The Migrant Pay Gap: Understanding Wage Differences Between Migrants and Nationals X

ILO The migrant pay gap: Understanding wage differences between migrants and nationals X

between migrants and nationals and migrants between wage differences Understanding pay gap: migrant The The migrant pay gap: Understanding wage differences between migrants and nationals

Silas Amo-Agyei

Publications of the International Labour Office enjoy copyright under Protocol 2 of the Universal Copyright Convention. Nevertheless, short excerpts from them may be reproduced without authorization, on condition that the source is indicated. For rights of reproduction or translation, application should be made to ILO Publishing (Rights and Licensing), International Labour Office, CH-1211 Geneva 22, Switzerland, or by email: [email protected]. The International Labour Office welcomes such applications.

Libraries, institutions and other users registered with a reproduction rights organization may make copies in accordance with the licences issued to them for this purpose. Visit www.ifrro.org to find the reproduction rights organization in your country.

ISBN 978-92-2-032372-4 (PRINT) ISBN 978-92-2-032371-7 (Web PDF)

The designations employed in ILO publications, which are in conformity with United Nations practice, and the presentation of material therein do not imply the expression of any opinion whatsoever on the part of the International Labour Office concerning the legal status of any country, area or territory or of its authorities, or concerning the delimitation of its frontiers.

The responsibility for opinions expressed in signed articles, studies and other contributions rests solely with their authors, and publication does not constitute an endorsement by the International Labour Office of the opinions expressed in them.

Reference to names of firms and commercial products and processes does not imply their endorsement by the International Labour Office, and any failure to mention a particular firm, commercial product or process is not a sign of disapproval.

Information on ILO publications and digital products can be found at: www.ilo.org/publns.

Printed in Switzerland

Produced by the Publications Production Unit (PRODOC) of the ILO. Graphic and typographic design, layout and composition, printing, electronic publishing and distribution. The ILO endeavours to use paper sourced from forests managed in an environmentally sustainable and socially responsible manner.

Code: JMB-REP iii

X Preface

Labour migration can be an important vehicle for real and growing. The crisis threatens to increase development, when it is fair, well-governed and inequalities and labour market differences between allows migrant workers to access decent work. The migrant workers and nationals, for example with world counts an estimated 164 million migrant respect to access to employment, types of work, workers, almost half of them women. They com- working conditions or skills development opportuni- prise 4.7 per cent of the global workforce and ties, which may in turn further deepen migrant pay contribute enormously to societies’ growth and gaps, leaving migrant workers further behind and development. Yet, migrant workers are too often countries in arrears in meeting their commitments treated unfairly and unequally in the labour market. to the 2030 Sustainable Development Agenda.

While the actual impact of the multiple crises trig- Many migrant workers, particularly women, are gered by the Covid-19 pandemic is still unknown, working in essential jobs and contributing greatly to anecdotal evidence suggests that migrant workers community well-being, particularly in the care and are some of the most affected. At the onset of the agriculture sectors, and yet as this Report shows, COVID-19 crisis, millions of migrant workers were the wage gap is strikingly high in these sectors. In forced to return home after losing their jobs. This some high-income countries migrant care workers has had a serious impact on their income and wages, earn over a fifth less than non-migrant care work- and on the support they can provide to their families. ers, despite their tireless efforts and contributions to a sector which is at the heart of humanity and The pandemic has exposed serious deficiencies in prosperity of our societies. Many of them are the labour migration governance, much of which have frontline workers in fighting the current health cri- been in place for years. In all countries, migrant sis, and well-designed action is needed to avoid a workers are facing problems of discrimination and deepening of the migrant pay gaps and to improve exclusion, but the COVID-19 pandemic has exac- conditions of work in this sector, in particular for erbated these deeply entrenched attitudes. This migrant workers. The Report provides an important Report, by examining prevailing wage differences way forward by showing how to measure working between migrant workers and nationals, shows that conditions, especially wages, of migrant workers, even prior to the pandemic wage inequalities were on the basis of which we can identify the extent of significant and even widening. the problem, and define the appropriate measures.

The Report on the Migrant Pay Gap: Understanding The Report finds that discrimination is likely to be wage differences between migrants and nationals – one of the main reasons that the migrant gap is analyses wage data of 49 countries that are avail- so large. If we can address this discrimination and able the latest year prior to the COVID 19 crisis. It eliminate the migrant pay gap we can help to bring provides evidence on how dire the situation actually an end to such inequalities, including the gender is with regard to pay – so vital to the daily life of pay gap, substantially reduce poverty, and allow workers and their families. The report finds that in migrant workers to access their fair share of the the years before the pandemic wage inequalities benefits of decent employment between migrant workers and nationals were of very high levels in many countries, and widening As countries emerge from the crisis, as borders in some. Women migrant workers are doubly dis- open and vaccines are being developed, human criminated against, especially with regard to pay. mobility will continue apace. How countries choose Therefore it is plausible to expect a further widening to move forward from the lessons we have learned of the wage gap between nationals and migrants. during the pandemic will test societiesʹ commitment to building back better. We can take a step in the Many migrant workers are in informal or low paid right direction by eliminating the migrant pay gap, employment, and concerns surrounding violations and the attendant discrimination and inequality of of the principle of equal pay between migrant treatment of migrant workers. There is no better workers and nationals for work of equal value, are nor more important time to do so than now.

Michelle Leighton Manuela Tomei Chief Director Labour Migration Branch Conditions of Work and Equality Department iv The migrant pay gap: Understanding wage differences between migrants and nationals

A migrant worker is seen on her day off in Jordan's Al Hassan Industrial Zone in the country's north. © Copyright ILO. Photographer: Marcel Crozet v

X Table of Contents

Preface iii

List of boxes vii

List of figures vii

List of tables ix

Acknowledgements x

Abbreviations and acronyms xi

Key findings xiii

Executive Summary xv

Chapter 1 – Background 1

1.1. Introduction 2

1.2. Related literature 6

Chapter 2 – Labour market characteristics of migrant workers and nationals 9

2.1. Share of migrant population 10

2.2. Geographical coverage 10

2.3. Weak negative correlation between wage inequality and the presence of migrant workers 16

2.4. Labour market participation, unemployment, and educational and occupational attainments of migrants and nationals 16

2.5. A focus on wage workers 27

2.5.1. Share of wage workers among total labour market participants 27 2.5.2. Gender and informality among wage workers 30 2.5.3. Education and occupations of wage workers 32 2.5.4. Differences between wage workers by economic sectors 50 2.5.5. Contractual conditions of wage workers 53 2.5.6. Public-sector employment 58 2.5.7. Migrant care workers 58 vi The migrant pay gap: Understanding wage differences between migrants and nationals

Chapter 3 – Measuring and understanding the migrant pay gap 63

3.1. The wage structure of migrant workers and nationals 64

3.2. The raw migrant pay gap 69

3.3. The migrant pay gap in different subgroups 70

3.3.1. Migrant women tend to pay a double penalty 70 3.3.2. The migrant pay gap in the formal and informal economies 76 3.3.3. The migrant pay gap is wider in the care economy than the overall migrant pay gap 79 3.3.4 Migrant workers have been among the hardest hit by the economic downturn associated with the COVID-19 pandemic 81

3.4. A complementary measure: Factor-weighted migrant pay gap 81

3.5. What factors lie behind the migrant pay gap? 85

3.5.1. Estimating the migrant pay gap across the hourly wage distribution 86 3.5.2. What part of the migrant pay gap can be explained by differences in the characteristics of migrant workers and nationals in the labour market? 87 3.5.3. Understanding the unexplained part of the migrant pay gap: Selected countries 93

Chapter 4 – Simulations 97

4.1. The counterfactual wage structure of migrant workers 98

4.2. The migrant pay gap before and after eliminating the unexplained part of the migrant pay gap 98

4.3. Measures to eliminate the unexplained part of the migrant pay gap can help reduce working poverty among migrant workers 100

4.4. Measures to eliminate the unexplained part of the migrant pay gap can help reduce wage inequalities and the aggregate gender pay gap in the economy 103

Chapter 5 – Conclusions 107

5.1. Key takeaways 108

5.2. Some limitations of the report 112

5.3. Policy implications and recommendations 113

References 117

Appendices 121

Appendix I. Description of methods 122

Appendix II. National data sources 125

Appendix III. Country and territory groups, by region and income 127

Appendix IV. Supplementary results 130 Table of Contents vii

List of boxes

X Box 1. Relevant concepts and definitions 4

X Box 2. List of countries covered in the estimates 15

X Box 3. Migrant workers have been significantly affected by the COVID-19 crisis 81

X Box 4. Decomposing the migrant pay gap: An illustrative explanation 88

List of figures

X Figure E-1 Summary of the migrant pay gap based on different estimation approaches xviii

X Figure E-2 Double penalties for migrant women and migrant care workers in HICs xix

X Figure E-3 The mean migrant pay gap and the pay gap at the top and bottom deciles of the wage distribution xx

X Figure E-4 The migrant pay gap and overall wage inequalities before and after eliminating the unexplained part of the migrant pay gap xxii

X Figure E-5 Working poverty among migrant workers before and after eliminating the unexplained part of the migrant pay gap xxiii

X Figure E-6 The aggregate gender pay gap before and after eliminating the unexplained part of the migrant pay gap xxiv

X Figure 1 The share of women and men among migrants and nationals, latest years 12

X Figure 2 The relationship between the Gini coefficient and the share of migrantsʼ population, latest year, age 16-70 years 17

X Figure 3 Labor force participation rates of migrants and nationals by sex, latest years 23

X Figure 4 The education of migrant and non-migrant wage workers by sex, latest years 34

X Figure 5 Occupations of migrant workers and nationals, latest years 38

X Figure 6 Occupations of nationals by sex, latest years 41

X Figure 7 Occupations of migrant workers by sex, latest years 44

X Figure 8 The association between education and occupational attainment of migrant and non-migrant workers, latest years 47

X Figure 9 The association between education and occupational attainment of migrant workers by sex, latest years 49

X Figure 10 The distribution of migrant and non-migrant wage workers by industrial sector, latest years 52

X Figure 11 The share of wage employees with temporary work contracts by sex, latest years 54

X Figure 12 Incidence of part-time work among migrant and non-migrant wage workers by sex, latest years 56 viii The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure 13 Incidence of public sector employment among migrant and non-migrant workers by sex, latest years 60

X Figure 14 The distribution of migrant wage workers between the care economy and other sectors by sex, latest years 61

X Figure 15 The shares of migrant wage workers and migrant care workers among total wage workers by sex, latest years 62

X Figure 16 The wage structure of migrant workers and nationals, selected countries, latest years 66

X Figure 17 Migrant pay gaps using hourly wages, latest years 71

X Figure 18 Migrant pay gaps using monthly earnings, latest years 72

X Figure 19 Comparing the mean and median migrant pay gaps 73

X Figure 20 Mean migrant pay gaps using hourly wages by sex, latest years 75

X Figure 21 The gender pay gap between men and women migrant workers and how it compares to the aggregate gender pay gap in the economy, using mean hourly wages, latest year 77

X Figure 22 The gender pay gap between non-migrant men and migrant women and how this compares to the aggregate gender pay gap in the economy, using the mean hourly wage, latest years 77

X Figure 23 Migrant pay gaps in the informal and formal economies, using the mean and median hourly wage, latest years 78

X Figure 24 A comparison between the aggregate migrant pay gap and the migrant pay gap in the care economy, latest years 80

X Figure 25 Factor-weighted migrant pay gaps using hourly wages, latest years 82

X Figure 26 Factor-weighted migrant pay gaps using monthly earnings, latest years 83

X Figure 27 Comparing the raw migrant pay gap with the factor-weighted migrant pay gap using mean hourly wages, latest years 84

X Figure 28 Decomposition of the migrant pay gap at the mean hourly wage into explained and unexplained parts, latest years 90

X Figure 29 The unexplained mean wage differentials between migrant workers and nationals, latest years 91

X Figure 30 The migrant pay gap before and after eliminating the unexplained part of the pay gap in studied economies (based on mean hourly wages), latest years 99

X Figure 31 The proportion of working poor among migrant workers before and after eliminating the unexplained part of the migrant pay gap, latest years 101

X Figure 32 The Gini coefficient after eliminating the unexplained part of the migrant pay gap, latest year 104

X Figure 33 The aggregate gender pay gap after eliminating the unexplained part of the migrant pay gap (based on mean and median hourly wages), high-income countries, latest year 105 Table of Contents ix

X Figure A-1 The Gini coefficient and the mean migrant pay gap using hourly wages from HICs, latest years 130

X Figure A-2 Industrial sectors of migrant and non-migrant wage workers by sex, latest years 131

X Figure A-3 The wage structure of migrant workers and nationals by sex, selected countries, latest years 133

X Figure A-4 The mean migrant pay gaps across the wage distribution, selected countries, latest years 136

X Figure A-5 The share of migrant workers and nationals by top and bottom centiles and intervening deciles of the hourly wage distribution, 48 countries, latest years 141

X Figure A-6 Decomposition of the migrant pay gap across the hourly wage distribution into explained and unexplained parts, selected countries, latest years 151

X Figure A-7 The proportion of migrant workers within occupations, education and the mean migrant pay gap, selected countries, latest years 156

X Figure A-8 The wage structure of migrant workers and nationals, with counterfactual wage distribution of migrant workers, selected countries, latest year 162

List of tables

X Table E-1. The 20 widest migrant pay gaps, latest years xvii

X Table 1. Share of migrant population among total working age population, latest years 11

X Table 2. Coverage of countries by income group 14

X Table 3. Coverage of countries by broad subregion 14

X Table 4. Labour market participation, unemployment and occupational attainment of migrants and nationals by sex, latest years 20

X Table 5. Education of migrants and nationals by sex, latest years 24

X Table 6. Share of wage workers among total labour market participants, latest years 28

X Table 7. Proportion of informal workers by sex, latest years 31

X Table 8. Observed labour market endowments, attributes and characteristics for the decomposition of the migrant pay gap 89

X Table 9. Summary: Migrant share of total working population, migrant share of wage workers, and the mean hourly migrant pay gap 94

X Table 10. The share of low-paid workers among migrant workers before and after eliminating the unexplained part of the migrant pay gap, by sex, latest years 102 x

X Acknowledgements

The report was commissioned by the Labour the conceptualisation and scope of the research. Migration Branch (MIGRANT) and produced as Theodoor Sparreboom from the ILO’s Decent Work a collaborative effort between MIGRANT and Team for Eastern and Southern Africa in Pretoria, the Inclusive Labour Markets, Labour Relations peer reviewed the full report. and Working Conditions Branch (INWORK), of The report was also enhanced by valuable com- the Conditions of Work and Equality Department ments and insights from (in alphabetical order) (WORKQUALITY) of the ILO. Kenza Dimechkie, Samia Kazi-Aoul, Michelle Silas Amo-Agyei was the main author of the Leighton, Mustafa Hakki Ozel, Natalia Popova, and report using the pay gap methodology developed Manuela Tomei. by Patrick Belser and Rosalia Vazquez-Alvarez We also wish to thank Janet Neubecker for copy- (INWORK) and which had been previously applied editing the drafts, and José Garcia and the team in the Global Wage Report 2018/19 (see ILO, 2018). of the ILO Document and Publications Production, Katerine Landuyt provided overall coordination of Printing and Distribution Branch (PRODOC). the report as well as substantial inputs in the main text. Rosalia Vazquez-Alvarez refereed the imple- This report would not have been possible with- mentation of the methodology and the analysis out the support of Manuela Tomei, Director of the data. Patrick Belser and Umberto Cattaneo of WORKQUALITY and Michelle Leighton, Chief of (Gender, Equality and Diversity and ILOAIDS Branch MIGRANT. (GED/ILOAIDS)) provided substantial insights on xi

X Abbreviations and acronyms

ASEAN Association of Southeast Asian Nations

CEACR Committee of Experts on the Application of Conventions and Recommendations (ILO)

CEO Chief executive officer

COVID-19 Coronavirus disease 2019

EU European Union

EU-SILC EU Statistics on Income and Living Conditions

GCC Gulf Cooperation Council

HIC High-income country

ICLS International Conference of Labour Statisticians

ICT Information and communication technology

ILO International Labour Organization

IOM International Organization for Migration

ISCO International Standard Classification of Occupations

ISIC International Standard Industrial Classification of All Economic Activities

KNOMAD Global Knowledge Partnership on Migration and Development (World Bank)

LMIC low- and middle-income country

OECD Organisation for Economic Co-operation and Development

SDGs Sustainable Development Goals

UN-DESA United Nations Department of Economic and Social Affairs

UNCTAD United Nations Conference on Trade and Development Seasonal migrant workers at the Noordoostpolder tulip fields, Netherlands. May 2017. © shutterstock.com xiii

X Key findings

This report presents a comprehensive global analysis of the migrant pay gap based on data covering 49 countries (33 High Income Countries (HICs) and 16 Low- and Middle-Income Countries (LMICs)) and about a quarter of wage employees worldwide. The 49 countries host nearly half of all international migrants and roughly 33.8 per cent of migrant workers worldwide. The report aims to contribute to efforts towards achieving the SDG targets 8.5 and 8.8, which respectively call for equal pay for work of equal value, and protected labour rights for all workers, including migrant workers, in particular women migrant workers and those in precarious employment in the framework of the United Nations agenda for 2030.

The following summarizes the key messages and conclusions from the study:

A. Migrant workers in HICs earn about 12.6 per cent less than nationals, on average. Notable variations, however, exist among countries and across different wage groups, with migrant workers earning as much as 42.1 per cent less than nationals on average (in Cyprus), and 71 per cent less than nationals among low-skilled workers.

B. Within a labour market already quite unfavourable to migrant workers in HICs, women migrant workers face a double wage penalty, both as migrants and as women. The pay gap between men nationals and migrant women in HICs, for example, is estimated at 20.9 per cent, which is much wider than the aggregate gender pay gap in HICs (16.2 per cent).

C. Migrant care workers in HICs (majority of whom are women) also face a double wage penalty for being migrants and care workers. The pay gap between migrant care workers and non-migrant care workers is about 19.6 per cent compared to the aggregate migrant pay gap of 12.6 per cent.

D. The migrant pay gap has widened in many HICs compared to ILO’s previous estimates. Among the 20 countries with the most significant migrant pay gaps, the estimated pay gap has widened in more than half of them compared to previous estimates reported in the ILO Global Wage Report 2014/15. The pay gap in these countries has increased by 1.3 to 26.4 percentage points.

E. Migrant workers have been among the hardest hit by the economic downturn associated with the COVID-19 pandemic, both in terms of employment losses and a decline in earnings for those who have remained in employment.

F. Despite similar levels of education, migrant workers in HICs tend to earn less than their national counterparts within the same occupational category.

G. Migrant workers in HICs are more likely to work in lower-skilled and low-paid jobs that do not match their education and skills. Higher-educated migrant workers in HICs are also less likely to attain jobs in higher occupational categories relative to non-migrant workers. This reflects the fact that migrants in HICs are likely to be affected by skills mismatch and have difficulties transferring their skills and experience across countries, in large part due to lack of adequate skills recognition systems for qualifications of migrant workers.

H. Among LMICs, migrant workers tend to earn about 17.3 per cent more than nationals on average, with notable exceptions. This is due, in part, to the significant proportion of temporary high-skilled expatriate workers among the total migrant population in some countries who tend to pool up the average wage of migrant workers.

I. A significant part of the migrant pay gap remains unexplained even when workers’ characteristics such as education, experience, age, or location are accounted for. About 10 percentage points of the estimated 12.6 per cent migrant pay gap in HICs remains unexplained by labour market characteristics of migrant workers and nationals. This may point to discrimination against migrant workers with respect to pay. xiv The migrant pay gap: Understanding wage differences between migrants and nationals

J. If the unexplained part of the migrant pay gap is eliminated, the migrant pay gap would nearly disappear in many countries and reverse in others. If wages were set based on factors such as education, experience and age, the migrant pay gap would stay very low in many countries and would even reverse in favour of migrant workers in some countries. K. The rate of working poverty among migrants, in particular migrant women would significantly reduce if the unexplained part of the pay gap is to be eliminated. Measures to eliminate the unexplained part of the migrant pay gap can reduce the proportion of low-paid migrant workers, by about 49 per cent in the sample of HICs and about 12 per cent in the sample of LMICs. L. In some selected countries (14 LMICs and two HICs), 62.4 per cent of migrant wage workers are informally employed compared to 50.8 per cent of nationals. Informal employment is higher among migrant women than among their men counterparts. M. In HICs, migrant workers are disproportionately represented in the primary sector and take far more jobs in the secondary sector than their national counterparts. More migrant workers, in particular migrant women, tend to work under temporary contracts and part-time. xv

X Executive summary

In many countries, men and women migrant work- limited. Nonetheless, understanding the migrant ers represent a significant share of the workforce pay gap is critical not just for ensuring protection and contribute importantly to societies and econo- of men and women migrant workers around the mies.1 According to the most recent ILO estimates, world and avoiding social dumping, but also, for there are 164 million migrant workers worldwide, avoiding unfair competition and labour market dis- of whom close to half are women. Despite the pos- tortions. Addressing the migrant pay gap, including itive migration experiences of many, migration is by affording migrant workers equality of treatment, frequently associated with abusive practices and will contribute to well-functioning labour markets, non-respect of fundamental rights at work. Migrant which will be particularly important as countries workers often face inequality of treatment in the seek to emerge and build back after the COVID-19 labour market, including with respect to wages, crisis. Further analysis is needed to understand the access to employment and training, conditions extent of the migrant pay gap around the world, of work, social security, and trade union rights. including differences in pay between migrant men Moreover, recruitment fraud and abuse can cause and migrant women. This report is a first attempt migrant workers, especially low- and semi-skilled to capture the migrant pay gap, including its gender workers, to face high recruitment fees and related dimension at the global level. costs depleting their wages and savings. One way The report uses recent available data from 49 to measure inequalities between migrant workers countries (where labour market data covering and nationals is by comparing the earnings of wages of migrant and non-migrant workers are migrant workers to that of non-migrant workers available) that span the five regions of the ILO and with similar labour market characteristics. which together represent about a quarter of wage The general principle of equal pay for work of employees worldwide.3 The 49 studied countries, equal value is set out in the preamble of the ILO comprising 33 High Income Countries (HICs) and Constitution and in ILO standards concerning 16 Low- and Middle-Income Countries (LMICs), equality and non-discrimination. The dedicated host nearly half (49.4 per cent) of all international ILO Conventions concerning migrant workers migrants and roughly 33.8 per cent of migrant also require ratifying States to ensure equal treat- workers worldwide. It is important to note that ment between migrant workers and nationals the quantitative data on labour market outcomes, with respect to remuneration. However, the ILO including data on wages of migrant and non-mi- supervisory bodies have noted on several occa- grant workers used for the analysis in this report sions non-compliance with this principle and have predate the COVID-19 crisis period. pointed to significant unlawful differences between Based on the data sets, the report discusses dif- migrant workers and nationals, in law or in practice. ferences in labour market outcomes of migrant Previous ILO research, including the ILO Global workers and nationals of the 49 countries, including Wage Report 2014/15,2 has also highlighted the gender differences. It highlights migrant pay gaps existence of significant wage differences (called across these countries with a view to facilitating the the migrant pay gap) between migrant workers adoption and implementation of evidence-based and non-migrant workers in some countries. At labour migration policies around the world, ensur- the national level, there have been attempts to ing that these are gender-responsive. The report analyse the migrant pay gap in several countries also contributes to the work towards achieving (some of which are documented in this report). SDG tartgets 8.5 and 8.8, which respectively call for However, global analysis of the migrant pay gap is “equal pay for work of equal value“ and “protected

1 For example, using the most recent wave of the European Union (EU) Labour Force Survey, Fasani and Mazza (2020a), in Immigrant Key Workers: Their Contribution to Europe’s COVID-19 Response, quantify the prevalence of migrant workers in key professions that the European Commission and Member States have identified and show that migrant “key workers” are essential for critical functions in European societies. They also highlight the contribution of migrant workers to the ongoing effort to keep basic services running in the European Union during the COVID-19 epidemic. 2 ILO. 2015. Global Wage Report 2014/15. Wages and Income Inequality, Geneva. 3 Majority of countries and territories do not have labour market data that include wages of both migrant and non-migrant workers. xvi The migrant pay gap: Understanding wage differences between migrants and nationals

labour rights for all workers, including migrant compares these latest estimates with those found workers, in particular women migrant workers, in the ILO Global Wage Report 2014/15. The list and those in precarious employment.” In addition, features 18 HICs and two LMICs (Costa Rica and the information contained in the report can help set Jordan). On top of the list is Cyprus where men and the basis for monitoring wage inequalities between women migrant workers earn as much as 42.1 per migrant workers and non-migrant workers around cent less than non-migrant workers, which is a the world, and between migrant men and migrant 7.3 percentage points increase from the estimated women; help support the case for closing these gap in 2010 (34.8 per cent) according to the ILO gaps in line with principles set out in the ILO instru- Global Wage Report 2014/15. Slovenia and Costa ments concerning migrant workers; and encourage Rica have the second and third widest migrant pay further research on policies and practices that are gaps (33.3 per cent and 30.1 per cent, respectively) effective for promoting change. while Italy and Jordan have the fourth and fifth widest gaps. While the migrant pay gap has reduced in For the purpose of this report, the migrant pay gap six countries (Argentina, Denmark, Estonia, Greece, – expressed in its simplest form – refers to the dif- Iceland, Spain), it has increased in the remaining ference in average wages between all non-migrant countries for which past estimates are available. workers and all migrant workers who are engaged in paid employment. Further differences arise when comparisons are done using monthly earnings rather than hourly wages. In fact, using four different combinations Migrant workers earn – mean hourly, median hourly, mean monthly, 12.6 per cent less per hour and median monthly – the report finds that the migrant pay gap in hourly wages is smaller than than nationals in high-income the gap in monthly earnings (reflecting inequalities countries and 17.3 per cent in working time), although the size of the gap varies across countries and across income groups. Figure more per hour than nationals E-1 shows the different estimates based on hourly in low- and middle-income wages and monthly earnings. For example, the weighted migrant pay gap in the sample of HICs countries ranges from about 12.6 per cent (in the case of Based on mean wages, the report estimates that mean hourly wages) to 18.4 per cent (in the case of migrant workers earn about 12.6 per cent and median monthly earnings) in favour of non-migrant 8.6 per cent less per hour than non-migrant work- workers. Similarly, the estimates for the EU ranges ers in the sample of 33 High Income Countries from about 8.6 per cent (in the case of mean hourly (HICs) and across the Member States of the EU,4 wages) to 16.8 per cent (in the case of median respectively, while in the sample of 16 Low- and monthly earnings) in favour of non-migrant work- Middle-Income Countries (LMICs) migrant workers. In the sample of LMICs, however, a different ers tend to earn about 17.3 per cent more per situation emerges. The estimates range from about hour than non-migrant workers (see figure E-1). 7.5 per cent (in the case of median hourly wages) Nevertheless, there are notable variations across to 19.1 per cent (in the case of mean monthly earn- countries. A possible reason migrant workers tend ings) in favour of migrant workers. to earn more on average than non-migrant workers Unlike the standard approach where the migrant in some LMICs, among others, is the likelihood of a pay gap simply looks at the difference between the relatively high proportion of temporary high-skilled average (or median) earnings of all non-migrant “expatriate” workers among the total migrant pop- workers and the average (or median) earnings of ulation in those countries. all migrant workers, a different picture emerges Table E-1 shows the list of the 20 widest migrant when education, age, and gender are used as pay gaps among the countries covered in the report factors to account for composition effects in esti- based on the latest available data. The table also mating the migrant pay gap. This results in what

4 The United Kingdom is included for the reporting period, while Germany is excluded due to unavailability of data at the time of writing this report. Executive Summary xvii

X Table E-1: The 20 widest migrant pay gaps, latest years

Migrant pay gap (latest year) Migrant pay gap 2014/15* Rank Country Income Group (%) (%)

1 Cyprus 42.1 34.8 HICs

2 Slovenia 33.3 6.9 HICs

3 Costa Rica 30.1 n/a LMICs

4 Italy 29.6 26.7 HICs

5 Jordan 29.5 n/a LMICs

6 Portugal 28.9 25.4 HICs

7 Spain** 28.3 29.9 HICs

8 Luxembourg 27.3 14.9 HICs

9 Austria 25.3 15.8 HICs

10 Greece** 21.2 29.9 HICs

11 Estonia** 21.0 22.7 HICs

12 Ireland 20.6 19.2 HICs

13 Netherlands 19.9 16.5 HICs

14 Argentina** 18.1 22.0 HICs

15 Iceland** 17.8 24.4 HICs

16 Denmark** 17.3 21.0 HICs

17 United States 15.3 n/a HICs

18 Latvia 15.1 9.0 HICs

19 Norway 15.0 12.2 HICs

20 Belgium 12.7 9.8 HICs

Notes: Estimates are based on mean hourly wages. HICs = High Income Countries. LMICs = Low- and Middle-Income Countries. * Retrieved from the ILO Global Wage Report 2014/15. “n/a” indicates that the estimate was not available in the ILO Global Wage Report 2014/15. Migrant pay gap decreased from the previous estimate based on the ILO Global Wage Report 2014/15. Migrant pay gap increased from the previous estimate based on the ILO Global Wage Report 2014/15.

is called the factor-weighted migrant pay gap.5 counterparts when the factor-weighted approach In comparison to the migrant pay gap based on is used. Relative to the standard approach, the pay the standard approach, the mean hourly migrant gap based on the factor-weighted approach is nar- pay gap based on the factor-weighted approach rower in the sample of HICs and the EU, and wider declines to approximately 9.5 per cent (in favour in the sample of LMICs because the latter accounts of nationals) in the sample of HICs and 7.8 per for composition effects in estimating the pay gap; cent (in favour of nationals) in the EU. However, effects that result from the existence of clusters in the sample of LMICs, migrant workers tend to of few workers – especially migrant workers – at earn about 23.8 per cent more than their national certain locations in the wage distribution.

5 The factor-weighted migrant pay gap reduces composition effects caused by the existence of clusters in the wage or earnings distribution of wage workers. In essence, migrant and non-migrant wage workers are somewhat grouped into homogeneous subgroups based on education, age and gender, and then the migrant pay gap is estimated for each of the subgroups. A weighted sum of all the subgroups’ specific migrant pay gaps is estimated to obtain the factor-weighted migrant pay gap, with the weights reflecting the size of each subgroup in the population. xviii The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure E-1: Summary of the migrant pay gap based on different estimation approaches

18.4 16.8 20 16.1 14.5 14.1 12.6 11.6 9.5 8.6 7.8 10

–10 –7.5 –9.1 Migrant pay gap (%) Migrant –20 –17.3–19.1 –23.8 –30 Factor-weighted Factor-weighted Factor-weighted Mean hourly wage Mean hourly wage Mean hourly wage Median hourly wage Median hourly wage Median hourly wage Mean monthly earnings Mean monthly earnings Mean monthly earnings Median monthly earnings Median monthly earnings Median monthly earnings

High-income European Union Low- and middle-income

Notes: Estimates are based on data from a relatively small sample of countries (33 HICs and 16 LMICs). The analysis yields opposing estimates for the sample of HICs and LMICs. Possible reasons for the opposing findings may include, among others, the relatively small sample of LMICs covered by the report; the relatively small proportion of migrant workers in LMICs; and the composition of jobs among migrant workers in LMICS (for example, the likelihood of a relatively high proportion of temporary high-skilled “expatriate” workers among the total migrant population in some countries).

Migrant women and migrant the average migrant worker in HICs. This finding corroborates previous ILO findings, which reveal care workers in HICs pay that due to the asymmetries between countries of origin and destination and often inconsistent law a double wage penalty 2000 and policy on migration and care, working conditions of migrant care workers tend to differ to a According to ILO global estimates, nearly half of the 2020 greater or lesser extent from those of their national world’s migrant workers today are women. Migrant counterparts.7 The care economy, though very women workers also represent a significant share broad, is defined in this report to include workers of those in domestic work, comprising 73.4 per in education, health and social work sectors, includ- cent (or 8.45 million) of all migrant domestic working domestic and personal care workers, and care ers around the world (in 2013). However, in HICs, workers in non-care sectors. migrant women workers tend to pay a double wage penalty for being both women and migrants, The pay gap between non-migrant men and a finding consistent with results from the OECD's migrant women (based on mean hourly wages) in International Migration Outlook 2020.6 Likewise, in the sample of HICs is estimated at approximately the care economy where work is often underval- 20.9 per cent, which is much wider than the esti- ued, migrant care workers – the majority of whom mated aggregate gender pay gap in HICs (16.2 per are women – pay a larger wage penalty relative to cent) (see figure E-2). The mean pay gap between

6 OECD (2020), International Migration Outlook 2020, OECD Publishing, Paris, https://doi.org/10.1787/ec98f531-en. 7 International Labour Organization (ILO). 2018. Care work and care jobs for the future of decent work (Geneva). King-Dejardin, A. 2019. The social construction of migrant care work. At the intersection of care, migration and gender (Geneva, ILO). Executive Summary xix

X Figure E-2: Double penalties for migrant women and migrant care workers in HICs

20.9 19.6 20 17.1 16.2 8 15

10 20.9

Percent (%) Percent 19.6 16.2 17.1 5

0 Non-migrant men Total men Care economy Total economy versus migrant women versus total women

Gender pay gaps Migrant pay gaps

Note: The aggregate gender pay gap is retrieved from the ILO Global Wage Report 2018/19.

migrant workers and non-migrant workers – men out. First, in some countries, there is a tendency and women combined – in the care economy is for the migrant pay gap to be strikingly high at approximately 19.6 per cent per hour as compared the bottom deciles but declines steadily from the to the aggregate pay gap between all migrant work- lower to upper points in the hourly wage distribu- ers and non-migrant workers of about 17.1 per cent tion. This could possibly imply non-compliance with in the sample of countries for which care workers or exclusion of migrant workers from minimum can be uniquely identified. Understanding the wage legislation. Exclusion from minimum wage underlying causes for these double wage penalties coverage can take many forms. In some countries, in the national context, and adopting measures to national provisions in force may explicitly provide eliminate them, would significantly contribute to for reduced minimum wage rates for migrant reducing wage inequalities. workers. Migrant workers could also be excluded because there is no minimum wage for the sector in which they are primarily employed. Likewise, Estimating migrant pay gaps migrants may not benefit from minimum wage 2000 at different points in the wage coverage because they are not members of a trade union that is a party to the collective agreement 2020 distribution provides insights covering the sector of activity concerned. on how targeted policies can Among the sample of HICs, this is the case in affect these gaps Austria, Cyprus, Denmark, France, Norway, Spain, and Sweden, where the migrant pay gap at the Migrant workers are often concentrated at certain first and/or second deciles of the hourly wage locations in the wage distribution, for example, distribution widens significantly. However, the gap around the minimum wage. To identify where in the shrinks as it moves from the lower to upper ends wage distribution the migrant pay gap is widest, the of the wage distribution. Figure E-3 reports the report estimates the hourly migrant pay gap at ten mean migrant pay gap in the economy together different locations in the wage distribution, that is, with the pay gap at the top and bottom deciles of the gap for the bottom 10 per cent wage earners up the wage distribution for the aforementioned coun- to the gap for the top 10 per cent earners. tries. Clearly, the pay gap at the bottom decile far The results show that the pay gap varies signifi- outweighs the mean migrant pay gap in each of cantly across the hourly wage distribution of each these countries. In the case of France, for exam- country. The following patterns appear to stand ple, although the mean gap is estimated at about xx The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure E-3: The mean migrant pay gap and the pay gap at the top and bottom deciles of the wage distribution

80 70 60 50 40 30 cent (% ) 20 Per 10 0 –10 –20 Cyprus France Spain Austria Sweden Denmark Norway United Finland Canada States

Mean migrant pay gap

Migrant pay gap at the bottom decile of the wage distribution

Migrant pay gap at the top decile of the wage distribution

Note: Estimates are based on mean hourly wages.

9.0 per cent, the gap at the bottom decile of the is common in countries such as Argentina, Belgium, wage distribution is approximately 71.1 per cent but Canada, Iceland, Luxembourg, the Netherlands, declines sharply to about 6.3 per cent at the ninth and the United States. For example, in the case of decile and eventually becomes negative at the tenth Canada, the migrant pay gap at the bottom and top decile. This magnitude of disparity has significant deciles of the hourly wage distribution are –0.6 per policy implications for poverty eradication and for cent and 0.4 per cent (Figure E-3), respectively, but it ensuring decent work among low-skilled migrant increases to about 6.5 per cent in the middle of the workers. For comparison, the figure also presents distribution (i.e. from the fifth to the eight decile). estimates for Canada, Finland, and the United Third, and particularly in some LMICs, the migrant States in which case the pay gap at the bottom pay gap widens and narrows, and reverses in favour decile is lower than the overall migrant pay gap. of nationals or in favour of migrant workers across Second, in other countries, although the migrant the hourly wage distribution. This pattern can give pay gap appears to be lower at the bottom and top an indication of where in the wage distribution tem- deciles of the hourly wage distribution, the gap is porary high-skilled “expatriate” workers are located very high in the middle of the distribution. This may in these countries. In Gambia for example, non-mi- possibly reflect underrepresentation of migrant grant workers tend to earn more than migrant workers in collective representation structures in workers from the bottom to the fourth decile of the middle of the distribution because of difficulties the wage distribution. However, the gap reverses in organizing or because nationals dominate the in favour of migrant workers from the fifth to the overall representation, a phenomenon that could top decile of the distribution, peaking at the ninth be exacerbated if migrants are perceived as a low- decile where migrant workers earn about 54.8 per wage employment threat to nationals.8 This pattern cent more than non-migrant workers.

8 Rubery, J. 2003. Pay equity, minimum wage and equality at work, InFocus Programme on Promoting the Declaration on Fundamental Principles and Rights at Work, Working Paper No. 19 (Geneva, ILO). Executive Summary xxi

A significant part istics have sizeable effects on the migrant pay gap in countries such as Austria, Canada, Luxembourg, of the migrant pay gap Norway, Portugal, Slovenia, the United Kingdom, remains unexplained and the United States, though a significant part still remains unexplained. Among LMICs, the same The report shows that in almost all the studied is true of Bangladesh, Costa Rica, Gambia, Jordan, countries there are wage gaps between non-mi- the United Republic of Tanzania, and Turkey. But in grant workers and migrant workers. These gaps most countries, a large part of the migrant pay gap arise for multiple and complex reasons that differ remains unexplained. For example, in Cyprus (which from one country to another and vary at different has the widest estimated pay gap in the sample of locations in the overall wage distribution. HICs), only about 4.4 percentage points of the pay gap of 42.1 per cent is explained by observed labour Similarly to the ILO’s Global Wage Report 2018/19, market characteristics of migrant workers and this report adapts the decomposition techniques nationals. Other countries with significantly higher 9 pioneered by Fortin, Lemieux and Firpo (2011) to levels of unexplained pay gaps include Argentina, divide the migrant pay gap (at different locations in Belgium, Denmark, Greece, Iceland, Ireland, Italy, the wage distribution) into two parts: an “explained” Jordan, Netherlands, and Spain. In the United part, which is accounted for by observed labour States, on the other hand, about 10 percentage market characteristics, and an “unexplained” part, points of the estimated migrant pay gap of 15.2 per which captures wage discrimination and includes cent is explained by observed labour market charac- characteristics that should in principle have no teristics. Chapter three, section 3.5 shows estimates effect on wages. Labour market characteristics here for all the countries covered in the report. are the so-called human capital characteristics (e.g. age, experience and education); the characteristics that define the jobs held by individuals (e.g. occupa- For similar levels of education, tional category, contractual conditions or working time); the characteristics that describe the work- migrant workers in HICs tend place where production takes place (e.g. industrial to earn less than nationals sector, size of enterprise, geographical location); and personal characteristics such as gender. within the same occupational category The report finds that, on average, education and other observed labour market characteristics Findings from the report show that, for a given explain a relatively small part of the migrant pay occupation, migrant workers’ education levels are gap at different locations in the wage distribution. similar (at the least) to that of nationals, in par- The unexplained part of the migrant pay gap largely ticular in the sample of HICs. In spite of this, the dominates the explained part in most countries, results show that, migrant workers tend to earn irrespective of income group. On the one hand, significantly less than non-migrant workers with the report shows that about 10 percentage points the same occupation in most of the studied coun- of the weighted migrant pay gap of approximately tries.10 For example, in the case of France, although 12.6 per cent (based on average hourly wages) migrant workers account for only 3.4 per cent of in the sample of HICs remains unexplained by professional positions across the country, they observed labour market characteristics of migrant have similar educational scores as nationals in this workers and nationals. On the other hand, nearly all occupational category (24.7 and 23.7, respectively). the 17.3 per cent of the pay gap in favour of migrant Nevertheless, these migrant workers still earn workers in LMICs is unexplained, on average. It is around 22 per cent less per hour than their national significant to add that there are notable excep- counterparts. This illustrates the fact that migrant tions, as well as wide variations across countries workers tend to have lower wage returns to their and across the wage distribution. Among HICs, education relative to nationals, even when they differences in observed labour market character- have the similar occupations as nationals.

9 Fortin, N.; Lemieux, T.; Firpo, S., 2011. "Decomposition methods in economics", in O. Ashenfelter and D. Card (eds): Handbook of Labor Economics (Amsterdam, Elsevier), pp. 1–102. 10 For example, in Argentina, Austria, Belgium, Canada, Cyprus, Denmark, Finland, France, Greece, Iceland, Ireland, Italy, Luxembourg, Netherlands, Norway, Portugal, Sweden, and the United States. xxii The migrant pay gap: Understanding wage differences between migrants and nationals

This phenomenon is compounded by the fact that account and any remaining unexplained pay gap is migrants in HICs are typically likely to be affected eliminated, among the sample of HICs, the migrant by skills mismatch and may have difficulties trans- pay gap would nearly disappear in countries like ferring their skills and experience across countries. Argentina, Belgium, Denmark, Finland, Italy, and Moreover, migrant workers’ skills may not be fully Sweden; and would reverse in favour of migrant recognized by employers and they may resort to workers in Chile, Cyprus, France, Greece, Hungary, continuous work in lower-skilled and low-paid jobs Ireland, Latvia, the Netherlands, and Spain. It that do not account for their higher skills. The report would decline substantially but remain positive in also finds that, given similar levels of education, the Austria, Canada, Luxembourg, Portugal, Slovenia, probability of being employed in a semi- or high- Switzerland, and the United States. On average, skilled job is much lower for migrant workers in HICs the migrant pay gap across the sample of HICs than for non-migrant workers in these countries. would decline substantially from approximately 12.6 per cent to about 0.2 per cent if wages were set according to observed labour market charac- Measures to eliminate the teristics. In the EU, the migrant pay gap would unexplained part of the reverse from about 8.6 per cent to about –7.9 per cent, on average (figure E-4). Among the sample migrant pay gap would keep of LMICs, the migrant pay gap would remain neg- pay differentials between ative in some countries, while it would be positive in others. nationals and migrant workers Additionally, measures to eliminate the unexplained low, and reduce overall wage part of the migrant pay gap can help to reduce over- inequalities all wage inequalities across countries. The report estimates that the Gini inequality coefficient – which Based on a counterfactual wage distribution expresses the level of wage inequalities within the of migrant workers, the report shows that the economy – would reduce from about 31.2 per cent migrant pay gap would generally stay narrow if to approximately 28.0 per cent on average in the migrant workers were equally remunerated as sample of HICs, from about 30.2 per cent to 29.6 per nationals for their labour market characteristics. cent in the EU, and from about 39.3 per cent to Once labour market characteristics are taken into 35.3 per cent in the sample of LMICs (figure E-4).

X Figure E-4: The migrant pay gap and overall wage inequalities before and after eliminating the unexplained part of the migrant pay gap

Mean hourly migrant pay gapWage inequalities

15 45 12.6 39.3 40 10 35.3 8.6 35 31.2 28.0 30.2 29.6 12.6 30 5 8.6 25 0.2 39.3 0.2 20 35.3 0 31.2 15 28.0 30.2 29.6 Gini coefficient (%) Migrant pay gap (%) Migrant –5 –7.9 10 5 –7.9 –10 0 Before After Before After Before After Before After Before After

HICs EU HICs EU LMICS Executive Summary xxiii

X Figure E-5: Working poverty among migrant workers before and after eliminating the unexplained part of the migrant pay gap

12 11.5

8 15.0 13.8 6 11.5 12.2 5 4 6.2 Rate of working poverty (%) Rate of working poverty 2 5.9

0 Before After Before After Before After

HICs EU LMICS

18 16.2 15.7 Thus, in countries16 where the unexplained part of Measures to eliminate the the migrant14 pay gap is significantly high, eliminating this gap would help enhance skills and jobs11.6 unexplained part of the 12 11.6 matching for men and women migrant workers, migrant pay gap can help and promote10 equality as well as economic productivity and development8 across countries. reduce the aggregate gender 6 pay gap in the economy Gender pay gap (%) 4 Measures to eliminate In addition to reducing the migrant pay gap, wage the unexplained2 part inequalities, and working poverty among migrant 0 workers, the report finds that measures that elim- of the migrant Beforepay gap can After inate the unexplainedBefore part of the migrantAfter pay gap help reduce workingMean poverty hourly can help to reduce the aggregateMedian hourly gender pay gap betweenHICs all men and all women in the economy, among migrant workers particularly in HICs. The report estimates that the aggregate gender pay gap in favour of men across Given the significant size of the unexplained part the sample of HICs would decline from around of the migrant pay gap, measures that eliminate 16.2 per cent to approximately 11.6 per cent when this part of the gap would help to reduce the rate using mean hourly wages, and from about 15.7 per of working poverty among migrant workers, espe- cent to 11.6 per cent when using median hourly cially among migrant women. By defining working wages (figure E-6). poverty (low-paid workers) as “the proportion of workers earning less than half of the median hourly wage”, eliminating the unexplained part of Other salient differences the migrant pay gap would reduce the proportion in labour market of low-paid migrant workers, by roughly 49 per cent in the sample of HICs (from about 11.5 per cent to characteristics of migrant 5.9 per cent), by about 59 per cent in the EU (from workers and nationals around 15.0 per cent to 6.2 per cent), and about 12 per cent in the sample of LMICs (from about Similarly to the ILO Global Estimates on 13.8 per cent to 12.2 per cent) (figure E-5). International Migrant Workers (2018b), this report 16

12 11.5

6 5 xxiv 4The migrant pay gap: Understanding wage differences between migrants and nationals

Rate of working poverty (%) Rate of working poverty 2

0 Before After Before After Before After

HICs EU LMICS X Figure E-6: The aggregate gender pay gap before and after eliminating the unexplained part of the migrant pay gap

18 16.2 15.7 16

14 11.6 12 11.6

10 16.2 8 15.7

6 11.6 11.6 Gender pay gap (%) 4

0 Before After Before After

Mean hourly Median hourly HICs

finds that migrants of working age in the sample of non-migrant workforce in the 14 studied countries HICs tend to have higher labour force participation are employed in the informal economy, compared than non-migrants, on average (72.1 per cent and to about 66.5 per cent of migrant workers. The 69.0 per cent, respectively), with notable variations gap among wage workers is even wider, with across countries. Among the sample of LMICs, how- about 50.8 per cent of non-migrant wage workers ever, migrants of working age tend to have lower employed in the informal economy compared to labour force participation than non-migrants, on 62.4 per cent of migrant wage workers. In terms of average (62.0 per cent and 64.6 per cent, respec- distribution by sex, informal employment is higher tively). In terms of distribution by sex, migrant men among migrant women than among migrant tend to have higher labour force participation rates men, on average (66.4 per cent and 65.7 per than their non-migrant counterparts, on average, in cent, respectively). Likewise, informality is higher the sample of HICs (83.1 per cent and 74.1 per cent, among women nationals than among their men respectively), but have lower participation rates counterparts (67.1 per cent and 60.9 per cent, than their non-migrant counterparts in the sample respectively).11 It is significant, however, to add that of LMICs (78.6 per cent and 81.7 per cent, respec- the estimates cover only two HICs (Argentina and tively), with variations across countries. Among Chile) and 12 LMICs. These countries host roughly women, migrant women tend to have lower labour only 5.3 per cent of international migrants and force participation rates than non-migrant women, about 3.0 per cent of migrant workers worldwide. on average, in both the samples of HICs (61.3 per By looking at the distribution of wage workers by cent and 64.0 per cent, respectively) and LMICs industrial sector, the report finds that, on average, (45.9 per cent and 48.4 per cent, respectively). migrant wage workers, compared to nationals, are The report finds that more active migrant parti disproportionately represented in the primary sector cipants (in the labour market), especially women – agriculture, fishing and forestry – in the sample migrants, in 14 of the studied countries – where of HICs (2.5 per cent and 1.5 per cent, respectively), data on informality is available – tend to be infor- while in the sample of LMICs, the proportions of both mally employed compared to the non-migrant groups are similar (10.6 per cent and 10.3 per cent, workforce. Notably, about 63.2 per cent of the respectively). In the sample of HICs, more migrant

11 The estimates are weighted to account for each country’s population size. Based on simple averages, about 70.3 per cent of the non-migrant workforce in the 14 studied countries have informal employment compared to about 70.4 per cent of migrant workers. In terms of sex, about 74.8 of migrant women active in the labour market engage in the informal economy compared to 66.4 per cent of migrant men. Executive Summary xxv

Cape Town, South Africa – August 2020: African business woman start her own small business, informal trading during the COVID-19 pandemic. © shutterstock.com

wage workers than nationals take up secondary Similar to findings from previous ILO research, the sector jobs – mining and quarry; manufacturing; report shows that migrant workers in both the sam- electricity, gas and water; and construction – (26.8 ples of HICs and LMICs are, on average, more likely per cent and 20.8 per cent, respectively), while in the than nationals to work under temporary contracts sample of LMICs, they (migrant wage workers) tend (27.0 per cent and 14.9 per cent, respectively in the to take up fewer secondary sector jobs, on average, sample of HICs, and 42.9 per cent and 41.7 per cent, than nationals (24.9 per cent and 32.6 per cent, respectively in the sample of LMICs), with few excep- respectively). However, while there is a tendency for tions including Australia, Canada, Chile, Hungary, fewer migrant workers to be employed in the tertiary Ireland, and Latvia (among the sample of HICs); and sector (i.e services) than nationals in HICs (70.7 per Bangladesh, Malawi, and Mexico (among the sam- cent and 77.7 per cent, respectively), they tend to ple of LMICs), and variations across countries. This take up more tertiary sector jobs than nationals in corroborates the findings of earlier ILO research12 the sample of LMICs, on average (64.6 per cent and according to which migrant workers are particularly 57.1 per cent, respectively), with few exceptions, prone to be employed in non-standards jobs. Entry including in Costa Rica, the Gambia, Jordan, Namibia, Nepal, and Turkey. In terms of distribution by gen- through temporary migration programmes or indi- der, migrant men wage workers tend to work more vidual characteristics are often one of the reasons. than their national counterparts in the primary and In addition, migrant workers tend to be overrepre- secondary sectors in the sample of HICs and the ter- sented in sectors with traditionally high incidence tiary sector in the sample of LMICs Similarly, migrant of non-standard jobs. As a consequence, migrant women wage workers tend to work more than their workers may also be more likely to suffer from the national counterparts in the primary and secondary disadvantages inherent to non-standards forms of sectors in the sample of HICs and the primary and employment, a fact that has become more evident tertiary sectors in the sample of LMICs. during the COVID-19 pandemic across the world.

