(0.034)(0.039)
 R-squared0.050.050.04
 N68,79768,79768,797

We begin Table 3 in column 1 with an examination of partnership probabilities. Recall that AB205 changed several aspects of the benefits and costs of partnership for gays and lesbians, not just potential access to a partner's ESI. These other changes included tax liability, parenting rights, and hospital visitation rights, among others. It is plausible, then, that partnership itself changed in response to AB205. Since one research design to test for differences in health outcomes surrounding AB205 and AB2208 would be to use partnership as a treatment and control group margin (i.e., to compare partnered gay men with nonpartnered gay men vs. partnered straight men with nonpartnered straight men before vs. after the law change), it is necessary to examine changes in partnership itself. If we found that the reforms changed partnership substantially, this would raise concerns about composition biases in triple differences estimates of the effects of the laws that used partnership as a treatment margin.

Indeed, we find some evidence that California's reforms increased partnership among lesbians. The bottom panel of column 1 indicates that in 2001–2003, lesbians were about 15 percentage points less likely to be in a partnership. The coefficient on the POST REFORM dummy indicates that for heterosexual women, partnership remained stable between this early period and 2005–2007. The coefficient on the interaction term, however, suggests that lesbians were 7.6 percentage points more likely to be in a partnership following the reform compared to the associated change for heterosexual women. This estimate is statistically significant at the 10 percent level; relative to the preexisting partnership figure for lesbians, it represents an increase of about 14 percent. The interaction term coefficient for gay men is also positive but smaller in magnitude (suggesting a 1.6 percentage point increase) and is statistically insignificant. However, because the relevant partnership figure for gay men was only around 40 percent, this point estimate does represent a nontrivial increase in partnership. These results suggest that it could be problematic to use partnership as a treatment margin in a triple differences framework to evaluate the effects of California's reform on outcomes, particularly for women.

In the next two columns of Table 3, we perform the related exercise for the probability an individual is employed at all (column 2) and the probability an individual is employed full-time (column 3). While we find small and statistically insignificant estimates for the coefficients of interest for men, we do find in the bottom panel of column 3 for women that lesbians were 7.1 percentage points less likely to be working full-time following the reform compared to the associated change for heterosexual women. This reduction in full-time employment is significant at the 10 percent level and could represent the alleviation of suboptimal labor force decisions associated with the new ability to access a partner's ESI.

In Table 4, we turn to a direct examination of the effects of California's reform on the sources of insurance coverage using a DD framework that compares changes in outcomes for gays and lesbians surrounding the law to the associated changes in outcomes for straight men and women. The format of Table 4 follows that of Table 3, except the outcomes differ in each column. In column 1, we show results for the probability an individual has any insurance, in column 2 for individually purchased insurance, in column 3 for Medicaid, in column 4 for own-name ESI, and in column 5 for dependent ESI. Again, the coefficients of interest are those on interaction terms of the GAY/LESBIAN indicator with the POST REFORM indicator. All models include the same demographic controls.

Table 4. California's reform and changes in health insurance and its sources, 2001–2007 CHIS
 (1) Any Insurance(2) Individually Purchased Insurance(3) Medicaid(4) ESI in Own Name(5) ESI in Someone Else's Name
Notes
  1. Robust standard errors below in parentheses. Each column within each panel is a separate regression. Models also include controls for age and its square, race, education, and urban location.

  2. *Significant at 10%; **significant at 5%; ***significant at 1%.

Males     
 Gay−0.0180.032***0.043***−0.029−0.068***
 (0.016)(0.012)(0.012)(0.021)(0.010)
 Post reform−0.006−0.006*0.004−0.013−0.001
 (0.005)(0.003)(0.003)(0.007)(0.004)
 Gay × post reform−0.0120.011−0.009−0.018−0.012
 (0.029)(0.022)(0.018)(0.035)(0.013)
 R-Squared0.060.010.050.040.01
 N49,63149,63149,63149,63149,631
Females     
 Lesbian−0.072***−0.008−0.0060.150***−0.223***
 (0.021)(0.014)(0.013)(0.028)(0.014)
 Post reform−0.010**−0.016***0.011***0.001−0.015***
 (0.004)(0.003)(0.003)(0.006)(0.005)
 Lesbian × post reform0.076**−0.0030.0240.0160.030
 (0.031)(0.018)(0.021)(0.042)(0.021)
 R-squared0.030.010.100.050.03
 N68,79768,79768,79768,79768797

