Rev Econ Household (2008) 6:327–346 DOI 10.1007/s11150-008-9041-6 Registered domestic partnerships among gay men and lesbians: the role of economic factors M. V. Lee Badgett Æ Gary J. Gates Æ Natalya C. Maisel Received: 1 March 2007 / Accepted: 1 July 2008 / Published online: 24 July 2008 Springer Science+Business Media, LLC 2008 Abstract In this paper, we predict the demand for a marriage-like status—registered domestic partnership–among same-sex couples. Domestic partnership in the state of California now comes with almost all of the rights and responsibilities of marriage that a state can provide. We use the LGBT (Lesbian, Gay, Bisexual, and Transgender) Tobacco Use Survey conducted by the California Department of Health and the Field Research Corporation in 2003. From this telephone survey, we use a probability sample of 1,002 lesbian and gay individuals in California. Using multinomial probit models of partnership status (single, not cohabiting, cohabiting, or registered), we find limited evidence of economic motivations in the choice to register. Gay men’s likelihood of registration rises with income; lesbians’ probability of registration rises with age. Couples with longer duration are more likely to register, suggesting that registration and duration are complementary signals of commitment and possibly of the need for rights and benefits of registration. Keywords Economics of household Sexual orientation Lesbian and gay issues Marriage Domestic partnership Same-sex couples M. V. Lee Badgett Center for Public Policy & Administration, University of Massachusetts, Amherst, MA, USA M. V. Lee Badgett (&) G. J. Gates The Williams Institute, UCLA School of Law, 405 Hilgard Avenue, P.O. Box 951476, Los Angeles, CA 90095, USA e-mail: lbadgett@pubpol.umass.edu; badgett@law.ucla.edu G. J. Gates e-mail: gates@law.ucla.edu N. C. Maisel Department of Psychology, University of California, Los Angeles, 1285 Franz Hall, P.O. Box 951563, Los Angeles, CA 90095, USA e-mail: maisel@ucla.edu 123 328 JEL Classifications M. V. Lee Badgett et al. J12 J16 J18 1 Introduction Social scientists in every discipline have studied marriage. Until relatively recently, the idea of a lifelong—or at least long-term—legal, personal, economic, and social commitment between two individuals centered exclusively on a heterosexual couple. Correspondingly, studies of marriage and relationships focused almost exclusively on different-sex couples. In the last twenty years, however, both legal and social institutions related to committed couples have evolved considerably. Social scientists have begun to learn about the social, economic, and psychological characteristics of same-sex couples (e.g. Blumstein and Schwartz 1983; Badgett 2001; Gates and Ost 2004; Andersson et al. 2006; Black et al. 2007). However, we know virtually nothing about which gay and lesbian couples decide to take advantage of the opportunity to legally recognize their relationship and why they make that choice, despite the fact that more than 23 percent of the U.S. population now lives in a state with a form of legal recognition of same-sex couples akin to marriage (e.g., marriage, civil unions, or domestic partnership). Both the legislative debate over the recognition of same-sex couples and the courtroom battles for the right to marry have highlighted the economic rights and benefits that come with marriage (see Baker vs. State of Vermont; Goodridge vs. Massachusetts). Since equality of access to those economic rights for same-sex couples appears to be a central goal of advocates seeking the expansion of rights, one might also expect economic considerations to play a central role in the decisions of couples who gain those rights. Therefore, in this paper, we use a new Californiabased data source to study the influence of economic factors on the demand for registered domestic partnership among same-sex couples. Registered domestic partnerships (RDP) in the state of California comes with almost all of the rights and responsibilities of marriage that a state can provide, and in recent litigation, the State of California argued that registered domestic partnership is legally equivalent to marriage in terms of such state-provided rights, benefits, and obligations. If lesbian and gay individuals perceive domestic partnership to be equivalent to marriage in terms of benefits and obligations, as we assume for our purposes, then we would expect them to approach the decision to register in the same way they would consider the decision to marry. To analyze the decisions about cohabitation and registration among same-sex couples, this study draws on the Lesbian, Gay, Bisexual, and Transgender (LGBT) Tobacco Use Survey, which was conducted by the California Department of Health and the Field Research Corporation in 2003–2004. This telephone survey included a probability sample of self-identified LGBT individuals in California with a total sample size of 876 gay men and 280 lesbians. More than half (51%) of these individuals lived with a partner, and 28% of partnered gay and lesbian people had registered their partnerships in a state or local registry. 123 Registered domestic partnerships among gay men and lesbians 329 We estimate multinomial probit models of partnership status to predict whether an individual is single, with a partner but not cohabiting, cohabiting with a partner but not registered, or cohabiting and registered. The multivariate approach distinguishes this paper from a recent study using the same dataset (see Carpenter and Gates 2008, which compares cross-tabulations on same-sex couples across datasets). We find limited evidence of economic motivations in the choice to register, though we do observe that gay men’s likelihood of registration rises with income. Couples with longer duration are more likely to register, suggesting that registration and duration are complementary signals of commitment and possibly of the need for rights and benefits of registration. Finally, we note that the relatively weak evidence of economic motivations found here is also consistent with the hypothesis that domestic partnership is viewed as an inferior status by lesbians and gay men when compared with marriage, even though domestic partnership is legally and economically equivalent to marriage for same-sex couples. 2 Conceptual framework Why do couples marry, and why might same-sex couples seek legal recognition through the marriage-like status of registered domestic partnership? Economic theories of utility maximization and household production argue that couples will marry when they are better off doing so (e.g. Becker 1991). Marriage may promote an individual’s and a couple’s economic well-being for several reasons: • • • • Promoting specialization of labor: In Becker’s household production model, marriage increases household efficiency by promoting the division of labor (Becker 1991). Specialization in particular tasks is risky for each individual, though, since the household might break up. A long-term commitment sealed by a marriage contract is important for providing security for both spouses and makes specialization more likely. Reducing ‘‘transaction costs’’: Robert Pollak (1985) argues that marriage has practical value by reducing transaction costs for couples, which are the costs in time and money involved in negotiating a legal relationship. The marital contract defines what would happen if the marriage ends by death or divorce, specifying who gets marital property and control of children. The long-term nature of marriage removes the need to renegotiate the terms of the legal relationship as couples experience changed circumstances. Providing social insurance: Wealth and