12 See: ILO. 2016. Non-standard employment around the world: Understanding challenges, shaping prospects (Geneva, ILO). xxvi The migrant pay gap: Understanding wage differences between migrants and nationals

The report also finds that incidence of part-time different migration corridors, where bilateral labour work is slightly higher among migrant workers than agreements are negotiated for different wages for non-migrant workers in HICs but lower than non-mi- a segment of the migrant population depending on grant workers in LMICs, on average. Migrant work- the migrants’ countries of origin.The following are ers have slightly higher part-time incidence rates some important policy implications emerging from than non-migrant workers in the sample of HICs, the findings of this report: on average (15.0 per cent and 14.6 per cent, respec- X Monitoring the impact of the ongoing COVID-19 tively), primarily due to the significantly higher inci- crisis on migrant workers is important in ad- dence of part-time work contracts among migrant dressing their specific vulnerabilities women compared to non-migrant women. While part-time incidence rates of migrant men is slightly While estimates presented in this report reflect lower than that of non-migrant men in the sample periods prior to the COVID-19 crisis, the findings of HICs (7.7 per cent and 8.3 per cent, respectively), bear enhanced relevance in the face of COVID-19. an average gap of 2.2 percentage points exists The ongoing worldwide COVID-19 crisis has put between the part-time rates of migrant women a spotlight on decent work deficits among men and non-migrant women in HICs (23.8 per cent and and women migrant workers around the world. 21.6 per cent, respectively), although the scale of the Experiences from previous economic crises sug- difference varies widely across countries. gest that the economic downturn associated with the COVID-19 pandemic may have dispropor- In the sample of LMICs, incidence of part-time work tionate and long-lasting negative effects on the tends to be lower among migrant workers than integration of migrants in their countries of desti- among non-migrant workers, on average (6.2 per nation.13 Recent survey data from Mexico and the cent and 8.7 per cent, respectively), with notable United States that covers up to the third quarter variations across countries. Both migrant men of 2020 shows that migrant workers have been and migrant women in LMICs tend to have lower among the hardest hit by the COVID-19 crisis, both part-time incidence rates than their national coun- in terms of employment losses and a decline in terparts, on average (3.9 per cent and 6.5 per cent earnings for those who have remained in employ- of migrant men and non-migrant men, respectively, ment. In view of these recent changes, the migrant and 10.3 per cent and 12.0 per cent of migrant pay gap estimates presented in this report are women and non-migrant women, respectively), likely to widen during and after the crisis.14 Analysis although part-time work is more prevalent among of the social and economic outcomes of men and women than among men in general. women migrant workers therefore remain most relevant in the immediate and long-term response What are the policy to the COVID-19 crisis. As countries safeguard their economies during and beyond the pandemic, implications? there is a need to monitor and protect the rights A major question emerging from the analysis in this of migrant workers. This should include covering report is, what can be done to progressively reduce them in national COVID-19 policy responses, such migrant pay gaps observed across countries, in as ensuring that migrant workers are covered by particular in HICs and in some LMICs, including measures relating to wage subsidies, and facilitat- through the effective application of the principle of ing their access to social security, including health “equal pay for work of equal value”. While there is care and income protection measures. a range of measures that can be taken to reduce X Reliable data, including data on wages of these pay gaps, the answer to this question will migrant workers and nationals, is needed on necessarily be country specific. This is because the other regions and countries of destination factors that drive and explain migrant pay gaps vary from country to country as well as in different parts Quality of data is key, notably availability of reli- of the wage distribution. They may also vary across able data on the distribution of wages amongst

13 See OECD, 2020a. Managing international migration under COVID-19, OECD Publishing, Paris. 14 For example, Fasani and Mazza (2020b) shows that migrant workers in the EU are more likely to be in temporary employment, earn lower wages and have jobs that are less amenable to teleworking during the COVID-19 crisis compared to non-migrant workers (see, https://ec.europa.eu/jrc/en/publication/ vulnerable-workforce-migrant-workers-covid-19-pandemic). The OECD’s International Migration Outlook 2020 finds that the COVID-19 crisis is reverting the trend of progress and jeopardising more than a decade of progress in migrant labour market inclusion in OECD countries (see, https://doi.org/10.1787/ ec98f531-en). Executive Summary xxvii

Guatemala agricultural migrant workers heading to Canada. © Copyright ILO

migrant workers and nationals, in particular for and production of labour migration statistics at other regions and countries of destination not national, regional and global levels, as well as covered in this report. This would help bridge the development of international concepts and the existing data gap, for example, with regard standards on labour migration statistics agreed to data on migration to Asia and the Arab States worldwide. (in particular the Gulf Cooperation Council (GCC) X There is a need to go beyond simple summary countries) and within North Africa, and South- measures of the migrant pay gap East Asia and the Pacific. It is important to go beyond simple summary Consideration could be given to reviewing and measures of the migrant pay gap (such as the modifying existing surveys across these coun- average (the mean) or median migrant pay gap) tries by introducing modules specifically related in order to understand the underpinning causes to migrant pay gaps into cross-sectional surveys. and thus identify the most effective policy mea- What the report recommends here is that these sures to reduce the gaps. This can be done by integrated modules should capture the labour examining in more detail the respective wage market outcomes of both migrant workers and structures of migrant workers and nationals, nationals, including information on wages and including their gender dimensions. In particular, it working conditions. is essential to analyse the migrant pay gap at different locations in the wage distribution (includ- The ILO is currently working towards filling a ing decomposing the gap into explained and part of this gap by implementing the Guidelines unexplained parts) as well as in different sectors Concerning Statistics on International Labour of the economy, and to calculate factor-weighted Migration (see ILO, 2018c), in particular focusing migrant pay gaps, which account for composition on appropriate methodologies for capturing and effects in estimating the pay gaps. collecting data on the main categories and subcategories of international migrant workers. This Computing migrant pay gaps at different points in is part of the ILO effort to improve the collection the wage distribution as well as in different sectors xxviii The migrant pay gap: Understanding wage differences between migrants and nationals

of the economy has important policy implications. ence across cou ntries, in large part due to lack of For example, a well-designed minimum wage with adequate skills recognition systems for qualifica- broad legal coverage – including those sectors tions of migrant workers. It also highlights the and occupations in which migrants are chiefly vulnerable position these migrant workers have employed - could reduce the migrant pay gap at in labour markets.15 Discriminatory practices may the lower end of the wage distribution. To maxi- also prevent migrant job seekers from obtaining mize the effect of minimum wages, setting lower employment in accordance with their education wage levels for sectors in which migrant workers and skills. Skills mismatch translates into migrant often predominate such as domestic work or workers being concentrated in lower-paid occu- agriculture should be avoided. Collective agree- pations, contributing to the observed migrant ments that include provisions on equal pay and pay gaps. pay transparency could have a similar effect in the middle and upper ends of the wage distribution. Educational or retraining programmes targeting Finally, policies and measures that promote train- men and women migrant workers who are more ing and equal opportunity for upward mobility for likely to be affected by skills mismatch, particu- migrant workers in the labour market, especially larly in countries where migrant workers earn for those with long-term residence, could have a significantly less than non-migrant workers, could positive effect on wage levels in senior positions. help to reduce the migrant pay gap. Reducing Likewise, eliminating discrimination and address- polarization and occupational segregation may ing occupational segregation of migrant workers require changing social and cultural perceptions in lower paid occupations and sectors may also and stereotypes contributing to discrimination help reduce the migrant pay gap. against migrants; creating opportunities for men and women migrant workers to enter into a wider Measures that promote the formalization of range of occupations, including managerial and the informal economy – such as extending to professional occupations, which offer better all workers, including migrant women, the right paid employment opportunities; and combating to a minimum wage and social security – can employer prejudice in hiring and promotion deci- also greatly benefit migrant workers, especially sions. More generally, labour market integration women, bringing them under the umbrella of measures can help reduce skills mismatch in legal and effective protection and empowering terms of access to jobs or recognition of foreign them to better defend their interests. qualifications. These measures can also help X Tackling the "explained" and "unexplained" counter discriminatory practices, including with parts of the migrant pay gap, including respect to pay, against migrant workers and through education, changing stereotypes, promote the principle of equal pay for work of and combating employer prejudice in hiring equal value, which would in turn help narrow and promotion decisions the unexplained part of the migrant pay gap and reduce working poverty among migrant workers, A significant share of migrant workers in paid especially among women. employment in many countries, in particular HICs, have higher levels of education and skills In any event, reducing the migrant pay will require relative to non-migrant workers but receive lower a broader strategy that includes also the adoption returns to these endowments. According to ILO of fair and effective labour migration policies that research these high levels of over-education address decent work deficits and ensure greater and skills mismatch among migrant workers is coherence across employment, education and consistent with the fact that immigrants have training, and other relevant policies at national, difficulties transferring their skills and experi- regional and global levels.

15 Sparreboom, T.; Tarvid, A. 2017. Skills mismatch of natives and immigrants in Europe (Geneva, ILO). Chapter 1 – Background 1 Chapter 1

Background

X 1.1. Introduction

International migration today is linked, directly or tainable. This is particularly true in view of today’s indirectly, to the world of work and the quest for aging population in many high-income countries decent work opportunities. Even if employment (HICs), a trend that is expected to grow in impor- may not be the primary driver for the initial move- tance over time17. ment, it usually features in the migration decision In many countries, men and women migrant work- process at some point. Family members joining ers represent a significant share of the workforce migrant workers abroad may also take up work, making important contributions to societies and either as employees or self-employed (ILO, OECD economies, and serving on the front lines carrying and World Bank, 2015). out essential jobs in health care and social work, According to United Nations (UN) estimates16, the transport, services, construction, and agriculture share of international migrants in the total world and agro-food processing (OECD and ILO, 2018)18. population was 2.8 per cent in 2000 and reached In spite of this, migrant workers often face disad- 3.5 per cent by 2019. The share of international vantage and unequal treatment in labour markets men migrants in the total male population across around the world including wage discrimination the world was 3.6 per cent and that of women and high recruitment fees and related costs (see, migrants was 3.4 per cent in 2019. In HICs, the e.g., ILO, 2016a, 2019a; ILO, 2020a, forthcoming).19 corresponding share has risen from 9.3 per cent in Furthermore, most migrant workers are concen- 2000 to 14.0 per cent in 2019, with the share of men trated in sectors of the economy with high levels of migrants in HICs estimated at 14.7 per cent in 2019 temporary, informal or unprotected work, charac- and that of women migrants standing at 13.2 per terized by low wages and lack of social protection, cent. Based on figures for 2017, the ILO estimates including in care work which in many countries is that there are 234 million international migrants of largely carried out by women migrant workers (ILO, working age worldwide, of whom about 164 million 2018d). One way to measure labour market differ- are workers; about 42 per cent of these workers are ences between migrant workers and nationals is by women (ILO, 2018b). comparing the earnings of migrant workers to that of non-migrant workers with similar labour market Owing to diverging demographics across coun- characteristics. This report is concerned with such tries, among other reasons, international labour a comparison. migration often creates a triple-win opportunity for countries of origin and destination, their nationals, The general principle of equal remuneration for and migrant workers and their families, at least work of equal value is set out in the Preamble of in the short run. The win for countries of origin the ILO Constitution,20 and is further embodied in is due to stabilized employment and remittances ILO Conventions, including the fundamental ILO as well as new skill sets from returnees. The win Conventions on equality and non-discrimination.21 for countries of destination stems from the fact The dedicated ILO Conventions concerning migrant that migrant workers are able to support a level workers also require ratifying States to ensure of economic activity that may otherwise be unat- equal treatment between migrant workers and

16 United Nations, Department of Economic and Social Affairs. Population Division (2019). International Migrant Stock 2019 (United Nations database, POP/ DB/MIG/Stock/Rev.2019), available at: https://www.un.org/en/development/desa/population/migration/data/estimates2/estimates19.asp. 17 See information on the World Health Organization website on Global Health and Aging, available at: https://www.who.int/ageing/publications/global_ health.pdf. 18 For example, using the most recent wave of the European Union (EU) Labour Force Survey, Fasani and Mazza (2020a) quantify the prevalence of migrant workers in key professions that the European Commission and Member States have identified and show that migrant “key workers” are essential for critical functions in European societies. They also highlight the contribution of migrant workers to the ongoing effort to keep basic services running in the European Union during the COVID-19 epidemic. 19 Recruitment fees and related costs are those which are incurred by workers in the recruitment process in order to secure employment or placement, regardless of the manner, timing, or location of their imposition or collection (ILO, 2019a). 20 ILO: Constitution of the International Labour Organisation (Geneva, 1919), available at: http://www.ilo.ch/dyn/normlex/en/f?p=1000:62:0::NO:62:P62 _LIST_ENTRIE_ID:2453907:NO. 21 Discrimination (Employment and Occupation) Convention, 1958 (No. 111), and the Discrimination (Employment and Occupation) Recommendation, 1958 (No. 111), para 2(f). The Equal Remuneration Convention, 1951 (No. 100) aims to promote the principle of equal remuneration between men and women workers, and covers all workers, including migrant workers. Chapter 1 – Background 3

nationals with respect to remuneration. The prin- 2020). Migrant workers are often the first to be ciple is embodied in the Migration for Employment laid-off, and often excluded from national COVID-19 Convention (Revised), 1949 (No. 97) and the Migrant policy responses, such as wage subsidies, unem- Workers (Supplementary Provisions) Convention, ployment benefits or social security and social 1975 (No. 143),22 as well as Migrant Workers protection measures (ILO, 2020d). In view of these Recommendation, 1975 (No. 151) which explicitly recent changes, the impact of COVID-19 on employ- refers, in its Paragraph 2(e) to equality of treatment ment opportunities and income losses may be with respect to remuneration for work of equal greater for migrant workers – especially women, value. The United Nations Sustainable Development who are often over-represented in the informal Goal (SDG) 8 Decent work and economic growth, tar- economy and more likely to work as personal gets 8.5 and 8.8, set out to achieve by 2030, “equal care and domestic workers (ILO, 2018d) – than for pay for work of equal value” and “protected labour non-migrant workers. rights for all workers, including migrant workers, Previous ILO research, including the ILO Global in particular women migrant workers, and those in Wage Report 2014/15, already highlighted the precarious employment” (UN, 2017a). existence of significant wage differences between While in most countries, national legislation rec- migrant workers and non-migrant workers in ognizes the principles of non-discrimination and some countries. While at the national level, there equality of treatment, challenges continue to exist have been attempts to analyse wage differences in ensuring these rights for migrant workers, in between migrant workers and nationals in several particular with respect to pay. Regardless of exist- countries (some of which are documented in sec- ing legislative provisions prohibiting all forms of tion 1.2 of this report), overall, global analysis of discrimination including with regard to remunera- the migrant pay gap, including its gender dimen- tion, unfair labour market differences exist between sions, has however been limited. Nonetheless, the migrant workers and nationals around the world. migrant pay gap continues to be of serious concern. The ILO supervisory bodies have noted situations Knowing its nature and extent is essential not only in which migrant workers face less favourable treat- for improving the protection of migrant workers’ ment compared to nationals including with respect rights and avoiding social dumping but also, for to remuneration, either due to non-inclusion of avoiding unfair competition and labour market dis- migrant workers in minimum wage legislation, or tortions. Addressing the migrant pay gap, including due to significant unlawful differences between by affording migrant workers equality of treatment, migrant workers and nationals, in law or in prac- will contribute to well-functioning labour markets, tice.23 These differences are likely to widen during which will be particularly important as countries and beyond the COVID-19 crisis, where men and seek to emerge and build back after the COVID-19 women migrant workers are among the hardest hit. crisis. The present report therefore aims to pro- Recent reports document rising levels of COVID-19 vide a first detailed and up-to-date analysis of the related discrimination and xenophobia against migrant pay gap worldwide, including its gender migrant workers and in some cases food insecu- dimension, using recent available data from 49 rity, layoffs, worsening working conditions includ- countries and covering nearly half (49.4 per cent) ing reduction or non-payment of wages, cramped of all international migrants and roughly 33.8 per or inadequate living conditions, and increased cent of migrant workers worldwide. restrictions on movements or forced returns (see, The report draws from the methodology used in the e.g., Fasani and Mazza, 2020b; Hubbard, 2020; ILO, ILO Global Wage Report 2018/19, which provides a 2020b, 2020c; OECD, 2020b; Pilling and England, detailed analysis of pay inequalities between men

22 Article 6(1)(a) of Convention No. 97 requires ILO member States to ensure that no less favourable treatment, without discrimination based on nationality, race, sex or religion, is applied to migrant workers lawfully in the country than that which is applied to nationals with regard to remuneration; The Migrant Workers (Supplementary Provisions) Convention, 1975 (No. 143) requires equality of treatment for migrant workers whose legal status cannot be regularized, with respect to rights arising out of past employment in respect of remuneration, social security and related benefits; Article 10 requires the adoption and implementation of a national equality policy for migrant workers lawfully in the country, while Article 12(g) provides for equality of treatment for these migrant workers with respect to conditions of work. In addition, the principle is referred to in the Domestic Workers Convention, 2011 (No. 189), Article 11, and the Domestic Workers Recommendation, 2011 (No. 201), Paragraph 14. 23 For example, in Hong Kong (China): Committee of Experts on the Application of Conventions and Recommendations (CEACR), Convention No. 97, observation 2014 and direct request 2020; and in Malaysia (Sabah): CEACR, Convention No. 97, observations, 2019; Saudi Arabia – CEACR, Convention No. 29, Observations 2016 and 2020; Argentina – CEACR, Convention No. 189, direct request, 2020;. See also: ILO. 2016. Promoting Fair Migration: General Survey concerning the migrant workers instruments. Report of the Committee of Experts on the Application of Conventions and Recommendations, Report III (Part 1B), International Labour Conference, 105th Session, 2016, paras 367–384 (ILO, 2016d). 4 The migrant pay gap: Understanding wage differences between migrants and nationals

X Box 1. Relevant concepts and definitions

The following concepts and definitions have been used for the purpose of this report. Migrant: In accordance with ILO’s guidelines concerning statistics of international labour migration which were adopted by the International Conference of Labour Statisticians in 2018 (ILO, 2018c), the term “international migrant” refers to a person of working age present in a country of measurement who is not a citizen of that country. The datasets from which estimates in this report are drawn, and which have been produced by national statistical agencies (see Appendix II for details on data sources) allow the determination of whether individuals are citizens or migrants in the country of measurement. The report focuses on those who have migrant status in the latest year for which data is available. Thus, individuals who migrate and later take up the nationality of their countries of destination (referred to as “foreign-born” citizens) are treated as nationals in this report. International migrant workers include all non-citizens who have labour market attachment (i.e. either employed or unemployed). By law, “foreign-born” citizens of a country should have similar treatment in the labour market as native-born citizens. Hence the focus of this report is on non-citizens of that country. It is acknowledged however that the use of country of birth is equally common in empirical analysis (and in accordance with international statistical standards) and that the UN-DESA numbers of migrants are generally produced on the basis of country of birth. In view of the definition of migrants, together with the focus on the working age population (rather than the total population), and the fact that the data sources for the report have been selected on the basis of availability of information on labour market outcomes including wages, the estimates of the size of the migrant population, and other estimates, may be different relative to estimates from other sources such as the UN-DESA. The data used in the report are derived from nationally representative labour force surveys whose sampling designs are not focused on migrants (by birth or by citizenship) but rather on geographical stratification and/or clustering. Subject to these limitations, our expectation is that the estimates provided in this report should provide at least a reasonable approximation of labour market gaps between nationals and migrant workers across the studied countries. National/non-migrant: Throughout the report, the term “national” or “non-migrant” refers to an individual who currently lives and/or works in his or her country of citizenship, regardless of whether the individual is native-born or a foreign-born citizen. Wage employees/workers: According to the International Classification of Status in Employment (ICSE-93), “employees” are workers who hold “paid employment jobs”, that is, jobs in which the basic remuneration is not directly dependent on the revenue of the employer. Employees include regular employees, workers in short-term employment, casual workers, outworkers, seasonal workers and other categories of workers holding paid employment jobs (ILO, 1993 and 2008). Care workforce: The care workforce is defined, where possible, to include care workers in care sectors (education, health and social work), care workers in other sectors, personal care and domestic workers, and non-care workers in care sectors, who support care service provision. Pay: Due to limitations in identifying basic pay or hourly wage in survey data, throughout the report the term “pay” refers to wages or earnings received by wage employees including basic pay and additional allowances. This is different from income received from other modalities of labour market participation, for example, self-employment. In this sense, the terms “migrant pay gap” and “migrant wage gap” are used interchangeably, irrespective of whether pay refers to hourly wages, monthly earnings, or any other way of describing earnings arising from wage employment. In order to minimize the impact of outliers on the estimates produced in the report (for example, the impact of the wages of legislators, senior officials and CEOs), the top percentile of wage earners in each country is excluded from the analysis. Gini coefficient: The Gini index or Gini coefficient is a statistical measure of distribution developed by Corrado Gini in 1912. It is often used as a gauge of economic inequality, measuring income distribution among a population. The coefficient ranges from zero to 1 (or 100 per cent), with zero -re presenting perfect equality and 1 (or 100 per cent) representing perfect inequality. Thus, a country in which every resident has the same income would have an income Gini coefficient of zero per cent. A country in which one resident earned all the income, while everyone else earned nothing, would have an income Gini coefficient of 1 or 100 per cent. The Gini index, however, should not be taken Chapter 1 – Background 5

as an absolute measure of income as high-income and low-income countries can have the same Gini coefficient if incomes are distributed similarly within each country. In this report, the Gini coefficient is computed based on hourly wages. High-income countries (HICs): The World Bank defines a high-income country as one that has a gross national income (GNI) per capita exceeding US$12,375 as of July 2019. By this national income per capita threshold, there are 80 economies classified by the World Bank as HICs as of July 2019. Of these, 33 representing about 41 per cent are covered in this report. Most of the 33 countries covered also double as “countries of destination” of international migrant workers. Low- and middle-income countries (LMICs): According to the World Bank, low-income countries are nations that have a per capita gross national income (GNI) of less than US$1,026 as of July 2019. Middle-income countries, on the other hand, are nations that have a per capita gross national income (GNI) between US$1,026 to US$12,375. This category is further divided into two: lower-middle income economies (US$1,026 to US$3,995) and upper-middle income economies (US$3,996 TO US$12,375). Among the 138 economies within the low-income and middle-income group classifications, this report covers only 16, which are combined to be one group (LMICs). Most of the countries covered under the LMIC group double as “countries of origin” for a substantial number of international migrant workers in HICs. In view of the small sample size of this big group of countries, together with the fact that the proportion of wage employees are usually low in comparison to the overall active labour market participants in LMICs, the analysis of this income group category covers only a small part of the labour markets of LMICs.

and women around the world (ILO, 2018a).24 The predate the COVID-19 crisis period. The impact of main objective of the report is to highlight labour the pandemic may, however, lead to a widening market differences between migrant workers and of labour market differences between migrant nationals, including migrant pay gaps across coun- workers and nationals, which may in turn deepen tries. It does so with a view to facilitating the adop- the migrant pay gaps presented in the report. tion and implementation of evidence-based labour migration policies around the world, and ensuring The report draws on a large volume of data. For this that these are gender-responsive. The report also reason it has several sections and subsections and contributes towards achieving SDG targets 8.5 and is organized as follows. Chapter 1 presents back- 8.8. In addition, the information contained in the ground information with an introduction and sum- report can help set the basis for monitoring wage mary of related literature on migration and labour inequalities between migrant workers and non-mi- market outcomes. Box 1 summarizes some relevant grant workers around the world, help support the concepts and definitions used throughout the case for closing these gaps, and encourage further report. Chapter 2 describes the data and provides research on policies and practices that are effective summary statistics of labour market characteristics for promoting change. of migrant workers and nationals, including gender dimensions. Chapter 3 evaluates the migrant pay For the purpose of this report, the migrant pay gap and decomposes the gap into explained and gap – expressed in its simplest form – refers to the unexplained parts. Chapter 4 presents simulation difference in average wages between all migrant results by analysing migrant pay gaps based on a workers and all non-migrant workers who are counterfactual wage distribution for migrant work- engaged in paid employment. The report focuses ers. The fifth and final chapter presents conclusions on migrants and non-migrants of working age, and recommendations. The Appendices contain the which is defined in this report as individuals of age methodology used in estimation the migrant pay 16-70 years across all the studied countries.25 It is gap, data sources, and supplementary statistical important to emphasize that the migrant pay gaps results from the analysis. presented in this report are based on data that

24 Appendix I explains the methodology used in estimating and decomposing the migrant pay gap. 25 Working age population is defined this way to make pay gap estimates comparable across countries. Nonetheless, we acknowledge the fact that variations exist in working ages across the countries covered in the report. 6

X 1.2. Related literature

The existing literature provides abundant research lower wages and have jobs that are less amenable on the flow of migration (see, e.g., ILO, 2018b; IOM, to teleworking during the COVID-19 crisis compared 2018; UN-DESA, 2019; UN, 2017b). There also exist to non-migrant workers (Fasani and Mazza, 2020b). ample evidence on recruitment fees and related The OECD’s International Migration Outlook 2020 costs associated with the general migration process. finds that the COVID-19 crisis is reverting the trend Although certain worker-paid migration costs have of progress and jeopardising more than a decade fallen in recent times following technological and of progress in migrant labour market inclusion in institutional changes,26 recruitment fees and related OECD countries (OECD, 2020b). costs remain considerably high, especially for low and semi-skilled workers27. The high economic There are different types of migration, each and social costs incurred by migrant workers are requiring different types of policy interventions increasingly recognized as serious impediments to to address gaps in labour market outcomes (ILO, realizing sustainable development outcomes from 2018c). Migration flows can be characterized by international migration (World Bank and ILO, 2019). their drivers (forced versus economic migration), their duration (permanent versus temporary), their In addition to high recruitment fess and related stance vis-à-vis the law (regular versus irregular), costs, migrant workers report substantial decent the basic characteristics of the migrants (age, gen- work deficits and short-falls in working conditions der, skills, among others), and the geography of such as those relating to contractual status, level the flows. These profiles determine how migration of wages and periodicity of wage payments, hours affects job opportunities, as well as the communi- worked, occupational safety and health, as well ties in countries of origin and destination. as trade union involvement and discrimination (Aleksynska, Kazi Aoul and Petrencu, 2017). The Migrant characteristics are especially important COVID-19 crisis has put a spotlight on the decent in understanding the labour market differences work deficits, including with respect to wages, in between migrants and nationals. Migration of the care sector, including domestic work, with men (internally and internationally) still dominates, migrant workers often doubly disadvantaged. though gender patterns and norms are shifting During the COVID-19 pandemic, men and women (Christiaensen, Gonzalez and Robalino, 2019; ILO, migrant workers, including domestic workers have 2018b). Women are not only migrating more than reported worsening working conditions including before, they are also migrating more for economic layoffs and reduction or non-payment of wages reasons (domestic and care work, in some cases), (ILO, 2020b, 2020c, 2020f). They are often excluded especially in Asia (Ingelaere et al., 2017; Lucas, from national COVID-19 policy responses, such as 2015). Migration by low-skilled workers is more wage subsidies, unemployment benefits or social often temporary (e.g. seasonal agricultural or con- security and social protection measures (ILO, struction workers), while migration by high skilled 2020d). Based on ILO’s rapid assessment survey workers is more likely to be permanent and further on experiences of migrant workers in ASEAN coun- away (World Bank, 2018a). Possibly, that low-skilled tries28 during the pandemic, migrant workers face, workers also move further from their home coun- among others, contract termination and reduced tries, for example, as is the case for low-skilled working days and pay, with the unemployed unable migrants moving within the EU (Farole, Goga and to access any social security support (ILO, 2020e). Ionescu-Heroiu, 2018), and migration of low-skilled In the European Union (EU), migrant workers are workers from Asia and Africa to the Middle East more likely to be in temporary employment, earn (ILO, 2017d). Unsurprisingly, the level of skills also

26 For example, studying real migration costs from India to the Cooperation Council for the Arab States of the Gulf (GCC), Seshan (2017) reports a 50 per cent decline in these costs on average between 1980 and 2014, driven by a reduction in visa fees and air ticket prices. This also seems to be due in part to the establishment of social migration networks. This has further been helped along by the ICT revolution, which has reduced communication costs and increased access to information. 27 For example, when using the SDG indicator 10.7.1: “Recruitment cost borne by an employee as a proportion of monthly income earned in country of destination”. For further details on this, see ILO (2019c) and World Bank and ILO (2019). For additional information on recruitment fees and related costs, and estimates of migration costs in particular, see, for example, Abella and Martin (2014), ILO (2016a), ILO (2020a, forthcoming) and Martin (2016). 28 Brunei Darussalam, Cambodia, Indonesia, Lao People's Democratic Republic, Malaysia, Myanmar, Philippines, Singapore, Thailand, and Vietnam. Chapter 1 – Background 7

affects the choice of destination. For example, high- and retention, and improve skills recognition and skilled migration rates are particularly high among certification of migrant care workers. developing countries. In the year 2000, one out of Kahanec and Zimmermann (2008, 2011), using every eight Africans with a university education data from the early 2000s, document differences in lived in a country in the OECD, the highest rate labour market characteristics – such as labour force among developing regions except the Caribbean, participation, unemployment, and occupational and Central America, and Mexico (World Bank, 2011, educational attainments – of nationals and migrant 2018b). According to the United Nations Conference workers in OECD countries. They argue that skilled on Trade and Development (UNCTAD), migration of labour migration has a large potential to reduce highly qualified professionals from the world’s 48 inequality in destination countries under standard least developed countries is quite stark, with about conditions. More recently, ILO Global Estimates on one in five university level educated professionals International Migrant Workers (2018b) show that: leaving for employment elsewhere (UNCTAD, 2012). (i) migrants worldwide tend to have higher labour These migration typologies should drive differences force participation rate than non-migrants (pri- in policy objectives as the constraints facing differ- marily due to the significantly higher labour force ent groups of migrants differ (Cho, 2017). participation rates of migrant women compared to Clear geographical patterns of international migra- non-migrant women); (ii) prime-age adults (ages tion emerge from the available data. Apart from 25-64) constitute about 87 per cent of international the much-emphasized pattern of the Global South- migrant workers; (iii) migrant workers are concen- North migration, there are three other important trated in HICs, with about 68 per cent of migrant migration corridors: (i) from South Asia to the workers worldwide employed in HICs; and (iv) Gulf countries (ILO, 2015a); (ii) within and from migrant workers are geographically concentrated, Latin America to North America (ILO, 2017a); and with about 61 per cent of all migrant workers found (iii) between countries in Africa, especially in West in three subregions: Northern America; Northern, Africa and southern Africa, with more than 80 per Southern and Western Europe; and the Arab States. cent of migration flows of African nationals taking In terms of the migrant pay gap, previous ILO place within the continent and the remaining flow research including the Global Wage Report 2014/15 taking place in spatially diversified locations beyond (2014a) highlighted significant pay gaps between colonial patterns (Flahaux and De Haas, 2016; ILO, nationals and migrant workers in some countries 2019b). in Europe and Latin America. Among European countries, for example, the Global Wage Report In relation to migration for jobs in the care economy, 2014/15 found that the migrant pay gap in 2010 the ILO’s Care Work and Care Jobs for Decent Work was about 9.8 per cent in Belgium, 14.9 per cent in Report (2018d) notes that health worker migration is Luxembourg, and 34.8 per cent in Cyprus. In Latin a feature of global health labour markets, driven by America, the migrant pay gap in 2012 was 22 per working conditions and income differentials across cent in Argentina, –113.8 per cent in Brazil, and countries. The report also highlights that most care –10.8 per cent in Uruguay. workers are women, often migrants and working in the informal economy under poor conditions and Apart from the ILO, other researchers have exam- for low pay (ibid.). Women migrant workers also ined the migrant pay gap for individual countries. represent a significant share of those in domestic Borjas (1990 and 1995) uses the 1970, 1980, and work, comprising 73.4 per cent (or 8.45 million) of all 1990 United States census data to examine the migrant domestic workers (ILO, 2015c). In The Social mean migrant-national earnings gaps and projects Construction of Migrant Care Work, King-Dejardin that migrant workers in the United States will earn (2019) finds that due to the asymmetries between between 15 and 20 per cent less than nationals countries of origin and destination and often incon- throughout much of their working lives. Similarly, sistent law and policy on migration and care, working using the census data, Butcher and DiNardo (2002) conditions of migrant care workers tend to differ to a investigate the migrant pay gap in the United States greater or lesser extent from those of their national at different deciles of the earnings distribution and counterparts. King-Dejardin (2019) also highlights find evidence, based partly on gender differences, the need for effective policies to help improve the that the minimum wage strongly widened the pay governance of labour migration for health-care gap in 1990. This could reflect the possibility that workers (and all migrant care workers, by extension), a significant proportion of migrant workers in the address decent work deficits for better recruitment United States is concentrated around the minimum 8 The migrant pay gap: Understanding wage differences between migrants and nationals

wage. Chiswick, Le and Miller (2008) also investigate characteristics. In addition, the fourth Belgian determinants of the earnings distribution for native- Socio-economic Monitoring Report 2019 29 shows that born workers and migrant workers in Australia and labour market differences – including with respect the United States. They find significant earnings to wages – between people of Belgian and foreign gaps with magnitudes and determinants varying origin remain significant, even with the same level by sex, year, and immigrant cohort, as well as across of qualification and field of study. the deciles of the earnings distribution. They find The migrant pay gap measured simply as the a pattern of higher earnings for migrant workers so-called “raw” or “unadjusted” pay gap can occur than for nationals at the lowest earnings decile in for many reasons, including the fact that migrant Australia and posit that this may reflect favourable workers’ personal characteristics, such as skills and selectivity in migration for Australia. education, may be advantageous or disadvanta- In more recent years, Antón, de Bustillo and Carrera geous to them in their destination countries. Part (2012), based on the Spanish Labour Force Survey of the migrant pay gap may also be unexplained. 2006 (LFS 2006) and the Wage Structure Survey Employer discrimination against migrant workers 2006 (WSS 2006), find that migrant women in due to factors such as prejudice or mistrust may Spain face double disadvantage of being women account for part of the unexplained wage gap and migrants, in particular women from develop- (ILO, 2014a; Solé and Parella, 2003). Other possible ing countries. The authors estimate that migrant reasons include differences in returns to foreign-ac- women in Spain earn 20 per cent less than quired skills and education of migrant workers, as non-migrant women. Using 2010 data from the employers may not fully recognize these (see, e.g., Netherlands, Siebers and van Gastel (2015) find Barrett, McGuinness and O'Brien, 2012), possibly that migrant workers’/ethnic minority employees’ due to the fact that skills recognition systems are lower levels of participation in work-related com- not prevalent and robust (see, e.g., Braňka, 2016; munication and the application of socio-ideological ILO, 2017c; Rosangela and Annavittoria, 2017). labour control widens the migrant earnings gap in Moreover, migrant workers, particularly singles, the Netherlands. In Germany, Ohlert, Beblo and may receive lower wages if they are perceived as Wolf (2016) find that non-German workers face having lower income needs than their national significantly lower wages in establishments covered counterparts with families to support (Rubery, by collective bargaining agreements, with the aver- 2003). In other cases, migrant workers may be age wage gap estimated at around 11.1 per cent under represented in collective representation (in favour of German nationals) based on a panel structures because of difficulties in organizing or data for the period of 2000-2010. However, using because nationals dominate the overall representa- the German Integrated Employment Biographies tion – this could be exacerbated if migrant workers (IEB) covering a cross-section of individuals up to are perceived as a low-wage employment threat to 2015, Brunow and Oskar (2019) report that the nationals (ibid.). The most appropriate mix of policy migrant wage gap in Germany is mostly explained responses will differ across countries, depending on by observable characteristics (endowments), espe- which factors have the largest impact on the pay cially location, labour market experience, and firm gap in each national context.

29 See https://employment.belgium.be/en/publications/4th-socio-economic-monitoring-labour-market-and-origin-2019. Chapter 1 – Background 9 Chapter 2

Labour market characteristics of migrant workers and nationals

X Chapter 2. Labour market characteristics of migrant workers and nationals

For a better understanding of the complexities of in the population of wage workers, distinguishing the migrant pay gap, it is important to first under- between migrants and nationals. On average, stand some key labour market characteristics that across the sample of HICs and LMICs covered in the contribute to the economic success of migrant report, migrant women account for a slightly higher workers and can drive the migrant pay gap. This proportion of the total working age migrant popu- chapter therefore provides an empirical evaluation lation than migrant men, with the share of migrant of differences in key labour market characteristics women being 50.6 per cent in HICs and 51.3 per of migrants and nationals based on recent data cent in LMICs, although there are notable variations from 49 countries covering about 25 per cent of the across countries. Similarly, women nationals have a world’s wage employees, nearly half (49.4 per cent) higher share of the total working age population of of all international migrants, and roughly 33.8 per nationals than men nationals, with women nation- cent of migrant workers worldwide.30 Basic empir- als accounting for 50.8 per cent of the total working ical facts at the country level are presented on the age population of nationals in HICs and 51.6 per differences between migrant workers and nationals cent in LMICs. However, the share of women (both for personal characteristics, employment status, job migrants and nationals) among wage employees characteristics, working conditions and workplace is low, on average, compared to men, which is characteristics, with a particular focus on gender consistent with findings in the ILO Global Estimates and informality. Report on International Migrant Workers (2018b). Only 43.0 per cent of migrant wage workers in the sample of HICs are women and 32.0 per cent in the 2.1. Share of migrant sample of LMICs are women, with large variations population across countries. Similarly, among non-migrants, about 47.9 per cent of wage workers in HICs are Table 1 shows the proportion of migrants of work- women and 34.2 per cent in LMICs are women. ing age (i.e. ages 16–70) among the total working age population as well as the total population of wage workers in the 49 studied countries. These 2.2. Geographical coverage two population shares vary significantly across It is important to understand how representative the countries. Among the countries for which data the findings in this report are in terms of world- is available, Luxembourg has the highest share wide coverage of international migrant workers. of migrants in the total working age population The majority of countries and territories do not and the total population of wage workers, with have labour market data that includes wages of 43.1 per cent and 47.0 per cent, respectively. Jordan, both international migrant workers and nationals. Switzerland, Australia, and Canada host the second, The reports’ benchmark databases were drawn third, fourth and fifth highest share of migrants. from 188 countries and territories (representing On average, the stock of migrants of working age about 99.9 per cent of the world population in is about 9.3 per cent in the sample of HICs and 2017), as used in the ILO Global Estimates Report 1.1 per cent in the sample of LMICs. In terms of on International Migrant Workers (2018b). Of the wage workers, the average share of migrant wage 188 countries in the benchmark databases, the workers among the population of wage workers is report covers 49 countries where labour market about 9.2 per cent in the sample of HICs and 1.2 per data including data on wages of migrant work- cent in the sample of LMICs. ers exist. International migrants living in these Figure 1 shows the respective shares of women 49 countries represent nearly half (49.4 per cent) of and men in the working age population as well as all international migrants31. Wage workers, includ-

30 Computation is based on ILO’s own estimation based on its databases on global wage employees and international migration data from UN-DESA. 31 See previous footnote. Chapter 2 – Labour market characteristics of migrant workers and nationals 11

X Table 1. The share of migrant population among the total working age population, latest years

Migrants' Migrants' Migrants' Migrants' Country Latest share Country Latest share Country population Country population Code year of wage Code year of wage (%) (%) workers (%) workers (%)

High-income LVA Latvia 2015 14.67 13.72

POL Poland 2015 0.15 0.21 EST Estonia 2015 14.83 13.39

SVK Slovakia 2015 0.16 0.14 CYP Cyprus 2015 16.58 19.53

HUN Hungary 2015 0.49 0.58 CAN Canada 2018 24.80 24.13

HRV Croatia 2015 0.59 0.56 AUS Australia 2017 27.35 26.37

LTU Lithuania 2015 0.59 0.53 CHE Switzerland 2016 28.71 29.31

CZE Czech Republic 2015 1.64 1.29 LUX Luxembourg 2015 43.13 46.99

PRT Portugal 2015 2.14 2.19 Weighted 9.27 9.22 Average FIN Finland 2015 2.93 2.22

Low- and middle-income NLD Netherlands 2015 2.96 2.64

ROU Romania 2015 0.10 0.14 MLT Malta 2015 3.13 2.64

BGD Bangladesh 2017 0.15 0.13 SVN Slovenia 2015 3.61 4.40

MDG Madagascar 2012 0.20 0.31 NOR Norway 2015 4.67 4.93

BOL Bolivia* 2017 0.31 0.31 CHL Chile 2017 4.71 6.71

NPL Nepal 2017 0.39 0.77 DNK Denmark 2015 4.86 4.43

MEX Mexico 2018 0.51 0.47 FRA France 2015 4.87 4.29

BGR Bulgaria 2015 0.60 0.39 SWE Sweden 2015 5.45 4.23

SLE Sierra Leone 2014 1.03 1.22 ISL Iceland 2015 5.91 5.92

MWI Malawi 2013 1.30 1.33 ARG Argentina 2018 6.35 6.36

ALB Albania 2013 1.96 1.57 GRC Greece 2015 6.46 8.84

TZA Tanzania** 2014 2.22 4.66 ITA Italy 2015 8.90 11.23

ESP Spain 2015 8.94 9.01 TUR Turkey 2017 2.80 3.10

USA United States 2018 9.06 8.95 NAM Namibia 2016 4.39 5.43

United GMB Gambia 2018 5.74 5.90 GBR 2015 9.70 9.28 Kingdom CRI Costa Rica 2018 10.80 12.82 BEL Belgium 2015 10.36 8.83 JOR Jordan 2016 33.89 44.34 IRL Ireland 2015 11.92 13.52 Weighted 1.07 1.22 AUT Austria 2015 14.44 12.90 Average

Note: Refer to Box 1 for definition of a migrant. * the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). 12 The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure 1: The share of women and men among migrants and nationals, latest years

Total working age population

Working age migrants Working age non-migrants

Poland Malta Slovenia Luxembourg Slovakia Switzerland Estonia Slovenia Belgium Netherlands Norway Sweden United States Iceland Finland Denmark Luxembourg Canada Sweden Spain Czech Republic Finland Latvia Austria Canada Cyprus Switzerland Italy Chile United Kingdom Australia Australia Ireland Belgium Austria Norway United Kingdom Croatia High-income Spain High-income Slovakia Greece France Iceland Ireland Lithuania Czech Republic Argentina Poland Hungary Greece France United States Italy Hungary Denmark Portugal Malta Argentina Croatia Lithuania Cyprus Estonia Portugal Chile Netherlands Latvia High-income High-income Nepal Costa Rica Madagascar Turkey Jordan Jordan Namibia Bulgaria Gambia Madagascar Sierra Leone Romania Mexico Bangladesh Tanzania** Bolivia* Albania Albania Costa Rica Tanzania** Bolivia* Mexico Turkey Gambia w- and middle income Bangladesh w- and middle income Namibia Lo Lo Romania Malawi Bulgaria Sierra Leone Malawi Nepal Low- and middle-income Low- and middle-income 020406080 100 020406080100 Share (%) Share (%)

Share of women Share of men Chapter 2 – Labour market characteristics of migrant workers and nationals 13

Total wage employees

Migrant wage workers Non-migrant wage workers

Slovenia Malta Slovakia Chile Poland Italy Lithuania Greece Hungary Argentina United States Luxembourg Finland Poland Estonia Netherlands Czech Republic Austria Luxembourg Cyprus Greece Czech Republic Sweden Spain Belgium Croatia Latvia Hungary Austria Slovenia Chile Australia Australia Switzerland Argentina Iceland Norway United States High-income Malta High-income Canada France Belgium Switzerland Slovakia Ireland Denmark Canada France United Kingdom Norway Italy Sweden Denmark Portugal Spain United Kingdom Iceland Lithuania Croatia Finland Netherlands Ireland Portugal Latvia Cyprus Estonia High-income High-income Nepal Jordan Jordan Bangladesh Madagascar Sierra Leone Sierra Leone Nepal Bangladesh Gambia Gambia Turkey Turkey Tanzania** Malawi Madagascar Tanzania** Costa Rica Albania Mexico Namibia Bolivia* Mexico Albania w- and middle income Bolivia* w- and middle income Malawi Lo Lo Costa Rica Romania Romania Namibia Bulgaria Bulgaria Low- and middle-income Low- and middle-income 020406080 100 020406080100 Share (%) Share (%)

Share of women Share of men

Note: Estimates are based on the working age population (i.e. adults with ages 16-70). The top panel shows the proportions of women and men among the entire working age population whereas the bottom panel shows the respective proportions among wage workers. High-income and low- and middle-income estimates are the averages of the sample of high-income countries and low- and middle-income countries, respectively. Averages are weighted by the number of wage employees in each country. * the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). 14 The migrant pay gap: Understanding wage differences between migrants and nationals

ing migrant wage workers from the 49 countries X Table 2. Coverage of countries by income represent about a quarter of wage employees group worldwide. Number of countries covered Income group Total countries* in the report Tables 2 and 3 indicate that the sample of 49 coun-

tries covered in the report represent about 26 per Total % cent of the total 188 countries and territories cited as benchmark data sources for the ILO global Low-income 31 6 19.4 32 estimates. Table 2 shows the coverage by income Lower middle- 50 2 4.0 group. All countries with labour force data on inter- income national migrant workers and nationals had disag- Upper middle- 48 8 16.7 gregated data based on sex. While about 55.9 per income cent of HICs are covered, only about 16.7 per cent and 4.0 per cent of upper middle-income and lower High-income 59 33 55.9 middle-income countries are covered, respectively. Total 188 49 26.1 The coverage for low-income countries (19.4 per cent) is higher than both categories of middle- Note: *These are countries and territories used for the income economies. ILO global estimation (ILO, 2018b).

Table 3 shows that by geographical region, North America and Northern, Southern, and Western X Table 3. Coverage of countries by broad Europe had the highest coverage (100 per cent subregion and 80 per cent, respectively), followed by Eastern Number Europe (60 per cent) and South Asia (22.2 per cent). Total of countries Region Broad subregion Broad subregions with no coverage included North countries* covered in the report Africa and East Asia. Subregions with the least coverage were South-East Asia and the Pacific Total % (about 4.5 per cent), followed by Arab States and sub-Saharan Africa (8.3 per cent and 12.8 per Africa North Africa 6 0 0.0

cent, respectively). Sub-Saharan Africa 47 6 12.8

Box 2 lists the sampled countries and notes any Americas Latin America 31 5 16.1 exceptions. Germany is excluded from the European and the Caribbean Union (EU) countries as labour market data on its North America 2 2 100.0 migrant workers were not available at the time of writing the report. The EU however includes the Arab States Arab States 12 1 8.3 United Kingdom, as it was a Member country at Asia and the East Asia 8 0 0.0 the period under review. Pacific South-East Asia 22 1 4.5 Labour migration is an increasingly complex and the Pacific and dynamic phenomenon taking place within and between all regions of the world. In certain South Asia 9 2 22.2

migration corridors, such as between Asia and Europe and Northern, Southern 30 24 80.0 the Arab States (particularly, GCC countries) and Central Asia and Western Europe within South-East Asia and the Pacific, the num- Eastern Europe 10 6 60.0 ber of international migrants, the large majority

of whom are migrant workers, has tripled since Central and Western 11 2 18.2 1990 (ILO, 2017b). Though the migration flows Asia in these corridors are important, coverage in the Total 188 49 26.1 report of countries of destination in these regions is low. This is mainly because existing labour force Note: *These are countries and territories used for the surveys of countries of destination in these regions ILO global estimation (ILO, 2018b).

32 See Appendix III for detailed lists of countries and territory groups, by ILO region and by World Bank income groups. Chapter 2 – Labour market characteristics of migrant workers and nationals 15

X Box 2. List of countries covered in the estimates

Low- and middle-income countries

Albania Jordan Romania Bangladesh Madagascar Sierra Leone Bolivia, Plurinational State of Malawi Tanzania, United Republic of Bulgaria Mexico Turkey Costa Rica Namibia Gambia Nepal

High-income countries

Argentina Finland Netherlands Australia France Norway Austria Greece Poland Belgium Hungary Portugal Canada Iceland Slovakia Chile Ireland Slovenia Croatia Italy Spain Cyprus Latvia Sweden Czech Republic Lithuania Switzerland Denmark Luxembourg United Kingdom Estonia Malta United States

European Union*

Austria France Netherlands Belgium Greece Poland Bulgaria Hungary Portugal Croatia Ireland Romania Cyprus Italy Slovakia Czech Republic Latvia Slovenia Denmark Lithuania Spain Estonia Luxembourg Sweden Finland Malta United Kingdom

Note: * This comprises all Member States of the European Union (EU) in 2015 with the exception of Germany, which had no available data at the time of writing the report. Throughout the report, averages for the European Union are therefore computed using estimates for the 27 out of 28 member countries for which data was available. Non-EU migrants and EU migrants within the European Union are not distinguished in the report. EU nationals who relocate from their home EU country to another EU country are treated equally as migrants from outside the European Union.

do not sufficiently capture migrant households. Though there have been wide-ranging efforts to For example, among the 22 countries in the South- produce reliable and comparable data on labour East Asia and the Pacific region, labour force migration, the low proportion of international data that sufficiently capture migrant workers migrants covered in this report relative to the was available only for Australia; and among the proportion covered in the ILO Global Estimates on 12 Arab States, Jordan was the only country with International Migrant Workers (2018b) reinforces reliable data. the fact that data gap remains significantly high, in 16 The migrant pay gap: Understanding wage differences between migrants and nationals

particular data on labour market outcomes includ- histories and economic institutions. The bottom ing wages of migrant workers, as noted by the ILO images, plots (c) and (d), describe the relationship supervisory bodies and the international communi- between inequality and the presence of migrants ty.33 There is a need, therefore, to augment efforts when the outliers in plots (a) and (b) (in circles) are to produce reliable data on labour migration that excluded. Both figures suggest a weak negative captures labour market outcomes of migrant work- correlation between the share of migrants and the ers, especially for the GCC countries, North Africa, level of wage inequality in destination countries. Latin America and the Caribbean, East Asia, and Countries with lower inequality are also countries South-East Asia and the Pacific. with a notable migrant population. Figure A-1 (see Appendix IV) replicates figure 2 by 2.3. Weak negative correlation showing the association between the Gini coefficient and the estimated migrant pay gap from Chapter 3 between wage inequality and using the sample of HICs. The figure shows that the size of migrants population there is no clear relationship between wage inequalities (represented by the Gini coefficient) and the Empirically, it is possible to show the association unadjusted migrant pay gap at the mean hourly between the presence of migrant workers and wage. However, a higher wage inequality index wage inequality in destination countries. On the appears to be weakly correlated with higher levels one hand, migrants carry different amounts and of the unexplained migrant pay gap (see Chapter 3 forms of capital34 with them and represent differ- for details of the migrant pay gap). ent types of labour, thereby directly affecting the distribution of income in destination countries. Migrant workers have an indirect impact through 2.4. Labour market changing the productivity of incumbent production participation, unemployment, factors as well as impinging on the redistributive policies in the destination countries (Kahanec and and educational and Zimmermann, 2008, 2011). On the other hand, dif- occupational attainments of ferent migrants may specifically choose to migrate to or seek employment in countries with different migrants and nationals degrees of equality. For example, countries with Labour market participation of working age migrant a high share of migrant population, such as the populations characterizes the economic activity of United States, also have higher income inequality. migrant workers and their earnings prospects. The Others like the Scandinavian countries have rel- extent to which migrants integrate in their desti- atively low shares of foreign population and low nation countries in terms of their labour market degrees of inequality. outcomes is linked directly to their participation in Figure 2 shows the correlation between the share of the labour market. Moreover, migrants’ education migrants’ population and (hourly) wage inequality and prior occupations mirror their skills set and nor- in destination countries. Wage inequality is repre- mally raise their chances of obtaining well-paid jobs. sented by the Gini coefficient (ILO, 2018a).35 Plot (a) There are a number of factors that determine indi- shows the relationship for all countries regardless viduals’ economic attainments and drive the eco- of differences in economic institutions, redistribu- nomic gap between migrant workers and nationals. tion policies, as well as the nature, type, and history Among these, perhaps, the most significant one is of immigration. Plot (b) characterizes this relation- human capital. Migrants' labour market success is ship using European countries (with the exception a function of their skills, including language skills, of Luxembourg and Switzerland, which are two as well as the transferability of these skills to their outlier countries) that share similar immigration new economic environment. In practice, however,

33 See ILO. 2016. General Survey concerning migrant workers instruments, paras 647-650. See also, UN: New York Declaration for Refugees and Migrants, General Assembly, 71st session, A/RES/71/1, 3 Oct. 2016, para. 40; UN: Declaration of the UN General Assembly High-level Dialogue on International Migration and Development, 68th session, A/RES/68/4, 21 Jan. 2014, para. 28. 34 The forms of migration capital may include cultural, social and economic capital that can be converted into advantageous positions in social fields (see Nohl et al., 2006; Erel and Ryan, 2018). 35 The Gini coefficient summarizes the relative distribution of wages in the population, with lower values (closer to zero) indicating lower levels of inequality and higher values (closer to 100 per cent) indicating higher levels of wage inequality (See box 1 for a detailed definition). Chapter 2 – Labour market characteristics of migrant workers and nationals 17

X Figure 2. The relationship between the Gini coefficient and the share of migrants' population, latest year, age 16-70 years

(a) All countries (b) Europe (without Luxembourg and Switzerland)

60 60 y = –0.2x + 35 y = –0.1x + 30 R2 = 0.069 R2 = 0.006 50 50

40 40

30 30

20 20 Gini coefficient (%) Gini coefficient (%)

10 10 0 10 20 30 40 50 0 5 10 15 20 Share of migrant population (%) Share of migrant population (%)

60 60 y = –0.4x + 36 y = –1.3x + 33 R2 = 0.053 R2 = 0.336 50 50

40 40

30 30

20 20 Gini coefficient (%) Gini coefficient (%)

10 10 0 10 20 0 2 4 6 8 10 Share of migrant population (%) Share of migrant population (%)

Notes: Data on the Gini coefficient is taken from the Global Wage Report 2018/19, which provides comprehensive estimates of within country wage inequality for high-, middle- and low- income countries. Plot (a) shows a scatter plot of the Gini coefficient as a function of the share of migrant population in all the 49 destination countries (see Table 1) with a linear line plot. Plot (b) shows a similar plot but using data from European countries with the exception of Luxembourg and Switzerland. Plots (c) and (d) shows same plots in (a) and (b) respectively, excluding the countries in the circles. Figure A1 in Appendix IV replicates this figure by showing the relationship between the estimated migrant pay gap in Chapter 3 and the Gini coefficient, using the sample of high-income countries. Source: ILO estimates based on survey data provided by national sources (see Appendix II).

migrants are typically likely to be affected by skills labour force participation, unemployment, and edu- mismatch and may have difficulties transfer- cational and occupational attainments of migrants ring their skills and experience across countries and nationals. Labour force participation refers to (Sparreboom and Tarvid, 2017), in large part due to the ratio of “active labour market participants”36 lack of adequate skills recognition systems for qual- to all individuals of working age. Educational ifications of migrant workers. Migrants’ skills are attainment is pooled together and defined by four often not fully recognized and migrants frequently categories: “less than or equal to primary educa- resort to continuous work in lower-skilled jobs that tion”, “lower secondary education without high do not account for their higher skills level (ibid.). school diploma”, “secondary education (completed), including vocational training”, and “university This section presents statistics for the working age education”. Occupational attainment here refers to population of a sample of 33 HICs and 16 LMICs on the proportion of individuals who attain high-skilled

36 These include individuals of working age who are either wage employees, employers, own-account workers, unpaid workers or unemployed (but actively seeking for employment). 18 The migrant pay gap: Understanding wage differences between migrants and nationals

occupations with job titles including senior official, Among the 16 LMICs covered in the report, migrants chief executive officer (CEO), manager, professional, tend to have lower labour force participation than associate professional, and technical officer.37 non-migrants, on average (62.0 per cent and 64.6 per cent, respectively), with variations across Table 4 presents country level estimates of labour countries (see figure 3 and table 4). For example, force participation along with unemployment and migrants’ labour force participation is lowest in proportions of migrant workers and nationals with Turkey (50.2 per cent) and highest in Romania high-skilled occupations, separating estimates for (95.1 per cent), whereas that of nationals is lowest women and men to highlight gender differences in Gambia (38.0 per cent) and highest in Madagascar in labour market attributes. Figure 3 compares (93.9 per cent). Both migrant men and women have the labour force participation of men and women. lower participation rates than their non-migrant Table 5, on the other hand, presents disaggregated counterparts on average (78.6 per cent for migrant estimates of levels of education of migrants and men versus 81.7 per cent for non-migrant men, and nationals, including its gender dimensions. This is 45.9 per cent for migrant women versus 48.4 per to highlight some of the salient stylized patterns of cent for non-migrant women). migrant-national labour market gaps and the role of human capital in driving these gaps. The higher labour force participation of migrants in the EU and in HICs, particularly of migrant men, Migrants tend to have higher labour is consistent with the hypothesis of positive-selection, whereby it is individuals with stronger labour force participation than nationals market potential, economic motives, and a desire to in HICs, but not in LMICs work who tend to migrate for economic purposes (see Chiquiar and Hanson, 2005; McKenzie and Similarly to findings from the ILO Global Estimates Rapoport, 2010; Parey et al., 2015). The lower labour Report on International Migrant Workers (2018b), force participation of migrant women relative to migrants of working age in the EU and in the sam- migrant men and non-migrant women in the EU ple of 33 HICs tend to have higher labour force par- and in HICs may be explained by: (i) the fact that ticipation than non-migrants, on average (73.9 per migrant women are more likely to engage in unpaid cent and 67.0 per cent, respectively in the EU, and care work, which is a major barrier to women’s 72.1 per cent and 69.0 per cent, respectively in the labour force participation (ILO, 2018c); (ii) and the sample of HICs), with notable exceptions (table 4). higher likelihood of women to migrate for reasons For example, among the sample of HICs covered other than employment (for instance, for family in this report, migrants’ labour force participation reunification or humanitarian reasons), as well as is lowest in Slovakia (56.1 per cent) and highest in possible discrimination against migrant women Iceland (95.2 per cent), whereas that of nationals that reduces their employment opportunities (see, is lowest in Greece (60.0 per cent) and highest in eg, ILO, 2018b; Kapur, 2010; OECD, 2009). Iceland (89.3 per cent). In terms of distribution by sex, migrant men in the Migrant workers tend to have sample of HICs tend to have higher labour force higher unemployment rates than participation rates than non-migrant men, on average (83.1 per cent and 74.1 per cent, respectively), nationals in both HICs and LMICs with some variations across countries (table 4 and In comparison to nationals, men and women figure 3). Estimates for migrant women and non-mi- migrant workers have higher unemployment grant women based on data from the sample of rates on average than nationals in the EU and the 33 HICs are however different from the findings sample of 33 HICs (13.5 per cent and 8.3 per cent, from the ILO Global Estimates Report on International respectively in the EU, and 7.7 per cent and 5.9 per Migrant Workers (2018b). The average labour force cent, respectively in HICs), with notable variations participation in the sample of HICs is estimated at across countries (table 4). Both migrant men and 61.3 per cent for migrant women, which is lower than migrant women in the sample of HICs tend to have the average labour force participation for non-mi- higher unemployment rates on average than their grant women (64.0 per cent), though variations do non-migrant counterparts (7.5 per cent for migrant exist across countries (see figure 3 and table 4). men versus 6.2 per cent for men nationals, and

37 See section 2.5.3 for detailed description of various occupational categories. Chapter 2 – Labour market characteristics of migrant workers and nationals 19