The results in Table 4 suggest that the reform had essentially no meaningful effect on health insurance outcomes overall or by source for gay men relative to straight men. All of the interaction terms are small and statistically insignificant. For females in the bottom panel of Table 4, in contrast, we find evidence that the laws helped to close the gap in health insurance coverage. We estimate that lesbians were 7.6 percentage points more likely to have insurance following the reform compared to the associated change for heterosexual women. This estimate is statistically significant at the 5 percent level. The coefficient on the LESBIAN times POST REFORM variable for ESI in someone else's name is positive and sizable in magnitude, though not statistically significant. The evidence broadly suggests that the reform increased health insurance coverage for lesbians relative to straight women and is consistent with the possibility that one of the mechanisms was an increase in access to a partner's ESI.

The results for women presented in Tables 3 and 4 suggest that the reform allowed some lesbians who were working full-time to qualify for health benefits to reduce their hours while taking up coverage through their partner's employer. In other models (not reported here but available on request), we further explored these results for women in two key ways. First, we examined outcomes that jointly examined health insurance coverage and employment. If California's reform induced some women to reduce work effort due to increased access to a partner's ESI, we should expect to see a relative reduction in the probability a lesbian works full-time and has ESI in her own name and relative increases in the probability a lesbian works part-time (or not at all) and has ESI in someone else's name.10 Second, we examined whether there were important differences in the effects for women with children compared to women without children, since the preference for shorter hours should be strongest among women with children (and thus any effects of the law should be concentrated in this group).11 Prior research on married women suggests that access to dependent insurance coverage does not represent as much of a constraint for women without children, presumably because they do not have the same preference for part-time hours (Buchmueller & Valletta, 1999).

Indeed, we found that the probability a woman reported working full-time and having ESI in her own name fell significantly for lesbians after the reform relative to heterosexual women, and this effect was only found for women with children.12 Moreover, we estimated that the probability a woman reported working part-time and having ESI in someone else's name increased for lesbians after AB205 relative to heterosexual women, and although the relevant interaction coefficient was not statistically distinguishable from zero, we found that it was sizably positive for women with children (0.059) and essentially zero for women without children (0.001). This pattern is again consistent with the hypothesis that California's reform alleviated labor market constraints for partnered lesbian women with children, allowing some women to reduce their hours while receiving ESI coverage through a working partner.13

Finally, we subjected our main findings on relative health insurance increases for lesbians to a variety of robustness checks. For example, although we are concerned about using partnership as a treatment margin due to possible composition biases suggested in Table 3 for women, we did estimate models that restricted attention to people in partnerships. These models produced very similar results to those in the bottom panel of Table 4: We continued to find a large, positive, and statistically significant increase in the likelihood of any insurance for lesbians compared to heterosexual women following the reform, and there is still evidence consistent with a possible role for dependent ESI. Also, while we chose not to include measures of employment, presence of children, or partnership directly in the health insurance models (due to concerns about endogeneity), we recognize that a large body of research suggests that all of these are important correlates of health insurance. In robustness analyses, we found qualitatively identical results to those in the bottom panel of Table 4 when we separately included either a dummy variable for being employed at all, a dummy variable for the presence of any children in the household, or a dummy variable for living with a partner. That is, we continued to find large and statistically significant increases in health insurance for lesbians compared to heterosexual women following the reform and plausible roles for dependent ESI. None of these additional robustness analyses indicated that the reform had any effects on insurance for gay men relative to heterosexual men, consistent with the findings in the top panel of Table 4. All of these results are available upon request.

CONCLUSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. LITERATURE REVIEW AND INSTITUTIONAL DETAILS
  5. DATA AND EMPIRICAL APPROACH
  6. RESULTS
  7. CONCLUSION
  8. ACKNOWLEDGMENTS
  9. REFERENCES
  10. Biographies

Sexual orientation-based disparities in health outcomes are well documented, and eliminating these disparities is an explicit goal of national health policy. A difficulty in reducing such disparities, however, is that the causes of sexual orientation-based differences in health are not well understood. Disparities in health insurance coverage by sexual orientation have been documented using a variety of samples and methods and may contribute to differences in health outcomes. One plausible cause of these health insurance disparities is that sexual minorities face barriers to accessing a same-sex partner's health insurance benefits due to the historic employer practice of covering heterosexual spouses but not same-sex partners. This possibility has received some attention in the literature (Ash & Badgett, 2006) and is consistent with previous descriptive research using couples-based samples (Buchmueller & Carpenter, 2010; Heck, Sell, & Gorin, 2006), but has not been directly evaluated using quasi-experimental methods.

Recently, several states have adopted laws giving gay and lesbian couples increased rights, including (in some states) the right to access a partner's ESI. Our study provides the first evaluation of such a reform: California's AB205 and AB2208. Specifically, we use DD methods to compare outcomes for gay men and lesbians before and after California's reform with the associated differences for heterosexual men and women, respectively. Our results for men provide no evidence that the reforms had economically or statistically significant effects on differences in insurance coverage overall or by source between gay and straight men. We similarly find no evidence that the laws affected partnership or work effort by gay men. These null findings are not particularly surprising given previous research showing low rates of partnership among gay men and low rates of official domestic partner registrations conditional on partnership among gay men (Carpenter & Gates, 2008; Badgett, Gates, & Maisel, 2008).

Among women, we do find some evidence that California's reform affected outcomes differentially for lesbians compared to heterosexual women. First, we estimate that partnership rates increased and full-time employment decreased for lesbians compared to heterosexual women coincident with the law. These effects are large in magnitude, which suggests that it is inappropriate to compare outcomes for partnered and nonpartnered lesbians before and after the reform due to composition problems associated with partnership. The differences in the effects of California's reform on partnership for gay men versus lesbians are consistent with the limited evidence from administrative sources that indicates that two-thirds or more of sexual minorities who take advantage of domestic partner registries are lesbians. Turning to health insurance outcomes, we find in the DD framework that lesbians were significantly more likely to have any health insurance coverage after the law compared to straight women, with suggestive increases in dependent coverage. Overall, our results are consistent with previous research on health insurance and married women's labor supply and suggest a potential role for policies that extend access to dependent coverage to reduce sexual orientation-based disparities in health insurance coverage among women.

ACKNOWLEDGMENTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. LITERATURE REVIEW AND INSTITUTIONAL DETAILS
  5. DATA AND EMPIRICAL APPROACH
  6. RESULTS
  7. CONCLUSION
  8. ACKNOWLEDGMENTS
  9. REFERENCES
  10. Biographies

We thank the editors, anonymous referees, Scott Hankins, and participants at the 2010 American Society of Health Economists, the 2010 American Psychological Association, and the 2009 Association for Public Policy Analysis and Management meetings for very useful comments. We thank Brandon Traudt, Chris Pankonin, Stone Shi, Kathleen Abanilla, and the staff at the UCLA Data Access Center for assistance with the CHIS data. These data are covered by a confidentiality agreement that restricts access; interested readers can contact the authors for more information. Carpenter thanks the University of California Office of the President Labor and Employment Research Fund, the Paul Merage School of Business, and the UC Irvine Council on Research, Computing, and Library Resources for research support, though results do not imply their endorsement. All errors and omissions are our own.

  1. 1

    Elimination of these disparities has been identified as a major public health goal by Healthy People 2020—the nation's health promotion and disease prevention initiative.

  2. 2

    Throughout this paper we will sometimes use the terms “gay” or “gay couples” to refer to both gay men and lesbians. Because of data limitations, we can say very little about health insurance concerns unique to bisexual or transgender persons, particularly in the context of domestic partner benefits; as such, we focus exclusively on gay men and lesbians.

  3. 3

    The practice varies across different branches of the U.S. federal government, but was made more complicated by the 1996 federal Defense of Marriage Act. California's government workers have enjoyed same-sex domestic partner health benefit eligibility since 2000.