income pooling by married couples and families provides insurance against bad times, such as the failure of a harvest or the loss of a job (Pollak 1985). An employed person usually has a legal obligation to support his or her spouse financially, if necessary. Signaling commitment: The willingness to marry is an important signal of commitment to a relationship (Grossbard-Shechtman 1982; Eskridge 1996), and it may underlie many other advantages of marriage. By agreeing to marry, each partner signals greater effort to maintain the relationship, a greater likelihood that the relationship will endure, and perhaps an agreement to make a fair 123 330 • • • M. V. Lee Badgett et al. settlement if, despite the good intentions of both individuals, the relationship should end. Taking advantage of economies of scale: By encouraging larger household sizes (more than one adult), marriage helps families take advantage of ‘‘economies of scale’’ (Nelson 1988). Economies of scale mean that doubling the inputs of time and other resources in some task results in more than double the output of family-related goods and services. Promoting the provision of caring labor: The long-term nature of the marital commitment promotes reciprocity and altruism, as partners take care of one another and any children they might be raising together. The unpaid work done in families is essential for the survival of healthy human beings (Folbre 1994). Access to institutional supports from third parties: Marriage also comes with economic and legal benefits and rights that often make couples better off than if they were to remain unmarried (e.g. Badgett 2001). Some are provided by the state, such as the right to sue for wrongful death or to file joint taxes; some are provided by third parties, such as employers who provide health insurance benefits for spouses of employees. Expectations of gaining from any or all of those advantages might enter into a couple’s deliberations about marriage. If such factors make a couple better off, and if none make a couple worse off (e.g. due to the absence of a marriage penalty in the tax code), we would expect these welfare-enhancing aspects of marriage to increase the likelihood that a couple would choose to marry. For instance, one study finds that same-sex couples would pay lower state taxes when filing joint state tax returns (Badgett and Sears 2005). Some—but not all—of these gains can be achieved by couples through cohabitation without legal recognition. To the extent that these economic gains require long-term relationships to encourage reciprocity and specialization, however, we would expect such motives to be more closely aligned with choosing marriage than with cohabitation because of the legal obligations that flow from marriage. And while cohabitation contracts could substitute for some of the contractual aspects of marriage (assuming that courts will enforce the agreements), private contracts cannot substitute for marriage when it comes to recognition by third parties, such as the state or employers. Cohabitation contracts might be viewed as inferior options to marriage also because they are likely to be more sensitive to individuals’ relative bargaining power. Finally, relationships united solely by private contracts are unlikely to enjoy the social and cultural recognition of a publicly defined marriage. Several other perspectives also suggest that the link between economic wellbeing and marriage is not so clear cut. Indeed, evidence from heterosexual couples suggests that the crucial difference between marriage and cohabitation may be more cultural than economic. In the United States, people are more likely to marry than cohabit if they are religious and not politically liberal, as well as when they have strong intentions to have children, have traditional ideas about gender roles, and do not value their individual freedom as highly (Rindfuss and VandenHeuvel 1990; Bumpass et al. 1991; Clarkberg et al. 1995; Smock 2000). Kiernan argues that the 123 Registered domestic partnerships among gay men and lesbians 331 choice to cohabit may involve a conscious desire to avoid dependency or a loss of power, particularly for women (K. Kiernan 2002, ‘‘Unpublished manuscript’’). Economic well-being might also be connected to marriage in other ways that complicate the standard perspective in economics. Some have found that married men, in particular, have higher wages in part because of selection of better-off individuals into marriage (Benham 1974; Nakosteen and Zimmer 1987; Grossbard-Shechtman 1986; Gray 1997; Hersch and Stratton 2000). Two recent studies add to the evidence for another effect of male income on different-sex marriage. A study of low-income women (Edin and Kefalas 2005) and another of working class cohabiting couples (Smock et al. 2005) suggest that some couples defer marriage until they have the economic resources (jobs and stable incomes) thought necessary for marriage. While economic theories of the family were developed to explain behavior of different-sex couples, these conceptual frameworks might help explain the decisions of same-sex couples, too, now that options for legal recognition have expanded (e.g. Badgett 2001; Black et al. 2007). Other approaches suggest some alternative hypotheses positing a more limited role for economic value in same-sex couples’ decision-making. First, some commentators have hypothesized that same-sex couples lack the complementarities and the child-bearing intentions that characterize different-sex couples and that enhance the value of marriage for such couples (e.g. Allen 2006), suggesting that applying an economic approach derived from the experience of different-sex couples would be inappropriate for same-sex couples. A second argument against a simple application of the economic model is that registered domestic partnership may not be an institution that is equivalent to marriage. As noted earlier, California’s registered domestic partnership was designed to be the legal equivalent of marriage: ‘‘Registered domestic partners shall have the same rights, protections, and benefits, and shall be subject to the same responsibilities, obligations, and duties under law… as are granted to and imposed upon spouses.’’ (California Family Code 297.5(a)) But procedures for entering a marriage and a domestic partnership differ. While marriage requires a couple to apply for a license in person and then go through a ceremony with an officiant and a witness, domestic partnership registration only requires that couples send a signed and notarized document to the California Secretary of State. Further, neither the federal government nor most other states recognize the domestic partner status, as is also the case with marriages of same-sex couples in Massachusetts. Other than the portability and federal recognition issues, though, marriage and domestic partnership are economically equivalent for same-sex couples (although not for differentsex couples).1 But the fact that some same-sex couples in California sued for (and won) the right to marry suggests that both couples and the California Supreme Court saw a distinction between marriage and domestic partnership. Indeed, the Court found that 1 Even the California Supreme Court noted that domestic partnership provides virtually all of the rights and obligations of marriage under California law. The Court ruled that calling legal recognition of relationships ‘‘marriage’’ for different-sex couples and ‘‘domestic partnership’’ for same-sex couples constitutes sexual orientation discrimination that violates the right to equal protection of the law. In re Marriage Cases, California Supreme Court (S147999), May 15, 2008, http://www.courtinfo.ca.gov/opinions/documents/S147999.PDF (accessed May 15, 2008). 