Max-Planck-Institut für demografische Forschung Max Planck Institute for Demographic Research Konrad-Zuse-Strasse 1 · D-18057 Rostock · GERMANY Tel +49 (0) 3 81 20 81 - 0; Fax +49 (0) 3 81 20 81 - 202; http://www.demogr.mpg.de MPIDR WORKING PAPER WP 2004-018 JUNE 2004 The Demographics of Same-Sex „Marriages" in Norway and Sweden Gunnar Andersson ([email protected]) Turid Noack ([email protected]) Ane Seierstad ([email protected]) Harald Weedon-Fekjær ([email protected]) This working paper has been approved for release by: Andres Vikat ([email protected]) Deputy Head of the Laboratory of Contemporary European Fertility and Family Dynamics. © Copyright is held by the authors. Working papers of the Max Planck Institute for Demographic Research receive only limited review. Views or opinions expressed in working papers are attributable to the authors and do not necessarily reflect those of the Institute. The Demographics of Same-Sex „Marriages“ in Norway and Sweden by Gunnar Andersson 1, Turid Noack 2, Ane Seierstad 2, and Harald Weedon-Fekjær 3 May 21, 2004 Abstract: The present study provides an investigation of the demographics of same-sex marriages, or registered partnerships (“registrerade partnerskap”), in Norway and Sweden. We give an overview of the demographic characteristics of such spouses, and study patterns in divorce risks. A comparison with similar dynamics of heterosexual marriages is provided. Our study is based on longitudinal information from the population registers of the two countries covering all persons in partnerships. Our demographic analyses involve information on characteristics such as age, sex, geographical background, experience of previous opposite-sex marriage, biological parenthood, and educational attainment of the partners involved. We find that in many respects the distributions of married populations over these characteristics differ by the sex composition of couples. Patterns in divorce risks are quite similar in same-sex and opposite-sex marriages, but divorce-risk levels are considerably higher in same- sex partnerships. The divorce risk in female partnerships is practically double that of the risk in partnerships of men. 1 Max Planck Institute for Demographic Research, Konrad-Zuse-Str. 1, 18057 Rostock, Germany. E-mail: [email protected] 2 Statistics Norway, POB 8131 Dep, 0033 Oslo, Norway. E-mail: [email protected] and [email protected] 3 Section of Medical Statistics, University of Oslo, Norway. E-mail: [email protected] 1. Registered partnerships: A new family type At present, the issue of granting legal recognition to same-sex couples is high on the political agenda in a large number of countries. In places where such a family type is not recognized, the debate tends to be intensifying. In many countries in Europe, it is already well established, and the discussion then more often concerns various amendments to existing rules. The first country at all to introduce a legal recognition of same-sex unions was Denmark in 1989, and the term “registered partnership” was invented for that purpose. In all Nordic countries, same-sex couples today have the possibility to contract a registered partnership, a civil status that in practice is not much short of a marriage. Such a family type was in the second place introduced in Norway in 1993, subsequently in Sweden in 1995, Iceland in 1996, and, finally, in Finland in 2002. By 2003, same-sex unions had been given legal recognition in one form or another also in Germany, France, Hungary, Portugal, Belgium, and the Netherlands1. In 2001, the latter country became the first in the world to amend its marriage act to give couples of the same sex admission to marry in the same manner as opposite-sex couples. In terms of innovation in family-demographic behavior, the Scandinavian countries are often singled out as forerunners, which other countries subsequently tend to follow in behavior. It might be debatable whether this really is true in a more general sense, but in the case of same-sex partnerships this certainly seems to be a correct description. Consequently, it might be worthwhile to have a closer look at the Nordic experience of same-sex family life. Several studies deal with the various political and legal aspects of the introduction of same-sex partnerships in Europe2. There is, however, still sparse knowledge about the demographic behavior that is related to this new family type. The purpose of our study is to provide some knowledge of that kind. Our study provides an overview of demographic characteristics and patterns in divorce risks of couples in registered partnerships in Norway and Sweden. The 1 In some further countries, like the USA, Canada, and Spain, same-sex unions had sometimes been legalised at the level of states and regions. 2 For a discussion of the passage of the partnership legislation in Denmark, see Søland (1998). Nielsen (1990) provides further evidence of legal aspects of the new family type. Noack (2000) discusses the introduction of registered partnerships in Norway, and Agell (1998) refers to the debate about the introduction of partnerships in Sweden. Martin and Théry (2001) discuss the introduction of another related family form, PACS, in France, which is open for same-sex and opposite-sex couples alike. For an overview of how the way to same-sex marriage got paved in the Netherlands, see Waaldijk (2001). 