Racial wage differentials in developed countries

Discrimination

Various studies have focused on discrimination as a source of racial wage differentials. Theoretical models suggest that minorities may be discriminated against because of negative social attitudes toward them (often referred to as “taste-based” discrimination), or because employers may infer the quality of a job applicant from a minority background based on the perceived average quality of people from that background (often referred to as “statistical discrimination”). Stereotyping, such as a widespread belief that people from a certain racial or ethnic background work less hard than others, may be a reason for employers not offering a job to racial minority applicants, or offering them lower wages, at least initially. Racial stereotypes may even become a self-fulfilling prophecy: Recent research on French supermarket workers suggests that those belonging to a racial minority work less hard if they have racially biased supervisors [2].

From an empirical perspective, the identification and measurement of discrimination is very difficult. Recent evidence suggests that job applicants with minority-sounding names are less likely to be recalled for an interview than similar job applicants with majority-sounding names. However, the evidence of the role that discrimination plays in racial wage differentials, and whether this is due to taste-based or statistical discrimination, is not very strong [3]. The debate on the existence and importance of discrimination and how it operates is still open.

Clearly, policies to reduce racial wage differentials should differ depending on whether discrimination is the result of a dislike of minorities or of lack of information. However, the understanding and elimination of discrimination is not the only avenue for policies that seek to reduce racial wage differentials. It is important to keep in mind that there may be racial wage differentials even in the absence of (wage) discrimination.

An important reason why racial minorities may receive lower wages than white natives is that they might be different in terms of characteristics. In the US, for example, Latinos have lower levels of education than white natives, as do black Caribbean people in the UK. On the other hand, black African Americans in the US and second-generation Pakistanis and Bangladeshis in the UK tend to have high levels of education, but they are also more likely to go to worse schools than white natives, which is in turn likely to have a negative effect on their wages. In addition, racial minorities tend to concentrate in poorly paid occupations.

The empirical evidence consistently shows that racial wage differentials decrease when characteristics are taken into account. However, because empirical analyses can often be constrained by data availability, different studies may take into account a different set of characteristics, thus potentially estimating different—but equally correct—levels of racial wage differentials. The question then arises as to what the various measures of racial wage differentials actually mean, and how they can best be employed for informing policy.

Measuring racial wage differentials

One commonly accepted way to measure racial wage differentials is to use models using only racial-minority type to explain hourly wages. When the models only include the racial-minority type and no other variables, they measure “unadjusted” differentials. Whereas the models measure “adjusted” differentials when they include other variables, such as education or experience. While unadjusted wage differentials can measure the average difference in wages between the majority and the minority group taken as a whole, the adjusted differentials measure the residual difference in average wages across groups, after taking into account some of the relevant characteristics.

When the other characteristics, such as level of education or work experience, are included in the model, then the racial variable can be interpreted as the remaining racial wage differential, which is not due to differences in education and work experience [4]. Some researchers employ a particular approach (known as the “Oaxaca Decomposition”) to measure how much of the racial wage differential remains unexplained. This approach divides racial wage differentials into two components: one which is explained by differences in average characteristics across groups (such as education and experience), and one which is due to differences in average returns to these characteristics. This second component is called the “unexplained” part, since it is not clear why returns to characteristics should differ between minorities and the majority.

Most studies focus on measuring how much the racial wage differential is reduced when certain characteristics are included in the model, and what proportion of the differential remains unexplained [5]. It is important to note, however, that the inclusion of certain characteristics may lead to the estimation of larger (rather than smaller) gaps (see below).

Some researchers interpret the unexplained part of the racial wage differential as “discrimination.” This, however, is incorrect. The unexplained part is a measure of how much, or how little, the chosen characteristics explain the racial wage differentials. The unexplained part includes the possible impact of discrimination, but also the impact of all other factors not included in the model. For example, a part of the reason why racial wage differentials remain unexplained may be due to differences across groups in the occurrence of career breaks, spells of self-employment or unemployment, or other factors that are not measured in the data and cannot therefore be taken into account. Unemployment spells may have a negative impact on wages, and may themselves be the result of discrimination or lack of opportunities. However, they are not a source of discrimination in wages. Since it is not possible to compare the importance of unmeasured causes, the unexplained part should hence not be all attributed to discrimination alone.

Although the role that discrimination may play (directly or indirectly) in shaping racial wage differentials should not be completely discounted, one useful direction for policy would be to focus on those characteristics that have been shown to be important determinants of racial wage differentials, such as education. These could be used as intermediate policy targets.

