Gay money

• Home
• Basic facts
• Full findings
• Marketing claims
• References
• Blog
• For journalists
• About

Full findings

What does the research really say about lesbian and gay incomes and earning? What are the facts about lesbian and gay economics?

New for 2017

1. Data from the U.S. Department of the Treasury show that gay marriage in the U.S. is a great thing to have if you’re already doing well. Details ☞

2. Two studies credibly show gay males earning more than hetero males

General statement of the findings

There are two ways you can describe the findings of most of the research on the topic:

• Weak statement: Neither gay males nor lesbians earn more than straight people.
• Strong statement: Gay males earn less than straight males, often much less. Meanwhile, lesbians earn more than straight females.

Both statements are the exact opposite of marketer’s claims that “the gay market” (less often, “the gay and lesbian market”) is “lucrative” or simply rich.

In case you aren’t clear on what I’m saying: When marketers tell you the gay market is a rich or lucrative one, they don’t have any evidence to back it up. The evidence shows the opposite – at least for gay males, the group they’re really talking about. Put harshly, they’re lying.

Things to remember about this project

• I read essentially all the research papers and most of the books on gay money. But not all of them were worth summarizing or excerpting from.

• Not all findings are statistically significant. That means not all findings are sure things; they could have happened by chance. I rarely report on when a finding is or is not statistically significant because this is not a peer-reviewed article on econometrics. A finding might turn out to be totally false, but it’s still a finding, and if I’m reporting it here it’s because it is of interest, not because it’s ironclad and provable. You can read individual papers to learn which figures are and are not statistically significant.

• All data come from U.S. sources unless otherwise specified.

Download a summary of findings

I have a summary of papers that arrived at quantitative findings available as Excel and OpenOffice spreadsheets.

General explanations

After reading all the research, the truth about gay money boils down to a few facts.

• Gay males work fewer hours than straight males. Lesbians work longer hours than straight females.

• Gay males often work in lower-wage, often female-dominated, jobs. The converse is true for lesbians, who often work in high-wage, typically-male-dominated jobs.

• Gay males and lesbians have kids less often than straight people do, though it’s by no means rare. It’s less likely that one partner will stay at home with the kids like a ’50s housewife.

• The group that earns the most money is married hetero males. They work the longest hours and are generally paid more by employers.

One table of numbers that sums up the phenomenon

If you’re the kind of person who likes to look at numbers, let me show you some. Blandford 2003 (p. 633) has an excellent table that, while deriving solely from that paper’s research, does a great job of summarizing the trends in lesbian and gay incomes. Remember, this is just one example, but it tells you a lot about the patterns found by most researchers. The highest percentages in each relevant column are emphasized.

Income distribution patterns for full-time workers,
by gender, sexual orientation, and marital status

Income category	Male workers				Female workers				All workers
	Heterosexual		Gay/ bisexual	All men	Heterosexual		Lesbian/ bisexual	All women
	Married	Unmarried			Married	Unmarried
≤$13,999	8.9%	19.3%	19.4%	12.7%	28.6%	27.4%	22.0%	27.9%	20.0%
$14,000 to $21,249	13.6%	17.8%	24.1%	15.5%	24.3%	25.9%	18.0%	24.9%	20.0%
$21,250 to $28,249	17.7%	21.0%	16.2%	18.8%	20.8%	21.2%	22.0%	22.0%	19.9%
$28,250 to $39,999	24.8%	24.0%	17.7%	24.3%	16.3%	17.0%	20.0%	16.7%	20.4%
≥$40,000	35.0%	17.9%	22.6%	28.7%	10.0%	8.5%	18.0%	9.5%	19.7%
Wealthiest 5%: ≥$63,325	11.9%	5.0%	3.2%	9.3%	2.3%	.7%	4.0%	1.6%	5.0%
Median income (1992 dollars)	$31,700	$25,042	$24,212	$28,937	$20,216	$20,216	$23,807	$20,216	$24,618

Singles as opposed to couples

Most studies use data on gay or lesbian couples, and most of those studies use some kind of inference to determine that people actually are gay or lesbian and/or that they actually are couples. As such, most of the research could be completely wrong in several different ways – people deemed gay might not be, people deemed in a partnership might be single, and everybody who’s single is left out completely.

But some papers have isolated gay and lesbian singles, i.e., gay males or lesbians who do not have partners of the same sex. (It is conceptually easier to define “being single” than “having a partner,” but definitively identifying the singles in large data sets turns out to be harder.)

• Carpenter 2007 used a data set that allowed him to restrict his examination to males with a number of family members equal to 1. This gave Carpenter the rare ability to isolate single gay and straight males inside a statistically valid set of data.

  Consistent with the general finding, though, is higher education for gay males (very much more common to have a college degree). Here, “the level of average income for men reporting same-sex sexual behaviour is slightly higher than for other men.”

  • Average income, nongay: $22,782

  • Average income, gay: $23,949 ($1,167 more)

  These are not strictly accurate numbers because Carpenter had to draw some inferences from gross household income. Respondents were not reporting their individual earnings to the dollar.

