Global Wage report de l'OIT - 4/12/2014

Global Wage Report 2014/15
Wages and income inequality

International Labour Organization
The International Labour Organization (ILO) was founded in 1919 to promote social justice and
thereby contribute to universal and lasting peace. The ILO is responsible for drawing up and
overseeing international labour standards. It is the only tripartite United Nations agency that
brings together representatives of governments, employers and workers to jointly shape policies
and programmes promoting decent work for all. This unique arrangement gives the ILO an edge
in incorporating “real world” knowledge about employment and work.

Global Wage Report
2014/15
Wages and income inequality
INTERNATIONAL LABOUR OFFICE • GENEVA

Copyright © International Labour Organization 2015
First published 2015
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 Publications (Rights and Permissions), International
Labour Office, CH-1211 Geneva 22, Switzerland, or by email: pubdroit@ilo.org. The International Labour Office welcomes such
applications.
Libraries, institutions and other users registered with reproduction rights organizations 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.
Global Wage Report 2014/15: Wages and income inequality
International Labour Office – Geneva: ILO, 2015
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 pub-lication
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.
ILO publications and electronic products can be obtained through major booksellers or ILO local offices in many countries, or
direct from ILO Publications, International Labour Office, CH-1211 Geneva 22, Switzerland. Catalogues or lists of new publica-tions
are available on our website: www.ilo.org/publns
ILO Cataloguing in Publication Data
This publication was produced by the Document and Publications Production,
Printing and Distribution Branch (PRODOC) of the ILO.
Graphic and typographic design, manuscript preparation, copy editing, layout
and composition, proofreading, printing, electronic publishing and distribution.
PRODOC endeavours to use paper sourced from forests managed
in an environmentally sustainable and socially responsible manner.
Code: DTP-WEI-CORR-SRO-REPRO-DISTR
ISBN 978-92-2-128664-6 (print)
ISBN 978-92-2-128665-3 (web pdf)
ISBN 978-92-2-128666-0 (epub)
International Labour Office
wages / income distribution / wage differential / wage policy / developed countries / developing countries
13.07
Also available in PDF in Arabic: ISBN 978-92-2-628665-8; Chinese: ISBN 978-92-2-528665-9; French: ISBN 978-92-2-228665-2;
Portuguese: ISBN 978-92-2-828665-6; Russian: ISBN 978-92-2-428665-0; and Spanish: ISBN 978-92-2-328665-1.
Also available as EPUB in French: ISBN 978-92-2-228666-9; and Spanish: ISBN 978-92-2-328666-8.
Also available in Kindle and iBook edition.

Preface
The Global Wage Report 2014/15 presents both the latest trends in average wages
and an analysis of the role of wages in income inequality. The first part of the
report shows that global wage growth in recent years was driven by emerging and
developing economies, where real wages have been rising since 2007 although wage
growth slowed in 2013 compared to 2012. In developed economies, wages generally
remained stagnant in 2012 and 2013, and in a number of countries wages remained
below their 2007 level. These trends are a matter of concern.
At the level of the individual worker or firm, the immediate impacts of higher
or lower wages are self-evident. At the national level, the effects of higher or lower
wages on aggregate demand and employment are context-specific and cannot be
predicted or evaluated without taking into account the level of wages relative to
productivity, the degree of openness of the country under consideration and the
relative size of the different components of aggregate demand. At the international
level, if too many countries pursue wage moderation policies, the outcome is likely
to be negative. In the current environment, in which the global economy risks
sliding back into a low-growth trap, higher wage growth would be desirable in
those countries where wages in the past have lagged behind productivity growth.
As the report demonstrates, in some countries policies have already started to shift
in that direction.
The second part of the report turns to the role of wages in income in-equality.
Inequality has become the subject of growing interest in recent years
across the world, and there has been a realization that growing inequality not only
undermines social justice objectives, but can also have adverse economic conse-quences.
Through the adoption of the 2008 Declaration on Social Justice for a Fair
Globalization, ILO Members renewed their commitment to pursue policies with
regard to wages and earnings designed to ensure a just share of the fruits of pro-gress
to all and recognized that for a fair outcome for all, it has become even more
necessary to achieve social cohesion and to combat poverty and rising inequalities.
In many countries, the distribution of wages and paid employment has been
a key factor in recent inequality trends. This highlights the importance of labour
market institutions and policies – including minimum wages and collective bar-gaining
– that have an effect on income distribution.
Wage gaps between men and women, and between nationals and migrants,
remain significant and are only partly explained by differences in experience, edu-cation,
occupation and other labour market characteristics. Implementing effective
anti-discrimination policies, alongside other policies that address the underlying
causes of these wage gaps, is a concrete way to progress toward greater social
justice and fewer inequalities.
The report also shows that lower-income groups rely disproportionately on
social transfers or income from self-employment. In fact, in almost all countries,

vi Global Wage Report 2014/15
income sources are more diverse at the top and bottom of the household income
distribution than in the middle, where households rely to a much larger extent on
wages. This illustrates the importance of ensuring coherence between wage pol-icies
and social protection alongside the creation of paid employment to reduce
inequality.
It is my hope that, in addition to the report itself, the approach presented here
will be of use for ILO constituents to identify, in their specific context, the appro-priate
mix of wages, employment, enterprise promotion, social protection and
other policies that are most conducive to combining increased economic growth
with decreased income inequality. It is also my hope that the report will assure
policy-makers of the importance of including wages as part of international efforts
to coordinate macroeconomic policies and to advance social justice.
Guy Ryder
ILO Director-General

Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
Executive summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
Part I. Major trends in wages
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2 Trends in average wage growth in the global economy . . . . . . . . . . . . . . . . . . 3
3 Developed economies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.1 Average wages: Some variation across countries . . . . . . . . . . . . . . . . . 6
3.2 Productivity outstrips wage growth . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.3 Labour income shares . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
4 Trends in emerging and developing economies . . . . . . . . . . . . . . . . . . . . . . . . 13
4.1 Higher wage growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4.2 Labour income share can decline despite high wage growth . . . . . . . 15
5 Towards a progressive convergence in average wages? . . . . . . . . . . . . . . . . . . 18
Part II. Wages and income inequality
6 Introduction: The role of wages in household income inequality . . . . . . . . . . 19
6.1 The context: Long-term inequality on the rise in many countries
since the 1980s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
6.2 The range of policy responses to inequality . . . . . . . . . . . . . . . . . . . . . 20
7 Mixed recent trends in income inequality . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
7.1 Measuring “top–bottom” and “middle-class” inequality . . . . . . . . . . 21
7.2 Developed economies: Mixed inequality trends in a context
of generally flat or declining incomes . . . . . . . . . . . . . . . . . . . . . . . . . . 23
7.3 Emerging and developing economies: Varying inequality
in a context of generally increasing incomes . . . . . . . . . . . . . . . . . . . . 25
8 Inequality: The role of wages and paid employment . . . . . . . . . . . . . . . . . . . 28
8.1 Analysing how wages affect changes in inequality . . . . . . . . . . . . . . . 28
8.2 The labour market effect: Wage effect plus employment effect . . . . . . 28
8.3 Developed economies: Job losses and wage cuts
as the major factors in changes in inequality . . . . . . . . . . . . . . . . . . . . 29
8.4 Emerging and developing economies: A major role
for labour market effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

viii Global Wage Report 2014/15
9 Relating inequality to sources of income . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
9.1 Developed economies: Disaggregating income sources . . . . . . . . . . . 35
9.2 Emerging and developing economies: A varied and evolving picture 40
10 Wage gaps: Which workers earn less than others, and why? . . . . . . . . . . . . . 44
10.1 The gender wage gap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
10.2 The migrant wage gap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
10.3 The informal economy wage gap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Part III. Policy responses to address wages and inequality
11 Introduction: The policy challenge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
12 Labour market policies to address wages and inequality (primary distribution) 59
12.1 The role of policies to affect wages and wage distribution directly . . 59
12.2 The role of policies to affect wages and wage distribution indirectly 62
13 Fiscal policies to address wages and inequality (secondary distribution) . . . . 63
13.1 The role of taxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
13.2 The role of social protection and its relationship to labour market
policies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
14 Conclusion: Combined policy actions can address inequality,
promote employment and support aggregate demand . . . . . . . . . . . . . . . . . . 65
Appendices
Appendix I Global wage trends: Methodological issues . . . . . . . . . . . . . . . . . 67
Appendix II Definitions, concepts and data issues . . . . . . . . . . . . . . . . . . . . . . 75
Appendix III Changes in household income inequality . . . . . . . . . . . . . . . . . . . 89
Appendix IV Decomposing wage gaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103

Contents ix
Figures
1. Annual average economic growth, 1995–2013 . . . . . . . . . . . . . . . . . . . . . . . 3
2. Annual average global real wage growth, 2006–13 . . . . . . . . . . . . . . . . . . . 5
3. Annual average real wage growth in the G20, 2006–13 . . . . . . . . . . . . . . . 5
4. Annual average real wage growth in developed economies, 2006–13 . . . . 6
5. Average real wage index for developed G20 countries, 2007–13 . . . . . . . . . 7
6. Average real wage index for selected European countries
most affected by the crisis, 2007–13 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
7. Trends in growth in average wages and labour productivity
in developed economies (index), 1999–2013 . . . . . . . . . . . . . . . . . . . . . . . . . 8
8. Labour productivity, real wages and estimated real compensation
per employee in developed economies (indices), 1999–2013 . . . . . . . . . . . . 9
9. Estimated real labour compensation per employee and labour
productivity growth in the largest developed economies, deflated
by the CPI and GDP deflator, 1999–2013 . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
10. Adjusted labour income share in developed G20 countries, 1991–2013 . . . 11
11. Adjusted labour income share in selected European countries most
affected by the crisis, 1991–2013 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
12. Average real wage growth in emerging and developing economies,
by region, 2006–13 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
13. Real wage growth in the largest emerging and developing
economies, 2012–13 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
14. Employees as a share of total employment, 1999 and 2013 . . . . . . . . . . . . . 14
15. Adjusted labour income share in Mexico and Turkey, 1995–2013 . . . . . . . 16
16. Unadjusted labour income share in China, 1992–2011 . . . . . . . . . . . . . . . . 16
17. Labour income share in the Russian Federation, 1995–2012 . . . . . . . . . . . 17
18. Evolution of labour productivity and wages in South Africa, 1994–2011 . . 17
19. Global average monthly wage distribution in 2000 and 2012 . . . . . . . . . . . 18
20. Measures of inequality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
21. Inequality in a sample of developed economies in the crisis years,
2006–10: (a) top–bottom inequality (D9/D1);
(b) middle-class inequality (D7/D3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
22. Inequality in a sample of emerging and developing economies
over the past decade: (a) top–bottom inequality (D9/D1);
(b) middle-class inequality (D7/D3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
23. Recent evolution of real household income in selected countries . . . . . . . . 27
24. The labour market effect: Wage effect and employment effect . . . . . . . . . . 29
25. Change in inequality between the top and bottom 10 per cent
(D9/D1) in developed economies, 2006–10 . . . . . . . . . . . . . . . . . . . . . . . . . . 30
26. Change in inequality within the middle class (D7/D3) in developed
economies, 2006–10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
27. Change in inequality between the top and bottom 10 per cent (D9/D1)
in selected emerging and developing economies, selected years . . . . . . . . . 33

x Global Wage Report 2014/15
28. Change in inequality within the middle class (D7/D3) in selected
emerging and developing economies, selected years . . . . . . . . . . . . . . . . . . 34
29. Share of wages in household income, latest year: Selected developed
economies and European average . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
30. Household income by group and source in selected developed
economies, latest year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
31. Real growth in household income by source of income for the top
and bottom 10 per cent, Spain and Romania, 2006–10 . . . . . . . . . . . . . . . . 38
32. Share of wages in household income: Selected emerging
and developing economies, latest year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
33. Household income by group and source in selected emerging
and developing economies, latest year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
34. Real growth in household income by income source for the top
and bottom 10 per cent, selected emerging and developing
economies, 2001/02 to 2010–12 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
35. Motherhood wage gap in selected emerging and developing economies,
latest year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
36. Explained and unexplained gender wage gaps in selected countries,
latest year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
37. Eliminating the unexplained gender wage penalty: Mean wage gap
before and after adjustment in selected economies, latest year:
(a) developed economies; (b) emerging and developing economies . . . . . . 49
38. Wage distribution and counterfactual wage distribution,
Russian Federation and United States, latest year . . . . . . . . . . . . . . . . . . . 50
39. Explained and unexplained migrant–national wage gaps
in selected countries, latest year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
40. Eliminating the unexplained migrant wage penalty: Mean wage gap
before and after adjustment in selected economies, latest year . . . . . . . . . 53
41. Wage distribution and counterfactual wage distribution, Cyprus
and Spain, latest year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
42. Explained and unexplained informality wage gaps in selected
Latin American countries, latest year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Boxes
1. What are wages? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2. The ILO Global Wage Database . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3. The composition effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
4. The relationship between wages, compensation, different deflators
and labour productivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
5. What is understood by the term “middle class”? . . . . . . . . . . . . . . . . . . . . . 22
6. The largest drop in income inequality in the European Union
during the Great Recession: The puzzling case of Romania . . . . . . . . . . . 39
7. The motherhood wage gap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
8. Informality and labour market segmentation in Argentina . . . . . . . . . . . . 55

Contents xi
Appendix figures
A1. The components of household income . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
A2. Illustration of the components of household income . . . . . . . . . . . . . . . . . 78
Appendix tables
A1. Regional groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
A2. Coverage of the ILO Global Wage Database, 2013 . . . . . . . . . . . . . . . . . . 68
A3. Coverage of the ILO Global Wage Database, 2006–13 . . . . . . . . . . . . . . . 69
A4. Equivalent scale parameters guide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
A5. Data sources for emerging and developing economies . . . . . . . . . . . . . . . . 80
A6. Reference years for emerging and developing economies . . . . . . . . . . . . . 80
A7. Distribution of the complete sample, by country and year . . . . . . . . . . . . 82
A8. The representativeness of the sample . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Appendix boxes
A1. Concepts and definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
A2. Measuring per capita household income . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

Acknowledgements
Main contributors
The report was prepared by staff of the Inclusive Labour Markets, Labour Relations
and Working Conditions Branch (INWORK) of the ILO with contributions from
other ILO colleagues in Geneva and in ILO field offices, under the responsibility
of Philippe Marcadent, Chief of INWORK. Patrick Belser was the principal editor
of the report. Rosalia Vazquez-Alvarez led the analysis for Part II of the report
and coordinated inputs from consultants as well as the technical peer review.
Kristen Sobeck updated the ILO Global Wage Database, produced the analyses
for Part I and wrote Part I of the report. Research assistance was provided by
Nicolas Maître. Chris Edgar and Charlotte Beauchamp coordinated the editing,
publication and translation, and also handled the anonymous peer review of the
report. Ksenija Radojevic Bovet was responsible for the proofreading. Christian
Olsen designed the cover page. Inês Gomes designed several figures.
Sandra Polaski, ILO Deputy Director-General for Policy, contributed actively
to the policy content of the report.
Specific contributions
Part II the report is based on a research project coordinated by Rosalia Vazquez-
Alvarez. The project included data analysis and contributions from Roxana
Maurizio (Universidad Nacional de General Sarmiento and CONICET, Argentina),
Aico Peter Van Vuuren (University of Amsterdam), Ingrid Woolard (University of
Cape Town), Youngjian Hu (Tianjin University of Finance and Economics, China),
Li Shi and Wan Haiyuan (School of Economics and Management, Beijing Normal
University, and Institute of Social Studies, National Development and Reform
Commission, China) and Uma Rani (ILO Research Department). Marianne
Furrer provided research assistance to Uma Rani for Part II of the report.
The technical peer review for Part II, carried out in addition to the general
peer review of the report, was undertaken by Miguel Portela (Universidade do
Minho, Braga, Portugal) and Andrea Regoli and Antonella D’Agostino (both of
University of Naples “Parthenope”). Excerpts on the middle class and the moth-erhood
pay gap were kindly contributed by, respectively, Alexandre Dormeier
Freire (Graduate Institute of International and Development Studies, Geneva)
and Damian Grimshaw and Jill Rubery (both of the University of Manchester
Business School). Global and regional estimates in Part I of the report are based
on the same methodology (described in Appendix I) as in earlier Global Wage
Reports. This methodology was formulated by Malte Luebker (ILO Regional
Office for Asia and the Pacific) in collaboration with Rafael Diez de Medina
and Monica Castillo (both ILO, Department of Statistics) on the basis of a pro-posal
by Farhad Mehran (ILO consultant). The methodology was peer reviewed

xiv Global Wage Report 2014/15
in 2011 by Yves Tillé (Institute of Statistics, University of Neuchâtel), Yujin Jeong
(American University, Washington, DC), Joseph L. Gastwirth (George Washington
University, Washington, DC) and Joyup Ahn (Korea Labor Institute).
Special thanks
Our special thanks go to Manuela Tomei, Director of the ILO’s Conditions of
Work and Equality Department, for her support and guidance. We also owe a
special thanks to all the national statistical offices who assisted us with our data
collection efforts.
We would further like to thank the entire team from ILO/SIALC (Information
System and Labour Analysis) in Panama, in particular Bolívar Pino, for providing
wage data on Latin America and the Caribbean, as well as Malte Luebker (ILO
Regional Office for Asia and the Pacific) for sharing wage data from Asia and for
providing new estimates of real wage growth in India. Special thanks are also due
to Sangheon Lee (ILO, Geneva), who was part of the team of main contributors to
previous editions of the report and who offered much guidance with this new edi-tion.
We also greatly appreciated the invaluable support we received from Edgardo
Greising, Francisco Guerreiro and Marie-Claire Sodergren (ILO, Geneva) with
respect to integrating the data used for this report into ILOSTAT. We would also
like to thank Ekkehard Ernst, Salwa Haidar and Christian Viegelahn for providing
data from ILO Global Employment Trends. Thanks also go to Najati Ghosheh for
continuously pointing us to the most up-to-date academic literature.
We would also like to thank the following for their valuable input and com-ments:
Laura Addati, Azita Berar Awad, Janine Berg, Duncan Campbell, Juan
Chacaltana, Harvey Clavien, Rigoberto Garcia, David Glejberman, Susan Hayter,
James Howard, Martine Humblet, Iyanatul Islam, Samia Kazi Aoul, Michele
Leighton, Nomaan Majid, Andres Marinakis, Adriana Mata-Greenwood, Jesse
Mertens, Irmgard Nübler, Martin Oelz, Shauna Olney, Natalia Popova, Stephen
Pursey, Wolfgang Scholz, Steven Tobin, Raymond Torres, Daniel Vaughan-
Whitehead and the four anonymous peer reviewers.
Data sources
This report is based in part on data from Eurostat, EU-SILC, 2003 to 2010 inclu-sive,
as well as on data from for the US PSID. The responsibility for all conclusions
drawn from the data lies entirely with the authors.

Executive summary
The context
Debates about the economic role of wages have intensified in recent years. At
the firm level, an increase or decrease in wages affects production costs and has
consequences for profitability, sustainability and competitiveness. At the country
level, the net effect of higher or lower wages depends on the direction and rela-tive
magnitude of the effects of wages on household consumption, investment and
net exports. In the Eurozone, concerns over deficits in aggregate demand arising
from insufficient household consumption have focused more attention on wages,
and many commentators have pointed to the decline or stagnation of wages as
increasing the risk of deflation. In some emerging and developing economies, more
attention has been devoted to wages as a key component of overall strategies to
reduce poverty and inequality.
Global wage growth decelerated in 2013 compared to 2012, and has yet
to rebound to pre-crisis rates
Global real wage growth dropped sharply during the crisis in 2008 and 2009, recov-ered
somewhat in 2010 and then decelerated again. Average monthly real wages
grew globally by 2.0 per cent in 2013, down from 2.2 per cent in 2012, and have
yet to rebound to pre-crisis rates of around 3.0 per cent growth in 2006 and 2007.
Global wage growth driven mostly by emerging and developing economies
Global wage growth in recent years was driven by emerging and developing
economies, where real wages have been rising – sometimes rapidly – since 2007.
However, there are major regional variations. While real wage growth in 2013
reached 6 per cent in Asia and nearly 6 per cent in Eastern Europe and Central
Asia, it amounted to less than 1 per cent in Latin America and the Caribbean (down
from 2.3 per cent in 2012). Tentative estimates also show that real wages grew by
almost 4 per cent in the Middle East, due to strong reported wage growth in Saudi
Arabia, but by less than 1 per cent in Africa. Real wage growth in emerging G20
economies slowed from 6.7 per cent in 2012 to 5.9 per cent in 2013.
Global wage growth cut in half when China is excluded
China accounted for much of global wage growth, because of its large size and high
real wage growth. Excluding China cuts global real wage growth almost in half,
from 2.0 per cent to 1.1 per cent in 2013, and from 2.2 per cent to 1.3 per cent in 2012.

xvi Global Wage Report 2014/15
Flat wages in developed economies
In the group of developed economies, real wages were flat in 2012 and 2013,
growing by 0.1 per cent and 0.2 per cent, respectively. In some cases – including
Greece, Ireland, Italy, Japan, Spain and the United Kingdom – average real wages
in 2013 were below their 2007 level. Composition effects (the effect on the average
wage due to the changing composition of workers in paid employment) played a
large role in crisis-affected countries.
Between 1999 and 2013, labour productivity growth in developed economies
outstripped real wage growth, and labour’s share of national income – also
a reflection of the link between wages and productivity – fell in the largest
developed economies
Overall, in the group of developed economies, real wage growth lagged behind
labour productivity growth over the period 1999 to 2013. This was the case before
the crisis in 2007 and – after a brief narrowing of the gap during the depth of the
crisis – labour productivity has continued to outstrip real wage growth since 2009.
Between 1999 and 2013, labour productivity growth exceeded wage growth
in Germany, Japan and the United States. This decoupling of wages and product-ivity
growth is reflected in the decline in the labour income share (the share of
GDP going to labour compensation) over the same period in these countries. In
other countries, such as France and the United Kingdom, the labour income share
remained stable or increased. Among emerging economies, the labour income
share increased in recent years in the Russian Federation, and declined in China,
Mexico and Turkey. It is important to note, however, that when real wage growth
is rapid, the welfare implications of a declining labour income share in emerging
and developing economies may be different from those in developed economies.
Average wages in emerging and developing economies are slowly converging
towards average wages in developed economies
Average wages are still considerably lower in emerging and developing economies
than they are in most developed economies. When measured in purchasing power
parity (PPP), the average monthly wage in the United States, for example, is more
than triple that in China.
While definitional and methodological differences make precise comparison
of wage levels across countries difficult, the average wage in developed economies
is estimated at approximately US$ (PPP) 3,000, as compared to an average wage
in emerging and developing economies of about US$ (PPP) 1,000. The estimated
world average monthly wage is about US$ (PPP) 1,600. However, the gap in real
wages between developed and emerging economies has narrowed between 2000
and 2012, based on strong wage growth in the latter, while in many developed
countries wages stagnated or contracted.

Executive summary xvii
Mixed trends in household income inequality
Growing inequality in many countries over the last few decades has commanded
increasing attention, as high levels of inequality can adversely affect well-being and
social cohesion as well as reducing medium- and long-term economic growth. The
report shows that recent trends in total household income inequality have been
mixed in both developed economies and emerging and developing economies. The
level of inequality is generally higher in the latter; however, progress has been
made in a number of such countries to reduce it, usually in a context of growing
incomes. In developed economies that experienced rising inequality, this has typ-ically
occurred in a context of stagnant or declining incomes.
Inequality starts in the labour market
In many countries, inequality starts in the labour market. Changes in the distribu-tion
of wages and paid employment have been key factors behind recent inequality
trends. In developed economies where inequality increased most, this was fre-quently
due to a combination of more wage inequality and job losses. In Spain
and the United States, the two countries where this inequality between the top and
bottom 10 per cent increased most, changes in the distribution of wages and job
losses accounted for 90 per cent of the increase in inequality in Spain and 140 per
cent of the increase in the United States. In developed countries where household
income inequality increased, other income sources offset about one-third of the
increase in inequality due to changes in wages and employment.
A number of emerging and developing economies experienced declines in
inequality. In these countries, a more equitable distribution of wages and paid
employment was a predominant factor. In Argentina and Brazil, where inequality
fell most, changes in the distribution of wages and paid employment accounted for
87 per cent of the decade-long reduction in top–bottom inequality in Argentina,
as they did for 72 per cent in Brazil.
Wages are a major source of household income
The important role of wages in household income inequality can be explained by
the fact that wages are a major source of household income in both developed
economies and emerging and developing economies. In developed economies,
wages represent about 70 to 80 per cent of total pre-tax, post-transfers income for
households with at least one member of working age, with substantial variation
across countries. In the emerging and developing economies studied in the report,
the contribution of wages to household income is smaller, ranging from about 50 to
60 per cent in Argentina and Brazil to about 40 per cent in Peru and 30 per cent in
Viet Nam. Self-employment income generally comprises a larger share of house-hold
income than in developed economies, particularly among low-income groups.
However, in both types of economies, income sources at both the top and
the bottom are more diverse than in the middle, where households rely mostly

xviii Global Wage Report 2014/15
on wages. In developed economies, social transfers play an important role in
supporting low-income households, whereas in many emerging and developing
economies low-income households rely mostly on self-employment. Among the
bottom 10 per cent of households, for example, wages represent roughly 50 per
cent of household income in the United States, 30 per cent in Italy, 25 per cent in
France, 20 per cent in the United Kingdom, 10 per cent in Germany and 5 per cent
in Romania. Among the middle- and upper-income groups, wages represent the
highest share of household income in almost all countries, reaching about 80 per
cent or more in Germany, the United Kingdom and the United States.
For emerging and developing countries, the share of wages among the bottom
10 per cent of households varies from about 50 per cent of household income in
the Russian Federation to less than 10 per cent in Viet Nam. In Argentina, Brazil,
China and the Russian Federation, the share of wages rises gradually across the
middle classes, before declining in the highest income groups.
Some groups suffer from discrimination and wage penalties
The report shows that in almost all countries studied there are wage gaps between
men and women as well as between national and migrant workers. These gaps
arise for multiple and complex reasons that differ from one country to another
and vary at different points of the overall wage distribution. These wage gaps can
be divided into an “explained” part, which is accounted for by observed human
capital and labour market characteristics, and an “unexplained” part, which
captures wage discrimination and includes characteristics (e.g. having children)
that should in principle have no effect on wages. The report shows that if this
un explained wage penalty was eliminated, the mean gender wage gap would actu-ally
reverse in Brazil, Lithuania, the Russian Federation, Slovenia and Sweden,
where the labour market characteristics of the disadvantaged groups should result
in higher wages. It would also nearly disappear in about half the countries in the
sample of developed economies.
A similar analysis is carried out to compare the wages of migrants with
those of national workers, showing that in various countries the mean wage gap
would reverse if the unexplained part of the gap was eliminated. Among developed
economies, this is the case in Denmark, Germany, Luxembourg, the Netherlands,
Norway, Poland and Sweden. In Chile, migrant workers earn more than their
national counterparts on average.
The report also finds a wage gap between workers in the formal and the
informal economy; this is shown in the report, for example, in wage gaps affecting
workers in the informal economy for selected Latin American countries. As with
gender and migrant wage gaps, the wage gap for workers in the informal economy
is generally lowest in the bottom deciles and increases for higher wage earners.
In addition, the observable labour market characteristics of informal economy
workers differ from workers in the formal economy across all points of the wage
distribution and for all countries (i.e. there is an explained gap across the entire
distribution). At the same time, however, the unexplained part of the wage gap
remains significant.

Executive summary xix
The policy challenge
Inequality can be addressed through policies that affect wage distribution directly
or indirectly and through fiscal policies that redistribute income through taxation
and transfers. However, increasing inequality in the labour market places a heavier
burden on efforts to reduce inequality through taxes and transfers, which is not
always possible or desirable. This suggests that inequality that arises within the
labour market should also be addressed through policies that have a direct effect
on the distribution of wages.
Minimum wages and collective bargaining
Recent research suggests that governments have considerable space for using
minimum wages as a policy tool. On the one hand, research shows that there is
either no trade-off between increased minimum wages and employment levels or
that such increases have very limited effects on employment, which can be either
positive or negative. On the other, it shows that minimum wages do contribute
effectively to reducing wage inequality. Minimum wage policies have been used as
effective policy tools by an increasing number of governments in recent years, in
both developed economies and emerging and developing economies. Importantly,
minimum wages should be set in a way that balances the needs of workers and
their families with economic factors.
Collective bargaining is another labour market institution that has long been
recognized as a key instrument for addressing inequality in general and wage in-equality
in particular. The extent to which collective bargaining can compress
overall wage inequality depends on the proportion of workers covered by collective
agreements and on the position of these workers in the wage distribution.
Promoting job creation
Job creation is a priority in all countries, and the report shows that access to, or
loss of, paid employment is a key determinant of income inequality. In developed
economies, job losses that disproportionally affected low-income workers contrib-uted
to increasing inequality. In emerging and developing economies, the creation
of paid employment for those at the bottom contributed to reducing inequality in
a number of countries. These findings confirm that the pursuit of full-employment
policies is an important aspect of reducing inequality. The promotion of sustain-able
enterprises is key in this regard. This involves establishing an environment
favourable to the creation and development of enterprises, as well as to innovation
and enhanced productivity. The resulting benefits can be shared equitably within
enterprises and in society more broadly.

xx Global Wage Report 2014/15
Special attention to disadvantaged groups of workers
Extending minimum wages and collective bargaining to low-paid workers will
generally be helpful in reducing inequality among women, migrants and vulner-able
groups, who are over-represented among these workers. However, these policy
tools alone will not eliminate all forms of discrimination or wage gaps, which
constitute a significant source of inequality. A wider range of policies is required
to overcome wage gaps across groups that are not explained by human capital
and labour market characteristics. For example, achieving equal pay between men
and women requires policies aimed at combating discriminatory practices and
gender-based stereotypes about the value of women’s work, effective policies on
maternity, paternity and parental leave, as well as advocacy for better sharing of
family responsibilities.
Fiscal redistribution through taxes and social protection systems
Fiscal policies can compensate to some extent for inequality in the labour market,
through both progressive taxation systems and transfers that tend to equalize
household incomes. Such policies are used more frequently by governments in
developed economies to address their income distribution objectives than in
emerging and developing economies, although there may be some convergence
under way. In emerging and developing economies, there appears to be space for
increased tax revenues through a variety of measures, including by broadening
the tax base through the transition of workers and enterprises from the informal
to the formal economy as well as by improving tax collection. Increased revenues
would in turn allow for the extension and upgrading of social protection systems,
which are often not fully developed in such economies.
The need for combined policy action
Wages constitute the largest single source of income for households in developed
and emerging economies alike, with only a few exceptions. At the same time, wages
make a smaller contribution to household income for the lowest income groups.
In developed economies, where social transfers are more important sources of
income for these groups, this calls for a combination of policies that help indi-viduals
in these households move into employment and measures that raise the
quality and compensation of the work they find. In some emerging and developing
economies, raising the income of low-income groups has been achieved through
direct employment programmes (as in India and South Africa) and cash trans-fers
(as in Brazil and Mexico, among many other countries). In the end, the most
effective and sustainable route out of poverty for the working-age population is a
productive, fairly paid job. Policies should be geared towards this objective.