Possible determinants of racial wage differentials

A comparison of unadjusted and adjusted wage differentials, and an analysis of which characteristics are most relevant in their determination, can be quite revealing and useful for policy. First, some characteristics can contribute more than others. Second, not all characteristics decrease wages (thus contributing to the racial wage differentials): Some characteristics may increase wages and contribute to a reduction of racial wage differentials [6], [7]. For example, the authors of one study find that the most important determinant of racial wage differentials in the UK is concentration in low-pay occupations, while the comparatively higher level of education of minorities partly compensates for the negative effect of occupational segregation [6]. This suggests that the problem may lie in the access to highly paid jobs, or in the lack of career progression, rather than in the lack of high levels of education.

The situation appears to be slightly different in the US, where evidence suggests that racial segregation in the workplace is largely due to segregation by skills and language proficiency, although this seems to be more relevant for Hispanics than for black people [7]. Note, however, that occupational and workplace segregation are slightly different concepts: while workplaces (or employers) may offer jobs in different occupations, the same—or similar—occupations may be available in different workplaces.

The literature rarely compares the relative importance of the various characteristics as determinants of racial wage differentials; instead, various strands of literature focus on one, or a few, possible determinants. The comparison of studies using different data and methodology, which are also designed to address different research questions—some not even focusing on wage differentials—is not straightforward, which means that generalization is not always possible. Nevertheless, by comparing the wider literature it is possible to draw conclusions on what factors may play a role in racial wage differentials, but without being able to directly compare their relevance.

Among the first characteristics considered in the literature are education, qualifications, and skills. Evidence for various countries, including the US, the UK, France, and Germany [1], [7], suggests that racial wage differentials are partly due to lower levels of education of many minority groups, especially when they include a large proportion of immigrants. However, there are important differences across minority groups. While some racial minorities are less qualified than the white majority, others tend to be more qualified. One explanation why racial wage differentials may persist, even after taking into account qualifications, is that minorities are more likely to attend lower-quality schools than the white native population, which may have a negative impact on their overall skill level. Skills, nevertheless, tend to increase with labor market experience. Evidence for Israel, focusing on university graduates, suggests that racial wage differentials tend to decrease with more time spent in the labor market [8].

Another important factor in the determination of racial wage differentials is that minorities tend to segregate in low-pay occupations [6]. Although one study points to the importance of skills and language proficiency (at least for some minorities in the US) for workplace segregation, the causes of occupational segregation have not been systematically analyzed [7]. Economic and sociological theories suggest that racial minorities may prefer to work in occupations where a large proportion of the workers are co-ethnics, and may therefore be prepared to accept comparatively lower wages to work in these occupations. It has also been suggested that, as the size of the minority in a country increases, people from the majority may increasingly fear competition for jobs and other opportunities. The majority would then react by preventing minorities from obtaining high-quality jobs, thereby increasing their political power. There is evidence of this from the US [9], [10]. As the size of the minority increases, and the majority adapts to the presence of minority groups, who in turn become more integrated into the host society, occupational segregation should decrease.

Following this strand of literature, various empirical studies have estimated the impact of the relative size of the various minority groups on their labor market outcomes, although the outcome of interest is often the level of wages of the minority group, rather than wage differentials compared to the majority. In addition, these studies analyze the impact of the size of the minority groups by comparing regions within one country, or neighborhoods within a city, and focus on the impact of residential segregation. These studies tend to find worse labor market outcomes, in terms of wages, for minorities living in more segregated areas in the US [10], although some argue that segregation has a positive impact on wages, once selection across areas is taken into account.

One of the mechanisms that may explain the importance of residential segregation on racial wage differentials is the quality of networks and the types of people that minority and majority groups are more likely to interact with, since these play a relevant role in the type of job people may find. A recent study focusing on the US and Estonia finds that racial wage differentials are larger in areas where people are more likely to choose friends from the same racial background [11]. However, besides being related to co-ethnic networks, this may also be the result of social class [6].

All these, as well as other, factors may contribute to racial wage differentials, and are also likely to interact with each other in a complex manner. In addition, it is likely that different factors may have a different degree of importance for each racial minority. For example, for some minorities, the main issue may be lack of appropriate qualifications, while for other minorities, qualifications may not be an issue and residential location may play a more crucial role. Policy should therefore take into account that the literature has found important differences across and within ethnic minorities: racial wage differentials as well as their determinants are likely to vary across groups. Therefore, adopting a unified approach to reducing racial wage differentials may be inappropriate.