  But when Carpenter took other factors into account, like never having been married, the data in fact showed an income penalty of 30%–37%. This is one of those cases where the raw data do not tell the whole story; you have to control for contributing factors.

• Badgett 2008b found that singles had much lower individual incomes than people in partnerships:

  • Single: $38,343

  • Partnerships of all kinds: $44,705

  • Partners not cohabiting: $39,152

  • Partners cohabiting: $44,367

  • Registered domestic partners: $48,821

• “Unattached lesbians earn about 10% less than lesbians with partners, and the difference becomes, if anything, more pronounced and more significant as additional controls are added” (Lafrance 2009).

Hours worked

Gay males work fewer hours than straight males, in turn partially explaining why they earn less. Various studies show results similar to Tebaldi 2006, so I’ll just quote that paper: “After controlling for other factors, the labour-supply analysis provides evidence that lesbian women supply, on average, 7% more hours of work per week than heterosexual married women do” (and more than unmarried hetero women). “[G]ay men work about 8% fewer hours per week than married men and about 6% fewer hours than heterosexual men who cohabitate with a woman.”

Is there a marriage premium?

Economics research has found that, in general, married men earn more than unmarried men (and in fact more than everyone else).

Now, why is that? There are some accepted theories.

• Even today, many people, including economists, assume that wives won’t work (because they’re at home looking after children). Or, if they do work, they don’t earn very much. Hence employers pay married men more on the assumption they have to support a family. They pay married women, or women they predict will marry, less because their loyalties will be divided.

• Men who intend to marry or who get married do the following to support their families: Amass greater education; and/or work higher-paying jobs; and/or work long hours.

Cases in point:

• Allegretto 2001 finds the marriage premium, all told, is about 14% for married men over gay-male partners.

• Black 2003 found no marriage premium for women, but a hefty one (+20%) for men.

• Carpenter 2005 found a premium for all married men. In the United States at the time of the study, Carpenter notes that “heterosexual marriage signals to employers that an individual is not gay,” which makes it easier for an employer to identify which employees to pay more. (This is on-the-job discrimination, of course, but we’re just describing it, not excusing it.)

• We really are talking about a premium for married hetero men, not a premium just for gay-male partners as compared to single gay males. Zavodny 2008 states that her “results provide little evidence that cohabiting gay men earn a ‘marriage premium’ analogous to that earned by married men.”

Balance between economic and non-economic work

The work you do can still be “economic,” and represent an economic choice, even if no money changes hands. This may seem counterintuitive at first, but it really is true. Housework and childrearing in the home are both economic choices, even though most homemakers don’t get paid. As such, those tasks are examples of home production. Meanwhile, working a job for wages is an example of market production.

Research on gay and lesbian economics, like other econometrics, has a hard time putting a dollar value on home production because it doesn’t actually involve dollars. But it’s still part of the whole picture.

• A theory often articulated in the research holds that gay males and lesbians realize certain truths about themselves early on and make choices accordingly. As Black 2003 describes it, “Lesbian women who realize early in life that they will not marry will invest more heavily in market-oriented human capital than heterosexual women do. They will subsequently be observed to earn more than other women.” (That means education and training.) But the decisions gay men make are not analogous to lesbians’: “Single men and men in gay relationships are predicted to earn less than married men to the extent that they specialize less intensely in market production than do married heterosexual men.” Thus gay males work less.

  Hence single straight males and coupled gay males realize they won’t have a wife and kids to support, so they have an incentive not to work longer hours or make other sacrifices on the job to earn higher pay.

• Black 2007 notes that “Gay and lesbian couples are less likely than their heterosexual counterparts to have a stay-at-home partner.... For gays, lesbians, and heterosexuals alike, the stay-at-home partner generally has the lower level of formal education.... Among households where both partners work, the gap between individuals’ income and hours worked is highest among ‘traditional’ heterosexual couples, by which we mean couples for whom the male is the highe[r] earner.”

Choice of occupation

Most studies that address this topic find that gay males are overrepresented in female-dominated occupations and underrepresented in traditionally male, often much-higher-paying, occupations. The converse is true for lesbians, who are found less often in female-dominated occupations and more often in male-dominated ones.

One paper (Black 2003) found no real difference in occupational categories between gay/lesbian and straight populations. A later paper (Black 2007) found that “gay men are in occupations that are more ‘typically female’ than other men while lesbian women are in occupations that are less ‘typically female’ than other women.”

These are noncontroversial findings when we think of our own experience. Almost no research makes the following observations, but gay males are generally uninterested in working on the factory floor or as a truck driver. Lesbians tend to have little or no interest in working as secretaries or hairdressers. But those AIDS hospices and rape crisis centres aren’t going to run themselves, now, are they? As Badgett 1997b plainly states: “Lesbians and gay men” – presumably in that order – “who have been politically motivated by the experience of prejudice, discrimination, and harassment may choose to work in nonprofit organizations dedicated to social change, where we also find women and ethnic minorities overrepresented.” (But Plug 2004 writes, working with Dutch data: “Do heterosexual and homosexual workers differ in their preferences for leisure and paid work? Or do they differ in their taste for public‑ versus private-sector jobs? We do not know.”)

