Europe's disappearing middle class?: Evidence from the world of work [1 ed.] 9789221303824, 9789221303817

While recent studies have highlighted the phenomenon and risks of increased inequalities between the top and the bottom,

309 32 18MB

English Pages 665 Year 2016

Report DMCA / Copyright

DOWNLOAD PDF FILE

Recommend Papers

Europe's disappearing middle class?: Evidence from the world of work [1 ed.]
 9789221303824, 9789221303817

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

EUROPE'S DISAPPEARING MIDDLE CLASS? EVIDENCE FROM THE WORLD OF WORK EDITED BY DANIEL VAUGHAN-WHITEHEAD

Elgar

Europe’s Disappearing Middle Class?

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 1

28/09/2016 08:24

Europe’s Disappearing Middle Class? Evidence from the World of Work

Edited by

Daniel Vaughan-Whitehead Senior Economist, International Labour Office, Geneva, Switzerland and Professor at Sciences Po, Paris, France

Cheltenham, UK • Northampton, MA, USA

International Labour Office Geneva, Switzerland

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 3

28/09/2016 08:24

© International Labour Organization 2016 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical or photocopying, recording, or otherwise without the prior permission of the publisher. Publications of the International Labour Office enjoy copyright under Protocol 2 of the Universal Copyright convention. Nevertheless, short excerpts from them may be reproduced without authorization, on condition that the source is indicated. For rights of reproduction or translation, application should be made to ILO Publishing (Rights and Licensing), International Labour Office, 4 route des Morillons, CH-1211 Geneva 22, Switzerland, or by email: [email protected]. The International Labour Office welcomes such applications. 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 studies and other contributions rests solely with their authors, and publication does not constitute an endorsement by the International Labour Office of the opinions expressed in them. Reference to names of firms and commercial products and processes does not imply their endorsement by the International Labour Office, and any failure to mention a particular firm, commercial product or process is not a sign of disapproval. Published by Edward Elgar Publishing, Inc. Edward Elgar Publishing Limited William Pratt House The Lypiatts 9 Dewey Court 15 Lansdown Road Cheltenham Northampton Glos GL50 2JA Massachusetts 01060 UK USA In association with International Labour Office 4 route des Morillons 1211 Geneva 22 Switzerland ISBN 978-92-2-130381-7 (paperback) A catalogue record for this book is available from the British Library Library of Congress Control Number: 2016935812 This book is available electronically in the Economics subject collection DOI 10.4337/9781786430601

ISBN 978 1 78643 059 5 (cased) ISBN 978 1 78643 060 1 (eBook) Typeset by Servis Filmsetting Ltd, Stockport, Cheshire

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 4

28/09/2016 08:24

Contents List of contributorsvii Foreword by Heinz Kollerxi   1. Is the world of work behind middle-­class erosion? Daniel Vaughan-­Whitehead, Rosalia Vazquez-­Alvarez and Nicolas Maître   2. Is the world of work stimulating middle-­class growth in the Baltic states? Jaan Masso, Inta Mierina and Kerly Espenberg   3. Social concertation and middle-­class stability in Belgium Sarah Kuypers and Ive Marx

1

62 112

  4. Transformation in the world of work and the middle class: The French experience Pierre Courtioux and Christine Erhel

160

  5. The erosion of the German middle class: The end of the ‘levelled-­out, middle-­class society’? Gerhard Bosch and Thorsten Kalina

198

  6. The middle classes in the Greek Great Depression: Dissolution or resilience? Maria Karamessini and Stefanos Giakoumatos

244

  7. Is Hungary still in search of its middle class? István György Tóth

279

  8. Middle incomes in boom and bust: The Irish experience Bertrand Maître and Brian Nolan

323

  9. The middle class in Italy: Reshuffling, erosion, polarization Annamaria Simonazzi and Teresa Barbieri

360

10. Stagnating incomes and the middle class in the Netherlands: Running to stand still? Wiemer Salverda

396

v

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 5

28/09/2016 08:24

vi

Europe’s disappearing middle class?

11. Still holding on? Inequality, labour market and middle-­income groups in Portugal Pilar González, António Figueiredo, Hugo Figueiredo and Luis Delfim Santos 12. Knocking on heaven’s door: Changes in the world of work and the middle class in Spain Rafael Muñoz de Bustillo and José-­Ignacio Antón 13. The rise and fall of the Swedish middle class? Dominique Anxo 14. How have middle-­income households fared in unequal Britain? A focus on work and employment trends Damian Grimshaw and Anthony Rafferty

441

495 543

586

Index633

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 6

28/09/2016 08:24

Contributors José-Ignacio Antón  Senior Researcher at the Institute of Economics of the Johannes Kepler University, Linz, Austria. Dominique Anxo  Professor of Economics at the Department of Economics, Linnaeus University, Sweden and Director of the Centre of Ageing and Life-­course Studies. Teresa Barbieri  PhD candidate in the European PhD in Socio-­ Economic and Statistical Studies programme at Sapienza University, Rome, Italy. Gerhard Bosch  Professor of Sociology and Senior Professor in the Institute of Work, Skills and Training (IAQ) at the University of Duisburg-­ Essen, Germany. Pierre Courtioux  Researcher at Edhec Business School and Associate Researcher at the Centre d’Economie de la Sorbonne, France. Christine Erhel  Associate Professor in Economics at the University of Paris 1 and Researcher at the Centre d’Economie de la Sorbonne, France. Kerly Espenberg  Project Manager and Analyst, Centre for Applied Social Sciences, University of Tartu, Estonia. António Figueiredo  Assistant Professor at the Faculty of Economics of the University of Porto (retired) and Director of Quaternaire Portugal, Portugal. Hugo Figueiredo  Assistant Professor at the Department of Economics, Management and Industrial Engineering of the University of Aveiro and Researcher at CIPES (Centre for Research in Higher Education), Portugal. Stefanos Giakoumatos  Professor of Statistics and Quantitative Methods, Department of Accounting and Finance, Technological Educational Institute of Peloponnese, Greece. Pilar González  Assistant Professor at the Faculty of Economics of the University of Porto and Researcher at CEF.UP, Portugal. Damian Grimshaw  Professor of Employment Studies and Director of vii

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 7

28/09/2016 08:24

viii

Europe’s disappearing middle class?

EWERC (European Work and Employment Research Centre), Manchester Business School, University of Manchester, United Kingdom. Thorsten Kalina  Sociologist and Senior Researcher at the Institute of Work, Skills and Training (IAQ) at the University of Duisburg-­Essen, Germany. Maria Karamessini  Professor of Labour Economics and Economics of the Welfare State at Panteion University, Athens and Governor of the Greek Public Employment Agency, Greece. Sarah Kuypers  Doctoral researcher at the Herman Deleeck Centre for Social Policy at the University of Antwerp, Belgium. Bertrand Maître  Senior Research Officer at the Economic and Social Research Institute, Dublin, Republic of Ireland. Nicolas Maître  Researcher, INWORK (Inclusive Labour Markets, Labour Relations and Working Conditions Branch), ILO, Geneva, Switzerland. Ive Marx  Professor of Social Policy at the Herman Deleeck Centre for Social Policy, University of Antwerp, Brussels. Jaan Masso  Senior Research Fellow, Faculty of Economics and Business Administration, University of Tartu, Estonia. Inta Mierina  Senior Researcher, Institute of Philosophy and Sociology, University of Latvia, Latvia. Rafael Muñoz de Bustillo  Professor of Applied Economics at the University of Salamanca, Spain. Brian Nolan  Professor of Social Policy and Director of the Employment Equity and Growth research programme at the Institute for New Economic Thinking, University of Oxford, United Kingdom. Anthony Rafferty  Senior Lecturer in Employment Studies at Manchester Business School, United Kingdom. Wiemer Salverda  Professor of Labour Market and Inequality, Amsterdam Centre for Inequality Studies (AMCIS) and Director Emeritus of the Amsterdam Institute for Advanced Labour Studies (AIAS), Netherlands. Luis Delfim Santos  Assistant Professor at the Faculty of Economics of the University of Porto and Researcher at CEF.UP, Portugal. Annamaria Simonazzi Professor of Economics, Economics and Law, Sapienza University, Rome, Italy.

