Legislative approaches to nondiscrimination at work: a comparative analysis across 13 groups in 193 countries

Background

There is international agreement on the fundamental importance of ending discrimination at work, with commitments to ending discrimination at work embedded in the Sustainable Development Goals (United Nations, 2016), the Universal Declaration of Human Rights and other UN human rights conventions (United Nations General Assembly, 1948, 1965; 1966, 1979; 2008) and fundamental conventions of the International Labour Organization (ILO, 1951). Yet, evidence continues to show that discrimination is a persistent problem around the world that impacts individuals at all stages of their working lives (ILO, 2011). Ending discrimination at work is imperative in its own right and because of the impact work has on income, social network, the nature and quality of individuals' daily experiences and opportunities for individuals to have impact on the communities and societies in which they live. Research suggests that an important first step to ending discrimination at work is by legally banning it (Bassanini and Saint-Martin, 2008). This paper uses a novel data set to identify whether countries have taken this first step to ending discrimination and highlights how data can be used to improve monitoring and accountability around advancing people's opportunities at work.

Existing discrimination in the workplace

There is substantial evidence on the persistence of discrimination in the workplace. Employment discrimination is found across demographic groups and across national borders, affects the ability of individuals to get a job in the first place, to receive adequate pay, to be promoted and to keep a job (Bassanini and Saint-Martin, 2008). Further, we focus on two areas with the most robust literature examining discrimination in countries around the world: hiring and pay. For hiring, researchers have been able to conduct callback studies, sending out large numbers of resumes that vary only based on demographic characteristics and assessing whether there are statistically significant differences in whether applicants are invited to interview for a job (Bertrand and Mullainathan, 2004). For pay, large data sets enable researchers to analyze disparities in pay using regression analysis that allows them to control for confounding factors (Mincer and Polachek, 1974).

Discrimination in hiring

Callback studies have identified discrimination in hiring based on race/ethnicity, religion, gender, age and class. Baert reviewed 37 callback studies in 18 countries across Europe (12 countries), the Americas (Mexico, Peru and the United States) and Asia Pacific (Australia, China and Malaysia) that had a focus on racial/ethnic discrimination [1]. 34 of these studies found discrimination in callback rates for interviews when the person did not have a racial/ethnic majority population name (2018).

Studies in France have found discrimination against Muslim applicants (Adida et al., 2010; Pierné, 2013), but religious discrimination is not limited to any one religion and depends on which group is in the minority. For example, a Greek study found that Pentecostal men and women were 18.7 and 27.2%, respectively, less likely to receive a callback (Drydakis, 2010), and a US study found discrimination against not only Muslims but also atheists, Catholics and pagans (Wright et al., 2013).

Evidence of sex discrimination in hiring affects both men and women. Studies show that women receive fewer callbacks in male-dominated professions (which are commonly higher paid), and men receive fewer callbacks in female-dominated professions (Riach and Rich, 2006; Booth and Leigh, 2010). Callback studies have documented lower callback rates for older women in administrative and sales jobs (Neumark et al., 2019).

Studies have also demonstrated discrimination and lower callback rates based on class. Caste-based studies in India have demonstrated that identical CVs with lower-caste names receive fewer callbacks for call center or entry-level jobs (Banerjee et al., 2009; Thorat et al., 2009). Similarly, a US-based study using combinations of items that signaled social class, such as different sports or personal interests (e.g. polo vs soccer), a financial-aid-based award or extracurricular activities that referenced status as a first-generation college student, found upper-class male applicants were the most likely to receive callbacks (Rivera and Tilcsik, 2016).

Discrimination in pay

Rigorous research has also found discrimination in pay across gender, race/ethnicity, class and sexual orientation. The International Labour Organization has found that globally women earn 60–80 cents on the man's dollar (International Labour Office, 2017). A study released in 2018 found that after human capital differences including education and experience were accounted for, women in Brazil still earned 24% less than men in the formal economy and 20% less in the in the informal economy – due largely to discrimination (Ben Yahmed, 2018). A growing body of literature has also examined intersectionality in the gender pay gap. That is whether the pay gap is not related to gender alone, but rather the intersection of gender and other identities. In this case, research has focused on whether the gender pay gap is driven by women's marital or family status. Evidence of a motherhood pay gap has been frequently found in higher-income countries(Waldfogel, 1998; Budig and England, 2001; Phipps et al., 2001; Anderson et al., 2003; Sigle-Rushton and Waldfogel, 2007), as well as evidence suggesting pay gaps in urban China driven by marital and motherhood status (Zhang, et al., 2008) and a working paper finding a negative relationship between family size and female earnings looking across 21 developing countries (Agüero et al., 2012).