2 analysis is based on information from Norwegian and Swedish population registers. For our purpose, we have managed to link information on various demographic and socio- economic characteristics of the same individuals from different other administrative registers. The study is an extension of previous work based on Norwegian data, where we, for example, found that the majority of partnerships were male and that the fraction of cross-national partnerships was fairly high (Noack 2000). A first analysis of divorce risk in same-sex partnerships showed that in Norway, lesbian couples had a considerable higher divorce risk than male couples. Another group with a high propensity to divorce were cross-cultural couples, i.e., couples in which one of the two partners was non-Nordic (Noack, Fekjær and Seierstad 2002). In the present study, we provide an elaborate comparison including similar data on partnerships in neighboring Sweden. In addition, we incorporate data on divorce risks of heterosexual married people. Such a thorough comparison of divorce risk patterns in opposite and same-sex marriages has thus far never been performed. The reason is, of course, that the legalization of same-sex partnerships is a recent development, and that the time available for observation has been brief. In our study, we thus manage to compare patterns and demographic behavior of a clearly defined total population of “married” same-sex couples to an equally defined population of opposite-sex couples. 2. Family dynamics of gays and lesbians: Previous research During the last decades family patterns of many countries have become more diverse. Although small in numbers and far from being accepted in most countries, legalization of same-sex marriages fits neatly into this development. The increasing diversity is often regarded as a part of a larger cultural change, implying an increase in freedom as well as an obligation for individuals to decide how to organize their lives in an individualized society (Beck and Beck-Gernsheim 1995). Another factor that might have paved the way for same-sex marriages is the increasing separation between reproduction and sexuality, in favor of a more plastic sexuality in the terminology of Giddens (1992). Sexuality has naturally always been separated from reproduction in homosexual relations, and this separation is becoming increasingly dominant also in heterosexual relationships. Thus, the disparity between homo- and heterosexual relationships is being diminished. The increasing acceptance 3 and legal legitimacy of homosexual practice may be the most important change regarding sexuality in the last decades, or as Giddens (1992:33) expressed it “... sexual diversity, although still regarded by many hostile groups as perversion, has moved out of Freud’s case-history notebooks into the everyday social world”. Moxness (1993), a Norwegian sociologist, has argued that same-sex marriages have become legalized not so much because homosexuality has become more accepted, but because marriage has become an increasingly empty institution and no longer is seen as a mandatory entrance to adult life, sexual life, and parenthood. New patterns of family life calls for new topics of research, and recent years have witnessed an increase in research on lesbian and gay lifestyles, and on same-sex families. Although the literature about same-sex relationships is abundant, most of it does not allow for the deduction of any firm demographic hypotheses. Many studies are based on small number of individuals. They have given interesting but often anecdotal information. Large-scale quantitative studies are rare. Many studies face serious problems related to sampling or representativity. In recent years, however, there has been an increasing recognition of the need to deal with these problems. As a result, more solid demographic studies have indeed appeared (Black et al. 2000). 2a. General problems in studying gays and lesbians Lack of representative samples is the most fundamental problem in quantitative studies on gays and lesbians. Self-recruited samples from an unknown population have been and still are very common in studies of homosexuals. Respondents are, for example, recruited by snowball methods, from the readers of particular magazines, from members of organizations for gays and lesbians, or more recently using those who are willing to fill in questionnaires presented at the Internet. Critical voices have also pointed out that much of the research on family life of gays and lesbians is done by studying white, well- educated, American middle-class people (Patterson 2000). In addition to such sampling problems, the question of how to identify homosexual people is increasingly debated. Should respondents be asked to self-identify themselves, or is it better to measure sexual practice, i.e., to ask about number of life- time same-sex partners, any such partner within a certain time period, the sex of the majority of partners, and so on? (Black et al. 2000). According to large-scale population studies carried out in the US, the proportion of men having had a male sex partner in a 4 last previous year is about 1-3 percent, as compared to 4-9 percent having had at least one male partner in the life time (Spira et al. 1993; Lauman et al. 1994; Black et al. 2000). The proportions of women having had a partner of the same sex are somewhat lower, well over 1 percent and about 4 percent, respectively. A different pattern is reported from a Norwegian study. In this survey, the proportions of respondents aged 19- 26 were slightly higher for women than for men when it concerns same-sex experience during the last 12 months as well as during life time (Pedersen and Kristiansen 2003:11). All the estimates referred to above are well below the often mentioned 10 percent benchmark of the famous report of Alfred C. Kinsey. This estimate however seems to be a misinterpretation of what Kinsey in fact had said (Sandfort et al. 2000). Kinsey’s study was based on information about life-time homosexual activity as well as homosexual desire, resulting in different levels of estimates. Notwithstanding, Kinsey’s sampling procedure also had its weaknesses. Not only the methodology, but also the view that individuals may be divided into gays, lesbians, bisexuals, and heterosexuals has met increasing criticism. The possibility that sexual identities may shift over time has attracted increasing attention (Patterson 2000). So far, little research may document such contentions. A recently published study of younger Norwegians finds, however, some signs of a confluent sexual culture, and more so among women than among men (Pedersen and Kristiansen 2003). To give a statistical portrait of any gay and lesbian