8.0 per cent for migrant women versus 5.6 per cent of migrant men with high-skilled jobs is estimated for women nationals). This finding is consistent with at 25.3 per cent, only 19.8 per cent of migrant recent findings in the OECD’s International Migration women tend to have higher skilled positions. Outlook 2020 (see, OECD, 2020b). This phenomenon is consistent with the hypothesis of slow adaptation A different picture emerges among the sample of migrants to the labour market of their countries of 16 LMICs covered in the report. More men and of destination (Kahanec and Zimmermann, 2008, women migrant workers among the total working 2011), holding that it takes some time for migrant age migrants in LMICs obtain high-skilled jobs than workers to retrain and integrate in a new labour nationals, with the average proportion of migrant market. Migrant workers may begin as unemployed, workers with high-skilled occupations estimated but experience substantial advancement and sub- at 15.8 per cent, while the corresponding average stantial wage progression later in the life-cycle. proportion of non-migrant workers is estimated at 8.8 per cent. Among the 16 LMICs, the proportion In the sample of LMICs, migrant workers on aver- of men and women migrant workers attaining high- age have slightly higher unemployment rates than skilled jobs is lowest in Jordan (1.9 per cent) and their national counterparts (6.8 per cent and 6.2 per highest in Romania (56.3 per cent), whereas that of cent, respectively). In terms of distribution by sex, nationals is lowest in the United Republic of Tanzania while the estimated unemployment rate for migrant (2.7 per cent) and highest in Bulgaria (23.4 per cent). men is slightly lower than that for non-migrant Both men and women migrant workers in the sam- men (5.4 per cent and 5.9 per cent, respectively), ple of LMICs have more high-skilled jobs than their the corresponding estimate for migrant women is non-migrant counterparts on average (20.8 per higher than that for non-migrant women (8.6 per cent for migrant men versus 10.7 per cent for men cent and 6.4 per cent, respectively). nationals, and 11.3 per cent for migrant women versus 7.1 per cent for women nationals). Migrant workers tend to have fewer high-skilled occupations than Migrants tend to have higher nationals in HICs but more high- education than nationals in both skilled occupations than nationals HICs and LMICs in LMICs The educational composition of migrant workers Among the sample of 33 HICs and in the EU, and may partly explain the observed labour market based on latest available data, fewer men and gaps between migrants and nationals, though women migrant workers among the total working table 5 provides only a limited support for this age migrants obtain high-skilled jobs (such as senior conjecture. On average, the level of education, in official, chief executive officer (CEO), manager, particular the proportion of men and women with professional, associate professional, and technical university education is considerably higher among officer positions) than nationals on average. The migrant workers than nationals in both the samples average proportion of men and women with high- of HICs and LMICs, and shows large variations by skilled occupations is estimated at only 25.1 per sex. While about 30.5 per cent and 29.0 per cent of cent for migrant workers compared to 31.6 per migrants in the EU and the sample of HICs, respec- cent for nationals in the EU, and only 22.7 per cent tively have university education, on average, the for migrant workers compared to 32.5 per cent for corresponding estimates for nationals in the EU and nationals in the sample of HICs (table 4). However, the sample of HICs are 27.4 per cent and 28.8 per there exists remarkable variations across the sample cent, respectively. Among the sample of HICs of HICs. For example, the proportion of men and covered in the report, the proportion of men and women migrant workers who have high-skilled women migrant workers with university education jobs is lowest in Greece (6.2 per cent) and highest is lowest in Slovenia (8.1 per cent) and highest in in Malta (57.0 per cent), whereas that of nationals Ireland (46.3 per cent), whereas that of nationals is is lowest in Argentina (16.1 per cent) and highest in lowest in Chile (12.7 per cent) and highest in Ireland Slovenia (71.0 per cent). (37.6 per cent). On average, the proportion with uni- However, in terms of distribution by sex, migrant versity education in the sample of HICs is higher for women hold even fewer high-skilled positions than both migrant women (30.8 per cent) and women migrant men in HICs. While the average proportion nationals (31.0 per cent) than the corresponding 20 The migrant pay gap: Understanding wage differences between migrants and nationals

X Table 4. Labour market participation, unemployment and occupational attainment of migrants and nationals by sex, latest years

Labour market participation by sex

Country Nationals Migrants Country Nationals Migrants

Total Men Women Total Men Women Total Men Women Total Men Women (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%)

High-income countries (HICs) Norway 73.0 76.4 69.7 81.1 85.7 76.3

Austria 66.3 72.4 60.3 67.9 80.2 56.9 Switzerland 79.0 82.8 75.1 81.1 84.8 77.7

Belgium 63.2 67.1 59.4 60.5 66.7 53.9 Australia 74.9 79.3 70.5 72.0 82.0 62.9

Croatia 60.7 66.1 55.4 65.2 70.9 61.0 United States 69.7 74.5 65.2 69.9 84.2 54.9

Cyprus 67.4 73.3 61.6 80.4 83.3 78.5 Canada 73.8 76.5 71.1 74.0 80.4 68.1

Czech Argentina 66.9 78.2 56.6 70.1 83.8 58.2 66.1 73.6 58.8 69.5 81.2 58.1 Republic Chile 64.8 77.1 53.7 83.6 93.1 75.0 Denmark 70.1 72.7 67.4 73.2 80.8 66.8 EU average 67.0 72.5 61.6 73.9 81.5 67.1 Estonia 72.4 75.3 69.7 70.8 78.8 61.7 High- 69.0 74.1 64.0 72.1 83.1 61.3 income Finland 71.6 72.9 70.3 67.7 77.6 57.7

France 65.7 68.9 62.6 64.2 70.6 58.8 Low- and middle-income countries (LMICs)

Greece 60.0 67.9 52.4 72.2 84.5 61.3 Bulgaria 66.7 71.3 62.1 50.6 46.4 53.9

Romania 62.7 72.7 52.9 95.1 89.1 100.0 Hungary 62.1 69.2 55.3 77.0 93.1 63.1

Turkey 57.1 77.3 36.8 50.2 74.0 29.3 Ireland 62.5 69.8 55.4 67.2 73.3 61.8

Albania 58.6 68.8 49.3 50.3 67.6 34.4 Italy 60.7 70.4 51.1 75.1 85.7 66.3

Jordan 40.0 65.0 15.0 52.9 75.6 16.8 Latvia 73.2 76.8 69.9 67.4 75.3 59.9

Bangladesh 60.8 84.6 37.9 54.0 79.7 32.9 Lithuania 69.6 73.6 65.9 70.7 78.8 63.6

Nepal 40.8 57.3 27.9 80.1 94.9 33.9 Luxembourg 60.1 65.6 54.4 72.0 79.2 64.7

Gambia 38.0 50.8 26.7 60.8 81.0 37.2 Malta 60.5 73.8 46.8 57.8 71.5 46.5

Madagascar 93.9 95.0 92.7 77.0 74.2 82.7 Netherlands 71.5 76.4 66.6 77.4 93.6 67.8

Malawi 80.4 85.9 75.6 75.5 89.9 65.6 Poland 67.4 74.6 60.4 92.9 96.0 86.9

Namibia 64.5 69.6 60.0 70.1 79.2 58.5 Portugal 67.6 72.0 63.5 74.2 82.8 68.5

Sierra Leone 60.6 61.3 60.0 59.2 69.1 48.5 Slovakia 68.7 75.0 62.6 56.1 53.4 59.8

Tanzania** 86.5 90.5 82.8 82.7 91.6 74.1 Slovenia 63.2 67.7 58.5 80.6 89.9 66.1

Bolivia* 71.1 83.2 59.8 60.2 75.2 46.8 Spain 70.1 74.9 65.3 79.8 87.9 72.6

Costa Rica 65.7 79.8 51.2 71.7 89.2 55.6 Sweden 79.5 81.2 77.9 68.0 77.2 58.9

Mexico 69.6 86.1 54.8 62.9 77.3 48.5 United 69.7 74.0 65.4 72.1 81.1 64.1 Kingdom Low- and middle- 64.6 81.7 48.4 62.0 78.6 45.9 Iceland 89.3 92.7 86.0 95.2 96.3 94.3 income Chapter 2 – Labour market characteristics of migrant workers and nationals 21

Unemployment by sex

Country Nationals Migrants Country Nationals Migrants

Total Men Women Total Men Women Total Men Women Total Men Women (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%)

High-income countries (HICs) Norway 4.3 4.6 3.9 7.5 7.1 8.0

Austria 5.6 5.6 5.6 15.2 18.9 11.9 Switzerland 1.9 1.9 1.8 4.0 2.6 5.4

Belgium 6.8 6.6 7.0 13.4 14.3 12.5 Australia 3.8 4.1 3.5 3.9 4.2 3.7

Croatia 15.1 15.7 14.5 22.6 17.7 26.2 United States 3.2 3.4 3.0 3.0 2.9 3.1

Cyprus 14.6 15.1 14.1 18.2 18.1 18.2 Canada 10.2 9.7 10.7 9.6 9.0 10.2

Czech 7.0 6.2 7.8 7.0 9.0 5.1 Argentina 6.3 6.5 6.1 5.9 7.2 4.8 Republic Chile 4.7 5.2 4.2 5.1 5.0 5.2 Denmark 4.5 4.7 4.3 12.5 10.4 14.3 EU average 8.3 8.7 8.0 13.5 12.5 15.4 Estonia 7.1 8.7 5.8 12.8 14.6 10.7

High- 5.9 6.2 5.6 7.7 7.5 8.0 Finland 6.6 7.6 5.7 17.0 16.0 18.0 income

France 6.3 6.4 6.2 12.5 11.4 13.4 Low- and middle-income countries (LMICs)

Greece 15.4 14.7 16.1 21.3 21.2 21.3 Turkey 7.2 8.7 5.7 6.6 9.3 4.3

Hungary 6.5 7.6 5.5 5.2 8.5 2.2 Albania 9.4 12.4 6.6 12.3 17.9 7.2

Ireland 11.8 15.0 8.8 12.5 12.2 12.9 Jordan 7.2 9.7 4.7 5.7 8.6 0.9

Italy 8.9 9.2 8.5 13.4 14.3 12.7 Bangladesh 2.7 2.8 2.5 1.4 1.2 1.5

Latvia 9.3 10.7 8.0 8.1 9.4 6.8 Nepal 5.0 6.4 4.0 1.1 0.3 3.5

Lithuania 8.9 10.1 7.7 12.0 2.7 20.1 Gambia 4.3 5.1 3.6 5.4 4.8 6.1 Luxembourg 2.6 2.8 2.4 6.0 6.6 5.5 Madagascar 4.0 3.2 4.8 2.6 3.8 Malta 4.5 5.1 3.9 6.0 4.2 7.6 Malawi 6.8 6.0 7.5 3.4 1.8 4.5 Netherlands 7.2 7.0 7.4 15.4 7.4 20.2 Namibia 16.1 16.4 15.8 10.1 8.8 11.8 Poland 11.2 12.1 10.3 8.2 5.6 13.2 Sierra Leone 5.5 6.1 5.1 8.8 13.4 3.8 Portugal 14.0 14.8 13.3 16.4 17.1 16.0 Tanzania** 9.3 8.7 9.8 7.7 7.4 7.9 Slovenia 9.1 9.1 9.2 18.6 11.7 29.2 Bolivia* 2.4 2.4 2.4 0.3 0.6 Spain 17.8 17.6 18.0 27.0 28.9 25.3 Costa Rica 6.6 6.7 6.4 5.8 5.8 5.8 Sweden 4.4 4.8 4.0 9.9 11.5 8.4 Mexico 7.9 6.0 9.5 7.1 6.4 7.9 United 4.3 5.2 3.5 6.2 6.6 5.8 Kingdom Low- and middle- 6.2 5.9 6.4 6.8 5.4 8.6 Iceland 3.6 3.8 3.4 10.1 6.4 13.3 income

(Table 4 continued on page 22) 22 The migrant pay gap: Understanding wage differences between migrants and nationals

(Table 4 continued from page 21) Proportion with high-skilled occupations by sex

Country Nationals Migrants Country Nationals Migrants

Total Men Women Total Men Women Total Men Women Total Men Women (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%)

High-income countries (HICs) Norway 44.8 43.8 45.7 38.3 40.8 35.7

Austria 35.4 37.1 33.6 21.6 23.1 20.2 Switzerland 43.3 47.6 39.0 39.5 42.1 37.0

Belgium 36.3 37.4 35.2 21.9 25.7 17.8 Australia 35.0 42.3 27.8 34.6 44.1 25.9

Croatia 22.2 22.3 22.1 20.3 15.8 23.6 United States 32.8 33.0 32.6 19.6 23.5 15.5

Cyprus 27.5 28.1 26.9 17.0 19.1 15.6 Canada 36.9 29.5 44.4 37.1 36.2 37.8

Czech Argentina 16.1 16.8 15.5 10.9 13.5 8.7 29.7 28.9 30.5 31.0 30.4 31.6 Republic Chile 17.1 17.6 16.6 17.3 20.9 14.1 Denmark 37.6 34.1 41.1 26.3 26.9 25.8 EU average 31.6 32.2 31.1 25.1 27.0 22.8 Estonia 38.4 33.4 43.0 19.6 19.0 20.4 High- 32.5 32.9 32.2 22.7 25.3 19.8 income Finland 37.2 35.9 38.4 27.6 29.4 25.7

France 35.8 38.3 33.3 22.9 22.7 23.1 Low- and middle-income countries (LMICs)

Greece 21.7 22.9 20.6 6.2 5.6 6.7 Bulgaria 23.4 20.8 26.0 27.9 39.3 19.0

Romania 14.9 15.3 14.5 56.3 70.8 44.5 Hungary 26.4 22.4 30.1 32.5 35.5 29.8

Turkey 10.4 13.9 6.9 10.6 12.5 8.8 Ireland 28.3 29.3 27.3 21.4 20.5 22.2

Albania 8.5 8.7 8.4 8.1 10.0 6.3 Italy 26.7 29.0 24.4 8.5 8.5 8.5

Jordan 10.4 12.9 7.8 1.9 2.5 0.9 Latvia 33.4 26.4 39.6 23.0 22.2 23.7

Bangladesh 4.7 7.0 2.4 6.9 13.2 1.8 Lithuania 32.9 26.3 38.9 41.3 44.4 38.7

Nepal 5.1 7.4 3.2 7.7 9.0 3.5 Luxembourg 40.0 41.5 38.4 30.8 34.1 27.5

Gambia 6.4 10.6 2.7 7.4 11.5 2.6 Malta 35.7 36.0 35.3 57.0 56.7 57.3

Madagascar 3.3 4.0 2.7 16.1 19.1 9.8 Netherlands 40.3 43.1 37.4 25.8 31.2 22.7

Malawi 2.8 4.0 1.7 5.4 11.6 1.1 Poland 24.2 19.5 28.8 40.4 48.2 25.3 Namibia 8.8 8.9 8.8 16.8 16.4 17.3 Portugal 25.5 26.3 24.8 13.8 16.6 12.0 Sierra Leone 2.9 4.3 1.7 3.5 5.4 1.5 Slovakia 28.3 25.4 31.2 13.8 12.4 15.8 Tanzania** 2.7 3.3 2.2 9.2 14.4 4.0 Slovenia 71.0 72.9 69.1 55.0 68.7 33.8 Bolivia* 11.9 13.4 10.5 18.1 21.9 14.7 Spain 23.7 25.0 22.4 10.8 11.5 10.2 Costa Rica 12.5 13.8 11.2 5.3 6.5 4.2 Sweden 43.3 41.7 44.9 30.4 29.4 31.4 Mexico 9.6 11.0 8.3 18.2 22.3 14.0 United 40.5 42.3 38.7 31.5 30.2 32.6 Kingdom Low- and middle- 8.8 10.7 7.1 15.8 20.8 11.3 Iceland 37.0 33.4 40.6 15.3 12.4 17.9 income

Note: EU, high-income, and low- and middle-income estimates are the averages of the European Union, the sample of high-income countries, and low- and middle-income countries, respectively. Averages are weighted by the number of wage employees in each country. Occupational attainment comprises the proportion of individuals with job titles such as: chief executive officer (CEO), manager, professional, associate professional, and technical officer. * the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). Chapter 2 – Labour market characteristics of migrant workers and nationals 23

X Figure 3: Labor force participation rates of migrants and nationals by sex, latest years

Migrant men vs. non-migrant men Migrant women vs. non-migrant women

Slovakia Malta Belgium Belgium France United States Croatia Austria Malta Finland Ireland Czech Republic Latvia Argentina Sweden France Finland Sweden Estonia Slovakia Lithuania Latvia Luxembourg Croatia Austria Greece Canada Estonia Denmark Ireland United Kingdom Australia Czech Republic Hungary Australia Lithuania Portugal United Kingdom High-income Cyprus High-income Luxembourg Argentina Slovenia United States Italy Greece Denmark Switzerland Netherlands Norway Canada Italy Portugal Spain Spain Slovenia Chile Hungary Norway Chile Switzerland Netherlands Cyprus Poland Poland Iceland Iceland High-income High-income Bulgaria Jordan Albania Turkey Sierra Leone Bangladesh Turkey Nepal Madagascar Albania Bolivia* Gambia Jordan Bolivia* Mexico Mexico Namibia Sierra Leone Bangladesh Bulgaria Gambia Costa Rica Romania Namibia w- and middle income Costa Rica w- and middle income Malawi Lo Lo Malawi Tanzania** Tanzania** Madagascar Nepal Romania Low- and middle-income Low- and middle-income 020406080 100 020406080100 Participating rate (%) Participating rate (%)

Migrant men Men nationals Migrant women Women nationals

Note: * the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). 24 The migrant pay gap: Understanding wage differences between migrants and nationals

proportions for migrant men (28.6 per cent) and highest in Mexico (35.0 per cent). In terms of sex, on men nationals (27.2 per cent). average, the proportion with university education in the sample of LMICs is higher for both migrant In the sample of LMICs, the educational gap in women (26.3 per cent) and women nationals favour of both men and women migrants is much (16.0 per cent) than the corresponding proportions wider than in the sample of HICs. While the pro- for migrant men (26.2 per cent) and men nationals portion of total migrants with university education (14.6 per cent). is estimated at 33.0 per cent in LMICs, on average, the corresponding proportion for nationals is esti- The findings corroborate earlier ILO research com- mated at 19.8 per cent. The proportion with uni- paring the education of men and women workers versity education is lowest in Nepal (3.3 per cent) (see, e.g., ILO, 2018a). While education is an import- and highest in Costa Rica (59.0 per cent) among ant determinant of labour market outcomes, it does migrants in the sample of LMICs, whereas that of not seem to be the sole driver of the observed nationals is lowest in Sierra Leone (0.9 per cent) and migrant pay gap as analysed in Chapter 3.

X Table 5. Education of migrants and nationals by sex, latest years

Nationals

Country Total Men Women

<=Primary Lower Upper University <=Primary Lower Upper University <=Primary Lower Upper University (%) Sec. Sec./Voc. (%) (%) Sec. Sec./Voc. (%) (%) Sec. Sec./Voc. (%) (%) (%) (%) (%) (%) (%)

High-income countries (HICs)

Austria 0.0 16.5 54.7 28.8 0.0 13.1 55.7 31.2 0.0 20.0 53.7 26.3

Belgium 1.7 23.2 38.1 36.9 1.7 23.8 40.5 34.1 1.8 22.6 35.8 39.8

Croatia 0.3 22.7 61.8 15.1 0.1 18.2 68.6 13.1 0.5 27.2 55.2 17.1

Cyprus 2.1 25.4 41.4 31.1 1.6 26.6 44.6 27.2 2.6 24.1 38.3 35.1

Czech Republic 0.2 11.8 71.1 17.0 0.2 9.5 73.9 16.4 0.2 14.0 68.3 17.5

Denmark 0.0 23.1 44.3 32.6 0.0 24.5 46.9 28.6 0.0 21.7 41.7 36.6

Estonia 0.1 17.9 47.6 34.4 0.1 22.8 53.2 23.8 0.1 13.5 42.6 43.8

Finland 0.0 21.1 44.3 34.6 0.0 23.7 47.2 29.1 0.0 18.5 41.5 40.0

France 1.0 21.4 48.2 29.4 1.0 20.1 51.6 27.2 1.0 22.6 44.8 31.5

Greece 1.4 29.2 42.8 26.6 1.2 27.8 44.5 26.5 1.7 30.5 41.2 26.6

Hungary 0.1 18.8 59.1 21.9 0.1 16.9 63.9 19.0 0.2 20.6 54.7 24.6

Ireland 0.0 31.0 31.3 37.6 0.0 35.4 27.3 37.3 0.0 26.7 35.3 38.0

Italy 0.8 41.1 41.3 16.8 0.7 41.6 41.7 16.0 1.0 40.5 40.8 17.7

Latvia 0.3 15.9 55.7 28.0 0.4 19.6 60.6 19.4 0.2 12.7 51.3 35.7

Lithuania 0.4 13.1 57.6 28.9 0.4 14.8 60.6 24.2 0.4 11.6 54.8 33.2

Luxembourg 0.9 30.3 45.0 23.7 0.9 26.4 49.0 23.7 0.9 34.4 41.0 23.7

Malta 0.1 58.0 24.9 17.0 0.1 57.9 25.3 16.7 0.1 58.1 24.5 17.3

Netherlands 0.6 23.4 42.7 33.3 0.6 22.3 42.8 34.2 0.6 24.5 42.5 32.4

Poland 0.3 13.6 63.3 22.8 0.3 13.4 66.6 19.6 0.2 13.9 60.0 25.9 Chapter 2 – Labour market characteristics of migrant workers and nationals 25

Country Total Men Women

<=Primary Lower Upper University <=Primary Lower Upper University <=Primary Lower Upper University (%) Sec. Sec./Voc. (%) (%) Sec. Sec./Voc. (%) (%) Sec. Sec./Voc. (%) (%) (%) (%) (%) (%) (%)

Portugal 3.8 59.2 20.7 16.3 2.8 63.4 21.0 12.8 4.7 55.2 20.4 19.7

Slovakia 0.1 11.6 67.4 20.9 0.1 10.5 71.2 18.1 0.1 12.7 63.6 23.5

Slovenia 0.0 17.4 56.0 26.6 0.0 15.8 62.0 22.1 0.0 18.9 49.9 31.2

Spain 4.7 41.1 22.8 31.4 4.1 43.2 23.0 29.6 5.3 39.0 22.5 33.2

Sweden 0.0 16.5 48.3 35.2 0.0 16.8 54.1 29.1 0.0 16.3 42.3 41.4

United King- 2.6 31.9 30.4 35.1 2.5 33.0 30.2 34.4 2.7 30.9 30.6 35.8 dom

Iceland 0.0 33.0 37.7 29.3 0.0 33.7 42.8 23.6 0.0 32.3 32.6 35.1

Norway 0.7 24.6 40.8 33.8 0.6 26.1 43.6 29.8 0.8 23.2 38.2 37.8

Switzerland 11.3 51.8 19.8 17.1 10.6 46.3 21.6 21.5 12.1 57.3 17.9 12.7

Australia 0.3 22.7 53.4 23.6 0.3 21.7 57.1 20.9 0.3 23.6 49.7 26.3

United States 0.9 2.5 64.5 32.1 0.9 2.7 66.2 30.2 0.9 2.3 63.0 33.9

Canada 2.4 11.4 63.6 22.6 2.7 12.7 65.1 19.4 2.0 10.1 62.2 25.8

Argentina 4.1 37.1 42.0 16.8 2.7 12.7 65.1 19.4 2.0 10.1 62.2 25.8

Chile 1.3 19.4 66.6 12.7 2.7 12.7 65.1 19.4 2.0 10.1 62.2 25.8

EU average 1.6 27.6 43.6 27.4 1.5 27.6 45.5 25.5 1.7 27.6 41.7 29.1

High-income 1.5 16.2 53.6 28.8 1.4 15.7 55.8 27.2 1.5 15.3 52.3 31.0

Low- and middle-income countries (LMICs)

Bulgaria 1.7 23.7 52.8 21.8 1.4 24.0 57.0 17.6 2.1 23.4 48.6 25.9

Romania 0.8 27.1 57.7 14.5 0.8 23.4 61.5 14.3 0.7 30.7 54.0 14.6

Turkey 44.2 30.6 9.1 16.0 36.1 35.1 11.2 17.5 52.3 26.1 7.0 14.5

Albania 5.6 47.5 34.1 12.8 4.9 44.1 38.7 12.3 6.3 50.5 29.9 13.3

Jordan 31.3 26.9 25.9 15.9 29.9 29.9 24.0 16.1 32.7 23.9 27.8 15.6

Bangladesh 53.5 16.2 26.2 4.1 51.6 14.6 28.1 5.8 55.3 17.8 24.4 2.4

Nepal 50.1 13.1 31.2 5.7 39.7 15.5 36.6 8.2 58.2 11.1 26.9 3.8

Gambia 53.4 16.9 26.9 2.5 47.2 17.9 31.3 3.4 58.8 16.1 23.0 1.7

Malawi 88.2 0.0 9.3 2.5 83.8 0.0 12.4 3.8 92.1 0.0 6.6 1.4

Namibia 39.6 24.1 22.1 5.5 40.4 24.0 22.5 5.1 39.0 24.2 21.7 5.9

Sierra Leone 8.6 14.9 54.2 0.9 9.2 18.5 52.7 1.5 8.1 11.7 55.5 0.5

Bolivia* 25.8 27.0 34.8 12.4 2.7 12.7 65.1 19.4 2.0 10.1 62.2 25.8

Costa Rica 9.9 53.6 19.7 16.8 2.7 12.7 65.1 19.4 2.0 10.1 62.2 25.8

Mexico 10.7 17.5 36.9 35.0 2.7 12.7 65.1 19.4 2.0 10.1 62.2 25.8

Low- and middle- 31.2 21.0 29.1 19.8 25.2 18.6 42.3 14.6 29.8 16.8 38.3 16.0 income

(Table 5 continued on page 26) 26 The migrant pay gap: Understanding wage differences between migrants and nationals

(Table 5 continued from page 25)

Migrants

Country Total Men Women

<=Primary Lower Upper University <=Primary Lower Upper University <=Primary Lower Upper University (%) Sec. Sec./Voc. (%) (%) Sec. Sec./Voc. (%) (%) Sec. Sec./Voc. (%) (%) (%) (%) (%) (%) (%)

High-income countries (HICs)

Austria 0.0 34.3 38.9 26.8 0.0 28.7 43.2 28.2 0.0 39.3 35.1 25.6

Belgium 6.9 30.3 31.4 31.5 6.4 31.4 32.4 29.8 7.4 29.0 30.2 33.4

Croatia 0.6 5.9 71.5 21.9 1.5 6.5 76.5 15.5 0.0 5.5 67.9 26.6

Cyprus 0.4 22.3 46.0 31.3 0.0 25.0 46.7 28.3 0.7 20.5 45.6 33.3

Czech Republic 0.0 14.9 50.1 35.0 0.0 12.5 52.0 35.6 0.0 17.3 48.2 34.5

Denmark 0.0 13.3 49.9 36.8 0.0 6.6 54.1 39.2 0.0 18.9 46.4 34.7

Estonia 0.2 13.2 58.8 27.9 0.0 14.4 61.1 24.5 0.3 11.7 56.2 31.8

Finland 0.0 3.0 58.1 38.8 0.0 3.6 62.1 34.3 0.0 2.4 54.1 43.5

France 8.3 40.0 28.4 23.3 7.6 42.2 32.1 18.1 8.8 38.2 25.3 27.6

Greece 2.5 36.7 43.6 17.1 3.2 39.1 44.8 12.9 2.0 34.7 42.5 20.9

Hungary 0.0 37.3 29.2 33.5 0.0 27.1 43.8 29.1 0.0 46.1 16.6 37.3

Ireland 0.0 13.8 40.0 46.3 0.0 14.5 40.5 45.0 0.0 13.2 39.5 47.4

Italy 2.6 36.1 46.0 15.3 2.3 42.3 42.8 12.6 2.8 31.0 48.7 17.5

Latvia 0.2 10.0 71.8 18.0 0.2 10.7 71.9 17.2 0.2 9.3 71.8 18.7

Lithuania 0.0 17.9 40.6 41.5 0.0 20.9 41.5 37.7 0.0 15.3 39.8 44.9

Luxembourg 1.9 42.5 27.3 28.4 1.6 43.4 27.0 28.0 2.1 41.5 27.6 28.8

Malta 0.2 38.1 29.8 31.9 0.0 39.6 28.5 31.9 0.4 36.9 30.8 31.9

Netherlands 5.1 20.6 47.9 26.4 6.5 15.9 51.4 26.2 4.2 23.3 45.9 26.6

Poland 0.0 2.0 55.1 42.8 0.0 0.0 54.0 46.0 0.0 6.0 57.3 36.7

Portugal 3.1 50.1 33.5 13.3 2.9 53.4 28.5 15.2 3.2 48.0 36.8 12.1

Slovakia 0.0 23.0 62.5 14.5 0.0 22.0 62.2 15.8 0.0 24.4 62.9 12.7

Slovenia 0.0 32.1 59.8 8.1 0.0 25.5 68.8 5.7 0.0 42.4 45.7 11.8

Spain 7.2 35.5 33.6 23.7 7.1 36.6 35.3 20.9 7.2 34.6 32.0 26.3

Sweden 0.0 27.8 28.4 43.7 0.0 28.7 31.7 39.5 0.0 26.9 25.2 47.9

United 15.0 18.2 21.6 45.3 14.3 21.2 22.9 41.6 15.6 15.5 20.5 48.5 Kingdom

Iceland 0.0 29.9 49.7 20.3 0.0 30.0 57.0 13.0 0.0 29.8 43.4 26.8

Norway 28.9 18.8 18.6 33.7 27.5 21.3 20.1 31.1 30.4 16.1 17.1 36.4

Switzerland 15.4 43.3 25.0 16.3 13.0 44.7 24.7 17.5 17.5 42.0 25.4 15.2

Australia 1.5 12.7 45.6 39.7 0.6 11.2 48.0 40.2 2.2 14.1 43.4 39.3

United States 12.7 10.2 50.0 27.1 13.2 10.0 50.2 26.5 12.2 10.4 49.7 27.6

Canada 3.7 6.7 49.8 39.8 3.3 7.5 49.1 40.1 4.2 5.9 50.4 39.6

Argentina 8.5 40.5 36.7 14.3 3.3 7.5 49.1 40.1 4.2 5.9 50.4 39.6 Chapter 2 – Labour market characteristics of migrant workers and nationals 27

Country Total Men Women

<=Primary Lower Upper University <=Primary Lower Upper University <=Primary Lower Upper University (%) Sec. Sec./Voc. (%) (%) Sec. Sec./Voc. (%) (%) Sec. Sec./Voc. (%) (%) (%) (%) (%) (%) (%)

Chile 0.8 8.2 65.8 25.2 3.3 7.5 49.1 40.1 4.2 5.9 50.4 39.6

EU average 7.9 26.7 37.4 30.5 7.8 30.3 40.9 27.6 8.3 27.1 36.1 32.8

High-income 10.0 18.4 43.8 29.0 10.0 18.8 44.6 28.6 9.9 17.3 43.1 30.8

Low- and middle-income countries (LMICs)

Bulgaria 0.0 4.6 50.4 45.0 0.0 5.2 55.5 39.3 0.0 4.1 46.6 49.3

Romania 0.0 30.5 36.8 32.7 0.0 0.0 81.7 18.3 0.0 55.5 0.0 44.5

Turkey 32.8 36.3 12.2 18.7 32.2 37.0 13.5 17.3 33.4 35.7 11.1 19.9

Albania 29.6 42.2 16.9 11.3 23.9 44.3 19.9 11.9 34.9 40.2 14.2 10.7

Jordan 57.3 13.6 22.1 7.0 52.1 14.4 26.9 6.6 65.6 12.3 14.5 7.7

Bangladesh 46.0 14.9 29.4 9.7 35.5 16.6 33.1 14.8 54.7 13.5 26.4 5.4

Nepal 56.8 19.1 20.8 3.3 60.8 19.4 17.6 2.2 44.3 18.2 30.8 6.7

Gambia 69.9 9.8 13.9 6.0 67.5 10.7 13.8 7.8 72.7 8.7 14.1 3.9

Malawi 79.4 0.0 12.9 7.7 73.1 0.0 14.9 12.0 83.7 0.0 11.6 4.7

Namibia 26.8 18.6 23.3 19.2 27.6 19.9 21.6 20.3 25.8 16.8 25.5 17.7

Bolivia 14.4 8.9 45.3 31.4 3.3 7.5 49.1 40.1 4.2 5.9 50.4 39.6

Costa Rica 24.1 48.9 16.2 10.8 3.3 7.5 49.1 40.1 4.2 5.9 50.4 39.6

Mexico 4.5 7.6 29.0 59.0 3.3 7.5 49.1 40.1 4.2 5.9 50.4 39.6

Low- and middle- 27.1 17.5 26.3 33.0 23.1 16.4 38.7 26.2 28.5 17.4 33.7 26.3 income

Note: EU, high-income, and low- and middle-income estimates are the averages of the European Union, the sample of high-income countries, and low- and middle-income countries, respectively. Averages are weighted by the number of wage employees in each country. “<=Primary“ is less than or equal to primary education; ”Lower Sec.“ is lower secondary education without high school diploma; ”Upper Sec./Voc.“ is secondary education (completed), including vocational training; and ”University“ is university education. Source: ILO estimates based on survey data provided by national sources (see Appendix II).

2.5. A focus on wage workers 2.5.1. Share of wage workers among total labour market participants From this part onwards, the sample used in the estimation is restricted to individuals of working age (i.e. adults with ages 16–70) who are wage The share of wage workers is lower workers only. In principle, however, the repre- among migrants than non-migrants in sented sample should include all statuses of HICs but higher among migrants than employment, except contributing family workers. non-migrants in LMICs Due to difficulties in collecting earnings data for self-employed categories (employers and own-ac- Table 6 reports the share of wage workers among count workers), the data is often limited to wage total labour market participants – including wage workers only. Nonetheless, from a policy stand- workers, employers, own-account workers, mem- point, wage workers are more likely to be subjected bers of cooperatives, unpaid (family) workers, and to minimum wage legislation, hence the one group the unemployed (who actively seek for employ- of interest for comparing wages between migrant ment) – comparing men and women migrant and non-migrant workers. workers to nationals. 28 The migrant pay gap: Understanding wage differences between migrants and nationals

Although migrant workers have higher labour force Of course, large variations exist across countries, participation in the EU and the sample of HICs on with for example, the share of wage workers among average (see table 4), in table 6, fewer migrants the total migrant labour market participants in the among the total migrant participants in the labour sample of HICs being lowest in Czech Republic market in the EU and the sample of HICs are wage (55.5 per cent) and highest in the United States workers relative to the corresponding share of wage (88.5 per cent), whereas the corresponding share workers among non-migrant labour force partici- for nationals is lowest in Greece (49.1 per cent) and pants in these countries (70.6 per cent and 76.5 per highest in Iceland (93.0 per cent). In terms of dis- cent, respectively in the EU, and 79.5 per cent and tribution by sex, on average, fewer migrant men 82.6 per cent, respectively in the sample of HICs). among the total migrant men participants in the

X Table 6. Share of wage workers among total labour market participants, latest years

Country Nationals Migrants

Total Men Women Total Men Women (%) (%) (%) (%) (%) (%)

High-income countries (HICs)

Austria 83.2 82.1 84.6 71.4 70.1 72.9

Belgium 80.3 78.3 82.5 70.3 68.1 73.1

Croatia 67.7 67.0 68.5 60.0 63.5 57.1

Cyprus 69.9 69.2 70.6 71.5 70.7 72.2

Czech Republic 74.5 73.0 76.4 55.5 55.6 55.3

Denmark 88.9 87.0 90.9 77.3 77.7 76.8

Estonia 87.6 85.3 89.9 79.5 80.1 78.6

Finland 84.5 81.9 87.2 67.4 70.5 63.0

France 84.6 83.5 85.8 75.8 78.8 72.7

Greece 49.1 49.1 49.2 57.3 58.2 56.3

Hungary 81.4 80.0 83.0 77.8 87.5 65.5

Ireland 68.5 60.2 78.7 73.6 73.2 73.9

Italy 67.6 65.1 71.0 70.7 68.8 72.8

Latvia 83.3 81.2 85.3 83.6 84.4 82.8

Lithuania 80.8 79.7 82.0 71.8 89.5 52.7

Luxembourg 90.5 90.0 91.2 88.3 89.1 87.3

Malta 84.0 81.7 87.7 73.8 69.8 79.0

Netherlands 81.1 81.1 81.2 66.6 70.4 63.6

Poland 69.6 68.9 70.5 70.9 70.0 72.9

Portugal 70.3 68.8 71.9 65.6 65.7 65.4

Slovakia 72.8 69.2 77.0 79.6 100.0 54.7

Slovenia 76.6 75.0 78.5 73.7 83.1 54.0

Spain 65.4 65.5 65.4 58.0 56.5 59.6

Sweden 91.6 90.3 93.0 82.1 80.8 83.8 Chapter 2 – Labour market characteristics of migrant workers and nationals 29

Country Nationals Migrants

Total Men Women Total Men Women (%) (%) (%) (%) (%) (%)

United Kingdom 82.9 78.4 87.9 76.3 73.4 79.6

Iceland 93.0 91.9 94.2 87.4 91.6 83.6

Norway 90.7 89.1 92.5 86.3 84.3 88.7

Switzerland 86.0 84.1 88.1 86.2 89.6 82.8

Australia 86.8 85.6 88.2 86.0 84.9 87.2

United States 89.9 89.0 90.8 88.5 88.5 88.4

Canada 74.5 73.6 75.5 71.7 69.7 73.8

Argentina 68.2 67.2 69.5 65.2 62.8 68.2

Chile 70.6 70.2 71.2 79.5 80.4 78.5

EU average 76.5 74.7 78.7 70.6 71.8 69.6

High-income 82.6 81.3 84.2 79.5 79.5 79.5

Low- and middle-income countries (LMICs)

Bulgaria 76.9 74.4 79.8 65.4 61.1 68.2

Romania 69.8 69.4 70.4 60.4 82.2 44.5

Turkey 59.7 62.5 53.7 75.4 76.1 73.8

Albania 33.6 35.1 31.7 31.3 32.0 30.1

Jordan 71.2 72.4 66.2 83.8 82.5 93.4

Bangladesh 37.1 41.1 28.6 35.7 41.7 23.8

Nepal 48.0 55.6 35.6 48.3 48.0 50.7

Gambia 44.4 50.3 34.4 28.5 29.3 26.6

Madagascar 11.0 14.0 7.9 20.5 26.9 8.6

Malawi 33.2 39.2 27.3 36.3 51.2 22.2

Namibia 52.6 57.6 47.5 60.3 61.3 58.7

Sierra Leone 10.3 16.3 5.0 12.5 17.5 4.7

Tanzania** 13.3 17.8 8.8 30.0 37.0 21.5

Bolivia* 37.3 39.4 34.4 44.0 46.9 39.9

Costa Rica 67.2 67.3 67.2 74.9 76.7 72.1

Mexico 65.3 67.2 62.6 67.1 69.0 64.0

Low- and middle- 54.6 57.2 49.9 57.8 61.9 52.1 income

Note: EU, high-income, and low- and middle-income estimates are the averages of the European Union, the sample of high-income countries, and low- and middle-income countries, respectively. Averages are weighted by the number of wage employees in each country. Labour force participants comprise all individuals who are active in the labour market. These include wage workers, employers, own-account workers, members of cooperatives, unpaid (family) workers, and the unemployed (who actively seek for employment). Source: ILO estimates based on survey data provided by national sources (see Appendix II). 30 The migrant pay gap: Understanding wage differences between migrants and nationals

sample of HICs tend to be wage workers relative to informal employment exceeds the share of men in the corresponding share of wage workers among informal employment. This is particularly the case the population of non-migrant men participants in for countries in sub-Saharan Africa, Southern Asia these countries (79.5 per cent and 81.3 per cent, and Latin America. In LIMCs, an estimated 50 per respectively). Similarly, fewer migrant women cent of all wage earners continue to work in the among the total population of migrant women informal economy, either in the informal sector labour force participants in the sample of HICs or as informal workers in the formal sector (ILO, are wage workers relative to the corresponding 2018d). Migrants, especially women, are more likely share of wage workers among the population of to work in the informal economy under poor con- non-migrant women participants (79.5 per cent and ditions and with low pay (ILO, 2018c). 84.2 per cent, respectively). Unfortunately, many existing labour force sur- In the sample of LMICs, however, the share of wage veys that capture wages of migrant workers and workers among total labour force participants is nationals do not cover the informal economy, in higher among migrants than non-migrants. On particular in most HICs. Therefore, data on the average, 57.8 per cent of migrant participants informal economy is available only for 14 of the tend to be wage workers, which is higher than 49 studied countries in this report39. These 14 coun- the corresponding percentage of non-migrant tries host roughly only 5.3 per cent of international participants (54.6 per cent). In Jordan, which has migrants and about 3.0 per cent of migrant work- a relatively large migrant share of the total pop- ers worldwide. Therefore, estimates for the infor- ulation (of about 33.9 per cent), the proportion mal economy generated from these 14 countries of wage employees among the total migrant are not representative of migrants in the informal participants is 83.8 per cent, whereas the cor- economy worldwide but specific only to migrants responding proportion for Jordan nationals is in these countries. 71.2 per cent. On the other hand, in Sierra Leone, which has just about 1 per cent of migrants in the More active migrant workforce, total population, the proportion of wage workers in particular women migrants among the total migrant participants is 12.5 per in studied countries tend to be in cent, whereas the corresponding proportion informal employment compared for Sierra Leonean nationals is 10.3 per cent. to the non-migrant workforce. 2.5.2. Gender and informality Table 7 presents the proportion of informal economy workers among the total active workforce as among wage workers well as among wage workers in 14 of the studied According to the most recent global figures, countries where data on informality is available. around 2 billion people worldwide work in the Except for Argentina and Chile, all the countries informal economy38, or 6 out of 10 workers, in covered are LMICs. On average, more migrant 2020 (ILO, 2020g). Out of these, just over 740 mil- workers (both active workforce and wage workers) lion are women. Informality concerns close to in the studied countries are in informal employment 9 out of 10 workers in sub-Saharan Africa and compared to non-migrant workers. While about Southern Asia. However, while globally the share 63.2 per cent of non-migrant workforce in the of women in informal employment is lower than 14 studied countries work in the informal economy, the share of men in informal employment, there about 66.5 per cent of migrant workers work in the are more countries where the share of women in informal economy. The gap among wage workers

38 According to the ILO, the informal economy refers to the set of all economic activities by workers and economic units that are – in law and practice – not covered or insufficiently covered by formal arrangements. See http://www.ilo.org/global/topics/employment-promotion/informal-economy/. 39 Although the existence of an informal economy is not unique to low- and middle-income economies, informality in the developed world is often negligible or statistically undetected in many existing labour force surveys. For example, in the case of European economies, the proportion of all employees operating without any type of contractual arrangement varied between 2.7 per cent in the Nordic countries and 9.5 per cent in southern Europe, and which exceeds 10 per cent of the workforce (Hazans, 2011). These estimates contrast with those found in developing economies. Based on statistics compiled by ILO from 47 developing economies, in more than half of these countries the proportion of people in informal employment in non-agricultural activities exceeds 50 per cent, whereas for about one-third of these countries informal employment accounts for at least 67 per cent of non-agricultural employment (ILO, 2013 and 2018f). The clear difference in the magnitude of informality between developed and developing economies suggests the need to develop policy tools specific to low-income and emerging economies to reduce informality worldwide. Although the magnitude of informality amongst developed countries is low – or highly negligible in some cases – it is still important to expand efforts to extend existing labour force surveys to include reliable and comparable information on both informality and migrant workers in developed economies in order to better understand the incidence of informality among migrant workers in these economies. Chapter 2 – Labour market characteristics of migrant workers and nationals 31

X Table 7. Proportion of informal workers by sex, latest years

Among the active workforce

Country Nationals Migrants

Total Men Women Total Men Women (%) (%) (%) (%) (%) (%)

High-income countries (HICs)

Argentina 67.9 61.7 73.6 75.3 69.7 80.2

Chile 28.7 26.8 31.3 24.9 19.4 31.2

Low- and middle-income countries (LMICs)

Turkey 34.2 28.5 46.8 46.0 52.3 31.8

Albania 60.7 58.4 63.4 63.8 65.2 61.5

Bangladesh 94.8 94.0 96.7 91.9 90.9 94.0

Nepal 80.7 77.3 86.4 85.4 84.7 92.3

Gambia 76.5 73.8 81.1 90.0 90.1 89.7

Madagascar 91.0 88.7 93.2 71.4 62.2 88.2

Malawi 85.1 81.1 89.0 78.6 67.3 90.1

Namibia 66.6 65.8 67.5 65.1 65.1 65.1

Tanzania 90.8 88.6 93.0 78.1 72.0 85.5

Bolivia 88.9 86.5 91.2 88.4 78.1 97.6

Costa Rica 62.2 51.2 73.6 65.2 51.2 78.0

Mexico 55.8 55.5 56.3 61.2 61.0 61.6

Weighted average 63.2 60.9 67.1 66.5 65.7 66.4

Among wage workers

Country Nationals Migrants

Total Men Women Total Men Women (%) (%) (%) (%) (%) (%)

High-income countries (HICs)

Argentina 31.5 29.5 34.0 48.3 45.9 51.0

Chile 15.5 12.3 19.4 17.2 11.5 23.8

Low- and middle-income countries (LMICs)

Turkey 17.2 16.3 19.4 43.1 51.3 24.4

Albania 18.1 23.1 10.9 31.3 38.3 17.7

Bangladesh 89.5 89.4 89.9 88.3 90.8 79.5

Nepal 85.2 84.7 86.4 91.7 92.3 86.4

Gambia 53.7 54.2 52.2 72.7 75.0 66.2

Madagascar 40.7 39.4 43.2 36.2 29.9 72.9

Malawi 82.1 76.7 89.6 73.3 64.3 93.1

Namibia 62.4 64.1 60.3 66.4 68.5 62.5

(Table 7 continued on page 32) 32 The migrant pay gap: Understanding wage differences between migrants and nationals

(Table 7 continued from page 31)

Country Nationals Migrants

Total Men Women Total Men Women (%) (%) (%) (%) (%) (%)

Tanzania 64.5 66.0 61.4 56.7 53.3 63.7

Bolivia 67.2 68.4 65.4 73.7 65.7 87.1

Costa Rica 25.9 22.4 31.6 43.0 37.9 51.0

Mexico 47.6 47.5 47.9 63.8 65.3 61.2

Weighted average 50.8 50.2 51.9 62.4 63.9 57.8

Note: Active workforce comprise all individuals who are active in the labour market. These include wage workers, employers, own-account workers, unpaid (family) workers, and the unemployed. Averages are weighted by the number of wage employees in each country. Source: ILO estimates based on survey data provided by national sources (see Appendix II).

is even wider, with 50.8 per cent of non-migrant as extending to all its workers, including men and wage workers working in the informal economy women migrants, the right to a minimum wage whereas the corresponding proportion for migrant and social security – can greatly benefit migrant wage workers is 62.4 per cent. There are of course workers, guaranteeing them legal and effective variations across the countries. Informality among protection, and can help reduce labour market the active workforce is higher among migrant work- gaps between migrant workers and nationals. Such ers than among nationals in seven of the 14 coun- measures can also help to minimize the enormous tries including Albania, Argentina, Costa Rica, the impact of the COVID-19 on workers in the informal Gambia, Mexico, Nepal, and Turkey. It is higher economy, including men and women migrant work- among migrant wage workers compared to non-mi- ers who are among the hardest hit40. grant wage workers in even more countries (ten of the 14 countries). For example, while only 17.2 per cent of non-migrant wage workers in Turkey works 2.5.3. Education and occupations in the informal economy, about 43.1 per cent of of wage workers migrant wage workers are informal. The share of wage workers with higher In terms of distribution by sex, informal employ- education is lower among migrants than ment is higher among migrant women workers than non-migrants in HICs but higher among migrant men workers on average (66.4 per cent and migrants than non-migrants in LMICs 65.7 per cent, respectively). However, in terms of wage workers, fewer migrant women than migrant As mentioned earlier, migrant workers’ education men are informal wage workers (57.8 per cent and normally mirrors their skills set and depicts their 63.9 per cent, respectively). This is particularly so chances of obtaining well-paid jobs in their desti- partly because fewer migrant women than migrant nation countries. Also, migrant workers’ education men are wage workers on average (see table 6). and prior occupations should partly determine their earnings prospects and their economic success. Compared to their male counterparts, women in However, in reality, migrant workers are typically the informal economy are more often found in the likely to be affected by skills mismatch and may have most vulnerable situations, such as domestic work- difficulties transferring their skills and experience ers, home-based workers or contributing family across countries (Sparreboom and Tarvid, 2017). workers. The over-representation of women in the informal economy is likely to exacerbate their vul- Unlike in table 5 where the education of working nerability to exploitation and abuse, including low age migrant workers and that of nationals are or non-payment of wages. Measures that promote compared, figure 4 compares the education of the formalization of the informal economy – such wage workers only. Panel A considers the share of

40 See ILO (2020g) for analysis of the impact of the COVID-19 crisis on the informal economy. Chapter 2 – Labour market characteristics of migrant workers and nationals 33

© shutterstock.com Total migrant and non-migrant workers wage employees with secondary school education education is higher among migrants in Bangladesh, (including vocational training), while Panel B focuses Gambia, Malawi, Mexico, and Namibia, for example, on individuals with university education. it is higher among nationals in countries like Costa Rica, Jordan, Nepal, Sierra Leone, and Turkey. In both panels, the average proportion of wage workers with higher education is lower among In terms of the distribution of wage workers’ migrant workers than among nationals in the sam- education by sex, the estimates in this report are ple of HICs (76.1 per cent and 88.4 per cent, respec- consistent with previous ILO research comparing tively in terms of secondary school education, and the education of men and women wage workers 32.2 per cent and 35.7 per cent, respectively in (see, e.g., ILO, 2018a). That is, the education of terms of university education), with wide variations women, in terms of the share with higher educa- across countries. For example, while the share of tion, outweighs that of men in both the sample wage workers with university education is higher of HICs and LMICs, on average. This is true for among nationals in countries such as Belgium, both migrant women and non-migrant women in Cyprus, France, Norway, and the United States, it almost all the studied countries. Among migrant is higher among migrants in Australia, Canada, wage workers, for example, while the share of Luxembourg, Sweden, and the United Kingdom. migrant women with secondary school education is about 78.5 per cent and 67.3 per cent in the Among the sample of LMICs, on the other hand, samples of HICs and LMICs, respectively, the cor- the share of wage workers with higher education responding share of migrant men is 75.7 per cent is higher among migrant workers, on average than in HICs and 66.2 per cent in the sample of LMICs. among nationals across these countries (63.4 per Similarly, in terms of university education, the cent and 57.4 per cent, respectively in terms of share of migrant women is estimated at 35.5 per secondary school education, and 43.0 per cent and cent in the sample of HICs and 38.2 per cent in 28.1 per cent, respectively in terms of university the sample of LMICs, on average, whereas that education), with notable variations across countries. of migrant men is 31.1 per cent in the sample of While the share of wage workers with university HICs and 31.2 per cent in the sample of LMICs. 34 The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure 4: The education of migrant and non-migrant wage workers by sex, latest years

Total migrant and non-migrant workers

Panel A: Proportion with secondary education Panel B: Proportion with university education

High-income countries

Portugal Slovenia Switzerland Slovakia France Portugal Greece Italy Argentina Greece Norway Switzerland Luxembourg Latvia Spain Argentina Hungary Iceland Italy Spain United Kingdom Croatia Iceland France Slovenia Netherlands Austria Estonia Malta Chile Belgium Austria Lithuania Cyprus Cyprus United States United States Luxembourg Sweden Hungary Netherlands Denmark Slovakia Poland Denmark Finland Estonia Norway Australia Lithuania Chile Canada Ireland Belgium Latvia Malta Canada Australia Croatia Sweden Poland Czech Republic Czech Republic United Kingdom Finland Ireland EU EU High-income High-income 020 40 60 80 100 0 20 40 60 80 100 Percent Percent

Low- and middle-income countries

Malawi Jordan Costa Rica Nepal Nepal Sierra Leone Jordan Costa Rica Gambia Malawi Tanzania** Turkey Gambia Bangladesh Bangladesh Namibia Namibia Albania Turkey Madagascar Albania Sierra Leone Bulgaria Bolivia* Romania Mexico Bulgaria Bolivia* Romania Mexico Low- and middle-income Low- and middle-income 0 20 40 60 80 100 020406080 100 Percent Percent

Migrant workers Non-migrant workers Chapter 2 – Labour market characteristics of migrant workers and nationals 35

Non-migrant men and women

Panel A: Proportion with secondary education Panel B: Proportion with university education

High-income countries

Switzerland Switzerland Slovenia Portugal Czech Republic Slovakia Malta Italy Portugal Iceland Portugal Italy Spain Croatia Greece Italy Argentina Switzerland United Kingdom Chile Latvia Luxembourg Canada Argentina Australia Austria Iceland Norway Slovakia Spain Netherlands Australia Croatia Austria Luxembourg France Ireland Malta Netherlands France Hungary Estonia Cyprus Iceland Chile Denmark United States Austria Croatia France Cyprus Belgium Netherlands United States Sweden Poland Luxembourg Hungary Slovenia Hungary Finland United Kingdom Denmark Greece Denmark Poland Slovenia Norway Finland Argentina Sweden Norway Chile Latvia Lithuania Canada Lithuania Canada Estonia Greece Belgium Latvia Finland Malta Czech Republic Spain Australia Poland Cyprus Sweden Slovakia Estonia Czech Republic Lithuania Ireland United Kingdom United States Belgium Ireland EU EU EU High-income High-income High-income 020 40 60 80 100 020 40 60 80 100 0 20 40 60 80 100 Percent Percent

Low- and middle-income countries

Malawi Malawi Bangladesh Bangladesh Tanzania** Sierra Leone Namibia Gambia Madagascar Namibia Nepal Turkey Nepal Gambia Costa Rica Sierra Leone Bolivia* Albania Mexico Jordan Romania Bulgaria Bulgaria Costa Rica Turkey Bolivia* Mexico Albania Romania Jordan Low- and middle-income Low- and middle-income 020 40 60 80 100 020 40 60 80 100 Percent Percent

Non-migrant women Non-migrant men

(Figure 4 continued on page 36) 36 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure 4 continued from page 35)

Migrant men and women

Panel A: Proportion with secondary education Panel B: Proportion with university education

High-income countries Slovenia Switzerland Slovenia Slovakia Portugal Portugal Portugal Hungary Switzerland Italy France Latvia Greece Slovenia Italy Switzerland Luxembourg Greece Latvia Norway Spain Argentina Spain Iceland Iceland Greece United States Spain Iceland Cyprus Croatia Austria Denmark France United Kingdom Austria Netherlands Belgium Croatia Estonia Italy Netherlands Chile Cyprus Estonia Austria Lithuania Luxembourg Cyprus Sweden Poland United States Malta France Luxembourg United States Finland Hungary Netherlands Denmark Denmark Norway Argentina Poland Estonia Finland Croatia Chile Canada Norway Poland Lithuania Australia Hungary Lithuania Canada Latvia Belgium Argentina Belgium Australia Malta Chile Australia Czech Republic Canada Sweden Malta Ireland Czech Republic United Kingdom Finland United Kingdom Slovakia Sweden Ireland Czech Republic Ireland EU EU EU High-income High-income High-income 020 40 60 80 100 020 40 60 80 100 0 20 40 60 80 100 Percent Percent Low- and middle-income countries

Jordan Malawi Malawi Jordan Madagascar Gambia Tanzania** Bangladesh Bangladesh Nepal Nepal Gambia Namibia Namibia Sierra Leone Albania Bulgaria Turkey Turkey Bulgaria Costa Rica Costa Rica Bolivia* Bolivia* Mexico Mexico Sierra Leone Albania Romania Romania Low- and middle-income Low- and middle-income 020 40 60 80 100 020 40 60 80 100 Percent Percent

Migrant women Migrant men

Note: EU, high-income, and low- and middle-income estimates are the weighted averages of the European Union, the sample of high-income countries, and low- and middle-income countries, respectively. Averages are weighted by the number of wage employees in each country. Secondary school education comprises upper secondary and post-secondary non-tertiary including vocational training. University education includes short-cycle tertiary, undergraduate, master and doctoral levels of education. * the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). Chapter 2 – Labour market characteristics of migrant workers and nationals 37

Occupations of migrant and non-migrant wage workers

This section of the report looks at the occupational attainment of wage workers, comparing migrant workers with nationals and separating HICs from LMICs. Occupational attainment is categorized into five broad jobs:41 high-skilled, semi-skilled, low-skilled, and unskilled. In countries where disaggregation of occupations allows for the indepen- Rome, Italy, March 2019. Young migrants working as apprentice chef in an industrial restaurant kitchen in Rome. © shutterstock.com dent identification of the group “personal care and domestic workers”, estimates are shown separately for them as the fifth category. High skilled jobs In contrast, the share of migrant wage workers with include (where possible) senior officials, chief exec- high-skilled jobs in the sample of HICs is almost utive officers (CEOs) and other managerial positions, everywhere lower than the share of non-migrant professional jobs, technical and associate profes- wage workers with high-skilled jobs in these coun- sional jobs. Semi-skilled jobs consist of clerical sup- tries (29.5 per cent and 45.0 per cent, on average, port workers, service and sales workers, and skilled respectively), except in Australia, Canada, Czech agricultural, forestry and fishery workers. Low-skilled Republic, Hungary, and Malta. For example, in the jobs comprise craft and related trades workers, plant cases of the United Kingdom and United States, and machine operators, and assemblers. Unskilled while respectively 40.4 per cent and 29.8 per cent jobs include elementary occupations. of migrant wage workers have high-skilled jobs, 50.1 per cent and 48.8 per cent of non-migrant Migrant wage workers predominantly wage workers have high-skilled jobs in these two occupy lower skilled jobs in HICs respective countries. Similarly, the share of migrant wage workers with semi-skilled jobs in HICs is lower Figure 5 shows how migrant wage workers’ occu- than the corresponding share of non-migrant wage pational attainment compares to that of nationals workers (17.2 per cent and 21.1 per cent, on aver- showing all the categories described above. In the age, respectively), except in Australia, Canada, Chile, sample of HICs covered in the report, the share of Netherlands, Slovakia, Slovenia and Spain. men and women migrant wage workers with jobs in lower occupational categories (personal care and The story is different in the sample of LMICs, most domestic workers, unskilled or low-skilled work- of which are also countries of origin. With the excep- ers) is almost universally higher than the share tion of Costa Rica, the Gambia, Jordan, Nepal, and of non-migrant wage workers with jobs in these Turkey, the proportion of migrant workers with categories, with few exceptions. On average, while high-skilled jobs is higher than the proportion of 24.3 per cent and 18.7 per cent of migrant wage nationals with similar positions (33.2 per cent and workers in the sample of HICs have low-skilled and 20.5 per cent, on average, respectively). However, unskilled jobs, respectively, only 19.6 per cent and proportionately fewer migrant wage workers than 8.2 per cent of non-migrant wage workers have nationals tend to have semi-skilled jobs (30.9 per jobs in these respective categories. In countries cent and 36.6 per cent, respectively), low-skilled jobs where the data allows for the identification of (17.8 per cent and 22.5 per cent, respectively), or personal care and domestic workers (Argentina, work as personal care or domestic workers (4.3 per Australia, Chile, and the United States), migrant cent and 5.6 per cent, respectively), with few excep- wage workers tend to predominantly have occu- tions. In terms of unskilled jobs, the share of wage pations in this category compared to non-migrant workers is slightly higher on average for migrants wage workers (19.3 per cent and 9.6 per cent, on than for non-migrants (16.2 per cent and 15.3 per average, respectively). cent, respectively).

