OIT - Horizontal inequalities in the labour market - A transnational review of the employment and wage outcomes of vulnerable groups

 ILO Brief 1 Horizontal inequalities in the labour market: A transnational review of the employment and wage outcomes of vulnerable groups  Research Brief May 2025 Horizontal inequalities in the labour market: A transnational review of the employment and wage outcomes of vulnerable groups Brief based on ILO Working Papers 107, 124, 131 and 141 1

 On the basis of survey data from the ILO Harmonized Microdata collection covering more than 160 countries at different levels of development, employment and wage outcomes were analysed for four specific vulnerable population groups: people living in rural areas, people with disabilities, persons living with a samesex partner and second-generation migrants. In total, survey microdata from 58, 65, 9 and 32 countries respectively were used to study each group.  The methodologies applied were designed to compare the labour market outcomes of each group studied with those of a reference group in the rest of the population, namely urban dwellers, people without disabilities, people in heterosexual relationships and the population without a migration background. Potential differences in socio-demographic composition between the groups studied and their reference counterparts were taken into account, especially with regard to age, education and occupational category.  Across countries, different employment patterns are observed, depending on the group in question. In particular, people with disabilities, second-generation migrants and coupled gay men are less likely to participate in the labour force and more likely to be unemployed than their counterparts of similar age and educational attainment. In contrast, coupled lesbians are more likely to be active in the labour market than comparable heterosexual women.  Most of the groups studied experience wage penalties that cannot be fully explained by age, education or occupational category. Rural employees earn 24 per cent less than their urban counterparts, with only half of the gap being explicable by differences in those socio-demographic characteristics. For people with disabilities, the pay gap vis-à-

that cannot be fully explained by age, education or occupational category. Rural employees earn 24 per cent less than their urban counterparts, with only half of the gap being explicable by differences in those socio-demographic characteristics. For people with disabilities, the pay gap vis-à- vis other workers is 12 per cent, with only a quarter of that gap being explicable. Coupled gay men tend to earn lower wages than their heterosexual counterparts with similar socio-demographic characteristics – up to 15 per cent less in the countries with statistically significant results. Conversely, coupled lesbians appear to enjoy a wage premium in many of the countries studied. The results are more country-specific for second-generation migrants, who were found to enjoy a wage premium in the United States of America but to face a wage penalty in many European countries.  International labour standards provide guidance on how to improve the employment and earnings outcomes of vulnerable groups. Relevant instruments in this regard include the Discrimination (Employment and Occupation) Convention (No. 111) and Recommendation (No. 111), 1958. Labour institutions such as minimum wages and collective bargaining are also very important in addressing the wage disadvantage faced by some groups.  Further research is needed to fully understand the observed gaps and country-specific differences, optimally involving the regular collection of data on labour market outcomes and a more systematic identification of vulnerable groups in surveys.

1 This brief has been prepared by Sevane Ananian, co-author with Giulia Dellaferrera, Lena Hassani -Nezhad, Utsoree Das and Nick Drydakis, of the research presented in ILO Working Papers 107, 124, 131 and 141. Key points ILO Brief 2 Horizontal inequalities in the labour market: A transnational review of the employment and wage outcomes of vulnerable groups  Introduction The ILO report Inequalities and the World of Work, submitted to the International Labour Conference at its

Key points ILO Brief 2 Horizontal inequalities in the labour market: A transnational review of the employment and wage outcomes of vulnerable groups  Introduction The ILO report Inequalities and the World of Work, submitted to the International Labour Conference at its 109th Session in 2021, stressed the relevance of assessing “horizontal” inequalities in the world of work as part of the overall diagnosis of inequalities (ILO 2021). The adjective “horizontal” referring to disparities observed across different groups in society,2 such inequalities occur when some groups find themselves disadvantaged and discriminated against on the basis of their visible identity, which includes such characteristics as gender, ethnicity, region and beliefs (Stewart 2016). Addressing horizontal inequalities in the world of work would not only help to promote the labour market inclusion of individuals from the population groups concerned, but also generate wider economic and social benefits. For instance, it has been estimated that within developing countries, where around 80 per cent of the world’s 1 billion people with disabilities live, approximately 7 per cent of gross domestic product is lost because of their exclusion from the labour market (ILO 2015). With regard to gender and ethnic background, a recent study focusing on the United States suggests that between 20 and 40 per cent of the growth in US gross domestic product per capita between 1960 and 2010 can be attributed to the reduction of the barriers faced by women and black men in the labour market (Hsieh et al. 2019). Findings such as these illustrate the benefits that can be expected for society as a whole from the implementation of policies aimed at strengthening the inclusiveness of labour markets. The increasing attention commanded by horizontal inequalities is reflected in a range of international instruments and commitments undertaken in recent years. Thus, the Sustainable Development Goals (SDGs) adopted by the United Nations General Assembly in 2015 feature several Goals and associated targets that specifically address eradicating the disparities faced by