PART I
Major trends
in wages
1 Introduction
Debates about the economic role of wages in the global recovery have intensified
since the publication of the last edition of the ILO Global Wage Report (ILO,
2012a). At the firm level, wages represent a motivating factor for workers, but also
a cost to enterprises. At the country level, however, the sum of all enterprise-level
wage adjustments can have complex effects. Since wages represent an important
source of household income, lower wages tend to reduce household consumption,
thereby depressing aggregate demand, unless that negative effect is offset by sub-stantial
gains in investment and net exports. The net effect of higher or lower wages
on aggregate demand thus depends on the direction and relative magnitude of the
effects of wages on its different components, including household consumption,
investment and net exports.1
This macroeconomic perspective is not new, but after virtually disappearing
from policy debates, it has received new attention in recent years, as many coun-tries
seek to achieve positive growth rates and create jobs while at the same time
trying to reduce imbalances between their imports and their exports. Wage policies
and wage trends play an important role in this equation. In countries with large
current account deficits, where imports exceed exports, wage moderation may con-tribute
to rebalancing trade accounts. This can have a positive effect on aggregate
demand and employment if the boost to the competitiveness of exports is large.
However, if wage cuts reduce domestic consumption by more than is offset by an
increase in exports, the result will be negative in terms of GDP growth, particu-larly
if government spending is reduced at the same time. In countries with large
trade surpluses, higher wage growth can contribute to a rebalancing of demand
towards domestic household consumption and away from exports – with the effect
on aggregate demand in these countries again depending on the magnitude of the
different elements of demand.
At the international level, a lack of policy coordination incurs risks. If too
many countries with close economic ties or in a large geographical area (such as
the European Union) simultaneously pursue a policy of “wage moderation”, the
result is likely to be a shortfall in aggregate demand and a reliance on consumers
in other regions. As the previous Global Wage Report (ILO, 2012a) has pointed
out, in the years before the global economic and financial crisis, wages failed to
increase in line with labour productivity in many countries, particularly – but not
only – in some of the large developed economies. The growing gap between wages
and productivity has translated into a declining labour income share in these coun-tries
(see EC, 2007; Guerriero, 2012; ILO, 2008a; IMF, 2007; Karabarbounis and

2 Global Wage Report 2014/15
Neiman, 2014), and may have contributed to structurally weak aggregate demand
in the years before the crisis (Wolf, 2014a).
In response, policies have started to shift. China, Germany, Japan and the
United States, for example, have recently introduced measures likely to boost wage
growth. Amongst these policy shifts, Germany has adopted a new minimum wage
which will come into effect in 2015, and Japan has encouraged its companies to
increase basic pay. Such policy shifts have been supported by IMF recommenda-tions.
2 In China, authorities are seeking to rebalance the economy towards more
domestic consumption. Meanwhile, the United States, which has long experi-enced
consumption sustained by credit more than wage growth, is also seeking
to encourage real wage growth to “make up for a decades-long trend of average
wages failing to keep pace with productivity gains” (USCEA, 2014, pp. 36–37).3
The IMF noted that “wages are stagnant” in the United States and called for an
increase in the minimum wage, combined with an expansion of the earned income
tax credit (EITC), both to reduce poverty levels and to provide some support to
aggregate demand (IMF, 2014a; Bloomberg, 2014).
In other countries, wage moderation (or even wage cuts) was seen as part of a
set of necessary adjustments to improve competitiveness and sustain employment
levels during the crisis. For example, the Bank of England considered that in the
United Kingdom a rising unemployment rate was prevented by wage moderation.4
European countries most affected by the sovereign debt crisis have also experi-enced
wage cuts. While these wage adjustments may have contributed to fostering
external competitiveness, their effects on aggregate demand – combined with those
of fiscal consolidation5 – have turned out to be more adverse than expected, and
further wage cuts in those countries are likely to be counterproductive (OECD,
2014b).
At the global policy level, the G20 ministers of finance and the ministers
of labour and employment called for the creation of “better jobs” (G20, 2014a,
paras 12–17) and for “policies to reinforce the links between productivity, wages
and employment” (G20, 2013a, para. 8.5).
It is within this global context that Part I of this report presents the latest
trends in average wages.

Part I 2. Trends in average wage growth in the global economy 3
2 Trends in average wage growth in the global economy
Wage trends take place within the broader macroeconomic context and the dif-ferent
experiences of developed economies and emerging and developing econ-omies.
The global economy contracted sharply between 2007 and 2009, quickly
recovered in 2010, but subsequently decelerated (figure 1). While growth rates after
2010 declined across the globe, they remained much higher in emerging and devel-oping
economies than in advanced economies. Several ILO reports 6 have pointed
out that labour markets are still in a state of disarray in many countries, and that
the global jobs deficit is worse today than before the onset of the global financial
and economic crisis in 2008. The challenge of absorbing the estimated 199.4 mil-lion
unemployed and integrating the 395.7 million new entrants into the labour
market in the next decade is enormous.
How have recent economic trends been reflected in average real wages?
Figure 2 provides two estimates (for a definition of wages, see box 1). The first is
a global estimate based on wage data for 130 economies using the methodology
described in Appendix I and the Global Wage Database (see box 2). The second
is also a global estimate, but omits China because of its large size (in terms of
number of wage earners) and high real wage growth, which remained in double
digits for most of the 2000s and accounted for much of the global wage growth.
As can be seen from figure 2, global real wage growth dropped sharply during the
crisis in 2008 and 2009, recovered somewhat in 2010 and then decelerated again.
It has yet to rebound to its pre-crisis rates.
Figure 3 shows estimates for the G20 as a whole and for its developed as
well as its emerging members.7 Together, the countries of the G20 produce about
three-quarters of world GDP 8 and employ more than 1 billion of the world’s
1.5 billion paid employees.9 Average real wage growth throughout the period
Figure 1 Annual average economic growth, 1995–2013 (GDP in constant prices)
World
Emerging market and developing economies
Advanced economies
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
10
8
6
4
2
0
–2
–4
Real GDP growth (%)
Note: Country groups are those used by the IMF as described in the appendix of the IMF’s World Economic Outlook, Apr. 2014.
Source: IMF, World Economic Outlook database, Apr. 2014. Data accessible at: www.ilo.org/gwr-figures

was driven by emerging and developing economies; the same pattern emerges
from the regional figures presented in figure 12 below. By contrast, in G20 devel-oped
economies average real wage growth remained low or negative throughout
the period.
Average wages are calculated using gross monthly wages (where available),
rather than the less frequently available hourly wages, and fluctuations therefore
reflect changes in both hourly wages and the average number of hours worked;
they are also net of consumer price inflation (i.e. deflated by the consumer price
index or CPI).10
Box 1 What are wages?
Wherever possible, in this report wages are defined according to the ILO definition of earnings
adopted by the 12th International Conference of Labour Statisticians (ILO, 1973).
(1) Direct wages and salaries for time worked, or work done, cover: (i) straight-time pay of time-rated
workers; (ii) incentive pay of time-rated workers; (iii) earnings of piece-workers (excluding
overtime premiums); (iv) premium pay for overtime, shift, night and holiday work; and (v) commis-sions
paid to sales and other personnel. Included are: premiums for seniority and special skills;
geographical zone differentials; responsibility premiums; dirt, danger and discomfort allowances;
payments under guaranteed wage systems; cost-of-living allowances; and other allowances.
(2) Remuneration for time not worked comprises: direct payments to employees in respect of
public holidays, annual vacations, and other time off with pay granted by the employer.
(3) Bonuses and gratuities cover: seasonal and end-of-year bonuses; additional payments in
respect of vacation periods (supplementary to normal pay); and profit-sharing bonuses.
Earnings include cash earnings and in-kind payments, but the two should be distinguished from
each other.
There are also related concepts which are broader. For example, while one element of labour
cost is earnings, it also includes other elements such as: food, drink, fuel and other payments in
kind, cost of workers’ housing borne by employers; employers’ social security expenditure; cost
of vocational training; cost of welfare services (e.g. canteen, recreational facilities); labour costs
not classified elsewhere (e.g. cost of work clothes), and taxes regarded as labour cost (e.g. taxes
on employment or payrolls). For a detailed description of these elements, see ILO, 1966.
Source: ILO, 1973.
Box 2 The ILO Global Wage Database
Data underlying the Global Wage Report are accessible at: www.ilo.org/ilostat/GWR
The ILO Global Wage Database covers four indicators from 1995 to 2013 for all ILO member
States (where available): minimum wages, average nominal wages, average real wages and
average real wage growth. In previous years, the data underlying the Global Wage Report
included many additional indicators (e.g. low pay, wage inequality by decile, wages by sex,
etc.). However, since the publication of the Global Wage Report 2012/13, the compilation
of these indicators has been transferred to the “Yearly indicators” collection of ILOSTAT at:
http://www.ilo.org/ilostat).

Part I 2. Trends in average wage growth in the global economy 5
Figure 2 Annual average global real wage growth, 2006–13
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0
Growth (%)
Global
Global (without China)
2.8
2.1
3.1
2.4
1.2
0.5
1.6
0.7
2.2
1.4
1.0
0.1
2.2
1.3
2.0
1.1
2006 2007 2008 2009 2010 2011 2012 2013
Note: Global wage growth is calculated as a weighted average of year-on-year growth in average monthly real wages in
130 countries, covering 95.8 per cent of all employees in the world (for a description of the methodology, see Appendix I).
Source: ILO Global Wage Database. Data accessible at: www.ilo.org/gwr-figures
Figure 3 Annual average real wage growth in the G20, 2006–13
G20 – emerging
–0.8
4.0
1.6
0.6 0.9
6.5
1.2
G20 – developed
8.3
2.6
0.7 0.8
G20
–0.4
6.7
2.4 2.2
0.3 0.3
9.2
5.9
4.8
5.9
3.0
2.4
1.2
2006 2007 2008 2009 2010 2011 2012 2013
10
8
6
4
2
0
–2
Growth (%)
Note: The estimate for the G20 uses the methodology specified in Appendix I, but is restricted to 18 out of 19 individual
countries for which data are available (Argentina identified some inconsistencies in its wage series for some years and
has been excluded).

3 Developed economies
3.1 Average wages: Some variation across countries
Looking at developed economies (for the composition of this and other groupings
used, see Appendix I), it is apparent from figure 4 that the growth rates of average
real wages have tended to fluctuate within a low and narrow range since 2006. This
pattern has become particularly pronounced in 2012 and 2013, years of virtually
flat wages, contributing in the current low inflation environment to concerns about
possible risks of deflation.
Figure 5 looks at the individual developed economy members in the G20,
which represent the largest developed economies in the world. It shows the variety
that exists within the overall trend depicted in figure 4. In France and the United
States, average wages are consistent with the pattern shown in figure 4, having
been relatively stagnant, with only minor fluctuations. However, Australia and
Canada show more positive growth in average wages partially attributed by some
to their natural-resource based growth during a boom in commodities (Downes,
Hanslow and Tulip, 2014; Statistics Canada, 2014). Conversely, notable declines are
observed in Italy and the United Kingdom, where the deep recession was accom-panied
by an unprecedented period of falling real wages. According to the Low
Pay Commission, British wages fell more sharply than at any time since records
began in 1964 (Low Pay Commission, 2014).
Figure 6 shows the extent to which wages changed in selected European coun-tries
most affected by the crisis. Most striking is the large decline in Greek wages,
resulting in part from a series of specific policy measures, including a 22 per cent
cut in the minimum wage for unskilled workers aged 25 and over and a 32 per cent
cut for those under 25 in 2012. Collective bargaining was also decentralized, with
priority given to enterprise-level agreements in cases of conflict with higher-level
agreements, which tended to facilitate downward wage adjustments (ILO, 2014a).
Figure 4 Annual average real wage growth in developed economies, 2006–13
0.9
1.0
–0.3
0.8
0.6
–0.5
0.1
0.2
2006 2007 2008 2009 2010 2011 2012 2013
1.2
0.8
0.4
0
–0.4
–0.8
Growth (%)
Note: Wage growth is calculated as a weighted average of year-on-year growth in average monthly real wages in
36 countries (for a description of the methodology, see Appendix I).

Part I 3. Developed economies 7
Figure 5 Average real wage index for developed G20 countries, 2007–13
110
106
102
98
94
90
Index (base year = 2007)
Germany (102.7)
United States (101.4)
Japan (98.7)
2007 2008 2009 2010 2011 2012 2013
Australia (108.9)
Canada (105.0)
France (102.3)
Italy (94.3)
United Kingdom (92.9)
Figure 6 Average real wage index for selected European countries
most affected by the crisis, 2007–13
110
100
90
80
70
Portugal (103.4)
Ireland (98.1)
Spain (96.8)
2007 2008 2009 2010 2011 2012 2013
Greece (75.8)
Box 3 The composition effect
Average real wages change from year to year not only when the wages of employees increase or
decrease over time, but also when the composition of the labour force changes. For example, if
low-paid employees are the first to be dismissed in a recession, the average wage of the remaining
employees who are employed automatically increases. Conversely, if newly hired employees are
paid less than others, then the average wage automatically decreases. This so-called composi-tion
effect creates a countercyclical bias in the data (understating wage growth in upswings, but
also understating wage declines in downturns), which has at times contributed to a perception
that real wages tend to be “inflexible”. However, some research with panel data (tracking only the
wages of individuals who remain employed) reveals that real wages generally adjust much more
substantially than had been thought during economic ups and downs (see e.g. Martins, Solon and
Thomas, 2012; Abraham and Haltiwanger, 1995; Solon, Barsky and Parker, 1994).

Public sector wages were further reduced several times during the crisis, contrib-uting
to a 23 per cent decline in government expenditure related to compensation
of employees between 2008 and 2012 (ILO, 2014a). In other crisis-affected coun-tries,
the general wage declines have been less dramatic, but the focus on average
wages tends to obscure the full extent of the wage turbulence in these countries.
In Spain, when the composition effect (see box 3) is discounted, the average
wages of those who remained employed seem to have fallen twice as much as the
trend shown in figure 6 (Banco de España, 2014). In Ireland, the proportion of
workers who experienced nominal pay cuts exceeded 50 per cent in both 2009 and
2010 (Doris, Sweetman and O’Neill, 2013). In Portugal, the wages of civil serv-ants
earning more than €1,500 per month and of blue-collar workers were cut,
and those of some 40 per cent of other wage earners were frozen (see ILO, 2013b;
Carneiro, Portugal and Varejao, 2013).
3.2 Productivity outstrips wage growth
Are differences in wage trends across countries a product of differences in labour
productivity growth? Figure 7 shows the relationship between wages and product-ivity
from 1999 to 2013 in the group of developed economies where labour product-ivity
refers to GDP (output) per worker. This definition captures how productively
labour is used to generate output, but also captures the contribution to output of
other elements such as changes in hours worked, changes in the skill composition
of labour, and the contribution of capital.11 While other measures of productivity
exist, labour productivity as defined here is used by the ILO as a decent work
indicator, and is the only one readily available for all countries up to and including
2013.12
Figure 7 shows that after a narrowing of the gap during the depth of the crisis
between 2008 and 2009, labour productivity has continued to outstrip real wage
Figure 7 Trends in growth in average wages and labour productivity in developed
economies (index), 1999–2013
118
115
112
109
106
103
100
Labour productivity index
Real wage index
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Note: Wage growth is calculated as a weighted average of year-on-year growth in average monthly real wages in 36 economies
(for a description of the methodology, see Appendix I). Index is based to 1999 because of data availability.
Sources: ILO Global Wage Database; ILO Trends Econometric Models, Apr. 2014. Data accessible at: www.ilo.org/gwr-figures

Box 4 The relationship between wages, compensation,
different deflators and labour productivity
Since wages represent only one component of labour
costs (see box 1), it may be more appropriate to com-pare
gains in labour productivity with increases in
average compensation per employee (as opposed to
wages). Compensation of employees includes wages
and salaries payable in cash or in kind and social
insurance contributions payable by employers (CEC,
IMF, OECD, UN and World Bank, 2009, para. 7.42).*
To address this argument, figure 8 compares the
change in labour productivity with the changes in
average real wages and in average real compensation
per employee; as can be seen, the gap still persists.
A second area of debate surrounds the most appro-priate
tool to be used to deflate wages or labour com-pensation:
the consumer price index (CPI) or the GDP
deflator. The GDP deflator captures the change in prices
of all goods and services produced in the economy.
In contrast, the CPI reflects the prices which affect
consumers (i.e. the prices of goods and services con-sumers
buy). Thus, the GDP deflator and the CPI can
differ, because the overall price of the basket of goods
bought by consumers is different from the overall price
of all goods produced domestically. So which deflator
is more appropriate in calculating changes in real
wages and/or labour compensation? It depends on the
end-use of the analysis. When the standard of living
is being assessed, the CPI more accurately reflects
changes in the purchasing power of consumers: this is
why the Global Wage Report uses the CPI to calculate
its global and regional estimates for average real wage
growth. However, for enterprises, the relevant measure
may be between productivity and labour compensation
deflated by the product price; in this case, the GDP
deflator may be more appropriate (Feldstein, 2008).
To address this second issue, two additional calcu-lations
are presented in figure 8, which compares
labour productivity growth with real compensation
per employee deflated by both the CPI and the GDP
deflator. The results show that, just as in the case of
the concept of wages/compensation used, it is like-wise
with the different deflators; whichever is used, a
gap remains.
A detailed study by the US Bureau of Labor Statistics
corroborates this finding for the United States: it
shows that while these measurement issues do affect
the exact estimates, they do not affect the overall
conclusion that productivity grew faster than wages
(Fleck, Glaser and Sprague, 2011). There was also a
similar finding for the United States in 2014 in the
Economic report of the President (USCEA, 2014).
* By definition, compensation of employees omits the income of the self-employed. To calculate the adjusted wage share,
there are several methods. See ILO, 2010a, Technical Appendix II.
Figure 8 Labour productivity, real wages and estimated real compensation per employee
in developed economies (indices), 1999–2013
118
115
112
109
106
103
100
Labour productivity index
Real compensation – GDP deflator
Real wage index – GDP deflator
Real compensation – CPI
Real wage index – CPI
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Note: Wage growth is calculated as a weighted average of year-on-year growth in average monthly real wages in 33 economies (for a description of
the methodology, see Appendix I). Iceland, Israel and Malta were excluded from this figure for reasons of data availability; these three countries
collectively contribute less than 1 per cent of the total employees in developed economies. Index is based to 1999 because of data availability.
Sources: ILO Global Wage Database; ILO Trends Econometric Models, Apr. 2014; IMF, World Economic Outlook, Apr. 2014; European Commission
AMECO database. Data accessible at: www.ilo.org/gwr-figures

Figure 9 Estimated real labour compensation per employee and labour
productivity growth in the largest developed economies,
deflated by the CPI and GDP deflator, 1999–2013
2
1
0
–1
Real compensation growth (%)
Compensation/CPI Compensation/GDP deflator
United Kingdom
United Kingdom
J
J
Canada
France
J J
J
J
Germany
Spain
J
Italy
J
J
J
J
United States
Australia United States
J J
J
Japan
J
J
J
Australia
France
45°
Canada
Germany
Italy
Japan
Spain
–1 0 1 2
Labour productivity growth (%)
Sources: European Commission AMECO database; IMF, World Economic Outlook, Apr. 2014; ILO Trends
Econometric Models, Apr. 2014. Data accessible at: www.ilo.org/gwr-figures
growth in this group of countries. Even when changes in real wages are calculated
using not the CPI but the GDP deflator, the trend presented in figure 7 persists
(see box 4).
The overall picture for developed economies is strongly influenced by the
largest economies in the group, in particular Germany, Japan and the United
States. Figure 9 shows the relationship between productivity and real compensa-tion
per employee (as opposed to real wages) for selected developed economies
between 1999 and 2013, using both the CPI and the GDP deflator. Real labour
compensation per employee is used instead of wages since it is more closely linked
to trends in the labour income share (see section 3.3 below). In several coun-tries,
labour productivity grew faster than labour compensation. However, in the
cases of France and the United Kingdom they grew fairly closely in line, while
in Australia, Canada and Italy the relationship between real compensation per
employee and labour productivity growth, during this particular period, depends
on the deflator used.13
3.3 Labour income shares
The labour income share measures the distribution of national income between
labour and capital; when it declines, this implies that a smaller share of national
income is going to workers. As pointed out in the Introduction, a considerable
number of studies have documented the decline in the labour share in many devel-oped
economies since the 1980s. This decline happened in part because of a shift in
employment from labour-intensive to more capital-intensive sectors. But the larger
part of the trend since the 1990s is explained by the fall of the labour income share

within industries, particularly in high- and medium-technology manufacturing,
and in financial services where profits soared.14 The available literature suggests
that this may have been due to a combination of pressure from financial markets
for high capital returns, globalization of international trade, technological change
and the simultaneous erosion of the redistributive power of labour market insti-tutions
(ILO, 2012a). Under most circumstances, a decline in the labour share
reflects a situation where labour productivity is growing faster than average wages.
The potential macroeconomic consequences of a continued decline in the labour
income share can be significant, including the constraining effect on household
income and consumption, which can contribute to the stagnation of aggregate
demand and undermine firms’ incentives to invest due to uncertain future sources
of demand, as recently highlighted in the joint ILO/OECD/World Bank report to
the G20 (ILO, OECD and World Bank, 2014).
Figure 10 shows how the labour income share has changed since 1991 in the
developed G20 countries. The unadjusted labour income only includes compen-sation
of employees, whereas the adjusted labour income share used in figure 10
makes an adjustment to account for the self-employed as well.15 In Canada (and
also in Australia), part of the decline is tied to the rise in commodity prices; profits
in the mining, oil and gas sectors in Canada doubled between 2000 and 2006
(Sharpe, Arsenault and Harrison, 2008; Rao, Sharpe and Smith, 2005). In Japan,
the decline is attributable in part to labour market reforms in the mid-1990s, when
more industries were allowed to hire non-regular workers; the consequent influx
of non-regular workers, who often earned less than regular workers, contributed
to the stagnation of wages over time (Sommer, 2009; Agnese and Sala, 2011). In
France, the labour income share remained relatively stable. In Italy and the United
Kingdom, the trend is unclear: while the labour income share declined in the early
part of the 1990s, since then wages and productivity have grown at a similar pace.
In the United Kingdom, the Low Pay Commission has estimated that employees’
Figure 10 Adjusted labour income share in developed G20 countries, 1991–2013
68
64
60
56
52
Adjusted labour income share (%)
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
United Kingdom
Japan
France
Germany
United States
Canada
Italy
Australia
Note: Adjusted wage share, total economy, as a percentage of GDP at current market prices (compensation per employee
as a percentage of GDP at market prices per person employed).
Source: European Commission AMECO database. Data accessible at: www.ilo.org/gwr-figures

Figure 11 Adjusted labour income share in selected European countries most affected
by the crisis, 1991–2013
65
60
55
50
45
Portugal
Spain
Ireland
Greece
1993 1995 1997 1999 2001 1991 1992 1994 1996 1998 2000 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
compensation and productivity have grown at more or less the same rate since
1964 (Low Pay Commission, 2014). In Italy, one factor contributing to the decline
in the labour income share at the beginning of the 1990s was a set of labour market
reforms that changed the wage bargaining system to curb wage growth (Lucidi
and Kleinknecht, 2010). In Germany, after years of wage moderation, the labour
income share has partly recovered in recent years.
Turning to European countries most affected by the crisis, figure 11 points to
the large decline in the Greek labour income share, to the sharp reversals of wage
shares in the Irish labour market, and to the continuously falling labour income
share in Spain since 2009.

Part I 4. Trends in emerging and developing economies 13
4 Trends in emerging and developing economies
4.1 Higher wage growth
In many emerging and developing economies, average real wages have tended to
increase more rapidly than in developed economies. Figure 12 presents the annual
average real wage growth by region, showing rates that are higher than those of
developed economies in most years.
Figure 12 Average real wage growth in emerging and developing economies, by region, 2006–13
0 0
2006 2007 2008 2009 2010 2011 2012
15
12
9
6
3
–3
–6
Growth (%)
Africa
2006 2007 2008 2009 2010 2011 2012 2013
15
12
9
6
3
–3
–6
Growth (%)
Asia
2013
2006 2007 2008 2009 2010 2011 2012 2013
15
12
9
6
3
–3
–6
Growth (%)
Eastern Europe and Central Asia
2006 2007 2008 2009 2010 2011 2012 2013
15
12
9
6
3
–3
–6
Growth (%)
Latin America and the Caribbean
2006 2007 2008 2009 2010 2011 2012 2013
15
12
9
6
3
–3
–6
Growth (%)
Middle East
0
0
0
3.1* 2.8*
3.7*
1.0*
5.8*
–5.5
1.2 0.9*
7.3 7.7
5.1
6.5 6.3
5.4 5.9 6.0
4.0
2.9
0.2 0.8 0.9 1.1
2.3
0.8
11.6
14.2
8.3
–3.4
6.0
4.5
8.4
5.8
1.0 1.3
–1.5
1.1* (1.1)
0.1*
5.0*
3.9*
* Growth rates published as “tentative estimates” (based on coverage
c. 40–74%).
( ) Growth rates published but likely to change (based on coverage of
less than 40%).
Note: Regional wage growth is calculated as a weighted average of year-on-
year growth in average monthly real wages. For a description of the
methodology, see Appendix I.
Source: ILO Global Wage Database. Data accessible at:
www.ilo.org/gwr-figures

Figure 13 Real wage growth in the largest emerging and developing economies,
2012–13
15
12
9
6
3
0
–3
Growth (%)
9.0
7.3
4.1
1.8
–0.5 –0.6
8.5
5.4
China Brazil Mexico Russian
Federation
2012
2013
14.4
8.2
0.2
3.1
–0.1
5.0
5.6
Ukraine Egypt* South Africa Saudi Arabia
* Data for 2013 are not available for Egypt.
Figure 14 Employees as a share of total employment, 1999 and 2013
90
75
60
45
30
15
0
Percentage of total employment
Africa Asia Latin America
and the Caribbean
1999 2013
Middle East Eastern Europe
and Central Asia
Source: ILO Trends Econometric Models, Apr. 2014. Data accessible at: www.ilo.org/gwr-figures
Developed
economies
Regional trends are influenced by the largest economies in the region.
Figure 13 shows data for the largest countries in the various regions for 2012–13. In
Asia, trends are driven largely by China, where average real wage growth slowed
in 2013 relative to 2012, but remained high. Real wage growth also slowed between
2012 and 2013 in Latin America and the Caribbean, where regional trends are
strongly driven by Brazil and Mexico. In Mexico, real wages declined in both 2012
and 2013, whereas in Brazil real wage growth slowed in 2013. Slowing real wage
growth in 2013 in the Russian Federation and Ukraine drove trends in Eastern
Europe and Central Asia. In Africa, there are considerable data constraints (for
this reason, certain years are marked as tentative in figure 12). Notwithstanding
these limitations, real wage growth declined in Africa in 2011, resumed in 2012 and
slowed in 2013, in significant measure because of reduced or negative real wage

growth in South Africa. Recent trends in the Middle East can be only tentatively
formulated, also owing to data constraints, but are driven in large part by Saudi
Arabia. As noted in box 2, the specific real wage growth rates for individual coun-tries
can be found in the ILO Global Wage Database.
In emerging and developing economies, data constraints make it difficult to
compare wage and labour productivity trends.16 In addition, labour productivity
refers to output per worker, while wages refer only to a subcategory of the working
population, namely employees. Employees typically represent about 85 per cent of
employment in developed countries, but in emerging and developing economies
this proportion is often much lower, and changes more rapidly (see figure 14). For
this reason, a more appropriate comparison in this group of countries would be
between wages and the labour productivity of employees only. Unfortunately,
such data are generally not available. All of these issues create some uncertainty in
analyses related to wages and productivity in emerging and developing economies.
As a result, subsequent analyses for this group of countries focus only on levels
and trends in the labour income share, for which data are more widely available.
4.2 Labour income share can decline despite high wage growth
This section highlights levels and trends in the labour income share for a selec-tion
of emerging and developing economies. As noted in the previous section,
their real wage growth has generally been higher than in developed economies.
However, one should not conflate high real wage growth with a growing labour
income share; the labour income share can decline despite high real wage growth if
labour productivity gains are even higher. In emerging and developing economies,
changes in the labour income share often take place in the context of rapid struc-tural
change and changing patterns of sectoral employment. Thus, composition
effects may be important for explaining changes in the labour income share.
Figure 15 shows the adjusted labour income share for Mexico and Turkey (two
of the larger countries among the group of emerging and developing economies
and for which data are available from the AMECO database). The decline in the
labour income share of Turkey since 1999 has been attributed to a combination of
higher capital inflows, financial liberalization and a decline in the labour income
share in manufacturing following trade liberalization and increased competition
from low-wage economies (Oyvat, 2011). In Mexico, productivity grew faster than
wages, and this is reflected in the decline in the labour income share since the
early 2000s.
Looking at the unadjusted labour income share, because the adjusted labour
income share is not available, the labour income share in China has decreased
since the early 2000s (figure 16). Various explanations have been offered for this
decline despite the rapid increase in wages. These include structural transform-ation
from the agricultural to non-agricultural sectors, where capital is typically
higher, and a decline in the labour income share in industry, attributed by some to
the restructuring of state-owned enterprises and expanded monopoly power (Bai
and Qian, 2010). Others interpret the declining labour income share as consistent
with surplus labour, where productivity gains lead to higher corporate profits and

Figure 15 Adjusted labour income share in Mexico and Turkey, 1995–2013
54
45
36
27
Mexico
Turkey
1995 1997 1996 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Figure 16 Unadjusted labour income share in China, 1992–2011
56
54
52
50
48
46
Unadjusted labour income share (%)
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Source: ILO calculations based on data from the China Statistical Yearbook, various years.
Data accessible at: www.ilo.org/gwr-figures
more investment, but where wage increases in the industrial sector are limited by
the availability of labourers from rural areas (Das and N’Diaye, 2013).
By contrast, in the Russian Federation the labour income share has increased.
The available data offer several measures of the labour income share; figure 17
presents two measures from official sources. The first measure (“without hidden
compensation”) is based on observed wages. The second is adjusted to include an
estimate of “hidden wages” and mixed income. Mixed income refers to the income
received by the owners of household unincorporated enterprises with and without
paid employees (CEC, IMF, OECD, UN and World Bank, 2009, para. 4.160). The
second measure is that used to generate the main official estimate of the aggregate
labour income share. While these two measures diverge in level, they follow each

Figure 17 Labour income share in the Russian Federation, 1995–2012
55
50
45
40
35
30
Labour income share (%)
Rosstat, including hidden wages and mixed income
Rosstat, without hidden wages and mixed income
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Source: Rosstat (see Lukyanova, forthcoming). Data accessible at: www.ilo.org/gwr-figures
Figure 18 Evolution of labour productivity and wages in South Africa, 1994–2011
Real earnings/month
Labour productivity/worker Labour productivity/hour
1995 2000 2005
140
130
120
110
100
90
80
140
130
120
110
100
90
80
Real earnings/hour
2010 1995 2000 2005
Source: Wittenberg, 2014. Data accessible at: www.ilo.org/gwr-figures
2010
other closely in trajectory. The shift away from traditional sectors (agriculture,
transport, manufacturing) towards services between 1995 to 2002 is advanced as
one explanation, although within-sector increases in the labour share appear to be
the dominant source of the overall growth in the aggregate labour income share
(Lukyanova, forthcoming).
In the case of South Africa, where the distribution of productivity gains
is a matter of debate, trends in wages and labour productivity using the Post-
Apartheid Labour Market Series (PALMS)17 show that on average – at least since
the year 2000 – wages and productivity have largely moved in tandem (figure 18)
(Wittenberg, 2014).18 However, national accounts data show a decline in the
un adjusted labour income share for the same period (UNdata, 2014).

5 Towards a progressive convergence in average wages?
Is there evidence of global convergence in average wages between developed econ-omies
and emerging and developing economies? As previously noted, comparing
wage levels across countries is particularly difficult given the variance between
definitions, survey coverage and methodologies.
A comparison of wages in the manufacturing sector across a range of coun-tries
by the US Bureau of Labor Statistics found that average wages are still con-siderably
lower in emerging and developing economies than they are in developed
economies, but also that the gap is slowly narrowing (BLS, 2012).
The persistent difference in wages between developed economies and emerging
and developing economies across the world is evident from figure 19, which shows
the shape of the world distribution of average wages if the abovementioned differ-ences
between countries’ wage data are disregarded and country wages in local
currency are converted to purchasing power parity dollars (PPP$), which capture
the difference in the cost of living between countries.19 The difference in wage levels
between the emerging and developing economies (on the left side of the distribu-tion)
and the developed economies (on the right) is quite substantial. For instance,
the average wage in the United States, measured in PPP$, is more than triple
that in China. However, the figure also shows that the difference in wage levels is
decreasing over time. Between 2000 (the red line) and 2012 (the blue line) the wage
distribution shifts to the right and becomes more compressed; this implies that in
real terms average wages grew across the world, but they grew by much more in
emerging and developing economies. This is consistent with trends in average real
wage growth presented in section 3 of this report. The average wage in developed
economies in 2013 lies at around US$ (PPP) 3,000 compared to an average wage
in emerging and developing economies of about US$ (PPP) 1,000. The estimated
world average monthly wage is about US$ (PPP) 1,600.
Figure 19 Global average monthly wage distribution in 2000 and 2012 (2012 PPP$)
0.0010
0.0008
0.0006
0.0004
0.0002
0
Density
China United States
2000 2012
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Average monthly wage in PPP$
Note: Refers to countries for which wage levels for 2000 and 2012 are available, and covers 83 countries which represent
73 per cent of the world’s employees. Wage levels have been converted to constant 2012 PPP$: PPP$1 is equivalent to the
purchasing power of US$1 in the United States.