41 Datasets from all 49 countries used in this report include a classification of occupations in accordance with the International Standard Classification of Occupations (ISCO), based on either the 1988 classification (ISCO-88) or its 2008 update (ISCO-08). The original classification separates individuals into several minor and major groups. The data sets provided for analysis further disaggregate the two-digit classification (i.e. the ten major groups) into a smaller number of groups (five to eight). In general, the following can be distinguished: legislators, senior officials, CEOs and other managerial positions, professional jobs, technical jobs, semi-skilled occupations, and unskilled occupations. In some instances, the disaggregation allows for the independent identification of the group “domestic and/or personal care workers” (ILO, 2008). CEOs cannot be identified independently but given that the top and bottom percentiles of wage earners are excluded from the analysis, the possibility that the presence of CEOs in the sample may distort the estimates is minimized. 38 The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure 5: Occupations of migrant workers and nationals, latest years

Total nationals and total migrant workers

High-skilled jobs Semi-skilled jobs

High-income countries

Greece United States Italy Switzerland Spain Hungary Portugal Lithuania Argentina Poland Iceland Canada Slovakia Estonia Cyprus Czech Republic Chile Finland Estonia Denmark Croatia Latvia Latvia Iceland Austria Luxembourg Ireland Norway France Sweden United States France Denmark Australia Netherlands United Kingdom Finland Austria Belgium Malta Poland Belgium Luxembourg Slovenia Sweden Cyprus United Kingdom Ireland Lithuania Croatia Hungary Greece Norway Italy Czech Republic Portugal Canada Netherlands Switzerland Spain Australia Argentina Slovenia Slovakia Malta Chile High-income High-income 020 40 60 80 100 0 20 40 60 80 100 Share with high-skilled jobs (%) Share with semi-skilled jobs (%)

Low- and middle-income countries

Jordan Nepal Costa Rica Romania Nepal Albania Malawi Madagascar Turkey Bangladesh Gambia Turkey Namibia Gambia Bangladesh Namibia Mexico Tanzania** Tanzania** Bolivia* Sierra Leone Bulgaria Madagascar Sierra Leone Albania Costa Rica Bulgaria Jordan Bolivia* Mexico Romania Malawi Low- and middle-income Low- and middle-income 0 20 40 60 80 100 0 20 40 60 80 100 Share with high-skilled jobs (%) Share with semi-skilled jobs (%)

Migrant workers Non-migrant workers Chapter 2 – Labour market characteristics of migrant workers and nationals 39

Low-skilled jobs Unskilled jobs

High-income countries

Chile Poland Argentina Czech Republic Denmark Slovakia Spain Switzerland Cyprus Norway Australia Canada Netherlands Australia Norway Argentina Sweden Sweden United Kingdom Estonia Belgium United Kingdom France United States Ireland Lithuania Croatia Netherlands Iceland Latvia Luxembourg Finland Greece Chile Portugal Luxembourg Hungary Croatia Austria Denmark Lithuania Belgium Finland Hungary Canada Austria United States Ireland Czech Republic France Italy Italy Switzerland Iceland Latvia Spain Slovakia Portugal Poland Cyprus Estonia Greece High-income High-income 020 40 60 80 100 0 20 40 60 80 100 Share with low-skilled jobs (%) Share with unskilled jobs (%)

Low- and middle-income countries

Romania Bulgaria Costa Rica Romania Gambia Albania Madagascar Bolivia* Bolivia* Mexico Malawi Jordan Mexico Sierra Leone Bulgaria Tanzania** Namibia Nepal Sierra Leone Turkey Tanzania** Namibia Bangladesh Malawi Jordan Madagascar Turkey Bangladesh Albania Costa Rica Nepal Gambia Low- and middle-income Low- and middle-income 0 20 40 60 80 100 0 20 40 60 80 100 Share with low-skilled jobs (%) Share with unskilled jobs (%)

Migrant workers Non-migrant workers

(Figure 5 continued on page 40) 40 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure 5 continued from page 39)

Personal care and domestic workers Women are over-represented in semi-skilled and personal care High-income countries and domestic work jobs

Australia Among the sample of HICs, more women wage workers tend to have semi-skilled occupations Chile than their male counterparts, on average. This is true for both migrant and non-migrant women in United States the sample of HICs. Among non-migrants, while about 27.9 per cent of women wage workers have

Argentina semi-skilled jobs, 14.8 per cent of men wage workers have semi-skilled jobs on average in HICs, with variations across countries (figure 6). Likewise, High-income the share of migrant women wage workers who 0 20 40 60 80 100 tend to have semi-skilled jobs is almost double Share with domestic/personal the respective share of their male counterparts in care workers the sample of HICs, on average (23.5 per cent and Low- and middle-income countries 12.6 per cent, respectively) (figure 7). In terms of high-skilled jobs, the shares of women and men Sierra Leone migrant wage workers are similar (29.7 and 29.4 per Bangladesh cent, respectively). However, for non-migrants, the Mexico share of women wage workers in high-skilled jobs is Bolivia* somewhat higher than the respective share of their Gambia male counterparts, namely about 47.5 per cent compared to 42.8 per cent for men wage workers Turkey in those jobs.42 In contrast, the share of men wage Albania workers (migrant and non-migrant men alike) in the Nepal low-skilled job categories is much higher than the Madagascar respective share of women in the sample of HICs Namibia (figures 6 and 7). These are job categories that are predominantly located in universally male domi- Tanzania** nated sectors such as construction, manufacturing, Costa Rica and mining and quarrying. For unskilled jobs, the Jordan share of men wage workers is only slightly higher Low- and middle-income (non-migrants) or almost similar (migrants) than the 0204060 80 100 share of their women counterparts in those jobs. Share with domestic/personal However, in terms of personal care and domestic care workers work jobs, which is among the lowest paid, the

Migrant workers Non-migrant workers share of women nationals who take these jobs is much higher than that of men nationals in almost all the countries where data is available (14.0 per

Note: High-income and low- and middle-income estimates are the cent and 5.6 per cent, on average, respectively). weighted averages of the sample of high-income countries and low- Among migrant wage workers, on the other hand, and middle-income countries, respectively. Averages are weighted the share of migrant women in personal care and by the number of wage employees in each country. High-skilled jobs domestic work jobs is more than triple of the share include, where possible, senior officials, CEOs and other managerial of men working in those jobs (33.8 per cent and positions, professional jobs, technical and associate professional jobs. Semi-skilled jobs consist of clerical support workers, service and sales 9.9 per cent, respectively). workers, and skilled agricultural, forestry and fishery workers. Low- In the sample of LMICs, the share of wage work- skilled jobs comprise craft and related trades workers, plant and machine operators, and assemblers. Unskilled jobs include elementary ers with high-skilled jobs is higher among women occupations. Personal care and domestic workers are grouped sep- nationals than among men nationals, on average arateyl in countries where independent identification of such occupations is possible. * the Plurinational State of Bolivia; ** the United Republic of Tanzania. 42 This should not be confused with the fact that fewer women than men Source: ILO estimates based on survey data provided by national hold senior management and leadership positions in most enterprises sources (see Appendix II). (see, e.g. ILO, 2015b, 2018a, 2019). Chapter 2 – Labour market characteristics of migrant workers and nationals 41

X Figure 6: Occupations of nationals by sex, latest years

Non-migrant men and non-migrant women

High-skilled jobs Semi-skilled jobs

High-income countries

Argentina Slovenia Chile Canada Portugal United States Spain Switzerland Croatia Lithuania Italy Poland Ireland Estonia Czech Republic Latvia Australia Hungary Greece Norway Slovakia Malta Hungary Denmark Austria Netherlands France Sweden Poland Portugal Iceland Belgium Cyprus Australia Finland Chile United Kingdom Luxembourg Netherlands Argentina Belgium France Sweden Ireland Denmark Finland Latvia Czech Republic Switzerland Austria United States Cyprus Estonia Slovakia Lithuania Iceland Luxembourg Croatia Norway Italy Canada United Kingdom Malta Spain Slovenia Greece High-income High-income 020 40 60 80 100 0 20 40 60 80 100 Share with high-skilled jobs (%) Share with semi-skilled jobs (%)

Low- and middle-income countries

Malawi Albania Bangladesh Bangladesh Mexico Jordan Namibia Nepal Costa Rica Madagascar Tanzania** Turkey Gambia Gambia Nepal Bolivia* Madagascar Sierra Leone Turkey Malawi Bulgaria Bulgaria Romania Romania Bolivia* Tanzania** Albania Namibia Sierra Leone Costa Rica Jordan Mexico Low- and middle-income Low- and middle-income 0 20 40 60 80 100 0 20 40 60 80 100 Share with high-skilled jobs (%) Share with semi-skilled jobs (%)

Non-migrant women Non-migrant men

(Figure 6 continued on page 42) 42 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure 6 continued from page 41)

X Figure 6: OccupationsLow-skilled of nationals jobs by sex, latest years Unskilled jobs

High-income countries

Luxembourg Norway Netherlands Canada Iceland Sweden Chile United States Belgium Switzerland Greece Australia Cyprus Netherlands United Kingdom Iceland Argentina Denmark Denmark Luxembourg Norway United Kingdom Finland Slovakia Sweden Finland France Czech Republic Spain Estonia Austria Ireland Latvia Italy Croatia Austria Estonia Argentina Italy Belgium Poland Lithuania Lithuania Greece Slovakia Cyprus Portugal Poland United States Latvia Czech Republic Croatia Hungary Hungary Ireland France Australia Spain Canada Chile Switzerland Portugal High-income High-income 020 40 60 80 100 0 20 40 60 80 100 Share with low-skilled jobs (%) Share with unskilled jobs (%)

Low- and middle-income countries

Bolivia* Jordan Gambia Albania Jordan Mexico Costa Rica Romania Malawi Sierra Leone Tanzania** Bolivia* Sierra Leone Turkey Namibia Costa Rica Nepal Bulgaria Turkey Bangladesh Mexico Tanzania** Madagascar Namibia Bulgaria Madagascar Romania Gambia Albania Nepal Bangladesh Malawi Low- and middle-income Low- and middle-income 0 20 40 60 80 100 0 20 40 60 80 100 Share with low-skilled jobs (%) Share with unskilled jobs (%)

Non-migrant women Non-migrant men Chapter 2 – Labour market characteristics of migrant workers and nationals 43

Personal care and domestic workers (25.2 per cent and 18.0 per cent, respectively) (figure 6), and higher among migrant women than High-income countries among migrant men, on average (36.3 per cent and 31.4 per cent, respectively). On average, the fraction Australia of women among the total women wage workers is also higher than the corresponding share of Chile men in terms of the unskilled job category (among migrant wage workers), and personal care and United States domestic work category (among both migrant and non-migrant wage workers). However, lower share

Argentina of women than that of men have semi-skilled and low-skilled jobs (among both migrant and non-migrant wage workers). High-income

0 20 40 60 80 100 Share with domestic/personal care workers

Low- and middle-income countries

Gambia Namibia Jordan Madagascar Sierra Leone Albania Nepal Costa Rica Tanzania** Bolivia* Mexico Turkey Bangladesh Low- and middle-income 020406080 100 Share with domestic/personal care workers

Non-migrant women Non-migrant men

Note: High-income and low- and middle-income estimates are the weighted averages of the sample of high-income countries and low- and middle-income countries, respectively. Averages are weighted by the number of wage employees in each country. High-skilled jobs include, where possible, senior officials, CEOs and other managerial positions, professional jobs, technical and associate professional jobs. Semi-skilled jobs consist of clerical support workers, service and sales workers, and skilled agricultural, forestry and fishery workers. Low- skilled jobs comprise craft and related trades workers, plant and machine operators, and assemblers. Unskilled jobs include elementary occupations. Personal care and domestic workers are grouped sep- arateyl in countries where independent identification of such occu- Portrait of domestic worker Hellina Desta. Seven years pations is possible. * the Plurinational State of Bolivia; ** the United ago, Hellina migrated from Ethiopia to work in Beirut, Lebanon. She has been working for her current Republic of Tanzania. employer for five years and plans to keep living and Source: ILO estimates based on survey data provided by national working in Lebanon. © shutterstock.com sources (see Appendix II). 44 The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure 7: Occupations of migrant workers by sex, latest years

Migrant men and migrant women

High-skilled jobs Semi-skilled jobs

High-income countries

Slovakia United States Greece Switzerland Spain Canada Italy Lithuania Portugal Malta Argentina Denmark Cyprus Iceland Chile Argentina Iceland Finland Estonia Sweden Latvia Ireland Belgium Hungary France Luxembourg Netherlands Czech Republic Austria Norway Ireland France Finland Estonia United States Portugal Denmark Latvia Luxembourg Cyprus Poland Austria Croatia United Kingdom Sweden Australia United Kingdom Belgium Australia Croatia Norway Greece Hungary Slovenia Lithuania Netherlands Czech Republic Spain Switzerland Italy Canada Chile Slovenia Poland Malta Slovakia High-income High-income 0 20 40 60 80 100 0 20 40 60 80 100 Share with high-skilled jobs (%) Share with semi-skilled jobs (%)

Low- and middle-income countries

Jordan Romania Malawi Madagascar Costa Rica Jordan Nepal Albania Bangladesh Nepal Tanzania** Bangladesh Gambia Bolivia* Madagascar Gambia Turkey Turkey Mexico Namibia Namibia Tanzania** Bulgaria Sierra Leone Albania Bulgaria Sierra Leone Malawi Bolivia* Costa Rica Romania Mexico Low- and middle-income Low- and middle-income 0 20 40 60 80 100 0 20 40 60 80 100 Share with high-skilled jobs (%) Share with semi-skilled jobs (%)

Migrant women Migrant men Chapter 2 – Labour market characteristics of migrant workers and nationals 45

X Figure 6: OccupationsLow-skilled of nationals jobs by sex, latest years Unskilled jobs

High-income countries

Croatia Hungary Slovakia Poland Poland Canada Cyprus Argentina Chile Australia France United States Spain Switzerland Greece Slovakia Luxembourg Czech Republic Belgium Netherlands Sweden United Kingdom Denmark Norway Argentina Chile Austria Sweden Norway Denmark Iceland Croatia Portugal Estonia United Kingdom Ireland Italy Finland Lithuania Latvia Netherlands Lithuania Czech Republic Austria Australia Luxembourg Latvia Belgium United States France Finland Italy Ireland Iceland Hungary Spain Estonia Cyprus Canada Greece Switzerland Portugal High-income High-income 020 40 60 80 100 020 40 60 80 100 Share with low-skilled jobs (%) Share with unskilled jobs (%)

Low- and middle-income countries

Gambia Bolivia* Madagascar Sierra Leone Bolivia* Madagascar Sierra Leone Jordan Nepal Mexico Jordan Albania Namibia Tanzania** Costa Rica Namibia Tanzania** Turkey Mexico Costa Rica Malawi Albania Bangladesh Turkey Malawi Bangladesh Gambia Bulgaria Nepal Low- and middle-income Low- and middle-income 020 40 60 80 100 020 40 60 80 100 Share with low-skilled jobs (%) Share with unskilled jobs (%)

Migrant women Migrant men

(Figure 7 continued on page 46) 46 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure 7 continued from page 45)

Personal care and domestic workers Association between educational and occupational attainments: Migrant High-income countries workers vs. nationals

Chile Could it be that the occupational attainment gaps observed between migrant and non-migrant workers, especially in HICs, exist because of different United States educational composition of migrant workers and nationals? Figure 8 illustrates that very often migrant workers are unable to attain the same types of job Argentina as their national counterparts, even when they have the similar levels of education. It appears from the High-income figure that, in the sample of HICs, migrant workers tend to receive less returns to their educational qual- 0 20 40 60 80 100 ifications relative to the qualifications of nationals. Share with domestic/personal This phenomenon reflects labour underutilization care workers and skills mismatch among migrant workers in HICs43. Migrant workers in HICs are typically likely Low- and middle-income countries to be affected by skills mismatch, in particular, Nepal mismatch by level of education, which reduces

Mexico their potential contribution to businesses and economies of their countries of destination. On the Gambia other hand, migrant workers in LMICs, on average Bangladesh tend to receive better returns to their educational Bolivia* qualifications, which is consistent with the notion that migrant workers from the Global North receive Turkey a premium for their labour market attributes and Namibia characteristics in the Global South. Albania The section below explores this association between Tanzania** educational and occupational attainments by look- Costa Rica ing at the proportions of migrant wage workers and nationals with secondary school education (Panel A) Madagascar and the proportion with university education Jordan (Panel B) against the proportions with high- and Low- and middle-income semi-skilled occupations.44 0 20 40 60 80 100 To better understand the illustration in figure 8, each Share with domestic/personal chart is divided into four quadrants. Quadrant 1 indi- care workers cates a situation whereby the proportion of workers Migrant women Migrant men with higher education is lower than the proportion of workers with high- and semi-skilled jobs within a country. Quadrants 2 and 3 show the cases where the Note: High-income and low- and middle-income estimates are the weighted averages of the sample of high-income countries and low- proportion of workers with higher education is simi- and middle-income countries, respectively. Averages are weighted lar to the proportion of workers with high- and semi- by the number of wage employees in each country. High-skilled jobs skilled jobs. Quadrant 4 indicates a scenario where include, where possible, senior officials, CEOs and other managerial the proportion of workers with higher education positions, professional jobs, technical and associate professional jobs. Semi-skilled jobs consist of clerical support workers, service and sales outweighs the proportion with high- and semi-skilled workers, and skilled agricultural, forestry and fishery workers. Low- jobs. Quadrants 2 and 3 represent “fair” scenarios, skilled jobs comprise craft and related trades workers, plant and ma- whereas Quadrants 1 and 4 represent “unfair” chine operators, and assemblers. Unskilled jobs include elementary occupations. Personal care and domestic workers are grouped sep- arateyl in countries where independent identification of such occupations is possible. * the Plurinational State of Bolivia; ** the United 43 See Sparreboom and Tarvid (2017 and ILO, 2018g). Republic of Tanzania. 44 High skilled jobs include senior officials and other managerial positions, professional jobs, technical and associate professional jobs while semi- Source: ILO estimates based on survey data provided by national skilled jobs consist of clerical support workers, service and sales workers, sources (see Appendix II). and skilled agricultural, forestry and fishery workers. Chapter 2 – Labour market characteristics of migrant workers and nationals 47

X Figure 8. The association between education and occupational attainment of migrant and non-migrant workers, latest years

Panel A: Comparing the share of wage workers with secondary education and that with high- or semi-skilled occupations

High-income countries Low- and middle-income countries High-income countries Low- and middle-income countries

Non-migrant wage workers Non-migrant wage workers

100 100 Quadrant 1 MLT Quadrant 2 SVN

LUX GBR BEL NOR GRC ISL CYP SWE BOL 75 NLD 75 MEX ESP IRL FRA / FIN CHE ITA AUS JOR ARG DNKAUTESTSVK LVA SLE PRT CHLHRV CZE CRI CAN HUN LTU GMB BGR ROU USA TZA TUR POL 50 50 NAM ALB MWI MDG NPL BGD 25 25 e with high- and semi-skilled jobs (% ) e with high- and semi-skilled jobs (%) Quadrant 3 Quadrant 4 0 0 Shar 0255075100 Shar 0255075100

Share with secondary education (%) Share with secondary education (%)

Migrant wage workers Migrant wage workers

100 100 MLT SVN ROU MEX BGR

BOL 75 AUS 75 SLE CHL CZE NOR GBR NLD SWE CAN MWI TZA CHE BEL SVK ARG LUX IRL FIN HUN LTU ALB MDG NAM 50 DNK HRV 50 FRA ESP AUT POL JOR BGD CYP LVA CRI GMB TUR PRT EST ITA USA GRC ISL 25 25

NPL e with high- and semi-skilled jobs (%) e with high- and semi-skilled jobs (% ) 0 0 Shar Shar 0255075100 0255075100

Share with secondary education (%) Share with secondary education (%)

Note: The 45-degree line is a line of equality. Points on this line indicate that the probability of attaining higher education equals the probability of attaining a high- or semi-skilled job. The horizontal and vertical lines divide the plot into four quadrants. The fitted line on the scattered dots shows the direction of the correlation between education and occupational attainment. High-skilled jobs include senior officials and other managerial positions, professional jobs, technical and associate professional jobs while semi-skilled jobs consist of clerical support workers, service and sales workers, and skilled agricultural, forestry and fishery workers. Source: ILO estimates based on survey data provided by national sources (see Appendix II).

(Figure 8 continued on page 48) 48 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure 8 continued from page 47)

Panel B: Comparing the share of wage workers with university education and that with high-skilled occupations

High-income countries Low- and middle-income countries High-income countries Low- and middle-income countries

Non-migrant wage workers Non-migrant wage workers

100 100 Quadrant 1 Quadrant 2 SVN

75 75

NOR CHE MLT LUX SWENLD GBR 50 USA EST 50 CAN AUS FIN BEL LVA LTU AUT FRA CYP IRL ALB CZE ITA ISL GRC SLE JOR HUN PRT ESP BOL SVK DNK GMB ROU CHL HRV POL BGR 25 ARG 25 TUR e with high-skilled jobs (%) e with high-skilled jobs (% ) NAM CRL BGD NPL MEX Shar Shar MWI Quadrant 3 Quadrant 4 0 0 0255075100 0255075100 Share with university education (%) Share with university education (%)

Migrant wage workers Migrant wage workers

100 100

ROU

75 MLT 75 SVN

BOL AUS BGR 50 50 CAN CHE NOR CZE ALB HUN SLE LTU GBR LUX POL SWE USA FIN BEL BGD MEX LVA NLD DNK NAM 25 FRA AUT IRL 25 GMB e with high-skilled jobs (% ) EST e with high-skilled jobs (%) TUR HRVCHL CYP SVK MWI ARG ISL NPL Shar PRT Shar ITA ESP CRI GRC 0 0 JOR 0255075100 0255075100 Share with university education (%) Share with university education (%)

Note: The 45-degree line is a line of equality. Points on this line indicate that the probability of attaining higher education equals the probability of attaining a high- or semi-skilled job. The horizontal and vertical lines divide the plot into four quadrants. The fitted line on the scattered dots shows the direction of the correlation between education and occupational attainment. High-skilled jobs include senior officials and other managerial positions, professional jobs, technical and associate professional jobs. Source: ILO estimates based on survey data provided by national sources (see Appendix II). Chapter 2 – Labour market characteristics of migrant workers and nationals 49

X Figure 9. The association between education and occupational attainment of migrant workers by sex, latest years

High-income countries Low- and middle-income countries

Migrant men Migrant men

100 100 Quadrant 1 MLT Quadrant 2 SVN BGR MEX BOL ROU 75 75 CHL MWI SLE AUS ARG TZA NOR BEL NLD CAN CHE CZE MDG LUX GBR IRL BGD 50 SWE 50 JOR LTU DNK FIN GMB FRA ESP SVK ALB HUN CYP CRI PRT AUT LVA TUR NAM USA EST POL ISL GRC 25 HRV 25 ITA

NPL e with high- and semi-skilled jobs (% ) Quadrant 3 Quadrant 4 e with high- and semi-skilled jobs (%) 0 0 Shar 0255075100 Shar 0255075100 Share with secondary education (%) Share with secondary education (%)

Migrant women Migrant women

100 100 SVN MLT ROU / SLE BOL SVK POL BGR AUS MEX 75 GBR HRV CZE 75 HUN NOR SWE CAN LUX NLD CHL NAM ALB CHE LTU AUT BEL LVA IRL DNK NPL TUR FRA ESP ITA 50 FIN 50 CYP EST GMB GRC MWI CRI PRT ISL USA ARG BGD MDG 25 25 NPL

JOR e with high- and semi-skilled jobs (%) e with high- and semi-skilled jobs (% ) 0 0 Shar Shar 0255075100 0255075100 Share with secondary education (%) Share with secondary education (%)

scenarios. The 45-degree line is a line of equality, with quadrant in more countries than migrant women points on this line indicating that the probability of are, even though their educational attainments are having a higher education equals the probability similar. In other words, given similar levels of edu- of attaining a high- and/or semi-skilled job. cation, migrant men tend to have fewer high- and semi-skilled jobs than migrant women, in particular, In Panel A, nationals across the sample of HICs are in the sample of HICs. This phenomenon is consis- located in the second quadrant, implying that the tent with estimates in figure 7, where the share of proportion of nationals with higher education is sim- migrant women workers with semi-skilled jobs far ilar to the proportion of nationals with jobs in higher outweighs the share of migrant men in those jobs. occupational categories. However, migrant workers tend to be located in the fourth quadrant in some The observed association between educational HICs. In Panel B, the proportion of migrant workers and occupational attainment, especially in HICs, is in the sample of HICs that attain high skilled occu- consistent with the literature that partly attribute pations is much lower than the nationals of these observed gaps in labour market outcomes of countries, although the proportion with university migrant workers and nationals to differences in education is similar for both groups. This illustrates returns to foreign-acquired education (see Barrett, that, given similar levels of education, the probability McGuinness and O'Brien, 2012). Not only are of attaining high- or semi-skilled occupation is much employers in HICs possibly reluctant to hire migrant lower for migrant workers than for nationals. This workers with foreign education, but also, they may means that higher educated migrant workers in be reluctant to hire migrant workers in general into HICs are less likely to attain jobs in higher occupa- high- or semi-skilled jobs, as many migrants update tional categories relative to non-migrant workers. and pursue higher education in their destination For example, in the United States and Finland, while countries (see Faini, 2005; Mahroum, 2000). the share of migrant workers with secondary school education is 78.7 per cent and 98.6 per cent, respectively, the shares of migrant workers with high- or 2.5.4. Differences between wage semi-skilled jobs are only 35.8 per cent and 50.6 per workers by economic sector cent, respectively. Among nationals, however, the proportion with high- or semi-skilled jobs is 61.0 per Migrant wage workers are over- cent (in the United States) and 72.4 per cent (in represented in the primary and secondary Finland) though their respective proportions with sectors, particularly in HICs secondary school education are not too different from that of migrant workers. When looking at the distribution of wage employees by industrial sector, the report finds that, on In the sample of LMICs, the story is more positive average, migrant wage workers, compared to for migrant workers, with higher educated migrant nationals, are disproportionately represented in the workers more likely to attain high- or semi-skilled primary sector – agriculture, fishing and forestry – occupations relative to nationals of these countries. in the sample of HICs, while in the sample of LMICs, It is also the case that migrant workers in middle-in- the proportions of both groups are similar. In the come countries such as the Plurinational State of sample of HICs, more migrant wage workers than Bolivia, Mexico, and Romania are more likely than nationals take up secondary sector jobs (mining nationals to have university education and they get and quarry; manufacturing; electricity, gas and rewarded accordingly (Panel B). water; and construction), while in the sample of A similar pattern of how migrant workers’ educa- LMICs, they (migrant wage workers) tend to take tional attainment is associated with their occupa- up fewer secondary sector jobs, on average, than tional attainment is exhibited in figure 9, showing nationals. However, while there is a tendency for the pattern separately for migrant men and migrant fewer migrant wage workers to be employed in women using the share of wage workers with sec- the tertiary sector (i.e. services) in HICs, they tend ondary school education. Although the pattern in to take up more tertiary sector jobs than nationals figure 9 is similar to the overall pattern for migrant in the sample of LMICs. In terms of distribution by workers in figure 8, it appears that migrant men gender, migrant men wage workers tend to work and migrant women in the sample of HICs face more in the primary and secondary sectors in the somewhat different treatment with respect to sample of HICs and the tertiary sector in the sample their occupational attainments given similar levels of LMICs than their national counterparts. Similarly, of education. Migrant men are located in the fourth migrant women wage workers tend to work more 51

in the primary and secondary sectors in the sample of HICs and the primary and tertiary sectors in the sample of LMICs than their national counterparts. The section below explores the types of industrial sectors in which wage employees take their main occupations by comparing migrant workers and nationals. The industrial categories for almost all countries follow the classifications given by the International Standard Industrial Classification of All Economic Activities (ISIC) (UN, 2008). This report groups industries into three main sectors: primary, secondary, and tertiary.45 Salinas, California - USA; July 1, 2015: Agricultural seasonal migrant farm workers harvest and package Romaine lettuce in the Figure 10 presents the proportion of wage employ- fields of Salinas Valley of central California. © shutterstock.com ees in each industrial sector by comparing migrant and non-migrant wage workers. The charts show that, on average, migrant workers, compared and 77.7 per cent, respectively), they tend to take up to nationals, are disproportionately represented more tertiary sector jobs than nationals in the sam- in the primary sector in the sample of HICs than ple of LMICs, with 64.6 per cent of migrant workers non-migrant workers (2.5 per cent and 1.5 per cent, in LMICs working in the tertiary sector compared to respectively), while in the sample of LMICs, the 57.1 per cent of nationals, on average. proportions of both groups are similar (10.6 per Figure A-2 in Appendix IV presents a similar pattern cent and 10.3 per cent, respectively). However, the of the distribution of migrant and non-migrant wage overall proportions of both migrant workers and workers by industrial sector, distinguishing between nationals in the primary sector is higher in LMICs men and women. The share of migrant men than in HICs, with some variation within income employed in the primary and secondary sectors is groups. For example, while about 11.6 per cent of higher than the corresponding share of non-migrant migrant wage workers in Greece work in the pri- men in the sample of HICs (3.4 per cent and 2.1 per mary sector, only about 1.2 per cent of nationals cent, respectively in the primary sector, and 38.1 per work in this sector. On the other hand, in Chile, only cent and 31.1 per cent, respectively in the secondary about 2.9 per cent of migrant wage workers take up sector). Similarly, on average, proportionately more primary sector jobs but more than 8.4 per cent of migrant women than non-migrant women are nationals work in this sector. employed in the primary (1.5 per cent and 0.8 per Similarly to the primary sector, more migrant wage cent, respectively) and secondary (10.9 per cent and workers than nationals in HICs take up secondary 9.7 per cent, respectively) sectors in HICs. However, sector jobs with about 26.8 per cent of migrant on average proportionately fewer migrant men and wage workers employed in this sector, compared migrant women attain employment in the tertiary to 20.8 per cent of non-migrant wage workers. sector compared to their national counterparts In contrast, proportionately fewer migrant wage in HICs (58.5 per cent and 66.8 per cent, respec- workers in the sample of LMICs are employed in the tively for migrant men and non-migrant men, and secondary sector compared to nationals of these 87.5 per cent and 89.5 per cent, respectively for countries (24.9 per cent and 32.6 per cent, respec- migrant women and non-migrant women). tively), with notable variations across countries and In the sample of LMICs, migrant men are less within income groups. represented in the primary and secondary sectors The tertiary sector, on the other hand, absorbs compared to non-migrant men, on average, but more wage employees than the primary and sec- over-represented in the tertiary sector. Migrant ondary sectors combined in both samples of coun- women in the sample of LMICs, on the other hand, tries (HICs and LMICs). While there is a tendency are over-represented in the primary and tertiary for fewer migrant workers to be employed in the sectors compared to non-migrant women but less tertiary sector than nationals in HICs (70.7 per cent represented in the secondary sector.

45 The primary industrial sector comprises agriculture, fishing and forestry. The secondary sector consists of mining and quarry; manufacturing; electricity, gas and water; and construction. The tertiary sector includes all services and industries not captured under the primary and secondary sectors. 52 The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure 10: The distribution of migrant and non-migrant wage workers by industrial sector, latest years

Primary sector Secondary sector

High-income countries

Netherlands Denmark Croatia Sweden Slovakia Spain Poland Malta Czech Republic Australia Lithuania Cyprus Argentina Netherlands Finland Chile Austria Belgium Belgium Canada Canada Switzerland Estonia Portugal Luxembourg Norway Australia France Ireland Switzerland Luxembourg United Kingdom United Kingdom Sweden Greece Norway Finland Latvia Croatia Slovenia United States Ireland Hungary United States Argentina Chile Italy Italy Czech Republic Iceland Austria Cyprus Latvia France Slovakia Portugal Iceland Denmark Estonia Spain Slovenia Hungary Lithuania Greece Poland High-income High-income 0 10 20 30 0 20 40 60 80 Share in primary sector (%) Share in secondary sector (%)

Low- and middle-income countries

Bulgaria Sierra Leone Romania Romania Albania Gambia Sierra Leone Bulgaria Turkey Mexico Tanzania** Mexico Madagascar Tanzania** Bolivia* Madagascar Costa Rica Nepal Namibia Bolivia* Bangladesh Gambia Jordan Costa Rica Albania Namibia Turkey Bangladesh Nepal Low- and middle-income Low- and middle-income 0 10 20 30 0 20 40 60 80 Share in primary sector (%) Share in secondary sector (%)

Migrant workers Non-migrant workers Chapter 2 – Labour market characteristics of migrant workers and nationals 53

Tertiarty sector 2.5.5. Contractual conditions of wage workers

High-income countries This section examines the contractual arrange- Poland ments under which migrant and non-migrant wage Lithuania workers are employed. In countries covered in the Slovenia Estonia report, most wage employee jobs are based on Iceland written contracts. However, significant variations Hungary exist across countries in institutional arrangements Latvia Greece regarding the duration of the contract and working Slovakia time. Taking account of these differences, the notion Italy Austria of a “temporary or limited-time job” is defined as a Czech Republic fixed-term contract or a pre-determined end date United States Argentina of work (usually within one year) including seasonal, Croatia daily, and non-contractual occasional workers. The France notion of “permanent job”, on the other hand, is United Kingdom Finland defined as a work contract of unlimited duration or Ireland no pre-determined termination date. Portugal Luxembourg Working time is derived from declared hours of Norway Spain work per week, similar to the ILO’s Global Wage Chile Report 2018/19, using the OECD definition of part- Switzerland Canada time workers as those who declare their usual Cyprus working time per week as 30 hours or fewer (van Belgium Bastelaer, Lemaître and Marianna, 1997). Netherlands Australia Denmark Sweden Migrant wage workers are more likely to Malta work under temporary contracts than High-income 0 20 40 60 80 100 non-migrant wage workers in both HICs and LMICs Share in tertiary sector (%) Figure 11 shows that migrant wage workers in both HICs and LMICs are more likely than nation- Low- and middle-income countries als to work under temporary contracts, on average Nepal (27.0 per cent and 14.9 per cent, respectively in the Bangladesh sample of HICs, and 42.9 per cent and 41.7 per cent, Turkey respectively in the sample of LMICs), with few excep- Albania tions including Australia, Canada, Chile, Hungary, Namibia Ireland, and Latvia (among the sample of HICs); Costa Rica and Bangladesh, Malawi, and Mexico (among the Jordan sample of LMICs), and variations across countries. Bolivia* Gambia Madagascar Note (figure 10): High-income and low- and middle-in- Tanzania** come estimates are the weighted averages of the sample of Mexico high-income countries and low- and middle-income coun- Bulgaria tries, respectively. Averages are weighted by the number Sierra Leone of wage employees in each country. Primary sector com- Romania prises Agriculture, Fishing and Forestry. Secondary sector Low- and middle-income consists of Mining and Quarry; Manufacturing; Electricity, 0 20 40 60 80 100 Gas and Water; and Construction. Tertiary sector includes all services and industries not captured under the Primary Share in tertiary sector (%) and Secondary sectors. Figure A-2 in Appendix IV shows this figure in terms of men and women. * the Plurinational Migrant workers Non-migrant workers State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). 54 The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure 11: The share of wage employees with temporary work contracts by sex, latest years

Total wage workers Migrant men and non-migrant men

High-income countries

Latvia Latvia Estonia Estonia Ireland Lithuania United Kingdom United Kingdom Hungary Ireland Luxembourg Malta Norway Hungary Lithuania Luxembourg Austria Norway Canada Austria Denmark Canada Malta Denmark Iceland Belgium Belgium Switzerland Switzerland Iceland Finland Cyprus France Finland Australia France Chile Sweden Italy Australia Czech Republic Italy Portugal Argentina Slovenia Chile Sweden Slovenia Cyprus Czech Republic Slovakia Portugal Croatia Greece Netherlands Croatia Argentina Spain Greece Slovakia Spain Netherlands Poland Poland High-income High-income 0 20 40 60 80 100 0 20 40 60 80 100 Share (%) Share (%)

Low- and middle-income countries

Bulgaria Bulgaria Costa Rica Costa Rica Romania Turkey Turkey Albania Albania Bangladesh Bangladesh Sierra Leone Sierra Leone Mexico Mexico Romania Gambia Malawi Namibia Gambia Malawi Madagascar Madagascar Namibia Tanzania** Tanzania** Nepal Nepal Low- and middle-income Low- and middle-income 0 20 40 60 80 100 020 40 60 80 100 Share (%) Share (%)

Migrant workers Non-migrant workers Migrant men Non-migrant men Chapter 2 – Labour market characteristics of migrant workers and nationals 55

Migrant women and non-migrant women For example, in Poland, while only about 27.3 per cent of nationals work under temporary contracts, about 97.8 per cent of migrant wage workers have High-income countries temporary work contracts. On the other hand, in

Slovakia Canada, about 13.7 per cent of nationals work Latvia under temporary contracts whereas the corre- Estonia sponding fraction of migrant wage workers is about Hungary Luxembourg 12.2 per cent. These finding are consistent with Ireland previous ILO research that shows that a greater United Kingdom Norway percentage of migrant workers find themselves in Denmark non-standard jobs, including under temporary work Slovenia contracts around the world (ILO, 2016b). Canada Iceland Austria Figure 11 also shows that, both men and women Czech Republic migrant wage workers in the sample of HICs are Finland more likely than men and women nationals to Switzerland Portugal work under temporary contracts (26.7 per cent Lithuania and 14.0 per cent of migrant men and non-migrant Belgium France men, respectively; and 27.4 per cent and 15.8 per Malta cent of migrant women and non-migrant women, Chile respectively). However, as compared to their male Australia Italy counterparts, both migrant women and non-mi- Netherlands grant women are more likely to work under tempo- Croatia rary contracts. In the sample of LMICs, on the other Cyprus Sweden hand, the reports finds that migrant men wage Greece workers, compared to non-migrant men, tend to Spain Argentina work more under temporary contracts, on average Poland (45.4 per cent and 41.1 per cent, respectively), while High-income the corresponding proportions of migrant women 0 20 40 60 80 100 and non-migrant women are close (43.0 per cent Share (%) and 44.0 per cent, respectively).

Low- and middle-income countries Incidence of part-time work contract Romania is higher among migrant workers than Sierra Leone Costa Rica non-migrant workers in HICs but lower Albania than non-migrant workers in LMICs Turkey Bulgaria Figure 12 shows that migrant workers have slightly Bangladesh higher part-time incidence rates than non-migrant Mexico workers in the sample of HICs, on average (15.0 per Namibia cent and 14.6 per cent, respectively), primarily due Gambia to the significantly higher incidence of part-time Malawi Tanzania** Note (figure 11): High-income and low- and middle-income estimates are the weighted averages of the sample Nepal of high-income countries and low- and middle-income Madagascar countries, respectively. Averages are weighted by the Low- and middle-income number of wage employees in each country. Temporary 02040 60 80 100 120 work contract comprises employees with a fixed-term contracts or a pre-determined end date of work including Share (%) seasonal, daily, and non-contractual occasional workers. Permanent work contract consists of employees with a Migrant women Non-migrant women work contract of unlimited duration or no pre-determined termination date. ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). 56 The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure 12: The share of wage employees with temporary work contracts by sex, latest years

Total wage workers Migrant men and non-migrant men

High-income countries

Croatia Croatia Poland Czech Republic Czech Republic Poland Slovenia Hungary Chile Slovenia Latvia Latvia Iceland Chile Estonia Portugal Hungary Luxembourg Cyprus Estonia Portugal Belgium Malta Iceland Sweden Denmark United States France Luxembourg Norway Denmark United Kingdom Lithuania Netherlands Finland Italy Norway Malta Slovakia United States Italy Cyprus United Kingdom Sweden Austria Lithuania Belgium Austria Canada Spain Australia Argentina Greece Canada Spain Australia France Finland Ireland Ireland Switzerland Switzerland Netherlands Greece Argentina Slovakia High-income High-income 0 510 15 20 25 30 35 0 10 20 30 40 Incidence of part-time work (%) Incidence of part-time work (%)

Low- and middle-income countries

Romania Romania Bulgaria Bulgaria Madagascar Madagascar Sierra Leone Sierra Leone Bangladesh Tanzania** Tanzania Bolivia* Jordan Bangladesh Turkey Jordan Mexico Mexico Nepal Turkey Gambia Costa Rica Namibia Nepal Costa Rica Gambia Bolivia Albania Albania Namibia Malawi Malawi Low- and middle-income Low- and middle-income 0510 15 20 25 30 35 40 0 10 20 30 40 Incidence of part-time work (%) Incidence of part-time work (%)

Migrant workers Non-migrant workers Migrant men Non-migrant men Chapter 2 – Labour market characteristics of migrant workers and nationals 57

Migrant women and non-migrant women work contracts among migrant women compared to non-migrant women. In most countries, women and men (migrant workers and nationals alike) differ sig- High-income countries nificantly with respect to working time. Specifically, Slovakia part-time work is more prevalent among women Croatia than men in most countries, which is consistent with Poland Czech Republic previous findings in the literature (see, e.g., Fagan et Chile al., 2014; ILO, 2016c and ILO, 2019b). While part-time Cyprus incidence rates of migrant men is slightly lower than Iceland Latvia that of non-migrant men in the sample of HICs (7.7 Slovenia per cent and 8.3 per cent, respectively), an average Malta Portugal gap of 2.2 percentage points exists between the Estonia part-time rates of migrant women and non-migrant Sweden Finland women in HICs (23.8 per cent and 21.6 per cent, United States respectively), although the scale of the difference Lithuania varies widely across countries. Denmark Hungary In the sample of LMICs, incidence of part-time work Norway Canada tend to be lower among migrant workers than Luxembourg among non-migrant workers, on average (6.2 per Greece Italy cent and 8.7 per cent, respectively), with notable Austria variations across countries. Both migrant men United Kingdom Australia and migrant women in LMICs tend to have lower Ireland part-time incidence rates than their national coun- Switzerland terparts, on average (3.9 per cent and 6.5 per cent Belgium Spain of migrant men and non-migrant men, respectively, France and 10.3 per cent and 12.0 per cent of migrant Netherlands Argentina women and non-migrant women, respectively), High-income although part-time work is more prevalent among 0 10 20 30 40 50 60 women than among men in general. Incidence of part-time work (%)

Low- and middle-income countries

Romania Bangladesh Bulgaria Madagascar Sierra Leone Jordan Tanzania** Turkey Namibia Mexico Nepal Albania Gambia Costa Rica Bolivia* Note (figure 12): High-income and low- and middle-in- Malawi come estimates are the weighted averages of the sample Low- and middle-income of high-income countries and low- and middle-income 010 20 30 40 50 60 70 countries, respectively. Averages are weighted by the number of wage employees in each country. Incidence Incidence of part-time work (%) of part-time is derived using the OECD definition of part- time workers as those who declare their usual working Migrant women Non-migrant women time per week as 30 hours or fewer. * the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). 58 The migrant pay gap: Understanding wage differences between migrants and nationals

2.5.6. Public-sector employment working in low-paid positions. While they have been on the front line in the fight against COVID- 46 Public sector employment is higher among nation- 19, the pandemic has exposed their particular als than migrant workers in all the sampled HICs vulnerability to wage cuts, non-payment of wages and LMICs for which recent data is available except or deteriorating working conditions, affecting their for Bangladesh, Malawi and Sierra Leone, all of income (ILO, 2020d, 2020h). The COVID-19 crisis which are LMICs (figure 13), with variations across has brought about the urgent need to improve 47 countries. For example, in the case of Jordan, while the working conditions, including wages, and rep- only 2.0 per cent of migrant workers are employed resentation of rights of workers in essential care in the public sector, about 41.7 per cent of nationals services (ILO, 2020i). have some form of public sector employment. In contrast, in the Plurinational State of Bolivia, the This section presents estimates of the proportion proportions of migrant workers and nationals in the of migrant wage workers who are employed in the public sector are similar (24.3 per cent and 24.6 per care economy as compared to migrant wage work- cent, respectively). This pattern is true for both men ers in other sectors of the economy, distinguishing and women, where both non-migrant men and between men and women. For the purpose of this women are more likely to have occupations in the report and following the definitions of the ILO, public sector than their migrant counterparts, with the care workforce is defined, where possible, to few exceptions including in Bangladesh, Malawi, and include care workers in care sectors (education, Sierra Leone (where the share of migrant men with health and social work), care workers in other public sector jobs outweighs that of non-migrant sectors, personal care and domestic workers, and men); and in Albania, Bangladesh, the Plurinational non-care workers in care sectors who support care State of Bolivia, and Sierra Leone (where the share service provision. of migrant women with public sector jobs outweighs that of non-migrant women). Migrant women are over-represented in the care economy compared 2.5.7. Migrant care workers to migrant men

The ILO estimates that most care workers are Although a substantial proportion of the overall women, often migrant women who work under migrant workforce work in the care economy, poor conditions and receive low pay (ILO, 2018d; the proportion of migrant women engaged in King-Dejardin, 2019). Domestic work, which is the care economy far outweighs that of migrant often informal, is situated at the lowest end of the men. Among the sample of countries for which care economy, and probably among the lowest reliable data sources exist, the proportion of men paying jobs. As indicated earlier in this report, the and women migrant workers engaged in the care ILO estimates there are about 67 million domes- economy ranges from about 3 per cent in Slovenia tic workers worldwide, 80 per cent of which are to about 41 per cent in Denmark (figure 14). women; about 11 million are migrant domestic Migrant women are more likely than migrant men workers (ILO, 2015c). They often lack recognition to take up employment in the care economy, with as workers, and labour and social protection law the proportion of migrant women employed in coverage, including minimum wage regulation, the care economy exceeding that of migrant men has been weak on them. Migrant workers have in all the 25 countries reported in figure 14. For often been fulfilling the need for care workers in example, while only 9.1 per cent of migrant men high income and middle-income countries typically work in the care economy in Italy, about 65.8 per cent of migrant women are employed in the care

46 We define public sector employment as all wage employment at institutions which are controlled and mainly financed by public authority. It is composed of a general government sector (government units, social security funds, and non-profit/non-market institutions controlled by public authority) and a public corporation sector (see Hammouya, 1999). 47 The data may reflect situations envisaged in Article 14 (c) of the Migrant Workers (Supplementary Provisions) Convention, 1975 (No. 143). These include situations where the protection of the interests of the State justifies the restrictions of certain employment or functions, by reason of their nature, to nationals of that State. The ILO supervisory bodies have noted that in most countries public service jobs are reserved to nationals. In some countries, restrictions on access to state employment are applicable to all state jobs, while in others, citizenship requirements apply to public sector recruitment processes at the federal and subnational level. The concept of ”public service“ and ”public enterprises“ may however cover a wide range of activities. In this regard, the ILO supervisory bodies have pointed out that general prohibitions as regards the access of foreigners to certain occupations, when permanent, are contrary to the principle of equal treatment unless they apply to limited categories of occupations or public services and are necessary in the interest of the State (see ILO General Survey concerning the migrant workers instruments, 2016, Geneva, para. 370). Chapter 2 – Labour market characteristics of migrant workers and nationals 59

economy. This finding corroborates previous ILO migrant wage workers employed in the care econ- findings (ILO, 2018d; King-Dejardin, 2019). omy represent a smaller proportion of overall wage workers in the economy (figure 15). For example, Migrant wage workers employed while migrant workers make up 47.0 per cent of all in the care economy represent a smaller wage workers in Luxembourg, only about 11.7 per proportion of overall wage workers cent of wage workers are migrant care workers. The proportions differ for migrant men and women in Figure 15 shows the proportions of migrant wage each country. In the example of Luxembourg, while workers and migrant care workers among total about 23 per cent of all women wage workers (of wage workers, distinguishing between men and whom about 46.6 per cent are migrant women) are women. Though the proportion of care work- migrant care workers, in contrast, only 2.8 per cent ers among migrant wage workers is substantial of men wage workers (of whom about 47.3 per cent compared to other sectors (as seen in figure 14), are migrant men) are migrant care workers.

