1. Diversity in OECD countries: Population diversity, labour market inclusion and acceptance of diversity

OECD countries and, by extension, their labour force, have become considerably more diverse in a rather short timeframe. Over the past two decades, the labour market participation of women has increased strongly; the population shares of migrants and their children are growing in almost all OECD countries and more LGBT people are open about their sexual orientation. These societal changes are occurring against the backdrop of ageing societies and a larger share of older workers than in past decades.

Ensuring that these groups are included in the labour market is therefore a key policy concern, not only for ethical reasons, but also in terms of economic development and social cohesion. In this context, the office of the OECD Secretary General allocated resources of the Central Priorities Fund (CPF) to assess how OECD countries can be equipped to make the most out of diversity and ensure equality of opportunity. Based on long-standing work by the OECD on groups traditionally disadvantaged in the labour market, a policy questionnaire among OECD countries on diversity policies as well as an online survey among HR professionals, this report addresses the question of how governments and businesses can set the conditions to make the most out of a diverse workforce.

The term ‘diversity’ is used as an umbrella term and in this report focuses on five overlapping groups that are widely seen as being disadvantaged and discriminated against in the labour market; women; immigrants, their descendants and ethnic minorities; LGBT people; older people; and people with disabilities. By its very nature, the choice of groups to include is challenging. Young people, for example, are disadvantaged in many countries, but are not included here since they are rarely included in national legislation listing groups against whom discrimination is prohibited, whereas the five groups listed above are, in most cases. Furthermore, while it may seem surprising to classify women as a diverse group – given that they represent half of the population – they are nevertheless included in this overview, reflecting both persisting inequalities in the labour market as well as the fact that diversity policies have long focused on providing equal opportunities for women. Finally, while this report analyses existing policies targeting certain groups that share common traits, it is equally important to acknowledge the influence that interactions of multiple advantages and disadvantages can have on individuals’ outcomes.

The report first discusses under which conditions diversity in the workforce may present a ‘business case’ for firms or the economy at large and presents evidence on how diversity can affect social cohesion more broadly. Chapter 3 shows how population diversity and labour market participation of women and minority groups in OECD countries have evolved over time. In addition, it provides an overview of attitudes towards diversity, equality and diversity policies and demonstrates how they have evolved. Chapter 4 evaluates some of the key existing diversity instruments, including both public policies and corporate practices, connects it with evidence on their effectiveness and, where possible, discusses differential effects on groups. Chapter 5 highlights the key challenges faced by governments and employers and Chapter 6 concludes.

Diversity is a broad term and includes many groups for which data are not available or are difficult to compare across countries or, in many cases, between different government agencies and data sources (e.g. household surveys, administrative data) in the same country. This is notably the case for ethnic identity (Box 1.1) as well as for LGBT people. Only a few population-based surveys include questions on sexual orientation and even within a given country, shares can differ markedly depending on whether survey questions ask for sexual self-identification or sexual behaviour and whether they are administered online or face-to-face (Valfort, 2017[1]). For example, estimates of the LGB population in the United States vary between 2.8% and 5.6%. For gender identity, Chile, Denmark and the United States have started to implement representative surveys on the transgender population (Balestra and Fleischer, 2018[2]). The following therefore looks at four dimensions of population diversity that can be more easily defined and compared across countries – the share of foreign-born, natives with at least one immigrant parent, older people, and people with disabilities (Table 1.1).

Table 1.1 shows considerable differences in population diversity across OECD countries. In about a quarter of OECD countries, immigrants make up less than 3.5% of the total population, whereas in eight OECD countries shares exceed 18%. Consequently, in countries with low immigrant shares, the population size of natives with at least one immigrant parent are also small, whereas the opposite is true for countries with high shares of immigrants.

For older people within working age, shares of the total working age population range between 10% and 20% in OECD countries. Shares of people in EU countries who report having a disability fall between 14% and 25%. Table 1.1 also shows that relatively few countries score high in more than two of the dimensions shown below. In Poland, the Slovak Republic and Hungary, for instance, the share of immigrants and the share of natives with at least one immigrant parent are among the lowest in the OECD, while for older workers they rank among the highest.

