Sexual Orientation Wage Gaps across Local Labour Market Contexts: Evidence from Canada

Denier, Nicole; Waite, Sean

doi:https://doi.org/10.7202/1043174ar

Introduction

Mounting evidence suggests that sexual orientation matters in the labour market (Klawitter, 2015). Research in Canada, specifically, points to a wage hierarchy not only by gender, but also by sexual orientation, with heterosexual men out-earning gay men, lesbians, and heterosexual women (Waite and Denier, 2015). Little research has looked at how residential concentration in large metropolitan areas factors into the creation of these wage differences. This is an important question, as sexual minorities, particularly gay men, tend to live in large urban centres. As we show, of the men in same-sex couples identified in our sample, 86% lived in metropolitan areas, with a full 57% in the three largest metropolitan areas of Toronto, Montreal, and Vancouver. Geographic concentration may shape the collective fortunes of sexual minorities, providing opportunities and constraints unique to a few local labour markets. At the same time, gay men and lesbian women living outside areas with larger LGBTQ+ communities may face added difficulty in the labour market, especially in areas where tolerance towards non-heterosexual individuals is limited.

This article aims to uncover how sexual orientation wage gaps vary across geographic areas in Canada and further document whether the mechanisms that contribute to those earnings disparities are similar across these contexts. As such, we provide the first exploration of sexual orientation wage gaps across Canada. We begin by documenting wage gaps across the metropolitan/non-metropolitan divide, further estimating wage gaps for the three largest Census Metropolitan Areas (CMAs): Toronto, Montreal, and Vancouver. We then examine how earnings differences vary across occupations and points in the earnings distribution to provide descriptive evidence of work settings that contribute to sexual orientation wage gaps. Finally, we ask to what extent public sector employment, which tends to be characterized by higher rates of unionization and bureaucratic hiring and promotion practices, potentially limiting differential valuation of employees, is associated with wage equality between heterosexual and sexual minority employees across areas (Mueller, 1998, 2002).

To answer these questions, we use data on same- and opposite-sex couples from the 2006 Census of Canada, the first since the federal recognition of same-sex marriage in Canada, and thus the first to nationally enumerate married and common-law same-sex couples.[1] We contrast the experiences of men and women in same-sex couples relative to heterosexual men in five places: Toronto, Montreal, Vancouver, and other metropolitan and non-metropolitan areas in Canada; areas that vary in terms of density of the sexual minority population, attitudes towards homosexuality, and labour market characteristics (Cutler and Jenkins, 2001; Heisz et al., 2005; McGrane, Berdahl and Bell, 2017). At the same time, each of these areas is subject to both federal and provincial anti-discrimination legislation outlined in the Canadian Charter of Rights and Freedoms and provincial human rights codes that prohibit discrimination in employment on the basis of sexual orientation—meaning that observed differences in pay gaps are not confounded with profound differences in policy environments. This stands in contrast to the United States, where employment protection for sexual minorities varies considerably across the country (Klawitter, 2011).

We find that wage gaps for gay men and lesbians are highest in non-metropolitan Canada. Gay men tend to be over-represented in high-paying occupations throughout Canada, but face wage gaps within these occupations. However, the extent of this disadvantage within occupations varies considerably across Canada. Yet, for both gay men and lesbian women, the highest observed wage gaps remain in non-metropolitan Canada. Only in the public sector do men and women in same-sex couples reach earnings parity with heterosexual men across Canada.

Literature Review

There is growing evidence in the United States, Canada, and a number of European countries that gay men earn less, and lesbians earn more, than their heterosexual counterparts (Ahmed and Hammarstedt, 2010; Antecol, Jong, and Steinberger, 2008; Badgett, 1995, 2001; Berg and Lien, 2002; Black et al., 2003; Clain and Leppel, 2001; Plug and Berkout, 2004; Waite and Denier, 2015).[2] In Canada, sexual orientation has only recently gained attention as a source of labour market stratification, in part because of difficulties identifying sexual orientation in surveys with employment information. To isolate the mechanisms generating labour force disadvantage, particularly discrimination, it is crucial to separate earnings from employment from other sources of total income. Unfortunately, no Canadian data simultaneously identify sexual orientation for the entire population and provide earnings data. For instance, the Canadian Community Health Survey (CCHS) has a question about sexual orientation but, as a health survey, limits earnings data to total income. The General Social Survey (GSS) also includes a direct question about sexual orientation, but income is coded categorically. Census data are the only data that have both earnings measures and allow researchers to identify sexual orientation based on conjugal status with a person of the same-sex; such data remain limited by failing to identify those not in conjugal relationships or to allow for self-identification. These challenges have complicated researchers’ attempts to estimate sexual orientation wage gaps and have produced some differences in estimates (Denier and Waite, 2016).

