Research and analysis

Outcomes in labour market for ethnic minorities by immigrant generation status

Published 17 April 2023

1. Main points

We examined the labour market outcomes of working age adults (16 to 64 year olds) living in the UK, comparing:

• people born abroad (‘first generation’)
• people with at least one parent born abroad (‘second generation’)
• white British people – both parents born in the UK and who identify with the ‘white British’ ethnic group

Controlling for age and date of the research interview, we found that the second generation of some ethnic minority groups perform better than white British people in unemployment and economic inactivity:

• second generation Bangladeshi men and second generation black African women are less likely to be unemployed compared with white British men and women respectively
• second generation Indian men are less likely to be economically inactive compared with white British men

Some ethnic groups still face disparities in the labour market, with higher likelihoods of being unemployed or economically inactive compared with white British people. For example:

• second generation black Caribbean men and women are more likely to be unemployed compared with white British men and women
• first generation Bangladeshi and both first and second generation Pakistani women are more likely to be economically inactive compared with white British women

Controlling also for educational qualification, health, marital status, geography, difficulties in speaking English and parenthood of a child below the age of 16, we found that:

• education and health are the major factors that affect labour market outcomes
• for men, being single is associated with higher likelihood of unemployment – while for women, having an dependent child is associated with higher likelihood of economic inactivity

Generation is a vital factor that affects labour market outcomes for ethnic minorities. Our analysis shows that:

• for black African women, Bangladeshi men, and Pakistani and Bangladeshi women, the second generation performs better than the first generation in the labour market
• however, the second generation of black Caribbean and Indian men is more likely to be unemployed compared with the first generation

2. Introduction

The purpose of this paper is to establish the social mobility journey of migrants coming to the UK and understand how migration status (that is, whether someone is a first or second generation migrant) has an effect on ethnic disparities in the UK.

Labour market statistics show ethnic minority workers are more likely than white or white British workers to be unemployed or economically inactive (Economic Inactivity, Ethnicity facts and figures). However, little attention is paid to the link between migration and ethnicity.

For example, the majority of first generation black Caribbeans migrated to the UK to fill labour shortages after World War 2. The majority of first generation black Africans migrated to the UK as students from the 1970s to the 1990s. Because of this, the labour market outcomes of these ethnic groups are different and affected by different factors.

This report explores the historical context and the reasons why the main ethnic groups moved to the UK. In addition, it uses the Understanding Society dataset to explore the likelihood of unemployment and economic inactivity by ethnicity, immigrant generation and gender.

We have used a multivariate logistic regression^{[footnote 1]} to understand the factors that may affect the likelihood of being unemployed or economically inactive by ethnic group and generation. Data on education, geography, health, ability to speak English, marital status and having a child below the age of 16 were included in the analysis. This was done to understand how these factors increase or decrease the likelihood of unemployment and economic inactivity, by ethnicity, generation and gender.

3. Literature review: migration, ethnicity and labour market

Migration in the UK has changed through the years since the end of World War 2. The UK government introduced mass migration policies to deal with labour shortages, but also to resettle political refugees that supported Britain during the war (Forced Migration Review). Commonwealth citizens helped Britain to rebuild during the years after the war and deal with labour shortages. New communities were established in the industrial areas of Lancashire and Yorkshire (Peach, GCK, Immigrants in the Inner City). But Commonwealth citizens also settled in London and other areas (Catney, G and Simpson, L, 2010)

Britain became the home of many refugees fleeing civil war and political unrest. Some notable events are the civil war in Pakistan and the resettlement of Ugandan refugees in the UK during the 1970s (Office for National Statistics (ONS)). Somalis and Sri Lankans sought asylum in Britain during the 1990s, while Iraqi, Syrian and Afghan refugees found homes in the UK trying to avoid conflict in their homelands (House of Commons Library).

The introduction of Freedom of Movement in the EU during the 1990s led to increased EU migration after 2000, especially during the Eurozone economic crisis where Europeans looked for better opportunities in Britain (ONS). But non-EU immigration has been rising in the UK in recent decades as more young people from China and India move to the country to benefit from the UK’s educational services (Migration Observatory).