However, these four broad groups mask within-group diversity, which is an aspect worth considering particularly for migrants given the large number of different origin countries. Table 1.2 shows this heterogeneity by classifying OECD countries according to diversity in country of birth, including both the native-born and foreign-born population. The diversity index is calculated as a Herfindahl-Hirschman Index – one of the most common indicators used to measure population diversity – and describes the likelihood that two persons who are randomly drawn from one country are born in the same country. The index ranges from 0 (all persons are born in the same country) to 10 (everyone is born in a different country). This implies that the index not only takes into account the number of countries of birth, but also the size of different groups. To illustrate, when comparing two hypothetical countries where people only come from country A or B, a country where 50% were born in A and 50% in B is ranked as more diverse than a country where the share is, for example, 80% and 20%. Therefore, the index largely mirrors the share of immigrant populations in Table 1.1; countries such as Poland, the Slovak Republic or Mexico score low on the index because their foreign-born population is comparatively small whereas countries with large immigrant populations, such as Luxembourg, Israel and Australia, score high. In most countries, population diversity based on country of birth has increased between 2000 and 2015, notably in Luxembourg, Spain and Norway, and decreased in only a few countries.

This picture changes when excluding the native-born and only considering country of birth diversity within the immigrant population. Table 1.3 shows that diversity among immigrants is particularly high in Denmark, the United Kingdom and Canada. In most countries, diversity among immigrants has changed little between 2000 and 2015, except for Poland, the Slovak Republic, the Czech Republic and Ireland, where diversity has increased by 1.5 to 2.5 index points as well as in Mexico, where the index decreased by 3.2 points.

However, it should be acknowledged that this indicator only provides a limited picture of population diversity. By being restricted to country of birth, the index does not account for native-born children of immigrants. In addition, the index does not show other dimensions of diversity within this group, such as religion (see Box 1.2) and ethnicity, and also treats migrants from countries that have a closer historical connection, e.g. Sweden and Norway or the Czech Republic and Slovakia, as ‘diverse’ as migrants born on other continents.

The “dashboard” in Table 1.4 shows the employment gaps of diverse groups and attitudes towards them, as well as their evolution over a ten-year period.

Employment gaps between men and women, native-born and immigrants, prime age and older workers and people with and without disabilities remain considerable in many OECD countries. Table 1.4 shows these gaps and their evolution over time. However, this should not be interpreted as an indication of how effective diversity policies are in a given country. For example, in countries with large shares of (highly skilled) labour migrants who arrive at the country with a job offer, employment gaps are likely to be smaller than in countries with high shares of family migrants or refugees – two groups that tend to struggle to find a foothold in the labour market (Dumont et al, 2016). Furthermore, it should be noted that fully closing the gaps for older people and people with disabilities is neither realistic, nor desirable, given the health concerns that may prevent individuals in these groups from working.

With these caveats in mind, Table 1.4 shows that both for women and older workers, employment gaps decreased markedly in the majority of OECD countries. Between 2007 and 2017, the gender employment gap decreased by at least 25% in two out of three OECD countries. In three countries, it was more than halved (Luxembourg, Latvia and Lithuania). Nevertheless, on average in the OECD, the employment gap between men and women still stood at 15 percentage points in 2017. In addition, in most countries with large gender gaps, e.g. Turkey, Mexico and Korea, the decline was considerably slower; gaps only decreased by around 10% in the past 10 years. A notable exception is Chile where gender gaps decreased by more than one-third.

As for the gender employment gap, the gap between older and prime age workers has decreased by at least 25% in two out of three OECD countries. Gaps decreased particularly strongly in the Netherlands, Germany, Denmark and Italy by more than 50%. Iceland was the only country where gaps widened, however this increase was small and Iceland remains the OECD country with the smallest employment gap between older and prime-age workers. Nevertheless, gaps remain large at an average 17 pp across OECD countries, ranging from under 10 pp in New Zealand, Korea, Chile and Sweden, to more than 30pp in Belgium, Hungary, Austria, Poland, Slovenia and Luxembourg.