123 332 M. V. Lee Badgett et al. marriage differs from domestic partnership in its symbolic meaning and common understanding, not in its legal rights and responsibilities. As one considers the relationship between economic factors and registering as domestic partners, Allen’s view that differences in child-bearing propensity mean that economic theories of different-sex coupling are not appropriately applicable to same-sex couples and the Court’s observation that domestic partnership and marriage are inherently unequal offer two perspectives that contribute to additional hypotheses: (1) economic factors should simply not matter in the decisions of samesex couples to register their partnership (a la Allen) or (2) the economic value of legal recognition might be diminished or overshadowed by the separate-but-equal nature of domestic partnership (a la the Court). In either case, the role of economic factors in the choice to register as domestic partners is a question well-suited for empirical testing. Few opportunities exist to examine the choices of same-sex couples to seek legal marriage or registration of their partnership, and few studies to date have analyzed the situations that do exist. In the United States, only Massachusetts and, quite recently, California have allowed same-sex couples to marry, but Vermont, New Hampshire, New Jersey, Connecticut, and Oregon also offer a status that is intended to provide all of the benefits of marriage that a state can provide. Canada, the Netherlands, Spain, and South Africa allow same-sex couples to marry, and many other European countries provide a status similar to civil unions. These situations provide researchers with opportunities to analyze the marriage or registration decisions of same-sex couples. Several qualitative studies of nonrandom samples of same-sex couples have analyzed the motives for seeking symbolic recognition (e.g. commitment ceremonies) or legal recognition (marriage, civil unions, or registered partnership) (Lewin 1988; Stiers 1999; Hull 2003; Schechter et al. 2005; Eskridge and Spedale 2006; Badgett 2008). According to these studies, the legal and material benefits of marriage played an important role in the decision of some same-sex couples to marry in Massachusetts, and similar reasons are cited for American same-sex couples’ stated desire to marry where marriage is not a legal option.2 For some European couples, practical benefits such as immigration rights or preferences appear to be attractive aspects of marriage or registered partnership (the European equivalent of civil unions or California’s registered domestic partnership).3 Marriage’s creation of a legal framework for organizing a couple’s life together is also an attractive feature of marriage for couples in both the United States and the Netherlands (e.g. Badgett 2008). These differing perspectives on the value of marriage suggest several more specific hypotheses that we will analyze using the data described in the next section. In some cases the existing conceptions of marriage lead to strong predictions about 2 See Schecter et al. 2005, on couples in Massachusetts. Stiers 1999, and Hull 2003, asked couples about their intentions to marry or prioritization of marriage as an issue. 3 For Denmark, see Eskridge and Spedale 2006, p. 102. Andersson et al. 2005, find that many male samesex couples who have registered as partners in Norway and Sweden are binational couples, unlike female same-sex couples and different-sex couples who married over the same period. 123 Registered domestic partnerships among gay men and lesbians 333 the effect of a variable, while in others the direction of an effect is an empirical rather than a theoretical matter. (1) (2) (3) (4) Income: Higher income individuals or couples might be more likely to marry, either because they are more desirable mates or because the higher levels of wealth held by high-income individuals create incentives and the financial ability to marry or register. Wealthier couples might gain more from registration, since a surviving partner will inherit when his or her partner dies intestate (without a will). (California no longer has an estate tax, so spousal advantages in that realm would not be relevant.) Since we do not have direct measures of wealth in the CTUS, income will serve as a proxy for wealth. Alternatively, Grossbard-Shechtman (1982) argues that higher income men might be less likely to marry in heterosexual relationships. In her model, men have a choice of giving risk averse women a high formal commitment (marriage) and a smaller share of income or a combination of a lower level of commitment and a higher share of income. Holding women’s characteristics constant, men with higher incomes will be less likely to marry than are lower income men, since the higher income men can offer women a higher share of male income to compensate their female partners for the economic risks associated with not marrying. If a similar dynamic operates for gay men or lesbians, perhaps because of varying risk aversion between partners, higher incomes might reduce the probability of registering a domestic partnership. Age: Older individuals might be more likely to register or marry for economic reasons. First, they might have a greater need to protect their assets because they have more of them. Second, they might also have a greater need to protect them because of vulnerability to serious illness or death. Children: Having a child in the household would increase the likelihood of coupling and registering. When one partner in a registered domestic (female) partnership gives birth, her partner is automatically considered to be the child’s legal parent. Accordingly, being registered domestic partners may reduce the legal expenses necessary to establish parental rights. Also, if domestic partners pool their income more, it might make it easier for one partner to take primary responsibility for childrearing duties. Need for social insurance or care: Recent events such as a serious illness or a spell of unemployment (or the expectation that such an event might occur) might increase the likelihood of marriage. Registered domestic partnership and marriage give partners legal rights with respect to making health care decisions for an incapacitated partner and provide hospital visitation rights. In the worstcase scenario, a partner’s death will give the surviving partner a stronger legal position vis-à-vis the deceased partner’s estate. In other words, the legal and economic value of registered domestic partnership is higher in these circumstances. So to the extent that the practical value matters, couples in which a partner has experienced serious illness or unemployment should be more likely to register. However, the hypothesis about selection into marriage suggests that an employed or healthy current or potential partner might not be willing to take on caring responsibilities, reducing the likelihood of continuing 123 334 (5) (6) (7) M. V. Lee Badgett et al. or entering a more legally binding relationship. In economic terms, the supply of individuals in precarious positions might be high in the marriage market, but the demand for such individuals as partners might be quite low (GrossbardShechtman 1982; see also Mukhopadhyay 2008). Sex: Female same-sex couples are more likely to be raising children than are male couples, likely raising the unconditional rate of registration among female couples. Even when controlling for current children in the household, women might be more likely to marry or register than men because of greater expectations of future childbearing