Department

of

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 8

28/09/2016 08:24



Contributors ­ix

István György Tóth  Director of TÁRKI Social Research Institute, Budapest, Hungary. Daniel Vaughan-­Whitehead Senior Economist, Inclusive Labour Markets, Labour Relations and Working Conditions Branch (INWORK), ILO, Geneva, Switzerland and Professor at Sciences Po, Paris, France. Rosalia Vazquez-­Alvarez  Econometrician, Wage Specialist, Wage Group at INWORK (Inclusive Labour Markets, Labour Relations and Working Conditions Branch), ILO, Geneva, Switzerland.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 9

28/09/2016 08:24

Foreword On 29 February 2016 in Brussels, on the occasion of the Conference jointly organized by the ILO and the European Commission on the ‘Long term in the world of work: Effects on inequalities and middle-­income groups’, I had the great privilege to host seven ministers and state secretaries in a round table about the future of the middle class in Europe. The policy debate was very rich and, as an introduction to this volume, we decided to publish a number of quotes from our honourable guests, extracted from their answers to three basic questions. 1. What are the main trends in the world of work which you believe have influenced inequalities in your country and, in particular, the evolution of the middle class? Mr Ivailo Kalfin, Deputy Prime Minister and Minister of Labour and Social Policy, Bulgaria ‘The reasons we have high inequalities at work in Bulgaria are related to a lack of qualification and education; they are also related to a very unfair tax system which is a flat-­rate tax for personal income of 10 per cent, and also regressive taxation with regard to social security.’ Mr Thorben Albrecht, State Secretary in the Federal Ministry of Labour and Social Affairs, Germany ‘In Germany, inequality is not as bad as in other countries, but that is only after redistribution through the tax and social systems. We have to look at primary wealth distribution and equality in the labour market. In contrast with some other countries, the middle class in Germany is not only concentrated in the public sector, but also in manufacturing, where we have middle-­skilled workers who are rather highly paid. But this model has come under pressure due to the growth over the past 25 years of non-­standard forms of employment, which are often associated with lower wages and social security. These trends have not been as effective as they should have been. Women’s employment increased, of course, but it is often confined to involuntary part-­time and low-­paid jobs. People with migrant backgrounds go through education but do not always succeed in xi

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 11

28/09/2016 08:24

xii

Europe’s disappearing middle class?

acquiring the skills that are necessary in the German labour market to really get into the middle class.’ Mr Georgios Katrougkalos, Minister of Labour, Social Security and Welfare, Greece ‘Inequalities have been exacerbated and the middle class has been squeezed in Greece for a series of reasons. Starting with the decline of the welfare state – and its provision of public goods in sectors such as education, social policy, and social security and health – and a decline in trade union membership which, of course, reduces the relative bargaining power of labour, we have experienced in the crisis the most massive deregulation of collective agreements in Europe – a reduction in the minimum wage and a higher share of part-­time and temporary employment. The progressivity of tax systems also declined from 59 to 30 per cent from 1980 to 2009.’ Ms Algimanta Pabedinskiene, Minister of Labour and Social Security, Lithuania and also a former President of the EU Labour Minister Council ‘One factor that is important for the growth of the middle class in Lithuania, as in other Central and Eastern European countries, is the education system. At the same time – and it is a paradox in Lithuania – although 60 per cent of the population has a high level of education, the middle class represents only 20−25 per cent. Other factors are also important, such as wages, and also the tax system – income tax but also taxes on SMEs as a constraint on the middle class. In Lithuania, signficant emigration has also limited the growth of the middle class.’ Mr Nicolas Schmit, Minister of Labour, Luxembourg and also former President of the Labour Minister Council class society. ‘Luxembourg is a small country, and a typical middle-­ Generally, we have been able to maintain this very strong middle class but, nevertheless, we also notice an increase in inequality; the gap between those who are unskilled and those who are at the upper level of the economy is increasing. There is a risk of an intergenerational gap, because young people suffer the most from unemployment, from precarity and from the increasing price of real estate, which makes it difficult to become a home owner. The model now is not one wage earner, but two; furthermore, these two wage earners are having more and more difficulties in buying a property and this is something which is contributing to the feeling: “if I am not able to buy my own apartment or house, do I belong to the middle class or not?”’

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 12

28/09/2016 08:24



Foreword ­xiii

Mr Miguel Cabrita, State Secretary in the Ministry of Labour, Solidarity and Social Security, Portugal ‘There are several reasons behind inequalities in Portugal: a structural deficit in the educational level of the workforce associated with low wages and a labour market segmentation due to the rapid growth of non-­ standard forms of employment, further exacerbated by fiscal consolidation programmes. The proportion of minimum wage earners has increased rather than decreased over recent years and collective bargaining has been weakened.’ Mrs Renata Szczech, Deputy Minister in the Ministry of Family, Labour and Social Policy, Poland ‘Inequalities in Poland are not coming much from the bottom but rather from the growth of temporary work and part-­time work. Today 80 per cent of young people work under a temporary contract and we would like to change this, with additional rules in terms of social security contributions associated to temporary contracts.’ 2. What are the policies in the world of work that you believe may help to reduce inequalities and favour the development of the middle class? Mr Ivailo Kalfin, Deputy Prime Minister and Minister of Labour and Social Policy, Bulgaria ‘To fight inequalities we first need a long-­term strategy for increasing wages that should start by stopping the limits on the increase in the minimum wage – they are low enough in Bulgaria and lead to low pay and poverty. This should then be complemented by collective bargaining on wages. We should also invest in professional and vocational training’. Mr Thorben Albrecht, State Secretary in the Federal Ministry of Labour and Social Affairs, Germany ‘Our government – also to stimulate domestic demand – decided to implement a number of reforms to strengthen wages and wage formation in Germany. First, and probably most important, is the introduction of a mandatory minimum wage, which was introduced last year to put a bottom floor into the wage-­formation system. Something that is less known by the public is that the same law makes collective agreements legally binding for a whole industry and sector, to strengthen wage collective bargaining. We are also trying to re-­regulate temporary agency work since this also influences wage formation in Germany.’

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 13

28/09/2016 08:24

xiv

Europe’s disappearing middle class?

Mr Georgios Katrougkalos, Minister of Labour, Social Security and Welfare, Greece ‘We must change our orientation. For us, the basic imperative is the return of collective bargaining – notably through a new law on collective ­bargaining – and social dialogue. Pension reforms are also key to guaranteeing a fairer redistribution despite generally smaller pension levels. But it will not be possible to have socially friendly solutions in one country if we cannot address the general social deficit in Europe, if we cannot balance the economic dimension of the European Union with equal respect for its social dimension.’ Ms Algimanta Pabedinskiene, Minister of Labour and Social Security, Lithuania ‘While the middle class is growing in cities, it is stagnating in less developed regions. In Lithuania, which has 60 municipalities, we are trying to develop a regional policy for local development in all sectors as one lever of middle-­class growth, with a combination of horizontal and vertical policies.’ Mr Nicolas Schmit, Minister of Labour, Luxembourg ‘We have to guarantee young people good education and good skills, and help them to enter the labour market on a good footing. And I think that we have done well with the youth guarantee and so it is a positive step in the European Union. We in Luxembourg have now decided to have free childcare from 2018 onwards, to give free and high-­quality childcare to everybody, because quality childcare is essential in terms of skills, of education and also of finding a good job later on. Coming back to some kind of wage indexation is not a bad idea to prevent the erosion of purchasing power; it has not destroyed the competitiveness of our Luxembourg companies, but has maintained the purchasing power. We also need to come back to some sort of a normal labour contract.’ Mr Miguel Cabrita, State Secretary in the Ministry of Labour, Solidarity and Social Security, Portugal ‘The Portuguese middle class is traditionally – even if not exclusively – connected to the public sector and this middle class has been affected in the past few years by early retirement agreements, wage cuts, heightened fiscal pressure, suspension of new recruitment and also freezes in career progression. Specifically for the middle classes, wages are the core revenue source. We must promote the middle class through a number of fundamental pillars: a