For immigrants, the picture is more mixed. In contrast to women and older people, immigrants have higher employment rates than the native-born in 11 OECD countries. In the majority of these countries, the employment rates of migrant women are higher than among native-born women. In all OECD countries, however, migrant women have lower employment rates than migrant men.. Across the OECD, in 2016/2017, the migrant employment gap was -0.8 pp – meaning that, on average, the employment rates for immigrants were almost 1pp higher. However, this is largely driven by non-EU OECD countries. In the EU, on average, there is a 3.5 pp difference, and in five EU countries (Belgium, Finland, Denmark, Sweden and the Netherlands) the gap is above 10 pp.

For a number of countries where foreign-born are already more likely to work, these gaps have further increased between 2006/07 and 2016/17, most notably in Chile, Israel and Portugal.

Among those countries where gaps have decreased by more than 50%, there are a number of countries where migrants used to have higher employment rates, but are now very close to the native-born rates, e.g. in Ireland, Lithuania and Greece. In some others, such as Spain and Estonia, migrants had slightly lower employment rates than natives in 2016/2017. In Italy, the gap has decreased but migrants still have higher employment rates than natives. Poland is the only country in this group where migrants were less likely to work in 2006/2017, but now have higher employment rates than the native-born (3.5pp). In New Zealand, the employment gap decreased from 6.5 to 1pp and in the United Kingdom from 5.5 to 2.5 pp.

Yet, in countries where migrants continue to have considerably lower employment gaps than natives, relatively little progress was made over the past ten years. In Mexico, France and Finland these gaps widened by around 2pp, while in other countries they have only decreased slightly.

While there are no data available using the same definition of disabilities, gaps are clearly the largest for this group; in 2012, for the 27 OECD countries with available data, the gap was only below 20pp in Italy, France, Switzerland and Germany, while exceeding 30pp in almost one third of countries.

Overall, this dashboard shows that employment gaps have remained substantial in most countries, particularly for older workers and people with disabilities. Iceland is the only OECD country where gaps are below 5% for three groups. For two groups, the gap is below 5% only in Lithuania and Latvia. While employment rates of LGB people cannot be compared over time and data are only available for a few countries,1 at first glance evidence indicates that gay men are less likely to be employed than heterosexual men, while the reverse occurs when lesbian women are compared with heterosexual women (Valfort, 2017[1]). However, these data, based on household surveys, identify LGB people indirectly, based on the gender of the respondent’s partner, thereby limiting the sample to LGB respondents who live in same-sex couples. Looking at same-sex households only is likely to overstate the impact of sexual orientation on employment rates. It does not take into consideration that in heterosexual households, women are less likely to work and also more likely to raise children and take care of domestic tasks than their male partners, whereas in same-sex households these heteronormative forms of division of labour hardly exist (ibid).

Survey data with a sufficiently large sample of homosexual and heterosexual singles is sparse, yet confirms this ‘household bias’. In the United Kingdom, for instance, partnered lesbian women are 27% points more likely to work than partnered heterosexual women, whereas single lesbian women are 9% less likely to be employed than heterosexual single women (Aksoy, Carpenter and Frank, 2016[8]).

Attitudes towards gender equality and openness towards minorities are an important component in assessing how OECD countries are faring and whether public opinion is supportive of strengthening diversity in the workplace.

In the context of overall acceptance of diversity, previous OECD work has shown that openness towards migrants, acceptance of homosexuality and support for gender equality are linked with each other (Valfort, 2017[1]). There is a positive relationship between acceptance of homosexuality and support for gender equality, possibly because homophobia is also linked with supporting traditional gender roles. In addition, in countries where acceptance of homosexuality is higher, the share of people agreeing that native-born should be favoured over migrants when jobs are scarce is lower. This may indicate that fostering openness towards one group may also positively affect attitudes towards other groups.