  4. 4

    Federal benefits were unaffected. Note also that the reform preceded California's short-lived period of gay marriage by 3.5 years. Our sample (2001 to 2007) entirely excludes this brief period when gay marriage was legal in the state.

  5. 5

    For examples of research that have examined sexual orientation-related public policy effects on other outcomes such as earnings, see Klawitter and Flatt (1998) and Klawitter (2011).

  6. 6

    An exception is Diamant et al. (2000), who use data from the 1997 Los Angeles Community Health Survey (LACHS) and find that sexual minority women are significantly less likely to be insured than heterosexual women.

  7. 7

    Because the NCS study, the HRC study, and the KFF study do not report availability of domestic partner benefits at the state level over time, we are unable to directly assess whether the reform changed employer benefits differentially in California relative to other states. Data from a confidential employer survey and reported in Ponce et al. (2010), however, do indicate that the percentage of California firms offering health benefits to same-sex couples increased substantially over this time period, from 22 percent in 2003 to 35 percent in 2004, 64 percent in 2006, and 70 percent in 2007 (see Appendix B to Ponce et al., 2010). Consistent with this, the NCS data indicate that among private industry workers, offers of health benefits to partners of same-sex employees is by far most common in states in the Pacific division (including California)—at 52 percent—compared to workers in the rest of the U.S. (with the Mountain, New England, and Middle Atlantic states being next closest at 39, 38, and 31 percent, respectively, and all other divisions at 22 percent or less). Note that even though some firms were offering domestic partner benefits prior to the reform, gays and lesbians may have viewed those benefits as uncertain or legally revocable prior to the reform (which they were).

  8. 8

    Note that the 2001 to 2007 CHIS does not ask whether the respondent is in a “registered domestic partnership,” which is the strict eligibility definition under laws like AB205 and AB2208. In practice, employers often consider partnerships that are not officially registered with the local or state government as long as other key criteria are met (e.g., cohabitation, shared financial arrangements, etc.). Using a marital status question to identify partnership status is less than perfect; for a discussion, see Carpenter and Gates (2008). We note that a small number of gay men (34) and lesbians (25) report being “married,” despite the fact that gay marriage was not legal in California until June 2008. Absent a household sex roster, we are unable to determine whether these are “closeted” heterosexuals, sexual minorities who were legally married in another jurisdiction (e.g., Canada), or sexual minorities who consider themselves to be effectively married. See Carpenter and Gates (2008) for a discussion of these issues. We do not exclude them from the current analysis, but we note they constitute a very small share of the gay and lesbian samples (2 and 2.5 percent, respectively).

  9. 9

    In some years the CHIS has asked questions about health insurance coverage of spouses. We are unable to use this information because 1) the questions changed over time; 2) the questions were not always asked of people living with partners; and 3) the questions specifically refer to spouses as opposed to partners.

  10. 10

    Note that in this scenario we would not expect to see overall increases in the probability of having any health insurance (which we do observe in Table 4), but instead would observe a shifting of sources. The increase in the probability of having any insurance could only therefore work through the channel of increased access to a partner's ESI if the partner was previously uninsured. Again, however, we do not observe the insurance status of the partner.

  11. 11

    In results not reported but available upon request, we found no evidence that the probability of having any children changed differentially for lesbians relative to heterosexual women after the law went into effect. The relevant coefficient estimate on the interaction term in a model where the dependent variable was an indicator variable for the presence of any children was .012 with a standard error of .033.

  12. 12

    The relevant coefficient on the interaction term for women with children was -.208 with a standard error of .099, while the relevant coefficient on the interaction term for women without children was .016 with a standard error of .048.