Differences between and within racial minorities

Although nowadays most countries host increasing proportions of racial minorities, either immigrant or from second or further generations, the size of the minority population, the history and size of immigration fluxes, as well as integration policies vary across countries [1], [5]. It is therefore not surprising that racial wage inequalities, even for second generations, vary across countries. The experience of black African Americans in the US is hardly comparable to the experience of second-generation black Africans in the UK. Similarly, because of differences in colonial ties, people of African origin who settled in the UK are likely to be very different from those who settled in France or in Portugal.

More important from a policy perspective is that racial wage differentials vary across groups, even within one country. While some minorities, such as Latinos or Mexicans in the US, experience large wage penalties, others, such as Asians, often receive higher wages than white natives. For example, Figure 1 and Figure 2 shows important differences in racial wage differentials across minority men in France and Germany. In France (Figure 1), the worst outcomes are experienced by Turkish and African people, while immigrants from Northern Europe appear to be paid more than French natives [1]. In Germany (Figure 2), it is immigrants from Italy and Greece who have the worst outcomes, while those from other EU countries appear to perform much better [1]. As shown in the illustration on page 1, in the UK, second-generation white immigrants have the best outcomes overall, while immigrants from Bangladesh and Pakistan have the worst [1]. Among graduates from Israeli universities it is those from Ethiopia who experience the largest wage penalties, while those from the former Soviet Union experience the smallest wage differentials [8].

Besides differences across racial minority groups, there is also considerable heterogeneity within each racial group. An interesting example is that of gender. The literature focusing on men consistently finds large racial wage differentials. In contrast, the literature analyzing racial wage gaps among women finds that, in general, women who belong to a racial minority tend to be paid the same—or even more—than white native women. This finding is consistent across various countries, including the UK, the US, and Israel [1], [3], [8], with some exceptions, such as France and Germany [1]. The smaller racial wage differentials among women may be the result of self-selection: i.e. women, especially those belonging to racial minorities, are often less likely than men to participate in the labor market, and those who do participate are likely to have qualities (such as higher education or motivation) that make them more likely to receive comparatively higher wages. Nevertheless, there is not much evidence on the reasons for smaller racial wage differentials among women than among men.

Most studies estimate gender-specific racial wage differentials, where wages of minority men are compared with wages of white native men, and wages of minority women are compared with wages of white native women. Nevertheless, it is also interesting to see how wages of ethnic-minority men and women compare to wages of white native men. This allows us to measure racial wage differentials and how they interact with gender wage differentials across minority groups.

Figure 3 shows an example for the UK that allows a comparison of wage differentials by race and gender. The figure shows that the gender wage gap for white British women was about 22% in the period 2013–2016. Indian men receive similar wages to white British men, Indian immigrant women (first generation) experience a gender wage gap, which is similar to that of white British women, while British-born Indian women (second generation and further generation) are paid more than white native British women and therefore experience a smaller gender wage gap. Bangladeshi men and women experience large racial wage gaps compared to white native British men and, on average, receive lower wages than white native British women. However, the racial wage gap is smaller among women and there seems to be no gender wage gap among British-born Bangladeshis. In contrast, Bangladeshi immigrant women seem, on average, to receive higher wages than Bangladeshi immigrant men, suggesting that the gender wage gap for this group may even be in favor of women. It is possible that self-selection into the labor market plays an important role for this group. Among black Caribbean immigrants, both men and women are paid on average similarly to white British women, i.e. they both experience racial wage gaps compared to white British men, but the gender wage gap between black Caribbean immigrant men and women is almost non-existent. Black Caribbean men and women, who are British-born, experience only small racial wage gaps and, on average, appear to be paid more than white British women, although still less than white British men.

Besides differences between men and women, there are also important differences between those born abroad and those born in the host country. Although, in general, racial wage differentials tend to be lower for people born in the host country, this is not always the case (e.g. Turkish men in France; see Figure 1). More recent literature also suggests that racial wage differentials may be larger among workers with low skills, while they may be essentially non-existent among high-skill workers, at least in the US [3], and generally tend to vary across the wage distribution and by type of salary. For example, in the US, in jobs paying performance-related wages the white–black wage differential is higher at higher levels of wages, while in jobs where pay is not related to performance the differential is lower at higher levels of wages [12]. In contrast, in the UK the wage differential is smaller in jobs paying performance-related wages than in jobs paying a time rate [13].