within the relationship. Marital history: Individuals who have married in the past are likely to have greater knowledge of the practical advantages (and perhaps disadvantages) of legal recognition. Those individuals might also possess marriage-positive attitudes or beliefs that heighten their interest in the status that is as close to marriage as same-sex couples can get in California. Thus this variable might reflect both economic and cultural attraction to legal recognition. Duration of relationship: On one hand, relationships that have lasted longer might increase the likelihood that a couple is willing to make the commitment embodied in registered domestic partnership. In that sense, personal and legal commitments are complementary. On the other hand, in the absence of a legal recognition option for same-sex couples, longevity of the relationship may act as a substitute commitment device for both the couple and for the couple’s family and community. Therefore, the effect of duration on the probability of registering is an empirical question. Studies show that duration of relationship is positively related to the pooling of income in heterosexual couples (Bonke and Uldall-Poulsenl 2007; Winkler 1989). While studies of heterosexual couples have found that longer cohabitation prior to marriage increases the risk of divorce, those studies may not be helpful in understanding the effect of duration on partnership registration for same-sex couples who have only recently had options for formal legal recognition. 3 Data 3.1 Sample To analyze the factors involved in the choice to legally register a same-sex relationship, we use the California Lesbian, Gay, Bisexual, and Transgender (LGBT) Tobacco Use Survey (or CTUS), which was conducted in 2003–2004 by the Field Research Corporation for the California Department of Health. This telephone survey asked a series of questions related to smoking and other matters to a probability sample of 2,287 primarily LGBT-identified individuals in California. We use a subsample of lesbians and gay men from this larger survey, as discussed below. The CTUS used a stratified sampling method to derive a representative sample of the LGBT population in California. (See Bye, Albright, and Krotki (2004) for more detail on weighting and stratification procedures.) The household-level survey 123 Registered domestic partnerships among gay men and lesbians 335 sampled six geographic strata that had low, average, and high proportions of samesex ‘‘unmarried partners’’ in Census 2000. (See Gates and Ost 2004, for details about Census counts of same-sex couples.) Since the Census data only allow identification of same-sex couples, the CTUS sampling strategy could bias the sample toward having more same-sex couples among the LGBT people surveyed than is representative among all LGBT people in California. Indeed, Carpenter and Gates (2008) showed that the proportion of lesbians and gay men partnered with someone of the same sex is relatively high in this sample when compared to the California Health Interview Survey, another smaller probability sample of California lesbians and gay men. However, the demographic characteristics of couples in Census 2000 and the CHIS are very similar to the CTUS couples, suggesting that the sampling procedure resulted in no serious bias in the characteristics of the couples.4 Respondents contacted through random-digit-dialing were asked a screening question about their sexual orientation and about sexual behavior with same-sex partners in order to assess each household for inclusion in the sample.5 The sample in this study includes only those men and women who self-identified as gay or lesbian on a later question. Our focus here is on lesbians and gay men because they are the most likely to partner with and register a relationship with someone of the same sex. We removed bisexuals from the sample because of the need for more complex modeling of the sex of a partner in addition to choices about cohabitation and registration. Two-thirds of the partnered bisexuals in this dataset report having a different-sex partner. In contrast, virtually no self-identified gay men or lesbians are partnered with someone of a different sex. We also removed transgender respondents because of the difficulty of identifying how both respondents and the state would classify transgender individuals for purposes of defining a same-sex relationship. We also did not include a small group of lesbians and gay men who said they were currently married. Unfortunately, the survey did not permit us to determine the sex of spouses in the cases where individuals said they were currently married. Finally, for consistency regarding gay and lesbian identity, we eliminated respondents who identified their sexual orientation as ‘‘no label’’, ‘‘queer’’, ‘‘questioning’’, or ‘‘other’’. After eliminating those with some missing data on variables studied, we have usable data from 768 gay men and 234 lesbians (N = 1002).6 4 However, the one exception to the similarities is that the rates of child-rearing were much lower in the CHIS than in the Census or the CTUS. 5 The screening questions were as follows: ‘‘We are interested in speaking with people who are not often studied in public health research: gay and bisexual men. Would you include yourself in one of these groups?’’ The question for women used ‘‘lesbian or bisexual women.’’ If the respondent answered no to this question, a second question was asked as follows (female example shown here): ‘‘Regardless of whether a person thinks of herself as lesbian, bisexual or heterosexual, we are also interested in speaking with women who have had sex with other women at any time in their life. Do you fall into this category?’’. 6 From the original sample of 2,287 individuals survey, we removed 239 individuals who did not answer the sexual orientation question, 438 men and women who identified as heterosexual, 328 who identified as bisexual, 55 transgender respondents, 30 individuals who said they were currently married, 53 respondents who chose ‘‘no label’’, and 54 who identified as queer, questioning, or other. 123 336 M. V. Lee Badgett et al. 3.2 Partnership and DP registration The CTUS asked respondents if they ‘‘currently have a primary partner,’’ with primary partner defined as ‘‘someone you love more than anyone else and feel a unique commitment to.’’ Subsequent questions asked about the sex of the partner, whether the couple lived together and whether their domestic partnership was registered with either the local or state government. Of those with a primary partner, 127 were living apart from their partner (27.7% weighted percentage of respondents with a partner), 233 were living together but were not registered (55.1% weighted), and 171 were living together and were registered (27.7% weighted). Unfortunately, the wording of the registration question does not distinguish between state and local registration. In 2003–2004, state registration as domestic partners entailed a much larger set of rights and responsibilities than did local registration in several cities in California. (Most local registration processes result in very few benefits or obligations.) In particular, the rights and responsibilities embodied in state domestic partner registration more closely resembled those of marriage as a result of legislation that was passed during the time the survey was in the field. Despite the ambiguity of the question, we argue that most of the people answering ‘‘yes’’ to the survey question on registration were registered with the state and that those who were only registered with a local government shared a personally (if not legally) equivalent commitment. First, the