population using traditional population surveys has also been considered difficult because of the mere size of the target groups. Or put another way, in standard demographic data sources, it may seem like looking for the needle in the haystack. In addition, the underlying assumption of most demographic surveys is heterosexual, and respondents often have no possibility to report on other types of family behavior than those suggested by the survey designers (Hoem et al. 2000: 87). The seemingly sensitive character of the topic has probably also made it difficult to include it in questionnaires where it otherwise might had appeared natural. Nevertheless, a number of existing data sources today allow for research on same-sex couples as defined by any co-residence of two persons of the same sex. 2b. Same-sex couples and same-sex co-residence For the United States, Black et al. (2000) have made a critical review and comparison of three sources available for systematic studies of the gay and lesbian populations: The 5 General Social Survey, the National Health and Social Life Survey, and the 1990 U.S. Census. Although documenting a number of measurement-error problems in the surveys and a considerable underreporting of same-sex couples in the census they conclude that the data sets seem good enough to allow for credible analyses of gays and lesbians in the US. Based on these data, they compare partnered gays and lesbians with the general population. They find that lesbians as well as gays have attained more education than married and non-married heterosexual partnered women and men. Partnered gays earn, however, less than men living in opposite-sex marriages. For women, the opposite is the case, partnered lesbians earn more than married women. These results appear when the comparisons are made between persons within similar age and educational categories. They conform to a related study by Black et al. (2001) that also included non-partnered individuals. A related study for the Netherlands, however, shows only negligible effects of sexual orientation on earnings (Plug and Berkhout 2004). For further research on the economic lives of lesbians and gay men see Badgett (1997, 2001). In addition, the US Census data indicates that 5 percent of male couples and nearly 22 percent of female couples live with children in the household. Although adoption and artificial insemination for lesbians and gays frequently are reported in the media, Black et al. (2000) conclude that most of the children of partnered gays and lesbians recorded in the census probably have been born while the parents lived in a previous opposite-sex marriage. 20 percent of partnered gays and 30 percent of partnered lesbians were previously married. The data also gives information on patterns in geographical settlement. Gay men seem to be concentrated to a selected number of urban areas, preferably big cities. Lesbian women are less concentrated, and more often live in smaller metropolitan areas. For a further discussion on why patterns in geographical concentration of gay men arise, see Black et al. (2002). Conventional demographic data have also been used to study the matching behavior of same-sex couples. Based on the US 1990 Census, Jepsen and Jepsen (2002) find positive assortative mating for four types of couples: married and cohabiting opposite-sex couples and male and female same-sex couples. Same-sex couples appeared more alike in their labor-market characteristics than did opposite-sex couples, while the opposite was the case for various non-labor-market traits. Evidently, census data that include information on household characteristics of surveyed individuals allow for the study of co-residing couples of the same sex. However, such data are not non-problematic; same-sex co-residential individuals have 6 not to be synonymous with gay and lesbian couples3 (Voon Chin Phua and Kaufman 1999). Such problems with ambiguity of data also appear when heterosexual cohabitation is studied. Co-residing persons of the opposite sex does not necessarily have to be sexual partners (Baughman et al. 2002). 2c. Family dynamics in same-sex marriages as compared to opposite-sex marriages A main purpose of our study is to provide information on the family dynamics in same- sex marriages as it can be measured in the manner of partnership-dissolution risks. In this respect we have not much of previous research to rely on. An overview of recent research on the family relationships of gays and lesbians by Patterson (2000) gives moderate information on the stability of gay and lesbian relationships. The study of duration of relationships typically requires a panel design or highly reliable retrospective data. So far, such data have been hard to establish for an appropriate study of couple dynamics of gays and lesbians4. Patterson (2000) concludes, however, that it seems reasonable to believe that some of the problems in homosexual relationships will stem from the same roots as problems experienced by opposite-sex couples. By contrast, the literature on divorce of heterosexual married couples is abundant. Considering the impact of various demographic variables, studies of such couples indicate that pairing at a very young age, low socio-economic status, low education, a considerable age difference between the spouses as well as socio-cultural differences are important risk factors for divorce (Clarke and Berrington 1999; Sayer and Bianchi 2000). For some of these factors, however, like that of a high risk for spouses with little formal education and for those in manual-worker occupation, the elevated divorce risk might decrease with the duration of marriage (Jalovaara 2002). 3 A recent German large-scale data source that includes information on same-sex and opposite-sex couples alike seems to be more precise in these aspects. The German Mikrozenus includes information on co-residence and also asks respondents to specify if they consider themselves living in a “Gleichgeschlechtliche Lebensgemeinschaft” (same-sex union) or any other type of family. Such self- identified same-sex couples are much fewer than the total of co-residing same-sex couples. Eggen (2002) suspects that problems connected with self-identification results in underreporting, and assumes that any “true” level of same-sex cohabitation in Germany would lie somewhere in between the numbers arising from the two possible definitions. 