• If we assume, based on our life experience, that tastes in activities are formed early on, it is no surprise that Carpenter 2007b found that expected tastes in activities carry through to college. “[S]tereotypes about gay men preferring the theat[re] and lesbians preferring athletics may have basis in fact.... These activity choices may also be related to future career prospects.” He found higher interest in arts and politics for gay males and for lesbians, but lesbians rate higher in interest in athletics. (Bisexual females disagree with lesbians just on that last point.)

• Antecol 2008 offers figures that are the easiest to understand. Gay males are overrepresented in jobs where less than half the employees are male, while lesbians are overrepresented in jobs where more than 60% of employees are male. “Specifically, gay men were less likely to be in the following occupations: Protective, transportation, architecture/engineering, installation/repair, and construction/extraction.... Interestingly, lesbian women were less likely to be in certain ‘pink-collar’ occupations, including office administration and sales.”

  For readers concerned with the fact the computer industry is overrun with antisocial male geeks, gay males and lesbians were in fact overrepresented in “computer/math” occupations in this study.

• Carpenter 2005 found gay males almost nonexistent in the fields of “precision craft repair, machine operator, material/equipment mover, and handler/equipment cleaner” jobs (about 5% versus about 25% for straight males). But gay males were overrepresented in “executive/administration/management” (21% versus 10%–14% for straight males). When males work in “administrative support” at all, they are usually gay (13% versus 4%–6%).

  Lesbians were almost equally likely to be in blue-collar trades and in the executive as straight females. The only discipline where lesbians are significantly less common than straight females is administrative support (8% versus about 17%).

• Berkhout 2008: “Our results are most consistent with the notion that, among men, gay/bisexual workers specialize less in marketable skills.... [Certain observed] gay/bisexual workers have invested more in their feminine traits, displayed by greater verbal ability and lower mathematical ability, and were more likely to graduate in typically female (and less [financially] rewarding) fields of study.”

• Clain 2001 found not only that lesbians are more likely to be in blue-collar occupations but that partnerless women are, too, suggesting that being a straight female with a partner means you avoid blue-collar work. The same study found gay males to be employed in blue-collar trades about one-third as often as partnered straight males and just under half as often as single straight males.

  [G]ay men and lesbians in highly-paid positions might be especially careful to avoid disclosure of their homosexuality without steering away from such positions altogether.... Because of the importance of the workplace environment, we expect that lesbians and gay men may enter less-financially-rewarding occupations than they would if they were straight, trading income and status for a job in which it is easier to pass or where the penalties for disclosure are relatively low, including jobs traditionally associated with lesbians and gay men or concentrated in areas with large gay communities.

  Queer jobs feel safer even if they pay worse. What Clain does not explore is the possibility that such fears are based on nothing and gay employees could actually have come out without problems. Econometrics is not the right science to evaluate that hypothesis, of course. She continues:

  However, we believe that losing a relatively-poorly-paid position is in fact likely to exact a relatively greater cost than losing a well-paid job. People in well-paid jobs generally have more options when looking for work and may be expected to have more financial security to weather a period of unemployment and job search. A lawyer who is fired can work as a waitress [sic], but a fired waitress cannot work as an attorney, and lawyers usually have more physical, social, and financial resources than waitresses.

  (I am not sure what she means by “physical” resources.)

• Allegretto 2001 finds, in effect, that the occupations most likely to be chosen by gay males have the largest wage penalties compared to married hetero men. Blandford 2003 finds something more or less similar:

  The disproportionately heavy concentration of openly gay and bisexual workers in the executive, administrative, or managerial positions and in the professional specialty category – occupational sectors typically associated with higher average wages – might lead one to expect higher average incomes for these men than for the group of men as a whole, holding all else equal.

  There is evidence, however, that these workers were situated in the more-poorly-remunerated occupations within the higher-paying occupational categories. For example... nearly two-thirds of the openly gay and bisexual men within the professional specialty category resided in jobs that were female-identified or in the arts.

When kids are present

• Black 2000 found that gay-male partners had one child 3% of the time compared to 18% of partnered heteros and 22% of married heteros. (There were other figures for two children and for three or more children.) For lesbians, the numbers are 12%. Most surveys show magnitudes in line with these. If you want very rough estimates, lesbians have kids less than half as often as straight couples do. Gay males have kids less than half as often as lesbians do. Again, these are approximate numbers that do not derive specifically from Black 2000; they’re useful just so you can get a grips on the general magnitudes involved.

• In a related issue, lesbians are much less likely to drop out of the workforce (e.g., because they plan to have kids). Labour-market attachment is much stronger for lesbians than straight women, Black 2007 found.

• Lesbians have kids about one-third less often than married women, according to Jepsen 2007, and about half again more often than cohabiting women.

  Black 2003 notes that “lesbian women may have had a stronger labour-market commitment, and for any given level of potential experience may have more actual labour-market experience” just by virtue of having not dropped out of the workforce to have kids.

• Black 2003 is also the only paper to find no significant earnings differences between gay males and lesbians all of whom have children. That finding also included controls for occupations, so it is not a straightforward conclusion that having kids nullifies the wage penalty for gay males and also nullifies the wage premium for lesbians.