In the first Canadian study, Carpenter (2008) used the 2003 and 2005 CCHS and found that individuals who self-identified as gay had incomes that were 12% lower than heterosexual men, while lesbians had incomes that were about 15% higher than heterosexual women, even after controlling for family situation, occupation, and labour force participation. Cerf (2016) pooled the 2003-2009 CCHS and found a 13% income penalty for gay men in couples and a 8% wage premium for lesbians with a partner. Mueller (2014) used the 2006-2010 GSS and found no difference in income between gay and heterosexual men and a 16% wage advantage for lesbians compared to heterosexual women when adjusting for education, experience, occupation, and industry position. Using 2006 Census data, Waite and Denier (2015) found that men in same-sex couples earned about 5% less and women in same-sex couples 9% less than heterosexual men, accounting for more detailed occupation and industry of employment. Lesbians earned on average 8% more than heterosexual women. Drawing on data from the 2001 and 2006 Censuses, as well as the 2011 National Household Survey (NHS), Waite (2015) found little evidence that sexual minority wage gaps had attenuated over the last decade. Dilmaghani (2017) used the 2008-2012 Canadian Alcohol and Drug Use Monitoring Survey and found a lesbian income premium, but no income gap for gay men who were employed full-time[3].

Sociologists have suggested that this stratification of earnings by gender and sexual orientation can be understood with insights from the larger gender wage gap literature (Waite and Denier, 2015). Theories of hegemonic masculinity posit that the ideal worker is seen as a stably employed, heterosexual male with children (Acker, 1990; Hodges and Budig, 2010). Labour markets grant dominance to this form of masculinity, which conveys authority, competence, and commitment to employers and coworkers (Connell and Messerschmidt, 2005; Ridgeway and Correll, 2004). Within this framework, gay men conform less to the ideal and, as a result, may have their productive abilities devalued. Conversely, lesbians may be perceived as closer to the ideal male, more committed to work than heterosexual women because they are less likely to be married with children (Klawitter, 2015). In a similar vein, lesbians who do have children may face added pressure to maximize wages in the absence of a higher earning heterosexual male in the household. Taken together, deviations from the hegemonic ideal worker type are associated with wage differences that result in a hierarchy of earnings by gender and sexual orientation. It remains possible that hegemonic ideals of masculinity and sexuality are localized, with norms and cultures specific to work contexts in different locations shaping notions of the ideal worker (Britton and Logan, 2008).

Pay Gaps across Local Labour Markets

To date, all Canadian studies have estimated sexual orientation wage gaps for the entire country. This masks an important aspect of the lives of sexual minorities in Canada: the role of space in influencing wellbeing across a number of domains, like safety, dating opportunities, and career options (Lewis, 2013, 2015). The emergence of gay enclaves following WWII has indeed shaped the residential patterns of the LGBTQ+ population in many countries—gay men, especially, have located in or near the “gaybourhoods” of large cities, like New York’s Chelsea, London’s Soho or Montreal’s Gay Village (Compton and Baumle, 2012; Gates and Ost, 2004; Ghaziani, 2014; Hinrichs, 2012; Nash, 2006; Nash and Gorman-Murray, 2014).[4] Historically, such enclaves stood apart from an otherwise intolerant society. But even today, feelings towards “homosexuals” are more favourable in urban and metropolitan areas than in rural and small town Canada (Cutler and Jenkins, 2001). Such tolerance may be the result of increased contact with or exposure to individuals of differing sexual orientations. Montreal, Toronto, and Vancouver are home to three of the largest gay enclaves in Canada: the Gay Village, Church-Wellesley Village, and Davie Village, respectively. The strong representation of LGBTQ+ populations in these areas, including the presence of anchor institutions and commemorative events, may help to solidify the collective identity and legitimacy of sexual minorities in these cities (Ghaziani, 2014). Indeed, in recent years major corporations (and employers) in these cities have sponsored commemorative events, like pride marches, to demonstrate some commitment to inclusion. Given these differences in attitudes and the density/visibility of the LGBTQ+ community, we may expect differences in wage gaps across Canada.