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 14

28/09/2016 08:24



Foreword ­xv

comprehensive national anti-­poverty strategy – also based on social allowances and benefits for the most vulnerable groups – a sustainable growth of the minimum wage to be determined through an agreement with social partners and job creation through a set of macroeconomic and sectorial policies, but also good quality jobs by restraining the use of temporary contracts and regulating self-­employment and other atypical forms of employment.’ Mrs Renata Szczech, Deputy Minister in the Ministry of Family, Labour and Social Policy, Poland ‘The Polish economy is taking part of its competitiveness from low wages. This is an orientation that we might need to modify to evolve towards a more modern economy, based on higher quality and value-­added products and services, and which will generate the future growth of the middle class. Ensuring the middle class in the future also requires strong demographic and family policies, alongside the reduction of the labour force and an increased number of pensioners. As in other countries, one major issue for the growth of the middle class in Poland is education. In 2014, we put in place a national training fund to support employers willing to train their labour force (80 per cent being paid by the state and 20 per cent by the employers), a fund that has achieved great success, with a rapid growth of workers benefiting from this system.’ 3. In this policy agenda, what role could social partners and social dialogue play in cooperation with the government? Mr Ivailo Kalfin, Deputy Prime Minister and Minister of Labour and Social Policy, Bulgaria ‘The income strategy and labour market issues should be decided together with the social partners. Beyond that, inequalities and the middle class are rarely discussed, at least in Bulgaria, when discussing different macroeconomic parameters such as poverty, wages and minimum wages. We thus thank the ILO and the European Commission for bringing this topic to the centre of the policy debate, because inequalities should always be taken into consideration when discussing social and economic policies.’ Mr Thorben Albrecht, State Secretary in the Federal Ministry of Labour and Social Affairs, Germany ‘Social dialogue should play a crucial role. While the digitalization of the workplace will require our labour markets to adjust to these new forms of work, tailor-­made solutions to combine flexibility with workers’ protection

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 15

28/09/2016 08:24

xvi

Europe’s disappearing middle class?

for different industries can best be made by the social partners and through collective agreements.’ Mr Georgios Katrougkalos, Minister of Labour, Social Security and Welfare, Greece ‘We hope to return to collective bargaining and social dialogue in Greece, notably through a new law on collective bargaining that we are preparing in consultation with the ILO.’ Ms Algimanta Pabedinskiene, Minister of Labour and Social Security, Lithuania ‘We also believe that social dialogue is key and we are changing the law to strengthen the role of the social partners in decision-­making in areas such as employment and social security policies.’ Mr Nicolas Schmit, Minister of Labour, Luxembourg ‘We have to rethink our wealth distribution model, especially concerning wages.’ Mr Miguel Cabrita, State Secretary in the Ministry of Labour, Solidarity and Social Security, Portugal ‘We believe that social dialogue should support the development of an income policy based on a decent work perspective, without undermining competitiveness. Although the dynamism of collective bargaining strongly depends on the engagement of the social partners, we understand and believe that the government can, and should, act in order to stimulate collective bargaining – for example, through the extension of collective agreements and other legal instruments – and make negotiations more appealing both to trade unions and companies.’ Mrs Renata Szczech, Deputy Minister in the Ministry of Family, Labour and Social Policy, Poland ‘We have set up a new tripartite social council with the aim of formulating constraining opinions on new legal provisions and initiatives.’ The above collection of ministerial quotations highlights three elements that we believe are essential for both national policies and the work of international organizations. First, middle-­income groups should become a policy target to be taken into account in policy discussions, including those on macroeconomic and budgetary issues.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 16

28/09/2016 08:24



Foreword ­xvii

Second, a new policy mix to reach that target, and to cover inequalities issues more broadly, is required; one which would include policy areas such as education, social protection, taxation and regional policy, but also real estate, capital markets, and public investment and services. Third, the world of work reveals itself to be a determinant for the future fate of the middle class in Europe, not only in terms of job availability, but also in terms of the quality of employment and work contracts, redistribution through wages and incomes policies, the enhanced participation of women in the labour market, collective bargaining and tripartite social dialogue, and the active involvement of the social partners in policy-­making. Obviously, the evidence and messages we could deliver on the role of the European Social Model in reducing inequalities and strengthening the middle class in Europe also provide useful lessons at the global level, as, at the moment, middle classes are developing in the BRICS and other ­countries in the world. We are happy that the present volume can help place this topic at the core of policy-­making, a debate that should continue, and that the ILO will strengthen its efforts with national actors and other international organizations. Heinz Koller Assistant Director-­General, Regional Director for Europe and Central Asia, ILO

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 17

28/09/2016 08:24

In memory of Jean-Jacques Paris, for his friendship and his commitment to a stronger Social Europe.

We would like to thank James Patterson for his copy-editing work and Boroka Gergely for her cover design.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 18

28/09/2016 08:24

1. Is the world of work behind middle-­class erosion? Daniel Vaughan-­Whitehead, Rosalia Vazquez-­Alvarez and Nicolas Maître 1. INTRODUCTION The international debate has recently focused on increased income ­inequalities and on the adverse effects that such inequalities may have on both social and economic developments (EC 2013, 2015; Dabla-­Norris et  al. 2015; OECD 2015). According to the Organisation for Economic Co-­operation and Development (OECD 2015: 3), ‘income inequality is at its highest level for the past half century’. Generally, this work on inequalities (with the possible exception of Gornick and Jänti 2013) has so far concentrated on the differences between the top and the bottom (see, for instance, EC 2015). The purpose of this volume is to address the question of income inequalities from three different angles: first, to identify what elements in the world of work may have contributed to increase inequalities, especially because recent reports have recognized that the roots of inequalities may well lie in the different labour market and industrial relations systems (OECD 2015). The second characteristic and value added of this chapter is to tackle inequalities from the perspective of all the different income groups (not only from those at the two extremes, the bottom and the top) and in particular from the ­perspective of middle-­income groups or what we generally categorize as the ‘middle class’ (which after all is the result of those going down and those going up). The third objective or value added of this work is to examine the ­evolution of such labour developments and their interaction with different income groups in the long term, over the past 20 years (and for some chapters even earlier); this long-­term perspective is reflected in the different chapters of this volume. At the same time, periods of crisis, with the radical changes they sometimes bring, can highlight the relationships we attempt to identify between the world of work and different income groups. We 1

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 1

28/09/2016 08:24

2

Europe’s disappearing middle class?

thus tried also to distinguish the particular effects of the ­financial and economic crisis and whether it may have strengthened or, on the contrary, countered some existing long-­term trends. This comprehensive and long-­term overview is important in enabling us to better identify whether some categories of people may have less access to social mobility and, for instance, lead to possible intergenerational or gender gaps. This overview is also relevant on the economic front, given the importance of the middle class for consumption and economic growth ­(Dabla-­Norris et al. 2015). Approaching inequalities from these two different but complementary perspectives – the world of work and middle-­income groups – also allows us to respond to such questions as: are increasing inequalities and increased polarization combined with a sort of erosion of the middle class? While the position of the bottom groups seems to have deteriorated over the past decade, especially since the financial and economic crisis of 2007–08, has this movement also affected middle-­income groups? If so, what are the driving forces in the world of work that may have contributed to such an erosion of bottom-­and middle-­income categories? What are the policies and institutions in the world of work that will be able to limit and even to reduce inequalities in the future? To answer these questions, a joint International Labour Organization (ILO) and European Commission project established a network of high-­level experts in 15 countries to explore in detail all the developments that may have characterized jobs and activities, the participation of different groups in the labour market, their employment conditions in terms of types of work contract, their wages and pay packages and systems, their ability to negotiate wages and working conditions through collective bargaining mechanisms and more generally within industrial relations systems and the specific roles of key actors such as employers, trade unions and national authorities. We have tried to follow the same structure throughout this volume. We first analyse in each chapter the trends pertaining to the ‘middle class’, looking at national developments over time and in the most recent period; we then identify the most relevant changes in the world of work that may ­contribute to explaining the trends in the middle class; finally, we try to capture – when possible – some of the effects in both the economic and the social domains of this evolving relationship between the world of work and the role and place of the middle class. In order to do this comprehensively, we have tried to provide all available data at national level, complemented in this introductory chapter with some cross-­country analysis using the database of the European Union Survey on Income and Living Conditions (EU-­SILC). While presenting the main changes and developments in both the middle class and the world of work in his or her country in detail, each expert

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 2

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­3

­ evelops the elements that he or she believes are the most relevant for d explaining the national story of the link between the world of work and income groups. The purpose of this introductory chapter is to present the overall framework for this work in terms of themes and methodology, then to summarize the main findings and policy issues presented in the different chapters and to c­ omplement them with some comparative data analysis at European level. This allows us to address policy issues that are also d ­ iscussed in ­individual chapters.