PART II
Wages and
income inequality
6 Introduction: The role of wages in household income
inequality
6.1 The context: Long-term inequality on the rise in many countries
since the 1980s
Part I of the report looked at one dimension of distribution: the distribution of
national income between labour and capital. Part II of the report examines how
the distribution of wages contributes to household inequality within countries.
Growing inequality has gained much attention in recent years.20 A large
and expanding literature has shown that inequality, measured in various ways,
has been increasing in a majority of developed economies in recent decades. The
OECD, for example, has documented in detail the rise in inequality in about two-thirds
of developed economies between the early 1980s and the financial crisis of
2008–09 (see OECD, 2008; OECD, 2011). In some countries, including the United
States and the United Kingdom, the rise in inequality has been particularly stark
(Salverda, Nolan and Smeeding, 2009b, p. 5), although top income shares have
increased in other countries too (Piketty, 2013).
Emerging and developing economies – where household income inequality
is often greater than in developed economies – have also not been immune from
this trend: between the 1990s and the 2000s inequality increased in a majority of
such countries for which household survey data are readily available (Ferreira
and Ravallion, 2009; see also Goldberg and Pavcnik, 2007). However, the trend is
not universal. In several Latin American countries, wage and household income
distributions have narrowed, often from a high level, between the mid-1990s and
the onset of the global economic crisis (Gasparini, Cruces and Tornarolli, 2009).
There appears to be a growing consensus that high levels of inequality can
be problematic for several reasons. While some level of inequality is a natural
outcome of heterogeneity in personal characteristics, providing incentives for
people to invest in education and make efforts at work, high and rising income
inequality can become an obstacle to “equality of opportunity” and lead to less
social mobility. With high inequality, economic advantage is more likely to be
inherited than earned (Corak, 2013).21 This can discourage individual effort as
well as damage perceptions of fairness in society and also adversely affect social
outcomes and social cohesion (d’Hombres, Weber and Elia, 2012; Jencks, 2002).
Inequality has also been shown to undermine economic growth. Recent
work by the IMF shows that more inequality reduces “the pace and durability
of growth” (Ostry, Berg and Tsangarides, 2014), while OECD research shows
that greater social and economic inclusion is strongly associated with longer and
stronger periods of sustained economic growth (OECD, 2014a). Although the
links between inequality and growth are complex,22 increasing attention has been

devoted to the adverse effects of inequality on health and education, on political
and economic stability, and on the social consensus required for well-functioning
societies. Inequality has also been highlighted as a factor increasing risk of crisis,
and as one of the possible causes of the 2008 financial crisis in the United States
(see for example Rajan, 2010; Palley and Horn, 2013; Sturn and Van Treeck, 2013;
Kumhof and Rancière, 2010; Krueger, 2012). By redistributing income from poorer
to richer households, growing inequality may have exerted downward pressure on
aggregate demand (as richer households have a lower propensity to consume their
income than poorer households) and encouraged many families to borrow beyond
their means to try to maintain their consumption levels.
6.2 The range of policy responses to inequality
Much of the debate on potential policy responses to growing inequality focuses on
redistribution through fiscal policies (OECD, 2014a; IMF, 2014b). The IMF con-siders
fiscal policy (which includes both taxes and transfers) as “the primary tool
for governments to affect income distribution”, and suggests that “both tax and
expenditure policies need to be carefully designed to balance distributional and
efficiency objectives” (IMF, 2014b, p. 1). It has been estimated that in developed
economies, fiscal policies reduce income inequality, by an average of about one-quarter
(OECD, 2011) to one-third (IMF, 2014b). The ILO has highlighted in par-ticular
the role of social protection systems in reducing inequality and contributing
to more inclusive growth (ILO, 2014f).23 Progressive personal income taxes and in-work
benefits (such as tax credits to low-income workers) play an important role,
but in practice most of the reduction is achieved on the expenditure side through
transfers. In emerging and developing economies, where tax revenues and social
transfers are lower, and indirect taxes dominate, fiscal policies play a much smaller
role in redistribution (IMF, 2014b). In both contexts, there is definitely scope for
reform to achieve more redistribution.24
However, a focus on the role of the labour market is important, too. The
OECD (2011) documented how in developed economies, in the decades before the
crisis, greater wage inequality had been the single most important driver of income
inequality.
Part II of this report extends the existing analysis of the relationship between
the distribution of wages and the distribution of household income. Section 7 looks
at recent trends in household income inequality in developed economies in the
crisis years, and also at trends in emerging and developing economies over roughly
the last decade. Section 8 then studies the role of wages in explaining these recent
trends. In particular, the report decomposes changes in income inequality with
a view to better understanding the role of wages, employment and other factors
in these changes. Seeking to better comprehend how wages can affect household
income inequality, section 9 of the report breaks down the income sources of
households located at different places in the distribution of income, in different
countries. Information on the income sources of different groups of households
can shed light on how different available policy instruments may be appropriate
to reduce inequality. Finally, in section 10, the report decomposes wage gaps for

Part II 7. Mixed recent trends in income inequality 21
women, migrants and workers in the informal economy between an “explained”
component and an “unexplained” penalty (which refers to factors that are “not
accounted for” by those observed human capital and labour market characteristics
that should in principle explain wage differences).
At the broadest level, increasing inequality in the labour market will place a
heavier burden on reducing inequality through secondary (tax and transfer) redis-tribution
mechanisms. These mechanisms have been declining in their impact
as inequality grows, and further redistribution may face political and other con-straints,
notably when directed to the working age population. Therefore, many
governments may wish to consider addressing the problem of rising inequality also
through policy measures that directly address distribution in the labour market.
Part III of this report thus turns to policy response and examines the range of
policies available.
7 Mixed recent trends in income inequality
7.1 Measuring “top–bottom” and “middle-class” inequality
This section documents recent trends in the distribution of total per capita house-hold
income inequality (hereinafter termed income inequality or simply inequality),
where total household income is defined following the Canberra Group as
including income from employment, property and transfers, before the payment
of taxes and other transfers paid (for definitions, see Appendix II).25 A sample
of developed and emerging and developing economies is covered. No single in-equality
measure is perfect, but two measures of inequality are used to assess:
(1) the distance between the top and the bottom 10 per cent within countries (also
referred to as deciles); and (2) the extent to which inequality has changed among
the middle class, within countries.26 Only households in which at least one member
is of working age are considered in the analysis, and income is adjusted for house-hold
size (see Appendix II).
A purely statistical definition of the term “middle class” is used with a view
to identifying whether the statistical middle group of the distribution has become
more “stretched” and dissimilar or, on the contrary, more “compressed” and
similar (for a more sociological discussion of the middle class, see box 5). Since
long-term trends already under way before the crisis are well documented, 27 this
report covers trends over the crisis period, with a first data point before the crisis
and a last point with the latest available data. For the most part, the analysis
covers the period from 2006 to 2010 in developed economies, and a slightly longer
period in emerging and developing economies. The estimates provide information
on the extent to which earlier trends have continued during the crisis, and give
some clues as to how the cost of the crisis has been distributed across different
social groups.
“Top–bottom” inequality is measured by comparing the top and the bottom
of the income distribution: see figure 20, where each person represents 10 per

Box 5 What is understood by the term “middle class”?
In developed economies, many people define themselves as middle class, and so the idea rep-resents
a shared sense of how most people in a society live. In emerging and developing econ-omies,
being middle class is typically more of an aspiration, representing how most people would
like to live.
Economists sometimes represent the middle class as an average class located in an interval of
75 per cent to 125 per cent of the median income (Thurow, 1987, cited by Ravallion, 2010).
But most authors agree that great heterogeneity characterizes the middle class. Consequently,
notions such as lower middle class, upper middle class and even old vs new middle class (Chauvel,
2006) have appeared, conveying a certain difficulty in providing an accurate and precise defin-ition.
When it comes to emerging and developing economies, authors are confronted with the
difficulty of applying a notion that is linked to the Western context. Ravallion simply considers
that “someone is middle class if that person would not be considered poor in any developing
country, implying a lower bound of $9 a day” (Ravallion, 2010, p. 452).
From an economic point of view, the middle class is considered the driver of modern consump-tion
societies (Kharas, 2010). From a sociological perspective, the formation of a social class
includes other factors: a specific economic position (possibly creating conflicts of interest with
other social classes) and a consciousness of having similar conditions (Chauvel, 2006, p. 31).
Today in Western societies, relatively high levels of consumption of goods and services define,
or are a major attribute of, the middle class. Some identify this increasing consumption as con-sumerism
and argue that this profoundly changes traditional social relations within the middle
class because people tend to be more individualistic than class-oriented (Trentmann, 2004), and
social relations are driven by more hedonistic and individualistic patterns (Migone, 2007). In this
view, both factors contribute to the erosion of the class identity in all societies.
In emerging and developing economies, such consumption is still perceived more as a privilege
of the upper middle class than as a unifying social force, as in Western-based societies.1
1 The example of Viet Nam is quite interesting: see Dormeier Freire, 2009.
Source: Alexandre Dormeier Freire, Graduate Institute of International and Development Studies, Geneva
(2014).
cent of the population. The measure of “top–bottom inequality” (also termed
the D9 / D1 ratio) is the ratio between two cut-off points: the threshold value
above which individuals are in the top 10 per cent and the threshold value below
which they are in the bottom 10 per cent of the distribution. Figure 20 also sets
out the boundaries of what is understood in this report as constituting “lower”,
“middle” and “upper” income groups. Middle-class inequality (D7/D3) is meas-ured
by cutting out the top and the bottom 30 per cent of the distribution and
comparing the “entry point” and the “exit point” of a statistical middle, com-prising
the 40 per cent of individuals grouped around the median (as shown in
figure 20).

Figure 20 Measures of inequality
10 per cent of
the population
Lower-income
group
7.2 Developed economies: Mixed inequality trends in a context
of generally flat or declining incomes
In our sample of developed economies, between 2006 and 2010 “top–bottom in-equality”
increased in about half of the countries, and decreased or remained stable
in the remaining countries. Figure 21(a) shows these trends with countries ordered
from left to right, from the countries where inequality decreased to those where
it increased. Using the methodology and data sources described in Appendix II,
inequality increased most in Spain and the United States (where inequality, meas-ured
by the D9/D1 ratio, is highest), and declined most in Bulgaria and Romania.
Over the same period, trends in middle-class inequality in developed econ-omies
have also been mixed, increasing in about half the countries where a
change can be observed and decreasing in the other half (figure 21(b)). Countries
are again ordered from left to right, starting with the countries where inequality
decreased most and moving to the countries where it increased most. We see that
according to our methodology, the country where inequality among the middle
class increased most is Ireland, followed by Spain. On the other side, Romania
and the Netherlands are the two countries in the sample where inequality among
the middle class fell most. The United Kingdom is one example of a country where
middle-class inequality increased while top–bottom inequality remained more or
less stable and even declined somewhat.
The “flattening effect” of the crisis on incomes
In developed economies, these mixed trends frequently took place in a context of
stagnating or declining household incomes between 2007 and 2009/10 (see figure 23).
With the exception of Spain, where inequality increased, some of the countries
most adversely affected by the crisis have seen a reduction in inequality as a result
of a general downward “flattening effect” of the crisis, meaning that incomes have
fallen more for high-income than for lower-income households. Thus, inequality
Bottom
10 per cent
Lower middle-income
group
Upper middle-income
group
Upper-income
group
D1 D3 MEDIAN D7 D9
Top
10 per cent

Figure 21 I nequality in a sample of developed economies in the crisis years, 2006–10:
(a) top–bottom inequality (D9/D1); (b) middle-class inequality (D7/D3)
12
10
8
6
4
2
Ratio D9/D1
(a) Top–bottom inequality
0
3.0
2.5
2.0
1.5
1.0
0.5
Ratio D7/D3
2006 2010
Italy
Greece
Czech Rep.
(b) Middle-class inequality
2006 2010
0
Bulgaria
Romania
Netherlands
Portugal
Norway
Denmark
Hungary
Luxembourg
Belgium
Slovenia
Germany
Note: For methodology, definitions and database, see Appendix II.
Source: ILO estimates. Data accessible at: www.ilo.org/gwr-figures
United Kingdom
Austria
Poland
Ireland
Finland
Estonia
France
Iceland
Slovakia
Latvia
Cyprus
Sweden
Lithuania
United States
Spain
Romania
Netherlands
Bulgaria
Italy
Portugal
Czech Rep.
Greece
Belgium
Hungary
Norway
Denmark
Austria
Slovakia
Finland
Slovenia
Poland
Germany
France
Luxembourg
Cyprus
Iceland
Latvia
Lithuania
United States
United Kingdom
Sweden
Estonia
Spain
Ireland
declined in Romania and Portugal and remained almost unchanged in Greece,
three countries severely hit by the crisis.28 A few countries, such as Denmark, the
Netherlands and Norway, have been able to combine growing household income
and falling inequality during this period.

7.3 Emerging and developing economies: Varying inequality in a context of
Figure 22 Inequality in a sample of emerging and developing economies over the past
decade: (a) top–bottom inequality (D9/D1); (b) middle-class inequality (D7/D3)
30
20
10
Ratio D9/D1
(a) Top–bottom inequality
4
3
2
1
Ratio D7/D3
Around 2000–02 Around 2006–08 Around 2010–12
Brazil Argentina China Peru Russian
Federation
Uruguay Mexico Chile Viet Nam South
(b) Middle-class inequality
0
0
Around 2000–02 Around 2006–08 Around 2010–12
Argentina Viet Nam Indonesia India
Africa
Indonesia India
Brazil Russian
Federation
Peru China Mexico Uruguay Chile South
Africa
generally increasing incomes
In our sample of emerging and developing economies, “top–bottom inequality”
has decreased over the last decade in about half of all countries, and has remained
stable or increased in the other half (figure 22(a)). Measuring inequality in emerging
and developing economies is sometimes difficult due to data limitations, and levels
of inequality may not be strictly comparable across countries (for more discus-sion
on the data, see Appendix II). A longer time frame is considered here since
the crisis did not affect most of these countries as much as it did the developed
economies. The longer time frame also captures some of the structural changes
countries may have undergone during the period. Our estimates suggest that top–
bottom inequality fell most in Brazil and Argentina, and also declined in China,29

Peru, the Russian Federation and Uruguay.30 Among the countries in our sample,
inequality increased most in India and Indonesia. It also increased in Viet Nam,
from a relatively low level, and in South Africa, from a very high level inherited
from the apartheid era. In South Africa, the rise in inequality during 2007–11
occurred because the income growth of bottom decile households stagnated in
real terms, while that of top decile households continued to increase at about the
same rate as in the earlier period.
Turning to the middle class, we see that inequality decreased in almost two-thirds
of our sample of emerging and developing economies, and remained stable
or increased in the remaining countries (figure 22(b)). In Argentina, Brazil and the
Russian Federation, middle-class inequality declined substantially. By contrast,
it remained unchanged in South Africa and increased in India, Indonesia and
Viet Nam.
In contrast to developed economies, in emerging and developing economies
these trends frequently took place in a context of increasing household incomes
(see figure 23). A comparison of figures 21 and 22 also shows that total inequality
remains higher in emerging and developing economies than in developed econ-omies
even after progress on reducing inequality in the former group. The dif-ference
is particularly marked in top–bottom inequality, while the middle class,
though more stretched, shows a proportionally smaller difference in inequality.

Figure 23 Recent evolution of real household income in selected countries
120
120
120
100 100 100
80
60
40
20
0
80
60
40
80
60
40
120
120
120
100 100 100
80
60
40
80
60
40
0
200
150
125
75
50
125
75
50
Index (base year 1999 = 100) India
160
140
120
80
60
Note: For methodology, definitions and databases, see Appendix II.
80
60
40
180
160
140
120
100
100
100
100
80
60
40
20
0
160
140
120
100 100
80
60
40
20
0
20
0
20
0
20
0
20
0
2004 2005 2006 2007 2008 2009 2010
Index (base year 2004 = 100)
Netherlands
2004 2005 2006 2007 2008 2009 2010
Norway
2004 2005 2006 2007 2008 2009 2010
Denmark
2004 2005 2006 2007 2008 2009 2010
20
Spain
2000 2008 2012
50
Russian Federation
2000 2008 2012
25
Brazil
2000 2008 2012
25
Mexico
2000 2008 2012
Viet Nam
1999 2004 2011
40
20
2000 2005 2010
South Africa
2004 2005 2006 2007 2008 2009 2010
Portugal
2006 2007 2008 2009 2010
Romania
0 0
0 0

8 Inequality: The role of wages and paid employment
What has been the role of wages in recent inequality trends? Total household income
inequality can change for many reasons. Household income is composed of income
from employment, property income, income from the production of household ser-vices
for own use, and transfers received.31 The overall change in income inequality
is therefore the sum of the changes of these components (for a detailed explanation,
see Appendix III). For the purposes of both analysis and policy-making, it is im-portant
to understand which of these components drives or dominates changes in
inequality. To what extent can recent trends in inequality be explained by changes
in the distribution of wages as compared to changes in the distribution of other
sources of income? Given the multiplicity of factors that can in theory explain
inequality, identifying the main forces at play is of importance for an informed
policy debate on how in some countries trends towards increased inequality can
be reversed, and on what lessons to draw from the reduction in inequality in other
countries.
8.1 Analysing how wages affect changes in inequality
The extent to which recent changes in inequality have been driven by changes
in wages, rather than by changes in other income sources, is explored through a
decomposition methodology. In general terms, the methodology introduced by
DiNardo, Fortin and Lemieux (1996) and further employed by Daly and Valletta
(2004) proceeds as follows: (1) decompose household income by income source;
(2) observe how the level of inequality changed during a given period; and (3) com-pare
the actual change in inequality to the change in inequality that would have
been observed if the distribution of wages had remained unchanged during the
same period (the “counterfactual”). The difference between the “counterfactual”
level of inequality and the actual level of inequality provides an estimate of the
effect of wages on the change in inequality.32 The same method (described more
fully in Appendix III) can be used to identify and estimate the contribution of
other income sources to the observed change in inequality.
8.2 The labour market effect: Wage effect plus employment effect
At the household level, income from wages depends on both the level of wages
received by members who are paid employees and on the number of members who
are paid employees. Indeed, when undertaking the decomposition, the distribution
of household wages is affected not only by the change in the distribution of wages
of paid employees, but also by the fact that wages fall to zero if some members of
the household lose their jobs as paid employees, or increase from zero to a posi-tive
value when a household member gets a job as paid employee. If a household
member loses his or her job and is eligible for unemployment benefits, wages fall
to zero and unemployment benefits turn positive. As illustrated in figure 24, the
labour market effect is therefore the sum of changes in the wages of those who are

Part II 8. Inequality: The role of wages and paid employment 29
Figure 24 The labour market effect: Wage effect and employment effect
Labour market effect
Wage effect
Employment effect
in paid employment at the beginning and end of the period (the “wage effect”) and
the change in the number of people who are paid employees (the “employment
effect”). In our analysis, we separate these two effects.
8.3 Developed economies: Job losses and wage cuts as the major factors
in changes in inequality
In developed countries, the labour market effect (i.e. wage plus employment effects)
would have increased inequality in two-thirds of countries if other income sources
had not offset the increase. In those countries where inequality did increase, other
income sources offset about one-third of the increase in inequality generated by
the labour market effect. Country-specific developments can be seen in figure 25,
which shows the findings from the decomposition of “top–bottom inequality”
(D9/D1) for developed economies. Countries are ranked from top to bottom,
starting with the country where inequality increased most, to the country where it
declined most, over the period 2006–10. The ranking of countries is thus the same
as in section 7, but figure 25 focuses on the change in (rather than the levels of)
top–bottom inequality. In addition to showing the actual change in inequality, the
figure shows how much of the change was due, respectively, to the wage effect, to
the employment effect and to changes in other sources of income in the household.33
Job losses and top–bottom inequality
When looking at countries where top–bottom inequality increased, labour market
effects (wage plus employment effects) were more important than other income
effects in explaining this increase in a majority of cases. In Spain and the United
States, the two countries where inequality increased most, the labour market
effect accounted for, respectively, 90 per cent and 140 per cent of the increase
in inequality – meaning that in Spain inequality was further increased by other
income sources, while in the United States (as in some other countries) other
income sources partially offset the increase in inequality caused by the labour
market effect. The employment effects dominate the wage effects in countries
where inequality increased the most, suggesting that job losses were the major
cause of top–bottom inequality in these countries during the crisis. (The bars in
figure 25 show that within the labour market effect, the wage effect contributed to

Figure 25 Change in inequality between the top and bottom 10 per cent (D9/D1)
in developed economies, 2006–10
Spain
United States
Lithuania
Sweden
Cyprus
Latvia
Slovakia
Iceland
France
Estonia
Finland
Ireland
Poland
Austria
United Kingdom
Czech Republic
Greece
Italy
Germany
Slovenia
Belgium
Luxembourg
Hungary
Denmark
Norway
Portugal
Netherlands
Romania
Bulgaria
Employment effect
Wage effect
Other income sources
–2.0 –1.5 –1.0 –0.5 0.5 1.0 1.5
0
Actual change in
inequality (D9/D1)

the overall increase in inequality in both Spain and the United States, but in these
two countries the employment effect was even larger, as many workers lost their
jobs and hence their wages.)
Among countries where top–bottom inequality declined, this was predom-inantly
a result of the labour market effect in Germany and Belgium. Note that in
Greece, Romania and Portugal, the wage effect contributed to less inequality; this
occurred because the whole wage distribution was flattened (i.e. wages have fallen
more for high-income than for lower-income households). In Bulgaria, Denmark,
the Netherlands and Norway, while the wage effect contributed to more inequality,
it was more than offset by other factors and inequality declined.
Wages matter more for the middle class
Looking at middle-class inequality (figure 26), the labour market effect contributed
to higher inequality in almost three-quarters of the countries in the sample. In
countries where inequality increased, other income sources offset only about 5 per
cent of the increase. Here again, countries are ranked from top to bottom, from
the country where household income inequality increased most, to the country
where it declined most, over the period 2006–10. As in the D9/D1 analysis (shown
in figure 25), here too the labour market effect is the dominating factor behind the
increase in inequality. It is notable, though, that other incomes offset the increase
in inequality much less among the middle class (as might be expected, since wages
are the major source of household income for the middle classes, as will be seen
later in this report).
When looking at middle-class inequality, labour market effect is dominated by
changes in the distribution of wages rather than by changes in employment in most
countries with increases in middle-class inequality, with Spain the most notable
exception. This was the case for example in Ireland, where middle-class inequality
increased most, but also in other countries where inequality increased, such as
Estonia, Iceland, Sweden and the United States. Considering the labour market
effect in those countries where inequality decreased, the decline in inequality was
exclusively due to the wage effect in Greece, Portugal and Romania. In Bulgaria
and the Netherlands, middle-class inequality fell even though the wage effect
pushed towards more inequality.
Taken together, the evidence shows that the labour market effect was the
largest force pushing towards more inequality over the period 2006–10; other
income sources offset some of these increases in some countries. In this sense, the
last few years have been no different from the three decades before the crisis, when
other evidence shows that increases in inequality were largely driven by changes in
the distribution of wages (see OECD, 2011; Salverda, Nolan and Smeeding, 2009b,
p. 11; Daly and Valletta, 2004). The difference is that during the crisis, employment
played a larger role in explaining changes in inequality.

Figure 26 Change in inequality within the middle class (D7/D3) in developed
economies, 2006–10
Ireland
Spain
Estonia
Sweden
United Kingdom
United States
Lithuania
Latvia
Iceland
Cyprus
Luxembourg
France
Germany
Poland
Slovenia
Finland
Slovakia
Austria
Denmark
Norway
Hungary
Belgium
Greece
Czech Republic
Portugal
Italy
Bulgaria
Netherlands
Romania
Employment effect
Wage effect
Other income sources
–0.21 –0.14 –0.07 0
0.07 0.14 0.21 0.28
Actual change in
inequality (D9/D1)

8.4 Emerging and developing economies: A major role for labour market effects
Among our sample of emerging and developing economies, the labour market
has also been an important factor in explaining inequality trends (even though
wages account for a smaller proportion of total income in those countries, as
will be seen in the next section of the report). In fact, no country in our sample
succeeded in reducing top–bottom inequality over the past decade without also
reducing inequality in the labour market. However, some countries saw declines
in wage inequality while experiencing growing income inequality due to changes in
other sources of household income.34
Top–bottom inequality: the importance of the labour market
Figure 27 shows the findings from the decomposition of top–bottom inequality
(D9/D1) for emerging and developing economies for which data allow the decom-position
technique to be applied, covering roughly the last decade. Countries are
ranked from top to bottom, from the country where inequality increased most to
the one where it declined most. Argentina and Brazil foot the list as the countries
where total inequality fell most: the labour market effect accounted for 72 per cent
of the decade-long reduction in top–bottom inequality in Brazil and 87 per cent
in Argentina. In both countries, other income sources (including transfers) also
contributed to the reduction in inequality, and both the wage and employment
effects contributed to the reduction in inequality attributable to the labour market.
In Peru, the Russian Federation and Uruguay, where inequality also declined, the
decline would have been even greater if other income sources had not contributed
towards more inequality. In Viet Nam, the decrease due to the wage effect was
Figure 27 Change in inequality between the top and bottom 10 per cent (D9/D1)
in selected emerging and developing economies, selected years
Viet Nam (2002–10)
Chile (2000–09)
Mexico (2000–12)
Uruguay (2004–12)
Russian Federation (2002–12)
Peru (2004–12)
Argentina (2003–12)
Brazil (2001–12)
–4.5 –3.0 –1.5 1.5 3.0
Employment effect Wage effect Other income sources Actual change in inequality (D9/D1)
0

Figure 28 Change in inequality within the middle class (D7/D3) in selected emerging and developing
economies, selected years
Viet Nam (2002–10)
Chile (2000–09)
Uruguay (2004–12)
Mexico (2000–12)
Peru (2004–12)
Russian Federation (2002–12)
Argentina (2003–12)
Brazil (2001–12)
–0.60 –0.48 –0.36 –0.24 –0.12 0
0.12 0.24
Employment effect Wage effect Other income sources Actual change in inequality (D9/D1)
insufficient to bring about an overall reduction in inequality. A similar outcome
occurred in South Africa (not shown in figure 27), where the data do not allow
for the decomposition exercise to be applied, but where data from other sources
show an increase in income inequality (see section 7 above) and a decline in wage
inequality.
In those emerging and developing economies where top–bottom inequality
declined, middle-class inequality also declined due to the labour market effect.
Figure 28 shows the decomposition analysis for middle-class inequality (D7/D3),
with countries ranked from top to bottom, from the country where total inequality
increased most to that where it declined most. Argentina, Brazil, Peru and the
Russian Federation show the greatest declines in inequality within the middle
class. In Brazil, the labour market effect accounted for 73 per cent of the fall in
middle-class inequality, and in Argentina for 78 per cent. By contrast, in Viet Nam
middle-class inequality increased despite a decrease due to the wage effect between
2002 and 2010.

Part II 9. Relating inequality to sources of income 35
9 Relating inequality to sources of income
To better understand the role of wages in household income, the report next
addresses the great variation in the weight of income sources across countries,
and across households located at different places in the distribution of income.
This is of key importance in order to: (a) understand how recent changes in wages
and employment have affected households at different parts of the income distribu-tion,
and how this, in turn, has affected income inequality; and (b) develop appro-priate
policy responses, for example with regard to the mix of minimum wages and
transfers. The link between wages and household income is not well documented
in the literature, either for developed economies or for emerging and developing
economies. This report provides some illustrations of the type of information that
policy-makers may find useful in designing policies to address inequality.
9.1 Developed economies: Disaggregating income sources
It is not surprising that, in most developed economies, wages are a major deter-minant
of changes in inequality, given that wages represent about 80 per cent of
household income in the United States and about 70 per cent – with some sub-stantial
variation between countries – in Europe. Figure 29 provides an estimate of
the respective percentages of total household income that, on average, come from
wages and from other income sources across a selection of developed economies.
In contrast to the previous section, this section disaggregates other income sources,
breaking them down into income from self-employment, capital gains, pensions,
unemployment benefits, other social transfers and remaining residual income. As
pointed out earlier, households where no member is of working age are excluded
from the analyses (for sample selection criteria and other aspects of the method-ology,
see Appendix II). In Germany and Sweden, wages represent at least 75 per
cent of household income, whereas in Greece and Italy they account for between
50 and 60 per cent, with self-employment and pensions playing a relatively larger
role than in other developed countries.35 Taken together, pensions, unemployment
benefits and other social transfers represent on average between 15 and 20 per
cent of household income in both Europe and the United States. In all countries,
reported capital gains are a relatively small proportion of reported incomes.36
Non-wage income sources are more important for lower earners
We have seen in section 8 that other (non-wage) income sources play a larger role
in changes in top–bottom inequality than in respect of middle-class inequality.
This reflects the fact that income sources at both the top and the bottom of the
income distribution are more diverse than in the middle, where households rely
mostly on wages. In figure 30, households are ranked in ascending order by their
per capita household income and divided into six groups: the “bottom 10 per cent”,
the “lower” income group (11th–30th percentiles), the “lower middle” class (31st–
50th percentiles), the “upper middle” class (51st–70th percentiles), the “upper”

Figure 29 Share of wages in household income, latest year: Selected developed economies
and European average
100
80
60
40
20
WG = wages; SEI = self-employment income; UB = unemployment benefits; ST = other social transfers; PEN = old-age pensions;
OI = residual income; CG= capital gains.
The European average includes: Austria, Belgium, Bulgaria, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece,
Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, Romania, Slovakia, Slovenia,
Spain, Sweden and the United Kingdom.
In the case of European economies, social transfers include old-age benefits, survivor benefits, sickness benefits, disability benefits and
education allowances (aggregated at the household level) and family and/or children allowances, housing allowances as well as social exclu-sion
allowances not elsewhere classified (e.g., incomes for destitute people, drug addicts, alcoholics or victims of criminal violence, among
others). The aggregation excludes unemployment benefits and old-age pensions which, for the purpose of illustration, are singled out in the
figure. Old-age benefits cover benefits that provide a replacement income when the aged person retires from the labour market, or guarantee
a certain income when a person has reached a prescribed age (private pension plans are included as part of capital gains). Capital gains
include individual private pension plans, dividends from incorporated business, interest and profits received from capital investment in an
unincorporated business in which the person does not work, and income from the rent of property or land. Residual income includes regular
intra-household transfers (e.g. alimony, child support, cash support from households in other countries), in-kind payments, value of goods
produced for own consumption, and income received from family members aged 15 or below with a non-working status.
In the case of the United States, all variables except residual income are defined as in the EU-SILC. Residual income includes income
received from other household members who are neither the head of household nor the spouse of the head (this is the case in about 6 per
cent of households). While the data set indicates that this particular income is not from social transfers, it does not indicate whether it is
from employment or from capital gains.
Source: ILO estimates; see Appendix II.
income group (71st–90th percentiles) and the “top 10 per cent”. As before, these
labels are formulated purely for practical purposes, to facilitate the description of
results, and do not have a sociological interpretation. For all the selected countries
shown in figure 30, it is for the poorest 10 per cent of households that wages repre-sent
the smallest source of household income, and in the middle classes and upper-income
groups that wages frequently make up the largest source of household
income. This pattern can in fact be observed in almost all developed economies.
There is also great variability across countries in the proportion of house-hold
income made up by wages in the top and bottom 10 per cent of households.
Figure 30 shows, for example, that among the bottom 10 per cent, wages represent
about 50 per cent of household income in the United States, more than 30 per in
Italy and about 25 per cent in France. By contrast, in the United Kingdom wages
represent less than 20 per cent of household income among the poorest households,
in Germany less than 10 per cent, and in Romania less than 5 per cent. In all coun-tries,
social transfers play an important role in supporting low-income households
0
Percentage
WG SEI UB ST PEN OI CG
Bulgaria Germany Denmark Greece Spain France Italy Portugal Romania Sweden United
Kingdom
Europe United
States

Figure 30 Household income by group and source in selected developed economies, latest year
100
80
60
40
20
Percentage
0
100
80
60
40
20
Percentage
(c) Romania
0
Lower Lower
100
80
60
40
20
Percentage
100
80
60
40
20
Percentage
0
100
80
60
40
20
Percentage
0
100
80
60
40
20
Percentage
(a) France
Bottom
10%
Lower Lower
middle
Upper
middle
Upper
Upper
Top
10%
(b) Germany
Bottom
10%
Lower
Lower
middle
Upper
middle
Upper
Upper
Top
10%
Bottom
10%
Lower
middle
Upper
middle
Top
10%
(d) United Kingdom
Bottom
10%
Lower
middle
Upper
middle
Top
10%
(e) Italy
Bottom
10%
WG SEI UB ST PEN OI CG WG SEI UB ST PEN OI CG
Lower Lower
middle
Upper
middle
Upper Top
10%
(f) United States
Bottom
10%
Lower Lower
middle
Upper
middle
Upper Top
10%
0
0
(a)–(d): wages represent less than 30 per cent of household income among the poorest 10 per cent of households.
(e)–(f): wages represent more than 30 per cent of household income among the poorest 10 per cent of households.
WG = wages; SEI = self-employment income; UB = unemployment benefits; ST = other social transfers; PEN = old-age pensions; OI = residual income;
CG= capital gains.
Note: “Bottom 10%” reflects up to the first decile; “lower” reflects from above the first decile to the third decile inclusive; “lower middle” reflects from above
the third decile to the fifth included; “upper middle” reflects from above the fifth decile and up to the seventh included; “upper” reflects from above the
seventh decile to the ninth included; “top 10%” reflects the tenth decile.
Source: ILO estimates.