According to the 2021 Census, 1 in 6 usual residents of England and Wales were born outside the UK. There was an increase of 2.5 million people born outside the UK since 2011, from 7.5 million (13.4%) to 10 million (16.8%) (International migration, ONS), with India the most common country of birth outside the UK in 2021. Non-UK passport holders (excluding those who also held a UK or Irish passport) went up from 4.2 million (7.4%) in 2011 to 5.9 million (9.9%) in 2021.

The migration flows over the last 70 years created a multi-ethnic population in Britain which increased both the population size and its ethnic diversity. According to the 2021 Census, 81.7% (48.7 million) of usual residents in England and Wales identified their ethnic group as white, a decrease from 86.0% (48.2 million) in the 2011 Census (Ethnic group, England and Wales: Census 2021, ONS). People from Asian ethnic groups made up the second largest percentage of the population (9.3%), followed by black (4.0%), mixed (2.9%) and other (2.1%) ethnic groups. The 3 largest increases since 2011 were seen in the number of people identifying as:

• ‘white: other white’ (6.2%, 3.7 million in 2021 – up from 4.4%, 2.5 million in 2011)
• ‘other ethnic group: Any other ethnic group’ (1.6%, 924,000 in 2021 – up from 0.6%, 333,000 in 2011)
• ‘black, black British, black Welsh, Caribbean or African: African’ (2.5%, 1.5 million in 2021 – up from 1.8%, 990,000 in 2011)

However, there are still challenges that ethnic minorities in Britain face in the labour market. For example, in 2021 the unemployment rate for all ethnic minority groups combined (excluding white minorities) was double the unemployment rate for white ethnic groups (8% and 4% respectively) (Unemployment, Ethnicity facts and figures). Ethnic minority groups also had higher rates of economic inactivity (27%) compared with white ethnic groups (21%) (Economic inactivity, Ethnicity facts and figures).

However, looking at ethnic minorities as one group does not take into account the diversity between them or the challenges each individual ethnic group may face in the labour market. For example, Pakistani and Bangladeshi groups (combined) had the highest unemployment rate among ethnic minority groups, while the Indian and white ‘other’ groups had the lowest rates (Unemployment, Ethnicity facts and figures).

It is also well documented that there are gender differences in the labour market (The gender gap in employment: International Labour Organization (ILO)). For example, 51% of Pakistani and Bangladeshi women (combined) were economically inactive in 2021, compared with 20% of Pakistani and Bangladeshi men (Unemployment, Ethnicity facts and figures). This example highlights the importance of looking at labour market analysis not only by specific ethnic group but by gender as well.

As mentioned, there have been many reasons that people migrated to the UK and people have different backgrounds and skills that affect their trajectory within the UK labour market. Some first generation immigrants in the UK are very highly skilled as they moved to the UK to study and as a consequence increase their employability. For example, first generation immigrants from India are more likely to be in highly-skilled occupations than the UK born (Migrants in the UK labour market: The Migration Observatory). There can be substantial differences in skills even within a single ethnic group. For example, Pakistani immigrants arriving today are much more highly skilled than those who arrived in the previous century (Luthra and Platt, 2017). This has intergenerational consequences. According to research from Yaojun Li, second generation immigrants have differences in social mobility depending on which ethnic group they belong to, the skills of their parents and the reasons their parents migrated to the UK.

In conclusion, by including migrants’ generation (for example, first or second) and separating the analysis between men and women, we can understand which ethnic groups are performing better in the labour market and which ones are still facing challenges and how we can address these challenges. This is the main focus of this paper and we have chosen to investigate more deeply the disparities in unemployment and economic inactivity.

To get a better understanding of the issues that ethnic minorities face in the labour market, we need to further explore other factors that affect the likelihood of individuals being unemployed or economically inactive. The level of educational qualification is one of the biggest factors in people’s employment status. According to a study from the Department for Education, higher levels of education lead to better labour market outcomes, including higher levels of employment.

Health is another major factor in employment, according to the Health Foundation. In many cases poor health leads people into unemployment or economic inactivity. Research from ONS revealed that there is an increasing number of working age people characterised as economically inactive because of long-term sickness, from around 2.0 million in spring 2019, to about 2.5 million in summer 2022.

Marital status and parenthood are also factors that influence the labour market outcomes between men and women. According to the International Labour Organization (ILO), marriage increases men’s labour participation, while married women are more likely to be unemployed or economically inactive. Having children reduces women’s labour market participation more so than getting married (ILO).