4 Kurdeck (1992, 1995) provides a study on the stability of gay and lesbian couples in the US. However it is based on such tiny data that it hardly offers any possibility to make generalizations to a wider population of gays and lesbians. 7 3. Data and methods The object of our study is registered partnerships in Norway and Sweden. Such a civil- status type has nearly the same legal consequences as a marriage. This means that registered partners have the same rights and duties as married heterosexual couples in relation to each other and to society. The acts are basically the same in all Nordic countries, but differ in the opportunity to adopt children, to have artificial insemination and to solemnize the partnership5. This being said, the legal rights and duties connected to marriage are less critical in Scandinavia than in other countries. (For an overview on family law and the consequences of marriage in countries in Europe, see Hamilton and Perry, 2002). In the context of the Nordic welfare state, social rights are largely based on individuals, regardless of their family status. Economic motives that may be important for marriage in the US, like those of the possibility of a common health- insurance coverage, are virtually non-existent in the universalistic welfare state. The data for our calculations on partnership dynamics are derived from the population-register systems of Norway and Sweden, which with a high degree of accuracy cover the populations of the two countries and their recordable vital events. Each change in civil status is recorded in the registers, and since each individual living in one of the two countries has a unique personal identity code we have been able to derive longitudinal histories of the family dynamics of each person who has ever registered a partnership formation in any of the two countries. Similar event histories can be collected for individuals who have married heterosexually, and we have managed to include such data for Sweden6. This allows for a proper comparison of our populations of same-sex partnerships with that of an equally defined population of opposite-sex marriages. The populations are defined by their civil status; there is no ambiguity in the categories we use. Individuals who have never lived in any of the two countries cannot be traced directly in the registers and some partnerships that 5 Churches are not available for ceremonies of partnership formation. In Norway the actual registration is performed by a Notarius Publicus, in Sweden by a court or a private person with special authorization. Medical assisted insemination is not given to women living in registered partnership either in Norway or Sweden. From 2003 registered partners in Sweden got the admission to jointly adopt children, including all types of international adoption. In Norway only admission to adopt the other partner's child is given (Waaldijk 2003). 8 involve persons living abroad cannot be incorporated properly into our analyses. In the case of Sweden, we had to exclude 100 same-sex couples from our analyses since we had no information at all on one of the two partners involved. The first part of our analysis is descriptive, where we display various demographic characteristics of individuals who have formed a partnership in Norway or Sweden. These characteristics are derived from various administrative registers and are measured at the time of partnership formation. We have defined our variables so that they give the characteristics at the couple level. Our demographic description involve information on characteristics such as age, sex, geographical background, experience of previous opposite-sex marriage, biological parenthood, and educational attainment of the partners involved. Our variables are defined as follows. We depict the age composition of persons registering a partnership by giving the mean age of the two partners at the time of registration. The distribution is given over the categories “mean age 30 or less”, “mean age 31-40”, and “mean age 41 or above”. In addition, we give the distribution over various categories of the age difference between the two partners involved. For both countries, we describe what fraction of partnerships that involve at least one person living in the capital area at the time of partnership formation7. For Norway, this is the City of Oslo, while for Sweden, we use the greater Stockholm metropolitan area as our geographical demarcation. We further describe the geographical background of the partners by giving the distribution over various national origins. We distinguish between couples where both partners are locals, and couples where at least one of the partners comes from abroad. In Norway, national origin is measured by citizenship at the time of partnership formation. In Sweden, it is instead measured by country of birth. We report on couples where at least one partner comes from another Nordic country, another European country (including the overseas Anglo-Saxon countries), a non-European country, or where the national 6 The data cover marriages contracted in 1993-1999, Swedish partnerships contracted in 1995-2002, and Norwegian partnerships contracted in 1993-2001. The minor discrepancy in the observation period of marriages as compared to that of registered partnerships in Sweden is due to data availability. 7 Most partners are likely to live together at the time of partnership formation, but need not necessarily be registered (yet) as living at the same address. In our data for Sweden, we found that about half of the partners involved had been registered as living together at the same address already for a period of at least two years prior to their partnership registration. 9