  13. 13

    Consistent with the lack of an effect on the likelihood of any employment in Table 3, we did not find significant interaction effects for the outcome of no employment and dependent ESI. This was true both for women with children and for women without children.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. LITERATURE REVIEW AND INSTITUTIONAL DETAILS
  5. DATA AND EMPIRICAL APPROACH
  6. RESULTS
  7. CONCLUSION
  8. ACKNOWLEDGMENTS
  9. REFERENCES
  10. Biographies
  • Abraham, J., & Royalty, A. (2006). Health insurance and labor market outcomes: Joint decision-making within families. Journal of Public Economics, 90, 15611577.
  • Ash, M., & Badgett, M. V. (2006). Separate and unequal: The effect of unequal access to employment-based health insurance on gay, lesbian, and bisexual people. Contemporary Economic Policy, 24, 582599.
  • Badgett, M. V. (1995). The wage effects of sexual-orientation discrimination. Industrial and Labor Relations Review, 48, 726739.
  • Badgett, M. V. (2001). Money, myths, and change: The economic lives of lesbians and gay men. Chicago: University of Chicago Press.
  • Badgett, M. V., Gates, G., & Maisel, N. (2008). Registered domestic partnerships among gay men and lesbians: The role of economic factors. Review of Economics of the Household, 6, 327346.
  • Black, D., Gates, G., Sanders, S., & Taylor, L. (2000). Demographics of the gay and lesbian population in the United States: Evidence from available systematic data sources. Demography, 37, 139154.
  • Black, D., Sanders, S., & Taylor, L. (2007). The economics of gay and lesbian families. Journal of Economic Perspectives, 21, 5370.
  • Buchmueller, T. (1996/1997). Marital status, spousal coverage, and the gender gap in employer-sponsored health insurance. Inquiry, 33, 308316.
  • Buchmueller, T., & Carpenter, C. (2010). Disparities in health insurance coverage, access, and outcomes for individuals in same-sex versus different-sex relationships, 2000 to 2007. American Journal of Public Health, 100, 489495.
  • Buchmueller, T., & Valletta, R. (1999). The effect of health insurance on married female labor supply. Journal of Human Resources, 34, 4271.
  • Bureau of Labor Statistics, United States Department of Labor. (2011). Employee benefits in the United States. Report #USDL-11-1112. Washington, DC: Author. Retrieved July 28, 2011, from http://www.bls.gov/news.release/pdf/ebs2.pdf.
  • Carpenter, C. (2003). Sexual orientation and body weight: Evidence from multiple surveys. Gender Issues, 21, 6074.
  • Carpenter, C. (2004). New evidence on gay and lesbian household incomes. Contemporary Economic Policy, 22, 7894.
  • Carpenter, C. (2005). Self-reported sexual orientation and earnings: Evidence from California. Industrial and Labor Relations Review, 58, 258273.
  • Carpenter, C., & Gates, G. (2008). Gay and lesbian partnership: Evidence from California. Demography, 45, 573590.
  • Cochran, S., & Mays, V. (2000). Relation between psychiatric syndromes and behaviorally defined sexual orientation in a sample of the US population. American Journal of Epidemiology, 151, 516523.
  • Cochran, S., Mays, V., Bowen, D., Gage, S., Bybee, D., Roberts, S., Goldstein, R., Robison, A., Rankow, E., & White, J. (2001). Cancer-related risk indicators and preventive screening behaviors among lesbians and bisexual women. American Journal of Public Health, 91, 591597.
  • Daling, J., Weiss, N., Hislop, T., Maden, C., Coates, R., Sherman, K., Ashley, R., Beagrie, M., Ryan, J., & Corey, L. (1987). Sexual practices, sexually transmitted diseases, and the incidence of anal cancer. New England Journal of Medicine, 317, 973977.
  • Denenberg, R. (1995). Report on lesbian health. Women's Health Issues, 5, 181191.
  • Diamant, A. L., Schuster, M. A., & Lever, J. (2000). Receipt of preventive health care services by lesbians. American Journal of Preventive Medicine, 19, 141148.
  • Diamant, A. L., Wold, C., Spritzer, K., & Gellberg, L. (2000). Health behaviors, health status, and access to and use of health care: A population-based study of lesbian, bisexual, and heterosexual women. Archives of Family Medicine, 9, 10431051.