Geography is another factor that should be taken into account. According to the Institute for Fiscal Studies (IFS), geographic disparities are persistent in the labour market. We need to bear in mind that they largely reflect the concentration of high-skilled workers, who would have better labour market outcomes wherever they live.

We also need to consider characteristics that affect labour market participation specifically for first generation immigrants. The ability to speak the English language is a factor that may affect labour market outcomes. Research from the Migration Observatory states that migrants whose main language at home is English are more likely to be employed and have higher average earnings. For working age immigrants, the length of time they spent in their destination country is a strong determinant of their employability.

Considering all the above will allow us to explore which ethnic minority groups perform better in the labour market and which ones still face disparities, by generation. Looking at characteristics, such as education and health, will allow us to understand where policy should focus on reducing unemployment and economic inactivity for all people.

4. Data and methodology

To identify the best quality data for our analysis, we explored a range of different data sources. We selected the Understanding Society (UK Household Longitudinal Study, or UKHLS) dataset as it is the only dataset in the UK that includes generation variables and has an ethnicity/migration boost (the survey added an extra 6,500 relevant households). Future research from Census 2021 may also provide potential to look at this topic area in future.

To define someone’s generation, we are using information on their country of birth, as well as their parents’ and grandparents’ countries of birth, as reported across different times they have been interviewed.

To improve sample size for the regression analysis, we merged waves 6, 7, 8, 10 and 11 of UKHLS to create a panel dataset with observations of people across different waves. We chose to start the analysis in wave 6 to include the immigrant and ethnic minority boost sample which was fielded at this time. This included:

• a boost for existing oversamples of Indian, Pakistani, Bangladeshi, black Caribbean and black African ethnic groups
• a new cross-sectional boost of all those born overseas

This made the data more representative of recent migration. Following advice from the UKHLS team we used the cross sectional weights (indscui_xw) for individuals. We rescaled the weights by having as a base the weights from wave 6.

Combining the waves and sample boosts for ethnic groups increases the sample size but it may create an overrepresentation of groups. To deal with that in the regression analysis we have clustered the standard errors by using the primary sampling unit (psu) and subsets of the population (strata). We have also followed further clustering by individuals using pidp.

After merging the data we had a weighted sample size of 220,308 person-year observations that had both ethnicity and generation variables. The generation variable defines people as first and second generation. It also has information for third, fourth, other (born UK, no data for any parents or grandparents), other (born UK, both parents born UK, no data for any grandparents) – but we removed them because of the small sample size. We also removed any observations of white British that were first generation immigrants to avoid skewing the results. That reduced the weighted sample size to 211,526 person-year observations.

We developed a model by using a multivariate logistic regression. Our analysis uses separate models for men and women, to capture variation by sex in the labour market. We are using 2 different regression models to understand the reasons for being unemployed or economically inactive.

The base model controls for:

• age
• time of interview (UKHLS wave)
• ethnicity and generation category (such as black African first generation, black African second generation)

Initially in the model we took into account the length of time spent in the UK for first generation immigrants, but we removed it from the model because of perfect multicollinearity with the generation variable.

The fully adjusted model takes into account further explanatory variables to explain the reasons we see the disparities. The second model controls for:

• geographical region
• difficulties in speaking English
• educational qualification
• health
• marital status
• someone being a parent of a child under 16 years old

We have limited the analysis to the working age people, 16 to 64 years old. We created a binary variable with the risk of being unemployed taking the value of 1, and being employed with the value of 0. Economically inactive people have been removed from the unemployment models as we study them in separate models in a similar way to unemployment. Also we performed the regression analysis on a model with complete information for all independent and dependent variables. The final weighted sample size for the unemployment model was 122,394 complete observations in the eligible age range, which is 56% of the original sample. Similarly for the inactivity model (comparing economic inactivity with employment) the final weighted sample size for the unemployment model was 148,243 complete observations in the eligible age range, which is 67.3% of the original sample.

In addition, to understand the effect of migrant generation on the ethnic minority outcomes in the labour market we also ran the regression analysis within specific ethnic groups (black Caribbean, black African, Pakistani, Bangladeshi and Indian ethnic groups), comparing the outcomes of the second generation with the first generation, rather than in comparison with the white British group.

5. Results

This section discusses the results from the regression models. It focuses on results that are statistically significant in both the base model and the fully adjusted model. Occasionally, results that are not statistically significant will be mentioned and will be caveated appropriately. The odds ratios^{[footnote 1]} mentioned are from the base model unless stated otherwise.