2.  OVERARCHING FRAMEWORK 2.1 The Existing Literature and Current Debates: Why Do We Need to Tackle Inequalities also from a Middle-­class Perspective? There have recently been a large number of studies on inequalities, especially from the OECD, the International Monetary Fund (IMF) and the European Commission (Dabla-­Norris et al. 2015; EC 2015; OECD 2015). This has helped to put increasing inequalities at the core of policy debates and has led to a number of policy reforms to start to reduce them. At the same time, these debates, as well as the existing literature on distribution, focus generally on inequalities between the top or the bottom, while the middle is often ‘forgotten’ (Atkinson and Brandolini 2013). However, the success and well-­being of the middle class is of crucial importance for whole societies as ‘it contributes to economic growth, as well as to social and political stability’ (Pressman 2007: 181). A large and healthy middle class provides a large labour force, increases consumption and serves as a sort of ‘buffer’ between the two extremes, mitigating class struggle and benefiting democracy (Pressman 2007; Foster and Wolfson 2010). A number of studies and articles in the press in Europe and North America have highlighted the importance of maintaining a strong middle class.1 For the OECD (2015: 26), ‘there is consistent e­vidence that the long-­term rise in inequalities of disposable income observed in most OECD countries has indeed put a brake on long-­term growth’. For the IMF (Dabla-­Norris et al. 2015: 4), ‘the income shares of the poor and the middle class are the main engines of growth’ and ‘stagnant incomes of the poor and the middle class have a causal effect on crises, and thus directly hurt short-­and long-­term growth’ (Dabla-­Norris et al. 2015: 8). According to Marianne Thyssen of the European Commission (Commissioner of DG Employment, Social Affairs and Inclusion), ‘Issues concerning poverty, inequality, and the weakening of the middle class have come to the forefront of the policy debate. It is now recognized that

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 3

28/09/2016 08:24

4

Europe’s disappearing middle class?

excessive inequality in a society reduces its growth potential’.2 The middle class, which takes nearly all its income from work – while the bottom may also rely on social assistance and the top also benefit from financial market revenues – also contributes substantially to state expenditure in terms of taxes and ­contributions and is thus also a key element in political stability. Inequalities between the top and the bottom seem to have been ­accompanied by a squeeze in the middle. An increasing number of authors argue that in recent decades the distribution has polarized, resulting in a declining middle class (for example, Foster and Wolfson 2010; Pressman 2007). At the same time, the study of either the top or the bottom, or both, does not provide the whole picture of what is going on in terms of distribution. Inequality measured by this top–bottom approach3 implies that a simultaneous regressive transfer from the middle to the top and an equally proportional progressive transfer from the middle to the bottom leave inequality unchanged, while it does not take into account the deterioration of the middle class position. In order to better grasp the distribution of wealth in a society it is thus essential to identify the shift of individuals and/or households across different income groups – for all income groups – rather than look only at the size of the respective groups that can be the result of many possible different movements. In this volume we analyse the shift from one income group to the other. The increase in the bottom class – except for new entrants in the labour market – is explained by a shift downward of middle-­class households, while the increase in the upper group is often the result of the opposite movement, of middle-­class households that have managed to move upward. Similarly, a significant decline of the bottom class is explained by households’ capacity to improve their income to join the middle class. For all these reasons it is important to look at what happens at the middle of the income distribution and its relationship to the rest of society. As such, it complements other studies that focus specifically on low incomes (for example, Corlett and Whitakker 2014; Kumar et al. 2014) and top incomes (for example, Alvaredo et al. 2013) and by doing so c­ ontributes to a stronger focus on the effects of work and employment trends – especially considering the strong interaction between the middle class and the world of work. 2.2 How to Define and Capture the Middle Class? Combining Four Different Approaches A key issue in the literature on the middle class is how to define it; any chosen definition will be somewhat arbitrary (Pressman 2007). The results on the long-­term evolution of the middle class will also probably depend on the definition that is adopted. Most often, some type of ‘objective’

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 4

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­5

measure is taken to classify the population. This is why most studies on the evolution of the middle class typically approach it from an economic perspective, with classes being defined based on income, asset holdings, expenditures, occupation and so on. Subjective measures, however, can reveal some interesting findings about class perception; for instance, how people experience their own place in society; whether they consider they belong to the middle class, or rather to the bottom or top of society. Four main approaches can be found in the literature, which we decided to combine in this volume. 2.2.1  The income approach: the dominant criteria The most common way of defining the middle class is to use income classes and thresholds to identify the ‘middle’ in the income scale. The middle class is often considered to be either the middle-­income deciles or a certain income space around the median. The middle class is generally defined as the income share going to the middle of the income distribution, those in deciles 3 to 8, as these represent the 60 per cent of the population that have an intermediate level of income (that is, they do not belong to either the 20 per cent with the lowest or the 20 per cent with the highest income level). Similarly, the OECD (2008) defines the ‘middle class’ as the distribution of real income in the middle three quintiles. The German Institute for Economic Research defines the middle class as households with an equivalized disposable net income of between 70 and 150 per cent of the median income. Other approaches fix the middle class at higher levels of the income distribution (Atkinson and Brandolini 2011). There are thus several possible definitions of the middle class using the income approach. Our group of experts took the decision to adopt a middle-­class definition based on the median equivalized household disposable income: ‘median’ because it allows us to examine the possible shift of the middle class;4 ‘household’ in order to take into account the ascent of households with more than one earner in the labour market and because it seems to be increasingly difficult for a household to belong to the middle class with only one income; ‘equivalized’ to take into account household composition because belonging to the middle class can clearly be influenced by the number of children or other dependents in the family; and, finally, ‘disposable income’ to take into account the role of income redistribution. We use a five-­group classification of the entire distribution and capture the middle class in terms of three middle-­income groups (lower middle class, core middle class and upper middle class), complemented by the two other extremes, namely, the lower-­income group and the upper-­income group, to be able to compare trends in the middle class with those in the top and the bottom.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 5

28/09/2016 08:24

6

Europe’s disappearing middle class?

Table 1.1  Selected range of income groups (percentage) Bottom Percentage of median income

Up to 60

Lower middle Core middle Upper middle class class class 60–80

80–120

120–200

Top Above 200

It was important to select the income categories that could best ­represent the middle class and serve as a basis for measuring the – evolving – gap between the middle class and the bottom and the top. Our experts’ group decided to retain 60 per cent of median income as the threshold below which to place the bottom income group, in order to match the low pay threshold used by the European Union (fixed at two-­thirds of median income by the Council of Europe). The top would represent those above 200 per cent of median income, while the middle class would represent the income ranges between those two extremes, that is, between 60 and 200 per cent of median income. We would also distinguish three categories within middle incomes, namely, the lower middle class (60–80 per cent of median income), the core middle class (80–120 per cent) and the upper middle class (120–200 per cent of median income). The top of the distribution begins at 200 per cent of median income, which corresponds to approximately the top 5 per cent wealthiest households. (See Table 1.1.) This approach allows us not only to capture the trends of the middle-­ income groups as such, but also to identify the shifts from one group to another, especially between the three possible categories of the middle class. The comparative analysis provided in this chapter is based on the SILC database available from 2004 to 2011 for 28 countries, including the 15 countries in our project. Some chapters complement this analysis of middle-­income groups as a percentage of median income by a deciles approach, measuring the middle six, four or two deciles of the income distribution. Given the difficulty of obtaining continuous time series with data for each year, we decided to track long-­term evolution through cut-­off points in time. The same classification based on five income groups is used throughout the volume. 2.2.2  The augmented approach: taking on board wealth and capital For a better grasp of whether the situation of the middle class has progressed or not, the evolution of their incomes is important, but may not reveal the whole story. It is important also to measure the wealth/capital of the middle class (home ownership, savings, investments, indebtedness