• Black 2007 found that gay couples almost never have children (90.3% do not). Lesbians usually don’t, either (77.9% do not). Straight couples usually do have children (only 38% don’t). “Costs of children are higher for gay and lesbian couples than for their heterosexual counterparts.”

Rich gay, poor gay

A paper with a disturbing title to match its findings – “Poverty Among Cohabiting Gay and Lesbian and Married and Cohabiting Heterosexual Families” (Prokos 2010) – looked at two-parent families with kids. The authors developed a ratio of income to poverty. How many of the different kind of families had the worst ratio, i.e., were in absolute poverty?

• Married heteros: 6% were in absolute poverty

• Two gay males: 12%

• Two lesbians: 12%

• Cohabiting straight couples (who “tend to be much younger”): 16%

“[L]esbian couples’ higher levels of education account for their lower poverty rates relative to cohabiting heterosexual families.” But married people are least likely to be poor, followed by gay and lesbian couples (attributable in some ways to higher education, but all the reasons are not known), then followed by cohabiting straight couples, the poorest of all.

Meanwhile, separate research shows the richest gays suffer the least: For a single stratum of gay and straight workers, gay marketers’ claims that gays are rich has some foundation. But these gays are merely as rich as comparable straight people. Antecol 2008 found that “top-earning gay men earned wages nearly identical to those of their married counterparts.” Even top-earning lesbians had only a small premium over top-earning married women.

Carpenter 2008 found similar results for Canada.

Education

Gays get more education than straights, but only lesbians end up earning more as a result, broadly stated.

Gay males and lesbians, nearly all papers find, have more education (in all definitions, from high school to Ph.D.) than straight people. In this respect the stereotype is correct: Gays do put a premium on education. But it doesn’t always pay. As Allegretto 2001 found, “Unlike most standard wage analyses, in this analysis the ‘disadvantaged’ group has higher levels of education than the ‘advantaged’ group.” That same paper also found huge wage penalties (compared to married hetero men) in sales and admin jobs, where gays outnumber straights. Antecol 2008 found that gay males and lesbians with less than a high-school education had much higher wage penalties than straight people did. Apart from the fact that we prefer education and academia, education really does pay.

One reading of these findings is as follows: We sign up for a lot of education but work low-end jobs where we’re paid less.

Rothblum 2007 hypothesized that “heterosexuals prioritize income (especially if they have children)” while gays and lesbians “may feel a greater sense of responsibility in helping to create a more just society and therefore may be more likely to have jobs explicitly aimed at creating social change or in helping the underserved.” That paper also suggests that heterosexuals have better networking options and can wrangle better or more conventional jobs that way.

• Black 2003 considered gay men’s and straight men’s educations compared to their fathers’. Gay males have enormously more education than their dads; straight men have marginally more education (except for “some college,” where it’s noticeably more). The educational backgrounds of the two groups’ dads are about the same, in part because gay sons and straight sons come from the same types of families (i.e., any family). (Carpenter 2007 also studied this issue and found similar results – gay and lesbian students went to college more often than their dads and moms did.)

  This means gays really do put a premium on education. They want more education than their parents had.

• Black 2003 notes that two things should not be true at once for gay males – having so much more education than straight males but so much less income. That paper doesn’t come out and say it, but surely the explanation is the jobs people choose, as explained above.

• Black 2007 found that barely any gay or lesbian partners had less than a high-school education (6.8% gay male, 7.7% lesbian, 10%–12% hetero). But numbers of gay or lesbian partners who have a bachelor’s degree or greater outpaced numbers of straight people with same (about 42% for gays and lesbians with BAs versus about 26% for straight people). Lesbians had a higher number of postgrad degrees than everyone but straight females.

• Peplau 2004: “Although lesbians are unquestionably vulnerable to prejudice and discrimination on the job, they nonetheless are overrepresented among America’s most-highly-educated and better-paid women workers.”

Discrimination

There seems to be an assumption that if gay males earn less than straight males, it must be due to on-the-job discrimination. I speculate that this theory is most prominent among journalists, who are liberal by nature and refuse to countenance the notion there is anything “different” about gay and lesbian job choices or outcomes. If the results are different, that must be due to outside forces, i.e., discrimination, according to this thinking. This soft bigotry of equal expectations underpins a lot of the popular-press discussion of the facts of gay-male wages (which are, as we know, lower).

The discrimination theory seems to maintain that employers can tell which of their male employees are gay, go right ahead and hire them anyway, then pay them worse than straight males. As we have found, the evidence shows that gay males earn less than straight males mostly because they work fewer hours and choose lower-paying jobs, not because of workplace discrimination.

And at any rate, if there were really widespread discrimination against gays and lesbians in wages and earnings, lesbians wouldn’t be earning more money than straight females. But they are! You can’t argue for across-the-board discrimination if one group (gay males) takes a wage hit while the other group (lesbians) earns a payday. This is not a statement that gay males could not specifically be subject to discrimination; it is a warning not to generalize. (Blandford 2003, pp. 638–639, is especially instructive here.)