inequalities is reflected in a range of international instruments and commitments undertaken in recent years. Thus, the Sustainable Development Goals (SDGs) adopted by the United Nations General Assembly in 2015 feature several Goals and associated targets that specifically address eradicating the disparities faced by different groups in society. In particular, SDG 5 calls for gender equality, while SDG 10 calls for inequality to be reduced within and among countries, including through “the social, economic and political inclusion of all, irrespective of age, sex, disability, race, ethnicity, origin,

2 In contrast to “vertical” inequalities, which are between individuals or households. religion or economic or other status” (SDG target 10.2). Addressing the aforementioned disparities requires the implementation of adequate policies and measures. The fact that these groups still suffer from economic inequality, even in high-income countries, is a sign that economic development alone is not enough to overcome the challenges faced by people in certain categories. Under the ILO framework, the Resolution concerning inequalities and the world of work adopted by the International Labour Conference at its 109th session in 2021 calls for urgent action and the implementation by the ILO of an integrated strategy to reduce inequalities. As part of such efforts, the Office has, in particular, been asked to conduct research and expand knowledge on inequalities. Quantifying disparities in the world of work and identifying their main drivers are key to the design of effective policies for tackling them. Global analyses of the labour market outcomes of individuals from vulnerable groups remain scarce. Moreover, the studies that do exist rarely seek to disentangle the various factors that may explain these outcomes, including the specificities of each group in terms of human capital characteristics. With the above considerations in mind, this brief presents the findings of a research project that examined the labour force participation, unemployment, status in employment and wages of four vulnerable groups across countries, namely, people living in rural areas, people with disabilities, persons

of human capital characteristics. With the above considerations in mind, this brief presents the findings of a research project that examined the labour force participation, unemployment, status in employment and wages of four vulnerable groups across countries, namely, people living in rural areas, people with disabilities, persons living with a same-sex partner and second-generation migrants. As part of the analysis, efforts were made to take into account the specific socio‑demographic composition of each group studied, especially with regard to education and experience. The detailed findings for each of the four groups were presented in ILO Working Papers 107, 124, 131 and 141 (Ananian and Dellaferrera 2024a, 2024b; Hassani-Nezhad, Ananian and Drydakis 2024; Das, Dellaferrera and Ananian 2025). This research was made possible by the national household survey data compiled by the ILO for a large number of countries as part of its Harmonized Microdata collection. Indeed, thanks to this collection it is becoming increasingly feasible to explore key labour market issues across regions and countries at different levels of development. ILO Brief 3 Horizontal inequalities in the labour market: A transnational review of the employment and wage outcomes of vulnerable groups  Identifying people living in rural areas, people with disabilities, persons living with a same-sex partner and second-generation migrants among the respondents of household surveys People living in rural areas were identified according to national definitions of rural areas.1 There are differences between countries in the way that they define rural areas, in particular with regard to the criteria applied to determine whether an area is integrated into an urban unit. Nevertheless, the available evidence suggests that the various national definitions of rural (and urban) areas are broadly consistent. The statistical definition of people with disabilities is guided by the International Classification of Functioning, Disability and Health, which considers impairments, activity limitations and participation restrictions under the umbrella term “disability”. In surveys, the identification of people with disabilities typically relies on respondents