On the particular issue of immigration, research has shown that in a number of European countries, such as Germany, Norway, Portugal and Spain, attitudes have become more favourable towards migrants over the past decade while in several others, such as Italy and Hungary, it has become more negative (Heath, Richards and Liebig 2018). That is, there was increasing polarisation between European countries in their attitudes. In addition to the growing divergence between countries, there were also a number of countries where internal polarisation occurred, with increasing numbers both of supporters of immigration and of opponents. Attitudes towards gender equality and LGBT people, however, have become steadily more positive over the past decade in the large majority of OECD countries (OECD 2019; OECD 2017).

However, assessing people’s attitudes towards diversity and gender equality by the means of opinion polls can lead to biased results. Particularly with socially sensitive topics or when a certain level of consensus has been reached, e.g. around gender equality, respondents are likely to provide an answer they think is expected from them, rather than their ‘true’ feelings towards an issue. This ‘social desirability bias’ can therefore distort the measurement of attitudes and provide an overly positive picture of attitudes towards diversity. In addition, positive attitudes may not necessarily be in line with actual behaviour. Therefore, rather than relying on estimates of personal, self-assessed attitudes, the following uses an indicator of how people assess their neighbourhood’s overall acceptance of diversity, rather than their personal one.

Generally, only a small majority of people in the OECD (55%) believe their neighbourhoods are good places to live for ethnic minorities, LGBT people and immigrants (Figure 1.2). Between 2008 and 2018, the share of respondents agreeing with this statement has increased in most OECD countries, yet at a very uneven pace across countries. In addition, in a number of countries shares had decreased considerably by in 2018 and in more than one-third of OECD countries, at least half of the population does not think that their neighbourhoods are good places for minorities to live. Looking at how people assess the quality of live for different groups shows that in very few countries the share of people thinking that their neighbourhoods are good places to live for LGBT people has decreased. For ethnic and racial minorities this is somewhat more likely to be the case, yet the decrease is most pronounced for immigrants; in almost half of OECD countries the share of people thinking that their neighbourhoods are good places for immigrants to live has decreased.

However, this indicator measures the perceptions of all residents, whereas the experiences of groups at risk of discrimination may be markedly different. For example, women have a different perception on gender-based discrimination than men. In 2015, on average in the EU, around one in three people agree that gender-based discrimination is fairly or very widespread, but in every EU country, women are more likely to hold this view than men (Eurobarometer, 2015[3]). Differences in perception are particularly large (more than 10 pp) in Estonia, Greece, Finland, France, Sweden and Slovenia.

Survey research among immigrants, their children and ethnic minorities in the EU show that almost one in four respondents felt discriminated against in the 12 months prior to the survey due to their ethnic or immigrant background (European Union Agency for Fundamental Rights, 2017[9]). Feelings of discrimination are most frequent in the area of employment; 12% report being discriminated while looking for work and 9% felt discriminated against at work. Levels are particularly high among respondents from North Africa (15% when looking for work and 14% at work) and for Roma respondents (16% and 5%, respectively). In addition, countries where a large share of the population thinks their community is a good place for minorities to live, e.g. in Finland, the Netherlands and Luxembourg, show some of the highest levels of perceived discrimination.

Experiences of discrimination or fear thereof, particularly at the workplace, also remains a considerable issue for LGBT people. In the EU, around one in three report that they have never disclosed their sexual orientation at work and another 23% state that they have rarely been open about their sexual orientation (European Union Agency for Fundamental Rights, 2014[10]). In addition, country-specific research consistently finds that people with disabilities report high levels of discrimination (see for example Krnjacki et al. (2018[11]) for Australia or Kassam, Williams and Patten (2012[12]) for Canada).

Thus, both the levels of perceived discrimination or fear thereof as well as relatively small changes in how people assess the openness of their neighbourhoods towards diverse groups indicate that although attitudes are slowly improving for some groups, there remains much prejudice.

When looking at the attitudes towards diversity policies in the workplace, a large majority in the EU believes that more should be done. On EU average, only 9% believe that enough is being done to promote diversity in their work place.2 Another 12% state that this is true to some extent.