relative numbers of registrations suggest that anyone who is ‘‘registered’’ in our data is much more likely to be registered with the state than with a city. By the end of June 2003, 19,734 couples had registered their partnerships with the state (Secretary of State 2006), and 94–95% were same-sex couples (Susan Cochran, personal communication 2005). That total is much larger than the typical number of local registrations, which numbered only in the hundreds (at most) for same-sex and different-sex couples in several large cities from which we were able to gather data (Sacramento et al.). Indeed, the CTUS tracks the data on partnership registrations reasonably closely. In 2004, the count of registered domestic partnerships was 32% of the number of same-sex couples counted in the Census, which is close to the CTUS weighted rate of 28%. Second, a comparison of a sample of registered same-sex couples in the city of Berkeley with the list of state-registered couples indicates that a sizable minority of those couples are registered with the state as well as with the city. Third, it is possible that couples equate the significance of state and local registration out of the mistaken belief that local registration gives them the same legal status as state registration. The California Secretary of State’s webpage on domestic partnership includes a clarification that local registration is not equivalent to state registration, suggesting that this is a common source of confusion. In the end, although we cannot conclusively say that individuals with registered domestic partners in our dataset are registered with the state, we argue that registration among couples in the CTUS is either in reality or in perception a legally meaningful registration at the state level. 123 Registered domestic partnerships among gay men and lesbians Table 1 Mean characteristics by sex of gay and lesbian sample, LGBT California tobacco use survey 337 All GL Gay men Lesbians Single (%) 35.8 42.8 24.5 Partner-not cohabiting (%) 11.0 11.2 10.7 Partner-cohabiting, not RDP (%) 35.4 35.2 35.7 Partner-RDP (%) 17.8 10.8 29.1 Age 42.3 42.5 42.0 College (%) 53.5 55.6 50.3 White (%) 75.0 77.1 71.6 Black (%) 5.7 2.7 10.4 Hispanic (%) 14.0 13.3 15.3 Other (%) 7.1 9.8 2.7 Race/ethnicity Total duration (among partnered) 9.7 10.8 8.4 Illness (%) 12.3 12.5 11.9 Unemployment (%) 20.1 19.4 21.2 Individual income ($) 42,426 43,694 40,379 Ever married (%) 16.4 12.7 22.2 Has child (%) 14.7 7.5 26.4 Female (%) 38.2 0.0 100.0 N 1,002 768 234 3.3 Other measures The CTUS also includes data on each respondent’s age, education, sex, household income (cut in half as a proxy for individual income for people in couples), presence of children under age 18 in the home,7 marital history (ever married), and race/ ethnicity. The survey also asked respondents to indicate whether they had lost a job in the last year or whether they or their partner had been diagnosed with a serious illness in the last year. In addition to the partnership status of each respondent, we have information on the duration of relationships. However, we have very little information about partners of respondents other than whether they have died or been diagnosed with a serious illness in the last year. Table 1 presents means from the CTUS for all gay and lesbian respondents and for gay men and for lesbians separately. We use sampling weights created by Field Research Corporation to better represent the characteristics of the population of gay and lesbian Californians. More than 40% of gay men (but only a quarter of lesbians) are single. A third of both groups live with a partner without registration, while lesbians are almost twice as likely as gay men to be in a registered domestic partnership. Gay men are slightly older and more likely to be white and college 7 Respondents are not asked about the age or origins of the children, e.g. whether they were born into the current relationship or a past relationship. 123 338 M. V. Lee Badgett et al. Table 2 Means by partnership status for gay and lesbian sample, LGBT California tobacco use survey Single All partnered Partnered–by status Not cohabiting Cohabiting only Registered Age 42.1 42.4 40.3 41.4 45.8 College (%) 49.7 55.6 48.0 54.1 63.3 White (%) 70.3 77.6 68.5 77.8 82.8 Black (%) 5.3 5.9 10.8 6.5 1.5 Hispanic (%) 14.4 13.9 12.4 14.7 13.1 Other (%) 11.4 4.7 8.4 4.4 3.0 Total duration 0 9.8 4.9 10.1 12.2 Illness (%) 15.2 10.6 13.7 9.7 10.7 Unemployment (%) 28.0 15.6 17.3 16.8 12.3 Individual income ($) 38,343 44,705 39,152 44,367 48,821 Race/ethnicity Ever married (%) 11.7 18.9 20.1 12.5 31.1 Has child (%) 8.8 18.0 20.7 15.2 22.0 Female (%) 26.2 45.0 37.2 38.6 62.5 N 471 531 127 233 171 educated than are lesbians. Lesbians have lower incomes and are more likely to be Hispanic, unemployed, ever married, and raising children than are gay men. Table 2 separates the sample by partnership status. Compared with single gay men and lesbians, partnered individuals look quite similar, with some exceptions. Partnered individuals are less likely to have been ill or unemployed recently, and partnered people have higher incomes and are more likely to be female. The last three columns of Table 2 split those in couples by their cohabitation and registration status. The comparison of couples suggests that people in registered domestic partnerships are older and are more likely to be white, female, and ever married than are people whose partnerships are not registered. Identifying predictors of registration more precisely requires a multivariate approach, however. 4 Method The decision to marry or to register as domestic partners involves at least three components, each of which tracks the data available from the CTUS. The first component is the decision to be in a serious romantic relationship. From that perspective, the empirical question involves comparisons between people who are in such a relationship with those who are not. The second component is the decision about cohabitation. This second component also implies a third component—the decision to marry/register or not. (In California, registered domestic partners are required to have a common residence, which makes cohabitation a prerequisite to registration.) While that third marriage/registration decision may or may not be 123 Registered domestic partnerships among gay men and lesbians 339 made explicitly by couples, remaining in the default status of legally single still constitutes a decision of a sort. One way to model these decisions is in a complex three-stage decision-making process: first decide whether to date, then whether to cohabit or not, and then (if cohabiting) whether to register. However, work by Seltzer (2000) and Smock (2000) suggests dating and cohabitation do not always represent the first steps toward marriage, even for heterosexual couples. Some heterosexual individuals choose not to marry and to instead live with unmarried partners or have a non-cohabiting partner. An alternative way of modeling the competing risks gives individuals all four options, which reflects the possibility that the decision to cohabit and marry/register might occur as a package rather than in sequential steps. Accordingly, we model individuals’ decisions with a multinomial probit procedure that allows for four possible states: single, partnering without cohabitation, cohabitation, or legal recognition.8 Since we are most interested in the contrast between cohabitors who are and are not registered domestic partners, we use the registered cohabitors as the base case for the multinomial probits for ease of comparison. 