national definitions of rural (and urban) areas are broadly consistent. The statistical definition of people with disabilities is guided by the International Classification of Functioning, Disability and Health, which considers impairments, activity limitations and participation restrictions under the umbrella term “disability”. In surveys, the identification of people with disabilities typically relies on respondents answering at least six key questions on whether they experience difficulties in seeing, hearing, walking, remembering, communicating and caring for oneself. This is based on the Washington Group short set of questions on functioning, where the response options are: “No – no difficulty”, “Yes – some difficulty”, “Yes – a lot of difficulty” and “Cannot do it at all”.2 Accordingly, individuals with disabilities were defined as those who reported having a lot of difficulties with, or not being able to manage at all, at least one of the aforementioned domains of functioning.3 The identification of persons living with a same-sex partner was based on individuals being part of a couple classified as a same-sex relationship, which is when the household head and the reported partner are both of the same sex. This approach proved to be quite accurate when it was also applied to a subsample of surveys that featured a direct identifier for same-sex partnership. Second-generation migrants were defined as individuals whose country of birth was the same as the host country, but whose mother was not born in the host country. Robustness checks were performed by applying the same criterion to the father's country of birth instead of the mother's and these yielded similar results overall. 1 In the harmonized microdata sets created by the ILO from national surveys, the name of the variable used to identify rural ar eas is “ilo_geo”, which corresponds to “geographical coverage”. 2 Where that is not possible, four core questions are used (on difficulties in seeing, hearing, walking and remembering). 3 The ILO harmonized variable “ilo_dsb_aggregate”, which denotes “disability status (aggregate)”, was used to identify people w ith disabilities in the surveys selected for analysis.  Methodology For each group studied, a sample of surveys was used containing information on individuals’ employment and wage outcomes

surveys selected for analysis.  Methodology For each group studied, a sample of surveys was used containing information on individuals’ employment and wage outcomes The empirical analysis of each vulnerable group’s labour market outcomes is based on microdata from a selection of labour force surveys and other household surveys conducted in both developed and developing countries.

3 For more details, see https://ilostat.ilo.org/about/data-collection-and-production/. 4 In surveys, data on income from self-employment tend to be collected less frequently and are usually of lower quality than data on wages. Only estimates related to wages are therefore reported in this brief. However, ILO Working Papers 107, 124 and 141 present estimates on self-employment income as well for a limited set of countries. All the countries studied were picked on the basis of whether a survey with relevant data was available in the ILO Harmonized Microdata collection, which now covers more than 160 countries. 3 The surveys selected provide information on the labour force participation and employment status of household members and on employees’ wages (albeit not for all countries). When analysing wage outcomes, the decision was taken to focus on gross hourly wages in order to eliminate variations in wages due to differences in working time.4 However, insights into the working time of each of the groups studied can be found in ILO Working Papers 107, 124, 131 and 141. ILO Brief 4 Horizontal inequalities in the labour market: A transnational review of the employment and wage outcomes of vulnerable groups The surveys selected also had to contain information that made it possible to identify respondents as belonging to the specific group of interest. The box below describes the identification strategy used for each group. Analysing differences in labour market outcomes between the group of interest and the rest of the population Individuals from each of the groups studied may differ from the rest of the population in terms of additional characteristics that have an impact on their labour market

identification strategy used for each group. Analysing differences in labour market outcomes between the group of interest and the rest of the population Individuals from each of the groups studied may differ from the rest of the population in terms of additional characteristics that have an impact on their labour market outcomes. For instance, the over-representation of older people among people with disabilities may result in lower labour market participation rates. Similarly, significant differences have been observed in educational attainment between urban and rural populations, which could partly explain the specificities of rural dwellers as a group when looking at labour market indicators based on raw data. It is therefore essential to analyse labour market outcomes using methods that are designed to take into account the socio-demographic composition of each group, especially with regard to the determinants of human capital. Such methods involve comparing individuals from the group of interest with people who have similar observed characteristics but who do not belong to that group. For this research project, the econometric technique of logistic regression was used, with age and education as control variables, to estimate the likelihood of labour market participation and unemployment of each group studied. When examining status in employment (that is, working as an employee versus being self-employed), broad occupational category was also included as a control variable. The crucial analysis of hourly earnings looked at the average pay gap between employees from the group of interest and the rest of employees. In particular, the aim was to estimate the gap in earnings that cannot be explained by factors shaping human capital, such as education and experience. To that end, two econometric techniques were applied, depending on the group under consideration: (a) the Blinder–Oaxaca decomposition,

5 See each of the aforementioned ILO Working Papers for the full specifications of the econometric models used to analyse wages . which decomposes the average wage gap into a part explained by socio-demographic characteristics such as age and education, and an unexplained part; and (b) the