The little international evidence available from individual states or local areas shows that the presence and magnitude of sexual orientation wage gaps may indeed be location-specific. Arabsheibani and Wadsworth (2004) show that wage gaps in the UK were found only in areas outside London. Similarly, Carpenter (2005) found no wage gap in California using the 2001 California Health Interview Survey. He offers that this null finding may be a result of more “liberal” views in the state, along with the presence of gay communities that have been integral to the passage of anti-discrimination legislation. A few studies in the United States have estimated the effect of state and local anti-discrimination policies on earnings gaps, and showed that there is significant variation in the effect of sexual orientation in earnings across states and metropolitan areas (Baumle and Poston, 2011; Klawitter and Flatt, 1998; Klawitter, 2011). These studies further indicated that the presence of local anti-discrimination policies, and a high density of same-sex couples in an area, decreased the wage disadvantage of gay men, but not that of lesbians (Baumle and Poston, 2011; Klawitter, 2011). In an innovative audit study, Tilcsik (2011) compared call back rates of gay male job applicants across U.S. states, and found hiring discrimination present in some, like Texas, but notably absent in more liberal states like New York. It remained unclear, though, whether this was a result of different attitudes or the presence of anti-discrimination legislation, which in the U.S. has been adopted in areas with larger gay communities.

Documenting differences in pay in the cities in which sexual minorities live is crucial to understanding and situating broader patterns of disadvantage. One part of this is methodological: national estimates may obscure disadvantage in the labour markets in which people actually work. For instance, suppose gay men and lesbians live in a few higher-paying urban areas, while heterosexual populations are spread evenly across higher- and lower-paying urban and rural areas; estimating the wage gap relative to the whole heterosexual population will underestimate the true wage disadvantage experienced in these labour markets[5]. It may even suggest that sexual minorities have higher wages on average.

Examining smaller geographic areas can further provide insight into the factors that lead to wage disparities, as they better identify common labour market chances and organizational cultures encountered by employees. Gender scholars have long-pointed to the role of occupational context in shaping ideal worker norms and thus labour market outcomes. This is partially because certain occupations are deemed more/less appropriate for women, and further because female dominated occupations often pay less (Britton and Logan, 2008; England, 1992, 2010; Reskin and Roos, 1990). Gay men and lesbians are more likely to sort into gender atypical fields of study and occupations (Ueno, Peña-Talamantes, and Roach, 2013; Ueno, Roach, and Peña-Talamantes, 2013). Such patterns could disadvantage gay men who sort into lower-paid feminine occupations, but advantage lesbians who sort into more highly-paid masculine jobs. Yet, occupational position does little to explain sexual orientation earnings gaps (Antecol, Jong, and Steinberger, 2008; Carpenter, 2008; Waite and Denier, 2015). This may be in part because wages are determined at a more local level than occupation, taking place within a workplace or firm (Cohen and Huffman, 2004; Fortin and Huberman, 2002; Huffman, 2004). Indeed, women face earnings disadvantage not only as a result of occupational segregation, but also because they are blocked from the highest-paying firms or jobs within these labour market structures—an effect commonly referred to as the “glass ceiling” in higher-paying professional occupations (Boudarbat and Connolly, 2013; Cotter et al., 2001; Roth, 2006). Evidence from linked employee-employer surveys has shown that women tend to be located more in lower-paying firms than men contributing to lower pay, but even within firms, women are paid less (Drolet and Mumford, 2012). Huffman (2004: 336) created occupation-industry-metropolitan area cells as proxies for jobs in the U.S., and found considerable geographic variation in the gender segregation of “jobs”. The author also found that the within-job gender wage gap is larger the higher up in the local wage hierarchy is the job. Consequently, differences in the location of the highest-paying firms or jobs within occupations or industries, or the characteristics of jobs across places, may influence earnings disparities by sexual orientation. Local labour market contexts narrow in on the local organizational cultures in which work is carried out.

Our analysis thus considers not only geographic variation in wage gaps, but also looks at whether earnings differences are similar in various occupations or points in the earnings distribution in different labour markets. This recognizes the diversity of earnings opportunities and character of wage inequality across Canada. For instance, in 2000, median annual earnings for full-time, full-year workers was $83,500 in Toronto, but only $65,000 in non-metropolitan Canada (Heisz et al., 2005). The ratio of earnings of full-time, full-year workers at the 90^th and 10^th percentiles in 2000 was 7.07 in non-CMA Canada, but 5.14 in Montreal (Heisz et al., 2005). Thus, a major part of an earnings gap in non-CMA Canada could be attributable to the larger relative penalty of employment in low-paying rather than high-paying jobs. A glass ceiling effect may be more pronounced in Toronto, where earnings at the top of the distribution are the highest. At the same time, certain types of employment may provide more equal opportunities uniformly across the country. Public sector employment, in particular, has been an avenue for women and minority groups to integrate into the labour force, as it tends to rely on clear rules in hiring and promotion practices that are often formulated in accordance with anti-discrimination legislation (Gunderson, 1979; Hou and Coulombe, 2010). The public sector also tends to be highly unionized, providing an additional mechanism to address potentially discriminatory actions (Mueller, 1998, 2002). Together, these forces may limit differential valuation of employees. Waite and Denier (2015) found that sexual orientation wage gaps were reduced and, in some cases, eliminated in the Canadian public sector nationally. To the extent that discretionary pay or taste-based discrimination produces earnings differences, public sector employment should uniformly reduce or eliminate wage gaps.