Migrant health workers. © Copyright ILO. Photographer: J. Reyes 60 The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure 13: Incidence of public sector employment among migrant and non-migrant workers by sex, latest years

High-income countries Low-middle-income countries

Total wage workers Total wage workers 40 50

30 40 30 20 yment (% ) yment (% ) 20 10 e of public-sector e of public-sector 10 emplo emplo Shar 0 Shar 0 y ico x Chile urke Nepal Jordan T Malawi Me Albania Canada zania** Gambia Bolivia* a Leone Namibia Australia n Argentina Costa Rica Ta Switzerland Bangladesh Madagascar Sierr United States

Migrant workers Non-migrant workers

Migrant men and non-migrant men Migrant men and non-migrant men 40 50

30 40 30 20 yment (% ) yment (% ) 20 10 e of public-sector e of public-sector 10 emplo emplo Shar Shar 0 0 Chile urk ey Nepal ambia n Jorda Leone T Malaw i o Mexic Albania Canada Bolivia * G Namibia Australia nzania* * Argentina Costa Rica Ta Switzerland Banglades h Madagasca r Sier ra United States

Migrant men Non-migrant men

Migrant women and non-migrant women Migrant women and non-migrant women 50 80 70 40 60 30 50 40 yment (% ) 20 yment (% ) 30 20 e of public-sector 10 e of public-sector emplo emplo 10 Shar 0 Shar 0 Chile urk ey Nepal ambia n Jorda Leone T o Mexic Malaw i Albania Canada Bolivia * G Namibia Australia nzania* * Argentina Costa Rica Ta Switzerland Banglades h Madagasca r Sier ra United States

Migrant women Non-migrant women

Note: Public sector employment refers to all wage employment at institutions which are controlled and mainly financed by public authority. It is composed of a general government sector (government units, social security funds, and non-profit, non-market institutions controlled by public authority) and a public corporation sector. * the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). Chapter 2 – Labour market characteristics of migrant workers and nationals 61

X Figure 14: The distribution of migrant wage workers between the care economy and other sectors by sex, latest years

Total migrant wage workers 100

40 Percent 20

0 ece nce land Italy e a Chile e Spain Latvia Jordan Fr Austria Cyprus Ir Gr Iceland Estonia Canada Norway Sweden Slovenia Belgium Australia embourg Denmark Argentina Costa Rica Switzerland Lux United States United Kingdom

Migrant men wage workers 100

40 Percent 20

0 ece Italy e Chile Spain ustri a Latvia Jordan France Cyprus A Gr Ireland Iceland Estonia ustralia Canad a Sweden Norwa y Slovenia Belgium A embourg Denmar k Argentina Costa Rica Switzerland Lux United State s United Kingdom

Migrant women wage workers 100

40 Percent 20

0 ece land Ital y e Chile Spai n e Latvia n Jorda France Cyprus Austria Gr Ir Estonia Iceland Canada Sweden Norwa y Slovenia Belgium Australia embourg Denmar k Argentina Costa Rica Switzerland Lux United States United Kingdom

The care economy Other sectors

Note: Care workforce includes care workers in care sectors (education, health and social work), care workers in other sectors, domestic workers and non-care workers in care sectors, who support care service provision. Other sectors include all sectors not captured under the care economy. Source: ILO estimates based on survey data provided by national sources (see Appendix II). 62 The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure 15: The shares of migrant wage workers and migrant care workers among total wage workers by sex, latest years

Total Sweden Women Men Total France Women Men Total Slovenia Women Men Total Denmark Women Men Total Norway Women Men Total Iceland Women Men Total Argentina Women Men Total Chile Women Men Total Belgium Women Men Total Greece Women Men Total United States Women Men Total Spain Women Men Total United Kingdom Women Men Total Italy Women Men Total Costa Rica Women Men Total Austria Women Men Total Estonia Women Men Total Ireland Women Men Total Latvia Women Men Total Cyprus Women Men Total Canada Women Men Total Australia Women Men Total Switzerland Women Men Total Jordan Women Men Total Luxembourg Women Men 0510 15 20 25 30 35 40 45 50 Percent

Share of migrant wage workers among total wage employees Migrant car workers

Note: Care workforce includes care workers in care sectors (education, health and social work), care workers in other sectors, domestic workers and non-care workers in care sectors, who support care service provision. Source: ILO estimates based on survey data provided by national sources (see Appendix II). Chapter 2 – Labour market characteristics of migrant workers and nationals 63 Chapter 3

Measuring and understanding the migrant pay gap

X Chapter 3 – Measuring and understanding the migrant pay gap

This chapter of the report explores the migrant pay gap48 using methods defined in the ILO’s Global 3.1. The wage structure of Wage Report 2018/19.49 The populations covered are migrant workers and nationals from 49 countries representing about a quarter of wage employees worldwide, and covering nearly Mean and median pay produce half (49.4 per cent) of all international migrants, migrant pay gap estimates with the and roughly 33.8 per cent of migrant workers worldwide. same signs but different magnitudes In summary, the findings in this chapter show that The terms “migrants’ pay” and “nationals’ pay” refer migrant workers in the sample of 33 HICs tend to to the full range of earnings received by all migrant earn significantly less than non-migrant workers workers and all nationals who are classified as paid in these countries, on average, and the observed or wage employees. This full range of wages is what pay gaps are largely unexplained by observed is referred to as “the wage distribution” or “the labour market characteristics. On the other hand, wage structure” of wage workers in the population. migrant workers in the sample of 16 LMICs tend The two measures that are commonly used to sum- to earn more than nationals of these countries, marize the information in such a distribution are on average. However, it is significant to add that the mean (the average of all the values covered) most of these LMICs are mainly countries of origin, and the median (the value located in the middle mostly with positive net emigration, where there of the distribution). Thus, the “mean migrant pay may be clusters of temporary high-skilled “expatri- gap” compares the average of the migrant work- ate” workers. There are of course few exceptions ers’ pay distribution to the average of the nationals’ among the sample of LMICs including Costa Rica, pay distribution, while the “median migrant pay Gambia, Jordan, Namibia, and Turkey where there gap” compares the value located in the middle of are relatively large migrant populations. Based on the migrant workers’ pay distribution to the value average hourly wages of nationals and migrant located in the middle of the nationals’ pay distribu- workers, the report estimates that the migrant pay tion. This may be a source of differences between gap is about 12.6 per cent (in favour of nationals) estimates. Visually examining and understanding in the sample of HICs and 8.6 per cent (in favour the wage structure is important because it explains of nationals) across the Member States of the EU, why using mean and median wages to estimate the while in the sample of LMICs migrant workers tend migrant pay gap may produce significantly differing to earn about 17.3 per cent more than nationals. estimates due to their sign (positive or negative) Nevertheless, there are notable variations across and magnitude (or size). countries as described in the sections that follow. Using hourly wages to estimate the migrant pay gap Before looking at the estimates of the migrant pay has the advantage of disentangling working time gap, it is important to understand how wages are from earnings. Conversely, the use of other measures distributed among wage workers in each coun- (monthly, weekly or daily pay) can reflect differences try. Section 3.1 examines the wage structures of not only in hourly pay but also in the number of hours migrant workers and nationals, showing also the worked over a period of time. In practice, though wage distribution by gender. monthly or weekly wages are more frequently available, most survey data from sources such as labour force surveys provide information that enables the derivation of hourly wages. In the sections that fol-

48 Migrant pay gap estimates presented in this report may differ from other sources owing to different ways that migrants are defined and differences in the choice of methodology 49 See Appendix I for description of the methods. Chapter 3 – Measuring and understanding the migrant pay gap 65

low, the migrant pay gap is analysed using hourly migrant workers have earnings in the neighbour- wages (to disentangle working time from earnings), hood of the minimum wage. However, the mean as well as using monthly earnings (to understand the hourly wage of migrant workers in Luxembourg is full extent of inequality of earnings between men and higher than their median hourly wage, and further women migrant workers and nationals). away from the minimum wage. This is because there are few clusters of highly paid migrant work- Figure 16 compares the hourly wage distribution of ers (illustrated by the small peaks in the upper men and women migrant workers to that of nation- ranges, and long tail of the migrant workers’ wage als for a selection of countries.50 With the exception distribution) whose hourly wages are pulling up of few countries, the probability densities of migrant the mean wage for all men and women migrant workers and nationals are bell-shaped51. This implies workers in Luxembourg. Thus, mean and median that, though the mean and median migrant pay migrant pay gaps in Luxembourg, while they have gaps may vary in size, they largely do not differ in the same sign, significantly differ in magnitude (see sign (i.e. both the mean and median migrant pay figures 17–19) because of irregularities in the way gaps have the same sign in almost all the countries in which wage employees are dispersed across the covered in the report). This implication is important range of hourly wages. from a policy standpoint. Whether or not the mean or the median hourly wage is used, the estimates We can also deduce from figure 16 that irregulari- for the migrant pay gap would be consistent, in ties in the probability distributions are more likely terms of sign for most of the countries.52 to occur in the case of migrant workers than in the case of nationals; that is, the peaks and troughs in From an equity standpoint, however, it is still the wage structure are more prevalent for migrant important to understand the magnitude of the workers than for nationals. The reason why this migrant pay gap, which is influenced by whether occurs across the wage distribution, especially for the mean or median wage is used. Using either the migrant workers, is that migrant workers cluster mean or median hourly wage only may underesti- around specific hourly wages. For example, in many mate or overestimate the migrant pay gap in coun- HICs, a clustering of migrant workers occurs around tries where the probability density functions display the minimum wage, which implies that the proba- peaks and troughs. Such differences in magnitude bility of finding migrant workers at the minimum resulting from using the mean or median wage can wage is higher than that of finding nationals (see pose an obstacle in advancing policies towards pay the cases of Cyprus and the Netherlands, for exam- equity between nationals and migrant workers. ple). What the clustering indicates is that migrant This may be illustrated by looking, for example, at workers are concentrated in specific ranges of the case of Luxembourg. In Luxembourg, as in most hourly wages reflecting occupational segregation HICs, the mean (solid vertical line) and median (bro- or “selective” labour market participation. In the ken vertical line) hourly wages of migrant workers case of the sample of LMICs, on the other hand, lie on the left of the mean and median hourly wages most of the clustering occurs at the top end of of nationals (the opposite is true for most of the the wage distribution, where there may be highly sampled LMICs). Luxembourg’s large proportion of skilled “expatriate” workers. migrant workers most likely receive the minimum One problem with estimates of the migrant pay wage which is reflected by the sharp rise of the gap generated by the classic mean and median migrant workers’ wage curve at the lower end of the measures is that they are distorted by these clus- wage distribution. In fact, the median hourly wage tering or composition effects, resulting in estimates of migrant workers in Luxembourg is not far from that vary in magnitude (or size) which makes mon- the tallest peak, which itself is close to the minimum itoring of trends difficult. Thus, the report goes wage, thus suggesting that a large proportion of

50 The hourly wage distribution is presented using the format of probability density function. A probability density function, usually called simply a “density function”, shows how individuals are dispersed across a range of values – in our case, the range of hourly wages. For a detailed description and interpretation of the probability density functions for the hourly wage, see the ILO’s Global Wage Report 2018/19 (ILO, 2018a). 51 The term "bell-shaped" originates from the fact that the graph used to depict a normal distribution consists of a line whose shape resembles that of a bell. The highest point on the curve, or the top of the bell, represents the most probable event in a series of data (containing the greatest number of a value, for example, the mean or the median), while all other possible occurrences are equally distributed around this most probable event (mean or median), creating a downward-sloping line on each side of the peak. This, thus creates a distribution that resembles a bell (hence the name, bell-shaped). The bell-shaped curve is symmetrical, in that, half of the data will fall to the left of the mean or median, and half will fall to the right. 52 Due to the consistency in the sign of the migrant pay gaps, both at the mean wage and at the median wage (as shown in section 3.2) more focus is placed on the mean migrant pay gap, in particular for decomposition and simulation exercises that follow in Chapter 4. 66 The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure 16: The wage structure of migrant workers and nationals, selected countries, latest years

High-income countries

Austria Belgium Croatia

.8 1 1 .8 .8 .6 .6 .6 .4 .4 .4

.2 .2 .2 Probability density Probability density Probability density Probability 0 0 0 –2 –1 0 1 2 3 4 5 6 0 1 2 3 4 5 0 1 2 3 Range of hourly wages Range of hourly wages Range of hourly wages

Cyprus Czech Republic Estonia

.8 1 .6 .8 .5 .6 .4 .6 .4 .3 .4 .2 .2 .2 .1 Probability density Probability density Probability density Probability 0 0 0 0 1 2 3 4 5 0 1 2 3 4 –1 0 1 2 3 4 5 Range of hourly wages Range of hourly wages Range of hourly wages

Finland France Greece

1 1 1

.8 .8 .8

.6 .6 .6

.4 .4 .4

.2 .2 .2 Probability density Probability density Probability density Probability 0 0 0 –1 0 1 2 3 4 5 6 7 –1 0 1 2 3 4 5 6 7 0 1 2 3 4 Range of hourly wages Range of hourly wages Range of hourly wages

Hungary Ireland Italy

1 .8 1

.8 .8 .6 .6 .6 .4 .4 .4 .2 .2 .2 Probability density Probability density Probability density Probability 0 0 0 –1 0 1 2 3 4 5 0 1 2 3 4 5 6 7 0 1 2 3 4 Range of hourly wages Range of hourly wages Range of hourly wages

Latvia Lithuania Luxembourg

.8 1 .8

.8 .6 .6 .6 .4 .4 .4 .2 .2 .2 Probability density Probability density Probability density Probability 0 0 0 –2 –1 0 1 2 3 4 –1 0 1 2 3 4 0 1 2 3 4 5 6 Range of hourly wages Range of hourly wages Range of hourly wages Chapter 3 – Measuring and understanding the migrant pay gap 67

High-income countries (continued)

Malta Netherlands Spain

.8 .8 .6 .5 .6 .6 .4 .4 .4 .3 .2 .2 .2 .1 Probability density Probability density Probability density Probability 0 0 0 0 1 2 3 4 5 0 1 2 3 4 5 6 7 –1 0 1 2 3 4 5 6 Range of hourly wages Range of hourly wages Range of hourly wages

Sweden United Kingdom Iceland

1 1 .8 .8 .8 .6 .6 .6 .4 .4 .4 .2 .2 .2 Probability density Probability density Probability density Probability 0 0 0 –1 0 1 2 3 4 5 6 0 1 2 3 4 5 6 7 –1 0 1 2 3 4 5 Range of hourly wages Range of hourly wages Range of hourly wages

Norway Switzerland Australia

1 .8 1 .8 .8 .6 .6 .6 .4 .4 .4

.2 .2 .2 Probability density Probability density Probability density Probability 0 0 0 –1 0 1 2 3 4 5 6 7 8 –3 –2 –1 0 1 2 3 4 5 –1 0 1 2 3 4 5 Range of hourly wages Range of hourly wages Range of hourly wages

United States Canada Argentina

1 .1 .8 .8 .8 .6 .6 .6 .4 .4 .4 .2 .2 .2 Probability density Probability Probability density Probability density Probability 0 0 0 –9 –8 –7 –6 –5 –4 –3 –2 –1 0 1 2 3 4 0 1 2 3 4 –1 0 1 2 3 4 5 6 7 8 9 Range of hourly wages Range of hourly wages Range of hourly wages

Chile

.4 Migrant workers Non-migrant workers

.2 Probability density Probability 0 0 2 4 6 8 10 12 14 Range of hourly wages

(Figure 16 continued on page 68) 68 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure 16 continued from page 67)

Low-middle-income countries

Turkey Jordan Bangladesh

1.5 1.2 1 1 .8 .8 1 .6 .6 .4 .5 .4 .2 .2 Probability density Probability density Probability density Probability 0 0 0 –2 –1 0 1 2 3 4 –2 –1 0 1 2 –6 –5 –4 –3 –2 –1 0 1 2 3 4 5 6 Range of hourly wages Range of hourly wages Range of hourly wages

Nepal Madagascar Malawi

.5 .5 .6 .4 .4 .4 .3 .3 .2 .2 .2 .1 .1 Probability density Probability density Probability density Probability 0 0 0 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 8 9 –2 –1 0 1 2 3 4 5 6 7 Range of hourly wages Range of hourly wages Range of hourly wages

Namibia Sierra Leone Tanzania

4 4 4

3 2 3

2 2 2

1 1 1 Probability density Probability density Probability density Probability 0 0 0 –6 –5 –4 –3 –2 –1 0 1 2 3 4 5 6 0 1 2 3 4 5 6 7 8 9 10 11 12 13 0 1 2 3 4 5 6 7 8 9 Range of hourly wages Range of hourly wages Range of hourly wages

Bolivia Mexico

8 8

6 6

4 4 Migrant workers Non-migrant workers 2 2 Probability density Probability density Probability 0 0 –1 0 1 2 3 4 5 6 –3 –2 –1 0 1 2 3 4 5 6 Range of hourly wages Range of hourly wages

Note: The green curve represents the hourly wage distribution of men and women migrant wage workers whereas the blue represents that of nationals. The solid vertical line in each of the charts indicates the mean hourly wage, whereas the broken vertical line indicates the median hourly wage. Figure A-3 in Appendix IV replicates this figure in terms of men and women separately, distinguishing between migrant workers and nationals. Source: ILO estimates based on survey data provided by national sources (see Appendix II). Chapter 3 – Measuring and understanding the migrant pay gap 69

beyond the classic mean and median migrant pay well as monthly earnings of migrant wage work- gaps to provide complementary estimates that ers and nationals. They include an overall average consider these composition effects by estimating migrant pay gap by income group based on weights what is called the factor-weighted migrant pay gap, that account for the population size (specifically, of and the pay gap across the entire wage distribution wage workers) of each studied country. of wage workers. An observation arising from these figures is that Similar patterns in the wage structures are exhib- regardless of whether hourly wages or monthly ited by looking at the wage structures of men and earnings is used, the migrant pay gap is signifi- women separately, distinguishing between migrant cantly positive in most HICs, implying that nation- workers and nationals. This is displayed in figure als in HICs earn much more than migrant workers A-3 (see Appendix IV). For a majority of the sample on average. It is negative in most LMICs, implying of countries, the mean hourly wage of migrant that migrant workers tend to earn more in LMICs women workers is located at the left of the mean than nationals on average. However, it is important hourly wage of the three other groups, indicating to note that these LMICs are mainly countries of that migrant women wage workers appear to be origin (mostly with positive net emigration), with a the most disadvantaged in terms of pay. few exceptions such as Costa Rica, Gambia, Jordan, Namibia, and Turkey. 3.2. The raw migrant pay gap Considering the mean migrant pay gap based on hourly wages, 26 of the 33 HICs covered in the The raw migrant pay gap, based on mean hourly report have a positive mean migrant pay gap, imply- wages of migrant workers and nationals, is esti- ing that nationals tend to earn more than migrant mated to be between 42.1 per cent in Cyprus (which workers in these countries. The same is true for 27 is a 7.3 percentage points increase from the esti- of the 33 HICs when looking at the median hourly mated gap in 2010 (34.8 per cent) according to the migrant pay gap (figure 17). However, among the ILO Global Wage Report 2014/15) and –12.6 per cent 16 sampled LMICs, only four countries (Costa Rica, in Slovakia in the sample of HICs, with a weighted Jordan, Nepal, and Turkey) have positive mean average of 12.6 per cent; and between 30.1 per cent hourly migrant pay gap estimates and only six in Costa Rica and –136.7 per cent in Madagascar in (Bulgaria, Costa Rica, Jordan, Namibia, Nepal, and the sample of LMICs, with a weighted average of Turkey) have positive median hourly migrant pay -17.3 per cent. gap estimates (figure 17). The raw migrant pay gap simply refers to the dif- In the case of monthly earnings in figure 18, the ference in pay between migrant wage workers prevalence of positive migrant pay gap is more or and nationals at a specific point in time. The gap is less the same as using hourly wages with 26 of the usually calculated as the margin by which migrant 33 HICs recording positive mean monthly migrant workers’ pay falls short of the pay of nationals. For pay gap, and 28 of the 33 HICs recording positive example, if migrant workers’ earn 85 per cent of median monthly migrant pay gap. Only three (Costa nationals’ earnings on average, then it is said that Rica, Jordan, and Turkey) of the 16 LMICs report pos- the migrant pay gap is 15 per cent. This part of itive mean monthly migrant pay gap while only five the report begins by showing the migrant pay gap (Bulgaria, Costa Rica, Jordan, Namibia, and Turkey) using the two common measures of summariz- have positive median monthly migrant pay gap ing pay (mean and median wages) to provide an when estimation is based on monthly earnings. insightful starting point for a detailed analysis and While estimates of the migrant pay gap is positive in understanding of pay differentials between migrant most HICs, it is negative in only few countries such wage workers and nationals. as in Australia (a finding consistent with estimate Figures 17 and 18 show estimates of the migrant from Chiswick, Le and Miller (2008))53 and Slovakia, pay gap for all the 49 sampled countries. Each with figures providing strong evidence of an over- of these figures presents estimates of mean and all pay gap in favour of nationals in HICs. On the median migrant pay gap, using hourly wages as other hand, the figures provide strong evidence of

53 As shown in Section 3.4, the raw migrant pay gap for Australia suffers a lot from composition effect, which when accounted for in the estimation, turns the negative estimate to a positive pay gap in favour of nationals. 70 The migrant pay gap: Understanding wage differences between migrants and nationals

an overall pay gap in favour of migrant workers in However, differences in magnitude (or size) alone LMICs, with a few exceptions including in Bulgaria, can become an obstacle in advancing policies Costa Rica, Jordan, Namibia, Nepal, and Turkey. towards pay equity between nationals and migrant Again, this trend in favour of migrant workers, on workers. Apart from presenting gender dimensions average, in the sample of LMICs may reflect the of the migrant pay gap as well as comparing the likelihood of clusters of highly skilled “expatriate” pay gap between informal and formal workers (sec- workers whose foreign-acquired education and tion 3.3), the analysis goes beyond the mean and experience may attract higher returns relative to the median to provide complementary estimates that of the nationals. that consider composition effects in estimating the migrant pay gap (called the factor-weighted The weighted averages in the sample of HICs migrant pay gap) in section 3.4, and to provide range from about 12.6 per cent (in the case of migrant pay gaps across the entire wage distribu- mean hourly wages) to 18.4 per cent (in the case of tion in section 3.5. median monthly earnings) in favour of nationals, and from about 8.6 per cent (in the case of mean hourly wages) to 16.8 per cent (in the case median 3.3. The migrant pay gap monthly earnings) in favour of nationals across the Member States of the EU. In the sample of LMICs, in different subgroups however, the weighted averages range from an estimated –7.5 per cent (based on median hourly 3.3.1. Migrant women tend to pay wages) to –19.1 per cent (based on mean monthly a double penalty earnings). However, there are wide variations among countries within income groups, with the Results from this section shows that migrant mean migrant pay gap, for example, ranging from women, particularly in HICs tend to pay a dou- 42.1 per cent in Cyprus to –12.5 per cent in Slovakia ble penalty for being both women and migrants in the sample of HICs and from 30.1 per cent in as compared to the average migrant worker, a Costa Rica to –136.7 per cent in Madagascar in the finding consistent with results from the OECD's sample of LMICs. International Migration Outlook 2020 (OECD, 2020b). Specifically, migrant women earn less than Another observation is that, although the mean migrant men (who in turn earn less than non-mi- and median estimates can generate different grant workers) in the sample of HICs. They also results in terms of magnitude (or size), they do earn less than non-migrant women and even far not differ much in terms of sign. This means that less than non-migrant men in the sample of HICs. whether mean and median hourly wages, or mean For example, the pay gap (based on mean hourly and median monthly earnings are compared, the wages) between non-migrant men and migrant difference in terms of sign is negligible. Figure 19 women in the sample of 33 HICs covered in this illustrates this by showing the mean and median report is estimated at around 20.9 per cent, which migrant pay gaps based on hourly wages in one is much wider than the estimated global aggregate chart and based on monthly earnings in another gender pay gap (based on mean hourly wages) of chart. These chats show similar trends in terms of 16.2 per cent among HICs. the direction of the estimates, with only few exceptions. The signs of both the mean and median migrant pay gaps are the same for all but only Migrant men and migrant women three countries (Bulgaria, Namibia and Poland)54 in HICs earn less than their non-migrant when based on hourly wages and four countries counterparts (Bulgaria, Namibia, Poland and Slovakia)55 when Figures 17 and 18 present estimates of the mean based on monthly earnings, implying that both and median migrant pay gaps at the aggregate (i.e. measures largely provide consistent estimates in for men and women combined) for the 49 studied terms of direction of the pay gap.

54 The migrant pay gap based on hourly wages switches sign from negative to positive only in Bulgaria (from –35.7 to 5.2 per cent), Namibia (from –28.0 to 34.1 per cent) and Poland (–4.9 to 6.2 per cent) when the pay gap is based on median hourly wages rather than mean hourly wages. The remaining 46 countries retain the same sign regardless of whether the migrant pay gap is based on mean or median hourly wages. 55 The migrant pay gap switches sign from –29.5 to 5.2 per cent in Bulgaria, from –32.0 to 44.4 per cent in Namibia, from –8.3 to 12.8 per cent in Poland, and from –11.2 to 1.5 per cent in Slovakia when it is estimated based on median monthly earnings rather than mean monthly earnings. The remaining 45 countries retain the same sign regardless of whether mean or median monthly earnings are used. Chapter 3 – Measuring and understanding the migrant pay gap 71

X Figure 17: Migrant pay gaps using hourly wages, latest years

Mean hourly migrant pay gap Median hourly migrant pay gap

Cyprus 42.1 Cyprus 42.9 Slovenia 33.3 Luxembourg 36.3 Italy 29.6 Slovenia 32.7 Portugal 28.9 Greece 31.7 Spain 28.3 Spain 31.1 Luxembourg 27.3 Italy 27.1 Austria 25.3 Netherlands 25.1 Greece 21.2 Austria 24.9 Estonia 21.0 Portugal 23.8 Ireland 20.6 Ireland 23.0 Netherlands 19.9 Iceland 22.5 Argentina 18.1 Estonia 20.1 Iceland 17.8 Hungary 19.5 Denmark 17.3 United States 17.6 United States 15.3 Argentina 17.0 Latvia 15.1 Latvia 15.3 Norway 15.0 Denmark 15.1 Belgium 12.7 Belgium 13.8 Finland 10.7 Norway 13.8

High-income Sweden 10.5 High-income France 12.9 France 8.7 Finland 11.5 Chile 6.4 United Kingdom 9.7 Hungary 5.3 Canada 7.1 Canada 4.0 Poland 6.2 United Kingdom 2.7 Switzerland 5.7 Switzerland 0.3 Sweden 5.4 Malta –2.7 Chile 3.7 Poland –4.9 Malta –2.9 Croatia –6.2 Australia –5.0 Lithuania –6.8 Czech Republic –10.9 Australia –6.9 Lithuania –21.7 Czech Republic –8.1 Slovakia –26.4 Slovakia –12.5 Croatia –29.4 EU 8.6 EU 14.1 High-income 12.6 High-income 16.1 Costa Rica 30.1 Namibia 34.1 Jordan 29.5 Jordan 33.3 Turkey 10.0 Costa Rica 17.1 Nepal 1.5 Nepal 9.9 Albania –6.2 Turkey 8.5 Tanzania** –15.2 Bulgaria 5.2 Mexico –17.2 Bangladesh –5.2 Malawi –23.1 Malawi –6.4 Bangladesh –26.3 Gambia –8.8 Bolivia* –28.0 Mexico –12.5 Namibia –28.0 Albania –13.4 Bulgaria –35.7 Romania –21.2 w- and middle income w- and middle income Gambia –39.9 Tanzania** –22.4 Lo Lo Sierra Leone –44.8 Bolivia* –40.4 Romania –53.6 Sierra Leone –50.5 Madagascar –136.7 Madagascar –115.0 Low- and middle income –17.3 Low- and middle income –7.5 0 0 20 40 60 20 40 60 –8 0 –6 0 –4 0 –2 0 –80 –60 –40 –20 –160 –140 –120 –100 –140 –120 –100 Percent Percent

Migrant men Men nationals Migrant women Women nationals

Note: EU, high-income, and low- and middle-income estimates are the averages of the European Union, the sample of high-income countries, and low- and middle-income countries, respectively. Averages are weighted by the number of wage employees in each country. * the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). 72 The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure 18: Migrant pay gaps using monthly earnings, latest years

Mean monthly migrant pay gap Median monthly migrant pay gap

Cyprus 41.0 Cyprus 41.0 Spain 35.5 Luxembourg 39.7 Slovenia 32.0 Spain 34.7 Italy 31.8 Slovenia 32.2 Portugal 30.2 Italy 31.5 Greece 30.1 Greece 31.5 Austria 26.5 Austria 29.0 Luxembourg 24.3 Ireland 25.4 Iceland 23.3 Hungary 24.6 Estonia 20.6 Portugal 23.8 France 19.2 United States 22.6 United States 17.3 Estonia 22.5 Ireland 17.0 France 20.9 Denmark 16.9 Netherlands 19.6 Argentina 16.8 Argentina 19.2 Netherlands 15.2 Argentina 18.8 Latvia 14.2 Iceland 17.6 Sweden 13.4 Latvia 16.7 Norway 11.5 Belgium 15.1

High-income Belgium 9.1 High-income Sweden 12.9 United Kingdom 6.9 Poland 12.8 Finland 6.5 Norway 12.8 Canada 3.0 Finland 10.8 Hungary 2.0 United Kingdom 9.8 Chile 1.2 Canada 7.7 Switzerland 0.3 Switzerland 3.1 Malta –6.4 Chile 2.5 Lithuania –6.5 Slovakia 1.5 Australia –7.6 Malta –0.6 Poland –8.3 Australia –12.7 Czech Republic –10.0 Czech Republic –13.2 Croatia –10.9 Lithuania –27.2 Slovakia –11.2 Croatia –29.3 EU 11.6 EU 16.8 High-income 14.5 High-income 18.4 Costa Rica 29.6 Namibia 44.4 Jordan 25.1 Jordan 37.5 Turkey 8.3 Costa Rica 14.0 Albania –2.4 Turkey 6.3 Nepal –12.7 Bulgaria 5.2 Bolivia* –15.8 Albania 0.0 Mexico –17.7 Mexico –9.5 Malawi –24.0 Nepal –11.1 Tanzania** –26.3 Bangladesh –11.1 Bulgaria –29.5 Bolivia* –17.0 Bangladesh –29.7 Tanzania** –19.0 Gambia –30.4 Romania –20.5 w- and middle income w- and middle income Namibia –32.0 Malawi –36.9 Lo Lo Romania –51.4 Gambia –47.2 Sierra Leone –71.6 Sierra Leone –55.6 Madagascar –153.2 Madagascar –131.6 Low- and middle income –19.1 Low- and middle income –9.1 0 0 50 20 40 60 –50 –8 0 –6 0 –4 0 –2 0 100 –200 –150 –100 –160 –140 –120 –100 Percent Percent

Migrant men Men nationals Migrant women Women nationals

Note: EU, high-income, and low- and middle-income estimates are the averages of the European Union, the sample of high-income countries, and low- and middle-income countries, respectively. Averages are weighted by the number of wage employees in each country. * the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). Chapter 3 – Measuring and understanding the migrant pay gap 73

X Figure 19: Comparing the mean and median migrant pay gaps

Mean and median migrant pay gaps Mean and median migrant pay gaps using hourly wages using monthly earnings Cyprus Slovakia Slovenia Croatia Italy Czech Republic Portugal Poland Spain Australia Luxembourg Lithuania Austria Malta Greece Switzerland Estonia Chile Ireland Hungary Netherlands Canada Argentina Finland Iceland United Kingdom Denmark Belgium United States Norway Latvia Sweden Norway Latvia Belgium Netherlands Finland Argentina

High-income Sweden High-income Denmark France Ireland Chile United States Hungary France Canada Estonia United Kingdom Iceland Switzerland Luxembourg Malta Austria Poland Greece Croatia Portugal Lithuania Italy Australia Slovenia Czech Republic Spain Slovakia Cyprus EU EU High-income High-income Costa Rica Madagascar Jordan Sierra Leone Turkey Romania Nepal Namibia Albania Gambia Tanzania** Bangladesh Mexico Bulgaria Malawi Tanzania** Bangladesh Malawi Bolivia* Mexico Namibia Bolivia* Bulgaria Nepal w- and middle income w- and middle income Gambia Albania Lo Lo Sierra Leone Turkey Romania Jordan Madagascar Costa Rica Low- and middle-income Low- and middle-income 0 0 50 20 40 60 –5 0 –80 –60 –40 –20 10 0 –200 –150 –100 –160 –140 –120 –100 Percent Percent

Mean migrant pay gap Median migrant pay gap

variations across countries. For example, in the case of Jordan, migrant men tend to earn 26.8 per cent less than non-migrant men, whereas in Namibia, migrant men earn about 29.2 per cent more than non-migrant men. In the case of the pay gap between non-migrant women and migrant women, the estimates change sign in nine countries (in reference to the overall migrant pay gap in figure 17), switching from negative to positive in Croatia, Lithuania, Madagascar, Malta, the United Republic of Tanzania, and Sierra Leone; and from positive to negative in Switzerland, Turkey and the United Kingdom. The weighted © aauw.org hourly pay gap between women nationals and migrant women is 12.4 per cent in the sample of HICs in favour of women nationals, 0.5 per cent in countries. However, figures 3 and 12, and table 7, the EU, and –21.7 per cent in the sample of LMICs, indicate that men and women differ significantly in with notable variations across countries. For exam- terms of their labour market participation, work- ple, in the sample of HICs, the pay gap between ing time status, and the incidence of informality. non-migrant women and migrant women ranges Women on average participate less than men in between –25.0 per cent in Slovakia to 44.4 per cent the labour market, and are more likely than men in Cyprus. In the case of Spain, the pay gap between to work in the informal economy and on part-time. non-migrant women and migrant women is esti- Therefore, figure 20 presents estimates of mean mated at 26.3 per cent, which is a 7.3 percentage hourly migrant pay gap, separating the sample to point increase in the estimated gap in 2006 (19.0 per distinguish between women and men, with the first cent) (Antón, de Bustillo and Carrera, 2012). column comparing migrant men and non-migrant men, and the second column comparing migrant Lower wages of migrant women relative to non-mi- women and non-migrant women. grant women is a clear indication that pay gaps are higher for migrant women than for women in gen- The pay gap between non-migrant men and eral who already earn less than men as reported in migrant men in figure 20 mimics the aggregate raw the ILO Global Wage Report 2018/19 (ILO, 2018a). migrant pay gap estimates in figure 17. The sign of the estimates stay the same for almost all the Migrant women earn less than migrant countries. With regards to size, the hourly pay gap between men nationals and migrant men tend to men in a large number of countries be wider in the sample of HICs and the EU than the Turning to the pay gap between men and women aggregate pay gap estimates. The weighted hourly migrant workers, figure 21 shows this and how the pay gap between men nationals and migrant men estimates compare to the aggregate mean gender is approximately 14.3 per cent in the sample of HICs pay gap at the country level.56 The figure shows that and 14.7 per cent in the EU (figure 20), compared the two types of gender pay gap exhibit a mixed to the aggregate estimates of 12.6 per cent in HICs pattern. In 17 of the 49 countries (including, among and 8.6 per cent in the EU (figure 17), with variations others, Australia, Canada, Estonia, Iceland, and the across countries. For example, while migrant men United States in the sample of HICs and Jordan, earn 38.3 per cent less than non-migrant men in Mexico, Sierra Leone, and the United Republic of Cyprus, they earn 1.3 per cent less than non-mi- Tanzania in the sample of LMICs), the gender pay grant men in Switzerland, and 9.8 per cent more gap between migrant men and migrant women is than non-migrant men in Australia. In the sample much higher than the aggregate gender pay gap of LMICs, on the other hand, the weighted aver- at the country level, reinforcing the double penalty age migrant pay gap between men nationals and hypothesis for migrant women in these countries. migrant men is estimated as –14.3 per cent, with For example, in the United States, the gender pay

56 Estimates of the aggregate gender pay gap at the country level are retrieved from the ILO’s Global Wage Report 2018/19. Chapter 3 – Measuring and understanding the migrant pay gap 75

X Figure 20: Mean migrant pay gaps using hourly wages by sex, latest years

Migrant men and non-migrant men Migrant women and non-migrant women

Cyprus 38.3 Cyprus 44.4 Slovenia 35.3 Estonia 33.8 Italy 31.7 Slovenia 30.3 Portugal 30.5 Luxembourg 29.0 Spain 30.4 Italy 27.5 Luxembourg 25.4 Austria 26.8 Austria 24.0 Portugal 26.5 Greece 21.8 Spain 26.3 Netherlands 20.2 Latvia 26.2 Argentina 18.3 Ireland 26.0 Denmark 17.7 Iceland 22.5 United Kingdom 16.7 Netherlands 20.8 Estonia 16.4 Greece 20.3 United States 15.7 United States 18.8 Norway 15.4 Lithuania 18.7 Ireland 14.5 Argentina 17.8 France 14.2 Denmark 16.8 Iceland 13.1 Belgium 15.1 Finland 10.2 Sweden 15.1

High-income Belgium 10.0 High-income Norway 14.8 Latvia 7.6 Finland 12.4 Sweden 7.5 Chile 6.6 Hungary 6.4 Canada 6.3 Chile 5.8 Malta 5.2 Canada 1.9 Hungary 4.3 Switzerland 1.3 Croatia 3.0 Poland –1.4 France 2.4 Slovakia –3.6 Switzerland –0.9 Czech Republic –8.6 Australia –2.8 Malta –9.6 Czech Republic –4.9 Australia –9.8 United Kingdom –12.7 Croatia –16.0 Poland –12.7 Lithuania –17.0 Slovakia –25.0 EU 14.7 EU 0.5 High-income 14.3 High-income 12.4 Costa Rica 27.4 Sierra Leone 67.0 Jordan 26.8 Jordan 44.1 Turkey 17.1 Costa Rica 34.1 Romania 6.6 Madagascar 25.9 Nepal 5.5 Tanzania** 20.4 Albania –5.3 Nepal 1.3 Malawi –13.8 Turkey –5.6 Bolivia* –15.6 Mexico –7.1 Gambia –16.5 Albania –7.9 Mexico –22.6 Bulgaria –25.8 Bangladesh –24.0 Namibia –28.6 Namibia –29.2 Bangladesh –34.1 w- and middle income w- and middle income Tanzania** –31.5 Malawi –40.9 Lo Lo Bulgaria –57.7 Bolivia* –48.6 Sierra Leone –66.9 Gambia –108.4 Madagascar –153.7 Romania –145.6 Low- and middle-income –14.3 Low- and middle-income –21.7 0 0 20 40 60 80 20 40 60 –8 0 –6 0 –4 0 –2 0

–80 –60 –40 –20 100 –180 –160 –140 –120 –100 –180 –160 –140 –120 –100 Percent Percent

Note: EU, high-income, and low- and middle-income estimates are the averages of the European Union, the sample of high-income countries, and low- and middle-income countries, respectively. Averages are weighted by the number of wage employees in each country. * the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). 76 The migrant pay gap: Understanding wage differences between migrants and nationals

A digital economy women migrant worker in Buenos Aires. © Copyright ILO Argentina

gap between migrant men and migrant women is Jordan, Madagascar, Sierra Leone, and the United about 20.1 per cent, compared to the aggregate Republic of Tanzania). The gender pay gap between gender pay gap (between all men and all women) men nationals and women migrant workers in the of 16.0 per cent. other 11 LMICs is in favour of women migrants, with an overall weighted average of about –15.1 per Migrant women earn far less than cent compared to the average aggregate gender pay gap of 2.4 per cent in these countries. non-migrant men in HICs

To further explore the double penalty hypothesis 3.3.2. The migrant pay gap in the for migrant women, figure 22 shows estimates of formal and informal economies the mean hourly gender pay gap between men nationals and migrant women and compares this Figure 23 presents estimates of the mean and gap to the aggregate gender pay gap at the coun- median hourly migrant pay gaps for 14 of the try level. As expected, the mean hourly pay gap 49 studied countries for which informality estimates between men nationals and migrant women in HICs are available. These 14 countries host roughly is positive everywhere (except in Poland, Slovakia, 5.3 per cent of international migrants and just and the United Kingdom) and exceeds the gender 3.0 per cent of migrant workers worldwide. In the pay gap at the country level in most of the sampled two HICs with available data, Argentina and Chile, 33 HICs, reinforcing the double penalty hypothesis both the mean and median migrant pay gaps are for migrant women. The gender pay gap between positive in the formal economy but negative in the men nationals and women migrant workers in the informal economy. sample of HICs is about 20.9 per cent compared to the aggregate gender pay gap of 16.2 per cent in In the case of LMICs, the results show a mixed pat- this sample, with variations across countries. tern. While the mean migrant pay gap is positive in Bangladesh, the Plurinational State of Bolivia, The trend is different among the sample of LMICs Costa Rica, and Malawi in the formal economy, it is where the mean hourly gender pay gap between positive only in Costa Rica, Namibia, Nepal, and the men nationals and migrant women exceeds the United Republic of Tanzania in the informal econ- overall country-level gender pay gap in only five omy. The median migrant pay gap in LMICs shows of the 16 LMICs covered in the report (Costa Rica, a similar pattern. Chapter 3 – Measuring and understanding the migrant pay gap 77

X Figure 21: The gender pay gap between men and X Figure 22: The gender pay gap between women migrant workers and how it compares to the non-migrant men and migrant women and how this aggregate gender pay gap in the economy, using compares to the aggregate gender pay gap in the mean hourly wages, latest year economy, using the mean hourly wage, latest years

United Kingdom Poland Luxembourg Slovakia Poland United Kingdom Netherlands Malta Spain Belgium Italy Hungary France France Belgium Croatia Slovakia Australia Argentina Netherlands Slovenia Luxembourg Greece Czech Republic Hungary Finland Denmark Argentina Ireland Canada Austria Switzerland Portugal Ireland Finland Chile Norway Sweden High-income High-income Malta Norway Cyprus Denmark Sweden Greece Chile Spain Canada Italy Switzerland Lithuania Australia Austria Iceland Iceland United States Latvia Croatia United States Czech Republic Slovenia Latvia Portugal Estonia Estonia Lithuania Cyprus High-income High-income Romania Romania Gambia Gambia Turkey Namibia Malawi Malawi Namibia Bangladesh Albania Turkey Bangladesh Albania Costa Rica Bulgaria Mexico Mexico Nepal Nepal Jordan Tanzania** w- and middle income w- and middle income Bulgaria Costa Rica Lo Lo Tanzania** Madagascar Madagascar Jordan Sierra Leone Sierra Leone Low- and middle-income Low- and middle-income 0 0 50 20 40 60 80 –5 0 10 0 –80 –60 –40 –20 –150 –100 –140 –120 –100 Percent Percent

Gender pay gap between migrant men and migrant women Gender pay gap between non-migrant men and migrant women

Aggregate gender pay gap at the country level Aggregate gender pay gap at the country level

Note: Aggregate gender pay gaps in the economy are retrieved Note: Aggregate gender pay gaps in the economy are retrieved from the ILO Global Wage Report 2018/19. High-income and low- from the ILO Global Wage Report 2018/19. High-income and low- and middle-income estimates are the averages of the sample of and middle-income estimates are the averages of the sample of high-income countries, and low- and middle-income countries, high-income countries, and low- and middle-income countries, respectively. Averages are weighted by the number of wage respectively. Averages are weighted by the number of wage employees in each country. ** the United Republic of Tanzania. employees in each country. ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by na- Source: ILO estimates based on survey data provided by national sources (see Appendix II). tional sources (see Appendix II). 78 The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure 23: Migrant pay gaps in the informal and formal economies, using the mean and median hourly wage, latest years

Formal economy Informal economy

Mean migrant pay gap Mean migrant pay gap

Argentina 17.6 Argentina –7.3

High- Chile High- Chile income 8.9 income –16.9

Costa Rica 30.0 Tanzania** 7.9

Malawi 22.7 Nepal 4.5

Bolivia* 10.8 Costa Rica 4.4

Bangladesh 9.1 Namibia 1.1

Albania –8.1 Turkey –4.9

Tanzania** –9.7 Albania –16.3

Turkey –11.2 Mexico –22.5

Mexico –36.9 Gambia –30.4

w- and middle income Nepal –50.6 w- and middle income Bangladesh –34.5 Lo Lo Namibia –57.5 Madagascar –40.6

Gambia –62.7 Malawi –40.8

Madagascar –146.0 Bolivia* –62.3

–20

–40

–60

–80

–10

–20

–30

–40

–50

–60

–70

–100

–120

–140

–160 –180 Percent Percent

Median migrant pay gap Median migrant pay gap

Argentina 16.8 Argentina –7.3

High- Chile High- Chile income 6.9 income –19.3

Costa Rica 18.5 Namibia 38.5

Bolivia* 12.5 Madagascar 18.0

Malawi 10.0 Tanzania** 6.5

Bangladesh 2.5 Costa Rica 2.8

Albania –4.4 Nepal –2.6

Turkey –11.1 Malawi –5.0

Tanzania** –17.2 Mexico –6.9

Gambia –31.9 Bangladesh –9.4

w- and middle income Mexico –58.3 w- and middle income Gambia –14.3 Lo Lo Namibia –107.4 Turkey –27.3

Nepal –118.8 Albania –31.5

Madagascar –207.0 Bolivia* –72.9

–20

–40

–60

–80

–50

–100

–150

–200 –250

Percent Percent

Note: The informal economy refers to the set of all economic activities by workers and economic units that are – in law and practice – not covered or insufficiently covered by formal arrangements. * the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). Chapter 3 – Measuring and understanding the migrant pay gap 79

The ILO Decent Work across Borders, a project funded by the European Union on migrant health workers and skilled professionals, launched a photo contest in 2013. The photo contest, in partnership with the Alliance of Young Nurse Leaders and Advocates (AYNLA) captured images related to migration from the perspective of young health professionals. © Copyright ILO

3.3.3. The migrant pay gap is wider tries, the mean pay gap within the care economy is in the care economy than similar to the overall pay gap for a number of countries (for example, Iceland, Portugal, Spain, and the overall migrant pay gap Sweden). In the case of Sweden, the overall mean migrant pay gap and the mean pay gap in the care This section compares estimates of the migrant economy stand at 10.5 per cent and 10.4 per cent, pay gap in the care economy with the overall respectively. Also, in countries like Finland, France, migrant pay gap. Using mean and median hourly Norway, Slovenia, Turkey and the United Kingdom, wages, migrant care workers in most countries migrant care workers tend to be better off than appear to have a higher wage penalty on average the average migrant worker. That is, the migrant than migrant workers in general (figure 24). For pay gap in the care economy is either negative or example, in Cyprus, migrant care workers receive lower than the overall migrant pay gap in these 25 percentage points less wages than the average countries. In the case of the United Kingdom, for migrant worker (67.2 per cent mean migrant pay example, while the overall mean migrant pay gap gap in the care economy compared to 42.1 per is estimated at 2.7 per cent, the migrant pay gap in cent overall mean migrant pay gap). In Malta, while the care economy is about –14.2 per cent. both the mean and median overall migrant pay gaps are negative (–2.7 per cent and –2.9 per cent, In sum, the mean migrant pay gap in the care econ- respectively), i.e., migrant workers earn more than omy is estimated at approximately 19.6 per cent com- non-migrant workers, the mean and median pay pared to the overall mean migrant pay gap of about gaps in the care economy are positive and wider in 17.1 per cent in the countries for which estimates are favour of non-migrant care workers (19.0 per cent available for the care economy. The corresponding and 13.5 per cent, respectively). weighted estimate of the median migrant pay gap in On the other hand, while migrant care workers tend the care economy is 22.5 per cent compared to about to pay a higher penalty in most of the studied coun- 18.6 per cent in the overall economy. 80 The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure 24: A comparison between the aggregate migrant pay gap and the migrant pay gap in the care economy, latest years

Using mean hourly wages Using median hourly wages

United Kingdom Slovenia

Slovenia Sweden

France Switzerland

Australia United Kingdom

Turkey Australia

Switzerland France

Canada Norway

Chile Turkey

Finland Canada

Sweden Finland

Norway Chile

Iceland Malta

Malta Denmark

United States Argentina

Denmark Belgium

Belgium United States

Estonia Iceland

Greece Latvia

Argentina Austria

Latvia Estonia

Netherlands Portugal

Spain Netherlands

Austria Costa Rica

Portugal Greece

Ireland Spain

Italy Luxembourg

Luxembourg Ireland

Costa Rica Italy

Jordan Jordan

Cyprus Cyprus

Weighted average Weigted average

–20 –10 01020304050607080 –40–20 020406080100 Percent Percent

Mean migrant pay gap in the care economy Median migrant pay gap in the care economy

Overall mean migrant pay gap Overall median migrant pay gap

Note: The care workforce includes care workers in care sectors (education, health and social work), care workers in other sectors, domestic workers and non-care workers in care sectors, who support care service provision. Source: ILO estimates based on survey data provided by national sources (see Appendix II). Chapter 3 – Measuring and understanding the migrant pay gap 81

X Box 3. Migrant workers have been significantly affected by the COVID-19 crisis

Survey data from Mexico and the United States that covers up to the third quarter of 2020, shows that at the onset of the pandemic, migrant and non-migrant workers in the two countries lost jobs, while among those that remained employed – both migrants and non-migrants – there was a decline in aggregate number of hours worked. Most of those that lost their jobs are low paid workers, which means that the average earnings of those that remained in paid employment during the pandemic is likely to be artificially higher compared to similar estimates at pre-pandemic periods. It is for this reason that comparing the migrant pay gap before and during the pandemic may not bring about much useful policy instrument. However, as an indication, one can compare trends in the aggregate real monthly earnings between the two groups. In the case of Mexico, the ratio of aggregate real monthly earnings between migrant and non-migrant workers before the onset of the pandemic was 0.75 per cent – i.e., migrant workers, who by then were about one per cent of the employed population, earned 0.75 per cent of real monthly earnings, in aggregate, compared to non-migrants. At the onset of the pandemic in April 2020 the ratio dropped drastically to 0.55 per cent, and although it had recovered gradually over the next few months, the ratio stood at 0.72 per cent by October 2020. It is important to point out that in Octo- ber 2020, the data from Mexico shows an increase of seven per cent more migrant workers compared to the same period in 2019 – while there are fewer non-migrant workers in employment as a consequence of the health crisis. Despite the increase in the number of migrant workers, the ratio between migrants’ and non-migrants’ aggregate real monthly earnings is lower at 0.72 per cent, thus showing that migrant workers in Mexico received less earnings mass – and less per worker – compared to that which they received in 2019. In the case of the United States, on the other hand, where migrants (including naturalized migrants) approximated about 18 per cent of workers in the population in pre-pandemic times, the ratio of total earnings between migrant and non-migrant workers dropped from 21.5 per cent to 18.1 per cent between March and April 2020. The ratio did recover somehow to reach 20 per cent by October 2020, although there were 8 per cent less migrants working in the United States by October 2020 compared to those observed in March 2020. This probably indicates that migrants that lost their jobs are low paid migrants since the total decline in aggregate real earnings would be greater if the number of migrant workers that lost their jobs (8 per cent) were spread across different locations of the wage distribution. Thus, in the United States the results suggest that the greater loss for migrant workers was in terms of employment loss – compared to the estimates for Mexico which shows that the loss was in terms of average earnings among those that remained in employment.

Notes: The results are preliminary and potentially subject to modification in a future independent ILO working paper.