As I read it, there is no proof in these papers that discrimination has occurred, nor that such discrimination resulted in lower wages. All that researchers can do is read between the lines and draw inferences. Plus the theory works well only for gay males; a couple of papers imply that gay males face earnings discrimination because employers find gay males distasteful but are OK with lesbians. But you can’t prove that by running numbers.

• Black 2003 could not find any statistically valid evidence of discrimination against lesbians when comparing them to straight women who have never been married.

• Plug 2004 makes the strongest statement – but really the only supportable statement – about what the wage premium for lesbians says about the discrimination theory: “Discrimination requires negative earnings effects for being homosexual. Although this is true for gay men, it does not hold for women.”

• Arabsheibani 2004 (U.K. data) found essentially no wage penalties for gay males over 40: “It may be that older homosexual workers have had longer to find a better job match and therefore may be less likely to experience discrimination.” This could another way of saying that older gay men have had no choice but to alter their careers to route around discrimination, if it even exists. More benignly, it says that gays self-sort into gay-friendly jobs, which we know already.

• Badgett 1995 paraphrases a theory by Escoffier 1975 (one of the few papers I could not get my hands on for this project). Badgett writes:

  [P]assing may require a conscious effort to avoid potentially awkward social interactions that contribute to job satisfaction or advancement for other workers. The isolation involved in many passing strategies could lead to higher absenteeism and job turnover, and the energy devoted to passing might reduce productivity. In this case, the behaviour is not an intrinsic characteristic of the worker but an effect of indirect discrimination within a workplace perceived as threatening. Two individuals with equal productive abilities would have differential productivity and, therefore, differential wages because of the work environment’s effect on the gay individual’s productivity.

  Again according to Badgett, you could pass for straight at the workplace another way – by becoming a workaholic. Gay workers, “driven by a half-conscious belief that if they just show themselves productive enough, worthy enough, good enough, [believe they] will overcome the invisible stigma.” Gays may try twice as hard just to be considered competent – a scenario familiar from other minority groups.

  Badgett 1995 also concludes, after controlling for education and job choice, that gay males earn less because of discrimination.

• Research from the U.K. includes discrimination-related papers by Jeff Frank, by far the least credible evidence reviewed in this entire project. The problems start with the subjects chosen (academics) and extend all the way down to wording (“LGB males”). His methodology deters straight males from responding, among other things.

Is there actual evidence of discrimination?

I found two papers that make a plausible case for discrimination as a cause of lower earnings.

• Carpenter 2008b found greater incidence of work harassment, distressing work environments, job loss, difficulty in finding a job, and “decreased income” for lesbians in Australia – the closest we have to research that measures discriminatory events.

• The most convincing purely econometric research on discrimination is by Elmslie 2007. Controlling for a few more factors than other papers did, this research finds “negative wage and compensation effects for gay men as compared with married men after controlling for the general marriage premium. These occupations are management, building and grounds cleaning and maintenance, construction and extraction, and production,” with transportation and material moving also possible though less statistically significant.

  Their explanation is “these occupations tend to be ones that do not involve direct contact with customers and are traditionally blue-collar or otherwise male-dominated.”

  This finding is a bit odd given that many other studies show gay males overrepresented in management and the whitest of white-collar professions. It could be that gay males earn less in very-high-level management ranks, where there’s a lot of machismo and so on, but that’s just my interpretation. My theory here is partially borne out by the finding (in the same paper) that the only statistically significant occupational field where lesbians suffer differential treatment is – again – management. Perhaps tough-guy managers are keen to keep women out of their midst, lesbians especially. But that finding is at odds with other research showing a greater acceptance of lesbians in male-dominated professions. My interpretation is that if tough-guy managers are biased against anyone, it’s gay males.

The cost of staying in the closet

It stands to reason there are gay-male and lesbian employees whose homosexuality is completely unknown to everyone in their workplace, or to everyone that matters, like bosses and the HR department. These people may not really be in the closet elsewhere in their lives, but they are at work. Of course, some people are completely or somewhat out in their workplace. But researchers are loath to admit that some employees are so obviously queer that nobody in the office has any illusions. This is another soft bigotry – an assumption that everyone is so truly equal as to be indistinguishable. People aren’t that stupid.

• Badgett 1997b states something that is more or less obvious – gays stay closeted where the penalties for coming out are greatest.

• Blandford 2003 was able to identify closeted gays, bisexuals, and lesbians. Somewhat unsurprisingly, 11% of the closeted gay/bi males worked in construction (vs. 2.6% out gays, 7%–8% heteros). Out lesbians were about ten times more likely to be cops (work in “protective services”) than all other females.

  But that paper also dodges the obvious conclusion – that gay men and lesbians have jobs they prefer to do. (“[T]here seems little basis for expecting differential occupational patterns and returns to market labo[u]r based solely on human-capital arguments.” All right – but what about patterns of temperament and occupational preferences?)

• For workers who are not out at work, is there a wage penalty? Berkhout 2008 found there was, at least in the Netherlands: “[G]ay/bisexual workers earn on average 9% less than other men, and that among gay/bisexual workers those with disclosed identity earn on average 8% more.... Independent of the specification used, the penalty for not being open... moves around the 5% [mark].”