5 Results Tables 3 and 4 present findings from the multinomial probit models for women (Table 3) and men (Table 4). Two separate models are estimated in each table. The first model in the left-hand columns includes all women (men). Thus for the first model, each panel of Tables 3 and 4 presents coefficients from the equation predicting being (1) single, (2) partnered but not cohabiting, or (3) partnered and cohabiting but not registered, the reference being registered domestic partners (RDPs). The second model, in the right-hand columns, limits the sample to people in couples, thus capturing the effect of the characteristics on a partnered person’s probability of cohabiting as opposed to being registered. The coefficients in the multinomial probit procedure reflect the probability of being in a particular status relative to being registered domestic partners. Positive coefficients mean that higher levels of that variable increase the probability of being in that status relative to being registered; negative coefficients mean that factor reduces the probability of being in the status relative to being registered. Here we focus mainly on the factors that distinguish RDPs from partnered respondents who live with their partners, having also allowed respondents to be in the other two categories. For women, few characteristics distinguish lesbians in registered domestic partnerships from those in other statuses. 8 We initially estimated multinomial logit models, which result in similar findings. However, tests of the required independence of irrelevant alternatives rejected the null hypothesis that IIA is an appropriate assumption in this situation. 123 123 Age 1.843* 0.966 * P \ 0.10, ** P \ 0.05, *** P \ 0.01 Note: Reference group is cohabiting women in a registered partnership 0.941 0.571 0.578 0.732 0.770 0.762 0.000 1.516* -0.376 -1.048* -0.003 -0.807 1.178 0.00001 0.133 -0.540 0.021 0.918 0.456 0.551 0.859 0.645 0.752 0.000 0.595 0.691 0.484 -0.008 -0.644 0.599 1.882** -0.858 2.008* 0.00001 0.393 -0.005 -0.194 1.553 0.456 0.706 0.658 -0.026 -0.586 Coeff. Constant -0.566 0.479 1.579** -0.477 1.397* 0.000001 0.382 -0.167 0.021 0.507 SE -0.557*** 0.488 0.539 0.901 0.674 0.739 0.000 -0.061*** -0.433 Coeff. Duration -1.189** 0.657 Other Any child 0.015 Hispanic -0.905* 1.201 Black Ever married 0.00001 Income indiv 0.580 0.707 0.652 -0.303 0.021 0.461 0.003 -1.336*** Unemploy Illness College SE 1.056 0.123 0.631 0.710 0.815 0.752 1.200 0.000 0.738 0.844 0.535 0.030 SE Coeff Coeff. SE Partnered, non cohabiting Partnered, non cohabiting Single (no partner) Partnered, cohabiting, not registered Partnered women only All women Table 3 Multinomial probit models predicting partnership status for women 1.783* -0.042 -0.767 -1.118* -0.808 -0.816 1.634 0.00002* 0.056 -0.686 -0.340 -0.031 Coeff. 1.057 0.040 0.480 0.600 0.891 0.692 1.104 0.000 0.620 0.714 0.549 0.031 SE Partnered, cohabiting, not registered 340 M. V. Lee Badgett et al. Other 0.000 0.000 -0.00001** 0.000 -0.594 0.112 0.563 1.182 -0.00001 0.522 0.553 0.576 0.694 0.967 0.000 0.612 * P \ 0.10, ** P \ 0.05, *** P \ 0.01 Note: Reference group is cohabiting men in a registered partnership 1.517** 0.684 0.995 0.028 0.780 0.553 0.475 0.609 0.520 0.800 1.053** -0.065 0.021 0.536 -0.094*** 0.083 0.527 -0.115 -0.068 0.762 -0.283 0.379 0.374 0.046** -0.024 -0.651 0.637 0.518 0.471 0.507 0.619 0.768 0.324 0.238 0.012 0.347 Duration 1.303** -0.304 -0.076 0.583 0.974 -0.00001 0.487 0.510 0.147 -0.012 Constant 0.540 0.416 0.452 0.477 0.705 0.374 -0.004 0.429 0.014 0.607 1.561** 0.351 0.331 0.216 0.000 SE 1.524** 0.514 Any child -0.459 0.185 0.666 Hispanic Ever married 0.238 Black -0.00001*** Unemploy Income indiv 0.709** 0.560* Illness 0.316 Coeff. 0.011 SE -0.009 0.352 Age College Coeff. Coeff. SE Coeff. SE Partnered, non cohabiting Partnered, non cohabiting Single (no partner) Partnered, cohabiting, not registered Partnered men only All men Table 4 Multinomial probit models predicting partnership status for men 1.938** -0.005 0.770 -0.239 0.238 0.762 -0.525 -0.00002** 0.345 0.699* 0.153 -0.009 Coeff. 0.926 0.024 0.644 0.533 0.700 0.617 1.020 0.000 0.443 0.404 0.424 0.020 SE Partnered, cohabiting, not registered Registered domestic partnerships among gay men and lesbians 341 123 342 • • • M. V. Lee Badgett et al. Single versus RDP: Compared with RDPs, single lesbians are less likely to have a college degree (a negative coefficient in the first panel of Table 3),9 less likely to have been ever married, and less likely to have had a child. Non-cohabiting versus RDP: Lesbians with non-cohabiting partners are younger than registered lesbians, as shown by negative coefficients on age in the second panel of Table 2. The non-cohabitants are also more likely to be black and other (primarily Asian American) than are RDPs. In the second model that predicts registration for lesbians only in couples, the racial patterns hold up. However, age is no longer statistically significant, while duration of relationship is negatively related to non-cohabitation. In other words, non-cohabiting couples have shorter relationships than registered couples, and older couples tend to have longer relationships. Cohabiting versus RDP: Lesbians who live with their partners are also younger than RDPs, but that difference is not statistically significant. In both models, lesbians in RDPs are more likely to have ever been married than unregistered cohabitants. (In unreported simple probits predicting registration only among couples who live together, the ever-married variable is still the only significant predictor of registration for lesbians who live with their partners.) Among men in Table 4, different factors distinguish RDPs from men in other partner status groups. • • • Single versus RDP: Having been diagnosed with a serious illness and having lost your job increases the probability of being single relative to being registered. Single men tend to have lower incomes than men with registered partnerships. Non-cohabiting versus RDP: Having children increases the likelihood of having a non-cohabiting partner. In the second model predicting registration only for men in couples, the presence of children also decreases the likelihood of registration (the coefficient shows children are a positive predictor of being a non-cohabitant). Older men in cohabiting partnerships and men who have recently experienced a serious illness are more likely to have non-cohabiting partners over registered partners. In contrast, men in longer relationships are more likely to be registered than non-cohabiting. Cohabiting versus RDP: Income is the main factor that distinguishes the registered and non-registered men who live with partners. In both models 9 In the specifications shown, we control for educational attainment with an indicator variable for having at least a college degree. We used this simple specification to provide a more intuitive interpretation of our findings. The current specifications suggest that, while not significant, women in registered partnerships do appear to have higher levels of education. We also estimated the models using indicators for some college, college, and graduate degrees. Comparing single women and those in registered partners, the specification suggests an even weaker association with education as none of the indicators are significant. Comparing non-registered partnered women (both cohabiting and non-cohabiting) to those in registered partners, the indicator for some college is positive and significant. This suggests that the ‘‘threshold’’ for finding that women in registered partnerships have higher levels of education is likely higher than the ‘‘some college’’ level. To test this, we also estimated the models with an indictor for a graduate degree only. In this case, we find