(as compared with other income groups), even though the type of transfers varies
across countries. In Germany, for instance, unemployment benefits and other
social transfers play an almost equally important role, whereas in other countries
unemployment benefits make up a much smaller share of household income in the
bottom 10 per cent.37 Among the middle and upper classes, wages represent the
highest share of household income in almost all countries, reaching about 80 per
cent or more in Germany, the United Kingdom and the United States. In Italy and
France, the richest 10 per cent of households draw a large share of their household
income from income sources other than wages, particularly from self-employment
income and capital gains (even though both of these household income sources are
likely to be underestimated in household surveys).
Figure 31 Real growth in household income by source of income for the top and bottom 10 per cent,
Spain and Romania, 2006–10
120
100
80
60
40
20
WG SEI UB ST PEN OI CG WG SEI UB ST PEN OI CG
0 0
2006 2007 2008 2009 2010
Bottom 10 per cent
(a) Country with increasing inequality: Spain
120
100
80
60
40
20
2006 2007 2008 2009 2010
Top 10 per cent
140
120
100
80
60
40
20
0 0
2006 2007 2008 2009 2010
Bottom 10 per cent
(b) Country with decreasing inequality: Romania
120
100
80
60
40
20
2006 2007 2008 2009 2010
Top 10 per cent
WG = wages; SEI = self-employment income; UB = unemployment benefits; ST = other social transfers; PEN = old-age pensions; OI = residual income;
CG= capital gains.
Note: The D9/D1 ratio used in the report cannot be compared to the ratio of the top to the bottom 10 per cent in this figure, since the former is a “threshold
measure” and this figure shows averages within deciles.

How information on income sources throws light on the sources of inequality
Figure 31 shows the change in income sources in two countries over the period
2006 to 2010 to provide an illustration of why top–bottom inequality (D9 / D1)
increased in Spain (the country in our sample where inequality rose most) and
why it declined in Romania (the country in our sample where inequality declined
most, together with Bulgaria). The figure shows the real change (i.e., adjusted for
inflation) in household income of the top and bottom 10 per cent, broken down
by source of income.
In Spain, growing inequality between 2006 and 2010 is the result of household
income falling more in real terms in the bottom 10 per cent than in the top 10 per
cent (the overall bars – where 2006 serves as the base year equal to 100 – shrink
more for the bottom 10 per cent across time than for the top 10 per cent). Looking
at the different components of the bars, we see that the share of household income
from wages declined in real terms between 2007 and 2010 for those in the bottom
10 per cent. Incomes from self-employment and from pensions also declined. For
the bottom 10 per cent, only income from unemployment benefits increased, but
not enough to prevent a sharp decline in overall real income. For the top 10 per
cent, household income from wages also declined, but by proportionally less than
at the bottom.
Box 6 The largest drop in income inequality in the European Union
during the Great Recession: The puzzling case of Romania
Romania registered the largest decrease in income
inequality among all EU Member States over the
recent recession: a 4.5 point drop in the Gini coef-ficient
between 2007 and 2010. During these years,
the country experienced a severe economic downturn
and some of the toughest fiscal consolidation meas-ures
applied anywhere in the EU, including a 25 per
cent wage cut for public sector employees. At the
same time, some categories of low-income households
saw modest increases in welfare payments. The drop
in inequality thus appears to reflect the impact of wage
cuts at the top of the income distribution and modest
gains for low-income households.
Romania nevertheless remains a country with a high
level of income inequality. This is in large part attribut-able
to the gap between wage earners and individuals
who are self-employed or contributing family workers.
In Romania, self-employed individuals and contributing
family workers represent over 90 per cent of employment
in agriculture, which itself makes up 29 per cent of total
employment – by far the highest level in any EU Member
State. The high share of employment in agriculture, and
the comparatively high proportion of self-employed and
contributing family workers, is a consequence of the
slow transition to a market economy. Only 37.5 per cent
of individuals living in rural areas are employed, while
34.6 per cent are self-employed and a further 27.3 per
cent are contributing family workers. Underemployment
in rural areas is also a key driver of economic inequality
in Romania, as rural inhabitants mix subsistence
farming with various forms of self-employment, informal
employment and cyclical employment abroad to supple-ment
their incomes. On average, households draw only
36 per cent of their income from self-employment in
agriculture, and 55 per cent from self-employment out-side
farming.
It should be pointed out that there are methodological
difficulties in measuring income inequality in Romania.
Unstable and often informal and undeclared forms of
income from self-employment are likely to be under-reported.
To reduce the problem of item non-response,
the National Statistical Office imputes these values.
This widespread imputation of own production of
agricultural products in Romanian household income
surveys significantly decreases income inequality
indicators.
Source: Domnisoru, 2014.

In Romania, a different story emerges: over the whole period 2006–10, top–
bottom inequality declined because household income, in real terms, fell at the top
(the overall size of the bar shrank) but increased slightly at the bottom. Looking
at the different components, wages accounted for a small proportion of house-hold
income in both 2006 and 2010 for households at the bottom: most household
income came from self-employment and from social transfers. In Romania, the
top 10 per cent rely to a much larger extent on wages, although this source of
income has been declining. The fall in inequality in the country may have been
due to fiscal consolidation measures affecting the top of the income distribution,
including public sector wage cuts, and modest gains, mostly from social transfers,
for low-income households (Domnisoru, 2014) (see box 6).
9.2 Emerging and developing economies: A varied and evolving picture
A higher share of self-employment
In emerging and developing economies, the contribution of wages to household
income is usually smaller than in developed economies, ranging from about
50–60 per cent in Argentina, Brazil, Chile, Mexico and the Russian Federation
(levels similar to Greece or Italy) to about 40 per cent in Peru and 30 per cent in Viet
Nam (see figure 32).38 Conversely, self-employment income generally comprises a
larger share of household income than in developed economies, varying from about
10 per cent in the Russian Federation to over 40 per cent in Viet Nam. The lower
average share of wages and higher share of self-employment income are a natural
reflection of the structural difference in employment status between developed econ-omies
and emerging and developing economies: the proportion of employees in total
employment is much lower – and the proportion of self-employed workers much
higher – in the latter (see figure 14 in Part I). The share of income accounted for by
social transfers ranges from about 5 to 20 per cent of household income, although
the content differs depending on the country, as explained in the note to the figure.
The share of wages among the bottom 10 per cent of households varies widely in
the group of emerging and developing economies shown in figure 33. For this group,
wages range from about 50 per cent of household income in the Russian Federation
to less than 10 per cent in Viet Nam. The share of wages in middle-income groups
is also mixed. In Argentina, Brazil, China and the Russian Federation, the share
of wages rises gradually across the middle classes before declining in the highest
income groups. By contrast, in South Africa, the share of wages rises sharply among
the three higher income groups, with social transfers representing the largest share
of income to those in the entire lower half of the distribution.
Strong growth in real wages helps to temper inequality
Contrary to the case of most developed economies, where the context has been
one of lower economic growth or persistent recession, the picture is much more
dynamic in many of the emerging and developing economies. Among countries
where inequality fell, the trend was frequently driven by strong growth in real

Figure 32 Share of wages in household income: Selected emerging
and developing economies, latest year *
100
80
60
40
20
Percentage
WG SEI ST OI
Argentina Brazil Chile Mexico Peru Uruguay Russian
Federation
Viet Nam Indonesia China
0
WG = wages; SEI = self-employment labour income; ST = social transfers; OI = other income.
Note: In the case of emerging and developing economies, self-employment income includes gains from employment as either self-employed
or own-account worker with employees. Whereas for the self-employed (employer) in developed economies it is possible to distinguish
between “income from labour” and their profits, this is not the case for emerging and developing economies: in this case, the reported
self-employment is likely to include profits generated by their self-employment (own account) business. Social transfers are defined for each
economy as described below in detail. Other income includes in-kind payments, capital gains, incomes received by family members aged
15 or below with a non-working status and the difference between the sum of income of all individual household members and the total
income reported for the household in the data set. Capital gains are included as part of other income for emerging and developing economies
because they cannot be distinguished from other income sources.
In the case of Indonesia, the data do not allow for a distinction between self-employment income and social transfers. In the case of China,
social transfers cannot be identified as a unique category by the data and are therefore likely to be included in “other income sources”. See
note 38 with reference to China’s social transfers as described in ILO, 2014f.
In the Russian Federation and Viet Nam, the data cannot uniquely isolate statutory pensions and unemployment benefits. Therefore, social
transfers refer to all incomes from the public social protection system including statutory pensions and unemployment benefits.
In all the Latin American countries represented in the figure (Argentina, Brazil, Chile, Mexico, Peru and Uruguay), social transfers are the
sum of all incomes from the public social protection system (including unemployment benefits, statutory old age pensions and all other
non-contributory social benefits) as well as benefits from private institutions (e.g., benefits received from religious institutions as is explicitly
in the data for Argentina).
* See Appendix III for a definition of the dates and data sources.
Source: ILO estimates based on national sources.
household income and especially by real wages at the bottom. This can be observed
in figure 34, which shows the real change (i.e. adjusted for inflation) of household
incomes of the top and bottom 10 per cent, broken down by source of income, for
selected emerging and developing economies. In the case of Brazil, between 2001
and 2012 real wages of the bottom 10 per cent increased more than those of the top
10 per cent; social transfers and other income sources also increased considerably
(in real terms) during that period at the bottom of the distribution. In the Russian
Federation as well, the decline in total inequality between 2002 and 2012 took
place in a context of much more rapidly increasing household incomes, with social
transfers growing much more at the bottom than at the top, and wages increasing
more at the bottom since 2006.
Income inequality may increase even where wage inequality falls
In some countries, inequality increased in spite of lower wage inequality. In the
case of Viet Nam, the share of wages in household income of those at the bottom
of the income distribution appears to have declined, with increases in household

Figure 33 Household income by group and source in selected emerging
and developing economies, latest year
100
80
60
40
20
Percentage
100
80
60
40
20
Percentage
100
80
60
40
20
Percentage
100
80
60
40
20
Percentage
100
80
60
40
20
Percentage
100
80
60
40
20
Percentage
(a) South Africa
Bottom
10%
Lower Lower
middle
Upper
middle
Upper Top
10%
(a)–(b): wages represent less than 30 per cent of household income among the poorest 10 per cent of households.
(c)–(f): wages represent more than 30 per cent of household income among the poorest 10 per cent of households.
(b) Viet Nam
Bottom
10%
Lower Lower
middle
Upper
middle
Upper
Lower Upper Lower Upper
Top
10%
(c) Brazil
Bottom
10%
Lower
middle
Upper
middle
Top
10%
(d) Argentina
Bottom
10%
Lower
middle
Upper
middle
Top
10%
(e) Russian Federation
Bottom
10%
Lower Lower
middle
Upper
middle
Upper Top
10%
(f) China
Bottom
10%
Lower Lower
middle
Upper
middle
Upper Top
10%
0
0
0
0
0
0
WG SEI ST OI
WG SEI ST OI
WG SEI ST OI
WG SEI ST OI
WG SEI ST OI
WG SEI ST OI

Figure 34 Real growth in household income by income source for the top and bottom 10 per cent,
selected emerging and developing economies, 2001/02 to 2010–12
Brazil Russian Federation
180
150
120
90
60
30
0 WG SEI ST OI WG SEI ST OI WG SEI ST OI WG SEI ST OI
0 0 0
2001 2006 2012
Bottom 10 per cent
120
100
80
60
40
20
2001 2006 2012
Top 10 per cent
280
240
200
160
120
80
40
2002 2006 2012
Bottom 10 per cent
2002 2006 2012
Top 10 per cent
180
150
120
Viet Nam Chile
140
120
100
80
60
40
20
WG SEI ST OI
WG SEI ST OI WG SEI ST OI WG SEI ST OI
0 0 0 0
2002 2006 2010
Bottom 10 per cent
200
160
120
80
40
2002 2006 2010
Top 10 per cent
120
100
80
60
40
20
2002 2006 2011
Bottom 10 per cent
90
60
30
120
100 100)
80
= 2002 year base (Index 60
40
20
2002 2006 2011
Top 10 per cent
Note: The D9/D1 ratio used in the report cannot be compared to the ratio of the top to the bottom 10 per cent in this figure, since the former is a “threshold
measure” and the present figure shows averages within deciles.
incomes depending instead on rising income from self-employment and other
income sources. One possible explanation might be that those with income from
wages have moved up to a higher income group on the income ladder, leaving a
growing share of self-employment income as a percentage of household income in
the bottom 10 per cent. Given the lower proportion of wage earners than in devel-oped
economies, income inequality in emerging and developing economies is linked
not only with wage inequality among paid employees, but also to a large extent
with the gap between wages and income from self-employment. In Chile, inequality
clearly diminished between 2002 and 2006, as transfers to the poorest households
increased, but increased between 2006 and 2011 as wages fell at the bottom.

10 Wage gaps: Which workers earn less than others, and why?
One element of the overall level of total inequality lies in the wage gaps between
different groups of workers. In order to close these wage gaps, it is important to
understand why they exist. The previous section emphasized the importance of
wages as a component of household income and showed how changes in wages
drive changes in total inequality. However, the varying characteristics of workers
across the household income distribution are also important because, although
some inequality can be explained with reference to general economy-wide factors,
such as returns to education, other factors are more specific to the groups that
cluster at the lower end of the wage distribution.
In this final section of Part II, special attention is devoted to groups which
often experience discrimination or other disadvantages in the labour market,
such as women, migrant workers and those who work in the informal economy.
Although labour market institutions and policies such as minimum wages that
aim to benefit all workers can be designed to cover all paid employees in principle,
reducing inequality often requires additional targeted policy action. This report
shows that women, migrants and workers in the informal economy sometimes incur
“wage penalties” for multiple and complex reasons that differ from one country to
another, and that the penalty occurs at different places in the overall wage distribu-tion.
Understanding the reasons for these wage penalties in the national context,
and adopting policies to eliminate them, could make a significant contribution to
reducing income inequality.
Women, migrants, and informal workers
To illustrate the importance of wage gaps, this report calculates some un adjusted
and adjusted wage gaps for women, migrants, and workers in the informal
economy, across the earnings distribution. Unadjusted wage gaps, for example,
refer to the earnings of men, or nationals, or formal economy workers, respectively
(at a given percentile in the wage distribution), minus the earnings of women,
or migrants, or informal economy workers (at the same percentile in the distri-bution).
The un adjusted gaps are then decomposed into an “explained” and an
“un explained” part. The explained part takes into account: experience; education
(grouped into four categories); occupational category (managerial, high-skilled,
semi-skilled, low-skilled and unskilled); economic activity (about ten categories,
including manufacturing, services, and public administration), location (urban,
rural); and work intensity (hours worked). When experience is missing, age is
considered as a proxy since education is also included in the specification. The
unexplained part – or wage penalty – is what remains after adjusting for these
observable labour market characteristics, which should in principle explain pay
differences.39 The full methodology is explained in Appendix IV.
From a policy standpoint, both the explained and unexplained components
are important as channels through which to redress differences in pay across
groups. From the onset, the heterogeneous nature of these groups should be

Part II 10. Wage gaps: Which workers earn less than others, and why? 45
emphasized and accordingly, appropriate policy responses vary. The drivers of
both the explained and unexplained components may overlap. In particular, the
unexplained component may capture discriminatory practices that can also influ-ence
variables considered in the explained component.40
10.1 The gender wage gap
Much research has attempted to interpret the gender wage gap, and the factors that
have been advanced by researchers include: (1) an undervaluation of women’s work;
(2) workplace characteristics (e.g. how substitutable workers are for each other, the
value of face time, etc.); (3) sex segregation channelling women into low value added
jobs; (4) the overall wage structure in a country – which may be shaped by wage-setting
mechanisms that may have been designed with a focus on workers in male-dominated
sectors); (5) the view of women as economic dependants; and (6) the
likelihood that women are in unorganized sectors or not represented in unions
(Goldin, 2014; Chen, Ge, Lai and Wan, 2013; Grimshaw, 2011; Rubery, Grimshaw
and Figueiredo, 2005; Heinze and Wolf, 2010; Rubery, 2003). Family behavioural
patterns may also influence the gap (see box 7 on the motherhood wage gap).
Unexplained factors remain important
Figure 36 shows the gender wage gap, calculated for each decile of the wage distri-bution
and split into an explained and unexplained component, for selected coun-tries.
Wage earners are ranked according to their level of wages, from the lowest
decile to the highest. The total unadjusted wage gap is the sum of the two bars:
the dark bar represents the proportion of the wage gap which can be explained by
observable labour market characteristics, and the light bar is the “unexplained”
gap. The gaps are provided in absolute values: for example, in the first decile in
Belgium there is an unadjusted gender wage gap of about €400, whereas in Estonia
it is about €50. The shapes of the decompositions vary across countries and across
groups. In Belgium and Estonia, women receive lower wages than men throughout
the distribution, but the unexplained part of the gap tends to be higher among
better-paid women. In the United States, the unexplained part is proportionally
small, and affects predominantly better-paid women. In Peru and Viet Nam, the
explained part tends to increase at higher wage levels of the wage distribution. By
contrast, in Sweden the unadjusted gender wage gap is very small (the light and
dark bars generally offset each other; the negative dark bars imply that women
would actually earn more than men if discrimination and other unexplained fac-tors
did not exist). A similar situation can be observed in Chile and in the Russian
Federation, where discrimination and other unexplained factors alone account for
differences in pay between men and women.
Figure 37 presents (1) the level of the average gender wage gap at the national
level for the countries included (the dark bar) and (2) a counterfactual estimate of
the contribution of the unexplained part of the wage gap to the overall un adjusted
wage gap (the light bar). The counterfactual wage gap is the gap which would exist
if men and women were equally remunerated entirely according to the observable

Box 7 The motherhood wage gap
What is the motherhood wage gap?
The motherhood wage gap measures the wage gap between mothers and non-mothers, the latter
being defined in most econometric studies as women without dependent children. This is dif-ferent
from the gender wage gap, which measures the wage gap between all women and all men
in the workforce. While there is a considerable international literature on the motherhood gap,
differences both in methodologies and in definitions of mothers and non-mothers create dif-ficulties
in comparing estimates. Moreover, in many countries, the data are often unsuitable for
analysis, typically because the questions posed in surveys make it difficult to establish a child’s
mother or father (particularly in developing countries where the nuclear family is less common).
Nevertheless, many studies draw on international harmonized pay and employment data that pro-vide
a useful basis for cross-country comparison, and others provide informative trend analyses
for single countries. Data on the motherhood wage gap are provided below for a selection of
countries; however, the figures are not directly comparable, for the reasons just mentioned.
From the available research, it appears that the unadjusted motherhood gap tends to be larger in
developing countries than in developed countries. Globally, the motherhood gap increases with
the number of children a woman has; in many European countries, for example, having one child
has only a small negative effect, but women with two and especially three children experience a
significant wage penalty. In developing countries, evidence suggests that the gender of the child
may matter, as daughters may be more likely than sons to help with household and caring tasks,
thereby reducing the motherhood gap. Whether the wage penalty associated with motherhood
accumulates over time also varies from one country to another. For example, in some countries,
mothers who have a strong job attachment are found to experience a wage decrease immediately
on return to employment, but soon catch up with non-mothers. In short, while the existence of
a motherhood gap seems universal, the magnitude of the effect that motherhood has on wages
varies from country to country.
Explanations for the motherhood wage gap
The main reasons advanced for the motherhood wage gap arise in different analytical frameworks.
An economics approach tends to emphasize: (1) reduced “human capital”, or knowledge, arising
from employment interruptions or reductions in working time, and possible lessened tendency
to seek out training or higher-paid positions with more responsibility; and (2) employment in
family-friendly jobs, such as some part-time positions, which are typically not as well paid.
A sociological approach tends to emphasize that: (1) employers may make hiring and promotion
decisions based on stereotypical expectations of the burdens imposed by families on mothers’
time and energy; (2) the undersupply or affordability of childcare and other work–family accom-modations
are a frequent concern; and (3) female-dominated occupations and workplaces tend
to be less rewarded than male-dominated occupations due to undervaluation of women’s work.
A comparative institutionalist approach emphasizes that: (1) countries provide very different op-portunities
for mothers to access jobs with higher wages through specific policies (e.g. childcare
provision, maternity and paternity leave); (2) a country’s tax and benefit system may confer
tax advantages on mothers who are identified as economically dependent; (3) the size of the
motherhood wage gap varies with the degree of inequality in a country’s overall wage structure;
(4) the cultural and family context matters, especially in countries with less developed policy
interventions; and (5) there may be implementation gaps, where protective or non-discrimination
policies exist but are not enforced or observed, for example in the informal sector or under non-regular
employment arrangements in the formal sector.

Figure 35 Motherhood wage gap in selected emerging and developing economies,
latest year
35
28
21
14
7
Russian
Federation
Argentina Brazil Chile Mexico Peru Uruguay
0
Note: The motherhood wage gap is defined as MPG = ((Enm – Em) ÷ Enm) × 100, where Enm is the average gross
monthly earnings of women who are not mothers and Em is the average gross monthly earnings of mothers.
Source: ILO estimates on trends in the motherhood wage gap. Data accessible at: www.ilo.org/gwr-figures
How to address the motherhood wage gap
The magnitude of the motherhood wage gap and the relevance of some of the abovementioned
explanations depend on the constellation of work–family laws, policies and measures, labour
market institutions, gender stereotypes and societal expectations in a given country. Nonetheless,
there are some general policy recommendations that can be used to address it:
• Job-protected parental leave of adequate duration and with income-related pay funded by
social insurance or public funds for both women and men, with specific provision for fathers.
• Accessibility of affordable and quality childcare services and flexible working arrangements
for all workers.
• Tax and benefit rules that treat mothers as economically independent adults.
• Addressing the implementation gap in work–family and social policies.
• Preventing and eliminating discrimination based on maternity and family responsibilities and
creating a family-friendly workplace culture.
• The right to regulated and flexible working hours, including the upgrading of part-time jobs and
access to them for women and men.
Source: Grimshaw and Rubery, forthcoming.
labour market characteristics taken into account in this report (i.e. education,
experience, economic activity, location, work intensity and occupation). Once
these adjustments are taken into account, in our sample of developed economies
(figure 37(a)) the mean gender wage gap nearly disappears (e.g. Austria, Iceland,
Italy) or even reverses (e.g. Lithuania, Slovenia, Sweden) in about half the coun-tries
in the sample. It declines substantially in other countries but remains largely
explained in Germany and the United States. Among our sample of emerging and
developing economies (see figure 37(b)), the gender wage gap reverses in Brazil and
the Russian Federation. In all other countries in the sample, the wage gap declines

Figure 36 Explained and unexplained gender wage gaps in selected countries, latest year
1000
800
600
400
200
Euros
Belgium
Unexplained
Explained
1st 2nd 3rd 4th Median
DECILE
6th 7th 8th 9th
0
350
280
210
140
70
Euros
Estonia
Unexplained
Explained
DECILE
6th 7th 8th 9th
0
7.5
6.0
4.5
3.0
1.5
0
–1.5
Local currency units (thousand)
Russian Federation
–3.0
Unexplained
Explained
DECILE
6th 7th 8th 9th
3.5
2.8
2.1
1.4
0.7
US dollars (thousand)
United States
0
–0.7
Unexplained
Explained
DECILE
6th 7th 8th 9th
200
150
100
50
Chile
0
–50
Unexplained
Explained
DECILE
6th 7th 8th 9th
350
280
210
140
70
Local currency units
Peru
Unexplained
Explained
DECILE
6th 7th 8th 9th
0
1.0
0.5
Euros (thousand)
0
–0.5
Sweden
–1.0
Unexplained
Explained
DECILE
6th 7th 8th 9th
500
400
300
200
100
Local currency units
Viet Nam
0
–100
Unexplained
Explained
DECILE
6th 7th 8th 9th
Note: In Europe in 2010, the bottom 10 per cent of women earned about €100 per month less than the bottom 10 per cent of men. Conversely, the top 10 per
cent of high-earning women earned close to €700 per month less than the top 10 per cent of men.

Figure 37 Eliminating the unexplained gender wage penalty: Mean wage gap before and
after adjustment in selected economies, latest year: (a) developed economies;
(b) emerging and developing economies
(a) Developed economies
Sweden
Lithuania
Slovenia
Poland
Hungary
Portugal
Romania
Latvia
Denmark
Luxembourg
Spain
Italy
France
Bulgaria
Finland
Europe
Belgium
Slovakia
Greece
Germany
Netherlands
Austria
Norway
Cyprus
Czech Republic
Iceland
Estonia
United Kingdom
Ireland
United States
Viet Nam
Mexico
Peru
China
Chile
Brazil
Uruguay
Argentina
Russian Federation
–20 –10 0
10 20 30 40
Actual wage gap
Explained wage gap
(b) Emerging and developing economies

Figure 38 Wage distribution and counterfactual wage distribution, Russian Federation
and United States, latest year
Gender w Gender wage gap: Russian Federation age gap: United States
100 100
80
60
40
20
Percentage
80
60
40
20
Percentage
(1.5×MED,MAX]
(MED,1.5×MED]
(2/3MED,MED]
(MW,2/3MED]
[0,MW]
0 0
G1 G0 G0–CTFL G1 G0 G0–CTFL
G1: Wage distribution for males; G0: Wage distribution for females; G0-CTFL: Wage distribution for females once the unexplained component
is eliminated and the gap is fully described by the explained component.
substantially, though less so in Argentina and Peru, where much of the gender wage
gap is also due to differences in education and other observable labour market
characteristics. The existence of negative “explained” gender wage gaps (i.e. nega-tive
light bars), in the presence of positive unadjusted wage gaps (i.e. positive dark
bars), points to the importance of gaining a better understanding of the factors that
influence pay for men and women with equal experience, qualifications and other
observable labour market characteristics, in order to address them effectively.
Figure 38 shows the results of applying the counterfactual estimation across
different wage levels for two countries with available data, the Russian Federation
and the United States. The first column shows the distribution of men by wage
level, the second column shows the distribution of women, and the third column
shows the distribution of women absent the unexplained wage gap. Consistent with
figure 36 – which showed that in the United States the unexplained wage gap
is small at the bottom – the elimination of the unexplained component brings
about the greatest increase in the proportion of women in the top category with
wages above one and a half times the median wage (where, according to figure 38,
the unexplained wage penalty is highest). In the Russian Federation, once the
un explained penalty is removed, the percentage of women on low pay declines
considerably, and the proportion earning higher wages equal to at least one and a
half times the median wage increases.
10.2 The migrant wage gap
A similar analysis can be carried out to compare the wages of migrants with those
of national workers. There are many reasons why wages of migrants may differ
from those of nationals, including the fact that their personal characteristics, such

Figure 39 Explained and unexplained migrant–national wage gaps in selected countries, latest year
8
6
4
8
6
4
–1
–2
7.5
5.0
8
6
4
10
8
6
4
2
–2
as skill level, may advantage or disadvantage them in their country of destination.
Part of the gap is also unexplained. Employer discrimination against migrants
because of factors such as prejudice or distrust may account for part of the unex-plained
wage gap (Solé and Parella, 2003). Some research also attributes this to
differences in returns to foreign-acquired education; employers may not be sure of
the quality of the education acquired abroad (Barrett, McGuinness and O’Brien,
2012). Migrants, particularly single migrants, may also receive lower wages than
nationals if they are perceived as having lower income needs than their national
counterparts with families to support (Rubery, 2003). In other cases, migrants
2
Euros (hundred)
Cyprus
10
DECILE
6th 7th 8th 9th
2
Euros (hundred)
Spain
10
DECILE
6th 7th 8th 9th
–3
Brazil
–4
DECILE
6th 7th 8th 9th
2.5
Euros (hundred)
Germany
–5.0
DECILE
6th 7th 8th 9th
2
Euros (hundred)
Norway
10
DECILE
6th 7th 8th 9th
Local currency units (hundred)
Argentina
DECILE
6th 7th 8th 9th
–2
–1
–2.5
–2
Unexplained
Explained
Unexplained
Explained
Unexplained
Explained
Unexplained
Explained
Unexplained
Explained
Unexplained
Explained
0
0
0
0
0
0

may be under- or unrepresented through collective representation structures
because of difficulties organizing or because nationals dominate the overall rep-resentation
– this could be exacerbated if migrants are perceived as a low-wage
employment threat to nationals (Rubery, 2003).
Figure 39 shows that in Germany, for example, high-wage migrant workers
earn less than high-wage nationals, even though they would earn higher wages
than nationals if they were remunerated according to their labour market
attributes (the dark bar is negative). In Argentina as well, the wage gap among
migrant and national top wage earners is exclusively due to the unexplained part.
In Cyprus, even though the overall unadjusted wage gap is higher at the top
than at the bottom of the wage distribution, the unexplained part accounts for
a larger share of the gap at the bottom. This implies that while the wage gap is
smaller at the bottom, migrant workers at the bottom would earn more than
their national counterparts if they were remunerated according to their observ-able
labour market characteristics alone. By contrast, among high wage earners
the gap is large, but can be attributed to migrants’ lower levels of education and
other observable labour market attributes. One exception to this pattern is Brazil,
where according to the available survey data, high-wage migrants (mostly uni-versity
graduates) earn more than high-wage nationals for both explained and
unexplained reasons.
Figure 40 shows what would remain of the wage gap if the unexplained
component were eliminated using the same counterfactual approach as employed
for the gender wage gap above. Among developed economies (figure 40(a)), in
Denmark, Germany, Luxembourg, the Netherlands, Norway, Poland and Sweden,
the mean wage gap reverses when the unexplained part is eliminated, implying
that on average migrant workers may have more education or experience, work in
higher-paid regions, or be more highly skilled, etc., than their national counter-parts.
In most other countries, the migration penalty declines but is not eliminated
after the adjustment. In the emerging and developing economies for which data
permit analysis (figure 40(b)), the results are similar, except in Chile. There, migrant
workers earn more than their national counterparts on average, although if they
were paid according to their observable labour market attributes, they would earn
slightly less than national workers (as shown by the increase in the light bar).
Figure 41 shows the counterfactual applied across the wage distribution for
two countries, Cyprus and Spain. The first column shows the wage distribution
of national employees, whereas the second column presents the same information
for migrant employees. The third column shows how migrants would be distrib-uted
in these groups if the “unexplained” wage gap were eliminated. We see that
in Cyprus, migrant workers are heavily represented in the lowest wage groups.41
However, this picture changes significantly once the unexplained wage penalty
is removed, with the migrant wage distribution becoming more similar to the
national wage distribution. This is consistent with figure 37(a), which shows the
unexplained component contributing more to the wage gap at the bottom of the
wage distribution. By contrast, the corresponding changes in Spain are smaller
because most of the wage gap between migrants and nationals is explained by a
difference in observable factors.

Figure 40 Eliminating the unexplained migrant wage penalty: Mean wage gap before
and after adjustment in selected economies, latest year
(a) Developed economies
Slovenia
Germany
Sweden
United Kingdom
Latvia
Belgium
Poland
Norway
Finland
France
Luxembourg
Austria
Netherlands
Europe
Ireland
Denmark
Estonia
Iceland
Portugal
Italy
Greece
Spain
Cyprus
Brazil
Chile
Uruguay
Argentina
0
Actual wage gap
Explained wage gap
20 30 40
–20 –10 10
(b) Emerging and developing economies
–120 –90 –60 –30 0
30
Figure 41 Wage distribution and counterfactual wage distribution, Cyprus and Spain, latest year
Cyprus Spain
100 100
80
60
40
20
Proportion (100%)
80
60
40
20
Proportion (100%)
(1.5×MED,MAX]
(MED,1.5×MED]
(2/3MED,MED]
(MW,2/3MED]
[0,MW]
0 0
G1 G0 G0–CTFL G1 G0 G0–CTFL
G1: Wage distribution for nationals, G0: Wage distribution for migrants, G0-CTFL: Wage distribution for migrants once the unexplained component
is eliminated and the gap is fully described by the explained component.

10.3 The informal economy wage gap
Analysing informal economy wage gaps is complex, not least because there are
so many different situations and phenomena that can fall into the broad category
of the “informal economy”. From a statistical point of view, informal employment
includes both employment in the informal sector and informal employment outside
the informal sector.42 In practice, workers in the informal economy differ widely in
terms of income, status in employment, sector, type and size of enterprise, and/or
degree of coverage by social and employment protection.
The role of segmentation and profitability
Wage gaps affecting those in the informal economy may arise for a variety of
reasons, but one common assumption is the existence of some form of labour
market segmentation between formal and informal jobs (see box 8 on the case of
Argentina). Formal sector workers may be covered by wage regulations, such as
minimum wages or collective bargaining provisions and social protection systems
to which informal workers may not have access, either because the relevant laws
do not cover them or are not effectively enforced. For example, a recent study using
household and labour force surveys for 11 emerging and developing economies
showed that one-third of employees who were in principle entitled to the minimum
wage were in fact paid less; the proportion paid less is also higher for females,
ethnic minorities, and for workers in the informal economy and rural areas (Rani,
Belser, Oelz and Ranjbar, 2013).
Another explanation is linked to the profitability of firms: if formal firms are
more profitable than informal firms, then they may be more likely to share their
profits with workers, resulting in a wage premium in the formal sector (Rand and
Torm, 2012). Part of the gap may therefore be eliminated through general as well
as tailored measures which facilitate “formalization”, namely, the incorporation
of workers and economic units of the informal economy into the formal economy.
Eliminating the unexplained gap is only part of the solution
Figure 42 shows wage gaps affecting workers in the informal economy for selected
Latin American countries. The observable labour market characteristics of
informal economy workers differ from those of workers in the formal economy
across all points of the wage distribution and for all countries in the sample (i.e.
there is an explained gap across the entire distribution). At the same time, however,
the unexplained part of the wage gap remains significant. In some countries, such
as Argentina (see box 8) and Mexico, the unexplained part of the wage gap is larger
than the explained part for most deciles. In Peru, the explained and unexplained
components contribute to the wage gap about equally, while in Uruguay most of
the wage gap is attributable to workers’ observable labour market characteristics.