5 See each of the aforementioned ILO Working Papers for the full specifications of the econometric models used to analyse wages . which decomposes the average wage gap into a part explained by socio-demographic characteristics such as age and education, and an unexplained part; and (b) the Mincer regression model, which can be used to estimate how much being part of the group under consideration affects wages on average, taking into account sociodemographic characteristics. In addition to age and education, the broad occupational category of individuals was also included in the socio-demographic characteristics taken into account in the econometric models constructed for the analysis of wage outcomes.5 While the reference groups against which the labour market outcomes of people living in rural areas, people with disabilities and second-generation migrants were compared did not exclude anyone on the basis of relationship status, members of gay and lesbian couples were compared to, respectively, men and women in heterosexual relationships (married or unmarried). Furthermore, the reference group for second-generation migrants did not include immigrants (that is, people born abroad), as their experiences and the challenges they face when it comes to labour market integration are likely to be very different. Finally, it should be noted that the scope of the analysis varies for each vulnerable group, depending on data availability and relevance. The studies on people living in rural areas and people with disabilities covered countries from all over the world and at different levels of development. In contrast, the study dealing with secondgeneration migrants was restricted to European countries, Australia and the United States, while the study on persons living with a same-sex partner looked only at nine countries in the Americas, Asia and Europe. For people living with same-sex partners, it was deliberately decided to examine several developing countries given that most previous studies have focused on developed countries. However, the sample was designed to include only countries where same-sex sexual activity is legal, which considerably limited the sample size. Table 1 below shows the number of countries included in the analysis for each vulnerable group, also broken down

previous studies have focused on developed countries. However, the sample was designed to include only countries where same-sex sexual activity is legal, which considerably limited the sample size. Table 1 below shows the number of countries included in the analysis for each vulnerable group, also broken down by region and level of development. ILO Brief 5 Horizontal inequalities in the labour market: A transnational review of the employment and wage outcomes of vulnerable groups  Table 1. Number of countries studied for each vulnerable group Group Number of countries in the overall analysis Countries included in the analysis of wages Africa Americas Arab states Asia and the Pacific Europe and Central Asia Middle and lowincome People living in rural areas 58 58 18 16 3 14 7 50 People with disabilities 65 30 14 7 2 12 30 37 Coupled gay men 9 9 0 7 0 1 1 6 Coupled lesbians 9 9 0 7 0 1 1 6 Secondgeneration migrants 32 29 0 1 0 1 30 2

Sources: Ananian and Dellaferrera (2024a, 2024b); Hassani-Nezhad, Ananian and Drydakis (2024); Das, Dellaferrera and Ananian (2025).  Main findings Labour force participation is lower for people with disabilities, second-generation migrants and coupled gay men On average across the countries studied for each group, people with disabilities, second-generation migrants and coupled gay men are less likely to participate in the labour force than their counterparts, that is, respectively, people without disabilities, those with native-born parents and

second-generation migrants and coupled gay men On average across the countries studied for each group, people with disabilities, second-generation migrants and coupled gay men are less likely to participate in the labour force than their counterparts, that is, respectively, people without disabilities, those with native-born parents and men in heterosexual couples (table 2). This pattern holds for each country income group considered in the case of people with disabilities, who are on average 25 percentage points less likely to participate in the labour force than their counterparts without disabilities. Second-generation migrants are, on average, 7.5 percentage points less likely to be active in the labour market than other native-born persons across the United States, Australia and all the European countries in the

6 Across the four country income groups considered, a statistically significant positive relationship between having a disability and being unemployed was absent only for low-income countries. On the other hand, a statistically significant positive relationship between being a coupled gay man and being unemployed was found in five of the eight countries for which estimates could be computed . Finally, being a second -generation migrant was found to significantly increase the likelihood of being unemployed in the United States and, on average, in Europe, this being the cas e in 12 countries studied there. sample. For men living with a same-sex partner, a statistically significant lower probability of participating in the labour force is observed in all nine countries studied. These findings suggest that belonging to one of the three aforementioned groups has a consistently strong effect on an individual's overall labour market experience, simply by virtue of being associated with lower labour force participation. On the other hand, people living in rural areas and coupled lesbian women appear to be more active in the labour market than urban dwellers and women in heterosexual couples respectively, almost certainly for very different reasons. In the first case, this is likely to be due to the overall income gap between rural and urban areas and the greater incentives for rural inhabitants to take any job in order to improve their situation. Coupled