  • Farber, H., & Levy, H. (2000). Recent trends in employer-sponsored health insurance coverage: Are bad jobs getting worse? Journal of Health Economics, 19, 93119.
  • Gates, G., Badgett, M. V. L., & Ho, D. (2008). Marriage, registration, and dissolution by same-sex couples in the US. Los Angeles: Williams Institute, UCLA School of Law, University of California. Retrieved July 23, 2010, from http://repositories.cdlib.org/uclalaw/williams/gates_1.
  • Harris Interactive/Witeck-Combs Communications. (2002). Fewer than half of all gay, lesbian, bisexual, and transgender adults surveyed say they have disclosed their sexual orientation to their health care provider. Press Release.
  • Human Rights Campaign. (2009). Domestic partner benefits: Prevalence among private employers. Washington, DC: Author. Retrieved October 15, 2009, from http://www.hrc.org/issues/workplace/benefits/11612.htm.
  • Heck, J., Sell, R., & Gorin, S. (2006). Health care access among individuals involved in same-sex relationships. American Journal of Public Health, 96, 11111118.
  • Institute of Medicine. (2011). The health of lesbian, gay, bisexual, and transgender people: Building a foundation for better understanding. Washington, DC: National Academy Press.
  • Kaiser Family Foundation. (2009). Among firms offering health benefits, distribution of whether employers offer health benefits to unmarried same-sex domestic partners, by firm size, region, and industry, 2009. Menlo Park, CA: Author. Retrieved October 15, 2009, from http://facts.kff.org/chart.aspx?ch=1034.
  • Kapinos, K. (2009). Changes in spousal health insurance coverage and female labor supply decisions. Forum for Health Economics and Health Policy, 12, Article 1.
  • Klawitter, M. (2011). Multilevel analysis of the effects of antidiscrimination policies on earnings by sexual orientation. Journal of Policy Analysis and Management, 30, 334358.
  • Klawitter, M., & Flatt, V. (1998). The effects of state and local antidiscrimination policies on earnings for gays and lesbians. Journal of Policy Analysis and Management, 17, 658686.
  • Laumann, E., Gagnon, J., Michael, R., & Michaels, S. (1994). The social organization of sexuality: Sexual practices in the United States. Chicago, IL: University of Chicago Press.
  • Olson, C. A. (1998). A comparison of parametric and semiparametric estimates of the effect of spousal health insurance coverage on weekly hours worked by wives. Journal of Applied Econometrics, 13, 543565.
  • Ponce, N., Cochran, S., Pizer, J., & Mays, V. (2010). The effects of unequal access to health insurance for same-sex couples in California. Health Affairs, 29, 15391548.
  • Solarz, A. (Ed.). (1999). Lesbian health: Current assessment and direction for the future. Washington, DC: National Academy Press.
  • Stall, R., & Wiley, J. (1988). A comparison of alcohol and drug use patterns of homosexual and heterosexual men: The San Francisco Men's Health Study. Drug and Alcohol Dependence, 22, 6373.
  • Stall, R., Greenwood, G., Acree, M., Paul, J., & Coates, T. (1999). Cigarette smoking among gay and bisexual men. American Journal of Public Health, 89, 18751878.
  • Wellington, A., & Cobb-Clark, D. (2000). The labor supply effects of universal health coverage: What can we learn from individuals with spousal coverage? In S. Polacheck (Ed.), Research in Labor Economics, Vol. 19 (pp. 315344). Greenwich, CT: JAI Press.
  • Zellers, W., McLaughlin, C., & Frick, K. (1992). Small-business health insurance: Only the healthy need apply. Health Affairs, 11, 174180.
  • Zimmer, D. M. (2009). Insurance arrangements among married couples: Analysis of benefit substitution and compensating differentials. Journal of Family and Economic Issues, 30, 428439.

Biographies

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. LITERATURE REVIEW AND INSTITUTIONAL DETAILS
  5. DATA AND EMPIRICAL APPROACH
  6. RESULTS
  7. CONCLUSION
  8. ACKNOWLEDGMENTS
  9. REFERENCES
  10. Biographies
  • THOMAS C. BUCHMUELLER is Professor of Business Economics and Public Policy at the Stephen Ross School of Business at the University of Michigan, Ann Arbor, MI 48109.

  • CHRISTOPHER S. CARPENTER is Associate Professor of Economics/Public Policy at the Paul Merage School of Business at the University of California, Irvine, CA 92697.