Figure 42 Explained and unexplained informality wage gaps in selected Latin American countries,
latest year
5
4
3
2
0.8
0.6
0.4
Mexico
7
6
DECILE
6th 7th 8th 9th
0.2
Peru
1.0
DECILE
6th 7th 8th 9th
2.0
1.6
1.2
Argentina
0
DECILE
6th 7th 8th 9th
8
6
4
Uruguay
10
DECILE
6th 7th 8th 9th
0
0
0.4
0
1
0.8
2
Unexplained
Explained
Unexplained
Explained
Unexplained
Explained
Unexplained
Explained
Box 8 Informality and labour market segmentation in Argentina
In Argentina, the rate of informal employment increased
during the 1990s and then decreased significantly
between 2003 and 2011, although it remains high.
Taken together, the share in total employment of self-employed
and informal paid employees (measured as
the share of workers who received wages but are not
registered with social security) declined from a peak of
more than 60 per cent in the early 2000s to slightly
below 50 per cent in 2011. Among paid employees,
the rate of non-registered employment dropped from
49.1 per cent in 2003 to 34.5 per cent in 2011. The
rate of informal employment among domestic workers
(a sector that accounts for 10 per cent of all paid
employees, mostly women) climbed from 89 per cent
in 1995 to 96.5 per cent in 2003, and then dropped
to 84 per cent in 2011.1 This decline in informality
took place in the context of a different macroeconomic
policy regime from that of the 1990s, and was achieved
in part through a series of policies designed to facili-tate
formalization and a reversal of earlier policies that
had sought to increase labour market flexibility.
In Argentina, results of a statistical analysis which con-trols
for the characteristics of workers show that there
was a significant and positive gap of 24.2 per cent in
2010–11 in transitions from informal to formal wage
employment, and a smaller, though still significant,
gap of 16.3 per cent when moving from informal wage
employment to formal wage employment in atypical
jobs – that is, on fixed-term or involuntary part-time
contracts.
1 This drop in the percentage of non-registered domestic employees is probably due, in part, to the implementation of tax
incentives to register this type of worker.
Source: Bertranou, Casanova, Jiménez and Jiménez, 2013.

PART III
Policy responses to address
wages and inequality
11 Introduction: The policy challenge
As noted in the introduction to Part I, debates about wage policy have shifted
in recent years and many governments have undertaken new or more ambitious
policies to address problems of low wages and wage inequality. This has occurred
in some emerging and developing economies as a key component of overall strat-egies
to reduce poverty and inequality. In some developed economies, concerns
over deficits in aggregate demand arising from insufficient household consumption
have also focused more attention on wages. Many commentators have pointed to
the decline or stagnation of wages in most of the Eurozone as increasing the risk
of deflation (Wolf, 2014b; OECD, 2014b). Such macroeconomic concerns about
aggregate demand and price stability have added to the intensity of the debate
over wages. In addition, renewed attention to the problem of increasing inequality
in many countries has produced evidence that high inequality slows or interrupts
medium- and long-term growth. In an era of slower than trend growth across
the world, this has only intensified the search for the factors that contribute to
inequality and for suitable policy responses.
In this context, this report has presented the recent evolution of wages across
countries and examined the link between wages and inequality. The report has
also shown that much of the recent increase in inequality in many countries is
driven by changes in wages and wage inequality. The report also highlighted the
important role of paid employment in the distribution of incomes. Identifying the
factors contributing to rising inequality provides an essential foundation for con-sidering
appropriate policy responses.
The report has examined the factor that has traditionally been cited as the
most fundamental explanation for the evolution of wages – productivity – and
found that in developed economies productivity growth has outstripped wage
growth over recent decades and that for the group as a whole, the gap has been
widening after a brief pause during the depth of the financial crisis. It is more diffi-cult
to identify this relationship over time for emerging and developing economies
due to insufficient availability of data, but the report finds that the share of GDP
going to labour, which also reflects the link between the growth of wages and
labour productivity, has declined over the last decade in China and Mexico, and
increased in the Russian Federation.
Another key factor traditionally cited to explain the evolution of wages is
returns to education. If higher skills are in greater demand than their supply,
for example due to increased use of technological innovations, wages will rise
more for the highly skilled, leading to increasing wage inequality. Research in
some developed economies has shown that wage premia for graduates have risen

since the 1980s – and remained constant in more recent years – despite an expan-sion
in the supply of graduates with tertiary education (Machin, 2009). However,
wages have particularly increased for those at the very top of the income distribu-tion,
which cannot be linked robustly to education. In emerging and developing
countries, the link between education and wages is mixed and in some countries
appears to have reversed as larger shares of the population gain higher education.
Other factors explaining the evolution of wages and wage inequality, including
globalization and pressure from financial markets for high returns on capital, have
been examined in other studies (see for example OECD, 2011; ILO, 2008b; ILO,
2012a). These factors may also have affected the functioning of labour market
institutions for redistribution, such as collective bargaining.
The report also looked at factors affecting the wage distribution for certain
groups, including women, migrant workers and workers in the informal economy,
to provide additional insights into the drivers of recent wage trends and wage
inequality.
Having contributed to an understanding of the different factors affecting
income and wage inequality, the report now turns to a discussion of appropriate
policy responses. As already noted, inequality can arise and be addressed either
through policies that affect labour market distribution directly (primary distribu-tion)
or through fiscal policies that redistribute through taxation and transfers
(secondary distribution). As noted in section 6.2, fiscal redistribution policies in
developed economies reduce income inequality by about one-quarter to one-third
on average, although the amount of redistribution achieved through those policies
appears to have declined in recent years (OECD, 2011). In emerging and developing
economies, where tax revenues and social transfers are lower and indirect taxes
dominate, fiscal policies tend to play a smaller role in redistribution, although
that has increased in countries where governments have pursued aggressive anti-poverty
and anti-inequality strategies. Across all countries where the labour
share is declining, and in light of increased mobility of capital, there appear to
be structural constraints on the degree of redistribution that can be accomplished
through fiscal policies. This suggests that inequality that arises within the distribu-tion
mechanisms of labour markets must also be addressed through policies that
reform those mechanisms to achieve desired reductions in inequality. With less
inequality in the labour market, there is less of a need for redistribution through
fiscal measures (Berg, forthcoming).
We examine each of these sets of policies in the next sections.

Part III 12. Labour market policies to address wages and inequality 59
12 Labour market policies to address wages and inequality
(primary distribution)
12.1 The role of policies to affect wages and wage distribution directly
Minimum wage policies back in the spotlight
Minimum wages can play an important role in reducing inequality and in sup-porting
the wages of low-paid workers. Recent research supports greater space for
utilizing minimum wages, in that new studies or meta-analyses of earlier studies
show minimum wages either have no negative effects on employment levels or
have very small effects that can be either positive or negative (Betcherman, 2014;
Belman and Wolfson, 2014). This is also the finding from meta-analyses carried
out for the United States and the United Kingdom (see in particular Doucouliagos
and Stanley, 2009, for the United States; Leonard, Stanley and Doucouliagos,
2013, for the United Kingdom).
Fewer studies and detailed meta-analyses exist for emerging and developing
economies (Nataraj, Perez-Arce, Srinivasan and Kumar, 2012). However, recent
studies have shown that minimum wages – sometimes combined with strengthened
collective bargaining – have contributed to reducing inequality in various Latin
American countries (see e.g. Keifman and Maurizio, 2012), as well as in other
emerging and developing economies. In the Russian Federation, for example, it
has been estimated that the reactivation of minimum wage policy explains much
of the observed upgrade in wages at the lower end of the wage distribution between
2005 and 2009, particularly for women (Lukyanova, 2011). In Turkey, the intro-duction
of a mandatory minimum wage in 2004 was also a key factor explaining
the wage growth at the low end of the wage distribution, which reduced wage in-equality
and also reduced excessive working hours (Bakis and Polat, 2013; Gönenç
and Rawdanowicz, 2010). To have the most positive effects, minimum wages must
be well designed and set in a way that balances the needs of workers and their
families with economic factors. This principle is reflected in the texts of the ILO
Minimum Wage Convention, 1970 (No. 131), and the associated Minimum Wage
Fixing Recommendation, 1970 (No. 135), which reflect agreement by governments,
employers’ organizations and workers’ organizations on the essential principles
for minimum wage setting and which have been widely utilized, as well as in the
recent tripartite discussion at the International Labour Conference (ILO, 2014d).43
The role of collective bargaining with respect to wages and wage inequality
Collective bargaining has long been recognized as a key instrument for addressing
inequality in general and wage inequality in particular. The ILO’s Right to
Organise and Collective Bargaining Convention, 1949 (No. 98), has served as a
useful guide to governments, employers’ organizations and workers’ organizations
in designing, developing and utilizing collective bargaining for 65 years.
In practice, countries where a large proportion of workers are covered by col-lective
agreements tend to have lower wage inequality (Visser and Checchi, 2009;

Hayter, forthcoming). This is because collective agreements lift wage floors and
compress wage distributions, reflecting the fact that union members tend to be
motivated both by relative earnings and normative concerns about inequality
(Checchi, Visser and van de Werfhorst, 2010). At the same time, the extent to which
unionization and collective bargaining affect the wage distribution also depends
on whether the collective bargaining system is narrow (where collective bargaining
takes place at the company or workplace level) or more inclusive and encom-passing
(where collective bargaining takes place at the national, industry and/or
branch level in multi-employer settings with coordination across levels) (Visser and
Checchi, 2009; Hayter, forthcoming). In countries with relatively narrow systems,
such as the United States, coverage by collective agreements tends to be lower and
top–bottom wage differentials (D9/D1) higher. In more inclusive and encompassing
systems, collective agreements tend to cover a larger proportion of workers and
top–bottom wage differentials (D9/D1) tend to be lower. Governments may also
take policy measures to extend the application of collective agreements to non-signatories,
thus reinforcing the equity-enhancing effects of collective bargaining.
The extent to which collective bargaining can compress overall wage inequality
depends on the position of unionized workers in the pay distribution, the outcome
of bargaining for different types of workers, and the degree to which collective
bargaining is centralized and coordinated (Bryson, 2007).
Minimum wage policies and collective bargaining should be seen
as complements, not alternatives
Institutions interact, and the relationship between statutory minimum wage pol-icies
and collective bargaining is perhaps of particular importance. In some coun-tries,
relatively high minimum wages can be attributed in part to underdeveloped
collective bargaining systems (Lee and McCann, 2014). In such instances, strength-ening
collective bargaining in addition to effective use of statutory minimum
wages is needed to address inequalities. Where collective bargaining is strong and
most workers are covered by collective agreements, there has been a perception
that there is less need for statutory minimum wages. Yet recent experience in a
number of countries with strong collective bargaining systems, such as Germany,
indicates that changing economic structures and practices can weaken the impact
of collective bargaining systems at low pay levels or in new sectors. In such cases,
both appropriate statutory minimum wage policy and robust collective bargaining
are needed.
Promoting equal pay across subgroups
As seen in section 10, a significant source of wage inequality can be found in exclu-sion
or discrimination against some categories of workers, including women and
migrant workers.
To reduce wage gaps and ensure equal pay across all groups, national legis-lation
must provide for the right to equal remuneration for work of equal value 44
and effective access to justice to claim this right. Labour market institutions and

Part III 12. Labour market policies to address wages and inequality 61
wage policies will only truly be effective in reducing inequality if they are inclusive
and cover workers from groups that are vulnerable, disadvantaged and/or sub-ject
to discrimination. This requires that the labour market institutions and wage
policies do not themselves directly or indirectly discriminate against vulnerable
groups of workers (e.g. by setting lower wage levels in sectors or occupations pre-dominantly
held by women or excluding migrants from coverage of minimum wage
laws). Extending minimum wages and collective agreements to low-paid workers
will not eliminate all forms of discrimination. But given the over-representation of
women, migrants and other vulnerable groups in low-paying jobs, minimum wage
setting and collective bargaining coverage can make a significant contribution to
social justice and more inclusive growth.
Overcoming unequal pay across groups requires sustained effort at various
levels and through a wide range of policy approaches. Equal pay between men and
women needs to be promoted through strong policies to promote gender equality,
including combating gender-based stereotypes about women’s roles and aspirations,
strengthening policies on maternity and paternity as well as parental leave, and
advocacy for better sharing of family responsibilities. Equal pay at the enterprise
level also requires job evaluation methods free from gender bias.
Equal pay between migrant and national workers can be improved by pro-moting
wider adoption of fair and effective labour migration policies that ensure
greater coherence across employment, education/training and development policies
at national, regional and global levels. Creating decent work opportunities in coun-tries
of origin is also essential to make migration a choice rather than a necessity.
To be effective, policies targeting vulnerable workers must be implemented.
This can pose particular challenges in the informal economy and in rural areas,
where labour market institutions such as labour inspectorates or collective bar-gaining
are weak.45 The pay gap between workers in the informal and formal
economies can be reduced by laws, policies and practices which facilitate transi-tions
from the informal to formal economy. The ILO is currently undertaking
a standard-setting process with a view to creating a Recommendation on facili-tating
transition from the informal to the formal economy, and will thus provide
policy guidance in this domain.
Finally, there continues to be a need for improved analysis to understand the
pay gaps in different countries and to monitor their evolution, including a better
understanding of the reasons underlying explained and unexplained components
of pay gaps.
Promoting job creation
Job creation is a priority in all countries, and this report has shown that access
to, and loss of, paid employment are a key determinant of income inequality. In
developed economies, job losses that disproportionately affect low-income workers
contributed to increasing inequality. In emerging and developing economies, the
creation of paid employment for those at the bottom helped reduce inequality in
a number of cases. These findings confirm that the pursuit of full-employment
policies is an important tool in reducing inequality (see also Berg, forthcoming).

In addition to hurting persons looking for paid employment, unemployment or
excess labour supply can also weaken the wage growth of those who are in paid
employment. A comprehensive and integrated approach to achieving the goal of
full employment is embedded in the ILO Employment Policy Convention, 1964
(No. 122), as well as the Global Employment Agenda (2003) and the conclusions
of the recurrent item discussion on employment by the International Labour
Conference (2010). The promotion of sustainable enterprises can be key in this
regard. This involves creating an enabling environment for enterprises to start
and grow, as well as encouraging innovation and enhancing productivity. The
resulting benefits can be shared equitably both within enterprises and in society
more broadly.
12.2 The role of policies to affect wages and wage distribution indirectly
Growing wage inequality in developed economies has been attributed to an
increase in the relative demand for skilled labour due to a combination of techno-logical
change biased in favour of skilled workers (skill-biased technical change)
and the growing exposure of unskilled workers to international competition
(Atkinson, 2007), although, as noted above, wage stagnation among workers with
higher education in recent years suggests that other factors may at times be more
important. In emerging and developing economies, there has been a dynamic evo-lution
of returns to education. In the case of Latin America, the decline in wage in-equality
from the mid-1990s, after an increase in the early 1990s, is often attributed
to the expansion in education and the associated decline in returns to education
(López-Calva and Lustig, 2010; Keifman and Maurizio, 2012; Azevedo, Davalos,
Diaz-Bonilla, Atuesta and Castañeda, 2013; Gasparini and Lustig, 2011; Birdsall,
Lustig and McLeod, 2011; Gasparini, Cruces and Tornarolli, 2009; Cornia, 2012).
In urban China, the rate of returns to education began to climb in the early 1990s
and continued to increase sharply thereafter (Naughton, 2007), leading to higher
average wages, but also to a rise in inequality.
While other factors have influenced these outcomes, sustained attention to
equal access to education, policies to increase the skills of the labour force, par-ticularly
those at the bottom of the distribution, and better analysis to improve
matching of labour skill demand and supply are appropriate in all countries.
The issue of appropriate skills is particularly relevant for current jobseekers.
Indeed, an important source of income inequality is the high incidence of un-employment
in many economies and the resulting downward pressure on wages
for most categories of workers. It is important to ensure that unemployed workers
remain attached to the labour market and retain or upgrade their skills to reduce
the risk that their earnings potential erodes, thereby exacerbating inequality. In this
regard, labour market analysis of skill needs and labour market policies, including
training programmes and job-matching services, are essential to improving re-employment
prospects and mitigating potential sources of future inequality.
In addition to education, many other factors have been highlighted as con-tributing
to wage inequality. Some of these arise in the context of labour mar-kets,
such as the “casualization” of part of the labour force through involuntary

Part III 13. Fiscal policies to address wages and inequality 63
temporary or involuntary part-time employment.46 Policies to address these sources
of inequality include regulatory reforms to avoid lower costs associated with non-standard
workers, such as when part-time workers fall under the threshold of social
security benefits or if fixed-term workers are exempt from severance pay (Gleason,
2006). Others arise at the margin of labour markets, for example changing norms
with regard to the salaries and bonuses of CEOs. While these may be amenable to
policy intervention, they will entail policies outside the labour market.
13 Fiscal policies to address wages and inequality
(secondary distribution)
As noted above, fiscal policies are also employed by governments to address their
income distribution objectives. These tools tend to be used more in developed
economies than in emerging and developing economies, although there may be
some convergence under way; the amount of redistribution achieved through those
policies in developed economies appears to have declined in recent years while a
number of governments in emerging and developing economies have stepped up
their use of transfers to address equity objectives and poverty reduction.
We now turn to the specific elements of the fiscal tools for redistribution:
taxes and transfers.
13.1 The role of taxes
We may be observing increasing structural constraints on the degree of redistribu-tion
that can be accomplished through taxation policies in developed economies.
In countries where the labour income share is declining, revenue mobilization,
through taxes on labour income, will also decline unless (1) the labour share is
increased or (2) tax rates on labour income are increased. In many developed
economies, taxation systems have become less progressive over recent years, and
this suggests that higher rates of taxation on wage income at the low and middle
ranges of the distribution would increase total inequality and be perceived as
highly unfair in light of the increased labour market inequality (OECD, 2012b).
This argues for addressing the declining labour share more directly through the
labour market policies and mechanisms discussed above.
In most emerging and developing economies, there appears to be space for
improved collection of taxes on labour income through a variety of measures.
Depending on country circumstances, these may include efforts to formalize
informal enterprises and workers to broaden the tax base (as well as to bring
them into the realm of contributory social protection schemes, discussed below);
improving the progressivity of tax schedules so that the highest earners pay a
larger share of the overall tax burden; and improved tax collection.

In both developed economies and emerging and developing economies, there
has been some successful experience with targeted tax reduction, elimination or
even positive tax credits to low-income households as ways to increase their overall
income and also to provide positive incentives for work. Policies such as earned
income tax credits for low-wage workers have increased the net income from work
for their households (Immervoll and Pearson, 2009; Immervoll, 2009). While these
mechanisms operate through the tax code, they are more correctly understood as
targeted transfers.
Progressive personal income taxes and directed tax efforts toward greater
equality play an important role, but in practice most of the reduction in inequality
through fiscal policy is achieved on the expenditure side through transfers, dis-cussed
below (IMF, 2014b; OECD, 2011).
The ability of governments to shift taxation to capital (as a source of revenues
for redistribution and a means to increase the progressivity of overall taxation
systems) is also constrained, in this case by the increased mobility of capital in the
context of globalization and reduction of barriers to cross-border capital mobility.
Taxation of capital has also been evaded by firms that use tax avoidance and cross-border
profit-shifting strategies to minimize payment of taxes to any government.
There has been an increasing recognition by governments and experts in recent
years that these avoidance and evasion strategies must be addressed through cross-border
policy coordination. The G20 in particular, supported by the OECD, has
made a priority of addressing these problems, beginning with a campaign for auto-matic
exchange of tax information (G20, 2014b). These efforts have been extended
to countries outside the G20 and OECD membership with some success.
13.2 The role of social protection and its relationship to labour market policies
The other policy instrument for post-market or secondary redistribution is the use
of transfers, which arise in many different forms. These include direct payments
to low-income households, such as: cash transfer systems; the creation of public
employment opportunities for low-income workers or households, such as the rural
employment guarantee scheme in India; direct or subsidized provision by govern-ments
to low-income households of food, fuel or production inputs, such as seeds;
pensions; publicly provided or financed health care; and public education, among
others. Collectively, these efforts are referred to as social protection or social pro-tection
systems. The ILO and others have demonstrated the role of social protec-tion
systems in reducing inequality and contributing to more inclusive growth.47
The appropriate design and mix of social protection measures will vary from
country to country, depending on the overall level of income and economic devel-opment,
the stage of structural transformation of the economy from agriculture to
manufacturing and services, the distribution of the poor between rural and urban
areas, the existence of vulnerable or excluded groups and other factors.48 However,
some general lessons can be drawn from the patterns of wage and total income
distribution presented in sections 7–10 of Part II.

Part III 14. Conclusion 65
14 Conclusion: Combined policy actions can address
inequality, promote employment and support
aggregate demand
As seen in figures 29 and 32, wages constitute the largest single source of income
for households in developed economies and emerging and developing economies,
with only a few exceptions. At the same time, wages make only a small contribu-tion
to household income for the lowest income groups (as seen in figures 30, 31, 33
and 34). In developed economies, this applies mainly to the bottom 10 per cent of
households, where social protection transfers are more important sources of income
for these households in the United States and northern Europe. This suggests that
policies to help individuals in these households to move into employment (through
measures such as active support for job search, training, childcare expenses, trans-portation,
etc.) combined with measures to make the work they find more remu-nerative
(through minimum wages, tax credits, etc.) will help these households to
gradually move up the income scale. In some developed economies, for example in
many southern European countries, self-employment makes up a larger source of
income for the bottom decile of households. Some of the same policies mentioned
above may be appropriate to help individuals in these households move into paid
employment, but in addition, other policies, such as access to financial and busi-ness
services, may be needed as well.
In some emerging and developing economies, wages also represent a relatively
minor share of income for households in the bottom decile, but in some countries
in this group, this pertains throughout the bottom third or more of the income
distribution. With fewer social transfers, income from self-employment and other
income sources (such as remittances) are more important factors. This suggests
that while improving wages will be a part of the policy response to addressing
inequality for these households, other measures to support their incomes will also
be needed. Here successes have been achieved through direct employment pro-grammes
(as in India and South Africa) and conditional or unconditional cash
transfers (as in Brazil and Mexico, among many others). Gradual formalization of
informal self-employment, through measures such as organization of cooperatives,
entrepreneurship training and tailored business development services to microen-terprises,
can also benefit such households. It is also important to recognize that
the most effective route out of poverty for these households is finding a job, and if
that job pays at least the minimum wage the household may move up to the next
decile of the income distribution.
All of these policies can be made more effective by conscious development of
strategies to address inequality and poverty that include coherence and comple-mentarity
between social protection, wage and taxation policies. Gradually, social
protection can lift households above subsistence level and allow them to invest in
their own productive potential, including through education of children, acqui-sition
of skills by adults and productivity-enhancing investment in agriculture
or other self-employment activities. Macroeconomic policies that favour expan-sion
of job creation as well as price stability can help create more employment

opportunities for individuals in these and other households in the lower half of the
income distribution and therefore allow them to benefit more from pro-equity wage
policies. Coherent policies can thus foster a virtuous cycle, whereby higher wages
and incomes lead to more consumption and therefore more aggregate demand and
investment. The same virtuous cycle will also produce higher tax payments and
government revenues, which allow for further policies to increase income equity
and growth without added debt, for example through additional investment in
infrastructure that benefits all segments of society, improvement in the quality of
education and health care for lower-income households, and so on.
Coherent strategies are also needed at international level, to coordinate macro-economic
policies as well as to jointly agree on approaches to wages, employment
and social protection policies to boost household income and consumption. If
individual countries try to maximize their growth through actions to increase
exports such as cutting wages, it may work for one country or several small coun-tries,
but at the global level it will reduce aggregate demand and therefore con-stitute
a beggar-thy-neighbour approach rather than a general welfare-enhancing
approach. If many countries were to follow this approach, the consequences could
be a serious contraction of output and trade.

Appendix I
Global wage trends: Methodological issues
The methodology to estimate global and regional wage trends was developed by
the ILO’s Inclusive Labour Markets, Labour Relations and Working Conditions
Branch (INWORK)49 for the previous Global Wage Reports in collaboration with
the Department of Statistics, following proposals formulated by an ILO con-sultant
and three peer reviews made by four independent experts.50 This appendix
describes the methodology adopted as a result of this process.
Concepts and definitions
y According to the international classification of status in employment (ICSE-93),
“employees” are workers who hold “paid employment jobs”, i.e. 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, 1993b).
y The word “wage” refers to total gross remuneration including regular bonuses
received by employees during a specified period of time for time worked as well
as time not worked, such as paid annual leave and paid sick leave. Essentially,
it corresponds to the concept of “total cash remuneration”, which is the major
component of income related to paid employment (ILO, 1998). It excludes
employers’ social security contributions.
y Wages, in the present context, refer to real average monthly wages of employees.
Wherever possible, we collected data that refer to all employees (rather than
to a subset, such as employees in manufacturing or full-time employees).51 To
adjust for the influence of price changes over different time periods, wages are
measured in real terms, i.e. the nominal wage data are adjusted for consumer
price inflation in the respective country.52 Real wage growth refers to the year-on-
year change in average monthly real wages of all employees.
Census approach
The methodology used for the global and regional estimates is a census method
with non-response. In the census approach, the objective is to find wage data for
all countries and to develop an explicit treatment in the case of total non-response
(see “Treatment of total non-response”, below). We have tried to collect wage data
for a total of 178 countries and territories, grouped into six separate regions.53

Table A1 Regional groups
Regional group Countries and territories
Developed economies Australia, Austria, Belgium, Bulgaria, Canada, Cyprus, Czech Republic, Denmark,
Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Israel, Italy,
Japan, Latvia, Lithuania, Luxembourg, Malta, Netherlands, New Zealand, Norway,
Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden, Switzerland, United
Kingdom, United States
Eastern Europe and Central Asia Albania, Armenia, Azerbaijan, Belarus, Bosnia and Herzegovina, Croatia, Georgia,
Kazakhstan, Kyrgyzstan, Republic of Moldova, Montenegro, Russian Federation, Serbia,
Tajikistan, The Former Yugoslav Republic of Macedonia, Turkey, Turkmenistan, Ukraine,
Uzbekistan
Asia and the Pacific Afghanistan, Bangladesh, Bhutan, Brunei, Cambodia, China, Fiji, Hong Kong (China),
India, Indonesia, Islamic Republic of Iran, Democratic People’s Republic of Korea,
Republic of Korea, Lao People’s Democratic Republic, Macau (China), Malaysia, Republic
of Maldives, Mongolia, Myanmar, Nepal, Pakistan, Papua New Guinea, Philippines,
Singapore, Solomon Islands, Sri Lanka, Taiwan (China), Thailand, Timor-Leste, Viet Nam
Latin America and the Caribbean Argentina,* 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, Nicaragua, Panama,
Paraguay, Peru, Puerto Rico, Suriname, Trinidad and Tobago, Uruguay, Bolivarian
Republic of Venezuela
Middle East Bahrain, Iraq, Jordan, Kuwait, Lebanon, Oman, Qatar, Saudi Arabia, Syrian Arab
Republic, United Arab Emirates, West Bank and Gaza, Yemen
Africa Algeria, Angola, Benin, Botswana, Burkina Faso, Burundi, Cameroon, Cabo Verde,
Central African Republic, Chad, Comoros, Congo, Côte d’Ivoire, Democratic Republic of
the Congo, Egypt, Equatorial Guinea, Eritrea, Ethiopia, Gabon, Gambia, Ghana, Guinea,
Guinea-Bissau, Kenya, Lesotho, Liberia, Libya, Madagascar, Malawi, Mali, Mauritania,
Mauritius, Morocco, Mozambique, Namibia, Niger, Nigeria, Réunion, Rwanda, Senegal,
Sierra Leone, Somalia, South Africa, Sudan, Swaziland, United Republic of Tanzania,
Togo, Tunisia, Uganda, Zambia, Zimbabwe
* Argentina identified some inconsistencies in its wage series for some years and has been excluded.
Table A2 Coverage of the ILO Global Wage Database, 2013 (%)
Regional group Country
coverage
Employee
coverage
Approximate coverage
of total wage bill
Africa 45.1 65.7 81.0
Asia 73.3 98.4 99.5
Developed economies 100.0 100.0 100.0
Eastern Europe and Central Asia 100.0 100.0 100.0
Latin America and the Caribbean 72.4 94.4 95.5
Middle East 75.0 73.1 87.3
World 73.4 95.8 98.6
Note: Country coverage refers to the number of countries for which we found wage data as a percentage of all the countries in the region;
employee coverage refers to the number of employees in countries with data available as a percentage of all employees in the region (as of
2013). The approximate coverage of total wages is estimated based on the assumption that wage levels vary across countries in line with
labour productivity (i.e. GDP per person employed, as of 2013), expressed in 2005 PPP$.