Our analysis has revealed that the second generation of some ethnic groups performs better than the white British group when we are looking at unemployment and economic inactivity.

For example, second generation Bangladeshi men are less likely than white British men to be unemployed (odds ratio of 0.53). There is evidence that Bangladeshi men born in the UK have a stronger probability to increase their job opportunities in the labour market (Ethnic minority disadvantage in the UK labour market, Joseph Rowntree Foundation). Bangladeshi boys also perform better in education compared with white British boys (GCSE results, Ethnicity facts and figures) that in combination with higher entry rates to university may have decreased their likelihood of being unemployed (Higher Education Statistics Agency (HESA)). It is worth mentioning that second generation Bangladeshi men have the same likelihood of being economically inactive as white British men, but that result is not statistically significant.

Second generation black African women are less likely to be unemployed compared with white British women (odds ratio of 0.37). One of the explanations may be the increased number of black African women getting into higher education over the last years (UK domiciled student enrolments, HESA). Black African women are also overrepresented in key worker jobs such as health and social care roles (NHS workforce race equality standard).

Second generation Indian men are less likely to be economically inactive compared with white British men (odds ratio 0.46). Generally Indian men have high employment rates (Labour market status by ethnic group, ONS) and are more likely to be in managerial roles and at the top of the income distribution (IFS Deaton Review). In addition, second generation Indian men are less likely to be unemployed compared with white British men (odds ratio 0.46), but this result is not statistically significant.

Despite these positive stories, some ethnic groups still face disparities in the labour market with higher likelihoods of being unemployed or economically inactive compared with white British people.

Second generation black Caribbean men and women are more likely to be unemployed (odds ratios of 2.94 and 3.38 respectively) compared with white British men and women in the base model. People from black Caribbean ethnic groups seem to struggle in the labour market. One of the reasons is their poor education outcomes (GCSE results, Ethnicity facts and figures) and the almost stable tertiary education entry rates over the last years (UK domiciled student enrolments by ethnicity, HESA). This is also confirmed by looking at the fully adjusted model, where when we take into account characteristics like education – the odds ratio is reduced to 2.46 for men and 2.68 for women.

First and second generation Bangladeshi and Pakistani women are more likely to be unemployed compared with white British women. At the same time, first generation Bangladeshi and both first and second generation Pakistani women are more likely to be economically inactive compared with white British women when we look at the base model. It is interesting that when we look at the fully adjusted model, the likelihood of being inactive for ethnic minority women is increasing. This means that explanatory variables such as level of education and health do not explain why Pakistani and Bangladeshi women are more likely to be economically inactive compared with white British women.

The reasons for the poor labour market outcomes for the above groups are complex, as the factors that we used in our model do not explain the disparities. According to qualitative work from the Department for Work and Pensions, this may be due to the effects of segregation and cultural attitudes, where women are expected to stay at home and care for younger and older members of the household.

Figure 1: likelihood (odds ratio) of unemployment for selected ethnic groups by generation and sex compared with white British ethnic group

Source: Equality hub analysis based on UKHLS data

Figure 2: likelihood (odds ratio) of economic inactivity for selected ethnic groups by generation and sex compared with white British ethnic group

Source: Equality hub analysis based on UKHLS data

Our analysis also confirms existing research on the relationship between the likelihood of unemployment and economic inactivity and general characteristics such as sex, education and health:

• both men and women are more likely to be unemployed or economically inactive if they have no education qualifications compared with people with a degree – for example, men with no qualification were more than 11 times more likely to be unemployed compared with men with a degree
• both men and women are more likely to be unemployed or economically inactive if they have poor health compared with people with good health – for example, women with poor health were more than 15 times more likely to be economically inactive compared with women with excellent health
• single men are more likely to be unemployed or economically inactive compared with married men (odds ratios of 3.36 and 2.05 respectively)
• men with children under the age of 16 are less likely to be economically inactive compared with men without a child (odds ratio 0.83) – while this phenomenon reverses for women, women with children are more likely to be economically inactive compared with women without (odds ratio 2.55)
• difficulty speaking English is one of the reasons that men are more likely to be unemployed (odds ratio 2.42) and women economically inactive (odds ratio 2.30) compared with men and women who are native speakers of English language

The results mentioned above reveal positive labour market outcomes among second generation ethnic groups, which suggests that certain second generation ethnic groups perform better than the first generation ethnic groups.