Research Questions

To summarize, there is reason to believe that sexual orientation pay gaps are not uniform throughout Canada, which may make national estimates misleading. Examining subnational variation further aids in breaking down the potential mechanisms that contribute to the sexual orientation pay gap, particularly within occupation and job wage inequality. This leads us to ask two questions:

Do wage gaps vary across metropolitan and non-metropolitan Canada?
Across local labour market contexts, do wage gaps differ by occupation, sector, or position in the wage distribution?

These questions are generative: research on sexual orientation in the Canadian labour market is in its infancy and requires documenting important dimensions of population dynamics and wage inequality to lay the groundwork for and motivate further research. Canada provides a particularly interesting case to examine subnational variability as the legislation governing gay marriage and labour market discrimination is uniform across the nation. As a result of this uniformity, observed differences in pay gaps across local labour markets are not confounded with vastly divergent legal contexts surrounding sexual orientation. Similarly, the federal legalization of same-sex marriage in 2005 spurred the enumeration of same-sex married and common-law couples across the nation, providing unique data in North America.

Data and Methods

To answer our questions, we use data from the master file of the 20% sample of the 2006 Census of Canada[6], the first since the federal legalization of gay marriage.[7] The Census provides information on the marital and common-law status of individuals, allowing us to identify same-sex married and cohabiting couples. The Census does not have a question on sexual orientation, so our sample is limited to couples in same- and opposite-sex relationships only. We refer to women in same-sex couples as lesbians, men in same-sex couples as gay, and individuals in opposite-sex couples as heterosexual, recognizing that these are not self-identified statuses. Exclusive focus on couples is increasingly standard practice as a result of the dearth of questions on sexual orientation in population data, although such an approach could bias estimates of true pay gaps if heterosexual and gay and lesbian couples systematically differ from the single population or those individuals in non-conjugal relationships. Currently, there is little evidence to suggest that selectivity into partnership based on observed human capital characteristics varies substantially by sexual orientation—a pressing topic for future research. Carpenter (2008) showed that gay men in couples in the 2003-2005 CCHS were about 1.25 times more likely than all gay men to have a bachelor’s degree or higher, similar to the ratio for partnered heterosexual men, who were 1.28 times more likely than all heterosexual men to possess such a degree. Likewise, coupled gay men were about 1.08 times more likely than all gay men to work full-time hours and heterosexual men in couples were about 1.09 times more likely than all heterosexual men to work full-time. Lesbians were similarly selected into couples based on higher levels of education and a higher propensity to work full-time.

To understand the process of labour market advantage, it is useful to adopt the most advantaged group as a yardstick. Research consistently shows that heterosexual men in partnerships are the most highly remunerated (Ahituv and Lerman, 2007; Chun and Lee, 2001); we thus compare the earnings of gay men and lesbian women to heterosexual men. Comparing lesbians to heterosexual men is important given the use of couple data and possible differences in unobserved selection into partnership. Unobserved heterogeneity is less likely an issue if the direction of selectivity into partnership is the same for the reference group—this is the case for lesbians, as discussed above. This is, however, not true for partnered heterosexual women: while heterosexual women in couples are more likely to have higher education, they are less likely to engage in full-time work, signaling a potential difference in labour force engagement (Carpenter, 2008). Observed “lesbian wage bonuses” then may be a result of unobserved heterogeneity, especially if heteronormative pressures to assume responsibility for caregiving work make partnered heterosexual women more likely to forgo the most lucrative labour market opportunities. We focus on models for lesbian women relative to heterosexual men, but provide estimates of wage inequality for lesbian women relative to heterosexual women below.

The sample is limited to employees aged 25-64, who have likely finished schooling and are attached to the labour market. Additionally, we exclude individuals residing in the Yukon, Northwest Territories, and Nunavut, more remote areas of Canada with significantly different labour markets.