  It pays to be out on the job.

Bisexuals

The findings for bisexuals, of which there aren’t a lot, are distinct from the findings for actual gays and for straight people.

• Carpenter 2005 suggests that true income variability happens with bisexuals. When researchers mix together gay males and bisexual males, the results from bisexuals that skew what’s really happening with gay males, which might be that they don’t suffer an income gap that can be statistically proven. “[T]he earnings difference between gay and bisexual persons appears to be on the same order as that between heterosexuals and non-heterosexuals.”

• Lafrance 2009 writes:

  [O]ne of the more disturbing findings of our analysis is the persistently poor labour-market experience of bisexuals. Studying gays and lesbians in stable partnerships will lead to a serious underestimate of the labour-market difficulties faced by those who are unable either to fit with the heterosexual norm or to achieve the high-education urban lifestyle pictured in New York Times marriage announcements.

---

Results from outside the U.S.

A few papers have been published looking at gay money in Canada, Sweden, the Netherlands, and the U.K. In all but the last country is the general finding true, and some of those U.K. results are simply unreliable.

Canada

The very richest gays may suffer less, at least in Canada: Carpenter 2008a found, for gay men, “that a minority sexual orientation was associated with a smaller personal-income penalty at higher income quantiles.... That is, for both gay men and lesbians the income differentials associated with sexual orientation are somewhat smaller at higher points in the income distribution.” Also, the income disparity is much worse for partnered gays and lesbians than singles.

Netherlands

Plug 2004 rather ingeniously looked at recent college graduates just entering the workforce. The paper admits that its method makes it almost impossible to measure on-the-job discrimination during the entire work life, but it has the advantage of showing that trends start early. The study found about a 4% wage penalty for gay males and a 4% wage premium for lesbians.

Sweden

Only one paper (Ahmed 2010) examines Swedish data and it supports the general finding.

The richest lesbians are much richer than straight females and the poorest gay males are much poorer than straight males. The lesbian earnings advantage “to a large extent is driven by relatively high earnings among lesbians at the top of the earnings distribution, while the earnings differential between lesbians and heterosexual females at the bottom of the earnings distribution is smaller and statistically insignificant.... It is a striking result that the earnings differential between gay and heterosexual men is considerably larger in the bottom than around the median and in the top of the earnings distribution.”

However, lesbian earnings premiums disappear when we factor in controls for education and so on. The most reliable explanation for the earnings gain for lesbians is the fact that lesbians have more education.

United Kingdom

Arabsheibani 2007 found about a 5% gain for gay males, which disappeared when controls were added. The lesbian wage premium starts out at an astounding 36%, then whittles down to 6% when controls are added. Effectively, then, findings with controls added approximately match U.S. findings, the study states.

Papers by Frank focus solely academics, making them instantly irrelevant even if there weren’t a raft of other reasons marking them as second-rate, which there are. Frank produced the worst research on gay and lesbian economics I read for this project. The less attention it gets, the better.

Australia

Carpenter 2008b found about a 30% gain in personal income for lesbians in Australia – an unusual finding, but still an earnings premium.

---

Do any studies contradict the general finding?

In other words, are there any studies that fail to show that gay males earn less than straight males and lesbians earn no less than straight women? Is there research that contradicts what otherwise would amount scientific consensus?

Yes, but not a lot of it.

• Black 2007 found that, for couples with adopted children, gay males earned the most money (average $68,491). Lesbians earned $63,996 on average. Both figures dwarfed the incomes of straight couples ($57,975). I assume this is a form of self-selection, because adoptive parents always self-select and only the richest adopt.

  That same paper also found that household incomes of gay-male partners dwarfed those of lesbian or straight couples ($91,676 gay males versus about $73,000 each for the others).

• Carpenter 2004 found earnings penalties for gay men and lesbians. It’s the latter point that is unusual: Most research finds lesbians earn more than straight women.

• Carpenter 2005 found that gay males earn more than unmarried straight males. Lesbians earn marginally more than all other women.

• Mueller 2007 could not find a significant earnings difference between lesbian couples and cohabiting hetero women, who in turn have almost no penalty compared to married women. (All these findings are unusual.)

• Phua 1999 found that by far the highest incomes of any couples were those of gay and lesbian cohabitors, with spreads of $10,000 to $25,000 over straight people in some comparisons.

• Ash 2006 found gay and lesbian couples earned much more than unmarried or married hetero couples – comparing gay-male couples to unmarried straight couples, almost twice as much.

• Research by Elmslie 2007 does not exactly contradict the general finding, but questions whether that finding is influenced by economists’ counting up earnings and wages without controlling for hours worked. When that is done, gay males in relationships earn 9% less than unmarried men in heterosexual relationships. No statistically significant findings for women were found.

• Carpenter 2007b studied about 1,800 college students and found that gay males had higher earnings from paid work than straight males; he could find no differences between lesbians and straight females. For income from “other sources,” the results were reversed – no difference for gay males, more for lesbians, both in relation to straight people.