To understand the ‘generation effect’ we have done further analysis by generation for the specific ethnic groups, following the same methodology. The models revealed that:

• second generation black African women are less likely to be unemployed compared with first generation black African women (odds ratio of 0.11 for the fully adjusted model)
• second generation of Bangladeshi men are less likely to be unemployed than first generation Bangladeshi men (odds ratio of 0.11 for the fully adjusted model), and also less likely to be economically inactive (odds ratio of 0.22 for the fully adjusted model)
• second generation Pakistani and Bangladeshi women are less likely to be economically inactive compared with first generation Pakistani and Bangladeshi women (odds ratio of 0.35 and 0.41 respectively for the fully adjusted model)
• but second generation Indian men are more likely to be unemployed compared with first generation Indian men (odds ratio of 3.50 for the fully adjusted model)

Table 1: regression analysis outputs for specific ethnic groups

Compared with first generation	Odds ratio of men being unemployed: base model	Odds ratio of men being unemployed: fully-adjusted model	Odds ratio of men being economically inactive: base model	Odds ratio of men being economically inactive: fully-adjusted model	Odds ratio of women being unemployed: base model	Odds ratio of women being unemployed: fully-adjusted model	Odds ratio of women being economically inactive: base model	Odds ratio of women being economically inactive: fully-adjusted model
African second generation	0.71	0.42	0.96	1.8	0.18 [1]	0.11 [1]	0.67	1.26
Bangladeshi second generation	0.3 [1]	0.11 [1]	0.63	0.22 [1]	0.57	0.71	0.20 [1]	0.41 [1]
Caribbean second generation	2.15	4.29 [1]	2.31	21.61 [1]	2.10	2.36	0.74	1.34
Pakistani second generation	1.74 [1]	1.97	1.33	1.36	0.28 [1]	0.30 [1]	0.29 [1]	0.35 [1]
Indian second generation	3.36 [1]	3.50 [1]	0.59	0.41	0.82	0.97	0.61 [1]	0.90

Notes: [1] indicates that results are statistically significant. The results for the black Caribbean ethnic group are not mentioned because of small sample size that may have affected the results.

6. Conclusion

Our analysis has revealed that there are generation differences in labour market outcomes within ethnic minority groups. First and second generation Bangladeshi and Pakistani women are more likely to be unemployed compared with white British women. At the same time, first generation Bangladeshi and both first and second generation Pakistani women are more likely to be economically inactive compared with white British women. However, second generation Pakistani and Bangladeshi women are less likely to be economically inactive compared with first generation Pakistani and Bangladeshi women. That reveals that migration and integration are factors that affect the disparities that ethnic minorities face in the labour market.

There are still disparities that some ethnic minorities are facing even in the second generation. For example, second generation black Caribbean men and women are more likely to be unemployed (odds ratios of 2.94 and 3.38 respectively) compared with white British men. Having a degree or good health are main determinants of labour market outcomes.

This is the first article of a series of analysis under Action 9 of Inclusive Britain. Equality Hub analysts will work with academia to provide further analysis to understand which factors are impacting these different outcomes among generations for ethnic minorities. For example, further analysis will seek to examine which socio-economic characteristics of first generation migrants are the most important in explaining the different outcomes of different migrant groups. This will also expand the analysis on social mobility, skill and role mismatching, and health outcomes.

7. Acknowledgments

This analysis was done as part of the Open Innovation Team fellowship scheme with the ESRC Research Centre on Micro-Social Change (MiSoC), with mentorship from Professor Renee Luthra and research assistance from Jonas Kaufmann.

The Open Innovation Team is a cross-government unit that works with experts to generate analysis and ideas for policy. The fellowship enables officials to access support using quantitative social science data and analysis to answer a specific policy question. Fellows propose a question they would like to research, and they receive guidance and mentorship from world-leading quantitative social scientists based at MiSoc.

The fellowship is currently in a pilot stage, but a new application window is planned for spring 2023.

1. An odds ratio for a particular ethnic group describes the relative difference in the likelihood of being unemployed or economically inactive in that group compared with a reference group (in this case, the white British ethnic group). An odds ratio higher than 1 indicates a greater likelihood, while an odds ratio less than 1 indicates a lower likelihood. ↩ ↩²