3.3.4 Migrant workers have been Mexico and the United States both before and after among the hardest hit by the the onset of the COVID-19 pandemic. economic downturn associated with the COVID-19 pandemic 3.4. A complementary

Few recent data that covers the COVID-19 pandemic measure: Factor-weighted period shows that migrant workers, likewise nation- migrant pay gap als, have suffered losses in the labour market, including employment losses, a decline in the aggregate To account for composition effects in estimating number of hours worked and a decline in average the migrant pay gap, the report follows the meth- earnings for those that remained in employment by ods used in the ILO Global Wage Report 2018/19 October 2020 compared to pre-pandemic periods. to generate a factor-weighted migrant pay gap57. Box 3 presents the situation of migrant workers in The factor-weighted migrant pay gap removes a

57 It is important to emphasize that the factor-weighted migrant pay gap is not equivalent to an estimate of adjusted (unexplained) migrant pay gap: the latter requires the use of other techniques, for example the identification of a counterfactual wage distribution, in order to identify and exclude the part of the gap arising from differences in labour market endowments of migrant workers and nationals. This issue is addressed in the next section (in particular, see section 3.5). 82 The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure 25: Factor-weighted migrant pay gaps using hourly wages, latest years

Factor-weighted mean migrant pay gap Factor-weighted median migrant pay gap

Cyprus 28.1 Cyprus 30.5 Estonia 23.7 Spain 23.4 Italy 20.3 Italy 19.1 Spain 18.6 Austria 17.8 Austria 18.1 Estonia 16.8 Portugal 17.4 Luxembourg 15.1 Netherlands 14.0 Canada 13.7 United Kingdom 11.8 Iceland 12.2 United States 11.6 Argentina 10.5 Denmark 11.5 United States 10.3 Canada 11.0 Denmark 10.0 Luxembourg 10.9 Portugal 8.7 Iceland 10.6 Belgium 8.6 Latvia 10.2 Switzerland 7.8 Argentina 10.2 United Kingdom 7.6 Norway 9.6 Greece 7.3 Malta 9.1 Norway 4.9 Belgium 7.0 France 5.4 Netherlands 4.6 Latvia 4.4

High-income Slovenia 4.9

Switzerland 4.2 High-income Australia 3.6 Czech Republic 4.0 Malta 2.3 Greece 3.2 Slovenia 1.5 Finland 2.4 Czech Republic 0.8 Australia 1.1 Chile –1.7 Sweden 0.4 France –2.2 Chile –2.7 Sweden –5.1 Slovakia –6.9 Slovakia –8.1 Hungary –7.2 Finland –8.6 Croatia –7.8 Lithuania –13.4 Poland –9.4 Hungary –15.8 Lithuania –14.1 Croatia –19.0 Ireland –17.7 Poland –20.3 EU 7.8 Ireland –25.5 High-income 9.5 EU 3.6 Jordan 17.5 High-income 7.4 Nepal 9.8 Jordan 17.2 Costa Rica 9.6 Costa Rica 9.4 Bangladesh 2.3 Nepal 6.6 Namibia 2.0 Bangladesh 1.2 Sierra Leone 0.9 Mexico –2.7 Romania –1.6 Romania –2.2 Bulgaria –6.7 Turkey –10.3 Turkey –14.3 Gambia –12.8 Tanzania** –16.5 Mexico –23.5 Namibia –18.3

w- and middle income Tanzania** –24.7 Bulgaria –18.5 w- and middle income Lo Albania –43.0 Gambia –35.9 Lo Madagascar –72.9 Madagascar –83.4 Bolivia* –148.9 Albania –132.4 Low- and middle-income –23.8 Low- and middle-income –17.9 0 50 40 –6 0 –1 0 –50 –200 –150 –100 –160 –110 Percent Percent

Note: EU, high-income, and low- and middle-income estimates are the averages of the European Union, the sample of high-income countries, and low- and middle-income countries, respectively. Averages are weighted by the number of wage employees in each country. * the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). Chapter 3 – Measuring and understanding the migrant pay gap 83

X Figure 26: Factor-weighted migrant pay gaps using monthly earnings, latest years

Factor-weighted mean migrant pay gap Factor-weighted median migrant pay gap

Cyprus 27.3 Cyprus 29.8 Spain 24.5 Luxembourg 28.3 Luxembourg 24.2 Greece 22.9 Greece 21.4 Spain 22.7 Estonia 21.1 Italy 21.5 Italy 20.9 France 18.9 France 18.9 Estonia 16.0 Iceland 15.9 Austria 14.9 Austria 15.6 Canada 13.1 Portugal 14.9 Denmark 12.5 Denmark 14.2 United States 10.6 United States 11.9 Argentina 10.4 Canada 10.9 Belgium 8.1 Latvia 9.7 Latvia 6.9 Netherlands 9.1 Portugal 6.7 United Kingdom 7.6 Ireland 5.7 Poland 7.5 Iceland 5.6 Argentina 7.4 Australia 4.8 Czech Republic 7.1 United Kingdom 4.8 5.4 4.0 High-income

Norway High-income Czech Republic Slovenia 4.8 Switzerland 3.5 Ireland 4.7 Poland 3.2 Belgium 4.4 Netherlands 2.1 Finland 2.9 Slovenia 1.1 Sweden 2.8 Norway 1.1 Switzerland 2.6 Finland 0.9 Australia 1.9 Sweden –1.1 Malta –4.0 Chile –4.8 Lithuania –4.5 Slovakia –7.5 Croatia –6.1 Malta –10.3 Chile –6.4 Croatia –13.2 Slovakia –7.4 Lithuania –13.2 Hungary –10.3 Hungary –16.0 EU 11.7 EU 9.7 High-income 11.3 High-income 10.1 Turkey 12.1 Namibia 19.1 Namibia 11.2 Turkey 14.0 Jordan 10.5 Jordan 12.8 Mexico 7.5 Mexico 10.4 Costa Rica 7.0 Costa Rica 6.7 Tanzania** 4.7 Romania –3.7 Albania 1.4 Bangladesh –4.8 Nepal –0.3 Tanzania** –7.3 Bangladesh –0.4 Bulgaria –7.4 Romania –2.5 Malawi –7.5 Bulgaria –3.8 Nepal –7.6 Malawi –8.6 Bolivia* –16.3 w- and middle income Sierra Leone –11.5 Madagascar –26.4

Lo Low- and middle income middle and Low- Bolivia* –12.6 Gambia –31.1 Madagascar –13.6 Albania –77.0 Gambia –43.9 Sierra Leone –178.6 Low- and middle-income 4.2 Low- and middle-income 3.2 30 80 –7 0 –2 0

-60 -40 -20 02040 –220 –170 –120 Percent Percent

Note: EU, high-income, and low- and middle-income estimates are the averages of the European Union, the sample of high-income countries, and low- and middle-income countries, respectively. Averages are weighted by the number of wage employees in each country. * the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). 84 The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure 27: Comparing the raw migrant pay gap with significant part of the composition effects caused the factor-weighted migrant pay gap using mean by the existence of clusters in the wage or earnings hourly wages, latest years distribution of wage workers. In essence, migrant and non-migrant wage workers are somewhat grouped into homogeneous subgroups based Slovakia on selected observed characteristics (or factors), Czech Republic Australia and then the migrant pay gap is estimated for Lithuania each of the subgroups. A weighted sum of all Croatia Poland the subgroups’ specific pay gaps is estimated Malta to obtain the factor-weighted pay gap, with the Switzerland weights reflecting the size of each subgroup in United Kingdom Canada the population. Hungary Chile Drawing from the human capital model (Mincer, France Sweden 1974) and the Global Wage Report 2018/19, educa- Finland tion, labour market experience, and gender are Belgium Norway chosen as three factors that together can pick up a Latvia major part of the composition effects in estimating United States the migrant pay gap. Education and experience Denmark High-income Iceland (with education mirroring occupational skills and Argentina age serving as a proxy for experience) are two Netherlands Ireland important indicators of the job profile of wage Estonia employees. It is acknowledged that women and Greece Austria men differ in their labour market participation and Luxembourg working hours (see table 4 and figures 3 and 12). Spain Portugal Furthermore, women wage employees are more Italy likely than men to work in the informal economy. Slovenia Cyprus Whereas “education” and “age” are in line with the EU human capital model, the inclusion of “gender” as a High-income Madagascar third factor incorporates a specific gender focus to Romania better capture the composition effects underlying Sierra Leone women’s and men’s respective modes of participa- Gambia Bulgaria tion and experience in the labour market. Namibia Bolivia* Bangladesh Mexico The factor-weighted approach Tanzania** Albania produces similar mean migrant Nepal w- and middle income Turkey pay gap estimates compared to the Lo Jordan Costa Rica standard approach Low- and middle-income Figures 25 and 26 present the results of apply- 40 –60 –10

–160 –110 ing the factor-weighted approach to the studied Percent countries using hourly wages and monthly earn-

Raw mean migrant pay gap ings. To see how the factor-weighted mean migrant

Factor-weighted mean migrant pay gap pay gap compares with the raw mean migrant pay gap based on the standard approach in section 3.2, the results from both approaches are plotted against each other. Figure 27 presents the result of Note: EU, high-income, and low- and middle-income estimates this comparison. are the averages of the European Union, the sample of high-income countries, and low- and middle-income countries, respec- Clearly, the factor-weighted method has an impact tively. Averages are weighted by the number of wage employees in each country. * the Plurinational State of Bolivia; ** the United on the migrant pay gap estimates in each coun- Republic of Tanzania. try, particularly in terms of size. In most HICs, the Source: ILO estimates based on survey data provided by na- factor-weighted migrant pay gap is similar to the tional sources (see Appendix II). migrant pay gap based on the standard approach. Chapter 3 – Measuring and understanding the migrant pay gap 85

On average, the report estimates that the fac- LMICs? The report finds that, education, years of tor-weighted migrant pay gap (based on mean experience, and other observed labour market hourly wages) is approximately 9.5 per cent in the characteristics explain a relatively small part of the sample of HICs and 7.8 per cent in the EU, which migrant pay gap. In many countries, the migrant are similar to the estimates based on the standard pay gap remains significantly high even after approach (12.6 per cent in the sample of HICs and accounting for the factors that normally explain 8.6 per cent in the EU) (figures 25 and 27). The fac- differences in wages between individuals. In tor-weighted estimates based on median hourly other words, the unexplained part of the migrant wages is estimated at 7.4 per cent in the sample pay gap largely dominates the explained part in of HICs and 3.6 per cent in the EU. Estimates based most countries, regardless of income group. On on monthly earnings are also similar to estimates average, the report shows that about 10 percent- based on the standard approach. There are of age points of the weighted migrant pay gap of course wide variations across the studied countries approximately 12.6 per cent (based on average and across the type of pay used. The mean hourly hourly wages) in the sample of HICs remains migrant pay gap based on the factor-weighted unexplained by observed labour market charac- approach, for example, is lowest in Ireland teristics, while nearly all the 17.3 per cent of the (–17.7 per cent, where the standard approach pro- pay gap in favour of migrant workers in the sample duces an estimate of 20.6 per cent) and highest in of LMICs is unexplained. Possible reasons for this Cyprus (28.1 per cent, where the standard approach (particularly in HICs), among others, include: (i) produces an estimate of 42.1 per cent), among the fact that migrant workers in HICs tend to have the studied HICs. The factor-weighted median lower wage returns to their education relative to migrant pay gap based on hourly wages ranges nationals, even when the migrant worker and the from –10.3 per cent in Hungary to 27.3 per cent in national work in the same occupational category; Cyprus, among the studied HICs. (ii) skills mismatch among migrant workers in HICs due to the fact that migrants face significant In the sample of LMICs, on the other hand, the barriers transferring their skills and experience factor-weighted approach produces an overall across countries; and (iii) discriminatory practices mean migrant pay gap (based on hourly wages) of against migrant workers, including non-compli- –23.8 per cent compared to –17.3 per cent when ance with the principle of “equal pay for work of based on the standard approach, and overall equal value”. median migrant pay gap (based on hourly wages) of –17.9 per cent compared to –7.5 per cent when based on the standard approach. In terms of The following sections in Chapter 3 attempt to monthly income, the factor weighted approach answer this question by first, estimating the produces an overall mean monthly migrant pay migrant pay gap at different points in the hourly gap of 4.2 per cent (compared to –19.1 per cent, wage distribution, which sheds light on the poten- when based on the standard approach) and an tial impact of different targeted policies on the overall median monthly migrant pay gap of 3.2 per overall migrant pay gap. It is thus important to cent (compared to –9.1 per cent, when based on know where in the wage distribution the migrant the standard approach). Clearly, there are large pay gap is widest. To complement this information, variations across the studied LMICs and across the chapter looks into the proportion of migrant the type of pay used, with the factor-weighted workers in different parts of the wage distribution, mean migrant pay gap, based on hourly wages, showing the extent to which migrant workers are for example, ranging from –148.9 per cent in the over-represented at the lower end of the wage Plurinational State of Bolivia (compared to –30.0 per distribution, or under-represented at the upper cent , when based on the standard approach) to end. Second, the migrant pay gap is decomposed, 17.5 per cent in Jordan (compared to 29.5 per cent at different parts of the wage distribution, into a from the standard approach). part that can be “explained” by differences in the labour market attributes of migrant workers and nationals and a part that is “unexplained” by such 3.5. What factors lie behind characteristics. Third and finally, factors contribut- the migrant pay gap? ing to the unexplained component of the migrant pay gap, including the lower returns to education Why do migrant workers generally earn less of migrant workers within the same occupations than nationals, especially in most HICs and some are examined. 86

hourly wage distribution is striking. However, the gap shrinks steadily as it moves from the lower to higher points in the wage distribution. In the case of France, for example, although the mean pay gap is estimated at about 9.0 per cent, the gap at the bottom decile of the wage distribution is approximately 71.1 per cent but declines sharply to about 6.3 per cent at the ninth decile and eventually becomes negative at the tenth decile. This magnitude of disparity have huge policy implications for poverty eradication and for ensuring decent work among low-skilled migrant workers. © shutterstock.com Second, in other countries, the migrant pay gap appears to be lower at the bottom and top deciles 3.5.1. Estimating the migrant of the hourly wage distribution but strikingly high pay gap across the hourly wage in the middle of the distribution. This may possibly distribution reflect under representation of migrant workers in collective representation structures in the middle of the distribution because of difficulties orga- This section analyses the migrant pay gap at dif- nizing or because nationals dominate the overall ferent points in the hourly wage distribution; in representation, a phenomenon that could be exac- particular, at each of the equally-sized ten deciles erbated if migrants are perceived as a low-wage of wage distribution (from the bottom 10 per cent employment threat to nationals (see, e.g., Rubery, wage earners up to the top 10 per cent earners). 2003). This pattern is common in countries such as This estimation is a useful tool that can shed light on Argentina, Belgium, Canada, Iceland, Luxembourg, the need for targeted policies to narrow the migrant the Netherlands, and the United States. For exam- pay gap. For example, introducing and/or enforcing ple, in the case of the Canada, the migrant pay gap minimum wages with broad legal coverage could at the bottom and top deciles of the hourly wage help reduce the pay gap at the lower end of the distribution are –0.6 per cent and 0.4 per cent, wage distribution; while collective agreements that respectively, but increases to about 6.5 per cent in include provisions on equal pay and pay transpar- the middle of the distribution (i.e. from the fifth to ency could have a similar effect in the middle and the eight decile). upper ends of the wage distribution. Third, and in particular in some LMICs, the migrant Figure A.4 (see Appendix IV) shows the migrant pay pay gap widens and narrows, and reverses in favour gap at each decile of the hourly wage distribution of nationals or in favour of migrant workers across for 48 of the 49 studied countries, comparing this the hourly wage distribution. This pattern can give with the mean hourly pay gap for each country. The an indication of where in the wage distribution are first observation is that the migrant pay gap varies temporary high-skilled “expatriate” migrant workers across the wage distribution in each country. The potentially located in those countries. In Gambia for following patterns appear to stand out. First, for example, non-migrant workers tend to earn more some countries, there is a tendency for the migrant than migrant workers from the bottom to the pay gap to be strikingly high at the bottom deciles, fourth decile of the wage distribution. However, the but it declines as it moves from the lower to upper gap reverses in favour of migrant workers from the ends in the hourly wage distribution. This could fifth to the top decile of the distribution, peaking at possibly imply non-compliance with or non-inclu- the ninth decile where migrant workers earn about sion of migrant workers in minimum wage legis- 54.8 per cent more than non-migrant workers. lation58. For example, in Austria, Cyprus, Denmark, France, Norway, Spain, and Sweden, the widening Given that the share of the migrant population of the pay gap at the first and second deciles of the is generally small across countries, and that the

58 Exclusion from minimum wage coverage can take many forms. National provisions in force in some countries may explicitly provide for reduced minimum wage rates for migrant workers. Migrant workers could also be excluded because there is no minimum wage for the sector in which they are primarily employed. Likewise, migrants may not benefit from minimum wage coverage because they are not members of a trade union that is a party to the collective agreement covering the sector of activity concerned (see ILO, 2014b). Chapter 3 – Measuring and understanding the migrant pay gap 87

sampling design of these datasets are not condi- points of the weighted migrant pay gap of approx- tioned on migration status, it could be that migrant imately 12.6 per cent (based on average hourly workers are not well represented at different points wages) in the sample of HICs remains unexplained in the hourly wage distribution. This could lead to by observed labour market characteristics. On the “small sample bias” in the estimates of the migrant other hand, nearly all the 17.3 per cent of the pay pay gap across the wage distribution. Verification of gap in favour of migrant workers in LMICs is unex- this is shown in figure A-5 (see Appendix IV) which plained. However, it is significant to add that there shows the respective shares of migrant workers and are large variations across countries and across the nationals at different locations in the hourly wage wage distribution. For example, in Cyprus (which distribution. Figure A-5 further disaggregates the has the widest estimated pay gap in the sample share of migrant workers by sex. In most countries, of HICs), only about 4.4 percentage points of the a sufficient proportion of migrant workers are pay gap of 42.1 per cent is explained by observed located in each decile of the hourly wage distribu- labour market characteristics, which is lower than tion. In a number of countries, migrant workers are the roughly 12 percentage points explained gap out disproportionately concentrated at the first and sec- of the estimated 34.8 per cent total migrant pay gap ond deciles, but as we move from lower to higher in 2010 by the ILO Global Wage Report 2014/15. In points in the wage distribution, their proportion the United States, on the other hand, about 10 per- declines, in some cases sharply. centage points of the estimated migrant pay gap of 15.2 per cent is explained by observed labour For example, migrant workers make up about market characteristics. 93 per cent of the bottom 1 per cent of wage earners in Cyprus (all of whom are migrant women), but only about 11 per cent of the top 1 per cent. In As previously defined, the unadjusted pay gap Italy, migrant workers account for more than 23 per refers to the earnings of nationals (at a given percent of the bottom 1 per cent wage earners but just centile in the wage distribution), minus the earnings about 3 per cent of the top 1 per cent. Estimates of of migrant workers (at the same percentile in the the migrant pay gap for these countries (figure A-4) distribution). On the one hand, the “explained” part indicate that the wage gap is much wider at the bot- refers to the part of the migrant pay gap that relates tom of the wage distribution, where migrant work- to differences in labour market attributes or char- ers are concentrated (many of whom are migrant acteristics of migrant workers and nationals, such women). However, in countries such as Australia, as human capital endowments, and job and work- Malta and Switzerland, the share of migrant workplace characteristics. In other words, the explained ers increases as we move from the bottom to top part takes into account the factors described in ends of the wage distribution. Estimating migrant table 8: age; experience; education (grouped into pay gaps at different points in the wage distribution four categories); occupational category (manage- can be more informative than using simple sum- rial, professional, technical, etc.); type of working mary measures such as the mean or median wage, contract (permanent versus temporary); economic especially when migrant workers are concentrated activity (about ten categories, including agriculture, at certain points in the wage distribution. construction, manufacturing, services, etc.); public versus private sector employment; regional location (e.g., urban versus rural); formality; gender; 3.5.2. What part of the migrant race (in the case of the United States); and work pay gap can be explained by intensity (hours worked). On the other hand, the differences in the characteristics unexplained part (or wage penalty) is what remains after adjusting for these observable labour market of migrant workers and nationals characteristics, which should in principle explain pay in the labour market? differences between migrant workers and nationals. It is important to emphasize that, in this context, Decomposing the migrant pay gap into the “unexplained” should be understood as not explained and unexplained parts accounted for by the observed labour market characteristics listed in table 8. Similarly, the “explained” The next step in understanding the migrant pay should be understood as only accounting for by the gap is to decompose the unadjusted pay gap into observed labour market characteristics listed in this “explained” and “unexplained” parts. On the one same table. The explained part may still reflect fun- hand, the report finds that about 10 percentage damental bias based on observed factors such as 88 The migrant pay gap: Understanding wage differences between migrants and nationals

X Box 4. Decomposing the migrant pay gap: An illustrative explanation

The decomposition of the migrant pay gap consists of three steps. First, a set of attributes (observed indicators in survey data) are selected on the basis of their relevance to the wage determination process. Table 8 shows the observed attributes and characteristics selected for the decomposition of the migrant pay gap. The selection is based on the availability of these indicators in each of the surveys described in Appendix II. Not all indicators are always available for all countries, and some are exclusive to particular economic contexts (see notes to table 8).

In the second step, econometric techniques are applied, using the observed attributes or characteristics, to generate a counterfactual wage distribution, which represents the wages that migrant workers would earn if they received the same returns to their attributes and characteristics as nationals. This results in three wage distributions: the wage distribution for nationals, the actual wage distribution for migrant workers and the counterfactual wage distribution for migrant workers. The three distributions can be compared at any of their quantiles, for example at the median. Suppose now that at the median, the hourly wage of nationals is 10 coins and that of migrant workers is 8 coins. This would mean that the median migrant pay gap is 20 per cent in favour of nationals. As- sume also that at the median, the counterfactual hourly wage is 9 coins. This represents the median wage that migrant workers would earn if, for their actual “average” endowments and attributes, they received the same returns as nationals with similar attributes. Then, the difference between what nationals get (10 coins) and the counterfactual (9 coins) is “explained” by differences in the observed attributes of nationals and migrant workers. The rest, namely the difference between the counterfactual (9 coins) and what migrant workers receive in reality (8 coins) is “unexplained” by the observed differences in attributes. Therefore, the unexplained part is attributable to the fact that migrant workers receive disproportionately lower returns to their labour market endowments and characteristics at the median. The unexplained part is also called the “structural” part of the migrant pay gap.

The construction of the counterfactual helps to identify the fact that migrant workers can have a different wage structure from nationals, not because they have different endowments but because they get different returns to such endowments – hence the word “structural” is sometimes employed to denote the unexplained part of the migrant pay gap. In the example above, the explained and unexplained parts of the migrant pay gap is 1 coin each. In sum, this hypothetical example illustrates that the total median migrant pay gap of 20 per cent can be decomposed into two parts: the explained part (10 per cent) and the unexplained part (10 per cent).

The third and final step in the decomposition consists of simply applying the counterfactual distribution to decompose the migrant pay gap at each quantile into that which can be explained and that which cannot be explained by differences in the observed attributes and characteristics of nationals and migrant workers, as explained in the hypothetical example above.

contractual conditions or personal characteristics income inequality and ensuring equal pay for work including gender or race/ethnicity. of equal value regardless of nationality or gender. Although labour market institutions and policies Similar to the ILO’s Global Wage Report 2018/19, (such as minimum wages that aim to benefit all this report adapts the decomposition techniques workers) can be designed to cover all paid employ- proposed by Fortin, Lemieux and Firpo (2011) to the ees in principle, reducing inequality often requires migrant pay gap. This allows for the identification additional targeted policy action. This part of the of explained and unexplained components of the report shows that migrant workers sometimes migrant pay gap across the hourly wage distri- incur “wage penalties” for multiple and complex bution. The technique involves three steps. First, reasons that differ from one country to another, a set of observed attributes and characteristics is and that the penalty can occur at different locations selected that typically explain differences in wages in the hourly wage distribution. Understanding the of migrant workers and nationals (table 8). The reasons for these wage penalties in the national second step comprises estimating a “counterfac- context, and adopting policies to eliminate them, tual” wage distribution for migrant workers. This could make a significant contribution to reducing counterfactual wage distribution represents the Chapter 3 – Measuring and understanding the migrant pay gap 89

X Table 8. Observed labour market endowments, attributes and characteristics for the decomposition of the migrant pay gap

Group Variables Notes

Endowments Age When experience is missing, it is proxied by the difference between age and years of schooling for individuals Education (categories) actively in employment. Years of experience Countries vary in terms of the number of educational categories, although most will identify four or five (e.g. no education; below primary; lower secondary; high school/vocational; university and above).

Job attributes Working time “Working time” can be a continuous variable or a dummy variable to identify full time versus part time (fol- or Characteristics Contractual conditions lowing the international definition given by the OECD). Occupational categories “Contractual conditions” implies a dummy variable to distinguish between permanent and temporary contracts. The occupational categories for all countries follow the international classification code ISCO-88 or ISCO-08. The following distinctions are made where possible: CEOs, Senior Officials, Managers; Professionals; Technical officers;; Clericals; Service and Sales; Skill Agro; Artisans; Low skills; and Elementary jobs.

Workplace Industrial category for The industrial categories for almost all countries follow the industrial classification characteristics production (principal given by ISIC Rev. 4. The size of the enterprise is usually declared in categories (micro, small, medium and economic activity) large). Size of the enterprise A dummy for informality is included independently for workers in the informal economy in countries where Public or private sector informality can be identified in the survey data. Regional location Urban versus rural area Formal versus informal employment

Personal Gender Race is available for the United States only. characteristics Race

Notes: ISCO = International Standard Classification of Occupations; ISIC = International Standard Industrial Classification of All Economic Activities, OECD = Organization for Economic Co-operation and Development. Not all variables are available for all countries in the data set (e.g. public sector workers are not identified in the countries from the European Union). Most of the following are observed everywhere: age, education, experience, working time, contractual conditions, occupational category, industrial code (principal economic activity) and rural/urban location. Exceptionally, in the case of the United States’ Current Population Survey, race is also identified, and a dummy for “white” versus all other races is used in the decomposition of the migrant pay gap for this country. In countries where the variable “occupational category” included a single category for “domestic worker” the latter was included independently as a category under occupations.

wages that migrant workers would earn if they as well as policies that aim to achieve better inte- received the same returns to their labour market gration for migrants in countries of destination. A endowments as nationals. The third step involves second issue related to policy is that the size of the using the counterfactual wage distribution to unexplained component can suggest that reduc- decompose the migrant pay gap into that which can ing the migrant pay gap may require additional be explained and that which cannot be explained measures to combat pay discrimination and wage by the observed attributes and characteristics. By penalties for migrant workers, especially migrant this way, the difference between nationals’ wage women, for instance through promoting legal and the counterfactual wage at each decile is the frameworks and policies conducive to ensuring explained gap and the difference between the equal pay for work of equal value. counterfactual and migrant worker’s actual wage Based on the decomposition approach used, both is the unexplained gap. Box 4 provides an explana- the explained and unexplained parts of the migrant tion of how these three steps work in practice and pay gap can be either positive or negative. The fol- Appendix I lays out the methodology in more detail. lowing interpretation therefore holds. A positive From a policy standpoint, both the explained and explained migrant pay gap implies that nationals unexplained components are important channels are rewarded higher than migrant workers and through which to redress differences in pay across this difference in pay is explained by differences in groups. First, identifying the part of the migrant pay their observed endowments. On the other hand, a gap that can be explained by labour market char- negative explained part means that based on their acteristics can inform policy-makers in designing observed endowments, migrant workers would policies that target differences in endowments and earn more than nationals if they (migrant workers) characteristics of migrant workers and nationals, were to receive same returns to their endowments 90 The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure 28: Decomposition of the migrant pay gap at the mean hourly wage into explained and unexplained parts, latest years

High-income countries Low- and middle-income countries

Hungary Sierra Leone Chile Netherlands Spain Namibia Australia France Costa Rica Ireland

Croatia Malawi Greece Latvia Madagascar Cyprus Italy Nepal Denmark Argentina Jordan Sweden Belgium United Kingdom Gambia Finland

Iceland Albania Switzerland Slovakia Bolivia* Lithuania Malta Mexico Estonia Norway Bulgaria Canada United States Poland Bangladesh Czech Republic Austria Tanzania** Portugal

Luxembourg Turkey Slovenia EU Low- and middle-income High-income 0 0 20 40 60 80 20 40 60 80 –80 –60 –40 –20 –80 –60 –40 –20 –140 –120 –100 –140 –120 –100 Migrant pay gap (%) Migrant pay gap (%)

Explaine d Unexplained

Note: EU, high-income, and low- and middle-income estimates are the averages of the European Union, the sample of high-income countries, and low- and middle-income countries, respectively. Averages are weighted by the number of wage employees in each country. * the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). Chapter 3 – Measuring and understanding the migrant pay gap 91

X Figure 29: The unexplained mean wage differentials between migrant workers and nationals, latest years

High-income countries Low- and middle-income countries

Czech Republic Madagascar Poland Slovakia Lithuania Tanzania** Malta Canada Bulgaria Switzerland

United Kingdom Bangladesh Slovenia Croatia Gambia Australia Luxembourg Turkey United States Finland Bolivia* Norway Austria Sweden Mexico Belgium

Portugal Albania Iceland Estonia Nepal France Denmark Malawi Argentina Latvia Namibia Greece Ireland Italy Sierra Leone Chile Netherlands Jordan Cyprus

Spain Costa Rica Hungary EU Low- and middle-income High-income 0 0 10 20 30 40 50 60 20 40 60 80 –8 0 –6 0 –4 0 –2 0 –40 –30 –20 –10 –100 Migrant pay gap (%) Migrant pay gap (%)

Note: The unexplained part of the migrant pay gap, using a counterfactual approach which compares wages of migrant workers to the wages that they would receive if they were nationals with similar characteristics. The same methodology is used in the ILO Global Wage Report 2018/19. Note: EU, high-income, and low- and middle-income estimates are the averages of the European Union, the sample of high-income countries, and low- and middle-income countries, respectively. Averages are weighted by the number of wage employees in each country. * the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). 92 The migrant pay gap: Understanding wage differences between migrants and nationals

as nationals (but in reality, migrant workers may average on the basis of their observed attributes earn less). The unexplained part is positive when the and characteristics. returns that migrant workers should get from their endowments are higher than what they receive. It is The explained and unexplained parts of negative, however, when migrant workers are paid above what they should get on the basis of their the migrant pay gap: A comparison across observed endowments. countries

This section shows the explained and unexplained A significant part of the migrant pay parts of the hourly migrant pay gap at the mean gap across the hourly wage distribution rather than across the entire wage distribution as remains unexplained shown in figure A-6. The results are displayed in figures 28 and 29, grouping countries according to Figure A-6 (see Appendix IV) shows the decompo- their income. Figure 28 presents the explained and sition of the migrant pay gap at each decile of the unexplained together while figure 29 presents the hourly wage distribution. Each chart shows two unexplained pay gap only. The two bars in Figure 28 components of the gap at each decile: the explained (explained and unexplained) should sum up to the and unexplained parts. The sum of the two parts at total unadjusted mean hourly pay gap in figure 17, each decile equals the raw gender pay gap at that in principle. decile, in principle (see figure A-4). The results show that, on average, the mean migrant The report finds that, on average, education and pay gap remains largely unexplained by observed other observed labour market characteristics differences in education and other observed labour explain a relatively small part of the migrant pay market characteristics of migrant workers and gap at different locations in the wage distribution. nationals. The unexplained part of the migrant pay Also, the results show that the relative importance gap largely dominates the explained part in most of the explained and unexplained parts of the countries, irrespective of income group. On the one migrant pay gap across the hourly wage distri- hand, the analysis of the data shows that about bution varies across countries. Among HICs, the 10 percentage points of the weighted migrant pay migrant pay gap is largely unexplained throughout gap of approximately 12.6 per cent (based on aver- the wage distribution in most countries (for exam- age hourly wages) in the sample of HICs remains ple, Argentina, Belgium, Cyprus and Italy). However, unexplained by observed labour market character- in few countries such as Austria, Australia, Canada, istics of migrant workers and nationals. On the other Luxembourg and the United States, a large part hand, nearly all the 17.3 per cent of the pay gap in of the pay gap is explained by differences in the favour of migrant workers in LMICs is unexplained, attributes and characteristics of migrant workers on average. However, there are several exceptions, and nationals. In others, for example, Norway and as well as wide variations across countries. Among the United Kingdom, the picture is quite different: HICs, differences in observed labour market charac- the migrant pay gap is largely unexplained at teristics have sizeable effects on the migrant pay gap lower deciles of the wage distribution, but appears in countries such as Austria, Canada, Luxembourg, to be partly explained in the upper parts of the Norway, Portugal, Slovenia, the United Kingdom, and distribution. the United States, though a significant part remains In the sample of LMICs, the picture is even more unexplained. Among LMICs, the same is true of mixed. On the one hand, the observed negative Bangladesh, Costa Rica, Gambia, Jordan, the United migrant pay gap is largely explained throughout Republic of Tanzania, and Turkey. But in most coun- the wage distribution in few countries such as tries, a large part of the migrant pay gap remains Albania and Malawi. On the other hand, the pay unexplained. For example, in Cyprus (which has the gap is unexplained in a number of countries (for widest estimated pay gap in the sample of HICs), example, Bangladesh, Jordan and Madagascar). In only about 4.4 percentage points of the pay gap of countries such as Bolivia, Mexico and the United 42.1 per cent is explained by observed labour mar- Republic of Tanzania, the unexplained gap is mostly ket characteristics of migrant workers and nationals, negative in the upper parts of the wage distribution. which is lower than the roughly 12 percentage points This reflects the possibility that migrant workers at explained gap out of the estimated 34.8 per cent the top of the distribution in those countries could total migrant pay gap in 2010 by the ILO Global receive returns above what they should get on Wage Report 2014/15. In the United States, on the Chapter 3 – Measuring and understanding the migrant pay gap 93

other hand, about 10 percentage points of the esti- migrant workers. This illustrates “vertical occupa- mated migrant pay gap of 15.2 per cent is explained tional segregation” – the clustering of nationals at by observed labour market characteristics. the top of occupational hierarchies and of migrant workers at the bottom. This pattern is true for almost all the sample of HICs with the exception 3.5.3. Understanding the of Australia, Canada, Chile and Luxembourg. These unexplained part of the migrant differences in occupations between migrant work- pay gap: Selected countries ers and nationals are part of the explained component of the migrant pay gap (see table 9). What lies behind the unexplained part of the However, the same figure shows that within occu- migrant pay gap? The decomposition exercise in pational categories and in almost all the HICs, the previous section shows that much of the gap migrant workers score just as high as or even much in earnings remains unexplained by the observed higher in education than nationals. The line show- differences in labour market attributes and char- ing the “score in education” for migrant workers acteristics of migrant workers and nationals. This is nearly always above the line for nationals in the section turns to the question of whether migrant majority of the sampled countries.59 Therefore, on workers obtain lower returns to their educational average, it appears that migrant workers’ educa- attainments relative to nationals within the same tional attainment is similar to or better than that occupation. Migrant workers are more likely than of nationals within each occupational category. In nationals to be attached to particular sectors spite of this, the charts also show that for almost and occupations in their countries of destination. all occupational categories and in the majority of In what follows, a deeper analysis explores the the sampled HICs, the migrant pay gap remains migrant pay gap in various occupational categories positive and sizeable, with few exceptions such and compares migrant workers’ educational levels as Australia, Chile, Switzerland, and the United with nationals’ within each occupational category. Kingdom. This indicates that within occupational Figure A-7 (see Appendix IV) explores this question categories, migrant workers obtain lower returns by comparing the proportions of migrant workers to their educational attainments in the majority and nationals with the pay gaps and educational of the HICs. This may be the result of a range of attainments within the same occupational cate- factors including pay discrimination at the work- gory. In the selected sample of 31 countries, the place, “horizontal segregation”, whereby at the datasets allow to separate workers into six distinct same occupational level (within occupational occupational groups: managers (MGR), profession- categories or even occupations themselves) als (PROF), technical (TECH), semi-skilled (SEMI), migrant workers and nationals have different job low-skilled (LOW), and the unskilled (UNS). The tasks. It would be interesting to look further into charts show that the share of migrant workers in returns to education within different sectors rather the lower occupational categories (unskilled, low- than occupations and to explore the differences in skilled or semi-skilled) is almost everywhere much returns to education for migrant men and women higher than the share of migrant workers in the top separately. occupational categories (managers, professionals Table 9 summarizes the results from the previous or technical). For example, in Italy, less than 3 per chapters by showing estimates of the migrant pay cent of managers are migrant workers, whereas gap (based on mean hourly wages), its decomposi- about 30 per cent of unskilled jobs are occupied by tion, and estimates of migrants’ population shares.

59 The “score in education” is a country-specific value that gives everyone a score to indicate his or her relative achievement in education in a given country. For the sample of countries for which “years of education” is not available, individuals declare their educational attainment as a categorical outcome. Typically, there will be about five categories: “no formal education”, “less than or equal to primary education”, “secondary education without high school diploma”, “high school completed, including those with some vocational education or training” and “university studies or above”. The “score in education” simply assigns to each individual a value that is related to these categories and increases exponentially for higher educational achievements. The assignment of values is done in a similar fashion as in the ILO’s Global Wage Report 2018/19 (2018a). Thus, individuals in the first and lowest category (no formal education) are assigned a value of 1; in the second category they are assigned a value of 6; and in the next three categories they are assigned values of 9, 16 and 25, respectively. This exponential increase simply aims at emulating the relative values that would have been given if data on the number of years spent in education to achieve a particular level of education existed for these countries. The exponential assignment helps to avoid assuming that the jump between one educational category and the next implies a constant and even effort (which is what the category number alone does). The assigned value is the score that an individual gets to quantify his or her education relative to other wage employees in a given country. For each of the occupational categories, we take the average of these scores in education for migrant workers and for nationals. The charts in the third column in figure A-7 (see Appendix IV) show these estimates. 94 The migrant pay gap: Understanding wage differences between migrants and nationals

X Table 9. Summary: Migrant share of total working population, migrant share of wage workers, and the mean hourly migrant pay gap

Factor- Mean Explained Unexplained Migrants' weighted Country Latest Migrants' hourly (mean) (mean) Country share of wage mean Code year population (%) migrant pay migrant pay migrant pay workers (%) hourly migrant gap (%) gap (%) gap (%) pay gap (%)

High-income

POL Poland 2015 0.15 0.21 -4.93 -9.35 13.36 -21.11

SVK Slovakia 2015 0.16 0.14 -12.55 -6.92 3.75 -16.93

HUN Hungary 2015 0.49 0.58 5.33 -7.20 -113.85 55.73

HRV Croatia 2015 0.59 0.56 -6.24 -7.81 -8.37 1.97

LTU Lithuania 2015 0.59 0.53 -6.76 -14.11 5.11 -12.51

CZE Czech Republic 2015 1.64 1.29 -8.09 4.04 15.72 -28.26

PRT Portugal 2015 2.14 2.19 28.91 17.38 17.71 13.62

FIN Finland 2015 2.93 2.22 10.69 2.42 2.80 8.12

NLD Netherlands 2015 2.96 2.64 19.86 14.00 -40.74 43.06

MLT Malta 2015 3.13 2.64 -2.67 9.09 6.12 -9.37

SVN Slovenia 2015 3.61 4.40 33.26 4.86 32.72 0.80

NOR Norway 2015 4.67 4.93 15.03 9.62 7.49 8.15

CHL Chile 2017 4.71 6.71 6.44 -2.74 -43.41 34.76

DNK Denmark 2015 4.86 4.43 17.32 11.48 -1.13 18.24

FRA France 2015 4.87 4.29 8.70 5.38 -9.93 16.95

SWE Sweden 2015 5.45 4.23 10.51 0.36 -0.37 10.84

ISL Iceland 2015 5.91 5.92 17.84 10.57 3.06 15.25

ARG Argentina 2018 6.35 6.36 18.07 10.20 -1.01 18.89

GRC Greece 2015 6.46 8.84 21.16 3.23 -5.65 25.37

ITA Italy 2015 8.90 11.23 29.59 20.33 -2.02 30.98

ESP Spain 2015 8.94 9.01 28.27 18.63 -39.71 48.65

USA United States 2018 9.06 8.95 15.27 11.58 10.00 5.86

GBR United Kingdom 2015 9.70 9.28 2.72 11.84 2.73 -0.01

BEL Belgium 2015 10.36 8.83 12.73 7.04 1.15 11.71

IRL Ireland 2015 11.92 13.52 20.58 -17.66 -8.75 26.97

AUT Austria 2015 14.44 12.90 25.26 18.08 17.29 9.64

LVA Latvia 2015 14.67 13.72 15.07 10.22 -5.62 19.59

EST Estonia 2015 14.83 13.39 21.00 23.68 6.16 15.82

CYP Cyprus 2015 16.58 19.53 42.06 28.12 -4.45 44.53

CAN Canada 2018 24.80 24.13 3.98 10.98 7.58 -3.89

AUS Australia 2017 27.35 26.37 -6.89 1.08 -10.21 3.01

CHE Switzerland 2016 28.71 29.31 0.32 4.23 3.33 -3.12

LUX Luxembourg 2015 43.13 46.99 27.34 10.88 23.96 4.44

EU average 6.00 6.03 8.61 7.82 -7.87 14.24

High-income 9.27 9.22 12.56 9.51 0.18 10.08 average Chapter 3 – Measuring and understanding the migrant pay gap 95

Low- and middle-income

ROU Romania 2015 0.10 0.14 -53.61 -1.59

BGD Bangladesh 2017 0.15 0.13 -26.35 2.27 3.44 -30.85

MDG Madagascar 2012 0.20 0.31 -136.72 -72.92 -29.55 -82.73

BOL Bolivia* 2017 0.31 0.31 -27.95 -148.95 -6.69 -19.93

NPL Nepal 2017 0.39 0.77 1.53 9.82 -23.32 20.15

MEX Mexico 2018 0.51 0.47 -17.23 -23.46 -3.77 -12.96

BGR Bulgaria 2015 0.60 0.39 -35.74 -6.71 1.48 -37.78

SLE Sierra Leone 2014 1.03 1.22 -44.82 0.88 -106.47 29.86

MWI Malawi 2013 1.30 1.33 -23.07 N/A -61.12 23.62

ALB Albania 2013 1.96 1.57 -6.19 -42.99 -6.96 0.72

TZA Tanzania** 2014 2.22 4.66 -15.18 -24.69 21.36 -46.47

TUR Turkey 2017 2.80 3.10 9.95 -10.33 25.24 -20.45

NAM Namibia 2016 4.39 5.43 -27.96 1.97 -69.54 24.52

GMB Gambia 2018 5.74 5.90 -39.90 -12.81 -11.29 -25.71

CRI Costa Rica 2018 10.80 12.82 30.06 9.61 -62.18 56.88

JOR Jordan 2016 33.89 44.34 29.50 17.54 -21.93 42.18

Low- and middle-income 1.07 1.22 -17.34 -23.83 0.61 -17.70 average

Note: The table combines estimates from table 1 and figures 17, 25 and 28. Estimates are based on the working age population only (i.e. adults with ages 16-70) and cover a sample of 49 countries for which micro data are available (see Section 2.2 for description of geographical coverage). EU, high-income, and low- and middle-income estimates are the averages of the European Union, the sample of high-income countries, and low- and middle-income countries, respectively. Averages are weighted by the number of wage employees in each country. N/A = Not available. * the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). Chapter 4 – Simulations 97 Chapter 4

Simulations

X Chapter 4 – Simulations

This chapter of the report focuses on the explained ences in labour market attributes or characteristics part of the migrant pay gap by filtering out the is narrower compared to the unadjusted wage gap. unexplained part and comparing the counterfactual For example, in countries such as Belgium, Cyprus, wage distribution of migrant workers to the wage Luxembourg, and the United States (among the distribution of non-migrant workers. sample of HICs), and Jordan (among the sample of LMICs), the counterfactual wage structure of The report finds that the migrant pay gap would migrant workers shifts almost entirely from the generally stay low if the unexplained part of the far left towards the wage structure of nationals. migrant pay gap is eliminated, and migrant work- In other countries such as Argentina, France, and ers were to receive similar returns to their labour Norway, there appears to be convergence of the market characteristics as nationals. Chapter 4 wage structures of nationals and migrant workers demonstrates that, measures that eliminate the indicating a near disappearance of the wage gap. unexplained part of the migrant pay gap can help to significantly reduce wage inequalities; working poverty among migrant workers, especially migrant 4.2. The migrant pay gap women; as well as the aggregate gender pay gap between men and women across countries. Thus, before and after eliminating in countries where the unexplained part of the the unexplained part of the migrant pay gap is significantly high, eliminating this gap would help enhance skills and jobs match- migrant pay gap ing for men and women migrant workers, and The migrant pay gap would generally stay narrower promote equality as well as economic productivity if migrant workers were to receive similar returns to and development across countries. their labour market attributes or characteristics as nationals. Once attributes and characteristics listed in table 8 are taken into account and any remaining 4.1. The counterfactual wage unexplained pay gap is eliminated, in the sample of structure of migrant workers HICs, the mean migrant pay gap nearly disappears (for example, in Argentina, Belgium, Denmark, Figure A-8 (see Appendix IV) compares the wage Finland, Italy, and Sweden) or reverses in favour structures of migrant workers with that of nationals. of migrant workers (as in Chile, Cyprus, France, The difference between this figure and figure 16 Greece, Hungary, Ireland, Latvia, the Netherlands, (which compares the actual wage distributions of and Spain) (figure 30). It declines substantially migrant workers and nationals), is that figure A-8 but remains largely explained in Austria, Canada, plots both the actual and counterfactual wage Luxembourg, Portugal, Slovenia, Switzerland, and distributions of migrant workers and compares the United States. On average, the migrant pay the two distributions with the wage distribution gap across the sample of HICs declines substan- of nationals. The counterfactual wage distribution tially from approximately 12.6 per cent to 0.2 per reflects the wage structure of migrant workers cent, when the unexplained part is eliminated. if they were equally remunerated as nationals, This implies that, on average, the large positive according to the observable labour market char- migrant pay gap in the sample of HICs is mostly acteristics described in table 8 (that is, taking into unexplained by observed labour market differences account attributes such as, education, experience, between nationals and migrant workers. In the EU, occupation, and gender). the migrant pay gap reverses from about 8.6 per cent to approximately –7.9 per cent, on average. The figure shows that, unlike the actual wage Migrant workers within the EU would in fact earn structure of migrant workers, the counterfactual about 7.9 per cent higher than nationals, on aver- wage distribution of migrant workers tends to be age, if wages were set according to observed labour closer to the wage distribution of nationals in most market characteristics. countries. This implies and reinforces the general trend in section 3.5.2, that the wage gap between Among the sample of LMICs, the pay gap remains nationals and migrant workers explained by differ- negative and partly explained in eight countries Chapter 4 – Simulations 99

X Figure 30: The migrant pay gap before and after eliminating the unexplained part of the pay gap in studied economies (based on mean hourly wages), latest years

High-income countries Low- and middle-income countries

Hungary Sierra Leone Chile Netherlands Spain Namibia Australia France Costa Rica Ireland Croatia Malawi Greece Latvia Cyprus Madagascar Italy Denmark Nepal Argentina Sweden Jordan Belgium United Kingdom Finland Gambia Iceland Switzerland Albania Slovakia Lithuania Bolivia* Malta Estonia Norway Mexico Canada United States Bulgaria Poland Czech Republic Bangladesh Austria Portugal Luxembourg Tanzania** Slovenia EU Turkey High-income 0 0 20 40 20 40 60 –8 0 –6 0 –4 0 –2 0 –80 –60 –40 –20 –160 –140 –120 –100 –140 –120 –100 Migrant pay gap (%) Migrant pay gap (%)

Explained mean migrant pay gap Raw mean migrant pay gap

pay gap is eliminated and migrant workers were to be remunerated equally as nationals based on their labour market attributes and characteristics.

Figure 31 shows the proportion of low-paid workers among the total migrant workers before and after eliminating the unexplained migrant pay penalty. The weighted proportion of low-paid migrant workers would decline by more than 42 per cent (from about 28.4 per cent to 16.4 per cent) in the sample of HICs, by more than half (from about 29.0 per cent to 13.1 per cent) among the EU Member States, and about 8 per cent (from about 21.5 per cent to 19.7 per cent) in the sample of LMICs, with variations across countries, when working poverty is defined by “the proportion of migrant workers earning less than two-thirds of the median hourly wage”. For example, the proportion of low-paid migrant workers would reduce from about 53.6 per cent to 24.4 per cent in Cyprus, from about 28.0 per Manolada, Ilia, Greece – March 3, 2016: Immigrant seasonal farm worker picks and packages strawberries directly into boxes cent in to 1.7 per cent in Portugal, and from about in the Manolada of southern Greece. © shutterstock.com 46.6 per cent to 11.4 per cent in Spain.

When a more extreme measure of working poverty is applied – the proportion of migrant workers (Albania, the Plurinational State of Bolivia, the earning less than half of the median hourly wage Gambia, Madagascar, Malawi, Mexico, Namibia, –, the decline in the proportion of low-paid migrant and Sierra Leone). It remains positive only in workers would be much higher than the former Turkey. It reverses from negative to positive in measure if the unexplained part of the migrant Bangladesh, Bulgaria, and the United Republic of pay gap is eliminated. The proportion of low-paid Tanzania; and from positive to negative in Costa migrant workers would decline, on average, by Rica, Jordan, and Nepal. roughly 49 per cent in the sample of HICs (from about 11.5 per cent to 5.9 per cent), by about 59 per cent in the EU (from around 15.0 per cent to 6.2 per 4.3. Measures to eliminate cent), and about 12 per cent in the sample of LMICs the unexplained part of (from about 13.8 per cent to 12.2 per cent). the migrant pay gap can Table 10 shows the estimates in figure 31, disaggregated by sex. In HICs and in the EU, eliminating help reduce working poverty the unexplained migrant pay gap reduces the among migrant workers proportion of working poverty among both men and women migrant workers, though the decline is This section analyses the effect of eliminating the more profound for migrant women. The incidence unexplained part of the migrant pay gap on low- of low-paid workers among migrant women in the paid migrant workers. The proportion of low-paid sample of HICs and the EU would decline from migrant workers before and after eliminating the about 35.1 per cent to 21.2 per cent and from unexplained part of the migrant pay gap can be about 30.3 per cent to 15.2 per cent on average, looked at using two measures: 1) the proportion of respectively, when working poverty is defined by migrant workers earning less than two-thirds of the earning less than two-thirds of the median hourly median hourly wage; and 2) the proportion earning wage. On the other hand, the incidence of low- less than half of the median hourly wage. The fig- paid workers among migrant men would decline ure and table below demonstrate that the rate of from about 23.9 per cent to 16.1 per cent in the working poverty among migrant workers, and in sample of HICs, and from about 27.5 per cent to particular women migrant workers, would decline 20.4 per cent in the EU, using the same measure of substantially if the unexplained part of the migrant working poverty. Chapter 4 – Simulations 101

X Figure 31: The proportion of working poor among migrant workers before and after eliminating the unexplained part of the migrant pay gap, latest years

Low-paid, as defined by two-thirds Low-paid, as defined by half of the median hourly wage of the median hourly wage Slovakia Czech Republic Hungary Slovakia Portugal Poland Czech Republic Hungary Croatia Canada Poland Portugal Sweden Croatia Spain Sweden Finland Chile Norway Belgium Malta United Kingdom Denmark Australia United Kingdom Malta Belgium France France United States Australia Luxembourg Canada Ireland Ireland Cyprus United States Norway

High-income Lithuania High-income Finland Italy Spain Chile Denmark Switzerland Argentina Luxembourg Greece Argentina Italy Cyprus Switzerland Greece Latvia Austria Lithuania Slovenia Estonia Iceland Austria Netherlands Iceland Latvia Netherlands Estonia Slovenia EU EU High-income High-income Romania Bulgaria Bulgaria Romania Bolivia* Bangladesh Turkey Turkey Bangladesh Bolivia* Gambia Jordan Sierra Leone Gambia Albania Costa Rica Jordan Malawi Mexico Sierra Leone Costa Rica Albania Nepal Nepal w- and middle income w- and middle income Malawi Mexico Lo Lo Tanzania** Tanzania** Namibia Namibia Madagascar Madagascar Low- and middle-income Low- and middle-income 0102030405060 0102030405060 Percent Percent

Proportion of low paid after the elimination Proportion of low paid before the elimination

Note: EU, high-income, and low- and middle-income estimates are the averages of the European Union, the sample of high-income countries, and low- and middle-income countries, respectively. Averages are weighted by the number of wage employees in each country. * the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). 102 The migrant pay gap: Understanding wage differences between migrants and nationals

X Table 10. The share of low-paid workers among migrant workers before and after eliminating the unexplained part of the migrant pay gap, by sex, latest years

A. Using less than two-thirds of the median hourly wage B. Using less than half of the median hourly wage as the definition of low pay as the definition of low pay

Before After Before After

Income Total Women Men Total Women Men Income Total Women Men Total Women Men Country Country group % % % % % % group % % % % % %

Slovakia 0.0 0.0 0.0 0.0 0.0 0.0 Slovakia 0.0 0.0 0.0 0.0 0.0 0.0

Poland 7.3 7.1 7.4 3.0 0.0 32.0 Croatia 0.0 0.0 0.0 2.1 5.0 0.0

Chile 10.1 11.9 8.5 21.8 28.5 15.3 Chile 2.5 2.8 2.1 2.2 2.4 2.0

Australia 10.8 14.9 7.2 15.3 22.6 7.8 Canada 2.6 3.3 2.0 1.0 1.0 0.9

Malta 16.4 21.3 12.0 13.2 12.0 14.7 Australia 2.8 3.7 1.9 4.2 6.9 1.3

Czech Republic 16.5 23.3 11.5 2.1 1.6 3.4 Poland 3.8 7.1 2.3 0.0 0.0 0.1

Croatia 17.0 30.7 2.4 2.2 5.0 0.2 Czech Republic 4.7 8.1 2.3 0.0 0.0 0.0

Lithuania 17.3 29.0 10.9 19.6 3.4 38.6 Belgium 5.3 7.0 3.9 2.4 3.2 1.8

Finland 21.9 22.5 21.6 11.8 14.7 9.1 Portugal 6.4 5.0 8.2 1.0 1.2 0.8

Canada 23.5 28.7 18.5 15.6 18.7 12.7 Malta 6.6 6.1 7.0 4.4 3.6 5.4

Denmark 25.2 28.2 22.3 13.8 11.0 17.9 United States 8.6 13.3 5.7 5.5 8.1 2.9

Switzerland 25.4 32.7 18.7 22.8 28.4 17.2 Finland 8.8 9.0 8.8 8.8 12.9 4.9

Belgium 26.1 30.2 22.8 14.9 20.8 10.3 United Kingdom 9.1 9.4 8.7 2.8 3.2 2.6

France 27.6 27.4 27.7 15.3 8.3 28.6 Lithuania 10.3 20.4 4.8 17.7 0.0 38.5

Portugal 28.0 35.8 18.0 1.7 1.5 1.9 Latvia 12.7 14.5 11.2 16.0 17.7 13.7

United States 28.7 41.6 20.7 18.2 25.8 11.0 Slovenia 13.1 13.9 12.8 28.3 52.0 2.7

Norway 29.2 31.1 27.5 12.9 15.6 10.8 Denmark 13.5 11.9 15.0 9.2 7.5 11.8

Sweden 29.3 31.6 27.5 7.3 2.9 17.6 Argentina 14.9 15.4 14.5 10.1 9.3 10.7 High-income High-income United Kingdom 30.2 32.5 27.9 13.9 14.3 13.6 France 15.3 14.4 16.2 5.2 4.1 7.3

Argentina 32.1 32.3 31.9 23.6 20.7 25.7 Iceland 15.4 17.5 13.2 22.4 43.0 2.0

Latvia 32.3 40.6 25.5 33.1 39.4 24.6 Switzerland 16.0 19.1 13.2 15.2 16.4 14.1

Iceland 36.5 41.2 31.6 30.6 55.6 5.9 Luxembourg 18.5 17.4 19.3 6.3 7.0 5.7

Estonia 38.0 51.5 29.0 33.7 51.5 22.5 Ireland 18.9 21.9 16.1 7.0 6.6 7.6

Ireland 39.0 35.1 42.7 15.9 13.0 19.2 Sweden 19.9 22.0 18.2 2.2 0.8 5.4

Austria 39.1 42.5 36.2 25.2 31.9 17.9 Norway 19.9 23.3 16.8 8.7 11.0 7.0

Netherlands 39.5 35.6 43.9 30.7 62.0 17.1 Greece 19.9 23.4 17.1 10.6 11.3 10.1

Luxembourg 41.7 35.4 46.7 23.0 25.8 20.3 Austria 25.3 24.2 26.2 18.5 23.2 13.4

Italy 43.8 46.4 41.3 21.3 22.9 19.9 Italy 25.4 26.7 24.3 10.7 11.2 10.3

Slovenia 44.8 42.2 45.6 28.9 52.2 3.6 Estonia 26.3 32.9 21.8 18.4 18.1 18.6

Greece 45.5 48.1 43.4 24.8 24.5 25.1 Netherlands 30.3 29.6 31.1 24.7 44.8 16.0

Spain 46.6 46.7 46.6 11.4 6.7 45.7 Cyprus 31.2 40.9 17.5 7.7 9.6 6.0

Cyprus 53.6 61.7 42.1 24.4 19.4 29.3 Spain 32.9 32.8 32.9 9.0 5.5 34.7

EU average 29.0 30.3 27.5 13.1 15.2 20.4 EU 15.0 15.8 14.5 6.2 8.9 8.6

High-income 28.4 35.1 23.9 16.4 21.2 16.1 High-income 11.5 14.1 9.9 5.9 8.4 5.8 average Chapter 4 – Simulations 103

A. Using less than two-thirds of the median hourly wage B. Using less than half of the median hourly wage as the definition of low pay as the definition of low pay

Before After Before After

Income Total Women Men Total Women Men Income Total Women Men Total Women Men Country Country group % % % % % % group % % % % % %

Romania 0.0 0.0 0.0 0.0 0.0 0.0 Bulgaria 0.0 0.0 0.0 0.0 0.0 0.0

Bangladesh 8.8 11.3 8.0 11.3 14.1 10.4 Romania 0.0 0.0 0.0 0.0 0.0 0.0

Bolivia* 10.5 28.1 0.0 6.6 32.0 0.6 Bangladesh 1.5 5.1 0.5 3.0 12.3 0.0

Bulgaria 14.5 23.2 0.0 0.4 0.5 0.0 Bolivia* 5.7 15.4 0.0 5.5 28.6 0.0

Albania 17.3 14.6 18.7 22.0 8.3 26.1 Costa Rica 7.7 11.8 5.2 13.4 13.0 13.7

Costa Rica 22.4 26.2 20.0 31.5 25.9 36.1 Albania 10.8 8.4 12.1 19.8 4.8 24.2

Sierra Leone 25.4 0.0 29.9 19.4 0.0 30.9 Jordan 11.3 24.0 9.3 9.6 3.8 11.9

Turkey 25.6 18.0 28.9 7.6 10.4 7.1 Nepal 13.1 57.4 7.7 20.0 42.3 2.8

Madagascar 26.1 72.9 18.0 49.8 98.4 1.2 Turkey 18.4 14.6 20.0 5.4 6.5 5.2

Mexico 28.6 34.8 25.0 31.3 44.0 22.1 Gambia 18.6 0.9 24.9 10.2 0.7 17.7

Jordan 29.1 56.9 24.7 24.4 6.4 31.3 Malawi 20.0 4.7 27.0 19.3 3.9 31.8 Low- and middle-income Low- Low- and middle-income Low- Gambia 29.4 18.5 33.3 14.1 5.9 20.7 Mexico 21.1 29.8 16.1 20.5 26.1 16.4

Malawi 31.0 18.6 36.6 31.9 19.3 42.0 Sierra Leone 22.5 0.0 26.4 19.4 0.0 30.9

Nepal 35.3 63.0 31.9 31.6 42.3 23.3 Madagascar 26.1 72.9 18.0 49.8 98.4 1.2

Tanzania** 35.7 54.1 26.9 39.0 53.7 31.7 Tanzania** 26.4 34.3 22.5 27.1 30.4 25.5

Namibia 49.9 44.2 53.1 42.8 49.7 37.5 Namibia 40.2 39.6 40.6 39.9 47.4 34.1

Low- and Low- and middle-income 21.5 25.3 19.7 19.7 26.9 15.2 13.8 19.3 11.7 12.2 18.0 9.0 middle-income average

4.4. Measures to eliminate reduce from about 31.2 per cent to approximately 28.0 per cent on average in the sample of HICs, the unexplained part of the from about 30.2 per cent to 29.6 per cent in the EU, migrant pay gap can help and from about 39.3 per cent to 35.3 per cent in the reduce wage inequalities and sample of LMICs (figure 32). the aggregate gender pay gap In terms of the aggregate pay gap between all in the economy men and all women in the economy, measures to eliminate the unexplained part of the migrant pay Measures to eliminate the unexplained part of gap can help reduce the aggregate gender pay gap the migrant pay gap can help reduce overall wage in favour of men across the sample of HICs from inequalities across countries, as well as the aggre- around 16.2 per cent to approximately 11.6 per cent gate gender pay gap in the economy. The Gini when using mean hourly wages, and from about inequality coefficient – which expresses the level 15.7 per cent to 11.6 per cent when using median of wage inequalities within the economy – would hourly wages (figure 33). 104 The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure 32: The Gini coefficient after eliminating the unexplained part of the migrant pay gap, latest year

Slovakia Belgium Czech Republic Denmark United States Australia Greece Croatia Canada Slovenia Italy Malta Hungary Sweden France Norway Spain Switzerland Netherlands

High-income Finland Austria Poland Iceland Cyprus Lithuania Latvia Argentina Luxembourg Portugal Ireland United Kingdom Estonia Chile EU High-income Jordan Bulgaria Bangladesh Albania Turkey Mexico Nepal Bolivia* Costa Rica Gambia Malawi w- and middle income Madagascar Lo Tanzania** Namibia Low- and middle-income 010203040506070 Gini coefficient (%)

Gini coefficient after the elimination Gini coefficient before the elimination

Note: This figure compares the unadjusted Gini coefficient with the Gini coefficient when the unexplained part of the migrant pay gap is eliminated. The Gini provides a comprehensive estimate of within country wage inequality. Data on the unadjusted Gini coefficient is taken from the Global Wage Report 2018/19. EU, high-income, and low- and middle-income estimates are the averages of the European Union, the sample of high-income countries, and low- and middle-income countries, respectively. Averages are weighted by the number of wage employees in each country. * the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). Chapter 4 – Simulations 105

X Figure 33: The aggregate gender pay gap after eliminating the unexplained part of the migrant pay gap (based on mean and median hourly wages), high-income countries, latest year

The gender pay gap based on mean hourly wages The gender pay gap based on median hourly wages