Because of the wide array of racial and ethnic groups around the world and the differences in who experiences discrimination in different countries, there is no single global estimate of wage disparities by race/ethnicity. However, there are country-specific studies. For example, in the United States, Rodgers and Spriggs estimated that 10–14% of the racial and ethnic wage gap for men was due to labor discrimination (Rodgers and Spriggs, 1996), which has been confirmed by numerous other studies (Darity et al., 1996; Neumark, 1998; Coleman, 2003). This is after considering other factors including parental investment in children's education and individual skills, education, tenure, age and metropolitan area.

A UK study found that “class ceiling” contributed to professionals and managers whose parents were working-class earning less than professionals and managers whose parents also worked in managerial or professional jobs (Laurison and Friedman, 2016). Drydakis used data on labor market outcomes in eight OECD countries to find a pay gap between gay and heterosexual men with comparable education and experience (2014).

Discrimination beyond hiring and pay

While the most rigorous evidence has been built around hiring and pay, these are not the only areas of the work where discrimination can negatively impact equal opportunities. Discrimination in promotions and demotions limits workers' abilities to progress in their careers over time. It can also lead to an underreporting of gaps in wages as workers who are otherwise qualified fail to progress to higher paying jobs. In addition to the immediate job and income loss associated with terminations, discrimination in terminations can also have longer-term impacts on career trajectories by increasing workers' time outside the labor force and in some cases reducing future employers' likelihood of hiring a worker due to short tenures at a particular job or too much time between jobs.

Similarly, harassment at work can limit workers’ opportunities to reach their full potential. Among both black and white women, general harassment was the second most common claim reported to the EEOC in the United States from 1998 to 2003 (Ortiz and Roscigno, 2009). Studies in Colombia and the United States have also found a high prevalence of harassment for sexual minorities, particularly compared to their heterosexual peers (Konik and Cortina, 2008; Álvarez et al., 2013). While few studies fully disentangle harassment from other forms of discrimination, evidence of harassment has also been found to be prevalent for workers with intersecting marginalized identities (Berdahl and Moore, 2006; Shaw et al., 2012).

Impact of antidiscrimination laws

While antidiscrimination laws alone do not eliminate discrimination in hiring, pay, promotions or firing, there are studies both across countries and across populations that demonstrate antidiscrimination laws make a difference. In the United States, studies have found that antidiscrimination laws contributed to wage and income increases for African Americans (Donohue and Heckman, 1991; Collins, 2003) and a narrowing of the racial/ethnic pay gap (Chay, 1999). Studies in the United Kingdom and Ireland have also found reductions in the gender pay gap that are attributable to equality in employment legislation (Zabalza and Tzannatos, 1985; Cassidy et al., 2002). Beller found that the equal employment opportunity laws led to a reduction in the gender pay gap of 9.6% over the period of 1967–1974 (1980). Moreover antidiscrimination legislation focused on sexual orientation has been shown to increase work and earnings (Klawitter, 2011; Martell, 2014).

Antidiscrimination legislation has not always been found to be effective. Studies of the Employment Standards Act in Ontario identified gaps in compliance and implementation, particularly for small firms, and did not show any significant impact (Gunderson, 1985; Baker and Fortin, 2004). This highlights the importance of implementation and enforcement alongside legal protections.

Ensuring implementation of strong legislative guarantees to nondiscrimination requires that these guarantees be linked to prohibitions of retaliation against employees who report experiences of workplace discrimination. Evidence shows that having legal protections in place against retaliation may increase reporting rates by reassuring workers that their careers will be protected if they report discrimination (Keenan, 1990; Bergman et al., 2002; Gorod, 2007; Pillay et al., 2018).