Analytic Strategy

Our analysis begins descriptively by documenting average differences in pay, occupation, and sector of employment by sexual orientation across five local labour market contexts. We proceed by separately estimating the average earnings difference between heterosexual and sexual minority employees for each place using Ordinary Least Squares (OLS) models. These OLS models control for a host of individual human capital and demographic characteristics, as well as occupation and sector of employment, discussed in detail below.[8] We then add to these separate OLS models an interaction term between sexual orientation and occupation of employment to gauge whether the impact of sexual orientation on wages is constant across occupations. To further elucidate whether there is a glass-ceiling effect, we estimate earnings differences by sexual orientation at different points in the earnings distribution using unconditional quantile regressions with a full set of controls. This amounts to predicting what effect sexual orientation has for high-wage vs. low-wage workers (Firpo, Fortin and Lemieux, 2009; Killewald and Bearak, 2014). Finally, we split our separate OLS models by industry sector, controlling for occupation, to assess whether the public sector may offer a check to differential valuation on the basis of sexual orientation across geographic areas.

Local Labour Market Contexts

We examine Statistics Canada Census Metropolitan Areas (CMA). A CMA consists of adjacent municipalities centred on a core metropolitan area; populations of CMAs must be greater than 100,000, with at least 50,000 residing in the core. This means that CMAs include not only downtown cores, but also well-integrated suburbs that would be within reasonable commuting distances. As a result, CMAs resemble local labour markets. We focus in on the largest three CMAs: Toronto, Montreal and Vancouver. We further compare these CMAs to the rest of other CMA Canada and non-CMA Canada, which includes smaller cities of less than 10,000 people, towns, and rural areas.

Labour Market Position

We are interested in three labour market positions: occupation, earnings class, and sector of employment. We use the 2006 National Occupational Classification for Statistics (NOC-S) broad categories to capture differences in occupation. We combine trades, transport, equipment operators, and related occupations, occupations unique to primary industry and occupations unique to processing, manufacturing, and utilities, which tend to have few sexual minority employees. The resulting classification is seven broad groups, which are roughly ranked by required levels of education and earnings. While these groups are broad, they allow large enough sample sizes of the gay and lesbian populations for meaningful analysis. To further explore sources of differentiation that are not the result of differential access to higher-paying occupations, we examine how well gay men and lesbians do in high- vs. low-paying employment. We rely on deciles of the earnings distribution to describe earnings classes. Industries as classified according to the 2002 North American Industry Classification System (NAICS-2002) are broken down into private and public sector, following Hou and Coloumbe (2010). Public sector industries include: federal government public administration; provincial and territorial public administration; local, municipal, and regional public administration; elementary and secondary schools; community colleges and CEGEPs; universities; and hospitals.

Earnings and Associated Determinants

Earnings are measured as total annual wage and salary income from 2005. This includes earnings from all paid work, as well as tips, commissions, and cash bonuses, before taxes and transfers. We exclude individuals with earnings less than $1000, and take the log transformation to account for outliers and individuals with intermittent labour force engagement[9]. We control for a number of individual characteristics that may account for differences in pay. For differences in human capital characteristics we control for level of education, work experience, and labour force participation. Education is a categorical indicator of highest degree obtained, including less than a high school degree; high school degree; college, CEGEP, certificate or apprenticeship; bachelor’s degree; master’s degree; or earned doctorate. The Census does not measure actual work experience, so we construct a Mincer proxy, to represent the potential number of years a respondent could have worked since completion of schooling (measured as Age – Years of Education – 6). The Mincer proxy is entered as a quartic function, which better represents the curvilinear relationship between earnings and experience (see Hamlen and Hamlen, 2012; Lemieux, 2006). Labour force participation is measured as annual weeks worked and whether or not the individual usually worked part-time or full-time. All models further control for demographic characteristics, including broad age group (25-34; 35-44; 45-54; 55-64), marital status, and presence of children in the household. As only couples are included in the sample, the reference category for marital status is cohabiting couples. We also control for immigration status and membership to a visible minority or Aboriginal group. Models for non-CMA areas additionally control for province of residence and residence in a rural area.

Results

Table 1 shows the distribution and mean earnings of the population, by gender, sexual orientation, and geographic area. Gay men and lesbians are more likely to live in metropolitan areas, supporting previous findings, but which cities they live in varies—gay men are particularly likely to live in Toronto, Montreal, and Vancouver, while lesbians are more likely to live in smaller cities. Around 31% of heterosexual men in couples live in a non-CMA area, but only 14% of gay men and 19% of lesbians live in these areas. Across Canada, both women and men in same-sex couples earn less than heterosexual men, with the largest gap for gay men in non-CMA areas, and the largest difference for lesbian women in Toronto.