Appendix I Methodological issues 69
To enable easier comparison to regional employment trends, our regional group-ings
are compatible with those used in the ILO’s Global Employment Trends
Model (GET Model) (see table A1). However, we have collapsed several GET
regions into a single region for Asia (which includes the GET regions East Asia,
South-East Asia and the Pacific, and South Asia) and for Africa (which comprises
North Africa and sub-Saharan Africa).
Overall, we found wage data from 130 countries and territories, with regional
coverage indicated in table A2. We have data from developed economies and all
countries in Eastern Europe and Central Asia. Although repeated attempts were
made to obtain wage figures from national statistical offices and/or international
repositories, in some instances wage data were not available. The coverage for
the remaining regions ranges from 45.1 per cent (Africa) to 75.0 per cent (Middle
East). However, since the database includes wage data for the largest and more
prosperous countries, the coverage in terms of employees and the total wage bill
is higher than the simple count of countries would suggest. In total, our database
contains information for 95.8 per cent of the world’s employees, who among them
account for approximately 98.6 per cent of the world’s wage bill.
Treatment of item non-response
In some countries for which we found data, the statistical series were incomplete, in
the sense that data for some years were missing. Table A3 provides coverage infor-mation
for each year from 2006 to 2013. As expected, the coverage of the database
becomes lower for the most recent years since some statistical offices are still pro-cessing
these data. Consequently, for 2013 we have real observations for only about
90.9 per cent of the world’s total wages, as compared to 96.8 per cent in 2011.
While the coverage in the most recent year is good in the developed economies
and in Eastern Europe and Central Asia, in other regions, such as the Middle
East and Africa, it is less so. For this reason, regional growth rates are flagged
Table A3 Coverage of the ILO Global Wage Database, 2006–13 (%)
Regional group 2006 2007 2008 2009 2010 2011 2012 2013
Africa 68.0* 66.9* 67.2* 68.0* 67.8* 79.7 77.7 42.7*
Asia 99.5 99.5 99.5 99.4 99.2 94.0 94.4 86.4
Developed economies 100.0 100.0 100.0 100.0 100.0 100.0 100.0 96.3
Eastern Europe and Central Asia 98.3 98.9 98.8 98.6 98.5 100.0 100.0 97.7
Latin America and the Caribbean 94.4 94.4 94.1 94.0 92.9 95.2 95.2 90.4
Middle East 87.5 87.8 88.0 67.4* (27.4) 64.9* 57.5* 52.0*
World 98.3 98.3 98.2 97.5 96.0 96.8 96.5 90.9
* Growth rates published as “tentative estimates” (based on coverage c. 40–74%).
( ) Growth rates published but likely to change (based on coverage of less than 40%).
Note: See text for estimation of coverage. A country is counted as covered only when a real observation is available, either from a primary or secondary source.

as “tentative estimates” when the underlying coverage of our database is between
40 and 74 per cent, to draw attention to fact that they might be revised once add-itional
data become available.
To address this kind of item non-response (i.e. gaps in countries for which we
have data) a “model-based framework” is used to predict missing values.54 This is
necessary in order to hold the set of responding countries constant over time and
thus avoid the undesired effects associated with an unstable sample. Depending
on the nature of the missing data points, several complementary approaches were
used; these are described in detail in Technical Appendix I of the 2010/11 edition
of the Global Wage Report.
Treatment of total non-response
Response weights
To adjust for total non-response (when no time-series wage data are available for a
given country), a “design-based framework” was used in which non-response was
considered as a sampling problem. Because non-responding countries may have
wage characteristics that differ from those of responding countries, non-response
may introduce a bias into the final estimates. A standard approach to reduce the
adverse effect of non-response is to calculate the propensity of response of different
countries and then weight the data from responding countries by the inverse of
their response propensity.55 This implies that no imputations are made for non-responding
countries.
In this framework, each country responds with a probability φj and it is
assumed that countries respond independently of each other (Poisson sampling
design). With the probabilities of response, φj , it is then possible to estimate the
total, Y, of any variable yj:
j∈U Σ
Y = y j
(1)
by the estimator
öY
=
yj
ϕ j j∈R Σ
(2)
where U is the population and R is the set of respondents. This estimator is un biased
if the assumptions are true (see Tillé, 2001). In our case, U is the universe of all
countries and territories listed in table A1 and R is those “responding” countries
for which we could find time-series wage data.
The difficulty, however, is that the response propensity of country j, φj,
is generally not known and must itself be estimated. Many methods are avail-able
in the literature to estimate the response propensity (see e.g. Tillé, 2001).
In our case, the response propensity was estimated by relating the response
or non-response of a given country to its number of employees and its labour
productivity (or GDP per person employed in 2005 PPP$). This is based on the
observation that wage statistics are more readily available for richer and larger
countries than for poorer and smaller countries. The number of employees and

labour productivity are used since these variables are also used for calibration
and size weighting (see below).56
For this purpose, we estimated a logistic regression with fixed effects as
follows:
prob(response) = Λ(α h +β1xj2006 +β 2nj2006 ) (3)
where xj 2006 is ln(GDP per person employed in 2005 PPP$) of country j in the
year 2006, nj 2006 is ln(number of employees) in 2006, and Λ denotes the logistic
cumulative distribution function (CDF).57 The year 2006 is chosen because it is
the midpoint between 1999 and 2013. The fixed effects, αh, are dummies for each
of the regions with incomplete data (Asia and the Pacific, Latin America and
the Caribbean, the Middle East, Africa), while the two remaining regions with
complete data form the omitted benchmark category. The logistic regression had
a universe of N = 177 cases and produced a pseudo R² = 0.230. The estimated
parameters were then used to calculate the propensity of response of country j, φj.
The response weight for country j, ϕj , is then given by the inverse of a coun-try’s
response propensity:
= 1
φj
ϕj (4)
Calibration factors
The final adjustment process, generally called calibration (Särndal and Deville,
1992), is undertaken to ensure consistency of the estimate with known aggre-gates.
This procedure ensures appropriate representation of the different regions
in the final global estimate. In the present context, a single variable “number of
employees”, n, in a given year t was considered for calibration. In this simple case,
the calibration factors, γit , are given by
jt = nht
γ , j h (5)
ˆ nht
where h represents the region to which country j belongs, nht is the known number
of employees in that region in year t, and nˆ ht is an estimate of total number of
employees in the region and the same year, obtained as a sum product of the
uncalibrated weights and the employment data from the responding countries
within each region.58
The resulting calibration factors for the year 2010 were 1.00 (developed econ-omies;
Eastern Europe and Central Asia), 1.01 (Asia and the Pacific), 0.972 (Latin
America and the Caribbean), 1.03 (Africa) and 1.14 (Middle East). Since all
calibration factors are either equal to or very close to 1, these results show that
estimates nˆ ht were already very close to the known number of employees, nht in
each region. Note the calibration process was repeated for each year so that the
weight of each region in the global estimate changes over time in proportion to its
approximate share in the global wage bill.

Calibrated response weights
The calibrated response weights, ϕ'jt, are then obtained by multiplying the initial
response weight with the calibration factor:
ϕ'jt = ϕj× γit (6)
The regional estimate of the number of employees based on the calibrated response
weights is equal to the known total number of employees in that region and a given
year. Thus, the calibrated response weights adjust for differences in non-response
between regions. The calibrated response weights are equal to 1 in the regions where
wage data were available for all countries (developed economies; Eastern Europe and
Central Asia). They are larger than 1 for small countries and countries with lower
labour productivity since these are under-represented among responding countries.
Estimating global and regional trends
One intuitive way to think of a global (or regional) wage trend is in terms of the
evolution of the world’s (or a region’s) average wage. This would be in line with
the concept used for other well-known estimates, such as regional GDP per capita
growth (published by the World Bank) or the change in labour productivity (or
GDP per person employed).
The global average wage, ȳt, at the point in time t can be obtained by dividing
the sum of the national wage bills by the global number of employees:
yt =
j njt yjt
nj jt
(7)
where njt is the number of employees in country j and ȳjt is the corresponding
average wage of employees in country j, both at time t.
The same can be repeated for the preceding time period t+1 to obtain ȳ*t+1,
using the deflated wages ȳ*jt+1 and the number of employees nt+1. It is then straight-forward
to calculate the growth rate of the global average wage, r.
However, while this is a conceptually appealing way to estimate global wage
trends, it involves some difficulties that we cannot at present overcome. In par-ticular,
aggregating national wages, as done in equation (7), requires them to be
converted into a common currency, such as PPP$. This conversion would make
the estimates sensitive to revisions in PPP conversion factors. It would also require
that national wage statistics be harmonized to a single concept of wages in order
to make the level strictly comparable.59
More importantly, the change in the global average wage would also be
influenced by composition effects that occur when the share of employees shifts
between countries. For instance, if the number of paid employees falls in a country
with high wages but expands (or stays constant) in a country of similar size with
low wages, this would result in a fall of the global average wage (when wage levels
stay constant in all countries). This effect makes changes in the global average
wage difficult to interpret, as one would have to differentiate which part is due to
changes in national average wages and which part is due to composition effects.

We therefore gave preference to an alternative specification to calculate
global wage trends that maintains the intuitive appeal of the concept presented
above but avoids its practical challenges. To ease interpretation, we also want to
exclude effects that are due to changes in the composition of the world’s employee
population. We therefore avoid the danger of producing a statistical artefact of
falling global average wages that could be caused by a shift in employment to low-wage
countries (even when wages within countries are actually growing).
When the number of employees in each country is held constant, the global
wage growth rate can be expressed as a weighted average of the wage growth rates
in the individual countries:
rt
= wj jt rjt (8)
where rjt is wage growth in country j at point in time t and the country weight, wjt,
is the share of country j in the global wage bill, as given by:
wjt = njt yjt / nj jt yjt (9)
While we have data for the number of employees, njt, in all countries and relevant
points in time from the ILO’s Global Employment Trends Model, we cannot esti-mate
equation (9) directly since our wage data are not in a common currency.
However, we can again draw on standard economic theory which suggests that
average wages vary roughly in line with labour productivity across countries.60 We
can thus estimate ȳj as a fixed proportion of labour productivity, LP:
ˆy
jt = LPjt (10)
where α is the average ratio of wages over labour productivity. We can therefore
estimate the weight as
wˆjt = njt LPjt / nj jt LPjt (11)
which is equal to
ˆ wjt = njt LPjt / j njt LPjt (12)
Substituting ŵjt for wjt and introducing the calibrated response weight, ϕ'j, into
equation (8) gives us the final equation used to estimate global wage growth:
rt
=
ϕ'j (13)
j wˆjt rjt
j wˆjt ϕ'j
and for regional wage growth:
rht =
j wˆjt rjt
j wˆjt
, j h
ϕ'j
ϕ'j
(13')

where h is the region of which country j is part. As can be seen from equations
(13) and (13'), global and regional wage growth rates are the weighted averages of
the national wage trends, where ϕ'j corrects for differences in response propensities
between countries.
Differences in global and regional estimates between editions
of the Global Wage Report
Since 2010, when the publication of regional and global wage growth estimates
using the methodology outlined above began, there have been slight revisions to
the historical estimates. While these revisions are relatively minor in some regions,
such as the developed economies and Eastern Europe and Central Asia, they are
more frequent and sometimes substantial in others. The revisions to regional esti-mates
can be explained by several factors, briefly highlighted here.
1. Improvements and revisions to surveys which collect wage data. Improvements
and revisions to existing wage data and surveys often occur. These may include
a change in the geographical coverage (e.g. from urban to national), a change in
sector coverage (e.g. from manufacturing to all sectors), a change in employee
coverage (e.g. from full-time employees only to all employees) etc. To the extent
these changes influence the growth in wages, they may also influence the
regional estimate.
2. Exclusions. In Latin America, Argentina has been excluded since the 2012 ver-sion
of the Global Wage Report (ILO, 2012a) because it identified inconsisten-cies
in its wage series.
3. Availability of new data from non-response and response countries. Particularly
in emerging and developing economies, there is often a lag in the process time
for data and/or their public availability. When new or older series are made
available, they are incorporated into the regional estimates.
4. Revision of other data sources used to calculate the estimates. Over time, revi-sions
to the CPI, total employment, total employees and labour productivity
can also influence regional and country estimates.

Definitions, concepts and data issues
Appendix II
Concepts, definitions and measures
As previously mentioned, wages and household income are two separate, but
related, concepts. While wages refer to gross remuneration in cash and in kind paid
to employees, household income is measured at the household level, and includes
all receipts received by the household or by its individual members (see definitions
in box A1). Wages are a key source of household income, but they are not the
only source of income, and sometimes not even the main source. Figure A1 shows
how different income components can be aggregated to produce total household
income. We see that employment-related income includes both employee income
(one component of which is wages) and income from self-employment. In add-ition,
total household income also comprises property income (such as interests,
dividends, and rents that arise from the ownership of financial or non-financial
assets), income of services for own consumption (such as imputed rents and unpaid
domestic services) and current transfers (which include government sponsored
social insurance schemes, employers sponsored benefits, social assistance benefits,
and transfers from other households or non-profit institutions such as unions). For
detailed definitions of each of these concepts, see box A1.
Figure A1 The components of household income
Household income
Income from
employment Property income
Source: ILO, 2003b.
Income from
the production
of household services
for own production
Employee income
Income from
self-employment
Current transfers
received

Box A1 Concepts and definitions
Household income consists of all receipts, whether monetary or in kind (goods and services),
that are received by the household or by individual members of the household at annual or more
frequent intervals, but excludes windfall gains and other such irregular and typically one-time
receipts. Household income receipts are available for current consumption and do not reduce
the net worth of the household through a reduction of its cash, the disposal of its other finan-cial
or non-financial assets or an increase in its liabilities. Household income may be defined
to cover: (i) income from employment (both paid and self-employment); (ii) property income;
(iii) income from the production of household services for own consumption; and (iv) current
transfers received.
Employee income comprises direct wages and salaries for time worked and work done, cash
bonuses and gratuities, commissions and tips, directors’ fees, profit-sharing bonuses and other
forms of profit-related pay, remuneration for time not worked, and free or subsidized goods and
services from an employer. It may include severance and termination pay as well as employers’
social insurance contributions. These items should be reported separately, when included. The
definition of these terms is consistent with their use in the resolution concerning the measure-ment
of employment-related income adopted by the 16th International Conference of Labour
Statisticians (ILO, 1998).
Employee income may be received in cash (monetary) or in kind as goods or services. Those
receipts in kind that are outputs of the employer’s production process should be included only
in so far as they are in line with the recommendations contained in the ILO’s Protection of
Wages Convention, 1949 (No. 95). Otherwise, they are imposed payments in kind that should be
excluded from employee income or valued at zero.
Wages are a component of employee income and for the purpose of this report refer to the stat-istical
concept of earnings (for a definition, see box 1 in Part I).
Income from self-employment is income received by individuals, over a given reference period, as
a result of their involvement in self-employment jobs as defined in the resolution concerning the
International Classification of Status in Employment adopted by the 15th International Conference
of Labour Statisticians (ILO, 1993b). In particular, income from self-employment concerns pri-marily
owners of unincorporated enterprises who work in these enterprises. It excludes profits
from capital investment of partners who do not work in these enterprises (“sleeping partners”),
dividends, and directors’ fees paid to owners of incorporated enterprises. Income from self-employment
includes the estimated value of goods and services produced for barter as well as
goods produced for own consumption, less expenses.
The basis for the measurement of income from self-employment is the concept of mixed income
defined by the System of National Accounts. Mixed income consists of the value of gross output
less operating costs and after adjustment for depreciation of assets used in production, where
these terms are as defined in the resolution concerning the measurement of employment-related
income adopted by the 16th International Conference of Labour Statisticians (ILO, 1998).
Source: ILO, 2003b.

Appendix II Definitions, concepts and data issues 77
Because income is measured at the household level, ranking households from
richest to poorest (in terms of household income) requires not only information
on total household income, but also information on the size of the household.
An annual household income of US$6,500 is not the same for a single-person
household as it is for a household with two adults and three children. In order to
rank households from richest to poorest, a per capita household income is calcu-lated.
Since there are economies of scale when several individuals live together,
and because children have lower material needs than adults (in terms of calories,
for example), estimating per capita income requires the use of equivalence scales.
The method used is explained in box A2.
The fact that household members’ economic well-being depends on per
capita, rather than total, household income implies that changes in household
income inequality depend not only on changes in the level of income per se from
different sources but also on demographic factors, such as changes in the popu-lation
age structure or family living arrangements. So, for example, the trend in
OECD countries towards smaller families – with a growing number of single-headed
households – tends to increase household income inequality and poverty
rates; similarly, the growing percentage of households with retired members cre-ates
a larger number of households with relatively smaller incomes.61 “Assortative
mating” (the tendency of high-wage earners to marry other high-wage earners)
also increases household income inequality, and seems to have contributed some-what
to widening inequality in OECD-type countries.62 However, these factors
generally operate over the medium to long term, and are in any event beyond the
scope of this report.
Other elements that contribute to households’ economic well-being, but are
beyond the scope of this analysis, include both household wealth (assets minus
liabilities) and the extent to which governments provide free or subsidized medical
services, education, or other goods and services that also affect people’s material
well-being. To take this latter element into account, household income would have
to be adjusted for the value of these social transfers in kind (STIK), but this is a dif-ficult
undertaking. Note also that this report decomposes total household income,
as opposed to disposable household income, because disposable household income
is not frequently available.63

Box A2 Measuring per capita household income
Household income can be measured as the sum of incomes from all different sources accruing
to a given household over a certain period of time. The link between total household income and
living standards depends on the number of persons who must live on this income: a total annual
income of, say, US$6,500 does not have the same meaning for a single-person household as it
does for a household with two adults and three children. To account for family size, and reflect
living standards, we could simply divide total household income by the number of household
members. In the hypothetical case illustrated in figure A2, we would divide US$6,500 by five,
and obtain a per capita household income of US$1,300.
Figure A2 Illustration of the components of household income
Employment-related income State
Wage
US$3,000
Self-employment
US$2,000
Child benefits
US$1,000
Dividends
US$500
Total household income US$6,500
Such a simple method, however, does not take into account the fact that there exist economies
of scale when people live together (e.g. only one dwelling instead of two), and that children need
fewer calories than adults. To take these further considerations into account, and to obtain an
adjusted measure of the “possibility of households to consume”, we follow Deaton and Zaidi’s
formula and calculate per capita household income as: E = (A + α K)θ where A represents the
number of adults, K is the number of dependent children, α represents the spending of a child
relative to an adult, and θ captures the economies of scale in a given household (Deaton and
Zaidi, 2002). We use the adjustment factors as specified in table A4, and illustrate the use of the
formula with our hypothetical example. We see that the effective per capita household income
would be set in a range from US$2,493 to US$2,728.
Table A4 Equivalent scale parameters guide
α θ In our example:
Advanced economies 0.75 0.6 PCHHI = 6 500 ÷ (2 + 0.75 × 3)0.6 = US$ 2 728
Middle-income or emerging economies 0.50 0.8 PCHHI = 6 500 ÷ (2 + 0.50 × 3)0.75 = US$ 2 540
Low-income or developing economies 0.30 0.9 PCHHI = 6 500 ÷ (2 + 0.30 × 3)0.90 = US$ 2 493

Data and country selection
The data set used in Part II of this report was constructed from micro-data from
developed economies and emerging and developing economies. Whenever pos-sible,
data sets used combined information on individual wage and total household
income. The Panel Study of Income Dynamics (PSID), a US longitudinal data set
that has collected data from households since 1968, is used for all US estimates,
while for European countries the EU Statistics on Income and Living Conditions
(EU-SILC) is employed. EU-SILC is supported by the European Commission and
has been progressively implemented since 2003. Both of these surveys provide stat-istics
on the income and living standards of workers and their families. Whereas
in the case of the EU-SILC we use data from the first available year for all coun-tries
(either 2003 or 2004), in the case of the PSID we use the available survey as
of 1997. In the case of emerging and developing countries, the analysis relies on
national data sources. Attempts were made to include emerging and developing
countries which are members of the G20 (given their relatively large size), as well
as a limited number of other countries, including poorer countries such as Peru,
the Philippines and Viet Nam. Coverage of Africa is limited to South Africa due
to difficulties in obtaining data sets for other countries with information on both
individual wages and household income.
The data cover a pre-crisis and a post-crisis period. In practice, the exact
time frame of the analysis varies by country, mostly due to constraints related to
data availability. While data are available for the United States for periods before
the year 2000 (in fact, we selected 1997 as the first period to be observed for the
PSID), comparable data from the EU-SILC survey are available only from 2003
or 2004 for most European countries in the sample, ending in 2010. The quality
of the data set also improved over time. In the analyses, 2006 is usually used as
the starting point for European economies and for the United States. In the case
of emerging and developing economies, where household surveys tend to be con-ducted
less frequently, the first data point usually (but not always) lies between
2000 and 2002; the second data point lies as close as possible to the beginning of
the global economic and financial crisis; the third data point is the latest available
year. Most data for this group of countries cover approximately a ten-year period
from 2000–02 to 2010–12.
The combined data set used for this report obviously has limitations. First,
there are differences in the definitions that are used. In particular, there are vari-ations
in how “wages” are defined across countries. Similarly, countries differ in
the way in which household income is defined, and in the extent to which infor-mation
is provided on the different sources of non-wage income – particularly on
employment-related income of the self-employed and on capital gains. Second,
surveys include measurement errors related to the accuracy of information col-lected
as well as errors in estimation that arise in the process of extrapolation
to the overall population. This is the case for the data from emerging and devel-oping
economies, but also for the EU-SILC.64 Tables A5 and A6 show the data
sources and years used for emerging and developing economies. The following text
describes in detail all data sources used.

Table A5 Data sources for emerging and developing economies
Region Country Data source
Asia India Employment–Unemployment Survey (EUS)
China (1) China Household Income Project (CHIP)
China (2) China Health and Nutrition Survey (CHNS)
Indonesia National Labour Force Survey (SAKERNAS) and National Socio-Economic Survey (SUSENAS)
Philippines Family Income and Expenditure Survey (FIES) and Labour Force Survey (LFS)
Viet Nam Household Living Standard Survey (HLSS)
Europe Turkey Turkish Household Labour Force Survey (HLFS) and the Household Budget Survey (HBS)
Russian Federation Russian Longitudinal Monitoring Survey (RLMS-HSE)
Africa South Africa Various
Latin
America
Brazil Pesquisa Nacional por Amostra de Domicilios (PNAD) and Pesquisa Mensal de Emprego (PME)
Argentina Encuesta Permanente de Hogares (EPH)
Peru Encuesta Nacional de Hogares sobre Condiciones de Vida y Pobreza (ENAHO)
Chile Encuesta de Caracterización Socioeconómica Nacional (CASEN)
Mexico Encuesta de Ingresos y Gastos de los Hogares (ENIGH) and Encuesta Nacional de Ocupación y Empleo (ENOE)
Uruguay Encuesta Continua de Hogares (ECH)
Table A6 Reference years for emerging and developing economies
Region Country Year 1 Year 2 Year 3
Asia India (EUS) 1999–2000 2004–05 2011–12
China (CHIP) 2002 2007 2009
China (CHNS) 2004 2006 2009
Indonesia (SAKERNAS) 2001 2005 2009
Indonesia (SUSENAS) 2001 2005 2010
Philippines (FIES) 2003 – 2009
Philippines (LFS) Jan. 2004 – Jan. 2010
Viet Nam (HLSS) 2002 2006 2010
Europe Russian Federation (RLMS-HSE) 2002 2006 2012
Turkey (HLFS) 2005 – 2010
Turkey (HBS) 2005 – 2007
EU-SILC 2003/04 All years 2010
Americas US (PSID) 1997 All years 2010
Argentina (EPH) 2003 2007 2012
Brazil (PNAD) 2001 2007 2012
Brazil (PME) 2001 2007 2012
Peru (ENAHO) 2004 2008 2012
Chile (CASEN) 2000 2006 2009
Mexico (ENIGH) 2000 2008 2012
Mexico (ENOE) 2000 2008 2012
Uruguay (ECH) 2004 2008 2012
Africa South Africa 2002 2007 2011
South Africa 2000 2005 2010

Sample selection criteria
All data sets have been cleaned of inconsistencies and outliers and have been
trimmed according to sample selection criteria applied equally to all data sets. The
selection criteria are summarized below, showing that we eliminate very young
dependent households and households whose members are elderly, in the sense
that they do not have links to labour market activities.
Selection at the level of the individual applies the following criteria:
y Anyone 15 years old or younger is excluded. However, if a child 15 or younger
has income, the income is included as part of the household income before they
are dropped from the sample. This ensures their income is not excluded from
total household income.
y Anyone classified as a full-time student or disabled is excluded. These indi-viduals
are usually classified as “out of the labour force” (OLF), but differ from
other people classified as “OLF” because they are not ready to become active
participants.
y Anyone 64 or older who is unemployed or out of the labour force is excluded
because they are not potential labour market participants.
These three categories of individuals are dropped in order to limit the sample to
individuals who are potential labour market participants. However, in so far as at
least one member of these individuals’ households is not excluded on the basis of
the above criteria, the household remains part of the sample and the individuals
are taken into account when estimating the equivalent scales which normalize
household income. For example, a household with two people over 64 is completely
excluded from the individual and household samples. By contrast, in a family with
two working adults and two children under 15, the children are dropped from the
individual sample, but a family of four is retained for the purpose of estimating
per capita values, for example, per capita household income.
Once the selection is completed at the individual level, sample selection cri-teria
at the household level are applied as follows:
y If all members of the household are at least 70 years old, the household is
excluded even if one or more household members claim to be employed.
y If all members of the household are at least 65 years old and all household
members are unemployed or out of the labour force, the household is excluded.
y If all members of the household are 18 years old or younger the household is
excluded even if one or more household members claim to be employed.
Europe (EU-SILC data)
For most countries in Europe, EU-SILC data are used. Data are available for
29 countries from either 2003 or 2004 to 2010, and available only for the last
year (2010) for Croatia and Switzerland. Table A7 shows the distribution of the
sample of individuals aged 16 and above by country and for each of the avail-able
years. Table A8 shows the same distribution once we have weighted the data

Table A7 Distribution of the complete sample, by country and year
2003 2004 2005 2006 2007 2008 2009 2010
Austria 7 944 8 969 10 044 11 120 9 037 9 007 9 357 9 258
Belgium 8 404 8 198 9 504 10 305 10 031 9 769 9 676 9 368
Bulgaria 8 714 8 568 10 757 11 805 12 349
Cyprus 7 911 7 681 7 390 7 023 6 440 7 805 8 164
Czech
Republic
7 140 12 138 15 669 18 240 15 875 14 638 14 022
Denmark 11 840 10 350 9 833 9 939 9 809 9 973 9 865 9 021
Estonia 7 871 8 472 11 436 10 552 9 477 9 796 9 604 9 473
Finland 20 264 20 436 19 513 19 221 18 499 17 376 18 659 15 827
France 15 851 15 464 16 080 16 671 16 407 16 395 16 975 17 246
Germany 20 336 20 585 20 337 18 475 18 132 17 985 18 366
Greece 11 578 10 085 10 260 9 976 11 348 12 017 11 652 9 711
Hungary 0 12 322 13 623 15 453 15 552 17 692 17 510 20 915
Iceland 6 124 6 125 5 993 6 033 6 061 5 956 6 198 6 332
Ireland 9 035 9 644 9 006 8 477 7 839 7 846 7 064
Italy 43 180 39 149 37 914 36 604 36 085 35 070 32 805 32 744
Latvia 6 697 7 627 7 712 9 126 10 144 10 698 11 004
Lithuania 0 8 574 8 502 8 996 8 523 9 086 9 508 8 860
Luxembourg 6 828 6 809 7 039 7 170 6 903 7 645 9 032 10 099
Malta 7 153 7 250 7 827
Netherlands 15 294 15 059 17 078 16 798 15 475 16 200 16 595
Norway 10 907 10 729 10 062 10 356 9 613 9 343 9 002 7 908
Poland 33 376 32 537 30 940 29 716 27 639 26 813 26 280
Portugal 9 936 9 000 8 503 8 230 8 218 8 789 8 783 9 606
Romania 14 040 13 368 12 910 12 633 12 352
Slovakia 11 504 11 118 11 171 12 513 12 361 12 567 11 911
Slovenia 21 976 24 853 22 596 22 614 22 837 22 597 21 916
Spain 26 024 25 980 23 963 24 429 25 490 26 178 26 143 24 274
Sweden 9 479 10 235 11 430 12 135 12 627 12 479 11 881 11 052
Switzerland 11 801
United Kingdom 16 425 15 000 13 968 13 400 12 239 11 648 11 663
Total 205 265 361 200 369 303 395 282 391 360 396 379 396 353 395 944

Table A8 The representativeness of the sample (age 16+ in the population)
2003 2004 2005 2006 2007 2008 2009 2010
Austria 5 597 232 5 649 314 5 670 950 5 687 390 5 726 730 5 716 189 5 723 961 5 753 396
Belgium 6 751 946 6 670 720 6 942 917 6 993 240 6 994 613 7 077 395 7 121 209 7 176 987
Bulgaria 5 684 380 5 667 859 5 609 350 5 573 473 5 504 147
Cyprus 523 401 545 389 557 187 565 793 574 370 590 138 611 008
Czech
Republic
7 322 981 7 365 339 7 372 667 7 379 933 7 428 375 7 499 729 7 482 708
Denmark 3 487 881 3 493 736 3 493 793 3 510 353 3 522 106 3 539 082 3 563 854 3 572 783
Estonia 942 109 951 610 957 181 955 684 952 350 952 029 942 034 937 957
Finland 3 429 718 3 445 352 3 438 060 3 438 522 3 469 620 3 465 694 3 486 422 3 490 489
France 37 851 332 38 310 760 38 668 663 38 943 543 39 397 381 39 407 615 39 710 294 39 567 631
Germany 54 518 208 54 283 658 54 206 957 54 282 903 54 003 247 53 752 484 53 826 717
Greece 7 493 085 7 544 536 7 718 533 7 790 411 7 808 242 7 772 145 7 800 858 7 719 678
Hungary 6 986 437 6 920 024 6 967 616 7 054 029 7 083 447 7 062 469 7 028 618
Iceland 188 631 192 043 197 221 202 988 208 924 211 886 205 694 204 745
Ireland 2 795 589 2 841 963 2 922 451 3 013 154 3 058 078 3 051 117 2 996 762
Italy 40 241 144 40 615 317 40 669 399 40 823 417 41 029 380 41 385 289 41 394 924 41 557 786
Latvia 1 577 329 1 614 839 1 604 107 1 625 830 1 657 945 1 636 205 1 470 164
Lithuania 2 351 874 2 351 535 2 392 374 2 404 195 2 396 573 2 368 299 2 282 816
Luxembourg 305 234 306 183 305 085 308 921 315 649 327 431 339 807 346 174
Malta 291 424 297 186 297 763
Netherlands 10 772 921 10 757 195 10 823 451 10 860 751 10 895 563 10 932 940 10 879 193
Norway 3 041 891 3 015 522 2 789 622 3 092 780 3 133 467 3 186 122 3 221 840 3 244 176
Poland 27 287 345 28 419 961 28 730 734 28 832 966 26 335 436 26 288 537 27 788 140
Portugal 7 496 615 7 538 436 7 566 542 7 563 214 7 579 329 7 476 115 7 468 328 7 427 838
Romania 15 824 847 15 525 148 15 299 138 15 643 709 15 620 096
Slovakia 3 934 644 3 889 183 3 996 472 4 005 149 4 084 888 4 126 486 4 115 668
Slovenia 1 438 142 1 479 054 1 431 650 1 420 372 1 443 380 1 432 298 1 422 322
Spain 30 224 324 31 393 743 31 875 154 32 453 037 33 099 238 33 495 288 33 499 766 33 232 906
Sweden 5 738 184 5 859 729 5 792 423 5 821 601 5 939 728 5 975 676 6 024 878 6 090 490
Switzerland 5 845 888
United Kingdom 39 153 157 40 093 360 40 186 920 40 707 465 40 971 270 41 132 755 41 345 583
Total 155 584 915 313 695 403 316 727 531 340 377 617 34 256 7228 34 111 3479 341 837 339 345 843 867

using cross-section weights that render the sample representative of the underlying
population. Both tables show the numbers of individuals (and their representation
in the population) after the sample was cleaned of inconsistencies and outliers, and
once the selected sample selection criteria mentioned above were applied.
EUROSTAT provides annual income measures and asks individuals to
declare the number of months they worked full time and part time during the
reference income period (RIP). The RIP is always the last calendar year. Using
Atkinson’s 65 conversion of part-time to full-time months, the number of full-time
equivalent months worked (FTMW) per year is estimated for each individual who
claims to have worked at least one month during the RIP. The FTMW equals the
sum of total full-time months and part-time months, where part-time months are
normalized by the relative weight of part-time hours to full-time hours. Dividing
total gross annual earnings by FTMW provides an estimate of gross monthly wage
for employees or monthly earnings in the case of the self-employed.
Europe (non-EU-SILC data)
For the Russian Federation, the analysis was based on the Russian Longitudinal
Monitoring Survey (RLMS-HSE).66 The RLMS-HSE is conducted by the Higher
School of Economics and ZAO “Demoscope” in cooperation with the Carolina
Population Center, University of North Carolina at Chapel Hill in the United States,
and the Russian Academy of Sciences’ Institute of Sociology. The RLMS-HSE is
nationally representative and is used because none of the regular official surveys
contains information on wages and household income. The Russian Labour Force
Survey provides information on employment but does not include questions on
wages. The Russian Household Budget Survey does not collect information on
household income: it focuses exclusively on household expenditure and does not
allow for different sources of income to be distinguished from each other. Data are
from 2002, 2006 and 2012. Gorodnichenko, Sabirianova and Stolyarov conclude
that the “RLMS appears to be a reliable data source for examining the inequality
trends in labor market outcomes, reported income, [and] consumption, with the
common caveats of income underreporting and underrepresentation of the super-rich”
(Gorodnichenko, Sabirianova and Stolyarov, 2010). The World Bank also
favours the RLMS over official data sources in a number of publications on in-equality
and poverty (e.g. World Bank, 1999).
For Turkey, two sources of data are used. The first is the Turkish Household
Labour Force Survey (HLFS), which is conducted quarterly. The years 2005 and
2010 are used for the analyses. The HLFS has data on wages alone. The second
source of data is the Household Budget Survey (HBS) for 2005 and 2007. These
data were also used by Tansel and Bircan and offer a much richer set of variables
on income than the HLFS (Tansel and Bircan, 2010).