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 6

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­7

and so on) and its evolution over time. Assets, savings and debt are important contributors to the economic perspective on class divisions as well. Owning or renting a good property or owning a small business, for example, have historically always been identified with middle-­class well-­being (Beckett 2010). Households faced by an income loss but who can smooth out their consumption by relying on savings and assets, loans or the financial help of others could still be regarded as middle class although their income level may suggest otherwise. By contrast, households that would be considered middle class based on their income may be positioned in the lower classes when all financial resources are accounted for. This shows the value of the augmented approach when combining the incomes approach with information on wealth. As noted in the Belgian chapter on the basis of Scott and Pressman (2011: 333), ‘it is now quite common for middle class households to rely on credit to help make ends meet’, which might be a sign of a deterioration of their status. On the other hand, access to credit continues to be one advantage of the middle class. National stories in this volume highlight how a facilitated or restricted access to property and housing may change the position of middle-­class categories, as shown in the chapters on the United Kingdom and Ireland. The experts in this volume thus complement the income approach by providing information on wealth because income inequality and income distribution are different from wealth inequality and wealth distribution (Nolan et al. 2014). 2.2.3  Occupational approach: following professional trajectories A third dimension addresses the situation of the middle class in terms of professions and categories of jobs generally considered to be characteristically ‘middle class’. A number of authors have criticized the definition of social classes exclusively based on income. In particular, Goldthorpe (for example, 2000, 2010) argues that it is also crucial to look at other elements such as economic security, economic stability (in the short run) and economic prospects (for the long term) that are typically all related to the type of job – its human capital requirements, autonomy over the working process and so on. The middle class can thus also be defined in terms of individuals’ different opportunities for development and social status, as shown by their occupation in the labour market. In this respect, the changing structure of jobs and occupations may have affected the probability of whether various types of employees working in different sectors belong to the middle class; this is why national chapters also try to take into account this occupational approach, by providing evidence on occupations, using for instance International Standard Classification of

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 7

28/09/2016 08:24

8

Europe’s disappearing middle class?

Occupations (ISCO) categories or nationally specific classifications, as in France. A number of national chapters also study individual professions or occupations through qualitative case studies. The occupational approach is also useful for studying mobility within the middle class, thus well complementing the income-­based approach. We should emphasize, however, that the occupational approach is based on an individual perspective, while the income approach generally takes households as the reference, a difference that often explains the different outcomes between the two approaches to the middle class and its evolution, as emphasized in the chapter on France. The case studies address the occupational dimension, especially to monitor what has been happening to certain occupations traditionally considered to belong to the middle class, such as teachers, doctors and public administration employees. In particular, we tried to provide evidence on the phenomenon of ‘job polarization’ which is described in the literature (Goos et al. 2009, 2014), referring to the fact that typical middle-­class jobs are vanishing, while high-­skilled and low-­skilled employment are increasing. This is a consequence of both computerization and offshoring to low-­wage countries, as described, for instance, in the chapter on Belgium. 2.2.4  Perceptual approach: capturing ‘relative’ position class individuals/­ Finally, presentation of the perceptions of middle-­ households complements the three dimensions discussed so far. People typically care a lot about how their own well-­being is positioned relative to that of others and attaining a middle-­class living standard is often considered to be a major accomplishment that is not easily given up (Pressman 2007). It was thus important in this volume to report also on perceptions of individual households concerning their capacity to retain or attain middle-­class status. We can see, for instance, in the chapter on Italy how middle-­class perceptions of growing income insecurity have increased over time, especially among younger generations, compared with results from similar surveys carried out among previous generations. Surveys thus help to capture intergenerational aspects related to the perception of middle class membership and the contributing factors in the world of work, such as later entry into the labour market, problems of qualifications or of the decreasing value of a higher education. The chapter on Portugal also highlights the important aspirations of the middle class and how they seem to have been depressed by the recent crisis. Other survey results among the middle class are presented in this volume, for instance in Greece and Germany (see also CREDOC 2011).

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 8

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­9

3. INEQUALITIES: WHAT IS HAPPENING TO MIDDLE-­INCOME GROUPS? Because the EU-­SILC database provides data from 2004 to 2011 it allows us to monitor the evolution of the middle class before the crisis of ­2007–08, in years representing a period of economic growth between 2004 and 2006 (Figure 1.1) and also the changes for the period covering the financial crisis, and in the years after the crisis, between 2008 and 2011 (Figure 1.2). 3.1 Before the Crisis, a Slow Erosion of the Middle Class alongside Increased Polarization In the first period – that is, before the financial crisis – ten countries experienced an increase in the size of their core middle class (see Figure 1.2), especially among some relatively recent European Union (EU) members from Central and Eastern Europe, such as Lithuania, Estonia, Hungary 2004–06 Lithuania Estonia Poland Spain France Belgium Hungary Czech Rep. Cyprus Sweden Italy Iceland Slovenia Austria Ireland Portugal Norway Europe Slovakia UK Latvia Denmark Luxembourg Greece Netherlands Germany –4.00%

Lower income class Lower middle-income class Core middle-income class Upper middle-income class Upper income class

–2.00%

0.00%

2.00%

4.00%

Figure 1.1 Changes (annualized percentage) in the size of the classes, 2004–06

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 9

28/09/2016 08:24

10

Europe’s disappearing middle class? 2008–11

Norway Italy Poland Denmark Austria Belgium Portugal Lithuania Ireland Iceland Hungary Luxembourg Latvia Germany Netherlands Slovakia Europe Sweden Czech Rep. France UK Greece Slovenia Cyprus Spain Estonia –4.00%

Lower income class Lower middle-income class Core middle-income class Upper middle-income class Upper income class

–2.00%

0.00%

2.00%

4.00%

Note:  We define here the different classes according to households’ position with regard to the median of the net equivalized income, taking all households with at least one person of working age. We also included in the sample households with one working adult even if above working age (pensioner working).

Figure 1.2  Changes in the size of the classes, 2008–11 and the Czech Republic and also some older members of the EU, such as Spain, France and Belgium. Three countries experienced a stable core middle class, while 13 countries had experienced an erosion of their core middle class between 2004 and 2006. The most significant erosion was in Germany, the Netherlands, Greece, Luxembourg, Denmark, Latvia and the United Kingdom. While this was accompanied by a growth of the size of the bottom group, it also led to an increase of the upper group, showing increased polarization. A typology can be drawn according to the patterns of the different income groups’ evolution. An increase in the size of the middle class seems to be correlated with a decrease in polarization for countries such as Lithuania, Estonia, Poland,

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 10

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­11

Table 1.2 Typology of scenarios in the relationship between polarization and the middle class Scenario

LC

LMC CMC UMC UC Example countries

Catching up

m

m

mk

k

k

Deterioration

k

k

m

m

m

Polarization

k

k

m

m

k

Decreased  inequalities

m

k

k

k

m

Mixed: catching  up and depolarization

m

k

k

m

m

Austria, Ireland, Portugal, Slovakia, Latvia (2004–06); Cyprus, Latvia (2008–11) Sweden (2004–06); Greece, Italy, Poland, Lithuania (2008–11) Europe, Germany, Greece, UK, Netherlands (2004–06); Spain, Slovenia, Estonia, Czech Rep. (2008–11) Lithuania, Estonia, Poland, Spain, France, Belgium (2004– 06); Norway, Luxembourg. (2008–11) Italy, Norway (2004–06), Ireland, Hungary (2008–11)