Ireland Netherlands Netherlands Spain Belgium Belgium Luxembourg

Spain Malta

Malta Italy

Poland Latvia Italy United Kingdom Austria Chile Finland

Latvia Estonia

Greece Australia Croatia Poland Sweden Slovenia Denmark

Slovenia Sweden

France France

Estonia Norway Norway Finland Lithuania Luxembourg Australia

Argentina Argentina

Chile Lithuania Cyprus Canada United Kingdom Switzerland Iceland

Portugal Portugal

Canada Slovakia

Slovakia Czech Republic Switzerland United States Czech Republic Cyprus United States

High-income High-income 5 0 5 0 –5 –5 10 15 20 25 10 15 20 25 30 –2 0 –1 5 –1 0 –20 –15 –10 Mean gender pay gap (%) Mean gender pay gap (%)

Aggregate gender pay gap after the elimination Aggregate gender pay gap before the elimination

Note: This figure compares the unadjusted gender pay gaps with the pay gaps when the unexplained part of the migrant pay gap is eliminated. Data on the unadjusted gender pay gap is taken from the Global Wage Report 2018/19. High-income is the average of the sample of high-income countries covered in the report. The average is weighted by the number of wage employees in each country. Source: ILO estimates based on survey data provided by national sources (see Appendix II). Chapter 4 – Simulations 107 Chapter 5

Conclusions

X Chapter 5 – Conclusions

This report is a first attempt to examine differences number of countries. In Chapter 3, the analysis in labour market outcomes of migrant workers and yields opposing results for the sample of HICs on nationals at the global level, including providing the one hand and the sample of LMICs on the other gender disaggregated estimates, and highlight- hand. Possible reasons for the opposing estimates ing the migrant pay gap. The report uses recent may include, among others, the relatively small available data from 49 countries that span the five sample of LMICs covered in the report; the relatively regions of the ILO and which together represent small proportion of migrant workers in LMICs; and about a quarter of wage employees worldwide. The the composition of jobs among migrant workers 49 countries, comprising 33 High-Income Countries in LMICs (for example, the likelihood of a cluster of (HICs) – representing about 56 per cent of all HICs temporary high-skilled ‘expatriate’ workers among as of July 2019 – and 16 Low- and Middle-Income the total migrant population). Countries (LMICs) – representing about 12.4 per cent of all LMICs–, host nearly half (49.4 per cent) of all international migrants and roughly 33.8 per cent of 5.1. Key takeaways migrant workers worldwide. It is important to note, The following points highlight the key findings from however, that the quantitative data on labour mar- the report: ket outcomes, including data on wages of migrant and non-migrant workers used in the analysis in this 1. International migrant workers constitute a report predate the COVID-19 crisis period. significant proportion of the global wage workforce, averaging approximately 9.2 per cent in The report drew from the methodology of the ILO the sample of HICs and 1.2 per cent in the sam- Global Wage Report 2018/19, which provides a ple of LMICs, with sizeable variations across detailed analysis of pay inequalities between men countries. Luxembourg has the largest propor- and women around the world. The main objective tion of migrant wage workers as a share of total of the report is to highlight labour market differ- wage workers with about 47.0 per cent share of ences between migrant workers and nationals, migrant wage workers. Jordan (44.3 per cent), including migrant pay gaps across countries. It Switzerland (29.3 per cent), Australia (26.4 per does so with a view to facilitating the adoption cent) and Canada (24.1 per cent) have the sec- and implementation of evidence-based labour ond, third, fourth and fifth largest shares of migration policies around the world, and ensuring migrant wage workers, respectively (table 1). that these are gender-responsive. It also contributes to the work towards achieving SDG tartgets 2. Although, on average, women (migrants and 8.5 and 8.8, which respectively call for “equal pay non-migrants alike) account for a slightly for work of equal value” and “protected labour higher proportion of the working age popula- rights for all workers, including migrant workers, tion than men across countries, the share of in particular women migrant workers, and those in women among total wage workers is low, on precarious employment.” Findings from the report average, compared to men. While the share of can also help set the basis for monitoring labour migrant women among the total working age market gaps, including wage inequalities, between migrant population is about 50.6 per cent in the migrant workers and non-migrant workers around sample of HICs and 51.3 per cent in the sample the world, and between migrant men and migrant of LMICs, only about 43.0 per cent of migrant women; support the case for closing these gaps wage workers in the sample of HICs are women as enshrined in the dedicated ILO Conventions and 32.0 per cent in the sample of LMICs are concerning migrant workers; and encourage fur- women, although there are notable variations ther research on policies and practices that are across countries. Similarly, while non-migrant effective for promoting change. women account for 50.8 per cent of the total working age population of non-migrants in It is important to emphasize that the conclusions the sample of HICs and 51.6 per cent of the drawn from the report are based on the estimates total working age population of non-migrants obtained by using the available data from a limited in the sample of LMICs, about 47.9 per cent Chapter 5 – Conclusions 109

of the total non-migrant wage workers in the reasons other than employment (for instance, sample of HICs are women whereas 34.2 per for family reunification or humanitarian rea- cent of the total non-migrant wage workers in sons), as well as possible discrimination against the sample of LMICs are women. This finding is migrant women that reduces their employment consistent with results from the the ILO Global opportunities (see, eg, ILO, 2018b; Kapur, 2010; Estimates Report on International Migrant OECD, 2009). Workers (2018b). The finding also reflects the generally low labour force participation rates 4. In 14 of the studied countries where data on of women, particularly migrant women who informality was available, the report finds that are more likely to engage in unpaid care work, a larger share of the active migrant workforce which is a major barrier to their labour force tend to be in informal employment compared participation (ILO, 2018d) (figure 1). to the non-migrant workforce, in particular women migrants. Notably, about 63.2 per cent 3. Similarly to findings from the ILO Global of the non-migrant workforce in the 14 stud- Estimates on International Migrant Workers ied countries are employed in the informal (2018b), this report finds that migrants of economy, compared to about 66.5 per cent working age in the sample of HICs tend to of migrant workers. The gap among wage have higher labour force participation than workers is even wider, with about 50.8 per non-migrants, on average (72.1 per cent and cent of non-migrant wage workers employed 69.0 per cent, respectively), with variations in the informal economy compared to 62.4 per across countries (table 4). Among the sample cent of migrant wage workers. In terms of of LMICs, however, migrants tend to have lower distribution by sex, informal employment is labour force participation than non-migrants, higher among migrant women workers than on average (62.0 per cent and 64.6 per cent, migrant men workers, on average (66.4 per respectively). In terms of distribution by sex, cent and 65.7 per cent, respectively) (table 7). It migrant men tend to have higher labour force is however important to take into account that participation rates than their non-migrant the estimates cover only two HICs (Argentina counterparts, on average, in the sample of HICs and Chile) and 12 LMICs. These countries (83.1 per cent and 74.1 per cent, respectively), host roughly only 5.3 per cent of international but have lower participation rates than their migrants and about 3.0 per cent of migrant non-migrant counterparts in the sample of workers worldwide. LMICs (78.6 per cent and 81.7 per cent, respectively), with some variations across countries. 5. The report shows that, given similar levels of Among women, migrant women tend to have education, the likelihood of attaining semi- or lower labour force participation rates than high-skilled occupations is lower for migrant non-migrant women, on average, in both the workers than for nationals in the sample of samples of HICs (61.3 per cent and 64.0 per HICs. In other words, highly educated migrant cent, respectively) and LMICs (45.9 per cent and workers in the sample of HICs tend to be less 48.4 per cent, respectively) (table 4 and figure likely to obtain jobs in higher paid occupational 3). The higher labour force participation of categories. This finding is consistent with the migrants in HICs, particularly of migrant men, literature that partly attributes observed gaps is consistent with the hypothesis of positive-se- in labour market outcomes of migrant work- lection, whereby it is individuals with stronger ers and nationals in HICs to differences in the labour market potential, economic motives, returns to foreign-acquired education, and to and a desire to work who tend to migrate for skills mismatch among migrant workers. Skills economic purposes (see Chiquiar and Hanson, mismatch translates into migrant workers 2005; McKenzie and Rapoport, 2010; Parey et being concentrated in lower paid occupations. al., 2015). The lower labour force participation Migrants’ skills are often not fully recognized of migrant women relative to migrant men in HICs and migrants frequently resort to con- and non-migrant women in the sample of HICs tinuous work in lower-skilled jobs that do not may be explained by: (i) the fact that migrant account for their higher skills level. In contrast, women are more likely to engage in unpaid migrant workers in the sample of LMICs, on care work, which is a major barrier to women’s average tend to receive better returns to their labour force participation (ILO, 2018c); (ii) and educational endowment, which is consistent the higher likelihood of women to migrate for with the notion that migrant workers from the 110 The migrant pay gap: Understanding wage differences between migrants and nationals

Global North tend to receive a premium for across countries (figure 11). This corroborates their labour market characteristics in the Global the findings of earlier ILO research (ILO, 2016b) South (figures 8 and 9). according to which migrant workers are particularly prone to be employed in non-standards 6. By looking at the distribution of wage workers jobs. Entry through temporary migration by industrial sector, the report finds that, on programmes or individual characteristics are average, migrant wage workers, compared to often one of the reasons. In addition, migrant nationals, are disproportionately represented workers tend to be overrepresented in sectors in the primary sector – agriculture, fishing with traditionally high incidence of non-stan- and forestry – in the sample of HICs (2.5 per dard jobs. As a consequence, migrant workers cent and 1.5 per cent, respectively), while in may also be more likely to suffer from the dis- the sample of LMICs, the proportions of both advantages inherent to non-standards forms groups are similar (10.6 per cent and 10.3 per of employment, a fact that has become more cent, respectively). In the sample of HICs, evident during the COVID-19 pandemic across more migrant wage workers than nationals the world. take up secondary sector jobs – mining and quarry; manufacturing; electricity, gas and 8. Similar to previous ILO research, the report water; and construction – (26.8 per cent and finds that the incidence of part-time work is 20.8 per cent, respectively), while in the sam- higher among migrant workers than non-mi- ple of LMICs, they (migrant wage workers) grant workers in HICs but lower than non-mi- tend to take up fewer secondary sector jobs, grant workers in LMICs, on average. Migrant on average, than nationals (24.9 per cent and workers have slightly higher part-time inci- 32.6 per cent, respectively). However, while dence rates than non-migrant workers in the there is a tendency for fewer migrant work- sample of HICs, on average (15.0 per cent and ers to be employed in the tertiary sector (i.e. 14.6 per cent, respectively), primarily due to the services) than nationals in HICs (70.7 per cent significantly higher incidence of part-time work and 77.7 per cent, respectively), they tend to contracts among migrant women compared to take up more tertiary sector jobs than nationals non-migrant women. While part-time incidence in the sample of LMICs, on average (64.6 per rates of migrant men is slightly lower than that cent and 57.1 per cent, respectively), with few of non-migrant men in the sample of HICs exceptions, including in Costa Rica, the Gambia, (7.7 per cent and 8.3 per cent, respectively), Jordan, Namibia, Nepal, and Turkey. In terms an average gap of 2.2 percentage points exists of distribution by gender, migrant men wage between the part-time rates of migrant women workers tend to work more in the primary and and non-migrant women in HICs (23.8 per cent secondary sectors in the sample of HICs and and 21.6 per cent, respectively), although the the tertiary sector in the sample of LMICs than scale of the difference varies widely across their national counterparts. Similarly, migrant countries. In the sample of LMICs, incidence women wage workers tend to work more in the of part-time work tends to be lower among primary and secondary sectors in the sample migrant workers than among non-migrant of HICs and the primary and tertiary sectors in workers, on average (6.2 per cent and 8.7 per the sample of LMICs than their national coun- cent, respectively), with notable variations terparts (figures 10 and A-2). across countries. Both migrant men and migrant women in LMICs tend to have lower 7. Similar to previous ILO research, the report part-time incidence rates than their national shows that migrant wage workers in both the counterparts, on average (3.9 per cent and samples of HICs and LMICs are, on average, 6.5 per cent of migrant men and non-migrant more likely than nationals to work under tem- men, respectively, and 10.3 per cent and porary contracts (27.0 per cent and 14.9 per 12.0 per cent of migrant women and non-mi- cent, respectively in the sample of HICs, and grant women, respectively), although part-time 42.9 per cent and 41.7 per cent, respectively work is more prevalent among women than in the sample of LMICs), with few exceptions among men in general (figure 12). including Australia, Canada, Chile, Hungary, Ireland, and Latvia (among the sample of 9. Care work remains an important source of HICs); and Bangladesh, Malawi, and Mexico employment for many migrant workers, in (among the sample of LMICs), and variations particular women migrants. The report cor- Chapter 5 – Conclusions 111

roborates previous findings in the ILO Care 14. The report uses education, age, and gender Work and Care Jobs for the Future of Decent Work as factors to remove a significant part of com- report (ILO, 2018d) by showing that migrant position effect in estimating the migrant pay women are more likely than migrant men to gap, effect that is caused by the existence of be employed as care workers, with the propor- clusters in the wage distribution of wage work- tion of migrant women employed in the care ers. This approach results in what is called the economy exceeding that of migrant men in factor-weighted migrant pay gap. Based on this all the studied countries (figures 14 and 15). approach, the report finds that, on average, migrant workers earn about 9.5 per cent less 10. Chapter 3 provides clear evidence of a migrant per hour than non-migrant workers in the pay gap across countries. The findings demon- sample of HICs, but earn about 23.8 per cent strate that, on average, migrant workers more per hour than non-migrant workers in earn about 12.6 per cent less per hour than the sample of LMICs. Relative to the standard non-migrant workers in the sample of HICs, approach (section 3.2), the factor-weighted with notable exceptions and variations across approach produces a migrant pay gap that is countries. However, in the sample of LMICs, lower in the sample of HICs, and wider in the migrant workers tend to earn about 17.3 per sample of LMICs because the latter approach cent more per hour than nationals, on average tend to account for the existence of clusters (figure 17) of few workers, especially migrant workers 11. The report also highlights that migrant women, at certain locations in the wage distribution particularly in HICs tend to pay a double pen- (figures 25 and 27). alty for being both women and migrants as compared to the average migrant worker, 15. Given that migrant workers are concentrated a finding consistent with results from the (clustered) at certain locations in the wage OECDʼs International Migration Outlook 2020 distribution, estimating migrant pay gaps at (OECD, 2020b). Specifically, migrant women different points in the wage distribution can be earn less than migrant men (who in turn earn more informative than using single summary less than non-migrant workers, on average) in measures like the mean or median migrant the sample of HICs. They also earn less than pay gap. The report finds that the migrant pay non-migrant women and even far less than gap varies significantly across the entire wage non-migrant men in the sample of HICs. For distribution for all the studied countries. In example, the pay gap (based on mean hourly some countries, the migrant pay gap may be wages) between non-migrant men and migrant larger at the top end of the wage distribution, women in the sample of 33 studied HICs is whereas in others it may be larger in the middle estimated at around 20.9 per cent, which is or at the bottom end of the wage distribution much wider than the estimated global aggre- (figures A-4 and A-5). gate gender pay gap (based on mean hourly wages) of 16.2 per cent among HICs (figures 21 16. Findings also indicate that a large part of the and 22). migrant pay gap is unexplained by observed 12. Similarly, migrant care workers (both men and differences in the labour market characteristics women) pay a larger wage penalty relative of migrant workers and nationals. That is, a sig- to the average migrant worker in the sample nificantly large migrant pay gap remains across of HICs for which care workers are uniquely countries after accounting for characteristics identified (19.6 per cent and 17.1 per cent, such as education, years of experience, and respectively), with notable exceptions. However, type of work contract (see table 8). The find- for migrant women working in the highly femi- ings show that, about 10 percentage points of nized care sector where work is often underval- the estimated mean hourly migrant pay gap ued, pay is even far less (figure 24). in the sample of HICs (12.6 per cent) remains unexplained, although there are notable excep- 13. Migrant workers have been among the hard- tions among the studied countries. On the est hit by the economic downturn associated other hand, on average, nearly all the –17 per with the COVID-19 pandemic, both in terms of cent estimated mean hourly migrant pay gap employment losses and a decline in earnings in the sampled of LMICs remains unexplained for those who have remained in employment. (figure 29). 112 The migrant pay gap: Understanding wage differences between migrants and nationals

17. Further, in spite of migrant workers’ education is significantly high, eliminating this gap would levels being similar to or higher than that of help enhance skills and jobs matching for men nationals for the same occupation, migrant and women migrant workers, and promote workers earn less than non-migrant work- equality as well as economic productivity and ers within the same occupation in several development across countries. countries, in particular in the sample of HICs 21. Given the significant size of the unexplained (figure A-7). component of the migrant pay gap, measures 18. There is a weak correlation between the share that eliminate this part of the gap would help of migrant’s population and the level of (hourly) to substantially reduce the rate of working pov- wage inequality in countries of destination. erty among migrant workers, especially among Countries with lower levels of (hourly) wage migrant women. By defining working poverty inequality are countries with a notable size (low-paid workers) as “the proportion of work- of migrant population, as a share of the total ers earning less than half of the median hourly working age population (figure 2). wage”, eliminating the unexplained part of the migrant pay gap would reduce the proportion 19. There appears to be no clear correlation of low-paid migrant workers, by roughly 49 per between wage inequalities and the unad- cent in the sample of HICs (from about 11.5 per justed migrant pay gap. However, higher wage cent to 5.9 per cent), by about 59 per cent in inequalities appear to be weakly correlated with the EU (from around 15.0 per cent to 6.2 per higher levels of unexplained (adjusted) migrant cent), and about 12 per cent in the sample of pay gaps (figure A-1). In view of this, policy mea- LMICs (from about 13.8 per cent to 12.2 per sures put in place to narrow the migrant pay cent) (figure 31 and table 10). gap, in particular the unexplained part should include measures that can help reduce overall 22. In addition to reducing the migrant pay gap levels of within-country wage inequalities. and the rate of working poverty among migrant workers, the report shows that measures that 20. Based on a counterfactual wage distribution eliminate the unexplained part of the migrant of migrant workers, the report shows that pay gap can help to reduce wage inequalities, the migrant pay gap would generally stay as well as the aggregate pay gap between low if migrant workers were to receive similar men and women in the economy. The report returns to their labour market characteristics estimates that the Gini inequality coefficient – as nationals. Once labour market characteris- which expresses the level of wage inequalities tics are taken into account and any remaining within the economy – would reduce from about unexplained pay gap is eliminated, among the 31.2 per cent to approximately 28.0 per cent sample of HICs, the migrant pay gap would on average in the sample of HICs, from about nearly disappear in countries like Argentina, 30.2 per cent to 29.6 per cent in the EU, and Belgium, Denmark, Finland, Italy, and Sweden; from about 39.3 per cent to 35.3 per cent in the and would reverse in favour of migrant work- sample of LMICs. The aggregate gender pay ers in Chile, Cyprus, France, Greece, Hungary, gap in favour of men across the sample of HICs Ireland, Latvia, the Netherlands, and Spain. It would decline from around 16.2 per cent to would decline substantially but remain positive approximately 11.6 per cent when using mean in Austria, Canada, Luxembourg, Portugal, hourly wages, and from about 15.7 per cent to Slovenia, Switzerland, and the United States. 11.6 per cent when using median hourly wages On average, the migrant pay gap across the (figures 32 and 33). sample of HICs would decline substantially from approximately 12.6 per cent to about 0.2 per cent if wages were set according to 5.2. Limitations of the report observed labour market characteristics. In the EU, the migrant pay gap would reverse from Foremost, estimates of the migrant pay gap and about 8.6 per cent to about –7.9 per cent, on other labour market differences between migrant average. Among the sample of LMICs, the workers and nationals presented in this report are migrant pay gap would remain negative in based on data that predate the ongoing COVID-19 some countries, while it would be positive in crisis. However, the crisis may widen the labour others (figure 30). Thus, in countries where market differences between migrant workers the unexplained part of the migrant pay gap and nationals, which may in turn further deepen Chapter 5 – Conclusions 113

the migrant pay gaps presented in this report. Therefore, a reassessment of the migrant pay gap would be useful when more recent labour market data covering the COVID-19 pandemic period become available.

The report does not distinguish between permanent and temporary migrant workers and does not consider the tenure of stay of migrant workers. Also, the report does not differentiate between migrant workers of different nationality of origin present in a particular country. Certainly, migrants’ status, origin and length of stay in countries of destination can influence labour market outcomes, including wages. Therefore, further research about labour market differences, including wage differences that may exist between groups of migrant workers of different status, origin or nationality, and with dif- Migrant workers from South and Southeast Asia are seen having their ferent length of stay in their destination countries, lunch break in a garment factory in Jordan. © shutterstock.com is encouraged.

Across the studied countries, a migrant is defined as a person of working age (16–70 years) present Why are migrant workers performing jobs that in a country of measurement who is not a citizen of may be different from nationals? This will require that country. Thus, foreign-born citizens, including further quantitative and qualitative analysis in naturalized citizens are treated equally as native- specific countries. born citizens in this report. The report, however, acknowledges that there may be differences in practice in the way in which foreign-born citizens 5.3. Policy implications are treated in the labour market as compared to native-born citizens. The examination of the differ- and recommendations ences in labour market outcomes for foreign-born and native-born nationals is outside the scope of A major question emerging from the analysis in this this report and is left for future studies. report is, what can be done to progressively reduce migrant pay gaps across countries, particularly in Non-EU migrants and EU migrants within an EU HICs and in some LMICs, including through the country are not distinguished in the report. In other effective implementation of the principle of “equal words, EU nationals who relocate from their home pay for work of equal value”? While there is a range country to another EU country are treated the same of policies and measures that can be taken to as migrants from outside the EU. Analysis of the dif- reduce these pay gaps, the answer to this question ferences in labour market outcomes for EU migrant will necessarily be country specific. This is because workers and non-EU migrant workers within the EU the factors that drive and explain migrant pay gaps is beyond the scope of this report. vary from country to country as well as in different This report highlights differences in returns to edu- parts of the wage distribution. They may also vary cation for migrant workers and nationals within the across migration corridors, where bilateral labour same occupational categories. It does not, however, agreements are negotiated for different wages for explore returns to education within sectors of the a segment of the migrant population depending on economy. Also, the report does not explore dif- the migrants’ countries of origin. ferences in returns to education for migrant men While the report is not intended to provide a detailed and women separately, which is recommended for analysis of the underlying causes of the differences future research. in labour market outcomes of migrant workers and Another important element for future research nationals (the report mainly highlights the stylized is the deepening of analysis at occupational level facts based on recent national labour force survey and sectors of the economy. To what extent are data), some important policy implications have migrant workers concentrated in certain jobs? emerged from the analysis as discussed below. 114 The migrant pay gap: Understanding wage differences between migrants and nationals

Kuala Lumpur, Malaysia – may 06, 2020: Foreign migrant workers line up to do COVID-19 screening. © shutterstock.com

Monitoring the impact of the (see, OECD. 2020b). Recent survey data from ongoing COVID-19 crisis on migrant Mexico and the United States that covers up to the third quarter of 2020 shows that migrant workers workers is important in addressing have been among the hardest hit by the COVID-19 their specific vulnerabilities crisis, both in terms of employment losses and a decline in earnings for those who have remained While estimates presented in this report reflect in employment. In view of these recent changes, periods prior to the COVID-19 crisis, the findings the migrant pay gap estimates presented in this bear enhanced relevance in the face of COVID-19. report are likely to widen during and after the cri- The ongoing worldwide COVID-19 crisis has put sis. Analysis of the social and economic outcomes a spotlight on decent work deficits among men of men and women migrant workers therefore and women migrant workers around the world. remain most relevant in the immediate and long- Experiences from previous economic crises sug- term response to the COVID-19 crisis. As countries gest that the economic downturn associated safeguard their economies during and beyond the with the COVID-19 pandemic may have dispro- pandemic, there is a need to monitor and protect portionate and long-lasting negative effects on the rights of men and women migrant workers. the integration of migrants in their countries of This should include covering migrant workers in destination (OECD, 2020a). Moreover, Fasani and national COVID-19 policy responses, such as ensur- Mazza (2020b) shows that migrant workers in the ing that they are covered by measures relating to EU are more likely to be in temporary employ- wage subsidies, and facilitating their access to ment, earn lower wages and have jobs that are social security, including health care and income less amenable to teleworking during the COVID-19 protection measures. crisis compared to non-migrant workers. Also, the COVID-19 crisis appears to be reverting the trend The report recommends a reassessment of the of progress and jeopardising more than a decade migrant pay gap across countries whenever of progress in migrant labour market inclusion national level labour market data covering the Chapter 5 – Conclusions 115

COVID-19 pandemic period and beyond become ing statistics of international labour migration (see available, in order to measure the impact of the ILO, 2018c), in particular focusing on appropriate COVID-19 crisis on the labour market outcomes of methodologies for capturing and collecting data on migrant workers. the main categories and subcategories of international migrant workers. This is part of the ILO effort Reliable data, including data to improve the collection and production of labour migration statistics at national, regional, and global on wages of migrant workers and levels, as well as the development of international nationals, is needed on other concepts and standards on labour migration statis- regions and countries of destination tics agreed worldwide.

Quality of data is key, notably availability of reliable data on the distribution of wages amongst migrant There is a need to go beyond workers and nationals, in particular for other simple summary measures of the regions and countries of destination not covered in migrant pay gap this report. This would help bridge the existing data gap, for example, with regard to data on migration It is important to go beyond simple summary mea- to Asia and the Arab States (particularly, the Gulf sures of the migrant pay gap (such as the mean or Cooperation Council (GCC) countries) and within median migrant pay gap) in order to understand North Africa, and South-East Asia and the Pacific. the underpinning causes and thus identify the most effective policy measures to reduce the gaps. One feasible option would be to review and modify This can be done by examining in more detail the existing surveys across these countries by intro- respective wage structures of migrant workers and ducing modules specifically related to migrant pay nationals, including their gender dimensions. In gaps into cross-sectional surveys. The use of mod- particular, it is essential to analyze the migrant pay ules to pick up specific information is an extended gap at different locations in the wage distribution practice when collecting survey data, with modules (including decomposing the gap into explained and integrated sporadically to pick up information on unexplained parts) as well as in different sectors a particular population group or particular events. of the economy, and to calculate factor-weighted What the report recommends here is not a module migrant pay gaps, which accounts for composition on matters related to migrant workers only, but the effects in estimating the pay gaps. This will be par- design and subsequent integration of modules that ticularly useful in countries where migrant workers are carefully thought out to cover matters that are cluster in particular sectors and occupations. identified as potential determinants of the migrant pay gap. The migrant pay gap, as in the case of the gender pay gap, is a slowly changing statistic, and Bridging the migrant pay gap for this reason the module could be administered only sporadically, not necessarily yearly. This could will require policies that target be a very cost-effective instrument that potentially different locations in the wage will produce sufficiently rich survey data that would distribution improve the understanding of factors contributing to the migrant pay gap. An important question is whether the migrant pay gap in a particular country is mostly driven In addition to modifying existing surveys that cap- by pay gaps at the bottom, in the middle or at the ture data on migrant workers, it is also important top of the wage distribution; or it is driven by pay to capture activities in the informal economy vis-à- gaps in specific sectors of the economy. Computing vis the formal economy. As this report highlights, migrant pay gaps at different locations in the wage migrant workers, particularly women migrants, distribution as well as in different sectors of the are more likely to work in the informal economy economy has important policy implications. For as compared to nationals. Capturing informality example, a well-designed minimum wage with in labour force surveys, both in LMICs and in HICs broad legal coverage, including in sectors in which would go a long way to provide a more reliable data migrant workers are mainly employed, could reduce source for understanding the migrant pay gap. the migrant pay gap at the lower end of the wage The ILO is currently working towards filling a part of distribution. To maximize the effect of minimum this gap by implementing the Guidelines concern- wages, setting lower wage level for sectors in which 116

migrant workers often predominate such as domes- managerial and professional occupations, which tic work or agriculture should be avoided. Collective offer better paid employment opportunities; agreements that include provisions on equal pay and combating employer prejudice in hiring and and pay transparency could have similar effects at promotion decisions. the middle and upper ends of the wage distribution. Finally, policies and measures that promote training and equal opportunity for upward mobility for Tackling the “unexplained” part migrant workers in the labour market, especially of the migrant pay gap for those with long-term residence, could have a The report finds that in many countries a substantial positive effect on wage levels in senior positions. part of the migrant pay gap remains unexplained Likewise, eliminating discrimination and addressing by differences in labour market characteristics of occupational segregation of migrant workers into migrant workers and nationals. There is, therefore, lower paid occupations and sectors may also help a need to focus attention on adopting and imple- reduce the migrant pay gap. menting in countries, where it is absent: (i) national Measures that promote the formalization of the legislation prohibiting pay discrimination against informal economy – such as extending to all work- migrant workers; and ii) measures that promote ers, including migrant women, the right to a min- equal pay for work of equal value between women imum wage and social security – can also greatly and men, and between migrant and non-migrant benefit migrant workers, especially women, bring- workers. This also includes implementing the full ing them under the umbrella of legal and effective principle of “equal remuneration for work of equal protection and empowering them to better defend value” beyond the narrower concept of “equal their interests. pay for equal work”, in line with ILO standards. Countries need to promote pay transparency to expose pay differences between migrant workers Tackling the “explained” part and nationals occupying similar positions or per- of the migrant pay gap forming work of equal value. Countries may also embrace proactive pay equity laws, which require The decomposition analysis in this report shows employers to regularly examine their compensation that part of the migrant pay gap can be explained practices, assess the gender and migrant pay gaps by differences in labour market characteristics of and take actions to eliminate the gaps that are due migrant workers and nationals, including edu- to discrimination in pay. cation and experience, and the fact that migrant More generally, labour market integration mea- workers are more likely than nationals to work in sures that offer migrant workers access to jobs, low-paid occupations or industries. The importance recognize their foreign qualifications, and provide of these factors varies from country to country, and the needed retraining programmes can help reduce across income groups. Migrant workers in paid skills mismatch. These measures must also counter employment who have higher levels of education discriminatory practices, including with respect than nationals, but receive relatively lower returns to pay, against migrant workers. Labour market to their education – partly due to skills mismatch integration measures can help narrow the unex- and non-recognition of their foreign qualifications plained part of the migrant pay gap and reduce – may benefit from educational or retraining poli- working poverty among migrant workers, especially cies that target men and women migrant workers, among women. This would not only be beneficial particularly in HICs. This may significantly reduce to migrant workers, but also it can be beneficial to the migrant pay gap in countries where migrant businesses and economies of destination countries workers earn significantly less than non-migrant through optimized migrant labour. workers. Reducing polarization and occupational segregation may require a series of measures In any event, reducing the migrant pay gap will including changing social and cultural perceptions require a broader strategy that includes also the and stereotypes contributing to discrimination adoption of fair and effective labour migration pol- against migrants; targeted efforts that create icies that address decent work deficits and ensure better chances for men and women migrant greater coherence across employment, education workers, especially for long-term residents, to and training, and other relevant policies at national, enter into a wider range of occupations, including regional and global levels. 117

X References

Abella, M.; Martin, P. 2014. Migration costs of low- States”, in Journal of Political Economy, Vol. 113, skilled labor migrants: Key findings from pilot sur- pp. 239–281. veys in Korea, Kuwait and Spain. Global Knowledge Chiswick, B.R., Le, A.T., Miller, P.W., 2008. “How Partnership on Migration and Development immigrants fare across the earnings distribution (KNOMAD) (Washington, DC, World Bank). in Australia and the United States”, in Industrial Aleksynska, M.; Kazi Aoul, S.; Petrencu, V. 2017. Labor Relations, Vol. 61, pp. 353–373. Deficiencies in conditions of work as a cost to labor Cho, Y. 2017. International labor mobility, presentation migration: Concepts, extent, and implications, at the jobs and migration core course (Washington KNOMAD (Washington, DC, World Bank). DC, World Bank). Antón, J.-I.; de Bustillo, R.M.; Carrera, M. 2012. Christiaensen, L.; Gonzalez, A.S.; Robalino, D.A. “Raining stones? Female immigrants in the 2019. Migration and jobs: Issues for the 21st cen- Spanish labour market”, in Estudios de economía, tury, Policy Research Working Paper, No. 8867 Vol. 39, Issue 1, pp. 53-86. (Washington, DC, World Bank). Barrett, A.; McGuinness, S.; O’Brien, M. 2012. “The Erel, U.; Ryan, L. 2018. “Migrant capitals: Proposing immigrant earnings disadvantage across the a multi-level spatio-temporal analytical frame- earnings and skills distributions: The case of work”, in Sociology. immigrants from the EU’s new Member States”, in British Journal of Industrial Relations, Vol. 50, Fagan, C.; Norman, H.; Smith, M.; González pp. 457–481. Menéndez, M.C. 2014. In search of good quality part-time employment, Conditions of Work and Borjas, G.J. 1990. Friends or strangers: The impact Employment Working Paper Series No. 43 of immigrants on the U.S. economy (New York, (Geneva, ILO). NY, Basic Books). Faini, R. 2005. “Migration, remittances and growth”, —. 1995. “Assimilation and changes in cohort quality in G.J. Borjas, G.J. and J. Crisp, J. (eds): Poverty, revisited: what happened to immigrant earnings International Migration and Asylum, Studies in in the 1980s?” in Journal of Labor Economics, Development Economics and Policy (London, Vol. 13, pp. 201–245. Palgrave Macmillan), pp. 171–187.

Braňka, J. 2016. “Understanding the potential Farole, T.; Goga, S.; Ionescu-Heroiu, M. 2018. impact of skills recognition systems on labour Rethinking lagging regions: Using cohesion policy markets: Research report”. International Labour to deliver on the potential of Europe’s regions, World Office, Skills and Employability Branch (Geneva, Bank Report on the EU (Washington, DC, World ILO). Bank).

Brunow, Stephan; Jost, Oskar. 2019. “Wages of Fasani, F.; Mazza, J. 2020a. Immigrant Key Workers: migrant and native employees in Germany: Their Contribution to Europe’s COVID-19 Response. new light on an old issue”, No 201910, IAB IZA Policy Paper No. 155 (Bonn, Institute for the Discussion Paper, Institut für Arbeitsmarkt- und Study of Labor (IZA)). Berufsforschung (IAB), Nürnberg [Institute for Employment Research, Nuremberg, Germany]. —. 2020b. A Vulnerable Workforce: Migrant Workers in the COVID-19 Pandemic. EUR 30225 EN, Butcher, K.F.; DiNardo, J. 2002. “The Immigrant and Publications Office of the European Union, Native-Born Wage Distributions: Evidence from Ispra, ISBN 978-92-76-18958-9 (online), United States Censuses”, in The International 978-92-76-18957-2 (print), doi:10.2760/316665 Labour Review, Vol. 56, pp. 97–121. (online),10.2760/914810 (print), JRC120730.

Chiquiar, D.; Hanson, G.H. 2005. “International Flahaux, M.-L.; De Haas, H. 2016. African migration: migration, self‐selection, and the distribution trends, patterns, drivers. Comparative Migration of wages: Evidence from Mexico and the United Studies, 4(1). doi: 10.1186/s40878-015-0015-6. 118 The migrant pay gap: Understanding wage differences between migrants and nationals

Fortin, N.; Lemieux, T.; Firpo, S., 2011. “Decomposition —. 2015c. ILO Global Estimates of Migrant Domestic methods in economics", in O. Ashenfelter and Workers: Results and Methodology (Geneva). D. Card (eds): Handbook of Labor Economics —. 2016a. The cost of migration: What low-skilled (Amsterdam, Elsevier), pp. 1–102. migrant workers from Pakistan pay to work in Saudi Hammouya, M. 1999. Statistics on public sector Arabia and the United Arab Emirates (Islamabad, ILO). employment: Methodology, structures and trends, Bureau of Statistics Working Paper (Geneva, ILO). —. 2016b. Non-standard employment around the world: Understanding challenges, shaping prospects Hazans, M. 2011. Informal workers across Europe: (Geneva). Evidence from 30 European countries, Discussion Paper No. 5871 (Bonn, Institute for the Study of —. 2016c. Women at work: Trends. (Geneva). Labor (IZA)). —. 2016d. Promoting Fair Migration, General Hubbard, B. 2020. Coronavirus Fears Terrify and Survey concerning the migrant workers Impoverish Migrants in the Persian Gulf, New York instruments. Report of the Committee of Times, 13 April. Experts on the Application of Conventions and Recommendations, Report III (Part 1B), Ingelaere, B.; Christiaensen, L.; De Weerdt, J.; International Labour Conference (ILC), 105th Kanbur, R. 2017. Why secondary towns can be Session (Geneva). important for poverty reduction: A migrant’s perspective. Policy Research Working Paper 8193 —. 2017a. Labour migration in Latin America and the (Washington DC, World Bank). Caribbean. Diagnosis, Strategy and ILO’s work in the Region. ILO Regional Office for Latin America and International Labour Office (ILO). 1993. Resolution the Caribbean (Lima). concerning the International Classification of Status in Employment (ICSE) (Geneva). —. 2017b. Addressing governance challenges in a changing labour migration landscape, Report IV, —. 2008. International Standard Classification of International Labour Conference. 106th Session Occupations (ISCO-08) (Geneva). (Geneva). —. 2013. Measuring informality: A statistical manual —. 2017c. How to facilitate the recognition of skills of on the informal sector and informal employment migrant workers? Guide for employment services (Geneva). providers (Geneva). —. 2014a. Global Wage Report 2014/15: Wages and —. 2017d. Common interests, shared goals: Promoting Income Inequality (Geneva). decent work from Asia and Africa to the Middle East. —. 2014b. Minimum wage systems, General Survey Conference paper (Beirut). of the reports on the Minimum Wage Fixing —. 2018a. Global Wage Report 2018/19: What lies Convention, 1970 (No. 131), and the Minimum behind gender pay gaps (Geneva). Wage Fixing Recommendation, 1970 (No. 135), Report of the Committee of Experts on the —. 2018b. ILO Global Estimates Report on International Application of Conventions and Recommendations, Migrant Workers: Results and Methodology, 2nd ed. Report III (Part 1B), International Labour (Geneva). Conference (ILC), 103rd Session (Geneva). —. 2018c. Guidelines concerning statistics on inter- —; OECD; World Bank. 2015. The contribution national labour migration, 20th International of labour mobility to economic growth, Joint Conference of Labour Statisticians Geneva, paper presented at the G20 Labour and 10–19 October 2018, 20/2018/Guidelines Employment Ministers’ Meeting, Ankara, Turkey, (Geneva). 3–4 September. 2015, pp. 3–4. —. 2018d. Care work and care jobs for the future of —. 2015a. Labour Market Trends Analysis and Labour decent work (Geneva). Migration from South Asia to Gulf Cooperation Council Countries, India and Malaysia. Report —. 2018e. World Employment and Social Outlook: (Kathmandu, Nepal). Trends 2018 (Geneva).

—. 2015b. Women in Business and Management —. 2018f. Women and men in the informal economy: Gaining Momentum (Geneva). A statistical picture. 3rd ed. (Geneva). References 119

—. 2018g. Guidelines concerning measurement of —; —. 2011. “International migration, ethnicity, and qualifications and skills mismatches of persons in economic inequality”, in W. Salverda, B. Nolan employment, 20th International Conference of and T. M. Smeeding (eds): Oxford Handbook of Labour Statisticians Geneva, 10–19 October 2018 Economic Inequality (Oxford, Oxford Handbooks). (Geneva). Kapur, R. 2010. Makeshift migrants and law: Gender, —. 2019a. General principles and operational guide- belonging, and postcolonial anxieties (London, lines for fair recruitment and definition of recruit- Routledge). ment fees and related costs (Geneva). King-Dejardin, A. 2019. The social construction of —. 2019b. Guiding labour migration and labour mobil- migrant care work. At the intersection of care, migra- ity in Africa (Geneva). tion and gender. International Labour Office, Labour Migration Branch and Gender, Equality —. 2019c. Operational Manual on Recruitment Costs and Diversity & ILOAIDS Branch (Geneva, ILO). – SDG 10.7.1 (Geneva). Lodigiani, R.; Sarli, A. 2017. “Migrants’ competence —. 2020a. [Forthcoming]. A Global comparative recognition systems: Controversial links between study on defining recruitment fees and related costs: social inclusion aims and unexpected discrimi- Inter-regional research on law, policy and practice nation effects”, in European Journal for Research (Geneva). on the Education and Learning of Adults, Vol. 8, —. 2020b. Impact of COVID-19 on migrant workers in Issue 1, pp. 127-144. Lebanon and what employers can do about it, press Lucas, R.E.B. 2015. Internal migration in developing release, 14 April. economies: An overview, KNOMAD Seminar Series —. 2020c. COVID-19: Impact on migrant workers (Washington, DC, World Bank). and country response in Malaysia, briefing note, Mahroum, S. 2000. “Highly skilled globetrot- 08 May. ters: Mapping the international migration of —. 2020d. Protecting migrant workers during the human capital”, in Research and Development COVID-19 pandemic: Recommendations for Policy- Management, Vol. 30, pp. 23–32. makers and Constituents, policy brief, 30 April Martin, P. 2016. What do migrant workers pay for for- (Geneva). eign jobs? KNOMAD Data and SDG Indicator 10.7.1, —. 2020e. Experiences of ASEAN migrant workers background paper presented at Improving Data during COVID-19: Rights at work, migration and on International Migration Towards Agenda 2030 quarantine during the pandemic, and re-migration and the Global Compact on Migration confer- plans, briefing note, 3 June (Bangkok). ence, Berlin, 2–3 Dec. 2016 (Geneva, IOM).

—. 2020f. Impact of the COVID-19 crisis on loss of jobs McKenzie, D.; Rapoport, H. 2010. “Self-selection and hours among domestic workers, fact sheet, patterns in Mexico-U.S. migration: The role of 15 June (Geneva). migration networks”, in The Review of Economics and Statistics, Vol. 92, pp. 811–821. —. 2020g. Impact of lockdown measures on the informal economy, briefing note, 05 May (Geneva). Mincer, J. 1974. Schooling, experience, and earnings (Washington, DC, National Bureau of Economic —. 2020h. Beyond contagion or starvation: Giving Research). domestic workers another way forward, fact sheet, Mohapatra, S.; Ratha, D. 2010. “Forecasting migrant 05 May (Geneva). remittances during the global financial crisis”, in —. 2020i. The COVID-19 response: Getting gender Migration Letters, Vol. 7, pp. 203–213. equality right for a better future for women at work, Nohl, A.-M.; Schittenhelm, K.; Schmidtke, O.; Weiß, 11 May (Geneva). A. 2006. “Cultural capital during migration: A International Organization for Migration (IOM). multi-level approach for the empirical analysis 2018. World Migration Report (Geneva). of the labor market integration of highly skilled migrants”, in Forum: Qualitative Social Research, Kahanec, M.; Zimmermann, K.F. 2008. International Vol. 7. No. 3, Art. 14. migration, ethnicity and economic inequality, Discussion Paper No. 5871 (Bonn, Institute for Ohlert, C.; Beblo, M.; Wolf, E. 2016. “Competition, the Study of Labor (IZA)). Collective Bargaining, and Immigrant Wage 120 The migrant pay gap: Understanding wage differences between migrants and nationals

Gaps Within German Establishments”, WiSo-HH United Nations. 2008. International Standard Working Paper Series No. 35. Industrial Classification of All Economic Activities, Revision 4 (New York, NY, Department of Organisation for Economic Cooperation and Economic and Social Affairs). Development (OECD). 2009. Is informal normal? Towards more and better jobs in developing coun- —. 2017a. The Sustainable Development Goals (SDGs). tries (Paris). (New York, NY). —; ILO. 2018. How Immigrants Contribute to —. 2017b. International Migration Report (New York, Developing Countries’ Economies, OECD Publishing, NY). Paris. United Nations Conference on Trade and —. 2020a. Managing international migration under Development (UNCTAD). 2012. The Least COVID-19, OECD Publishing, Paris. Developed Countries Report 2012: Harnessing Remittances and Diaspora Knowledge to Build —. 2020b. International Migration Outlook 2020, Productive Capacities (New York and Geneva, OECD Publishing, Paris. UNCTAD). Parey, M.; Ruhose, J.; Waldinger, F.; Netz, N. 2015. The United Nations, Department of Economic and Social selection of high-skilled migrants. IZA Discussion Affairs (UN-DESA). 2019. International migrant Paper No. 9164 (Bonn, Institute for the Study of stock. (New York, NY). Labor (IZA)). Van Bastelaer, A.; Lemaître, G.; Marianna, P. 1997. Pilling, D.; England, A. 2020. Saudi Arabia repatriating The definition of part-time work for the purpose of thousands of migrants back to Ethiopia: Such mass international comparisons. OECD Labour Market deportations likely to exacerbate spread of COVID-19, and Social Policy Occasional Paper, No. 22 (Paris, UN warns, Irish Times, 12 April. OECD). Rubery, J. 2003. Pay equity, minimum wage and equal- World Bank. 2011. Leveraging migration for Africa: ity at work, InFocus Programme on Promoting the Remittances, skills and investments (Washington, Declaration on Fundamental Principles and Rights DC). at Work, Working Paper No. 19 (Geneva, ILO). —. 2018a. Moving for prosperity: Global migration and Siebers, Hans; van Gastel, Jilles. 2015. “Why migrants labor markets (Washington, DC). earn less: in search of the factors producing the ethno-migrant pay gap in a Dutch public orga- —. 2018b. Migration and remittances: Recent develop- nization” in Work, Employment & Society, Vol. 29, ments and outlook – Transit migration, Migration Issue 3, pp. 371–391. and Development Brief, April (Washington, DC). Seshan, G. 2017. Dynamics of migration cost, World —; ILO. 2019. Statistics for SDG indicator 10.7.1: Bank Jobs and Migration Core Course. (Washington Guidelines for their Collection (Geneva, ILO). DC, World Bank). Sparreboom, T.; Tarvid, A. 2017. Skills mismatch of natives and immigrants in Europe (Geneva, ILO). References 121 Appendices

X Appendix I. Description of methods

migrant pay gap. The third step, which has to do The migrant pay gap with the application of unconditional quantile The (raw) migrant pay gap at the mean or quantile v regression to further identify how an individual’s of the wage distributions of migrant workers and labour market characteristics contribute to the for- nationals of country i is estimated as how much the mation of the migrant pay gap is beyond the scope country’s migrant workers’ earnings fall short of the of this report. earnings of its nationals at that mean or quantile in percentage terms. This can be expressed in level Step 1: Identifying the form as counterfactual wage distribution v v PAY(NATIONALS) i – PAY(MIGRANTS) i ∆v i = v x 100 The counterfactual wage distribution for migrant PAY(NATIONALS) i workers is the wage structure that would be and in logarithm form as observed among migrant workers if they received the same returns as nationals to their (migrant ∆v v v workers’) labour market endowments and attri- i = log [ PAY(NATIONALS) i ] – log [ PAY(MIGRANTS) i ], butes. Fortin, Lemieux and Firpo (2011) propose the use of a “weighting factor” to elicit such a counter- v where ∆i is the migrant pay gap at the mean or factual distribution. Intuitively, the weighting factor quantile v of the wage distribution of country i. assigns higher weights to migrant wage workers whose endowments and attributes make them more similar to nationals in the labour market, Decomposing the migrant while migrant workers whose characteristics make pay gap them less similar to nationals in the labour market are assigned a lower weight. Chapter 3 of the report follows the ILO’s Global For each wage worker i in the sample, we observe Wage Report 2018/19 and adapts the method a set of indicator ( X ) that describes the charac- proposed by Fortin, Lemieux and Firpo (2011) to teristics of nationals ( Ti = 1) and migrant workers identify, measure and decompose the explained ( Ti = 0) in the labour market; for example, X can and unexplained components of the migrant pay include age, education, contractual arrangements, gap. The decomposition exercise of Fortin, Lemieux gender, place of work, among others (see Table 8). and Firpo (2011) attributes a weight to each of the The information can be used to estimate the prob- variables that are assumed as determinants of the ability of having a particular set of attributes, where migrant pay gap and consists of three steps. Step 1 a wage employee is a national, that is P ( X|T = 1) estimates a counterfactual wage distribution for or a migrant worker, that is, P ( X|T = 0). It can be migrant workers – that is, the wage distribution that shown that would characterize migrant workers if they were to receive the same returns to their labour market P ( T = j|X ) P ( X|T = j) = , for j = 0,1 characteristics as nationals. Step 2 consists in using P ( T = j ) the counterfactual wage distribution to separate where P ( T = j ) = P ( j ) simply indicates the probability the explained and unexplained parts of the migrant of being a national ( j = 1) or a migrant worker ( j = 0) pay gap at each quantile of the pay distribution (in in the population. Based on this, the individual’s this report, the hourly wage distribution). The third specific weighting factor ω( ) can be constructed as and final step consists in applying unconditional i follows: quantile regression to estimate the weight attached to each variable that contributes to determining the P ( T = 1|X) P (0) ω i . i = (1) migrant pay gap. P ( Ti = 0|X) P (1)

What follows aims to provide a heuristic under- The terms P ( Ti = 1|X) and P ( Ti = 0|X) in expres- standing of steps 1 and 2, with reference to the sion (1) can be regarded as propensity scores and Appendix I. Description of methods 123

can be estimated using either a probit or a logit In practice, once the re-weighting factor has been specification. The estimation of either one of these estimated, standard software packages can be specifications produces coefficients for each of the employed to draw distributional statistics – for variables in X. These coefficients can be employed example quantiles – directly by simply applying to project the conditional probabilities for each the appropriate weights to the wages of nationals national and each migrant wage employee in the (wi, i∈n(N )), migrant workers (wi, i∈n( M )) and the sample. Thus, an estimate of P ( Ti = 1|X ) projects counterfactual ((ωi . wi), i∈n( M )), respectively. What the conditional probability of being a national for is more important is to make sure that the corre- each wage employee (that is, for both migrant sponding propensity scores are well approximated workers and nationals). When the estimated value by including as much information as possible (indi- of P ( Ti = 1|X) is high for a migrant worker this cators in X and several interaction terms between means that his/her labour market attributes are them). This should guarantee that the counterfac- very similar to nationals who are wage employees tual wage distribution of migrant workers is well in the population. It also means that the weighting captured by the re-weighting factor. factor constructed using expression (1) will be high. In summary, once the re-weighting factor is con- Once the weighting factor has been constructed, structed it is possible to draw quantiles from each this can be used to “re-weight” the wages observed of the three empirical wage distributions, namely, N for migrant wage workers. These re-weighted val- from that of nationals (qv ), from that of migrant M ues – where the wages of migrant workers who are workers (qv ) and from that of the counterfactual C more similar to nationals are given a higher weight, wage distribution of migrant workers (qv ). The suffix and those of migrant workers less similar to nation- “v” indicates each one of the deciles of the hourly als are given a lower weight – serve to construct wage distribution, that is, v = {1, 2, 3,…, 9, 10} as an empirical distribution that emulates the wage displayed in figures A-4 and A-6 (Appendix IV). structure of migrant workers if they had received the same returns to their labour market characteristics as nationals. This is what is referred to as a Step 2: Using the counterfactual counterfactual wage distribution. Thus, if the cumu- wage distribution to identify lative density function for migrant wage employees ( M ) in the population can be expressed as the explained and unexplained components of the migrant pay gap

FM ( y) = wi . I { Yi ≤ y } g ∑ i∈n(M) Let yi be the natural logarithm of wages (e.g. hourly wages) observed for group g in the population, where Yi denotes the wage of a migrant worker where g = N, M, C follows the notation outlined in i∈n( M ) in the sample of migrant workers n(M), the previous step. Drawing quantiles from each of and wi is the population (frequency) weight, then the three distributions of the natural logarithmic the counterfactual wage distribution for migrant transformation, the (raw) migrant pay gap at the workers can be similarly expressed as: vth quantile (∆v ) can be expressed as follows:

Fc ( y) = (ωi . wi ) . I { Yi ≤ y } ∑ i∈n(M) v N M ∆ = qv – qv (2)

This shows how the re-weighting factor enters the estimation of the counterfactual wage distribution Expression (2) shows the distance between two for migrant wage employees. Likewise, we can esti- quantiles that have been drawn from two wage dis- mate the cumulative distribution function for wages tributions of the (natural logarithms of) wages: that N M of nationals as, of nationals (qv ) and that of migrant workers (qv ). We can also draw the vth quantile from the counter- C factual distribution of migrant workers, that is, (qv ). FN ( y) = wi . I { Yi ≤ y } ∑ i∈n(N) This represents the hourly wage at that quantile that migrant workers would earn if they were to where in the case of nationals, Yi denotes the wage receive the same returns as nationals for having of a national i∈n( N ) in the sample of nationals who similar labour market endowments and attributes. are wage employees n( N ), and wi is the population Using this counterfactual quantile, the following can (frequency) weight. be constructed: 124 The migrant pay gap: Understanding wage differences between migrants and nationals

∆v N C C M ∆v ∆v = qv – qv + qv – qv = X + U (3) to differences between nationals’ and migrant EXPLAINED UNEXPLAINED workers’ wage structures once we control for PART PART = COMPOSITION = STRUCTURAL differences in their labour market characteristics. EFFECT EFFECT Since the unexplained part is due to a difference in wage structures, ∆v is also referred to as the Since the counterfactual emulates what migrant U “structural effect”. workers should receive as returns for sharing the same endowments and attributes as nationals, In practical terms, the decomposition of the migrant the distance between what nationals get and what pay gap as expressed in equation (3) requires the migrant workers should receive if they have the following stages. First, transformation of wages in same endowments and attributes as nationals is the sampled population into logarithmic scales. explained by any differences in endowments and Second, construction of the re-weighting factor as v labour market characteristics. This is why ∆X is described in (1). Third, appropriate application of referred to as the “explained” part of the migrant weights, that is, allowing for the re-weighting factor pay gap, also known as the migrant pay gap due on migrant workers’ wages to draw the (logarith- to “composition effects”. On the other hand, the mic) wage distribution for nationals, migrant work- distance between what migrant workers should ers and the counterfactual. Fourth, drawing the receive (for their endowments and attributes and quantiles of interest. Fifth and final stage, applying as emulated by the counterfactual) and what they the simple distance as expressed in (2) to estimate actually get (for these endowments and attri- the (raw) migrant pay gap, and its decomposition v butes) cannot be explained. This is the part ∆U that as expressed in (3), at each selected quantile of the remains “unexplained”, that is, the part that is due hourly wage distribution. 125

X Appendix II. National data sources

Latest years Country Subregion for which data Data type Data source is available

Albania Northern, southern and Western 2013 Labour force survey Instituti i Statistikave Albania Europe (INSTAT)

Argentina Latin America and the Caribbean 2018 Encuesta Permanente de NSO – latest data from ILO Hogares repository or SIALC (Sistema de información y análisis Laboral de América Latina y el Caribe)

Australia South-East Asia and the Pacific 2017 Household, Income and Labour Melbourne Institute of Statistics, Dynamics in Australia The University of Melbourne

Austria Northern, southern and Western 2015 EU-SILC Eurostat* Europe

Bangladesh South Asia 2017 Labour force survey Bangladesh Bureau of Statistics

Belgium Northern, southern and Western 2015 EU-SES Eurostat* Europe

Bolivia (Plurinational Latin America and the Caribbean 2017 Encuesta Continua de Empleo NSO – latest data from ILO State of) repository or SIALC

Bulgaria Eastern Europe 2015 EU-SES Eurostat*

Canada North America 2018 Labour force survey NSO – data from ILO repository

Chile Latin America and the Caribbean 2017 Encuesta Nacional de Empleo NSO – latest data from ILO repository or SIALC

Costa Rica Latin America and the Caribbean 2018 Encuesta Continua de Empleo NSO – latest data from ILO repository or SIALC