American Community Survey 2010 findings

The data that seem to most resoundingly contradict the general finding came out in October 2010, just before the release of this project in November 2010. The U.S. Bureau of the Census’s American Community Survey reports data from localities with populations of 65,000 and greater. It found much higher incomes for gay-male couples – by far the highest, in fact.

Now, these findings were reported in the popular press and by academics, but incompletely. I must say I found it suspicious that the first mainstream press coverage came from the anti-gay Washington Times; American right-wingers are happy to insist that gays are rich, hence undeserving of “special rights.” This is not a claim that the Washington Times coverage was wrong or inaccurate.

But a legitimate research insitution, the National Center for Family & Marriage Research at Bowling Green State University, produced a report (PDF) entitled “Same-Sex Couple Households in the U.S., 2009.” It stated:

Male same-sex couple households... have the highest average household income of all couple households at nearly $117,000, while unmarried different-sex couple households have the lowest average household income at approximately $64,000....

The Williams Institute produced a press release (PDF) stating:

While the average household income of same-sex couples exceeded that of different-sex married couples ($104,048 vs. $93,351, respectively) in 2009, the incomes of same-sex couples fell by 3% between 2008 and 2009. This compares to a 1.8% decline among different-sex married couples. Same-sex male couples were particularly affected as evidenced by the 4.4% decline in their household incomes. This figure also exceeds the 3.6% decline in income among all households in the U.S.

Williams Institute researcher Gary Gates produced a PowerPoint presentation (typically information-poor, as PowerPoints almost invariably are; PDF) showing that same-sex couples receive public assistance much more often than straight couples.

So let’s recap: A single data set is claimed to report that gay-male couples are richer than everybody else and also that gay and lesbian couples receive public assistance much more often. Could both these claims be correct? In a word, yes.

Let me be the first to report all the relevant data from the ACS, not just whatever fits an agenda. To save you the trouble of looking around for the raw data, you can just download my version (.XLS) of the original Census Bureau Excel spreadsheet. (The original is found under “Data from the American Community Survey: 2009 Tables” on a lengthy page of other spreadsheets.)

Average household incomes:

• Married hetero: $93,351
• Unmarried hetero: $64,005
• Gay males: $116,749
• Lesbians: $92,213

Percentage earning less than $35,000 vs. $100,000 or more:

• Married hetero: 17.1% vs. 31.7%
• Unmarried hetero: 31.8% vs. 15.7%
• Gay males: 13.2% vs. 44.1%
• Lesbians: 17.9% vs. 34.0%

Put all these figures together, and the impression they leave coincides tightly with that of 1990s gay marketers – that gay males (not “the gay community”) are well-off. More so than straight people, in fact.

Now, what about the DINK stereotype? What proportion of these couples have children in the household?

• Married hetero: 42.0%
• Unmarried hetero: 42.1%
• Gay males: 11.8% (not “no kids”)
• Lesbians: 23.9% (also not “no kids”)

So that part of the marketers’ stereotype is disconfirmed, consistent with absolutely all the other data that looks at presence of children.

Do these ACS findings mean all the other research is wrong?

Actually, yes, they might – but not that other findings are categorically wrong across the board.

• ACS does not remove outliers. Other research eliminates, for example, “those with hourly wages of less than $1 and [of] more than $500” (Mueller 2007). But with ACS, we’re dealing with 100 times as many straight couples as gay couples (61,732,298 vs. 581,300). The presence of just a few millionaires could skew the averages. (The average of {1, 2, 3, 4, 5, 6, 10⁶} is about a million even though the real trend of those figures is less than 10.)

  I don’t know that these millionaires exist; it’s just my hypothesis. But the presence of those millionaires would prove merely that gay and lesbian millionaires exist, not that the entire gay community is well-off or a lucrative market. And their outlier wages mask the true wages of the rest of the sample.

  How would one fix this problem? Re-run all the other studies with high-end outliers included. Show the ACS data points so we could determine how many high-end outliers are in the gay and straight sets.

• The data on income spread (how many couples earn less than $35,000) show poverty levels similar across the board – except for unmarried heteros who, consistent with other research, are poorest of all. Again, this goes to show that gay and lesbian couples are not all overrepresented at the high end of the wage scale. Remember, the gay-marketing angle did not hold that there are also a lot of poor gays mixed in with the rich ones; we were told that the entire gay community (really just gay males) was affluent.

  ACS data show a higher proportion of gay and (surprisingly) lesbian couples with high incomes. But this too is consistent with my hypothesis that a small number of very high earners are skewing the data. (Two millionaires out of 10,000 are not many millionaires. But neither are two out of 100, and those are the orders of magnitude we’re dealing with here.)

• Again, singles aren’t counted, because the methodology to do that is difficult, expensive, and arguably intrusive. (But necessary.) Prediction: These individuals’ incomes are lower.

• Recall that ACS covers localities with populations of 65,000 or more. Since gays live overwhelmingly in cities (Dan Black’s classic paper [2002] has an equally classic title: “Why Do Gay Men Live in San Francisco?”), even without double-checking I promise you this survey captured most of the gay and lesbian couples in the U.S. (Gay gentleman farmers may make for a cute TV show, but as a reflection of reality they come up short.) Consequently, the data probably really do capture all or almost all high-earning gays. Adding their incomes to the mass of incomes distorts that mass more than adding the quantity of high earners distorts the full quantity of people studied.