Spain, France, Hungary and Cyprus, with a decline in the size of both the richest and the poorest classes (a case of ‘decreased inequalities’ in the typology of Table 1.2). Polarization and slight erosion of the core middle class also characterized the average for Europe in Figure 1.1. By contrast, a decrease in the size of the middle class seems to be correlated with an increased ‘polarization’ for countries such as Germany, the Netherlands, Greece, Luxembourg and the United Kingdom. We can observe that for these countries the decrease in the middle class is associated with an increase in the size of the bottom and upper classes. The chapter on the United Kingdom confirms how high inequality appeared well before the crisis, with implications for the middle class. There does seem to be a direct relationship between polarization and the middle class, the middle class inversely reflecting the evolution of inequalities. Figure 1.3 clearly shows a direct correlation (as shown by the trend line and by an R2 of 0.70 for the period 2004–06) between the trends in inequalities and the size of the middle class. On this basis, we can predict that the recent increase in inequalities in developed countries – but also ­elsewhere – described recently by several international organizations (OECD, IMF, ILO) must have been accompanied by an erosion of the middle class, although to date this aspect has remained relatively unexplored, with the debate on inequalities focusing more on the lowest and the highest income groups.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 11

28/09/2016 08:24

12

Europe’s disappearing middle class? a) Changes between 2004 and 2006 (R2 = 0.70) 0.8

Annual change in inequality (P90/P10) ratio

0.6 0.4

DE EL NO NL LV UK DK Europe PT AT SE LU CY BE SI SK IS CZ IE FR ES IT HU

0.2 0 –0.2 –0.4

LT

–0.6 PL

–0.8 –1.0

EE

–1.2 0 2 3 –4 –3 –2 –1 1 Annual change in the size of the middle class (%) b) Changes between 2008 and 2011 (R2 = 0.48)

Annual change in inequality (P90/P10) ratio

0.5 0.4

ES

0.3 EE

0.2

LT

SI

0.1

EL CY

0

UK

–0.1

IT

IS AT FR SK PL LV BE CZ HU Europe DE IE PT SE LU NL

–0.2

DK

–0.3 –2

–1.5

–1

–0.5

0

NO

0.5

1

Annual change in the size of the middle class (%)

Figure 1.3 Changes in inequalities (P90/P10 ratio) and changes in size of the middle class, 2004–2011 3.2  Trends in the Middle Class Influenced by its Initial Size Another clear relationship emerges from the data, between the decline in the middle class and its original size. The size of the core middle class in Europe varies considerably from country to country, as shown in Figure 1.4. In 2006, the size of the core

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 12

28/09/2016 08:24



29

29

35

36

36

37

37

38

42

44

45

16

15

17

16

18

16

15

15

13

14

13

16

13

14

12

12

1

11

1

9

France

Czech Rep.

Austria

Slovakia

Iceland

Slovenia

Norway

Denmark

Sweden

26

27

30

28

26

29

28

27

28

25

28

28

29

27

27

27

28

28

28

28

29

30

30

31

32

33

34

35

16

14

12

15

14

14

16

15

15

16

15

18

19

17

22

16

20

22

19

19

21

18

20

18

17

17

16

11

12

12

10

27

26

29

29

23

25

25

26

13

12

15

13

22

22

20

Greece

27

15

Romania

25

Cyprus

28

9

Netherlands

28

8

Europe

29

9

Hungary

28

9

Luxembourg

28

9

Spain

28

9

Germany

29

9

Italy

2

8

Poland

4

12

UK

5

9

Ireland

6

10

Estonia

6

11

Lithuania

6

12

Portugal

6

12

Bulgaria

7

16

15

Latvia

5

Belgium

Is the world of work behind middle-­class erosion? ­13

Lower-income class

Lower middle-income class

Middle class

Upper middle-income class

Upper class

Figure 1.4  Size of income classes, Europe, 2006 (percentage) middle class across Europe ranged from 23 per cent in Latvia to 45 per cent in Sweden. The map of Europe also shows a more substantial middle class in the Nordic countries – Denmark, Iceland, Norway and Sweden – and a smaller middle class in both the south (Greece, Portugal) and in central and eastern Europe (Estonia, Latvia, Romania). These distinctions are consistent with the findings of the Centre de recherche pour l’étude et l’observation des conditions de vie (CREDOC) (2011). We can see that countries with a smaller middle class before the crisis, such as Lithuania, Spain, Poland and Estonia, are precisely those that experienced the greatest increase in their middle class (countries having experienced an increase in their middle class are on the right of the ­vertical line), while countries with a stronger middle class, such as Germany, Denmark and Norway, experienced a decline in the size of their middle class during that period (Figure 1.5). There was thus a sort of catching-­up process in the years before the crisis between those initially with a small middle class and those that had already achieved a rather large middle class. Interestingly, this process is not visible after the crisis with, first, a progression of the middle class which is very limited in percentage and in number of countries. The crisis thus seems to have stopped the ­catching-­up process of countries with a small middle class. By contrast, the middle class slightly increased in countries such as Denmark and Norway. There is thus a sort of upper social mobility or ‘catching-­up’ process: countries with a smaller middle class – generally less developed countries in transition – have experienced a rapid economic growth that has allowed more people to shift up to the middle group and also to the upper-­income

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 13

28/09/2016 08:24

14

Europe’s disappearing middle class?

Initial size of the middle class (%), 2004

a) Changes between 2004 and 2006 50 DK

45

SE NO

40 SK

35

NL

DE

LU

Europe

30

UK EL

25 20 –4

IS SI AT

CZ CY BE FR HU

IE PT IT

ES

LV

–3

–2

–1

0

PL

1

LT EE

2

3

Annual change in size of the middle class (%) b) Changes between 2008 and 2011

Initial size of the middle class (%), 2008

50 45

CZ

40 SI

35 30

CY

SK FR

AT HU NL LU

EL

IT IE PL

UK DE LV

20 –2

–1.5

–1

NO

BE

IS

Europe EE

ES EE

25

DK

SE

–0.5

LT

0

0.5

1

1.5

2

Annual change in size of the middle class (%)

Figure 1.5 Initial size of the middle class (level at the beginning of the period) and the annual change in size during the period ranges. We will see later the driving forces behind such a development. Rapidly developing economies, especially in their first stage of development, can also experience a rapid growth of inequality and therefore a further erosion of their middle class along with economic growth. Another scenario is when there is a general deterioration, with a decrease of both the middle and the upper classes, which shift down towards lower-­income groups (that,

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 14

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­15

as a result, increase in size); this scenario occurs in cases of deep and prolonged recession (we will see a few examples of it in the post-­crisis period). There can also be other combinations. Table 1.2 summarizes this typology. 3.3 Effects of the Crisis: A Reduction of the Middle Class (Size and Share in Total Income) The crisis has halted or accelerated a number of trends concerning income groups. While Figure 1.5 shows a number of European countries with an increasing core middle class before the crisis, this scenario is no longer present after the crisis. The difficulties of exiting from the economic slowdown and some adverse labour market developments (lower participation, increased unemployment) have affected middle-­income groups. There is thus a halt in the progress of the middle class in most countries. Figure 1.2 shows 19 countries that experienced a decline in the size of their core middle class in 2008–11, while it remained stable in two countries and increased slightly in only five countries. Even countries where the middle class had rapidly grown before the crisis have experienced a rapid fall, such as Estonia, Spain and Cyprus. Other European countries continued to experience the erosion of their middle class, such as Greece, the United Kingdom and Germany. Countries that were still registering an increase in their core middle class in 2004–06, such as the Czech Republic, France and Sweden, experienced a reverse trend in 2008–11. Figure 1.6 presents an overview of national trends in the size of the middle class and also in the middle-­class share in total household income. It first shows that the number of countries that experienced a decrease in both indicators were much more numerous in the 2008–11 period than in 2004– 06. This was typically the case of the United Kingdom, Spain, Lithuania and Estonia, but also Slovenia and Sweden. In fact, almost all countries have shifted towards the bottom left quarter in Figure 1.6, showing that they have experienced a decrease in both the size of their middle class and the share of total income going to the middle class. Moreover, even when there was a progression it was very small, as in Romania, Portugal, the Czech Republic and the Netherlands. The chapter on the United Kingdom confirms that the distribution of households moved to the left with an increased concentration around the lower part of the core middle-­income band with a majority of households being centred around 70–80 per cent of median household income. The above results are confirmed by a recent OECD (2015) conclusion, that ‘uncertainty and fears of social decline and exclusion have reached the middle class in many societies’. It is thus important to identify the drivers of such adverse developments.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 15

28/09/2016 08:24

16

Europe’s disappearing middle class?