Currently, rigorous studies of the impact of antidiscrimination legislation on employment outcomes has been largely limited to higher income countries. More research is needed to assess the impact of these laws in other settings, particularly countries where a large proportion of workers work in the informal economy.

This study

Given the substantial evidence of discrimination across countries and given the powerful evidence that antidiscrimination laws can make a difference, we carried out a study to examine how common antidiscrimination laws were and which groups were best covered. We used international agreements as a framework.

International agreement on ending discrimination at work

As a global community, all UN member states have agreed to the principle of equal opportunities at work for all. The Universal Declaration of Human Rights (UDHR), which was passed in 1948 and binds all UN members clearly specifies that “everyone is entitled to all the rights and freedoms set forth in this declaration, without distinction of any kind, such as race, color, sex, language, religion, political or other opinion, national or social origin, property, birth or other status” (United Nations General Assembly, 1948). Among the rights guaranteed in the UDHR is the right to work. The UDHR was followed by detailed guarantees of equality at work in the International Covenant on Economic, Social and Cultural Rights (United Nations General Assembly, 1966) and a series of conventions delineating in greater detail equal rights and prohibitions of discrimination based on race/ethnicity (United Nations General Assembly, 1965), sex/gender (United Nations General Assembly, 1979) and disability status (United Nations General Assembly, 2008), among others.

Ending discrimination at work is also a core labor standard of the International Labour Organization. All countries agreed in 1998 to four overriding core labor standards to ensure decent work and social justice, including the elimination of discrimination at work (International Labour Organization, 1998), building on prior conventions to eliminate sex discrimination and broader protections from discrimination in any aspect of employment or occupation on the basis of “race, color, sex, religion, political opinion, national extraction or social origin” (International Labour Organization, 1951).

For all 193 UN countries, we examined the extent of antidiscrimination legislation covering hiring, pay, promotions/demotions, terminations and harassment and protections from retaliatory action for reporting harassment. We examined each area for 13 groups. These included those groups explicitly covered by the UN Declaration of Human Rights (gender, race/ethnicity, religion, nationality, social class and political affiliation) [2] and further specific agreements, those covered by separate human rights conventions (disability, migration), and associated areas where there is compelling research evidence of discrimination; sexual orientation and gender identity (addressed under CEAW general comments), marital and family status (associated with gender discrimination) and age (for which a convention is under consideration).

We examined differences in legal framework across region and country income level. In assessing differences across income level, we empirically examine the question of whether legislative differences are due to constraints affecting anticipated ability to implement. This data provides a new tool to monitor national action toward fulfilling international commitments to equal rights and nondiscrimination at work.

Data source

The database was created using labor codes, antidiscrimination legislation, equal opportunity legislation and penal codes covering private sector workplaces in force as of August 2016. Relevant original, full-text national-level legislation was identified using the International Labour Organization (ILO)'s NATLEX database. This database focuses on national-level legislative protections; in countries where antidiscrimination measures are legislated subnationally, we examined legislation for each subnational unit and captured the minimum level of protection when provisions varied across subnational units.

Defining social groups

Terminology used to describe protected groups varies across legislation. We captured protections based on 13 different statuses. Protection from discrimination based on gender was coded when there were protections based on “gender” or “sex” or specific protections for “female” or “women” employees. Protections based on race or ethnicity includes references to “color,” “clan,” “ethnic origin,” “indigenous,” “aboriginal,” “tribe” or “ethnic groups.” Religion includes protections based on “creed,” “religious belief,” “religious opinion,” “religious adherence” or “confession.” Sexual orientation includes references to “sexual preference” and “homosexuality.” Gender identity was coded when legislation referenced “gender expression,” “gender reassignment,” “transgenderism” or “Hijra/Kothi.” Age includes broad references to age and specific nondiscrimination protections for minors, the elderly or individuals above a certain age. Marital status was captured when legislation referenced “civil status,” “married,” “relationship status,” “personal status” or whether employees have a husband or wife. Family status includes references to “parenthood,” “having children,” “single parents,” workers with “family responsibilities” and gender specific references to both “motherhood” and “fatherhood” or “maternity” and “paternity.” Migrant status includes references to “internal migrants,” “foreign migrant workers,” “immigrant status” and “economic migrants.” Foreign national origin includes “ancestry,” “citizenship or origin of parents,” “country or place of birth,” “homeland,” “national descent” or “national origin.” Social class includes references to “caste,” “social background,” “economic standing,” “economic status,” “social condition,” “social origin,” “socioeconomic status” or “disadvantaged.” Political affiliation was coded when legislation referenced “political beliefs,” “political convictions” or “political party affiliation.” Disability includes references to general disability (“handicap,” “impaired” or “special needs”) or specific to mental or physical disabilities.