Table 1

**Distribution of Population and Mean Earnings, by Geographic Area**

Table 2 shows the extent of segregation by occupation and sector of employment in each place. The occupational distribution across CMAs is quite similar for heterosexual men; management and finance occupations are relatively more common in Toronto, while sales and service occupations are relatively more important in Montreal and Vancouver, but not by much. In non-CMA Canada, however, heterosexual men are much more likely to work in manufacturing, trades, and primary industry: over half of heterosexual men in non-CMA Canada work in the latter three occupational groups.

Table 2

**Distribution of Occupation and Industry Sector, by Sexual Orientation and Geographic Area**

Are gay men able to reach high-paying occupations, relative to heterosexual men? Gay men were more likely than heterosexual men to work in management occupations in all areas except Montreal. The comparatively high proportion of gay men in managerial positions in Toronto and Vancouver could contribute to lower earnings disadvantage in these labour markets, particularly in Toronto, where wages at the top of the earnings distribution are the highest. Gay men are also more likely than heterosexual men to work in business, finance, and administrative occupations. In more middle class occupations, like those in social science, education and government, and those in arts, culture, recreation, and sport, gay men are over-represented in all areas. In higher-paying health occupations, gay men are under-represented in the three largest CMAs, especially Toronto. Across Canada, gay men are over-represented in sales and services and under-represented by large margins in manufacturing, trades, and primary industry, two of the lowest-paying occupational groups. To summarize, if anything, gay men work in high-paying occupations. When gay men do work in lower-paying occupations, they are more likely to work in sales and services rather than manufacturing, trades, and primary industry. The relative over-representation of gay men in the highest-paying broad occupations is weakest in non-CMA Canada and Montreal. The general pattern, however, seems to leave much room for inequality within occupations in all areas.

For lesbian women, on the other hand, lack of equal representation in higher-paying occupations is a likely source of lower pay. In all areas, lesbian women are less likely than heterosexual men to work in management. On the other hand, women are over-represented in business, finance, and administrative occupations, and those in social sciences, education, government service and religion, especially in Vancouver. Lesbian women were just are likely as heterosexual men to work in sales and services, but more likely to work in these occupations in other CMA and non-CMA Canada. Lesbians were under-represented in manufacturing, trades and primary industry, although to the least extent in Montreal. For lesbians, there is then a larger middle class tilt in the occupational distribution relative to both heterosexual and gay men.

If the highest paying jobs are located in the private sector, over-representation of sexual minorities in the public sector may limit wage opportunities. For gay men, such over-representation is common everywhere, but especially so in Montreal and non-CMA Canada. For lesbians, relative over-representation is greatest in non-CMA Canada, followed by Vancouver.

Figure 1 presents estimated wage gaps from OLS models controlling demographics, human capital characteristics and occupation and sector of employment. Wage gaps remain across Canada even after taking into account vastly different work situations. These OLS models further show that wage gaps tend to be highest in non-CMA Canada. For gay men, earnings gaps in Montreal, Toronto, and Vancouver are all close to the national estimates. In Toronto, the wage gap is actually lower, but the wage gap for gay men in non-CMA areas is more than double the size. For gay men, living in a large urban area is associated with relatively lower-wage disadvantage. For lesbians, there does not seem to be the same general benefit to living in one of Canada’s largest cities. The national gap is similar to that observed in Vancouver and non-CMA Canada, but wage gaps are lower in Montreal and Toronto for lesbians. Given these patterns, it is likely that the differences are the result of variation in wage inequality within occupations or at similar points in the earnings distribution.

Figure 1

**Sexual Orientation Wage Gaps**

Table 3 presents the interaction of sexual orientation and occupation of employment.[10] The average difference in wage gaps for those in same-sex couples relative to heterosexual individuals in different occupations is calculated by adding the main effect of sexual orientation to the occupation interaction term, with sales and service occupations as the reference category. The penalty for working in low-wage occupations is uniformly the largest in all places. Working in sales and services or manufacturing, primary industry, and trades brings similar wage gaps across CMAs. The penalty for these occupations is highest in non-CMA Canada. In higher-paying occupations, however, there is more variation in how well gay men do. In Montreal, men working in management occupations face the same average wage penalty as gay men working in lower-paying occupations. At the same time, there is only a small difference in average wages in business, finance, and administrative occupations. The exact opposite bears out in Toronto: gay men come close the wage obtained by heterosexual men in management, but face steep disadvantage in business, finance, and administrative occupations. In Vancouver, gay men are able to close the gap in both managerial and finance occupations. In non-CMA Canada, the large gap remains in management, but closes by half in business finance and administrative occupations. In middle class occupations, Vancouver stands out as the only place in which gay men do not reach parity or out-earn heterosexual men working in health and sciences or social science, education and government. In Montreal, gay men working in arts, culture, recreation, and sport have a sizable advantage over heterosexual men.