Annual change in the income share of the middle class (%)

4

a) Changes between 2004 and 2006 (R2 = 0.85)

3

EE LT

2

HU

1

PL

CZ FR Europe SE BE IT LV UK AT SI SK IS CY NO LU IE PT DK

0 –1

ES

EL NL

–2 –3 –4 –5 –4

DE

–3

–2

–1

0

1

2

3

Annual change in the size of the middle class (%)

0.6

b) Changes between 2008 and 2011 (R2 = 0.73)

Annual change in the income share of the middle class (%)

0.4 0.2

NL HU PT DE

0 SE

–0.2 EL

CZ

–0.6 –0.8

SI

–1

IT PL AT

DK

LU

Europe

–0.4

IE LT

SK LV

IS

FR UK

–1.2 –1.4 –2

NO

BE

EE ES

–1.5

CY

–1

–0.5

0

0.5

1

Annual change in size of the middle class (%)

Figure 1.6 Correlation between change in middle class size and change in its household income share, 2004–11

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 16

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­17

4. EXPLANATORY FACTORS IN THE WORLD OF WORK In this section we explore the specific effects that trends in the world of work may have had on inequalities and middle-­income groups. Table 1A.1 in the appendix to this chapter summarizes some of the main findings in the national chapters. We distinguished between two periods – before and after the crisis – when relevant and when the effects were distinct and sometimes even going in the opposite direction. We complement this overview with some analysis based on the EU-­SILC database, which also provides a few variables of the world of work, such as belonging to the public or private sector, employment status and different skill categories. In the course of this comparative analysis, we remain conscious of its limitations owing to the difficulty of isolating the relative impact of elements of the world of work. Within the world of work itself we also distinguish as far as possible structural changes, such as in labour participation (for instance, of women) and the generalization of the dual-­earner model, and changes in the occupational structure, from institutions/policies (­educational upgrading, health, social protection, taxes, social dialogue/ unionization and so on), and from the distinctive influence of the economic and social context (globalization, recession and crises). Despite such complexity, the data we present confirm that elements in the world of work are likely to have played a key role in middle-­class growth and erosion. 4.1  Evolving Labour Market Participation Labour market participation has increased over time and represents an important factor in the middle-­class increase. In particular, higher labour market participation among women from the 1980s and 1990s has totally changed the number of working adults in households and thus has also increased the probability of belonging to the middle class. For instance, in Spain, the female participation rate grew from 29 per cent in 1986 to 53 per cent in 2014 alongside fairly regular growth. Figure 1.7 shows that in all countries the more working adults there are in the household the higher will be the opportunities for this household to belong to the middle class and even the upper class. This is particularly strong in countries such as Belgium and Ireland. A number of chapters confirm that the development of a dual and even multiple earners household model helped to increase the middle class, as in Portugal, Spain, the Netherlands and many others. This is confirmed by Figure 1.8, which presents the relationship between the average number of working adults in the household and the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 17

28/09/2016 08:24

18

Europe’s disappearing middle class?

120% 100% 80% 60% 40%

Lower income class

Lower middle-income class

Upper middle-income class

Upper income class

Sweden

Norway

Iceland

Netherlands

Germany

Denmark

UK

Austria

France

Cyprus

Slovenia

Europe

Estonia

Belgium

Italy

Czech Rep.

Portugal

Latvia

Luxembourg

Spain

Lithuania

Poland

Slovakia

Greece

0%

Hungary

20%

Core middle-income class

Figure 1.7 Average share of working adults in households, by income classes, Europe, 2011

0.50

Size of the middle class

0.45 0.40

DK NO

CZ

0.35

PL LU

0.25

EL

AT

BE

HU

0.30

SE

SI

SK

IT PT ES LT LV

CY Europe

FR

EE

IS NL

DE UK

0.20 0.15 0.10 0.50

0.55

0.60

0.65

0.70

0.75

0.80

0.85

0.90

0.95

Average share of working adults in the HH among MC

Figure 1.8 Size of middle class and average share of working adults in households among middle-class households (HH), Europe, 2011

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 18

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­19

size of the middle class.5 The countries with a higher share of working adults in the household – and thus with higher participation of women – are also those with a larger middle class (here we regrouped the three middle-­class ­subgroups) and the upper class. Both indicators are very high in Scandinavian countries (Sweden, Norway and Denmark), but also in the Netherlands, Iceland, France and Germany. Female employment rates increased steadily in those countries but also in Belgium, the United Kingdom and many others. Both indicators are lower in southern European countries, such as Greece, Spain, Italy and Portugal, but also in some central and eastern European countries, such as Poland, Lithuania and Latvia. Ireland is similar. This might be explained by an increased participation of women in the labour market that occurred later – and thus remains below average – than in Scandinavian countries, for instance. However, even in these countries upper-­class membership is positively related to the number of working adults in the household. The relatively late increase in the labour market participation of women in Greece, for instance (from 40 per cent in 1993 to 55 per cent in 2008), helped to increase household living standards and enabled them to shift up the income ladder. This also contributed to maintaining low unemployment rates in Greece prior to the 2008 financial crisis. One result that appears rather clearly in some of the chapters in this volume – for instance, in the chapters on the Netherlands and Italy – is that today there is a need for two working adults in a family to ensure that it belongs to the middle class, while in the 1990s a good income on the part of the breadwinner was often sufficient. On the other hand, the rise of dual-­earner households has enabled them to acquire more income and to attain or sustain middle class status. Similarly, the context of unemployment may influence the probability of belonging to or remaining in the middle class. Different chapters in this volume – notably Italy and Greece – show how very high youth unemployment rates could explain a lower probability of being part of the middle class in the future and would thus lead to a sort of intergenerational gap. By contrast, lower youth unemployment rates in Belgium or Germany – made possible in Germany thanks to the use of vocational training and short-­time working schemes – could help younger generations to achieve or maintain middle-­class status. The situation of older workers also has an impact. Employment rates for older workers (55–64 years) have increased during the past decade, as in Sweden. The postponement of exit at the end of one’s career entails a lengthening of working life and provides a means of sustaining ­middle-­class status.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 19

28/09/2016 08:24

20

Europe’s disappearing middle class?

4.2  A Change in the Structure of Jobs Since income groups can be characterized by specific types of jobs, the middle class can also be defined according to professions and types of jobs. For instance, overall, the share of professionals, managers and associate professionals is much higher among the upper-­income class and still high among the upper middle-­income class and – depending on the country – among the core middle class. Conversely, elementary occupations tend to be clustered among the lower-­income households. Beyond labour participation, the shift in the structure of employment that has taken place over recent decades also had an impact on the middle class. In particular, while the years of growth led to an increase in middle-­income jobs – notably in the public sector and in services – a sort of employment polarization is observed in many European countries. This is due to a falling demand for jobs with intermediate skills, whether owing to routinization of job tasks and/or the skill upgrading bias of technical change. This declining share of middle-­ skilled occupations relative to low-­and high-­skilled occupations is described in the literature for the United States (Autor et al. 2007) and Europe (Goos et al. 2009) and is confirmed by national experiences presented in this volume. In Spain, for instance, we observe two employment restructuring movements: first an increase in professionals that led to an increase of the upper-­middle class and even upper class, and at the same time the reduction of skilled manufacturing jobs that led to a substitution of jobs typical of the core middle class by low-­middle class and even lower-­class jobs. Job polarization is also reported in the United Kingdom, with a long-­term decline of manufacturing jobs during the 2000s, with job losses clustered in the middle three quintiles, mainly in the high-­technology and low-­ technology parts of manufacturing. In the middle quintile, services job losses added to the decline in manufacturing. During 2001–10 there was a collapse in employment in mid-­quintile jobs of around 700 000 in a context of overall net employment expansion. This job polarization – which may correspond to a U-­shaped workforce – is identified by Damian Grimshaw and Anthony Rafferty (in this volume) as a major explanation of the deeper stagnation in real earnings among middle-­wage groups of workers compared with other workers. This job polarization has been aggravated since the recent financial and economic crisis. In Spain, an intense destruction of employment in the middle of the jobs distribution is reported. 4.3  New Forms of Work Contract Labour market reforms in the 1990s in Europe led to a progressive growth of non-­standard forms of employment (NSFE) that are defined