Defining areas of work

We examined legal protections across five areas to ensure comprehensive protection from discrimination at work: hiring, pay, promotions/demotions, terminations and harassment. Our measure of protection from discrimination in hiring includes whether there are specific guarantees for protection from discrimination for job applicants. Our measure of protection from discrimination in pay includes whether workers are guaranteed equal remuneration, compensation, pay, salary or wages for equal work or equal pay for work of equal value. Our measure of protection from discrimination in promotions/demotions includes whether countries explicitly protect from discrimination in promotions or career advancement or whether there are protections from discrimination in discipline or demotions. Our measure of protection from discrimination in terminations includes guarantees of equal job stability, job security and continuance of employment. Our measure of harassment captures whether workers are protected from actions or behaviors that create a hostile, intimidating or humiliating work environment.

We separately analyzed whether workers are protected from retaliatory action for reporting discrimination at work. Our measure of protection from retaliation captures whether or not laws prohibit of any form of retaliation linked to reporting. This often, but not exclusively, takes the form of prohibitions of dismissal as reprisal for reporting workplace discrimination or harassment.

Defining types of protection

To assess the prevalence of protection for each social group, we first assessed whether there was at least some form of protection for each social group. This measure captured whether workers were explicitly protected from discrimination in any of the aforementioned five specific areas (hiring, pay, promotions/demotions, terminations and harassment) or were broadly protected from discrimination in the workplace based on their social group. Broad protection includes provisions that broadly protect individuals from discrimination based on a particular status in the workplace, but do not explicitly provide protection in any of the areas listed earlier. For example, “there shall be no discrimination on the basis of gender in employment.”

We then assessed the comprehensiveness of protections by counting the number of specific areas for which a group was explicitly protected from discrimination. Countries were categorized as having protections in one of the five specific areas (hiring, pay, promotions/demotions, terminations and harassment) if they explicitly protected the group from discrimination in that specific area. Countries were categorized as having “no protections” if there was no broad protection for the specific group or no explicit protection for the specific group in any of the five areas. For example, “no employer shall discriminate against any prospective employee on the grounds or gender, race, religion, or any other grounds” would be captured as having no protections on the basis of age. In some countries case law or regulations may play an important role in extending coverage to more groups in practice. However, not only does absence of a global repository of relevant case law preclude such a systematic assessment for all countries, these protections are likely to be less permanent and more likely to change than explicit legislative protection.

Analysis

Differences in levels of protection were examined across social groups and areas of work, as well as by country income level and region. We used the McNemar's Chi-square test to assess whether there were statistically significant differences in the percentage of countries with protections for each pair of social groups or each pair of work area. For example, we tested whether there were differences in the proportion of countries guaranteeing at least some protection from discrimination at work based on gender compared to disability status. We separately compared each social group to the other 12 social groups in this paper. We also calculated the percentage of countries that protect workers from retaliation when reporting discrimination at work and used the McNemar's Chi-square test of significance to test for differences across social groups. Finally, we examined differences in protection for each social group by country income level, using the Pearson's Chi-square statistics to test for significance. For example, we tested whether there were differences in protection based on gender for low-income, middle-income and high-income countries.

Country income level and region were categorized according to the World Bank's country and lending groups as of 2016 (World Bank, 2016) [3]. All analyses were conducted in Stata 14.

