Home-ownership among opposite- and same-sex couples in the US

Abstract

This paper explores the home-ownership implications of legal issues pertaining to marital status and of discrimination based on sexual orientation or marital status using United States 2000 Decennial Census data. Interesting differences are found between couple types in the effects of several variables. Same-sex couples are more likely than unmarried opposite-sex couples to own rather than rent homes but less likely to do so than married couples. In particular, the effects of household income and of a black householder are smaller for married couples than for same-sex and unmarried opposite-sex couples. Also, same-sex couples are not more likely to own homes in center city areas than elsewhere; married couples, however, are less likely to own homes in the city. Among high-income same-sex households, women are more likely than men to own homes but less likely to do so in US center city areas.

Keywords:

• Home-ownership
• same-sex couples
• unmarried couples
• JEL Codes: D1, D

Acknowledgments

The author thanks the Associate Editor and three anonymous referees for their helpful comments.

Notes

¹ Because of these federal restrictions and because the Census Bureau re-categorizes any mention of “spouse” and “marriage” for same-sex partners, for the purposes of this paper, I refer to married opposite-sex couples simply as married couples. This phraseology should not be interpreted as opposition to legal marital status for same-sex couples.

² For discussion of first-time living arrangements and the home-ownership status of young adults after leaving the parental home, see Clara Mulder (2003). For analysis of home-ownership and housing behavior of the elderly, see Isaac F. Megbolugbe, Jarjisu Sa-Aadu, and James D. Shilling (1997).

³ However, if the property is sold, each spouse's share of the gain is taxable. The share of the gain or loss is determined by state law governing ownership of property.

⁴ The average duration of marriages ending in divorce is fourteen years in Canada (Statistics Canada 2004) and in Scotland (General Register Office for Scotland 2005). In the US, the average duration of first marriages ending in divorce is eight years (Rose M. Kreider 2005). For Scottish unmarried opposite-sex couples, the median duration of relationship is two to three years (Fran Wasoff and Claudia Martin 2004). At a meeting of the Scottish Parliament Equal Opportunities Committee (2003), it was reported that “the average length of same-sex relationship[s] was similar to the average length of relationship reported by heterosexual couples of similar age groups. For people between 18 and 25, relationships might last for 18 months to two years … . as people get to 25 and 35 … they become more settled and established.”

⁵ In cases in which one member of the couple already owns a home when the second member joins the household, the dissolution of the couple's relationship is less likely to end home-ownership of the original member. However, the income of a partner makes ownership easier to continue, and the less likely the dissolution of the relationship is, the more secure the couple's ownership position. Adding a partner to the title of a home, however, may impose costs. Besides fees associated with paper work and attorneys, federal and state laws may view the change as a partial transfer of assets subject to taxation in the case of same-sex couples and unmarried opposite-sex couples.

⁶ Other couples may have postponed marriage because of insufficient income, in which case they are likely to have delayed home-ownership as well. The Earned Income Tax Credit (EITC) in the US federal tax code may have discouraged some unmarried couples from marriage. If a low-income parent marries another low-income individual, their total family earnings may be pushed just high enough to make them ineligible for the EITC.

⁷ Differences in home-ownership may also be influenced by the non-random selection of cohabitating and non-cohabitating couples. People who choose to live with other individuals in an intimate relationship may be more inclined to establish stable living arrangements than those who are not partnered. Based on small samples, Black et al. (2000) found a higher rate of monogamous relationships (although not necessarily cohabitation) among lesbians than among gay men. If similar differences exist in cohabitation rates and in interest in permanent living arrangements, greater home-ownership might be expected among the lesbian population than among the gay male population. However, such differences may not apply when only cohabiters are considered.

⁸ The twenty-five largest metropolitan areas in alphabetical order are:

• Atlanta, GA    Boston, MA    Chicago, IL

  Cincinnati, OH   Cleveland, OH   Dallas, TX

  Denver, CO    Detroit, MI    Houston, TX

  Kansas City, MO/KS  Los Angeles, CA  Miami, FL

  Minneapolis/St. Paul, MN New York, NY   Philadelphia, PA

  Phoenix, AZ    Pittsburgh, PA   Portland, OR

  Sacramento, CA   St. Louis, MO   San Diego, CA

  San Francisco, CA  Seattle, WA    Tampa/St. Petersburg, FL

  Washington, DC

⁹ While the number of multi-racial individuals in the US is increasing, over 90 percent of the householders in this study were either white only, black only, or Asian only. Thus, including white as an explanatory variable resulted in considerable collinearity, and the white variable showed up as insignificant. Therefore, the white variable was omitted.

¹⁰ Ideally, calculations would include a measure of the strength of a couple's relationship. However, the Census data have no such variable. The data even lack a measure of how long the couple has been living together. There are two measures of duration at current residence, YRMOVED and MOB. YRMOVED indicates when the householder moved into the residence but gives no information on the partner. MOB is provided separately for the two partners and indicates whether the person lived in the same residence five years ago or not. However, if the two partners did not live in the current residence five years ago, that would not necessarily imply that they were not together five years ago. Furthermore, owners tend to have longer durations of residence than renters, partly because of large transactions costs associated with buying and selling a home. Hence, a duration-at-current-residence variable would capture primarily the tendency for owners to live in their homes longer than renters and is therefore not used.

¹¹ Sixteen states and Washington, DC were represented in the sample. Laws prohibiting housing discrimination on the basis of sexual orientation were in effect in 2000 in California, Massachusetts, Minnesota, New York, and Washington, DC. States that did not have such a law were Colorado, Florida, Georgia, Illinois, Maryland, Michigan, Missouri, New Jersey, Ohio, Pennsylvania, Virginia, and Washington.

¹² The low incomes of unmarried opposite-sex couples have important implications. As McCarthy, Van Zandt, and Rohe (2001) explain, since lower-income households hold larger shares of their wealth in housing, they are more vulnerable to extreme price movements and housing investment losses. They may be more highly leveraged, and high interest payments and mortgage insurance may substantially reduce their financial benefits. In recessions, these households can be at serious risk of defaulting on their mortgages and losing their homes. McCarthy, Van Zandt, and Rohe (2001), therefore, recommend that policies be implemented to preserve home-ownership. Such policies are likely to be of greater help to unmarried opposite-sex couples because of their low income levels.

¹³ Marginal probabilities are shown in Tables 3 through 6 since they more clearly show the effects of the explanatory variables than logit coefficients do. Without the original coefficients, standard errors are not helpful. Therefore, p-values are provided instead to enable the reader to determine the significance of the coefficients. Supplementary tables containing the coefficients and the standard errors on which the p-values are based are available upon request.

¹⁴ For assessing the within-sample performance of the fitted model, Mars (J. S.) Cramer (1999) recommended a cut-off point to produce a prediction that maximizes the fit of the predicted outcome to the given estimated probability. That cut-off point is α, the proportion of observations in the sample that are home-owners. Since home-ownership propensities differ by couple type and by age and income group, different cut-off points α_i were used for each of the twelve subgroups.