At the same time, and somewhat paradoxically, only a minority is supportive of concrete diversity measures in their workplace (Figure 1.3). On average in the EU, only around 36% would support training on diversity issues for employees and employers and 35% would support policies monitoring recruitment to ensure that all candidates with equal skills and qualifications have the same opportunities. Support for monitoring the composition of the workforce is generally lower (32%), and particularly low in Latvia, Estonia and Germany. Only in five countries (Ireland, Sweden, Spain, Portugal and Greece) is the average support for all three measures above 40%.

These seemingly contradictory findings – thinking that more needs to be done in the workplace, yet not supporting concrete diversity policies – shows that more abstract support for diversity and equality of opportunity may not necessarily translate into support for diversity policies, possibly reflecting fears that such measures may affect themselves negatively somehow.


[8] Aksoy, C., C. Carpenter and J. Frank (2016), “Sexual orientation and earnings: new evidence from the UK”, European Bank for Reconstructing and Development Working Paper No. 196.

[2] Balestra, C. and L. Fleischer (2018), “Diversity statistics in the OECD: How do OECD countries collect data on ethnic, racial and indigenous identity?”, OECD Statistics Working Papers, No. 2018/09, OECD Publishing, Paris, https://dx.doi.org/10.1787/89bae654-en.

[4] Bureau, U. (2018), https://www.census.gov/about/our-research/race-ethnicity.html.

[3] Eurobarometer (2015), Eurobarometer 83.4: Climate Change, Biodiversity, and Discrimination of Minority Groups, May-June 2015, European Commission.

[9] European Union Agency for Fundamental Rights (2017), EU MIDIS II - Second European Union Minorities and Discrimination Survey - Main results, Publications Office of the European Union, Luxembourg.

[10] European Union Agency for Fundamental Rights, F. (2014), EU LGBT survey – European Union lesbian, gay, bisexual and transgender survey – Main results, Publications Office of the European Union, Luxembourg.

[12] Kassam, A., J. Williams and S. Patten (2012), “Perceived Discrimination among People with Self-Reported Emotional, Psychological, or Psychiatric Conditions in a Population-Based Sample of Canadians Reporting a Disability”, The Canadian Journal of Psychiatry, Vol. 57/2, pp. 102-110.

[11] Krnjacki, L. et al. (2018), “Disability-based discrimination and health: findings from an Australian-based population study”, Australian and New Zealand Journal of Public Health, Vol. 42/2, pp. 172-174.

[5] Mathews, K. et al. (2017), 2015 National Content Test: Race and Ethnicity Analysis Report, United States Census Bureau. U.S Department of Commerce. Economics and Statistics Administration.

[6] OECD/EU (2018), Settling In 2018: Indicators of Immigrant Integration, OECD Publishing, Paris/European Union, Brussels, https://dx.doi.org/10.1787/9789264307216-en.

[7] Pew Research Center (2014), Global Religious Diversity. Half of the Most Religiously Diverse Countries are in the Asia-Pacific Region, http://www.pewresearch.org/wp-content/uploads/sites/7/2014/04/Religious-Diversity-full-report.pdf.

[1] Valfort, M. (2017), “LGBTI in OECD Countries: A Review”, OECD Social, Employment and Migration Working Papers, No. 198, OECD Publishing, Paris, https://dx.doi.org/10.1787/d5d49711-en.


← 1. Australia, Canada, France, Sweden, United Kingdom and United States.

← 2. In the Eurobarometer Survey, diversity includes: Ethnic origin, sexual orientation, being over 55 years old, being under 30 years old, religion or beliefs, disability, gender identity and gender.

Metadata, Legal and Rights

This document, as well as any data and map included herein, are without prejudice to the status of or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city or area. Extracts from publications may be subject to additional disclaimers, which are set out in the complete version of the publication, available at the link provided.


© OECD 2020

The use of this work, whether digital or print, is governed by the Terms and Conditions to be found at http://www.oecd.org/termsandconditions.