Some form of protection against discrimination

Nearly all (89%) countries guarantee some protection from discrimination in the workplace on the basis of gender, a significantly greater percentage than countries with some protection for the other social groups (p < 0.01). Around three-quarters legislate some protection on the basis of disability (79%), religion (77%) and race/ethnicity (76%), with no significant difference between the proportions of countries offering some protection to these three groups (Figure 1).

Legal protections from discrimination at work based on migrant status, foreign national origin, gender identity or sexual orientation are significantly less common (p < 0.01). Less than 40% of countries in the world provide any form of protection from discrimination at work for migrants (38%, p < 0.01) and workers of different foreign national origins (38%, p < 0.01). Less than a third of countries (32%, p < 0.01) provide at least some protection from discrimination at work on the basis of sexual orientation and far fewer (10%) do so on the basis of gender identity. Protections from discrimination based on gender identity are significantly lower than protections for every other group (p < 0.01).

While the majority of countries provide at least some protection against discrimination based on gender in the workplace (89%), significantly fewer provide some protection on the basis of identities that often intersect with gender: marital status (49%, p < 0.01) and family status (38%, p < 0.01).

Comprehensive protection from discrimination

While a majority of countries provide at least some protection from discrimination at work, only a minority of countries provided comprehensive protections by explicitly protecting all five work areas (hiring, pay, promotions/demotions, terminations and harassment) to any group. Here too, the groups that were most likely to receive comprehensive protections from discrimination were those based on gender (23%), disability (19%), religion (17%) and race/ethnicity (17%). Although the prevalence of at least some protection from discrimination at work based on sexual orientation was among the lowest, it is among the most comprehensively protected with 16% of countries guaranteeing protections in all five work areas. Together these findings indicate that while countries are less likely to protect from discrimination based on sexual orientation, those that do are more likely to provide comprehensive protections (Table 1).

Areas most likely to receive protection

While 65% countries guarantee nondiscrimination based on gender in hiring, significantly more, 82%, guarantee equal pay for equal work or equal pay for work of equal value (p < 0.01). Workers with disabilities are protected from discrimination in hiring (61%) significantly more often than pay (46%, p < 0.01) or terminations (55%, p = 0.05). However, no statistically significant differences are seen between protections for hiring and pay or hiring and terminations for discrimination based on religion, political affiliation, social class or migrant status (Table 2).

For each social group, significantly fewer countries protect from discrimination in promotion and/or demotion and from workplace harassment than countries with protection in hiring (p < 0.02). The only exception is for sexual orientation and gender identity based discrimination where only a few countries guarantee protection, but those that do are likely to provide widespread protection.

Protection from discrimination and country income

For most social groups, low-, middle - and high-income countries offered similar levels of protection, suggesting that legally protecting individuals from discrimination at work is not a question of a country's available resources. For example, 93% of high-income countries explicitly protect from gender discrimination in at least one work area compared to 87% of low-income and 87% of middle-income countries. High-income countries, however, were significantly more likely to protect against discrimination based on sexual orientation (58% compared to 25% of middle-income and 6% of low-income, p < 0.01 for both) and gender identity (25% compared to 5% of middle-income and 3% of low-income, p < 0.01 and p = 0.01, respectively). The reverse was true in the case of social class and political affiliation, where significantly more low- and middle-income countries have protections from discrimination than high-income countries (social class: 71% of low-income and 70% of middle-income compared to 35% of high-income, p < 0.01; political affiliation: 77% of low-income and 65% of middle-income compared to 51% of high-income, p = 0.02 and p = 0.07, respectively). For marital status, low-income countries were less likely to explicitly protect at least one area (39%) compared to high-income countries (58%, p = 0.09) (Table 3).

Protection from discrimination and country region

Differences in levels of protections were more evident across country region than by income. When looking at how many groups countries prohibited discrimination for in at least one area, protections were most common in Europe and Central Asia. No country in Europe and Central Asia failed to protect all groups compared to 2% in the Americas, 4% in Sub-Saharan Africa, 11% in the Middle East and North Africa, 12% in South Asia and 19% in East Asia and Pacific (see Plate 1). However, in every region, at least one country protected the vast majority of groups (10–13), suggesting that more comprehensive protections from discrimination at work are feasible across a range of settings. More than half of countries in Europe and Central Asia protected 10–13 groups compared to 31% in the Americas, 21% in Sub-Saharan Africa, 16% in East Asia and Pacific, 12% in South Asia and 11% in the Middle East and North Africa.