Asian countries
Data for China are from two separate sources. The first source is the China
Household Income Project (CHIP) for 2002, 2007 and 2009. The surveys are
nationally representative and samples were randomly drawn from the larger
annual national household income survey conducted by the National Bureau of
Statistics (NBS). The purpose of these surveys is to estimate wages, employment,
consumption and related economic issues in both rural and urban areas of China.
Data on migrants are included in this survey. The second source is the China
Health and Nutrition Survey (CHNS), a longitudinal survey jointly conducted
by the Carolina Population Center at the University of North Carolina at Chapel
Hill in the United States, and the National Institute of Nutrition and Food Safety
at the Chinese Center for Disease Control and Prevention in China. The survey
covers nine provinces of varying levels of development (Helongjiang, Liaoning,
Shangdong, Henan, Hubei, Hunan, Jiangsu, Guangxi, Guizhou provinces) for
2004, 2006 and 2009.
The analysis for India is based on the Employment–Unemployment Survey
(EUS) carried out by the National Sample Survey Office (NSSO) of India. It
covers all major Indian states. The years (known as rounds) considered for the
analyses are the 55th (July 1999 to June 2000), 61st (July 2004 to June 2005) and
68th (July 2011 to June 2012). The 55th and 68th rounds do not contain informa-tion
on income from self-employment. In the 61st round there is no direct ques-tion
related to self-employment income, but two questions related to remuneration
were introduced: (1) whether respondents regarded their current income from self-employment
as remunerative (as opposed to being produced for self-consumption),
based on different income brackets; and (2) what amount per month they would
consider remunerative (as opposed to being produced for self-consumption). From
these questions, it would have been possible to compute approximate averages
for each income bracket, but this procedure was considered inappropriate for the
analyses undertaken for this study. Therefore, income from self-employment and
total employment-related income are not computed.
The statistics for Indonesia are based on two different sources. The main
source is the national labour force survey (Survei angkatan kerja nasional,
SAKERNAS). This provides the basis for calculation of all statistics related to
employment, wages, income from self-employment and household employment-related
income. The years 2001, 2005 and 2009 are used for the analyses. Even
though more recent SAKERNAS data are available, these could not be consid-ered
since the questionnaire has undergone changes and the self-employed are no
longer questioned about their income. SAKERNAS does not contain income from
sources other than employment and consequently household income cannot be
calculated. Therefore, the national socio-economic survey (Survei social ekonomi
nasional, SUSENAS), a household survey, was taken into consideration as a
second source of data. It contains information on household income based on a
consumption approach and is available for 2001, 2005 and 2010.
Data for the Philippines are taken from two data sets, the Family Income
and Expenditure Survey (FIES) and the Labour Force Survey (LFS). The years
2003 and 2009 are used from the FIES, and for the LFS the second (July 2003),

third (October 2003), and fourth (January 2004) quarters of 2003; the second (July
2009), third (October 2009), and fourth (January 2010) quarters of 2009; and the
first (April 2011), second (July 2011), and third (October 2011) quarters of 2011 are
employed. The LFS contains data on wages but not income from self-employment.
Income from self-employment is covered in the FIES. Thus, wages are taken from
the LFS, whereas income from self-employment is taken from the FIES. Total
employment-related income is computed using both surveys. Statistics on house-hold
income are taken from the FIES, which collects income data at the household
level. Given that FIES data were unavailable for 2011, the 2011 analysis covers
wages, but excludes income from self-employment and household income.
Data on Viet Nam are from the Household Living Standard Surveys (HLSS)
for 2002, 2006 and 2010. The Labour and Employment Survey (LES) is not used
because in recent years (i.e. 2011 and 2012) the self-employed have not been asked
about their income from self-employment.
Americas
For Argentina, we use the Encuesta Permanente de Hogares (EPH). The survey
includes demographic and socio-economic characteristics of the population and
is linked to the labour force. It is implemented by INDEC (Instituto Nacional de
Estadísticas y Censos) and is used in our analyses for 2003, 2007 and 2012. Micro-data
are available for 31 urban areas (aglomerados urbanos).
For Brazil, micro-data are used from two surveys: the Pesquisa Nacional por
Amostra de Domicilios (PNAD) and the Pesquisa Mensal de Emprego (PME).
Both surveys are conducted by the Instituto Brasileiro de Geografia e Estadística
(IBGE). Given that the PME collects information only on wages and on income
from self-employment, other forms of income were imputed using the PNAD in
order to estimate total household income (following the methodology by Machado
and Perez Ribas, 2010). Data are used for 2001, 2007 and 2012.
For Chile, data are from the Encuesta de Caracterización Socioeconómica
Nacional (CASEN), which is carried out every two or three years. Data are used
for 2000, 2006 and 2011.
For Mexico, the Encuesta de Ingresos y Gastos de los Hogares (ENIGH) is
used; this collects information on the distribution, levels and structure of house-hold
incomes and expenditures, together with information on the economic activ-ities
of these households. We use data for 2002, 2008 and 2012.
For Peru, the Encuesta Nacional de Hogares sobre Condiciones de Vida
y Pobreza (ENAHO) is used. It has been conducted since 1995 by the Instituto
Nacional de Estadística e Informática (INEI) and is national. Data are used for
2004, 2008, and 2012.
For the United States, the Panel Study of Income dynamics (PSID) is used.
This is a longitudinal dataset, compiled annually from 1968 to 1996 and biannu-ally
from 1997 onwards. In relation to previous years, the 1997 sampling strategy
changed significantly in order to reduce the cost of collecting the data. Thus, 1997
was the year that was both as far back as possible in time and also the best in
terms of attaining a sample comparable to years ahead in time, e.g. the post-crisis

periods close to 2010. Despite becoming biannual, the PSID has kept the main
characteristics of a longitudinal study in that it keeps on tracking families previ-ously
surveyed, new families that branch out from previously surveyed families
(e.g. children who later form families of their own, divorced parents who create
new households, etc.). However, our interest is not in tracking the same families
but in interpreting income distributions over periods. This is why we have chosen
to use the cross-sectional data of the PISD, which is also available by selecting
the appropriate sample weights. Each household in the sample is represented by
the head and (where applicable) the spouse. All other household members are
accounted for (e.g. their wages, social security contributions, capital gains etc.);
although we cannot identify their individual characteristics, this is not a drawback
and has no adverse effect for our analytical purposes. The core sample includes
socio-economic variables collected continuously over time with extensive and
complete information on labour market outcomes and required household char-acteristics,
among others. The PSID is directed by the University of Michigan,
and the data are used by researchers, policy analysts and teachers around the
globe. According to the PSID website, the National Science Foundation recog-nized
the PSID as one of the 60 most significant advances funded by the NSF in
its 60-year history.
For Uruguay, the Encuesta Continua de Hogares, implemented by the
Instituto Nacional de Estadística (INE), is used for 2004, 2008 and 2012. However,
while the data for 2008 and 2012 cover both urban and rural areas, those for 2004
cover only urban areas.
Africa
For South Africa, different data sets are used for the labour market and household
income indicators since there is no single nationally representative survey for the
appropriate period which includes enough detailed information on all variables.
The labour force survey is used for 2002 (Q2), 2007 (Q2) and 2012 (Q4), and the
income and expenditure survey for 2000, 2006 and 2010/11.

Appendix III
Changes in household income inequality
Inequality estimates using counterfactuals
To isolate the impact of changing wage inequality on household income inequality,
the analysis relies on estimating a counterfactual outcome where the latter holds
inequality constant in the first period (e.g. in the pre-crisis period). Counterfactuals
simulate the unknown and aim to answer the question “What would have hap-pened
if …?”. For example, in this exercise, one question the simulation aims to
answer is what would have happened to household income inequality between two
periods of time (e.g. before and after the crisis) if wage inequality had remained
constant between the two periods. The method of counterfactuals isolates wage
inequality while identifying how changes in other sources of income (social trans-fers,
income from self-employment, etc.) contribute to the observed (as opposed
to simulated) change in household income inequality between two periods. To see
this more clearly, the difference in the D9/D1 between 2010 (T2) and 2006 (T1) can
be interpreted with the following expression:
T1,T 2D91= T1,T 2D91(wages) + T1,T 2D91(other incomes) (1)
Expression (1) shows that different sources of income can contribute to changes
of the D9/D1 ratio between two periods. Let ΔT1,T2D91C,WG be the change in the
D9/D1 ratio between the actual outcome and the counterfactual – based on per
capita household income.67 In other words, ΔT1,T2D91C,WG is the ratio that would
have resulted if wage inequality in 2010 (T2) was measured allowing for wage in-equality
to be the same in 2006 (T1):
(2)
T1,T 2D91= T1,T 2D91(wages) + T1,T 2D91(others incomes)
T1,T 2D91C,WG = T1,T 2D91(wages)C + T1,T 2D91(others incomes)
( T1,T 2D91 T1,T 2D91C,WG )= T1,T 2D91(wages) T1,T 2D91(wages)C
Expression (2) shows that by construction, the effect of wage inequality has been
isolated from household income inequality between two periods. In practice and
by construction the distance ΔT1,T2D91(wages)C should equal zero (because changes
in wage inequality are held constant over time) such that the final expression in (2)
uses (ΔT1,T2D91 – ΔT1,T2D91C,WG) to evaluate changes in household income inequality
which result from changing wage inequality.68 The same exercise can be applied
to any of the other income components identified at individual level to isolate
their effect on household income inequality, for example, to isolate the effect of
changing dispersion in unemployment benefits or self-employment income.

Decomposing wage gaps
To calculate the wage gaps for different pairs of population subgroups (e.g. women
and men, migrants and nationals, informal economy and formal economy workers),
the following methodology was employed.
Calculating the explained and unexplained parts of unadjusted wage gaps
First, the 1st, 2nd, 3rd, … , 8th and 9th deciles of the wage distribution are
calculated using wage employees only and for each subgroup separately (males,
females, migrants, nationals, formal economy workers, informal economy
workers). In each case, the wage distribution is based on a full-time equivalent
measure of monthly wages for countries where the data is provided annually
(European Union economies and the United States) and declared monthly wages
for all other data sets (emerging and developing economies). In all cases, our
analysis applies only to the last available year. The last year is always 2010 for
developed economies and 2011 or 2012 for emerging and developing economies.
The nine estimated deciles are pairwise compared between each set of mutu-ally
exclusive subgroups of employees. In our comparison, and for each paired
subgroup, we define one of the subgroups to be the “disadvantaged” group in
relation to their returns in the labour market. Among the subgroups consid-ered,
females, migrants and informal economy workers are considered disadvan-taged
whereas males, nationals and formal economy workers are considered,
on average, to fare better in the labour market when compared to their corres-ponding
disadvantaged groups.
Let the disadvantaged subgroup be G0 and the advantaged subgroup be G1.
The decile-specific “raw” wage gap is the wage decile difference between G0 and
G1. Thus, if we define a decile of the wage distribution as Vg(α), g = 1, 0 such
that α = 1,2,3,…,8,9 then v0,1(α) = v1(α) – v0(α) is the raw difference of the α-decile
between pairs of mutually exclusive subgroups.
The explained and unexplained parts of the gaps are then calculated. In gen-eral
terms, the “explained part” takes into account differences between the two
groups being compared based on labour market characteristics observed in the
data, namely, experience, education (four categories), occupational category (man-agerial,
highly skilled, semi-skilled, low-skilled and unskilled), economic activity
(from manufacturing to services, about ten categories including public administra-tion),
location (urban, rural) and work intensity (hours worked). When experience
is missing, age is considered as a proxy.69 The unexplained part of the decomposi-tion
then captures wage difference that cannot be explained by the previous list
of variables.
Appendix IV

How are the explained and the unexplained parts technically identified? Let
vc(α) be the “counterfactual” decile to the subgroup G0 (i.e. the α-decile from a
hypothetical distribution of wages that would be observed for the disadvantaged
subgroup G0 if they were to receive the same returns to their observed labour
market characteristics as those received by their counterfactuals in the advan-taged
subgroup G1). To identify the counterfactual distribution for subgroup G0,
we employ propensity score matching. In other words, a conditional probability
model is employed to compare each member of disadvantaged subgroup G0 to
all members in the advantaged subgroup G1 in terms of their observable char-acteristics.
Then each individual in the G0 subgroup is assigned the wages of the
individual in the advantaged subgroup G1 that is closest to him or her in terms
of propensity-score weighted labour market characteristics. The distribution of
assigned wages becomes the counterfactual wage distribution for the sub-group G0,
i.e., the disadvantaged one.70
For each decile the following difference is estimated:
The distance {v1(α) – vc(α)} is the decile-specific difference between the wages
of the advantaged group and the wages that the disadvantaged group would have
obtained if they had had the same returns to all of their observable character-istics.
In other words, this is the wages of the advantaged subgroup minus the
counterfactual wages of the disadvantaged subgroup and identifies the explained
part of the wage gap. To calculate the unexplained wage gap, the following cal-culation
is made: {vc(α) – v0(α)}. This equation is the counterfactual wages of
the disadvantaged subgroup minus the actual wages of the subgroup. Together,
{v1 (α) – vc(α)} + {vc(α) – v0(α)}, i.e. the explained and the unexplained parts, add up
to the raw (overall) decile-specific wage gap. The explained and/or unexplained
parts can be positive or negative. When the explained part is negative it indicates
that at that specific decile, and in accordance with the labour market character-istics
X, the wages of the disadvantaged subgroup should be higher relative to the
labour market characteristics of the advantaged subgroup in the same decile, i.e.
the disadvantaged subgroup is overqualified with regard to X in that decile. When
the unexplained part is negative, it indicates positive discrimination towards the
disadvantaged subgroup with regard to the set X of covariates. The sum of the
explained and unexplained parts yields the total unadjusted pay gap.

Notes
1 For detailed description of macroeconomic effects, see Lavoie and Stockhammer,
2013.
2 The IMF considered that in Germany, in light of the strong labour market, “it
would not be inappropriate for real wages to rise, and therefore help improve
the labor share of national income. This would help spur domestic demand, and
make the economy less vulnerable to external shocks, while not endangering
Germany’s competitiveness” (IMF, 2013a, p. 24). The IMF similarly called for
increases in basic wages in Japan, “which would be important to stimulate infla-tion
expectations and support purchasing power once inflation starts to rise”
(IMF, 2013b, p. 12). Underlying this call is a view that the dwindling labour share
of income has forced Japanese workers to dip into their savings to finance con-sumption
growth, and that higher wages are required for a transition towards
sustainable growth led by the private sector (Botman and Jakab, 2014).
3 In the light of the cyclical and structural factors affecting the labour market,
the US Federal Reserve Bank is also looking towards changes in compensa-tion
as part of its overall evaluation of labour market slack (and subsequently
monetary policy). See Yellen, 2014.
4 “Indeed, real product wages – a measure of wages relative to the prices that
companies charge for their output – fell broadly in line with output per worker.
That could reflect companies responding to faltering output per worker by
bearing down on pay. But the observed combination of weak wages and
lower output per worker could also reflect people becoming more willing to
accept pay restraint, encouraging companies to employ more people” (Bank
of England, 2013, p. 34).
5 “Forecasters significantly underestimated the increase in unemployment and
the decline in private consumption and investment associated with fiscal con-solidation”
(Blanchard and Leigh, 2013, p. 5).
6 See e.g. ILO, 2014e.
7 The G20 includes: Argentina, Australia, Brazil, Canada, China, France,
Germany, India, Indonesia, Italy, Japan, the Republic of Korea, Mexico, the
Russian Federation, Saudi Arabia, South Africa, Turkey, the United Kingdom,
the United States and the European Union.
8 This is calculated as the sum of GDP of the 19 individual countries that are
members of the G20 (the 20th member being the European Union) as a share
of the world’s GDP, based on purchasing power parity, as estimated in IMF,
2013c.

9 According to recent ILO estimates, in 2014 there are 3.2 billion employed per-sons
in the world, of which 1.6 billion (50.3%) are waged and salaried workers
(ILO, 2013a).
10 Data are available from the ILO Global Wage Database (see box 2). The average
wage has been adopted as an ILO “decent work indicator” (ILO, 2012b).
11 These elements can also be captured by other measures of productivity, such
as output per hour worked (another definition of labour productivity), capital
productivity and multifactor productivity. Capital productivity is another par-tial
measure of productivity which reflects capital’s contribution to changes
in production. Multifactor productivity “helps disentangle the direct growth
contributions of labour, capital, intermediate inputs and technology. This is an
important tool for reviewing past growth patterns and for assessing the poten-tial
for future economic growth” (OECD, 2001, p. 20). However, some research
questions the accuracy and relevance of using multifactor productivity (Felipe,
2008; Felipe and McCombie, 2013).
12 Moreover, labour productivity “forms a direct link to a widely used measure of
living standards, income per capita”, and is “important as a reference statistic
in wage bargaining” (OECD, 2001, p. 15). It is also a decent work indicator.
13 Considering the full sample of 33 countries used in figure 8, trends in labour
compensation and wage growth vis-à-vis productivity growth changed (based
on choice of “wage” variable and deflator) for slightly more than one-third of
countries. However, as shown in the figure, the measure used does not affect
the final results. Using wages and the CPI, about half the countries experienced
wage growth that exceeded productivity growth over the period. When com-pensation
and the GDP deflator are used, the findings are similar (See Sobeck,
forthcoming).
14 See OECD, 2012a, and the shift-share analysis in ILO, 2010a. In the finan-cial
sector, rising profits were accompanied by higher bonuses and wages, but
not for everyone. Bell and Van Reenen showed that the premium attached to
being employed in the UK financial sector increased only for the top 10 per
cent of earners in the sector (Bell and Van Reenen, 2013). Looking at over 400
UK listed companies from 1999 to 2010, the authors also showed that higher
shareholder returns are strongly associated with the level of pay of CEOs and
senior executives, an order of magnitude smaller for managers, and practically
non-existent for other workers. According to the authors, the median pay of
a FTSE-100 CEO in the United Kingdom is now about 116 times that of the
median worker, compared to a ratio of 11 in 1980. In the United States, the
median pay of an S&P 500 CEO has risen to 240 times that of the median
worker in 2008 from 26 times in 1970.
15 For a more detailed description, see ILO, 2010a, Technical Appendix II.
16 For example, in South Africa the only source of data on the average wage is the
Quarterly Employment Statistics, whose most recent available data relates to
March 2006 and covers only employees in the formal, non-agricultural business

Notes 95
sector. In China, the wage series refers to so-called urban units, a category that
comprises mostly state-owned and collective-owned enterprises, and excludes
part of the private sector; a new series for wages in private enterprises was
started in 2008, but no series currently combines both the public and the pri-vate
sectors. In India, representative data are available only through special
computations using the National Sample Survey Office (NSSO) employment/
unemployment data. Different official sources of wage data exist, including the
Annual Survey of Industries (ASI), the Employment–Unemployment Survey
and the Wage Rate Index, which is computed from the occupational wage
survey with panel data from about 1,256 units covering 14 manufacturing
industries, four mining industries and three plantations. This source was used
in the ILO Global Wage Report 2012/13 after consultation with the Ministry
of Statistics and Programme Implementation; this edition of the report has
switched to the NSSO due to its broader coverage.
17 PALMS, a collaboration between the University of Cape Town and the
University of Michigan, has painstakingly assembled and harmonized the
entire sequence of existing labour market surveys carried out by Statistics
South Africa since 1994, including the October Household Surveys (1994–99),
the biannual Labour Force Surveys (2000–07) and the Quarterly Labour Force
Surveys (since 2008) (see Wittenberg, 2014).
18 A recently adopted report on productivity measurement produced by a
task team from the National Economic Development and Labour Council
(NEDLAC) in South Africa put forth measures to improve labour productivity
measurement.
19 PPP$1 is equivalent to the purchasing power of US$1 in the United States.
For example, if you were to travel to the United States from China and walk
into a bank to exchange 3,000 yuan for US dollars, the bank would give you
about US$490. However, in terms of purchasing power in China, 3,000 yuan is
equivalent to about US$1,140, because it is less expensive to live in China than
in the United States.
20 The OECD has published a number of reports on the subject: see e.g. OECD,
2008; OECD, 2011; OECD, 2014a. Taking a perspective from developing coun-tries,
at least two of the World Bank’s last ten World Development Reports cen-tred
on issues related to inequality: see World Bank, 2006; World Bank, 2012.
The IMF also recently studied the link between income inequality, economic
growth and fiscal policy (see e.g. Berg and Ostry, 2011; IMF, 2014b). The ILO
itself has issued various publications on inequality: see e.g. ILO and IILS,
2008; ILO, 2014b. These institutional reports have been published alongside
a growing number of academic articles and books on inequality, including
(to name just a few) Piketty, 2013; Milanovic, 2011; and Galbraith, 2012. The
growing academic research has also been summarized in large volumes such
as Salverda, Nolan and Smeeding, 2009a.

21 This is also true within countries. Chetty, Hendren, Kline and Saez show for
example that in the United States, areas of high social mobility are also places
which have, among several things, less income inequality: see Chetty, Hendren,
Kline and Saez, 2014.
22 See, for example, the literature reviews in Neves and Silva, 2014. In principle,
inequality may accelerate growth, for instance by providing incentive for in-novation
and entrepreneurship (Lazear and Rosen, 1981) or by allowing some
individuals to accumulate enough capital to start a business and invest in edu-cation
(Barro, 2000). At the same time, inequality may have a negative impact
on growth by negatively affecting human capital investment of underprivileged
individuals (Perotti, 1996) or by generating economic and political instability
that reduces investment (Alesina and Perotti, 1996; Rodrik, 1999).
23 See in particular page 155.
24 However, fiscal redistribution is also limited in as much as the labour market
is dominated by the informal sector. For example, in the case of Turkey, the
fiscal reforms in 2007 and 2008 reduced wage inequality, as they targeted low
income earners by reducing the tax wedge and social security costs among low
income earners (Gönenç and Rawdanowicz, 2010). However, despite the effect
such reforms had among workers in the formal sector, the size of the informal
sector in Turkey remains significant at about 40 per cent of the working age
population (Tansel and Kan, 2012).
25 According to the Canberra Group, an expert group mandated to provide
conceptual and definitional guidance to national statistical offices, “total
income” and “disposable income” are the main income aggregates produced
(UNECE, 2011, p. 17). The components of total household income are set out
in figure A1 (see Appendix II); disposable income is obtained by deducting
from total income all current transfers paid, such as direct taxes and social
insurance contributions. Disposable income is usually the preferred measure to
analyse income distribution, but it is less frequently available than total income,
including in our own data set. Hence this report studies trends in total house-hold
income rather than disposable household income.
26 For practical purposes, this report has selected as thresholds the 30th and 70th
percentiles of the household income distribution to identify households in the
middle class. Whereas the proportion of households between these thresholds
remains constant at 40 per cent, the incomes that define the upper and lower
limits can change to become either wider apart (thereby stretching the income
values associated with the middle class) or closer together (thereby compressing
the income range identified with middle-class households). Another interesting
measure of inequality includes the top 10 per cent versus the bottom 40 per cent,
which was found to be quite robust and highly correlated with another possible
measure of inequality: the Gini coefficient (Cobham and Sumner, 2013).
27 Since as early as the 1980s, most developed economies have experienced
increasing household income inequality: the gap between the top and bottom

Notes 97
10 per cent widened in countries where inequality was already high, such as
the United Kingdom and the United States. This was also the case in coun-tries
with low inequality, such as Germany and the Nordic countries. Thus,
between 1985 and 2007/08 the Gini coefficient increased from 34 to 44 in the
United States and from 32 to 35 in the United Kingdom; likewise, in the same
period the Gini coefficient increased from 20 to 26 in Sweden, from 23 to 25 in
Norway and from 21 to 26 in Finland. In the case of emerging and developing
economies, there has been no universal trend: estimates from 40 emerging
and developing economies show that in 13 countries inequality has increased,
in six countries inequality has not changed significantly, and the remaining
21 countries have experienced a significant decline in inequality (Ferreira and
Ravallion, 2009). Some examples of emerging and developing economies where
inequality has decreased substantially from the mid-1990s to the mid-2000s are
(as measured by the Gini coefficient) Chile (3.3 per cent), Mexico (4.1 per cent)
and Turkey (8.1 per cent) (OECD, 2011).
28 Note that in Greece inequality has increased in 2011 and 2012, according
to Eurostat, http://appsso.eurostat.ec.europa.eu/nui/show.do?dataset=ilc_
di12&lang=en
29 Note that for China, different data sources indicate different trends particularly
in the years between the early 2000s and 2006/07, where data from the China
Health and Nutrition Survey (CHNS) show a decline in inequality while data
from the China Household Income Project (CHIP) point towards increasing
inequality. Both data sources suggest that inequality declined in most recent
years.
30 Although estimates from Turkey could not be directly integrated in the data
due to significant difference in the structure of the data, our estimates for
Turkey also suggest a decrease in wage inequality among formal sector wage
employees when comparing both D9/D1 and D5/D1. The decrease, however,
had only a mild effect on household income inequality.
31 For the technical definition of household income, see Appendix II.
32 In the estimation of the “counterfactual”, we hold wages constant and allow for
the dispersion (inequality) of other incomes and total income to vary between
periods. It is important to point out that in this case we are assuming exogen-eity
between changes in the dispersion of wages and the dispersion of all other
income components. Clearly, if other incomes (in particular, social transfers)
change endogenously with wages at each decile of the distribution, our counter-factual
would not be a pure way of keeping the distribution of wages constant:
the movement of other components would have some of the effect of movement
in wages (Belser and Vazquez-Alvarez, forthcoming).
33 These other sources include self-employment income, social transfers, private
intra-household transfers and capital gains. For detailed definitions of these
other sources of household income, see notes to figure 29.

34 A decline in the wage effect does not necessarily imply a decline in wage in-equality.
However, ILO analyses conducted for this report show that for these
countries it is the case.
35 Given that our sample excludes households where no member is of working
age, the higher share of pensions in total household income in Italy and Greece
may reflect the fact that in these countries more pensioners live with other
family members who are of working age.
36 It must be kept in mind that households with very high incomes tend to be
under-represented in household surveys, and the results presented may there-fore
be more representative of the incomes of the bottom 99 per cent of the
population, as pointed out in Brandolini and Smeeding, 2009, p. 77. Since
high-income respondents may be underestimated in survey samples or may
under-report capital gains, this is likely to exacerbate underestimation of the
proportion represented by such gains in household income (see for example
the case of South Africa, documented by Wittenberg, 2014). Note also that the
reported capital gains in France, according to calculations from the SILC data
set, jump from a small share – similar to that in other countries – until 2006 to
around 10 per cent or more in 2007 and subsequent years. It is unclear if this
is a real change or a data-related issue.
37 Note that in Germany, the Hartz reforms appear to have lessened the dualism
in social policies through the merger of means-tested unemployment as-sistance
programmes with the social assistance programme, which gave social
assistance beneficiaries who were out of the labour market greater access to
training programmes and job placement services, according to Clasen and
Goerne, 2011.
38 In the case of China, figure 32 shows no information on social transfers. This
is because in the case of China, the data set does not identify this category
alone at the level of the household. Thus, all items that refer to social trans-fers
are included in the category “other incomes”. In practice, during the past
12 years, China has significantly extended various elements of their social
protection system (including child and family benefits, maternity benefits,
household income support, old age pensions and health care coverage). The
significant increase has been reflected in a 3 per cent increase in public social
protection expenditure (as a percentage of GDP) since the year 2000, with
a view to moving to a more inclusive and mores sustainable growth pattern
(ILO, 2014f).
39 In this context, the “unexplained” should be understood as not accounted for by
the observed labour market characteristics listed in the paragraph. Similarly,
the “explained” should be understood as only accounting for the observed
labour market characteristics listed in the paragraph.
40 For example, the explained component of the gender wage gap may exist
because women have less education; this raises further questions as to why this
is the case. If, for example, this is because girls face gender discrimination in

Notes 99
education that manifests itself through lower access to schooling and reduced
attention from teachers (as compared to boys), education – captured by the
explained component – is grounded in discrimination that may contribute to
the unexplained component as well.
41 Half the median wage is used in Cyprus, since there is no national minimum
wage in place.
42 See ILO and DSS, 2004. The 15th International Conference of Labour
Statisticians (ICLS) defined the informal sector in terms of the characteristics
of the production units (household unincorporated enterprises), rather than the
characteristics of the jobs or the employed persons (ILO, 1993a). Employment
in the informal sector was thus composed of all persons employed in informal
sector enterprises. To better capture the “informalization” of employment,
the 17th ICLS defined informal employment as comprising all informal jobs,
whether carried out in formal sector enterprises, informal sector enterprises
or households. Informal jobs in formal sector enterprises include employment
that is not subject to standard labour legislation, income taxation, social pro-tection
or entitlement to certain social benefits (ILO, 2003a).
43 See http://www.ilo.org/global/standards/lang--en/index.htm
44 ILO’s Equal Remuneration Convention, 1951 (No. 100), represents a widely
agreed standard.
45 Looking at compliance rates with existing minimum wage laws in four Latin
American countries (Chile, Costa Rica, Peru and Uruguay), Marinakis shows,
for example, that the level of non-compliance is particularly high in rural areas,
where productivity is lower and where labour inspectorates and trade unions
have a much weaker presence (Marinakis, 2014).
46 ILO, forthcoming.
47 ILO, 2014f.
48 The ILO’s Social Protection Floors Recommendation, 2012 (No. 202), provides
guidance to member States in building comprehensive social security systems
and extending social security coverage by prioritizing the establishment of
national floors of social protection. To achieve universal coverage of popula-tions,
the ILO’s extension strategy recommends ensuring at least basic levels
of income security and access to health care as a priority through nationally
defined social protection floors while progressively ensuring a wider scope and
higher levels of protection.

Appendix I
49 Formerly known as the Conditions of Work and Employment Programme
(TRAVAIL).
50 ILO-commissioned report: Mehran, 2010. Peer reviews: Tillé, 2010; Jeong and
Gastwirth, 2010; Ahn, 2010.
51 Aiming for the broadest possible coverage is in line with the idea that decent
work and hence adequate earnings are a concern for all workers, and that stat-istical
indicators should cover all those to whom an indicator is relevant. See
ILO, 2008b.
52 We do so on the basis of the IMF’s consumer price index (CPI) for each country.
In cases where our national counterparts explicitly provide a real wage series,
the real wage series is used in place of the nominal series deflated by the IMF
CPI.
53 Our universe includes all countries and territories for which data on employment
are available from the ILO’s Global Employment Trends Model (GET Model),
and thus excludes some small countries and territories (e.g. the Channel Islands
and the Holy See) that have no discernible impact on global or regional trends.
54 This is in line with standard survey methodology, where a model-based frame-work
is generally used for item non-response, while a design-based framework
is used for questionnaire non-response.
55 For a discussion of the missing data problem, see also ILO, 2010b, p. 8.
56 An alternative specification with GDP per capita and population size produced
very similar results.
57 Data for the number of persons employed and the number of employees are
from KILM (ILO, 2013a), and data on GDP in 2005 PPP$ from the World
Bank’s World Development Indicators.
58 The estimate, ˆnh, of the number of employees in region h is obtained by mul-tiplying
the number of employees in countries from the region for which we
have wage data with the uncalibrated weights, and then summing up across
the region.
59 See for example the work done mainly for industrialized countries by the
International Labor Comparisons programme of the US Bureau of Labor
Statistics (2009) (http://www.bls.gov/fls/). Since we do not compare levels, but
focus on change over time in individual countries, data requirements are less
demanding in our context.
60 See also ILO, 2008a, p. 15 for the association between wage levels and GDP
per capita. Notwithstanding this, wage developments can diverge from trends
in labour productivity in the short and medium term.

Notes 101
Appendix II
61 Burtless, 2009.
62 OECD, 2011, p. 34.
63 According to the Canberra Group, an expert group mandated to provide con-ceptual
and definitional guidance to national statistical offices, “total income”
and “disposable income” are the main income aggregates produced (UNECE,
2011, p. 17). While total household income is reflected in figure A1, disposable
income is obtained by deducting from total income all current transfers paid,
such as direct taxes, social insurance contributions or other transfers. Disposable
income is usually the preferred measure in analysing income distribution, but it
is less frequently available than total income, including in our own data set.
64 Verma and Betti consider that comparability in the EU-SILC could be improved
through greater standardization across countries in the manner in which nega-tive,
zero and very large values are treated. They also found that non-response
is high to very high in various countries, which diminishes the quality of parts
of the data set (Verma and Betti, 2010).
65 Atkinson uses the “current number of hours worked and declared” by each
individual and estimates the median of this variable by country and gender and
separately for full time and part time, i.e. med(hours, ft)g,c and med(hours,
pt)g,c for each of the c countries and for g = male and female. Using these esti-mates,
by subgroup, he constructs atk – scale = med(hrs, pt)g,c ÷ med(hrs, ft)
g,c and multiplies – for each individual – the total number of months worked
as part time by atk – scale which is a number in the (0,1) range.
66 See http://www.cpc.unc.edu/projects/rlms-hse, http://www.hse.ru/org/hse/rlms.
Appendix III
67 We define household income as the sum of income from wages and other
incomes. The breakdown helps to illustrate the method where we hold wage
inequality constant at a given period of time and allow “other” sources of
household income (i.e. the sum of self-employment income, social benefits,
capital gains and intra-household private transfers) to vary between periods.
As above, wages refer to income received by waged employees in the house-hold.
The counterfactual for each individual is estimated using wage distribu-tions
at individual level and the result is incorporated into the corresponding
distributions of household incomes. For more detail, see DiNardo, Fortin and
Lemieux, 1996, and Daly and Valetta, 2004. For a more detailed explanation of
the components of household income and our estimation of per capita house-hold
income, see Appendix II.
68 That is, the measure ΔT1,T2D91(wages)C is equal to ΔD91(wage counterfactual to
T1) – ΔD91T1 where the counterfactual “mimics” the ranking and dispersion of
D9/D1 in T1. In other words, the ratios D9/D1T1 and D9/D1)T2(wage counterfactual to T1)

should be equal. In practice they can vary slightly in a ranking process that
maps workers ranked in T2 to a population in T1 that differs in size in the
sample.
Appendix IV
69 The selected set of variables to explain returns to participation follow the
classic Mincer selection in the empirical evaluation of labour market outcomes
(see Mincer, 1974).
70 We use propensity score matching. The method consists in defining a set of
characteristics X in the data that describe the labour market endowments of
individuals (age, education, experience, skills, occupation and work inten-sity
measures). Using as an example the gender wage gap, these character-istics
are used to estimate the conditional probability of being a female, i.e.
P(Female = 1 | X). The value of the propensity score for each female is com-pared
in a balancing process to the propensity score estimate of all males in
the sample. Those males that have an identical or similar propensity score to
a female are the closest match (or matches) to the female in terms of labour
market characteristics (education, experience, location, industry, etc.). Once
the closest male is located for each female (the nearest neighbour), the observed
male’s wage serves as the counterfactual wage to that female. We tried either
the closest member or functions of sets of close members (e.g. the average of
the n-set of closer males to each female in terms of the propensity score) but
the final results did not vary significantly. We selected “the closest”.