Table 3

**Interaction Terms of Occupation and Sexual Orientation, OLS Estimates of Log Annual Earnings**

For lesbians, there is considerably less variability in pay penalties in different occupations. In Montreal, the average pay gap is only mitigated in art, culture, recreation and sport, where lesbians enjoy a wage advantage over heterosexual men. In Toronto, there is no statistically or substantively significant pay gap for lesbians in sales and service occupations. Wage gaps emerge at the heights of the earnings distribution—in management, business, finance and administrative, and health and science occupations. In Vancouver, the interaction term is not jointly significant, indicating that lesbians face similar earnings gaps wherever they work. Finally, in other CMA and non-CMA Canada, the wage gap is reduced only in health and sciences and social sciences, education and government.

These patterns are borne out again when we look at the wage gap at different parts of the wage distribution (see Figures 2 and 3). For gay men in Toronto, the largest gaps are observed at the bottom of the earnings distribution, whereas in Vancouver, they are most prevalent in the middle of the distribution. In Montreal, the gap grows at the upper ends of the distribution. In non-CMA Canada, there are gaps everywhere, but the penalty is the largest in lower-paying jobs. For lesbians, on the other hand, in Toronto and Montreal wage gaps tend to be smaller at the bottom end of the distribution and higher at the top. In non-CMA Canada and Vancouver, the gap is highest at the lower end of the distribution, but on the whole fairly consistent across the distribution.

Is public sector employment associated with wage equality between heterosexual and gay and lesbian employees? The answer is overwhelmingly yes. Figures 4 and 5 present OLS wage gaps in the public and private sectors. For gay men, in all areas of Canada, wage gaps are reduced or become not significant in the public sector. Private sector disadvantage actually worsens in Toronto, suggesting that public sector employment there provides an avenue through which gay men are able to overcome aggregate disadvantage (i.e. they out-earn similar heterosexual men). For lesbians, the gap in public sector employment is eliminated in all places, with the exception of Montreal.

Figure 2

**Wage Gaps Across Earning Classes, Gay Men**

How does the view of lesbian women’s labour market experience change when compared to heterosexual women? Table 4 presents estimates from models comparing lesbian women to heterosexual women. We observe wage advantages in Montreal, Toronto, and other CMA and non-CMA Canada, but not in Vancouver. In all areas where there exists a lesbian wage advantage, it is the most pronounced in lower-paying occupational groups, like sales and services, and manufacturing, trades and primary industry. In Montreal, lesbians out-earn heterosexual women in all occupational groups except social science, education and government. Conversely, in Toronto, where the advantage is lower in aggregate, lesbians are actually paid less than heterosexual women in management, business, finance and administrative, and health and science occupations, but earn significantly more in sales and services, arts, culture, recreation and sport, and manufacturing, trades and primary industry. Together, the lesbian wage penalty relative to heterosexual men results from a penalty within higher-paying positions, while the lesbian wage bonus results from higher pay in lower-paid occupations.

Figure 3

**Wage Gaps Across Earning Classes, Lesbian Women**

Discussion and Conclusion

Building on previous findings, this study examined how sexual orientation wage gaps varied across local labour market contexts and organizational settings within local labour markets. We drew attention to the importance of examining within-occupation inequality in each setting in light of previous findings showing that occupational segregation has little traction in explaining sexual orientation wage gaps (Antecol et al., 2008; Waite and Denier, 2015). Additionally, we focused on how organizational settings, like public sector employment, interact with broader social environments to shape wage gaps. Data from the 2006 Census of Canada allowed us to generate estimates that are not endogenous to the adoption of different anti-discrimination legislation across contexts.

Figure 4

**Wage Gaps by Sector of Employment, Gay Men**

Our results lent support to our first question: there are geographic differences in sexual orientation wage gaps, particularly between metropolitan and non-metropolitan Canada. We also documented variation in occupational representation: differences between places are not large, except for greater over-representation in lower-paying sales and services in non-CMA Canada. These results are consistent with recent findings that show at the national level gay men and lesbian women have higher occupational attainment than heterosexual men, and sort into gender atypical occupations (Ueno et al., 2013; Verbakel, 2013). Where they diverge is showing that the extent of this representational advantage varies widely across metropolitan and non-metropolitan areas, and seems to depend on the composition of occupations in the area.