Looking at protections for specific groups by region, while all or nearly all countries prohibited gender discrimination in at least one work area in the Americas (97%), Europe and Central Asia (100%), the Middle East and North Africa (89%) and Sub-Saharan Africa (92%), protections were significantly less in other regions. Only half of countries banned gender discrimination in South Asia (p < 0.01 compared to Europe and Central Asia) and only three-fifths did so in East Asia and Pacific (63%, p < 0.01 compared to Europe and Central Asia) (see Table 4). While gender discrimination was generally the most commonly protected status, in East Asia and Pacific and South Asia, protections based on disability were more common (67 and 63%, respectively). Other regional exceptions to global trends included social class and political affiliation being covered nearly as frequently in Sub-Saharan Africa (81%) as protections based on race/ethnicity (85%).

Retaliation

For most social groups, a majority of countries do not pair protection from discrimination with a prohibition of retaliation for reporting workplace discrimination (Table 5). For example, 79% of countries protect workers from discrimination based on disability; however, among these countries that make discrimination on the basis of disability illegal, less than half (49%) protect workers who report discrimination from retaliation. In contrast, for the group with the lowest level of protection from discrimination (gender identity), nearly all countries that extend protection from discrimination pair this protection with a prohibition against retaliation for reporting.

Global agreements and policy implications

While further research is needed that focuses on implementation mechanisms, this study highlights a series of areas that are important for policy action. It is important to work toward universal protections. 187 UN member states have ratified the Convention on the Elimination of Discrimination Against Women (United Nations General Assembly, 1979), yet 19 countries that are parties still do not have national laws preventing gender discrimination in the workplace. The same can be said for the Convention on the Rights of Persons with Disabilities (United Nations General Assembly, 2008). 157 countries had ratified the convention by the end of 2015, yet 26 states parties did not have laws prohibiting discrimination on the basis of disability of the workplace in 2016. This disparity between commitments countries have made through international agreements and the extent to which they have passed the national laws to put these commitments into practice needs to be addressed. Moreover, countries that provide partial protection (covering only certain aspects of work) should move to full protection.

Economy-wide benefits

There is strong evidence that reducing discrimination also benefits the economy as a whole. Discrimination violates fundamental principles that all people should be treated equally and violates all global human rights agreements; it is also economically inefficient. Discrimination leads to having fewer people in the workforce and leads to selection for positions on bases other than merit. Both reduce productivity. These principles hold across groups.

The most work estimating the size of the economic loss due to discrimination has occurred in the area of gender. A series of studies have been carried out by leading consulting firms such as McKinsey, by banks including Goldman Sachs and by academics. These studies have found that gender differences in labor force participation cost the global economy as much as $28 trillion (Woetzel et al., 2015, p. 2).

Further steps

While legal protections against discrimination are the first step, it is crucial that laws contain mechanisms that support successful implementation. Important areas for future study include among others: examining the mechanisms that laws around the world contain regarding implementation, responding to cases of discrimination and monitoring. Understanding whether legislation and enforcement mechanisms protect workers from discrimination on the basis of two or more intersecting identities is also critical to ensuring that the most marginalized workers have full protection under the law. Evaluating the extent to which different legal approaches taken by countries influence their effectiveness is also greatly needed.

Transparency can play an important role in accelerating equality in the workplace. Each global agreement should be linked to maps that allow citizens around the world to readily see whether their country has not only ratified the relevant international agreements to prevent discrimination on the basis of gender, disability, race ethnicity religion or other characteristics, but also whether their country has passed the national laws that prevent discrimination from being experienced in central areas of life such as work or education.

The global agreements containing guarantees of equal rights for all human beings have the potential to be transformational but only if they are linked to clear action within countries. That action is far more likely to occur if transparency supports citizens and civil society alike rewarding policymakers who act in concordance with these agreements and holding accountable those policymakers who have yet to take the needed steps.