Croatia Northern, southern and Western 2015 EU-SILC Eurostat* Europe

Cyprus Central and West Asia 2015 EU-SES Eurostat*

Czech Republic Eastern Europe 2015 EU-SES Eurostat*

Denmark Northern, southern and Western 2015 EU-SILC Eurostat* Europe

Estonia Northern, southern and Western 2015 EU-SES Eurostat* Europe

Finland Northern, Southern and Western 2015 EU-SES Eurostat* Europe

France Northern, southern and Western 2015 EU-SES Eurostat* Europe

Gambia Sub-Sahara Africa 2018 Labour force survey Gambia Bureau of Statistics

Greece Northern, southern and Western 2015 EU-SILC Eurostat* Europe

Hungary Eastern Europe 2015 EU-SES Eurostat*

Iceland Northern, southern and Western 2015 EU-SILC Eurostat* Europe

Ireland Northern, southern and Western 2015 EU-SILC Eurostat* Europe

Italy Northern, southern and Western 2015 EU-SES Eurostat* Europe

Jordan Arab States 2016 Labour force survey NSO – latest data from ILO repository or SIALC

Latvia Northern, southern and Western 2015 EU-SES Eurostat* Europe 126 The migrant pay gap: Understanding wage differences between migrants and nationals

Latest years Country Subregion for which data Data type Data source is available

Lithuania Northern, southern and Western 2015 EU-SES Eurostat* Europe

Luxembourg Northern, southern and Western 2015 EU-SES Eurostat* Europe

Madagascar Sub-Sahara Africa 2012 Enquête Nationale sur l'Emploi Institut National de la et le Secteur Informel (ENESI) Statistique, Ministry of Economy of Madagascar

Malawi Sub-Sahara Africa 2013 Labour force survey National Statistical Office of Malawi; Ministry of Labour

Malta Northern, southern and Western 2015 EU-SES Eurostat* Europe

Mexico Latin America and the Caribbean 2018 Encuesta Nacional de Instituto Nacional de Ocupación y Empleo Estadísticas y Géographia de México (INEGI)

Namibia Sub-Sahara Africa 2016 Labour force survey Namibia Statistics Agency

Nepal South Asia 2017 Labour force survey Central Bureau of Statistics

Netherlands Northern, southern and Western 2015 EU-SES Eurostat* Europe

Norway Northern, southern and Western 2015 EU-SES Eurostat* Europe

Poland Eastern Europe 2015 EU-SES Eurostat*

Portugal Northern, southern and Western 2015 EU-SES Eurostat* Europe

Romania Eastern Europe 2015 EU-SES Eurostat*

Sierra Leone Sub-Sahara Africa 2015 Labour force survey Government of Sierra Leone

Slovakia Eastern Europe 2015 EU-SES Eurostat*

Slovenia Northern, southern and Western 2015 EU-SES Eurostat* Europe

Spain Northern, southern and Western 2015 EU-SES Eurostat* Europe

Sweden Northern, southern and Western 2015 EU-SES Eurostat* Europe

Switzerland Northern, southern and Western 2016 Swiss Household Panel Survey Swiss Federal Statistics Office Europe

United Republic of Sub-Sahara Africa 2014 Integrated labour force survey National Bureau of Statistics Tanzania

Turkey Central and Western Asia 2017 Turkish Labour force survey; Turkish Statistical Institute

United Kingdom Northern, southern and Western 2015 EU-SES Eurostat* Europe

United States North America 2018 Current Population Survey Bureau of Labor Statistics

Note: EU-SILC: EU Statistics on Income and Living Conditions; INSTAT = Instituti i Statistikave Albania; NSO = National Statistics Office; EU-SES= Structure of Earnings survey, SIALC = Sistema de información y análisis Laboral de América Latina y el Caribe * Part of this report is based on data from Eurostat. We acknowledge and thank Eurostat for providing data from the Structure of Earnings Survey under contract number RPP 252/2015-SES-ILO, and data from EU-SILC under contract number 52/2013-EU-SILC. The responsibility for all conclusions drawn from these data lies entirely with the authors. 127

X Appendix III. Country and territory groups, by region and income

ILO country groupings by region

Region Subregion - broad Countries

Africa North Africa (6) Algeria, Egypt, Libya, Morocco, Sudan, Tunisia

Sub-Saharan Africa (47) Angola, Benin, Botswana, Burkina Faso, Burundi, Cameroon, Cabo Verde, Central African Republic, Chad, Comoros, Congo, Democratic Republic of the Congo, Côte d’Ivoire, Djibouti, Equatorial Guinea, Eritrea, Ethiopia, Gabon, the Gambia*, Ghana, Guinea, Guinea-Bissau, Kenya, Lesotho, Liberia, Madagascar*, Malawi*, Mali, Mauritania, Mauritius, Mozambique, Namibia*, Niger, Nigeria, Rwanda, São Tomé and Principe, Senegal, Sierra Leone*, Somalia, South Africa, South Sudan, Swaziland, United Republic of Tanzania*, Togo, Uganda, Zambia, Zimbabwe

Americas Latin America and the Caribbean (31) Argentina*, the Bahamas, Barbados, Belize, Plurinational State of Bolivia*, Brazil, Chile*, Colombia, Costa Rica*, Cuba, Dominican Republic, Ecuador, El Salvador, Guadeloupe, Guatemala, Guyana, Haiti, Honduras, Jamaica, Martinique, Mexico*, Netherlands Antilles, Nicaragua, Panamá, Paraguay, Peru, Puerto Rico (unincorporated territory of the United States), Suriname, Trinidad and Tobago, Uruguay, Bolivarian Republic of Venezuela

North America (2) Canada*, United States*

Arab States Arab States (12) Bahrain, Iraq, Jordan*, Kuwait, Lebanon, Occupied Palestinian Territory, Oman, Qatar, Saudi Arabia, Syrian Arab Republic, United Arab Emirates, Yemen

Asia and the Pacific East Asia (8) China, Hong Kong (China), Japan, Democratic People's Republic of Korea, Republic of Korea, Macau (China), Mongolia, Taiwan (China)

South-East Asia and the Pacific (22) Australia*, Brunei Darussalam, Cambodia, Fiji, French Polynesia, Guam (United States), Indonesia, Lao People's Democratic Republic, Malaysia, Myanmar, New Caledonia (France), New Zealand, Papua New Guinea, Philippines, Samoa, Singapore, Solomon Islands, Thailand, Timor-Leste, Tonga, Vanuatu, Viet Nam

South Asia (9) Afghanistan, Bangladesh*, Bhutan, India, Islamic Republic of Iran, Maldives, Nepal*, Pakistan, Sri Lanka

Europe and Central Asia Northern, southern and Western Europe (30) Albania*, Austria*, Belgium*, Bosnia and Herzegovina, Channel Islands (United Kingdom), Croatia*, Denmark*, Estonia*, Finland*, France*, Germany, Greece*, Iceland*, Ireland*, Italy*, Latvia*, Lithuania*, Luxembourg*, Malta*, Montenegro, Netherlands*, Norway*, Portugal*, Serbia, Slovenia*, Spain*, Sweden*, Switzerland*, The Former Yugoslav Republic of Macedonia, United Kingdom*

Eastern Europe (10) Belarus, Bulgaria*, Czech Republic*, Hungary*, Poland*, Republic of Moldova, Romania*, Russian Federation, Slovakia*, Ukraine

Central and Western Asia (11) Armenia, Azerbaijan, Cyprus*, Georgia, Israel, Kazakhstan, Kyrgyzstan, Tajikistan, Turkey*, Turkmenistan, Uzbekistan

*Countries covered in the report 128 The migrant pay gap: Understanding wage differences between migrants and nationals

Country groupings by income level

Developed countries Emerging countries Emerging countries Developing countries (High-income) (Upper-middle income) (Lower-middle income) (Low income)

Andorra Albania Angola Afghanistan

Antigua and Barbuda Algeria Armenia Benin

Australia Argentina Bangladesh Burkina Faso

Austria Azerbaijan Bhutan Burundi

Bahamas Belarus Bolivia, Plurinational State of Central African Republic

Bahrain Belize Cambodia Chad

Barbados Bosnia and Herzegovina Cameroon Comoros

Belgium Botswana Cabo Verde Congo, Democratic Republic of the

Brunei Darussalam Brazil Congo Eritrea

Canada Bulgaria Côte d'Ivoire Ethiopia

Channel Islands China Djibouti Gambia

Chile Colombia Egypt Guinea

Cyprus Cook Islands El Salvador Guinea-Bissau

Czech Republic Costa Rica Eswatini Haiti

Denmark Croatia Georgia Korea, Democratic People's Republic of

Estonia Cuba Ghana Liberia

Finland Dominica Guatemala Madagascar

France Dominican Republic Honduras Malawi

French Guiana Ecuador India Mali

French Polynesia Equatorial Guinea Indonesia Mozambique

Germany Fiji Jordan Nepal

Greece Gabon Kenya Niger

Greenland Grenada Kiribati Rwanda

Guam Guadeloupe Kyrgyzstan Senegal

Hong Kong, China Guyana Lao People's Democratic Republic Sierra Leone

Hungary Iran, Islamic Republic of Lesotho Somalia

Iceland Iraq Mauritania South Sudan

Ireland Jamaica Micronesia, Federated States of Tanzania, United Republic of

Israel Kazakhstan Moldova, Republic of Togo

Italy Lebanon Mongolia Uganda

Japan Libya Morocco Zimbabwe

Korea, Republic of Malaysia Myanmar

Kuwait Maldives Nicaragua

Latvia Marshall Islands Nigeria

Liechtenstein Mauritius Occupied Palestinian Territory

Lithuania Mexico Pakistan Appendix III. Country and territory groups, by region and income 129

Developed countries Emerging countries Emerging countries Developing countries (High-income) (Upper-middle income) (Lower-middle income) (Low income)

Luxembourg Montenegro Papua New Guinea

Macau, China Namibia Philippines

Malta Nauru São Tomé and Principe

Martinique Panama Solomon Islands

Monaco Paraguay Sri Lanka

Netherlands Peru Sudan

Netherlands Antilles Romania Syrian Arab Republic

New Caledonia Russian Federation Tajikistan

New Zealand Saint Lucia Timor-Leste

Norway Saint Vincent and the Grenadines Tunisia

Oman Samoa Ukraine

Palau Serbia Uzbekistan

Poland South Africa Vanuatu

Portugal Suriname Viet Nam

Puerto Rico Thailand Western Sahara

Qatar The Former Yugoslav Republic of Yemen Macedonia

Réunion Tonga Zambia

Saint Kitts and Nevis Turkey

San Marino Turkmenistan

Saudi Arabia Tuvalu

Seychelles Venezuela, Bolivarian Republic of

Singapore

Slovakia

Slovenia

Spain

Sweden

Switzerland

Taiwan, China

Trinidad and Tobago

United Arab Emirates

United Kingdom

United States

United States Virgin Islands

Uruguay 130

X Appendix IV. Supplementary results

X Figure A-1. The Gini coefficient and the mean migrant pay gap using hourly wages from HICs, latest years

The Gini vs. the unadjusted mean migrant pay gap The Gini vs. the unexplained mean migrant pay gap

60 60 y = –0.06x + 11 y = 0.37x + 1 HUN 2 2 R R = 0.0005 50 = 0.0084 50 ESP NLD CYP 40 40 CHL 30 ITA IRL GRC 30 DNK LVA 20 ARG EST SWE BEL ISL FRA 20 10 PRT AUT NOR AUS HRV LUX USA 0 SVN 10 GBR CAN CHE –10 0 MLT SVK LTU –20 al mean migrant pay gap (%) al mean migrant POL lained mean migrant pay gap (%) lained mean migrant t

–10 p CZE

To –30

–20 Un ex –40 15 20 25 30 35 40 45 15 20 25 30 35 40 45

Gini coefficient (%) Gini coefficient (%)

Notes: Data on the Gini coefficient is taken from the Global Wage Report 2018/19, which provides comprehensive estimates of within country wage inequality for high-, middle- and low- income countries. The left-hand side plot shows a scatter plot of the Gini coefficient as a function of the unadjusted mean migrant pay gap (based on hourly wages) in the sample of 33 high-income economies (see table 1) with a linear line plot. The right-hand side plot shows a scatter plot of the Gini coefficient as a function of the unexplained mean migrant pay gap in the sample of 33 high-income economies with a linear line plot. Source: ILO estimates based on survey data provided by national sources (see Appendix II). Appendix IV. Supplementary results 131

X Figure A-2. Industrial sectors of migrant and non-migrant wage workers by sex, latest years

Migrant men and non-migrant men

High-incomeHigh-income countries countries(HICs)

Primary sector Secondary sector Tertiary sector Netherlands Denmark Poland Belgium Sweden Slovenia Austria Spain Iceland Finland Australia Lithuania Croatia Malta Estonia Slovakia Ireland Italy Poland Canada Hungary Czech Republic Chile Slovakia Lithuania Finland Greece Argentina Switzerland Croatia Canada Cyprus Argentina United Kingdom Czech Republic Austria Sweden United Kingdom France Estonia Belgium Latvia Australia Norway Portugal Luxembourg Netherlands United States Portugal Cyprus Norway Luxembourg Luxembourg Switzerland United States Spain Ireland Greece Norway Latvia France Netherlands Slovenia Latvia United Kingdom United States Austria Switzerland Chile Hungary Belgium Italy Argentina Chile France Croatia Czech Republic Iceland Italy Finland Portugal Slovakia Ireland Cyprus Iceland Canada Denmark Estonia Australia Hungary Slovenia Denmark Spain Lithuania Malta Greece Poland Sweden High-income High-income High-income 0481216 020406080 100 020406080 100 Share (%) Share (%) Share (%)

Low- and middle-incomeLow- and middle-income countries (LMICs)countries

Primary sector Secondary sector Tertiary sector Romania Sierra Leone Nepal Bulgaria Bulgaria Bangladesh Albania Romania Turkey Nepal Gambia Namibia Sierra Leone Tanzania Albania Mexico Costa Rica Madagascar Madagascar Bolivia* Turkey Bolivia Jordan Tanzania Namibia Madagascar Mexico Costa Rica Gambia Gambia Bangladesh Mexico Bolivia Jordan Tanzania** Costa Rica Albania Sierra Leone Namibia Turkey Bulgaria Bangladesh Nepal Romania Low- and Low- and Low- and middle-income middle-income middle-income 0510 15 20 25 30 01020304050607080 020406080 100 120 Share (%) Share (%) Share (%)

Migrant men Non-migrant men

(Figure A-2 continued on page 132) 132 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure A-2 continued from page 131)

Migrant women and non-migrant women

High-income countries (HICs) High-income countries

Primary sector Secondary sector Tertiary sector Netherlands Slovakia Lithuania Denmark Poland Iceland Slovenia Hungary Estonia Croatia Malta Czech Republic Slovakia Sweden Latvia Poland Spain Ireland Lithuania Belgium United Kingdom Czech Republic Denmark Greece Argentina France Finland Finland Luxembourg Austria Austria Australia United States Luxembourg Cyprus Slovenia Canada Chile Canada Iceland Italy Portugal Croatia Norway Belgium Argentina Switzerland Estonia Switzerland Chile Switzerland Netherlands Italy Italy Greece Netherlands Australia Norway Argentina France Portugal Cyprus Latvia Canada Spain Cyprus Slovenia Croatia United States United States Australia Norway United Kingdom Hungary Chile Austria France Portugal Finland Luxembourg United Kingdom Ireland Sweden Sweden Latvia Denmark Ireland Czech Republic Belgium Spain Estonia Malta Greece Iceland Poland Hungary Lithuania Slovakia High-income High-income High-income 012345678 0102030405060 020406080 100 120 Share (%) Share (%) Share (%)

Low- and middle-incomeLow- and middle-income countries (LMICs)countries

Primary sector Secondary sector Tertiary sector Albania Sierra Leone Nepal Bolivia Madagascar Bangladesh Romania Romania Turkey Bulgaria Jordan Albania Sierra Leone Gambia Gambia Namibia Madagascar Mexico Costa Rica Tanzania** Tanzania Nepal Bulgaria Turkey Mexico Bolivia* Costa Rica Tanzania Mexico Namibia Bolivia Namibia Bangladesh Bulgaria Costa Rica Gambia Bangladesh Jordan Madagascar Albania Romania Nepal Turkey Sierra Leone Low- and Low- and Low- and middle-income middle-income middle-income 01020304050607080 0510 15 20 25 30 35 40 45 020406080 100 120 Share (%) Share (%) Share (%)

Migrant women Non-migrant women

Note: High-income and low- and middle-income estimates are the weighted averages of the sample of high-income countries and low- and middle-income countries, respectively. Averages are weighted by the number of wage employees in each country. Primary sector comprises Agriculture, Fishing and Forestry. Secondary sector consists of Mining and Quarry; Manufacturing; Electricity, Gas and Water; and Construction. Tertiary sector includes all services and industries not captured under the Primary and Secondary sectors. *the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). Appendix IV. Supplementary results 133

X Figure A-3: The wage structure of migrant workers and nationals by sex, selected countries, latest years

High-income countries (HICs)

Austria Belgium Cyprus

1 1.2 1 .8 1 .8 .8 .6 .6 .6 .4 .4 .4 .2 .2 .2 Probability density Probability density Probability density Probability 0 0 0 –2 –1 0 1 2 3 4 5 6 0 1 2 3 4 5 0 1 2 3 3 5 Range of hourly wages Range of hourly wages Range of hourly wages

Czech Republic Denmark Estonia

1 1.5 .8 .8 .6 1 .6 .4 .4 .5 .2 .2 Probability density Probability density Probability density Probability 0 0 0 0 1 2 3 4 0 1 2 3 4 5 6 7 –1 0 1 2 3 4 5 Range of hourly wages Range of hourly wages Range of hourly wages

Finland France Greece

1.2 1 1 1 .8 .8 .8 .6 .6 .6 .4 .4 .4 .2 .2 .2 Probability density Probability density Probability density Probability 0 0 0 –1 0 1 2 3 4 5 6 7 –1 0 1 2 3 4 5 6 7 0 1 2 3 4 Range of hourly wages Range of hourly wages Range of hourly wages

Hungary Ireland Italy

1 1 1 .8 .8 .8 .6 .6 .6 .4 .4 .4 .2 .2 .2 Probability density Probability density Probability density Probability 0 0 0 –1 0 1 2 3 4 5 0 1 2 3 4 5 6 7 0 1 2 3 4 Range of hourly wages Range of hourly wages Range of hourly wages

Latvia Luxembourg Malta

1 1 1 .8 .8 .8 .6 .6 .6 .4 .4 .4 .2 .2 .2 Probability density Probability density Probability density Probability 0 0 0 –2 –1 0 1 2 3 4 0 1 2 3 4 5 6 0 1 2 3 4 5 Range of hourly wages Range of hourly wages Range of hourly wages

Migrant men Men nationals Migrant women Women nationals

(Figure A-3 continued on page 134) 134 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure A-3 continued from page 133)

High-income countries (HICs) (continued)

Netherlands Spain Sweden

1 .8 1 .8 .6 .8 .6 .6 .4 .4 .4 .2 .2 .2 Probability density Probability 0 density Probability density Probability 0 0 0 1 2 3 4 5 6 7 –1 0 1 2 3 4 5 6 –1 0 1 2 3 3 5 6 Range of hourly wages Range of hourly wages Range of hourly wages

UK Iceland Norway

.8 1 1.2 .8 1 .6 .8 .6 .4 .6 .4 .4 .2 .2 .2 Probability density Probability density Probability density Probability 0 0 0 0 1 2 3 4 5 6 7 –1 0 1 2 3 4 5 –1 0 1 2 3 4 5 6 7 8 Range of hourly wages Range of hourly wages Range of hourly wages

Switzerland Australia United States

1 1.2 1.2 .8 1 1 .8 .8 .6 .6 .6 .4 .4 .4 .2 .2 .2 Probability density Probability density Probability density Probability 0 0 0 –3 –2 –1 0 1 2 3 4 5 –1 0 1 2 3 4 5 –9 –8 –7 –6 –5 –4 –3 –2 –1 0 1 2 3 4 Range of hourly wages Range of hourly wages Range of hourly wages

Canada Argentina Chile

.8 1.2 1 1 .6 .8 .8 .6 .6 .4 .4 .4 .2 .2 .2 Probability density Probability density Probability density Probability 0 0 0 0 1 2 3 4 –1 0 1 2 3 4 5 6 7 8 9 0 2 4 6 8 10 12 14 Range of hourly wages Range of hourly wages Range of hourly wages

Migrant men Men nationals Migrant women Women nationals Appendix IV. Supplementary results 135

Low- and middle-income countries (LMICs)

Turkey Albania Jordan

1.2 1 1.5 1 .8 .8 .6 1 .6 .4 .4 .5 .2 .2 Probability density Probability density Probability density Probability 0 0 0 –2 –1 0 1 2 3 4 –4 –3 –2 –1 0 1 2 3 4 5 6 7 8 9 –2 –1 0 1 2 Range of hourly wages Range of hourly wages Range of hourly wages

Bangladesh Nepal Gambia

.8 1 .6 .8 .6 .5 .4 .6 .4 .3 .4 .2 .2 .2 .1 Probability density Probability density Probability density Probability 0 0 0 –6 –5 –4 –3 –2 –1 0 1 2 3 4 5 6 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 Range of hourly wages Range of hourly wages Range of hourly wages

Madagascar Malawi Namibia

5 5 4 4 4 3 3 2 2 2 2 1 1 1 Probability density Probability density Probability density Probability 0 0 0 0 1 2 3 4 5 6 7 8 9 –2 –1 0 1 2 3 4 5 6 7 –6 –5 –4 –3 –2 –1 0 1 2 3 4 5 6 Range of hourly wages Range of hourly wages Range of hourly wages

Siera Leone Tanzania Bolivia

8 .5 1

6 .4 .8 .3 .6 4 .2 .4 2 .1 .2 Probability density Probability density Probability density Probability 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 0 1 2 3 4 5 6 7 8 9 –1 0 1 2 3 4 5 6 Range of hourly wages Range of hourly wages Range of hourly wages

Costa Rica Mexico

1.2 8 1 6 .8 Migrant men Men nationals .6 4 Migrant women Women nationals .4 2 .2 Probability density Probability density Probability 0 0 0 1 2 3 4 5 6 7 8 9 10 11 –3 –2 –1 0 1 2 3 4 5 6 Range of hourly wages Range of hourly wages

Note: Mean hourly wage: Solid vertical line. *the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). 136 The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure A-4: The mean migrant pay gaps across the wage distribution, selected countries, latest years

High-income countries (HICs)

Austria Belgium 60 20 50 15 40 30 10 20

ant pay gap (% ) ant pay gap (% ) 5 10

Migr 0 Migr 0 Mean D10D9D8D7D6D5D4D3D2D1 Mean D10D9D8D7D6D5D4D3D2D1 gap gap

Croatia Cyprus 10 100 80 0 60 –10 40 ant pay gap (% ) ant pay gap (% ) –20 20

Migr –30 Migr 0 Mean D10D9D8D7D6D5D4D3D2D1 Mean D9D8D7D6D5D4D3D2D1 D10 gap gap

Czech Republic Denmark 0 30 25 –5 20 15 –10 10 ant pay gap (% ) ant pay gap (% ) 5

Migr –15 Migr 0 Mean D10D9D8D7D6D5D4D3D2D1 Mean D9D8D7D6D5D4D3D2D1 D10 gap gap

Estonia Finland 30 20 25 15 20 15 10 10 ant pay gap (% ) ant pay gap (% ) 5 5

Migr 0 Migr 0 Mean D10D9D8D7D6D5D4D3D2D1 Mean D10D9D8D7D6D5D4D3D2D1 gap gap

France Greece 80 40

60 30 40 20 20

ant pay gap (% ) ant pay gap (% ) 10 0

Migr –20 Migr 0 Mean D10D9D8D7D6D5D4D3D2D1 Mean D10D9D8D7D6D5D4D3D2D1 gap gap Appendix IV. Supplementary results 137

High-income countries (HICs) (continued)

Hungary Ireland 30 40

20 30 10 20 0 10 –10 ant pay gap (% ) ant pay gap (% ) –20 0

Migr –30 Migr –10 Mean D10D9D8D7D6D5D4D3D2D1 Mean D10D9D8D7D6D5D4D3D2D1 gap gap

Italy Latvia 60 25 50 20 40 15 30 10 20 5 ant pay gap (% ) ant pay gap (% ) 10 0

Migr 0 Migr –5 Mean D10D9D8D7D6D5D4D3D2D1 Mean D10D9D8D7D6D5D4D3D2D1 gap gap

Lithuania Luxembourg 30 50

20 40 10 30 0 20 –10 ant pay gap (% ) ant pay gap (% ) –20 10

Migr –30 Migr 0 Mean D10D9D8D7D6D5D4D3D2D1 Mean D10D9D8D7D6D5D4D3D2D1 gap gap

Malta Netherlands 2 50

0 40 –2 30 –4 20 –6 ant pay gap (% ) ant pay gap (% ) –8 10

Migr –10 Migr 0 Mean D10D9D8D7D6D5D4D3D2D1 Mean D10D9D8D7D6D5D4D3D2D1 gap gap

Poland Portugal 20 50

10 40 0 30 –10 20 –20 ant pay gap (% ) ant pay gap (% ) –30 10

Migr –40 Migr 0 Mean D10D9D8D7D6D5D4D3D2D1 Mean D10D9D8D7D6D5D4D3D2D1 gap gap

(Figure A-4 continued on page 138) 138 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure A-4 continued from page 137)

High-income countries (HICs) (continued)

Slovakia Slovenia 20 60 10 50 0 40 –10 30 –20 20

ant pay gap (% ) –30 ant pay gap (% ) –40 10

Migr –50 Migr 0 Mean D10D9D8D7D6D5D4D3D2D1 Mean D10D9D8D7D6D5D4D3D2D1 gap gap

Spain Sweden 60 40 50 30 40 30 20 20 ant pay gap (% ) ant pay gap (% ) 10 10

Migr 0 Migr 0 Mean D10D9D8D7D6D5D4D3D2D1 Mean D10D9D8D7D6D5D4D3D2D1 gap gap

United Kingdom Iceland 20 40

15 30

10 20

5 10 ant pay gap (% ) ant pay gap (% ) 0 0

Migr –5 Migr –10 Mean D10D9D8D7D6D5D4D3D2D1 Mean D10D9D8D7D6D5D4D3D2D1 gap gap

Norway Switzerland 40 10

0 30 –10 20 –20

ant pay gap (% ) 10 ant pay gap (% ) –30

Migr 0 Migr –40 Mean D10D9D8D7D6D5D4D3D2D1 Mean D10D9D8D7D6D5D4D3D2D1 gap gap

Australia United States 0 25

20 –5 15 –10 10

ant pay gap (% ) –15 ant pay gap (% ) 5

Migr –20 Migr 0 Mean D10D9D8D7D6D5D4D3D2D1 Mean D10D9D8D7D6D5D4D3D2D1 gap gap Appendix IV. Supplementary results 139

High-income countries (HICs) (continued)

Canada Argentina 8 30

6 25 20 4 15 2 10 ant pay gap (% ) ant pay gap (% ) 0 5

Migr –2 Migr 0 Mean D10D9D8D7D6D5D4D3D2D1 Mean D9D8D7D6D5D4D3D2D1 D10 gap gap

Chile 20

–10 ant pay gap (% ) –20

Migr –30 Mean D10D9D8D7D6D5D4D3D2D1 gap

Low- and middle-income countries (LMICs)

Bulgaria Turkey 20 15 0 10 –20 –40 5 –60 ant pay gap (% ) ant pay gap (% ) 0 –80

Migr –100 Migr –5 Mean D10D9D8D7D6D5D4D3D2D1 Mean D9D8D7D6D5D4D3D2D1 D10 gap gap

Albania Jordan 20 50

–20 40

–60 30

–100 20 ant pay gap (% ) ant pay gap (% ) –140 10

Migr –180 Migr 0 Mean D10D9D8D7D6D5D4D3D2D1 Mean D9D8D7D6D5D4D3D2D1 D10 gap gap

Bangladesh Nepal 0 40

–10 20

–20 0

ant pay gap (% ) –30 ant pay gap (% ) –20

Migr –40 Migr –40 Mean D10D9D8D7D6D5D4D3D2D1 Mean D9D8D7D6D5D4D3D2D1 D10 gap gap

(Figure A-4 continued on page 140) 140 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure A-4 continued from page 139)

Low- and middle-income countries (LMICs) (continued)

Gambia Madagascar 10 0 0 –10 –40 –20 –80 –30

ant pay gap (% ) –40 ant pay gap (% ) –120 –50

Migr –60 Migr –160 Mean D10D9D8D7D6D5D4D3D2D1 Mean D10D9D8D7D6D5D4D3D2D1 gap gap

Malawi Namibia 0 60 –10 40 20 –20 0 –30 –20 ant pay gap (% ) ant pay gap (% ) –40 –40

Migr –50 Migr –60 Mean D10D9D8D7D6D5D4D3D2D1 Mean D9D8D7D6D5D4D3D2D1 D10 gap gap

Sierra Leone Tanzania** 0 10

–50 0

–100 –10

–150 –20 ant pay gap (% ) ant pay gap (% ) –200 –30

Migr –250 Migr –40 Mean D10D9D8D7D6D5D4D3D2D1 Mean D9D8D7D6D5D4D3D2D1 D10 gap gap

Bolivia* Costa Rica 0 60 –10 50 –20 40 –30 30 –40 20 ant pay gap (% ) ant pay gap (% ) –50 10

Migr –60 Migr 0 Mean D10D9D8D7D6D5D4D3D2D1 Mean D10D9D8D7D6D5D4D3D2D1 gap gap

Mexico 50

ant pay gap (% ) –25

Migr –50 Mean D10D9D8D7D6D5D4D3D2D1 gap

Note: D1-D10 are deciles that split the hourly wage distribution into ten equally sized groups (from the bottom 10 per cent wage earners up to the top 10 per cent wage earners). “Mean gap” repeats the estimated raw mean migrant pay gap in the economy based on hourly wages from figure 17. *the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). Appendix IV. Supplementary results 141

X Figure A-5: The share of migrant workers and nationals by top and bottom centiles and intervening deciles of the hourly wage distribution, 48 countries, latest years

High-income countries (HICs)

Migrant workers and non-migrant workers Migrant men and migrant women

Austria Austria 100 40 80 60 20 40 Percent Percent 20

0 0 1st cntl 1st cntl 3rd decile 3rd decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 100th cntl 100th cntl 2nd decile 2nd decile 2–10th cntl 2–10th cntl 91–99th cntl 91–99th cntl Cyprus Cyprus 100 100 80 80 60 60 40 40 Percent Percent 20 20 0 0 1st cntl 1st cntl 3rd decile 3rd decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 100th cntl 100th cntl 2nd decile 2nd decile 2–10th cntl 2–10th cntl 91–99th cntl 91–99th cntl Czech Republic Czech Republic 100 5 80 4 60 3 40 2 Percent Percent 20 1 0 0 1st cntl 1st cntl 3rd decile 3rd decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 100th cntl 100th cntl 2nd decile 2nd decile 2–10th cntl 2–10th cntl 91–99th cntl 91–99th cntl

Migrant workers Non-migrant workers Migrant men Migrant women

(Figure A-5 continued on page 142) 142 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure A-5 continued from page 141)

High-income countries (HICs) (continued)

Migrant workers and non-migrant workers Migrant men and migrant women

Denmark Denmark 100 20 80 15 60 10 40 Percent Percent 20 5

Migrant workers Non-migrant workers Migrant men Migrant women Appendix IV. Supplementary results 143

High-income countries (HICs) (continued)

Migrant workers and non-migrant workers Migrant men and migrant women

Hungary Hungary 100 2 80 60 1 40 Percent Percent 20

Migrant workers Non-migrant workers Migrant men Migrant women

(Figure A-5 continued on page 144) 144 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure A-5 continued from page 143)

High-income countries (HICs) (continued)

Migrant workers and non-migrant workers Migrant men and migrant women

Luxembourg Luxembourg 100 100 80 80 60 60 40 40 Percent Percent 20 20 0 0 1st cntl 1st cntl 3rd decile 3rd decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 100th cntl 100th cntl 2nd decile 2nd decile 2–10th cntl 2–10th cntl 91–99th cntl 91–99th cntl Malta Malta 100 10 80 60 5 40 Percent Percent 20

Migrant workers Non-migrant workers Migrant men Migrant women Appendix IV. Supplementary results 145

High-income countries (HICs) (continued)

Migrant workers and non-migrant workers Migrant men and migrant women

Slovakia Slovakia 100 1.0 80 60 0.5 40 Percent Percent 20

Migrant workers Non-migrant workers Migrant men Migrant women

(Figure A-5 continued on page 146) 146 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure A-5 continued from page 145)

High-income countries (HICs) (continued)

Migrant workers and non-migrant workers Migrant men and migrant women

Iceland Iceland 100 15 80 60 10 40 5 Percent Percent 20

Migrant workers Non-migrant workers Migrant men Migrant women Appendix IV. Supplementary results 147

High-income countries (HICs) (continued)

Migrant workers and non-migrant workers Migrant men and migrant women

Canada Canada 100 40 80 60 20 40 Percent Percent 20

Migrant workers Non-migrant workers Migrant men Migrant women

Source: ILO estimates based on survey data provided by national sources (see Appendix II).

(Figure A-5 continued on page 148) 148 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure A-5 continued from page 147)

Low- and middle-income countries (LMICs)

Migrant workers and non-migrant workers Migrant men and migrant women

Bulgaria Bulgaria 100 2 80 60 1 40 Percent Percent 20

Migrant workers Non-migrant workers Migrant men Migrant women Appendix IV. Supplementary results 149

Low- and middle-income countries (LMICs) (continued)

Migrant workers and non-migrant workers Migrant men and migrant women

Nepal Nepal 100 5 80 4 60 3 40 2 Percent Percent 20 1 0 0 1st cntl 1st cntl 3rd decile 3rd decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 100th cntl 100th cntl 2nd decile 2nd decile 2–10th cntl 2–10th cntl 91–99th cntl 91–99th cntl Gambia Gambia 100 20 80 15 60 10 40 Percent Percent 20 5

0 0 1st cntl 1st cntl 3rd decile 3rd decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 100th cntl 100th cntl 2nd decile 2nd decile 2–10th cntl 2–10th cntl 91–99th cntl 91–99th cntl Madagascar Madagascar 100 5 80 4 60 3 40 2 Percent Percent 20 1 0 0 1st cntl 1st cntl 3rd decile 3rd decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 100th cntl 100th cntl 2nd decile 2nd decile 2–10th cntl 2–10th cntl 91–99th cntl 91–99th cntl Malawi Malawi 100 5 80 4 60 3 40 2 Percent Percent 20 1 0 0 1st cntl 1st cntl 3rd decile 3rd decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 100th cntl 100th cntl 2nd decile 2nd decile 2–10th cntl 2–10th cntl 91–99th cntl 91–99th cntl Namibia Namibia 100 25 80 20 60 15 40 10 Percent Percent 20 5 0 0 1st cntl 1st cntl 3rd decile 3rd decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 100th cntl 100th cntl 2nd decile 2nd decile 2–10th cntl 2–10th cntl 91–99th cntl 91–99th cntl

Migrant workers Non-migrant workers Migrant men Migrant women

(Figure A-5 continued on page 150) 150 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure A-5 continued from page 149)

Low- and middle-income countries (LMICs) (continued)

Migrant workers and non-migrant workers Migrant men and migrant women

Sierra Leone Sierra Leone 100 5 80 4 60 3 40 2 Percent Percent 20 1 0 0 1st cntl 1st cntl 3rd decile 3rd decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 100th cntl 100th cntl 2nd decile 2nd decile 2–10th cntl 2–10th cntl 91–99th cntl 91–99th cntl Tanzania** Tanzania** 100 10 80 60 5 40 Percent Percent 20

0 0.0 1st cntl 1st cntl 3rd decile 3rd decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 100th cntl 100th cntl 2nd decile 2nd decile 2–10th cntl 2–10th cntl 91–99th cntl 91–99th cntl Costa Rica Costa Rica 100 25 80 20 60 15 40 10 Percent Percent 20 5 0 0 1st cntl 1st cntl 3rd decile 3rd decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 100th cntl 100th cntl 2nd decile 2nd decile 2–10th cntl 2–10th cntl 91–99th cntl 91–99th cntl Mexico Mexico 100 1.0 80 60 0.5 40 Percent Percent 20

Migrant workers Non-migrant workers Migrant men Migrant women

Note: *the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). Appendix IV. Supplementary results 151

X Figure A-6: Decomposition of the migrant pay gap across the hourly wage distribution into explained and unexplained parts, selected countries, latest years

High-income countries (HICs)

Austria Belgium 60 20

50 15 40 10 30 5 20 ant pay gap (% ) ant pay gap (% ) 10 0

Migr 0 Migr –5 D10D9D8D7D6D5D4D3D2D1 D10D9D8D7D6D5D4D3D2D1

Croatia Cyprus 60 100 40 80 20 60 0 40 –20 20 ant pay gap (% ) ant pay gap (% ) –40 0

Migr –60 Migr –20 D10D9D8D7D6D5D4D3D2D1 D10D9D8D7D6D5D4D3D2D1

Czech Republic Denmark 40 40

20 20 0

–20 0 ant pay gap (% ) ant pay gap (% ) –40

Migr –60 Migr –20 D10D9D8D7D6D5D4D3D2D1 D10D9D8D7D6D5D4D3D2D1

Estonia Finland 40 40

20 20 0 0 ant pay gap (% ) ant pay gap (% ) –20

Migr –20 Migr –40 D10D9D8D7D6D5D4D3D2D1 D10D9D8D7D6D5D4D3D2D1

France Greece 80 60 60 40 40 20 20 0 ant pay gap (% ) ant pay gap (% ) 0 –20

Migr –40 Migr –20 D10D9D8D7D6D5D4D3D2D1 D10D9D8D7D6D5D4D3D2D1

Explaine d Unxplained

(Figure A-6 continued on page 152) 152 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure A-6 continued from page 151)

High-income countries (HICs) (continued)

Ireland Italy 40 60

40 20 20

0 0 ant pay gap (% ) ant pay gap (% ) –20

Migr –20 Migr –40 D10D9D8D7D6D5D4D3D2D1 D10D9D8D7D6D5D4D3D2D1

Latvia Lithuania 40 80 60 20 40 20 0 0

ant pay gap (% ) –20 ant pay gap (% ) –20 –40

Migr –40 Migr –60 D10D9D8D7D6D5D4D3D2D1 D10D9D8D7D6D5D4D3D2D1

Luxembourg Malta 50 20

40 0 30

20 –20 ant pay gap (% ) ant pay gap (% ) 10

Migr 0 Migr –40 D10D9D8D7D6D5D4D3D2D1 D10D9D8D7D6D5D4D3D2D1

Netherlands Poland 120 80

80 40

40 0

0 -40 ant pay gap (% ) ant pay gap (% ) –40 -80

Migr –80 Migr -120 D10D9D8D7D6D5D4D3D2D1 D4D3D2D1 D10D9D8D7D6D5

Portugal Slovenia 60 120

80 40 40 20 0

ant pay gap (% ) 0 ant pay gap (% ) –40

Migr –20 Migr –80 D10D9D8D7D6D5D4D3D2D1 D10D9D8D7D6D5D4D3D2D1

Explaine d Unxplained Appendix IV. Supplementary results 153

High-income countries (HICs) (continued)

Sweden United Kingdom 150 30

100 20 50 10 0

ant pay gap (% ) ant pay gap (% ) 0 –50

Migr –100 Migr –10 D10D9D8D7D6D5D4D3D2D1 D10D9D8D7D6D5D4D3D2D1

Iceland Norway 40 60

40 20 20

0 0 ant pay gap (% ) ant pay gap (% ) –20

Migr –20 Migr –40 D10D9D8D7D6D5D4D3D2D1 D10D9D8D7D6D5D4D3D2D1

Switzerland Australia 10 10

0 0 –10

–20 –10 ant pay gap (% ) ant pay gap (% ) –30

Migr –40 Migr –20 D10D9D8D7D6D5D4D3D2D1 D10D9D8D7D6D5D4D3D2D1

United States Canada 25 12

20 8

15 4

10 0 ant pay gap (% ) ant pay gap (% ) 5 –4

Migr 0 Migr –8 D10D9D8D7D6D5D4D3D2D1 D10D9D8D7D6D5D4D3D2D1

Argentina Chile 30 80

20 40

10 0

ant pay gap (% ) 0 ant pay gap (% ) –40

Migr –10 Migr –80 D10D9D8D7D6D5D4D3D2D1 D10D9D8D7D6D5D4D3D2D1

Explaine d Unxplained

Note: D1-D10 are deciles that split the hourly wage distribution into ten equally sized groups (from the bottom 10 per cent wage earners up to the top 10 per cent wage earners). Source: ILO estimates based on survey data provided by national sources (see Appendix II).

(Figure A-6 continued on page 154) 154 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure A-6 continued from page 153)

Low- and middle-income countries (LMICs)

Bulgaria Turkey 40 80

0 40

–40 0

ant pay gap (% ) –80 ant pay gap (% ) –40

Migr –120 Migr –80 D10D9D8D7D6D5D4D3D2D1 D10D9D8D7D6D5D4D3D2D1

Albania Jordan 40 80 0 –40 40 –80 –120 0 ant pay gap (% ) ant pay gap (% ) –160

Migr –200 Migr –40 D10D9D8D7D6D5D4D3D2D1 D10D9D8D7D6D5D4D3D2D1

Bangladesh Nepal 20 80

0 40

–20 0

ant pay gap (% ) –40 ant pay gap (% ) –40

Migr –60 Migr –80 D10D9D8D7D6D5D4D3D2D1 D10D9D8D7D6D5D4D3D2D1

Gambia Madagascar 60 60 40 30 20 0 –30 0 –60 –20

ant pay gap (% ) ant pay gap (% ) –90 –40 –120

Migr –60 Migr –150 D10D9D8D7D6D5D4D3D2D1 D10D9D8D7D6D5D4D3D2D1

Malawi Namibia 30 200 150 0 100 50 0 –30

ant pay gap (% ) ant pay gap (% ) –50 –100

Migr –60 Migr –150 D10D9D8D7D6D5D4D3D2D1 D10D9D8D7D6D5D4D3D2D1

Explaine d Unxplained Appendix IV. Supplementary results 155

Low- and middle-income countries (LMICs) (continued)

Sierra Leone Tanzania** 100 80

40 0 0 –100 ant pay gap (% ) ant pay gap (% ) –40

Migr –200 Migr –80 D5D4D3D2D1 D10D9D8D7D6 D10D9D8D7D6D5D4D3D2D1

Bolivia* Costa Rica 50 150

100 0 50

0 –50 ant pay gap (% ) ant pay gap (% ) –50

Migr –100 Migr –100 D10D9D8D7D6D5D4D3D2D1 D10D9D8D7D6D5D4D3D2D1

Mexico 50

25 Explaine d Unxplained 0

ant pay gap (% ) –25

Migr –50 D10D9D8D7D6D5D4D3D2D1

Note: D1-D10 are deciles that split the hourly wage distribution into ten equally sized groups (from the bottom 10 per cent wage earners up to the top 10 per cent wage earners). * the Plurinational State of Bolivia; ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II). 156 The migrant pay gap: Understanding wage differences between migrants and nationals

X Figure A-7: The proportion of migrant workers within occupations, education and the mean migrant pay gap, selected countries, latest years

High-income countries (HICs)

Proportion of migrant workers Migrant pay gap Educational score

Austria 40 30 30

30 20 20 20 e in education ant pay gap (% ) 10 10 10 Scor Migr e of migrant workers (% ) e of migrant 0 0 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Belgium 30 15 30

20 10 20 e in education

10 ant pay gap (% ) 5 10 Scor Migr e of migrant workers (% ) e of migrant

0 0 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Cyprus 60 60 30

40 40 20 e in education 20 ant pay gap (% ) 20 10 Scor Migr e of migrant workers (% ) e of migrant 0 0 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Denmark 16 40 30

12 30 20 8 20 e in education ant pay gap (% ) 10 4 10 Scor Migr e of migrant workers (% ) e of migrant 0 0 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Estonia 30 30 30

20 20 20 10 e in education 10 ant pay gap (% ) 10 0 Scor Migr e of migrant workers (% ) e of migrant 0 –10 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Nationals Migrant workers Appendix IV. Supplementary results 157

High-income countries (HICs) (continued)

Proportion of migrant workers Migrant pay gap Educational score

Finland 8 80 30

60 6 20 40 4 20 e in education ant pay gap (% ) 10 2

0 Scor Migr e of migrant workers (% ) e of migrant 0 –20 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

France 12 30 30

9 15 20

6 0 e in education

ant pay gap (% ) 10 3 –15 Scor Migr e of migrant workers (% ) e of migrant

0 –30 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Greece 40 40 30

30 30 20 20 20 e in education ant pay gap (% ) 10 10 10 Scor Migr e of migrant workers (% ) e of migrant 0 0 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Ireland 30 30 30

15 20 20 0 e in education 10 ant pay gap (% ) 10 –15 Scor Migr e of migrant workers (% ) e of migrant 0 –30 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Italy 40 80 30

30 40 20 20 e in education ant pay gap (% ) 0 10 10 Scor Migr e of migrant workers (% ) e of migrant 0 –40 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Nationals Migrant workers

(Figure A-7 continued on page 158) 158 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure A-7 continued from page 157)

High-income countries (HICs) (continued)

Proportion of migrant workers Migrant pay gap Educational score

Latvia 30 15 30

10 20 20 5 e in education 10 ant pay gap (% ) 10 0 Scor Migr e of migrant workers (% ) e of migrant 0 –5 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Luxembourg 80 30 30

60 20 20

40 e in education

ant pay gap (% ) 10 10 20 Scor Migr e of migrant workers (% ) e of migrant

0 0 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Netherlands 10 40 30

8 20 20 6 0 4 e in education ant pay gap (% ) 10 –20

2 Scor Migr e of migrant workers (% ) e of migrant 0 –40 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Portugal 8 50 30

40 6 20 30 4 20 e in education ant pay gap (% ) 10 2

10 Scor Migr e of migrant workers (% ) e of migrant 0 0 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Spain 30 40 30

20 20 20 0 e in education 10 ant pay gap (% ) 10 –20 Scor Migr e of migrant workers (% ) e of migrant 0 –40 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Nationals Migrant workers Appendix IV. Supplementary results 159

High-income countries (HICs) (continued)

Proportion of migrant workers Migrant pay gap Educational score

Sweden 20 20 30

15 10 20 10 0 e in education ant pay gap (% ) 10 5 –10 Scor Migr e of migrant workers (% ) e of migrant 0 –20 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

United Kingdom 20 30 30

15 15 20

10 0 e in education

ant pay gap (% ) 10 5 –15 Scor Migr e of migrant workers (% ) e of migrant

0 –30 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Iceland 30 20 30

10 20 20 0 e in education 10 ant pay gap (% ) 10 –10 Scor Migr e of migrant workers (% ) e of migrant 0 –20 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Norway 20 30 30

20 15 20 10 10 0 e in education ant pay gap (% ) 10 5

–10 Scor Migr e of migrant workers (% ) e of migrant 0 –20 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Switzerland 50 10 30

40 0 20 30

20 e in education ant pay gap (% ) –10 10

10 Scor Migr e of migrant workers (% ) e of migrant 0 –20 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Nationals Migrant workers

(Figure A-7 continued on page 160) 160 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure A-7 continued from page 159)

High-income countries (HICs) (continued)

Proportion of migrant workers Migrant pay gap Educational score

Australia 40 10 30

30 0 20 20 e in education ant pay gap (% ) –10 10 10 Scor Migr e of migrant workers (% ) e of migrant 0 –20 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

United States 30 12 30

9 20 20

6 e in education

10 ant pay gap (% ) 10 3 Scor Migr e of migrant workers (% ) e of migrant

0 0 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Canada 30 20 30

20 10 20 e in education 10 ant pay gap (% ) 0 10 Scor Migr e of migrant workers (% ) e of migrant 0 –10 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Argentina 15 15 30

10 10 20 e in education 5 ant pay gap (% ) 5 10 Scor Migr e of migrant workers (% ) e of migrant 0 0 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Chile 10 20 30

8 10 20 6 0 4 e in education ant pay gap (% ) 10 –10

2 Scor Migr e of migrant workers (% ) e of migrant 0 –20 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Nationals Migrant workers

NOTE: MGR = Manager; PROF = Professionals; TECH = Technical; SEMI = Semi-skilled; LOW = Low-skilled; UNS = Unskilled. Source: ILO estimates based on survey data provided by national sources (see Appendix II). Appendix IV. Supplementary results 161

Low- and middle-income countries (LMICs)

Proportion of migrant workers Migrant pay gap Educational score

Turkey 5 30 30

4 15 20 3

2 e in education ant pay gap (% ) 0 10

1 Scor Migr e of migrant workers (% ) e of migrant 0 –15 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Jordan 80 40 30

60 0 20

40 e in education

ant pay gap (% ) –40 10 20 Scor Migr e of migrant workers (% ) e of migrant

0 –80 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Gambia 10 40 30

8 0 20 6 –40 4 e in education ant pay gap (% ) 10 –80

2 Scor Migr e of migrant workers (% ) e of migrant 0 –120 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Namibia 10 40 30

8 0 20 6 –40 4 e in education ant pay gap (% ) 10 –80

2 Scor Migr e of migrant workers (% ) e of migrant 0 –120 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Tanzania** 20 40 30

15 0 20 10 –40 e in education ant pay gap (% ) 10 5 –80 Scor Migr e of migrant workers (% ) e of migrant 0 –120 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Nationals Migrant workers

(Figure A-7 continued on page 162) 162 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure A-7 continued from page 161)

Low- and middle-income countries (LMICs) (continued)

Proportion of migrant workers Migrant pay gap Educational score

Costa Rica 20 20 30

15 15 20 10 10 5 e in education ant pay gap (% ) 10 5

0 Scor Migr e of migrant workers (% ) e of migrant 0 –5 0 Shar MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS MGR PROF TECH SEMI LOW UNS

Nationals Migrant workers

NOTE: MGR = Manager; PROF = Professionals; TECH = Technical; SEMI = Semi-skilled; LOW = Low-skilled; UNS = Unskilled. ** the United Republic of Tanzania. Source: ILO estimates based on survey data provided by national sources (see Appendix II).

X Figure A-8: The wage structure of migrant and non-migrant workers, with the counterfactual wage distribution of migrant workers, selected countries, latest year

High-income countries (HICs)

Austria Belgium Cyprus

1 1.2 .8

.8 1 .6 .8 .6 .6 .4 .4 .4 .2 .2 .2 Probability density Probability density Probability density Probability 0 0 0 –2 –1 0 1 2 3 4 5 6 0 1 2 3 4 5 0 1 2 3 3 5 Range of hourly wages Range of hourly wages Range of hourly wages

Estonia Finland France

.6 1 1 .5 .8 .8 .4 .6 .6 .3 .4 .4 .2 .1 .2 .2 Probability density Probability density Probability density Probability 0 0 0 –1 0 1 2 3 4 5 –1 0 1 2 3 4 5 6 7 –1 0 1 2 3 4 5 6 7 Range of hourly wages Range of hourly wages Range of hourly wages

Non-migrant workers Migrant workers Counterfactual of migrant workers Appendix IV. Supplementary results 163

High-income countries (HICs) (continued)

Greece Luxembourg Netherlands

1 .8 1.2 .8 1 .6 .6 .8 .6 .4 .4 .4 .2 .2 .2 Probability density Probability 0 density Probability density Probability 0 0 0 1 2 3 4 0 1 2 3 4 5 6 0 1 2 3 4 5 6 7 Range of hourly wages Range of hourly wages Range of hourly wages

United Kingdom Iceland Norway

1 1.2 1 .8 1 .8 .8 .6 .6 .6 .4 .4 .4 .2 .2 .2 Probability density Probability density Probability density Probability 0 0 0 0 1 2 3 4 5 6 7 –1 0 1 2 3 4 5 –1 0 1 2 3 4 5 6 7 8 Range of hourly wages Range of hourly wages Range of hourly wages

Switzerland Australia United States

.8 1 1 .8 .6 .8 .6 .6 .4 .4 .4 .2 .2 .2 Probability density Probability density Probability density Probability 0 0 0 –3 –2 –1 0 1 2 3 4 5 –1 0 1 2 3 4 5 –9 –8 –7 –6 –5 –4 –3 –2 –1 0 1 2 3 4 Range of hourly wages Range of hourly wages Range of hourly wages

Canada Argentina Chile

1 .8 1 .8 .8 .6 .6 .6 .4 .4 .4 .2 .2 .2 Probability density Probability density Probability density Probability 0 0 0 0 1 2 3 4 –1 0 1 2 3 4 5 6 7 8 9 0 2 4 6 8 10 12 14 Range of hourly wages Range of hourly wages Range of hourly wages

Non-migrant workers Migrant workers Counterfactual of migrant workers

(Figure A-8 continued on page 164) 164 The migrant pay gap: Understanding wage differences between migrants and nationals

(Figure A-8 continued from page 163)

Low- and middle-income countries (LMICs)

Jordan Bangladesh Nepal

1.5 1 .6 .8 1 .6 .4 .5 .4 .2 .2 Probability density Probability 0 density Probability density Probability 0 0 –2 –1 0 1 2 –6 –4 –2 0 2 4 6 0 1 2 3 4 5 6 7 Range of hourly wages Range of hourly wages Range of hourly wages

Gambia Madagascar Malawi

1.2 .5 .5 1 .4 .4 .8 .3 .3 .6 .2 .2 .4 .2 .1 .1 Probability density Probability density Probability density Probability 0 0 0 0 1 2 3 4 5 6 0 1 2 3 4 5 6 7 8 9 –2 –1 0 1 2 3 4 5 6 7 Range of hourly wages Range of hourly wages Range of hourly wages

Namibia Sierra Leone Tanzania

.5 .8 .5

.4 .6 .4 .3 .3 .4 .2 .2 .2 .1 .1 Probability density Probability density Probability density Probability 0 0 0 –6 –4 –2 0 2 4 6 0 1 2 3 4 5 6 7 8 9 10 11 12 13 0 1 2 3 4 5 6 7 8 9 Range of hourly wages Range of hourly wages Range of hourly wages

Costa Rica Mexico

1.2 .8 1 .6 .8 .6 .4 .4 .2 .2 Probability density Probability density Probability 0 0 0 1 2 3 4 5 6 7 8 9 10 11 –3 –2 –1 0 1 2 3 4 5 6 Range of hourly wages Range of hourly wages

Non-migrant workers Migrant workers Counterfactual of migrant workers

Source: ILO estimates based on survey data provided by national sources (see Appendix II). The migrant pay gap: Understanding wage differences between migrants and nationals

Based on recent data from 49 countries, this report analyzes differences in wages between migrant workers and nationals, providing a global overview of how migrant women and men fare in labour markets in low-, middle- and high-income countries. The report compares the labour market characteristics of migrants and nationals that contribute to their economic success and the migrant pay gap, with special attention to gender differences within and among these groups. Focusing on wage workers, it studies the raw and the factor-weighted migrant pay gaps, shedding light on the “explained” and “unexplained” parts of the raw migrant pay gap. In highlighting the persistent differences in wages between migrants and nationals, the report points to the urgency of implementing fair, evidence-based labour migration and labour market policies that contribute to more just societies, in line with the principles embodied in international labour standards.

ISBN: 978-92-2-032372-4 and nationals between migrants pay gap: Understanding wage differences The migrant International Labour Office 4, route des Morillons CH-1211 Geneva 22, Switzerland ILO