that single and non-cohabiting partnered women are significantly less likely than women in registered partnerships to hold a graduate degree. There is not a significant difference between cohabiting registered and non-registered women. There were not significant results for men. 123 Registered domestic partnerships among gay men and lesbians 343 estimated for men, higher income individuals are more likely to be registered. Individuals in a cohabiting couple where at least one partner has experienced a serious illness are less likely to be registered. (In simple probits predicting registration, having children also decreases the likelihood of registration.) Similar multinomial probit models (not shown) that combine the male and female subsamples find that being female is positively related to being in a registered partnership in all comparisons. This finding fits with the gender pattern seen in Vermont civil unions (Solomon et al. 2003), but contrasts sharply with higher proportions of male couples than female couples initially marrying or registering in European countries (Waaldijk 2001; Rothblum 2005). Overall, these findings suggest several conclusions about the economic motivations for registration as domestic partners in California: evidence of economic motivations is present but limited. First, the most direct economic measure—income—appears to matter only for men, increasing the probability of registration. Income might be an effective proxy for wealth, and hence for the value of the legal framework that guides the distribution of assets in the case of death or relationship dissolution. Higher incomes might allow for more public celebrations that could (but do not necessarily) go along with the simple bureaucratic process of creating a registered domestic partnership. Or men with higher incomes might seek out registered domestic partnership as a legal status that also has a higher social status. However, it is not clear why income would not matter for women in the same way if these economic factors are important. Interpreting the gender difference in the context of Grossbard-Shechtman’s model suggests that gay men might be more risk averse than lesbians, with high income gay men using their income to convince their partners to marry in order to create more stable relationships (GrossbardShechtman 1982). It may be that within gay male relationships, the lower earning partner receives a higher income transfer from his higher earning partner than their female counterparts in lesbian or heterosexual relationships receive from their partners. Second, economic factors may be at work in several less direct measures of the economic value of registration. Perhaps age better captures these economic considerations for women than does income. Older women are more likely to be registered (though the differences are not always statistically significant), perhaps because they have more assets to protect or feel more vulnerable to illness or death compared with younger women, although these effects disappear when controlling for relationship duration. Third, we find a positive impact of duration on the likelihood of being registered versus being in a non-cohabiting relationship for both men and women. This suggests that longevity of relationships is also a measure of commitment, and that longevity is complementary to registration rather than being a substitute for legal status. Since domestic partnership registration was a very recent option at the time of the survey, the causal connections are likely to be clearer here: longer duration makes registration more likely rather than the other way around. Longtime couples have had time to engage in symbolic or practical steps to achieve some legal protection or cultural recognition for their relationships. The length of the 123 344 M. V. Lee Badgett et al. relationship provides information to each individual in the couple that the other is seriously committed to the relationship. Longtime couples have also had time to achieve the kinds of economic gains that some economic theories predict are likely tied to marriage, such as economics of scale, social insurance, and even specialization. Long duration couples have had time to establish themselves socially and economically, so the relative gains to registration for them are likely smaller than the gains for newer couples. In this case, one might expect registered couples to have shorter duration relationships than their non-registered counterparts. But we find the opposite. Long duration couples are in fact more likely than newer couples to seek registration. Clearly, they see a distinct value to registration that goes beyond what longevity can facilitate. This finding is consistent with an economic explanation whereby couples see a clear value in the specific legal and economic benefits that only a formalized status can provide. The economic relevance of other meaningful characteristics that distinguish registered gay men and lesbians from other individuals is harder to infer, however. Take, for example, the role of having ever been married. If having been married conveys information about the economic value of registration, then we would also expect ever married men and women to be more likely to register. While this is true for women, it does not hold for men. Of course, we know that the general economic circumstances for men and women are quite different, so perhaps it is not a surprise that they would respond differently to some economic incentives. We would also expect that the experience of a serious illness by a partner would lead couples to register as a way of providing additional economic protection. But it does not. In fact, for men, illness is negatively associated with registration. Women’s greater demand for registered domestic partnership compared with men also creates ambiguities in relation to an economic explanation. On the one hand, if gender were simply a proxy for different levels of wealth, we would expect women to be less likely to register since they have fewer financial assets that registration and accompanying formal dissolution/divorce laws could protect. On the other hand, lesbians are more likely to have children than gay men and this could certainly complicate such a simplistic explanation since domestic partnership offers potential economic incentives associated with parenting and children. For example, registration can provide easier access to health care for all members of a family. It can also make it easier for both partners to gain parental rights, meaning that both partners would be able to make important decisions in the case of a medical emergency affecting a child. Curiously, the presence of children in the home has a positive influence on lesbians’ propensity to partner and register but a negative influence on gay men’s propensity to cohabit. One could certainly make an argument that the positive association between children and registration for lesbians is evidence that they are responding to some of the economic incentives above, but given that gay men do not respond in the same way, there might be other factors at play. Perhaps lesbians with children gain more from having a partner’s involvement and household labor than do gay men with children. Or perhaps children are a more attractive characteristic for potential lesbian partners than for potential gay male partners. Census data 123 Registered domestic partnerships among gay men and lesbians 345 suggest that many of the children being raised by same-sex couples are likely to have been born into prior heterosexual relationships (Gates and Romero 2008). Evidence also suggests that women are more likely than men to have sole