The ACS data are extremely interesting and are a welcome addition to the corpus. But in and of themselves, I don’t see how they can be interpreted as validation of the stereotype of rich gays. They certainly show that high-earning gay males (and – less so – lesbians) exist in the United States.

Could all these results put together just be totally wrong?

It’s unlikely but possible, for several reasons.

• Sample sizes are usually small – sometimes only a few dozen gay, lesbian, and/or bisexual subjects.

• A great many individual findings are not statistically significant, though no paper surveyed here reports only non-significant data (because then there would be nothing to report). Having this many non-significant results could mean all the results are inaccurate.

• In most cases, researchers considered gay and lesbian couples, which leaves out large populations that could materially change the results. (Stated another way, there might be one set of results for couples and another for singles.) Lafrance 2009: “The majority of homosexual men, and 40.6% of lesbians, are single/never married, suggesting that studies based on gays and lesbians cohabiting with same-sex partners exclude a broad population.”

  Income disparity is much worse for partnered gays and lesbians than singles, Carpenter 2008a found, which means “studies relying only on partner-based samples... may have systematically overstated the magnitude of the average sexual-orientation gap.”

• A very serious issue is defining gay, lesbian, or bisexual. Nobody at all seems to think that defining heterosexual is a problem, least of all heterosexuals, who are never in any doubt about it whatsoever. Few population surveys have the honesty to ask respondents what their sexual orientation is and also which gender their partners have been. (You need both facts to cross-check and eliminate little white lies.) Researchers are forced to draw inferences from, say, statistical surveys.

• Very low earners and very high earners are usually eliminated (e.g., Mueller 2007: “those with hourly wages of less than $1 and [of] more than $500”). Since the myth of gay affluence holds that gay men are affluent, eliminating the most affluent subjects may skew the results. Studies sometimes go out of their way to exclude income that comes from something other than wages. This includes not only social assistance (Lafrance 2009) but self-employment, including high-end consultants with six-figure incomes. Here again, what are probably the poorest and the richest subjects are excluded.

Ridiculous explanations

A couple of papers propose ridiculous explanations for why gay males and lesbians make certain occupational choices. The last thing researchers want to do is admit that gay men like certain jobs (not just “occupational sectors”), that lesbians like different jobs, and that neither group would be caught dead working certain jobs. Some researchers float explanations that strain credulity.

• Black 2007 notes that gay males “who realize early in life that they are unlikely to form traditional households with children may plan on specializing less intensely in market production.” I really don’t think that young gay adults sit themselves down one find day and congratulate themselves on how they dodged a bullet in not having kids, which frees them up to work less.

• Carpenter 2005 states that “researchers evaluating labour-market discrimination must contend with the fact that disclosure of one’s gay, lesbian, or bisexual orientation in the workplace is a necessary (but not sufficient) condition for the existence of... discrimination.” Some people are obviously, even screamingly, gay or lesbian. They couldn’t hide if they tried. The assumption that everyone is equal and indistinguishable until we explicitly out ourselves at work doesn’t make sense.

• Clain 2001 essentially sets up “the stereotypic heterosexual male worker” as an ideal. The paper suggests that lesbians are more like that ideal but gay men are less like it. Hence lesbians have greater self-esteem in the workplace, gay men less. This somehow leads lesbians to earn more than straight females and gay men to earn less than straight males, as that same paper also confirms.

• Badgett 1995 actually states (“suggest[s]”) that there is no link between being gay or lesbian and choice of education: “[S]exual orientation is unlikely to influence educational decisions.” We know from real life and from the preponderance of evidence that gays and lesbians are more educated. (The study in question found equal years of study for all groups.) We know gays and lesbians are temperamentally influenced to choose more education. Gays don’t like dropping out of school!

Other facts

• Several studies find that gay-male couples’ incomes and lesbian couples’ incomes are similar to each other. Never mind what straight people make – according to these results, all gay and lesbian couples earn roughly the same amount on average. The interesting thing is these earnings make gay couples lower earners than straight couples but – again, with the same earnings – make lesbian couples higher earners. Everything, it seems, really is relative.

• “[G]ay men report relatively higher levels of investment income, while home ownership is relatively lower for both gay and lesbian couples” (Black 2007).

• “[L]esbian couples have somewhat more expensive homes than their heterosexual counterparts, and gay couples appear to have much more expensive homes” (Black 2000).

• Tebaldi 2006 wonders if “gay men have a higher reservation wage,” i.e., a floor wage they won’t go below. Though the paper calls this “a less reasonable explanation,” what it’s actually saying is gay males won’t work for peanuts. This is probably actually true for a lot of gay males, myself included.

• Gluckman and Reed 1997 get it quite wrong when they claim “it is clear that women’s lower incomes place lesbian households at a unique disadvantage.” (Lesbians earn more than straight women.)

Posted: 2010.11.16