Striking differences in occupational attainment do not explain away the wage disadvantage in any of the labour markets, pointing to differences in inequality within occupations and earnings classes across Canada. Sexual minorities face steep wage penalties in certain labour markets, but relative successes in others. That pay gaps within the same occupational groups vary widely across local labour markets potentially points to the role of firm-level organizational or cultural differences within a labour market in shaping the valuation of sexual orientation in the labour market. We further asked whether the internal organization of public sector employment is associated with wage equality, potentially through rule-based hiring and promotion procedures, which are often designed to comply with federal and provincial anti-discrimination legislation. For both lesbians and gay men, wage gaps are eliminated in the public sector in many parts of Canada. This was true even in non-CMA Canada where private sector wage gaps and within occupation earnings inequality are greatest.

Figure 5

**Wage Gaps by Sector of Employment, Lesbian Women**

There are a number of potential mechanisms that underlie these patterns. One candidate is attitudinal differences towards homosexuality across metropolitan/non-metropolitan areas (Cutler and Jenkins, 2001; McGrane, Berdahl and Bell, 2017). In research on sexual orientation wage gaps, discrimination, whether conscious or unconscious, looms as a possible factor underlying wage penalties. This possibility has received support from both audit studies and legal proceedings, yet remains difficult to measure in population surveys. That we found higher sexual orientation wage gaps, even within similar occupations and earnings-classes, in areas with higher levels of intolerance towards “homosexuals” may bolster claims that sexual minorities are devalued relative to heterosexual men in the Canadian labour market. That pay gaps are reduced in the public sector even where pay gaps are larger in the private sector highlights the role of clear, rule-based hiring and promotion procedures in reducing wage inequality. We found divergent results across sectors, even though sexual orientation is protected in anti-discrimination legislation in Canada, which may suggest that legislation alone may not ensure the equal valuation of sexual minorities. The institutional structure of the public sector provides evidence of potential mechanisms through which pay gaps may be reduced in private sector employment.

Table 4

**OLS Estimates of Log Wage Gaps for Lesbian Women Relative to Heterosexual Women**

Alternatively, couples in all areas may choose to divide work hours and commitment differently amongst household members. More or less egalitarian household specialization may contribute to differences in wage gaps. Some have also suggested that gay men may forgo wage opportunities to enjoy higher amenity cities—although at face value this is the exact opposite of the pattern we find (Black et al., 2002). Gay men have the greatest occupational attainment in the three largest CMAs. It is in largely rural non-CMA Canada where gay men face the least favourable labour market. More precisely identifying and systematically testing these mechanisms is an important avenue for future research.

This study remains limited by available data: at present no Canadian data provide self-reported sexual orientation and earnings measures. The Census data only allow us to identify sexual orientation through partnership with someone of the same-sex. We thus do not know how individuals self-identify and do not have data for individuals who don’t reside with a partner. Should selection into partnership systematically vary by sexual orientation in ways that influence labour market success, our estimates may be biased. Our study also represents a unique historical moment in Canada. The 2006 Census occurred just about a year after the federal recognition of same-sex marriage. It is possible in the decade after that the meaning of marriage and common-law partnership has changed: a wave of people may have gotten married during 2005 in response to the legislation, the legislation may have made marriage more appealing, or the legislation may have even encouraged people to come out. While evidence indicates that same-sex partnership is becoming more common, it remains unclear how selection into partnership has changed over time in ways that would impact our findings. Waite (2015) used data from the 2001 and 2006 Censuses and 2011 NHS and found consistent wage gaps for gay men and lesbian women relative to heterosexual men across all years. We opted not to use the 2011 NHS for this paper since, as a voluntary survey, it was a departure from previous long-form Censuses. This generated some data issues, notably lower response rates in certain communities and provinces (Statistics Canada, 2013). Further, in 2012 Statistics Canada announced that they may have over-counted same-sex couples in some provinces.[11] Given the geographic focus of this paper, and the novelty of the question, we focused on the 2006 data to avoid any data issues that have an explicitly geographic dimension.

Notwithstanding these limitations, the geographic approach highlights the concentration of same-sex couples in large urban centres, the differences in sources of wage disadvantage across areas, and the steeper earnings penalties faced by gay men and lesbians relative to heterosexual men in non-metropolitan areas. Future research should hone in on how local cultural contexts shape earnings inequality throughout Canada.

Summary

Résumé

Resumen

Introduction