custody of children (Cancian and Meyer 1998; Cook and Brown 2005). Single gay men may perceive a greater risk of a custody challenge if they were to partner with another man than single lesbians perceive if they were to partner with a woman. It is also possible that the gender difference in registration reflects women’s greater intentions to have children in the future, or even a greater biologically-based preference for commitment. Finally, we offer a quite different potential explanation for the lack of strong predictive power of likely covariates of registration. If same-sex couples see domestic partnership as an inferior legal and symbolic status when compared with marriage, they might not react in the way predicted by economic theory. Registered domestic partnership is simply not marriage. The federal government does not recognize the status and it is generally not portable across state lines (with the exception of a handful of states that recognize other states’ civil union and civil union-like relationships). Entering domestic partnership requires no ceremony or public vows. In short, couples’ actions (or, perhaps more appropriately, inactions) may indicate the lesser economic and cultural value of domestic partnerships. Our data suggest that only about a third of same-sex cohabiting couples in California are registered. In contrast, 91% of cohabiting different-sex couples in California are married.10 By heterosexual standards, lesbians and gay men appear to have a relatively low take-up rate for formal recognition of their relationships. This suggests that policymakers who have created separate statuses to recognize the relationships of lesbians and gay men have not created a status that lesbians and gay men view as equivalent to marriage. As more states begin to offer marriage to same-sex couples, researchers will be presented with better conditions under which they can assess the relative value of marriage to lesbians and gay men and the potential differences in how they value marriage compared to other forms of legal recognition. References Allen, D. W. (2006). An economic assessment of same-sex marriage laws. Harvard Journal of Law & Public Policy, 29(3), 949–980. Andersson, G., Noack, T., Seierstad, A., & Weedon-Fekjaer, H. (2006). The demographics of same-sex ‘marriages’ in Norway and Sweden. Demography, 43(1), 79–98. Badgett, M. V. L. (2001). Money, myths, and change: The economic lives of lesbians and gay men. Chicago: University of Chicago Press. Badgett, M. V. L., & Sears, R. B. (2005). Putting a price on equality? The impact of same-sex marriage on California’s budget. Stanford Law and Policy Review, 16, 85–197. Badgett, M. V. L. (2008). Something old, something new: Will gay people change marriage? Manuscript. Becker, G. (1991). Treatise on the family. Cambridge: Harvard University Press. Benham, L. (1974). Benefits of women’s education within marriage. Journal of Political Economy, 82(2, Part 2: Marriage, Family Human Capital, and Fertility), S57–S71. Black, D. A., Sanders, S., & Taylor, L. (2007). The economics of lesbian and gay families. Journal of Economic Perspectives, 21(2), 53–70. 10 Author calculations using Census 2000, SF-1 tables P-18 and PCT-14. 123 346 M. V. Lee Badgett et al. Blumstein, P., & Schwartz, P. (1983). American couples. New York: William Morrow & Co. Bonke, J., & Uldall-Poulsen, H. (2007). Why do families actually pool their income? Evidence from Denmark. Review of Economics of the Household, 5(2), 113–128. Bumpass, L., Sweet, J., & Cherlin, A. (1991). The role of cohabitation in declining rates of marriage. Journal of Marriage and the Family, 53(4), 913–927. Bye, L., Albright, V., & Krotki, K. (2004). Survey methodology report for the 2003 California LGBT tobacco survey. San Francisco, CA: Field Research Corporation. Cancian, M., & Meyer, D. R. (1998). Who gets custody? Demography, 35(2), 147–157. Carpenter, C., & Gates, G. (2008). Gay and lesbian partnership: Evidence from multiple surveys. Demography, 45(3). Clarkberg, M., Stolzenberg, R., & Waite, L. (1995). Attitudes, values, and entrance into cohabitational versus marital unions. Social Forces, 74(2), 609–632. Cook, S. T., & Brown, P. (2005). Recent trends in children’s placement arrangements in divorce and paternity cases in Wisconsin. University of Wisconsin-Madison: Institute for Research on Poverty. Edin, K., & Kefalas, M. (2005). Promises I can keep: Why poor women put motherhood before marriage. Berkeley: University of California Press. Eskridge, W. N., Jr. (1996). The case for same-sex marriage. New York: Free Press. Eskridge, W. N., Jr., & Spedale, D. R. (2006). Gay marriage: For better or for worse? Oxford: Oxford University Press. Folbre, N. (1994). Who pays for the kids? Gender and the structures of constraint. London: Routledge. Gates, G. J., & Ost, J. (2004). The gay and lesbian atlas. Urban Institute Press: Washington, D.C. Gates, G. J., & Romero, A. (2008). Parenting by gay men and lesbians: Beyond the current research. In E. Peters & C. M. Kamp Dush (Eds.), Marriage and family: Multiple complexities and perspectives. Columbia University Press. Grossbard-Shechtman, A. (1982). A theory of marriage formality: The case of Guatemala. Economic Development and Cultural Change, 30(4), 813–830. Grossbard-Shechtman, A. (1986). Marriage and productivity–an interdisciplinary analysis. In B. Gilad & S. Kaish (Eds.), Handbook of behavioral economics. Greenwich: JAI Press. Hull, K. E. (2003). The cultural power of law and the cultural enactment of legality: The case of same-sex marriage. Law and Social Inquiry, 28(3), 629–657. Lewin, E. (1998). Recognizing ourselves: Ceremonies of lesbian and gay commitment. New York: Columbia University Press. Mukhopadhyay, S. (2008). Do women value marriage more? The effect of obesity on cohabitation and marriage in the USA. Review of Economics of the Household, 6, 111–126. Nelson, J. A. (1988). Household economies of scale in consumption: Theory and evidence. Econometrica, 46, 1301–1314. Pollak, R. (1985). A transaction cost approach to families and households. Journal of Economc Literature, 23, 581–608. Rindfuss, R. R., & VandenHeuvel, A. (1990). Cohabitation: A precursor to marriage or an alternative to being single. Population and Development Review, 16(4), 703–726. Rothblum, E. (2005). Same-sex marriage and legalized relationships: I do, or do I? Journal of GLBT Family Studies, 1(1), 21–31. Schechter, E., Tracy, A., Page, K., & Luong, G. (2005). Doing marriage: Same-sex relationship dynamics in the post-legalization period, in Same-Sex Marriage Study Group, ‘‘What I did for love, or benefits, or…: same-sex marriage in Massachusetts’’. Wellesley Centers for Women, Paper No. 422, 2005, pp. 17–25. Smock, P. J. (2000). Cohabitation in the United States: An appraisal of research themes, findings, and implications. Annual Review of Sociology, 26, 1–20. Smock, P. J., Manning, W., & Porter, M. (2005). Everything’s there except money: How money shapes decisions to marry among cohabitors. Journal of Marriage and Family, 67(3), 680–696. Solomon, S. E., Rothblum, E. D., & Balsam, K. F. (2003). Money, housework, sex, and conflict; same-sex couples in civil unions, those not in civil unions, and heterosexual married siblings. Journal of Family Psychology, 18, 275–286. Stiers, G. (1999). From this day forward: Commitment, marriage, and family in lesbian and gay relationships. New York: St. Martin’s Griffin. Waaldijk, K. (2001). Small change; How the road to same-sex marriage got paved in the Netherlands. In R. Wintemute & M. Andenaes (Eds.), Legal recognition of same-sex partnerships: A study of national, European and international law (pp. 437–464). Oxford: Hart Publishing. 123