heterosexual couples respectively, almost certainly for very different reasons. In the first case, this is likely to be due to the overall income gap between rural and urban areas and the greater incentives for rural inhabitants to take any job in order to improve their situation. Coupled lesbians’ higher probability of participating in the labour force may be partly ascribed to the fact that same-sex couples do not follow the same patterns of specialization based on the comparative advantages of partners that are typically observed within heterosexual couples (Siminski and Yetsenga 2022; Hofmarcher and Plug 2022; Martell and Roncolato 2016). Conversely, this could also explain to some extent the lower labour force participation of men living with a same-sex partner. In most of the groups studied, unemployment patterns mirror those observed for labour force participation For most of the vulnerable groups studied, trends in unemployment follow a similar pattern to labour force participation, leading to a cumulative effect on employment (table 2). In particular, people with disabilities, second-generation migrants and men in gay relationships appear to be more likely to be unemployed than their counterparts in most cases.6 Discrimination is one of the reasons put forward to explain the higher unemployment rates observed for these groups. In addition to behaviours that merely involve avoiding the recruitment of individuals from ILO Brief 6 Horizontal inequalities in the labour market: A transnational review of the employment and wage outcomes of vulnerable groups groups that are disapproved of (known as “taste-based discrimination”), discrimination can also stem from employers’ beliefs regarding the productivity of job applicants, particularly when imperfect information is available about their labour market characteristics. Such discrimination likewise occurs when employers have ill-founded views about the cost of workers from certain groups. For instance, the cost of providing appropriate workplace accommodations for people with disabilities is frequently overstated by employers. With regard to individuals in same-sex relationships, studies have shown that some employers may perceive homosexuality as

ill-founded views about the cost of workers from certain groups. For instance, the cost of providing appropriate workplace accommodations for people with disabilities is frequently overstated by employers. With regard to individuals in same-sex relationships, studies have shown that some employers may perceive homosexuality as carrying a higher risk of HIV infection, which they associate with reduced productivity and increased absenteeism. Unemployment patterns among coupled lesbians do not appear to be as clear-cut as those observed for labour force participation. This is in line with several previous studies that have reported mixed findings for women in lesbian relationships. Certain trends in employment status are also noticeable for some groups Patterns of status in employment can be observed for some groups, even after controlling for occupational category. On average across countries, people living in rural areas and those with disabilities are more likely to be self-employed than their counterparts living in urban areas and those without disabilities, while secondgeneration migrants are more likely to be wage employees than people with native-born parents. For persons living with a same-sex partner, the findings appear to vary considerably across countries. The reasons prompting people to take up either wage employment or self-employment may in some cases be related to the specific circumstances of the group to which they belong. For instance, studies have highlighted that people with disabilities tend to choose to be self-employed because of the greater flexibility and the better balance between disability status and working life that self-employment may facilitate. Employer discrimination, “pushing” workers with disabilities into self-employment, is another possible explanatory factor. On the other hand, the considerable role played by social networks in achieving success as an entrepreneur has been advanced as a reason for the tendency of second-generation migrants to seek wage employment instead.  Table 2. Effect of belonging to a vulnerable group on the likelihood of labour force participation, unemployment and wage employment (versus self-employment) Group Labour force participation Unemployment Wage employment People living in

migrants to seek wage employment instead.  Table 2. Effect of belonging to a vulnerable group on the likelihood of labour force participation, unemployment and wage employment (versus self-employment) Group Labour force participation Unemployment Wage employment People living in rural areas + - - People with disabilities - + - Coupled gay men - + Varies across countries Coupled lesbians + Varies across countries Varies across countries Secondgeneration migrants - + + Note: Average effects across countries are reported in this table, except for persons in a same-sex relationship: for these, the most frequent trends across the nine countries studied are reported.

Sources: Ananian and Dellaferrera (2024a, 2024b); Hassani-Nezhad, Ananian and Drydakis (2024); Das, Dellaferrera and Ananian (2025).

People with disabilities, those living in rural areas and coupled gay men tend to earn lower hourly wages Substantial pay gaps were observed between most of the vulnerable groups studied and the rest of the population, even after controlling for their specific socio-demographic composition. Specifically, employees living in rural areas earn on average 24 per cent less than their urban counterparts across the sampled countries, with only half of that gap being attributable to rural–urban discrepancies in education, experience and occupational category (figure 1). The average pay gap between people with and without disabilities is 12 per cent, with three quarters of that gap unexplained (figure 2). Finally, in three of the nine countries studied, coupled gay men face a wage penalty of between 10 and 15 per cent after controlling for socio-demographic factors, while in only one country do they appear to enjoy a wage premium relative to their heterosexual counterparts (figure 3a). Arriving at a clear delineation of the factors behind the pay disparities experienced by these groups of