Bibliography
Abraham, K.; Haltiwanger, J. 1995. “Real wages and the business cycle”, in
Journal of Economic Literature, Vol. 33, Sep., pp. 1215–64.
Agnese, P.; Sala, H. 2011. “The driving forces behind the falling labour share
and persistent unemployment”, in Pacific Economic Review, Vol. 16, No. 5,
pp. 577–603.
Ahn, J. 2010. Responses to draft ILO report “Estimation of global wage trends:
Methodological issues, mimeo (Seoul, Korea Labor Institute).
Alesina, A.; Perotti, R. 1996. “Income distribution, political instability, and
investment”, in European Economic Review, Vol. 40, No. 6, pp. 1203–28.
Alesina, A.; Rodrik, D. 1994. “Distributive politics and economic growth”,
in Quarterly Journal of Economics, Vol. 109, No. 2, pp. 465–90.
Atkinson, A.B. 2007. “The distribution of earnings in OECD countries”,
in International Labour Review, Vol. 146, No. 1–2, pp. 41–60.
Azevedo, J.P.; Davalos, M.E.; Diaz-Bonilla, C.; Atuesta, B.; Castañeda, R.A.
2013. Fifteen years of inequality in Latin America: How have labor markets
helped? Policy Research Working Paper 6384 (Washington, DC, World
Bank Gender and Equity Unit).
Bai, C.-E.; Qian, Z. 2010. “The factor income distribution in China: 1978–2007”,
in China Economic Review, Vol. 21, pp. 650–70.
Bakis, A.; Polat, S. 2013. Wage inequality in Turkey: 2002–2010, GIAM
Working Papers 13-9 (Istanbul, Galatasaray University Economic
Research Center).
Banco de España. 2014. “Un análisis de los efectos composición sobre la
evolución de los salarios”, in Boletín Económico, 02/2014, pp. 57–61.
Bank of England. 2013. Inflation Report, Aug. (London).
Barrett, A.; McGuinness, S.; O’Brien, M. 2012. “The immigrant earnings
disadvantage across the earnings and skills distributions: The case of
immigrants from the EU’s new member states”, in British Journal of
Industrial Relations, Vol. 50, Issue 3, Sep. 2012, pp. 457–81.
Barro, R.J. 2000. “Inequality and growth in a panel of countries”, in Journal of
Economic Growth, Vol. 5, No. 1, pp. 5–32.
Bell, B.D.; Van Reenen, J. 2013. “Extreme wage inequality: Pay at the very top”,
in American Economic Review: Papers and Proceedings, Vol. 103, No. 3,
pp. 153–57.
Belman, D.; Wolfson, P.J. 2014. What does the minimum wage do? (Kalamazoo,
MI, W.E. Upjohn Institute for Employment Research).

Belser, P.; Vazquez-Alvarez, R. Forthcoming. The re-distributive effect of taxes
on household income inequality: A counterfactual approach, working paper
(Geneva, ILO).
Berg, J. Forthcoming. “Introduction”, in J. Berg (ed.): Labour markets,
institutions and inequality: Building just societies in the 21st century (Geneva
and Cheltenham, ILO and Edward Elgar).
Berg, A.; Ostry, J.D. 2011. Inequality and unsustainable growth: Two sides of the
same coin? Discussion Note SDN/11/08 (Washington, DC, IMF).
Bertranou, F.; Casanova, L.; Jiménez, M.; Jiménez, M. 2013. Informality and
employment quality in Argentina: Country case study on labour market
segmentation, Conditions of Work and Employment Series 49 (Geneva, ILO).
Betcherman, G. 2014. “Labor market regulations: What do we know about their
impacts in developing countries?”, in World Bank Research Observer online,
16 Apr., doi: 10.1093/wbro/lku005.
Birdsall, N.; Lustig, N.; McLeod, D. 2011. Declining inequality in Latin America:
Some economics, some politics, Working Paper 251 (Washington, DC,
Center for Global Development).
Blanchard, O.; Leigh, D. 2013. Growth forecast errors and fiscal multipliers,
Working Paper WP/13/1 (Washington, DC, IMF).
Bloomberg. 2014. “Lagarde: US should increase the minimum wage”, Market
Makers, Bloomberg Television, 16 June, http://www.bloomberg.com/video/
lagarde-u-s-should-increase-the-minimum-wage-kYnLLpMbQLGPxKD79_
DYZQ.html [23 Sep. 2014].
BLS (Bureau of Labor Statistics, US Department of Labor). 2012. International
comparisons of hourly compensation costs in manufacturing, USDL-12-
2460, 19 Dec. Available at: http://www.bls.gov/news.release/pdf/ichcc.pdf
[23 Sep. 2014].
Botman, D.; Jakab, Z. 2014. “Abenomics: Time for a push from higher wages”,
IMFdirect, 20 Mar. Available at: http://blog-imfdirect.imf.org/2014/03/20/
abenomics-time-for-a-push-from-higher-wages/ [30 Sep. 2014].
Brandolini, A.; Smeeding, T. 2009. “Income inequality in richer and OECD
countries”, in W. Salverda, B. Nolan and T. Smeeding (eds), pp. 71–101.
Bryson, A. 2007. “The effect of trade unions on wages”, in Reflets et perspectives
de la vie économique, Vol. 46, Nos 2–3, pp. 33–45.
Burtless, G. 2009. “Demographic transformation and economic inequality”, in
W. Salverda, B. Nolan and T. Smeeding (eds), pp. 435–55.
Carneiro, A.; Portugal, P.; Varejao, J. 2013. Catastrophic job destruction,
Discussion Paper 7670 (Bonn, IZA).
CEC (Commission of the European Communities); IMF (International
Monetary Fund); OECD (Organisation for Economic Co-operation and
Development); UN (United Nations); World Bank. 2009. System of National
Accounts 2008 (New York). Available at: http://unstats.un.org/unsd/
nationalaccount/docs/SNA2008.pdf [24 Sep. 2014].

Bibliography 105
Chauvel, L. 2006. Les classes moyennes à la dérive (Paris, Seuil).
Checchi, D.; Visser, J.; van de Werfhorst, H.G. 2010. “Inequality and union
membership: The influence of relative earnings and inequality attitudes”,
in British Journal of Industrial Relations, Vol. 48, No. 1, Mar., pp. 84–108.
Chen, Zhihong; Ge, Ying; Lai, Huiwen; Wan, Chi. 2013. “Globalization and
gender wage inequality in China”, in World Development, Vol. 44, Apr.,
pp. 256–66.
Chetty, R.; Hendren, N.; Kline, P.; Saez, E. 2014. Where is the land of
opportunity? The geography of intergenerational mobility in the United States,
Working Paper 19843, Jan. (Cambridge, MA, National Bureau of Economic
Research).
Clasen, J; Goerne, A. 2011. “Exit Bismarck, enter dualism? Assessing
contemporary German labour market policy,” in Journal of Social Policy,
Vol. 40 (4), pp. 795–810.
Cobham, A.; Sumner, A. 2013. Is it all about the tails? The Palma measure of
income inequality, Working Paper No. 343 (Washington, DC, Center for
Global Development).
Corak, M. 2013. “Income inequality, equality of opportunity, and
intergenerational mobility”, in Journal of Economic Perspectives, Vol. 27,
No. 3, pp. 79–102.
Cornia, G.A. 2012. Inequality trends and their determinants, United Nations
University World Institute for Development Economics Research (UNU-WIDER),
Working Paper 2012/09 (Helsinki).
Daly, M.C.; Valletta, R.G. 2004. “Inequality and poverty in the United States:
The effects of rising dispersion of men’s earnings and changing family
behaviour”, in Economica, Vol. 73, No. 289, pp. 75–98.
Das, M.; N’Diaye, P. 2013. Chronicle of a decline foretold: Has China reached the
Lewis turning point? Working Paper WP/13/26 (Washington, DC, IMF).
Deaton, A.; Zaidi, S. 2002. Guidelines for constructing consumption aggregates for
welfare analysis, Living Standards Measurement Study Working Paper 135
(Washington DC, World Bank).
d’Hombres, B.; Weber, A.; Elia, L. 2012. “Literature review on income inequality
and the effects on social outcomes”, in European Commission Joint
Research Center Scientific and Policy Report.
DiNardo, J.; Fortin, N.M.; Lemieux, T. 1996. “Labour market institutions and
the distribution of wages, 1973–1992: A semi-parametric approach”, in
Econometrica, Vol. 64, No. 5, pp. 1001–44.
Domnisoru, C. 2014. The largest drop in income inequality in the European Union
during the Great Recession: Romania’s puzzling case, Conditions of Work
and Employment Series 51 (Geneva, ILO).
Doris, A.; Sweetman, O.; O’Neill, D. 2013. Wage flexibility and the Great
Recession: The response of the Irish labour market, Discussion Paper 7787
(Bonn, IZA).

Dormeier Freire, A. 2009. “Motorbikes against Ho Chi Minh? Or the
consumption icons of a social transformation in Vietnam”, in Copenhagen
Journal of Asian Studies, Vol. 27, No. 1, pp. 67–87.
Doucouliagos, H.; Stanley, T.D. 2009. “Publication selection bias in minimum-wage
research? A meta-regression analysis”, in British Journal of Industrial
Relations, Vol. 47, No. 2, June, pp. 406–28.
Downes, P.; Hanslow, K.; Tulip, P. 2014. The effect of the mining boom on the
Australian Economy, Reserve Bank of Australia Research Discussion Paper
RDP2014-08.
EC (European Commission). 2007. “The labour income share in the European
Union”, in Employment in Europe 2007 (Brussels, Directorate-General for
Employment, Social Affairs and Equal Opportunities), pp. 237–72.
Feldstein, M. 2008. “Did wages reflect growth in productivity?”, in Journal of
Policy Modeling, Vol. 30, No. 4, pp. 591–94.
Felipe, J. 2008. “What policy makers should know about total factor productivity”,
in Malaysian Journal of Economic Studies, Vol. 45, No. 1, pp. 1–19.
Felipe, J.; McCombie, J.S.L. 2013. The aggregate production function and the
measurement of technical change (Northampton, MA, Edward Elgar).
Ferreira, F.H.G.; Ravallion, M. 2009. “Poverty and inequality: The global
context”, in W. Salverda, B. Nolan and T. Smeeding (eds), pp. 599–636.
Fleck, S.; Glaser, J.; Sprague, S. 2011. “The compensation–productivity gap:
A visual essay”, in Monthly Labor Review (Washington, DC, US Bureau of
Labor Statistics), Jan., pp. 57–69.
G20. 2013. Labour and employment and finance ministers’ communiqué, Moscow,
19 July. Available at: http://www.fin.gc.ca/n13/13-097-eng.asp [30 Sep. 2014].
G20. 2014a. G20 Labour and Employment Ministerial Declaration, Melbourne,
10–11 September 2014.
G20. 2014b. Communiqué Meeting of G20 Finance Ministers and Central Bank
Governors, Cairns, 20–21 September 2014. Available at: https://www.g20.
org/sites/default/files/g20_resources/library/Communique%20G20%20
Finance%20Ministers%20and%20Central%20Bank%20Governors%20
Cairns.pdf [30 Oct. 2014].
Galbraith, J.K. 2012. Inequality and instability: A study of the world economy just
before the crisis (Oxford, Oxford University Press).
Gasparini, L.; Cruces, G.; Tornarolli, L. 2009. Recent trends in income inequality
in Latin America, Working Paper 2009-132 (Verona, Society for the Study
of Economic Inequality).
Gasparini, L.; Lustig, N. 2011. “The rise and fall of income inequality in Latin
America”, in J.A. Ocampo and J. Ros (eds): Oxford handbook of Latin
American economics (Oxford, Oxford University Press), ch. 27.
Gleason, S.E. 2006. The shadow workforce: Perspectives on contingent work in the
United States, Japan, and Europe (W.E. Upjohn Institute for Employment
Research, Kalamazoo, Michigan).

Bibliography 107
Goldberg, P.; Pavcnik, N. 2007. “Distributional effects of globalization in
developing countries”, in Journal of Economic Literature, Vol. 45, No. 1
(Mar.), pp. 39–82.
Goldin, C. 2014. “A grand gender convergence: Its last chapter”, in American
Economic Review, Vol. 104, No. 4, pp. 1091–1119.
Gönenç, R.; Rawdanowicz, L. 2010. Regulatory reforms to unlock long-term
growth in Turkey, Economics Department Working Paper 77 (Paris, OECD).
Gorodnichenko, Y.; Sabirianova, P.K.; Stolyarov, D. 2010. “Inequality and
volatility moderation in Russia: Evidence from micro-level panel data on
consumption and income”, in Review of Economic Dynamics, Vol. 13, No. 1,
pp. 209–37.
Grimshaw, D. 2011. What do we know about low-wage work and low-wage workers?
Analysing the definitions, patterns, causes and consequences in international
perspective, Conditions of Work and Employment Series 28 (Geneva, ILO).
Grimshaw, D.; Rubery, J. Forthcoming. The motherhood pay gap: A review of the
issues, theory and international evidence (Geneva, ILO).
Guerriero, M. 2012. The labour share of income around the world. Evidence from
a panel data set, Development Economics and Public Policy Working Paper
Series, WP No. 32/2012 (Manchester, University of Manchester).
Hayter, S. Forthcoming. “Unions and collective bargaining”, in J. Berg (ed.):
Labour markets, institutions and inequality: Building just societies in the 21st
century (Geneva and Cheltenham, ILO and Edward Elgar).
Heinze, A.; Wolf, E. 2010. “The intra-firm gender wage gap: a new view on
wage differentials based on linked employer-employee data”, in Journal of
Population Economics, Vol. 23, pp. 851–79.
ILO (International Labour Office). 1966. Resolution concerning statistics of labour
cost, adopted by the 11th International Conference of Labour Statisticians
(Geneva).
ILO (International Labour Office). 1973. Resolution concerning an integrated
system of wages statistics, adopted by the 12th International Conference of
Labour Statisticians (Geneva).
ILO (International Labour Office). 1993a. Resolution concerning statistics of
employment in the informal sector, adopted by the 15th International
Conference of Labour Statisticians (Geneva).
ILO (International Labour Office). 1993b. Resolution concerning the International
Classification of Status in Employment (ISCE), adopted by the 15th
International Conference of Labour Statisticians (Geneva).
ILO (International Labour Office). 1998. Resolution concerning the measurement
of employment related income, adopted by the 16th International Conference
of Labour Statisticians (Geneva).
ILO (International Labour Office). 2003a. Guidelines concerning a statistical
definition of informal employment, adopted by the 17th International

ILO (International Labour Office). 2003b. Resolution concerning household
income and expenditure statistics, adopted by the 17th International
ILO (International Labour Office). 2008a. Global Wage Report 2008/09:
Minimum wages and collective bargaining – Towards policy coherence
(Geneva).
ILO (International Labour Office). 2008b. World of Work Report 2008: Income
inequalities in the age of financial globalization (Geneva).
ILO (International Labour Office). 2010a. Global Wage Report 2010/11: Wage
policies in times of crisis (Geneva).
ILO (International Labour Office). 2010b. Trends Econometric Models: A review
of the methodology, 19 Jan., Employment Trends Unit (Geneva).
ILO (International Labour Office). 2012a. Global Wage Report 2012/13: Wages
and equitable growth (Geneva).
ILO (International Labour Office). 2012b. Decent Work indicators: Concepts
and definitions (Geneva). Available at: http://www.ilo.org/wcmsp5/groups/
public/---dgreports/---integration/documents/publication/wcms_229374.pdf
[23 Sep. 2014].
ILO (International Labour Office). 2013a. Key Indicators of the Labour Market
(KILM), 8th edition (Geneva).
ILO (International Labour Office). 2013b. Portugal: Tackling the jobs crisis in
Portugal (Geneva).
ILO (International Labour Office). 2014a. Greece: Productive jobs, Studies on
Growth with Equity (Geneva).
ILO (International Labour Office). 2014b. International Journal of Labour
Research: The challenge of inequality, Vol. 6, Issue 1 (Geneva).
ILO (International Labour Office). 2014c. Provisional Record 13, Part I.
Third item on the agenda: Information and reports on the application
of Conventions and Recommendations. Report of the Committee on the
Applications of Standards. International Labour Conference, 103rd Session,
Geneva. Available at: http://www.ilo.org/wcmsp5/groups/public/---ed_norm/
---relconf/documents/meetingdocument/wcms_246781.pdf
ILO (International Labour Office). 2014d. Provisional Record 13, Part I.
Third item on the agenda: Information and reports on the application of
Conventions and Recommendations. International Labour Conference,
103rd Session, Geneva.
ILO (International Labour Office). 2014e. Employment policies for sustainable
recovery and development: Recurrent discussion under the ILO Declaration
on Social Justice for a Fair Globalization. Report VI, International Labour
Conference, 103rd Session, Geneva. Available at: http://www.ilo.org/
wcmsp5/groups/public/---ed_norm/---relconf/documents/meetingdocument/
wcms_240032.pdf [30 Sep. 2014].

Bibliography 109
ILO (International Labour Office). 2014f. World Social Protection Report
2014/15: Building economic recovery, inclusive development and social
justice (Geneva). Available at: http://www.ilo.org/wcmsp5/groups/public/---
dgreports/---dcomm/documents/publication/wcms_245201.pdf; http://www.
social-protection.org/gimi/gess/ShowTheme.action?th.themeId=3985.
ILO (International Labour Office). Forthcoming. Non-standard forms of
employment, Report for the Tripartite Meeting of Experts (Geneva).
ILO (International Labour Office); DSS (Department for Statistics and
Sociology of the Republic of Moldova). 2004. Employment in the informal
economy in the Republic of Moldova, Working Paper 41 (Geneva).
ILO (International Labour Office); IILS (International Institute for Labour
Studies). 2008. World of Work Report 2008: Income inequalities in the age of
financial globalization (Geneva).
ILO (International Labour Organization); OECD (Organisation for Economic
Co-operation and Development); World Bank. 2014. G20 labour markets:
Outlook, key challenges and policy responses. Report prepared for the G20
Labour and Employment Ministerial Meeting, Melbourne, Australia,
10–11 Sep.
IMF (International Monetary Fund). 2007. “The globalization of labor”, in
World Economic Outlook, Apr.: Spillovers and cycles in the world economy
(Washington, DC), pp. 161–92.
IMF (International Monetary Fund). 2013a. Germany: 2013 Article IV
consultation, IMF Country Report No. 13/255, Aug. (Washington, DC).
IMF (International Monetary Fund). 2013b. Japan: 2013 Article IV consultation,
IMF Country Report No. 13/253, Aug. (Washington, DC).
IMF (International Monetary Fund). 2013c. World Economic Outlook
Database, Oct.
IMF (International Monetary Fund). 2014a. United States: 2014 Article IV
consultation, IMF Country Report No. 14/221, July (Washington, DC).
IMF (International Monetary Fund). 2014b. Fiscal policy and income inequality
(Washington, DC).
Immervoll, H. 2009. Minimum-Income Benefits in OECD Countries: Policy
Design, Effectiveness and Challenges, IZA DP No. 4627.
Immervoll, H.; Pearson, M. 2009. A good time for making work pay? Taking
stock of in-work benefits and related measures across the OECD, IZA Policy
Paper No. 3.
Jencks, C. 2002. “Does inequality matter?”, in Daedalus, Vol. 131, No. 1, Winter,
pp. 49–65.
Jeong, Y.; Gastwirth, J.L. 2010. Comments on the draft ILO report “Estimation of
global wage trends: Methodological issues”, mimeo (Washington, DC, HEC
Montreal and George Washington University).
Karabarbounis, L.; Neiman, B. 2014. “The global decline of the labor share”,
in The Quarterly Journal of Economics, 129(1), pp. 61–103.

Keifman, S.N.; Maurizio, R. 2012. Changes in labour market conditions and
policies: Their impact on wage inequality during the last decade, United
Nations University World Institute for Development Economics Research
(UNU-WIDER), Working Paper 2012/14 (Helsinki).
Kharas, H. 2010. The emerging middle class in developing countries, Working
Paper 285 (Paris, OECD Development Centre).
Krueger, A. 2012. “The rise and consequences of inequality in the United States”,
speech at the Center for American Progress, Washington, DC, 12 Jan.
Kumhof, M.; Rancière, R. 2010. Inequality, leverage and crises, Working Paper
10/268 (Washington, DC, IMF).
Lavoie, M.; Stockhammer, E. (eds). 2013. Wage-led growth: An equitable
strategy for economic recovery (Geneva and Basingstoke, ILO and Palgrave
Macmillan).
Lazear, E.P.; Rosen, S. 1981. “Rank-order tournaments as optimum labor
contracts”, in Journal of Political Economy, Vol. 89, No. 5, pp. 841–64.
Lee, S.; McCann, D. 2014. “Regulatory indeterminacy and protection in
contemporary labour markets: Innovation in research and policy”, in
D. McCann, S. Lee, P. Belser, C. Fenwick, J. Howe and M. Luebker (eds):
Creative labour regulation: Indeterminacy and protection in an uncertain
world (Geneva and Basingstoke, ILO and Palgrave Macmillan), pp. 3–28.
Leonard, M.; Stanley, T.D.; Doucouliagos, H. 2013. “Does the UK minimum
wage reduce employment? A meta-regression analysis”, in British Journal of
Industrial Relations, Vol. 52, No. 3, pp. 499–520.
López-Calva, L.; Lustig, N. (eds). 2010. Declining inequality in Latin America:
A decade of progress? (New York and Washington, DC, UNDP and
Brookings Institution Press).
Low Pay Commission, 2014. National minimum wage, Low Pay Commission
Report 2013, Cm 8816, Apr. (London, The Stationery Office).
Lucidi, F.; Kleinknecht, A. 2010. “Little innovation, many jobs: An econometric
analysis of the Italian labour productivity crisis”, in Cambridge Journal of
Economics, Vol. 34, No. 3, pp. 525–46.
Lukyanova, A. 2011. Effects of minimum wages on the Russian wage distribution,
Working Paper WP BRP 09/EC/2011 (National Research University Higher
School of Economics, Moscow).
Lukyanova, A. Forthcoming. Analysis of trends in the distribution of wages in the
Russian Federation, Conditions of Work and Employment Series (Geneva,
ILO).
Machado, A.; Perez Ribas, R. 2010. “Do changes in the labour market take
families out of poverty? Determinants of exiting poverty in Brazilian
metropolitan regions”, in Journal of Development Studies, Vol. 46, No. 9,
pp. 1503–22.
Machin, S. 2009. “Education and inequality”, in W. Salverda, B. Nolan and
T. Smeeding (eds), pp. 406–31.

Bibliography 111
Marinakis, A. 2014. Incumplimiento con el salario mínimo en América Latina:
El peso de los factores económicos e institucionales (Geneva, ILO).
Martins, P.S.; Solon, G.; Thomas, J.P. 2012. “Measuring what employers do
about entry wages over the business cycle: A new approach”, in American
Economic Journal: Macroeconomics, Vol. 4, No. 4, pp. 36–55.
Mehran, F. 2010. Estimation of global wage trends: Methodological issues, mimeo
(Geneva, ILO).
Migone, A. 2007. “Hedonistic consumerism: Patterns of consumption in
contemporary capitalism”, in Review of Radical Political Economics,
Vol. 39, No. 2, pp. 173–200.
Milanovic, B. 2011. The haves and the have-nots: A brief and idiosyncratic history
of global inequality (New York, Basic Books).
Mincer, J. 1974. Schooling, experience and earnings (New York, National Bureau
of Economic Research).
Nataraj, S.; Perez-Arce, F.; Srinivasan, S.; Kumar, K.B. 2012. What is the impact
of labour market regulations in LICs? How does it vary by gender? RAND,
Labour and Population Working Paper No. 957 (Santa Monica, CA, RAND).
Naughton, B. 2007. The Chinese economy: Transitions and growth (Cambridge,
MA, and London, MIT Press).
Neves, P.C.; Silva, S.T. 2014. “Inequality and growth: Uncovering the main
conclusions from the empirics”, in Journal of Development Studies, Vol. 50,
No. 1, pp. 1–21.
OECD (Organisation for Economic Co-operation and Development).
2001. Measuring productivity: OECD manual (Paris).
OECD (Organisation for Economic Co-operation and Development). 2008.
Growing unequal? Income distribution and poverty in OECD countries
(Paris).
OECD (Organisation for Economic Co-operation and Development).
2011. Divided we stand: Why inequality keeps rising (Paris).
OECD (Organisation for Economic Co-operation and Development). 2012a.
Employment Outlook 2012 (Paris).
OECD (Organisation for Economic Co-operation and Development). 2012b.
Economic Policy Reforms: Going for growth 2012 (Paris).
OECD (Organisation for Economic Co-operation and Development). 2014a.
All on board: Making inclusive growth happen (Paris).
OECD (Organisation for Economic Co-operation and Development). 2014b.
Employment Outlook 2014 (Paris).
Ostry, J.D.; Berg, A.; Tsangarides, C.G. 2014. Redistribution, inequality, and
growth, Discussion Note SDN/14/02 (Washington, DC, IMF).
Oyvat, C. 2011. “Globalization, wage shares and income distribution in Turkey”,
in Cambridge Journal of Regions, Economy and Society, Vol. 4, No. 1,
pp. 123–38.

Palley, T.I.; Horn, G.A. (eds). 2013. Restoring shared prosperity: A policy agenda
from leading Keynesian economists (CreateSpace Independent Publishing).
Perotti, R. 1996. “Growth, income distribution, and democracy: What the data
say”, Journal of Economic Growth, Vol. 1, No. 2, pp. 149–87.
Piketty, T. 2013. Le capital au XXI e siècle (Paris, Seuil).
Rajan, R. 2010. Fault lines: How hidden fractures still threaten the world economy
(Princeton, NJ, Princeton University Press).
Rand, J.; Torm, N. 2012. “The benefits of formalization: Evidence from
Vietnamese manufacturing SMEs”, in World Development, Vol. 40, No. 5,
May, pp. 983–98.
Rani, U.; Belser, P.; Oelz, M.; Ranjbar, S. 2013. “Minimum wage coverage and
compliance in developing countries”, in International Labour Review,
Vol. 152, No. 3–4, Special Issue, Informality across the global economy –
subsistence, avoidance and violation (Geneva, ILO).
Rao, S.; Sharpe, A.; Smith, J. 2005. “An analysis of the labour productivity
growth slowdown in Canada since 2000”, in International Productivity
Monitor, No. 10, Spring, pp. 3–23.
Ravallion, M. 2010. “The developing world’s bulging (but vulnerable) middle
class”, in World Development, Vol. 38, No. 4, pp. 445–54.
Rodrik, D. 1999. “Where did all the growth go? External shocks, social conflict,
and growth collapses”, in Journal of Economic Growth, Vol. 4, No. 4,
pp. 385–412.
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 DECLARATION/WP/19/2003 (Geneva,
ILO).
Rubery, J.; Grimshaw, D.; Figueiredo, H. 2005. “How to close the gender pay
gap in Europe: Towards the gender mainstreaming of pay policy”, in
Industrial Relations Journal, Vol. 36, No. 3, pp. 184–213.
Salverda, W.; Nolan, B.; Smeeding, T. (eds). 2009a. The Oxford handbook of
economic inequality (Oxford, Oxford University Press).
Salverda, W.; Nolan, B.; Smeeding, T. 2009b. “Introduction”, in W. Salverda,
B. Nolan and T. Smeeding (eds), pp. 3–23.
Särndal, C.-E.; Deville, J.-C. 1992. “Calibration estimators in survey sampling?”, in
Journal of the American Statistical Association, Vol. 87, No. 418, pp. 376–82.
Sharpe, A.; Arsenault, J.-F.; Harrison, P. 2008. The relationship between labour
productivity and real wage growth in Canada and OECD countries, Research
Report 2008-8 (Ottawa, Centre for the Study of Living Standards).
Sobeck, K. Forthcoming. Trends in wages and productivity in developed
economies, Conditions of Work and Employment series (Geneva, ILO).
Solé, C.; Parella, S. 2003. “The labour market and racial discrimination in
Spain”, in Journal of Ethnic and Migration Studies, Vol. 29, No. 1, pp. 121–40.

Bibliography 113
Solon, G.; Barsky, R.B.; Parker, J.A. 1994. “Measuring the cyclicality of real
wages: How important is composition bias?”, in Quarterly Journal of
Economics, Vol. 109, No. 1, Feb., pp. 1–25.
Sommer, M. 2009. Why are Japanese wages so sluggish?, Working Paper
WP/09/97 (Washington, DC, IMF).
Statistics Canada. 2014. Table 281-0063 – Employment and average weekly
earnings including overtime (SEPH), seasonally adjusted, for all employees
by industries classified using the North American Industry Classification
System (NAICS), annual (persons unless otherwise noted) [29 Oct. 2014].
Sturn, S.; Van Treeck, T. 2013. “The role of income inequality as a cause
of the Great Recession and global imbalances”, in M. Lavoie and
E. Stockhammer (eds), pp. 125–52.
Tansel, A.; Bircan, F. 2010. “Wage inequality and returns to education in Turkey:
A quantile regression analysis”, Discussion Paper 5417 (Bonn, IZA).
Tansel, A.; Kan, E.O. 2012. “The Formal/Informal Employment Earnings Gap:
Evidence from Turkey”, Discussion Paper 6556 (Bonn, IZA).
Tillé, Y. 2001. Théorie des sondages: Echantillonage et estimation en populations
finies (Paris, Dunod).
Tillé, Y. 2010. Expertise report on the “Estimation of global wage trends:
Methodological issues”, mimeo (Neuchâtel, Institute of Statistics, University
of Neuchâtel).
Trentmann, F. 2004. “Beyond consumerism: New historical perspectives
on consumption”, in Journal of Contemporary History, Vol. 39, No. 3,
pp. 373–401.
Tronti, L. 2010. “The Italian productivity slow-down: The role of the bargaining
model”, in International Journal of Manpower, Vol. 31, Issue 7, pp. 770–92.
UNdata. 2014. National accounts data. Available at: http://data.un.org/
[8 Aug. 2014].
UNECE (United Nations Economic Commission For Europe). 2011. Canberra
Group handbook on household income statistics, 2nd edn (Geneva, United
Nations). Available at: http://www.unece.org/fileadmin/DAM/stats/groups/
cgh/Canbera_Handbook_2011_WEB.pdf [22 Sep. 2014].
USCEA (United States Council of Economic Advisers). 2014. The economic
report of the President together with the annual report of the Council of
Economic Advisers, 2014 (Washington, DC, US Government Printing
Office).
Van Biesebroeck, J. Forthcoming. How tight is the link between wages
and productivity? A survey of the literature, Conditions of Work and
Employment series (Geneva, ILO).
Verma, V.; Betti, G. 2010. “Data accuracy in EU-SILC”, in A.B. Atkinson and
E. Marlier (eds): Income and living conditions in Europe (Luxembourg,
Publications Office of the European Union), pp. 57–77.

Visser, J.; Checchi, D. 2009. “Inequality and the labor market: Unions”, in
W. Salverda, B. Nolan and T. Smeeding (eds), pp. 230–57.
Wittenberg, M. 2014. Analysis of employment, real wage, and productivity trends
in South Africa since 1994, Conditions of Work and Employment Series 45
(Geneva, ILO).
Wolf, M. 2014a. The shifts and the shocks: What we’ve learned – and have still to
learn – from the financial crisis (London, Penguin).
Wolf, M. 2014b. Reform alone is no solution for the Eurozone, Financial Times,
21 Oct.
World Bank. 1999. Russia: Targeting and the longer-term poor, Report 19377-RU
(Washington, DC).
World Bank. 2006. World Development Report 2006: Equity and development
(Washington, DC, and Oxford, World Bank and Oxford University Press).
World Bank. 2012. “Overview”, in World Development Report 2012: Gender
equality and development (Washington, DC), pp. 2–40.
Yellen, J.L. 2014. “Labor market dynamics and monetary policy”, speech
given at the Federal Reserve Bank of Kansas City Economic Symposium,
Jackson Hole, Wyoming, 22 Aug. Available at: http://www.federalreserve.
gov/newsevents/speech/yellen20140822a.htm [22 Sep. 2014].

Global Wage report de l'OIT - 4/12/2014

More Related Content

What's hot (20)

Viewers also liked (20)

Similar to Global Wage report de l'OIT - 4/12/2014 (20)

More from FactaMedia (20)

Recently uploaded (20)

Global Wage report de l'OIT - 4/12/2014