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 20

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­21

as covering ‘work that falls outside the scope of a standard employment relationship, which itself is understood as being work that is full-­time, indefinite employment in a subordinate employment relationship’ (ILO 2015: 1). Included under this definition are temporary employment, temporary agency work, ambiguous employment relationships and part-­time employment. Non-­standard forms of employment progressively extended to all income categories and thus started to affect the middle class, too. We add in this section self-­employment as another form of employment that affects middle-­class groups. These forms of employment can have advantages for both employers – in terms of providing more flexibility in hiring and also planning work, especially in seasonal activities – and workers, to accommodate family or education obligations or to serve as a stepping stone towards permanent employment. On the other hand, these forms of employment often lead to lower wage or non-­wage costs (ILO 2015: 3).6 Temporary employment is particularly high in Spain and Poland (25 per cent of waged employees in 2013) but has also been growing slowly in Portugal and the Netherlands (up to 20 per cent), France (16.5 per cent), Slovenia and Sweden (16 per cent), Germany (13.5 per cent) and Italy (to 13.2 per cent) (ILO 2015: 5). This has sometimes led to increasing duality in the labour market. In Belgium, Greece, France, Italy, Sweden and Germany this type of work has had a stronger impact on the lower class, young people, migrants and women. In Belgium, for instance, the increase in the share of temporary contracts remained in the bottom lower-­middle class; temporary employment is especially high among young and migrant workers, a trend that could explain why a significant part of them struggle to attain middle-­class status. In Italy and the United  Kingdom, women seem to face difficulties in moving out of part-­time and non-­standard forms of employment. The rise of temporary employment in Spain – which is one major change of the Spanish labour market in recent decades – has altered one major attribute of the middle class, which was to have a full-­ time job with an open-­ended contract and with good coverage by employment protection legislation. In the early onset of the financial crisis, a last-­in-­first-­out principle prevailed, as firms found it more difficult to fire tenured employees than to stop renewing temporary contracts and, as a result, total temporary contracts in the 17 EU countries of the euro area (EUROSTAT) declined in proportion to total employment from 16.6 per cent in 2007 to 15.4 per cent in 2009. As the crisis deepened, however, the easiest way for companies to adjust to the shock and retain employees was through an increase in non-­ standard and part-­time employment, which impacted all cohorts in the world of work. In France, the share of fixed-­term contracts doubled for the middle class between 1996 and 2011 (from 4 to 9 per cent).

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 21

28/09/2016 08:24

22

Europe’s disappearing middle class?

The growth of such forms of contract – in all sectors, including the public sector – with their greater uncertainty and sometimes employment and income disruption have had an impact on the middle class; 21 per cent of the labour force in Portugal are now on a temporary contract. In Italy, non-­standard employment accounts for almost 40 per cent of total employment (60 per cent among young workers) (OECD 2015) and around a third of total employment in the OECD in 2013 shared roughly equally between temporary jobs, permanent part-­time jobs and self-­employment (OECD 2015: 29). In Spain, in 2014 nearly 20 per cent of the two professions representing the middle class – professionals and ­technicians – were on temporary contracts. Temporary work contracts have brought to the middle class a higher risk of slipping down the social class scale. In fact, studies have shown that non-­standard forms of employment through lower wages and stagnant or falling wages contribute to polarization (ILO 2015: 32). In Greece, Italy, Ireland, Spain and Portugal, youth unemployment has not only increased significantly, but more young people have become exposed to being trapped in non-­standard forms of employment and of experiencing discontinuous careers, low pay and inadequate social protection. The chapter on Sweden shows that low-­skilled young people were also particularly affected there, which further strengthened class polarization during the crisis. In 2012, half of young employees aged 20–24 years were on short-­term contracts, compared with less than 10 per cent among prime age adults. Involuntary part-­time work also increased significantly especially after the crisis. In Greece, part-­time work doubled between 2008 and 2014 (from 4 to 9.3 per cent) and 70 per cent of part-­time employment is involuntary. Similarly, more than 50 per cent of part-­time is involuntary in Bulgaria, Cyprus, Italy, Romania and Spain (ILO 2015: 19). Involuntary part-­ time work has also grown in the United Kingdom and France – where working-­time flexibility represented a practical way for firms to adjust to the crisis. A decrease of involuntary part-­time work was witnessed in Belgium. We should add that temporary employment is also often involuntary: the percentage of workers who reported holding a temporary job because they could not find a permanent job was, for instance, 40 per cent in the Netherlands, 67 per cent in Sweden and over 90 per cent in Greece, Portugal and Spain (OECD 2014). Ireland and the UK also experienced a massive rise in the use of so-­ called ‘zero-­hours contracts’, leading to a decrease in job quality and income insecurity. The chapter on the United Kingdom shows how the standard employment relationship has been eroded and has hollowed out prospects for relatively secure and stable middle-­income occupations. One

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 22

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­23

case study analyses the effects of non-standard forms of employment – such as zero-­hours contracts – for middle-­income households. Conversely, self-­employment is often related to middle-­class well-­being (Goldthorpe 2000, 2010) and in countries such as France, Italy and the United Kingdom specific social measures, such as tax incentives and employment protection legislation, were put in place to encourage the development of self-­ employment. The appearance of a new type of self-­employment (so-­called ‘bogus self-­employment’ or ‘dependemployment’) in recent years – imposed by the employer to ent self-­ replace previous employee status to reduce costs, for instance, social ­contributions – may also bring more risk and vulnerability and thus hinder rather than strengthen their belonging to the middle class. This is the case in Italy where, within the same category of self-­employed, there are both true, high-­income ‘professional’ workers and low-­paid, precarious workers. One of the most striking features of the post-­2010 jobs recovery in the United Kingdom has been the significant growth in self-­employment. The share of workers in self-­employment was stable prior to the crisis (2003–09), but then rapidly accelerated so that by 2014 it accounted for nearly one in six workers (15 per cent). Overall, jobs for the self-­employed accounted for 55 per cent of net job expansion during 2011–14 (540 000 of 990 000 net job growth). The increased incidence of self-­employment was associated with lower median income, tougher welfare entitlement rules, new technologies that enable employers and customers to access ‘crowdsourced’ employment and straightforward income tax advantages. 4.4  Wages and Incomes The evolution of wages is certainly a major factor explaining the long-­ term development of the middle class. While households in the bottom class often depend strongly on social transfers because they do not participate in the labour market and the top largely benefits from financial income, households in the middle class acquire their well-­being based on their wages. The chapter on the Netherlands, for instance, highlights how income from earnings is not only by far the most important category, but also directly responsible for the contrasting evolution of the middle class. While before the crisis most countries were experiencing a growth in median income within each class – and in particular within the ‘middle-­ class’ income groups – this progression was halted by the crisis, with a majority of countries experiencing first a decline in the median income (see Figure 1.9) (which identifies the different income groups with regard to that median) and then a fall in the median income within each class (Figures 1.10a and 1.10b).

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 23

28/09/2016 08:24

24

Estonia

Lithuania

Latvia

Poland

Slovakia

Iceland

Czech Rep.

Ireland

Cyprus

Slovenia

Hungary

Italy

Spain

Sweden

UK

Belgium

Norway

Luxembourg

Greece

Denmark

Portugal

France

2008–11

Austria

Europe

Germany

2004–06

Netherlands

35 30 25 20 15 10 5 0 –5 –10 –15

Europe’s disappearing middle class?

Figure 1.9  Changes in net median income, Europe, 2004–06 and 2008–11

At the same time, the erosion of median incomes occurred mainly after the crisis in countries such as Ireland and Greece and affected mainly low-­ income groups, while in Germany it started before the crisis for low and middle-­income groups (Figure 1.10a). The Netherlands is characterized by a massive long-­term stagnation of real incomes, especially at the bottom, which combined with growth at the top has led to increased inequality. The United Kingdom has been characterized by falling and stagnant pay levels, allied with low productivity growth and the absence of appropriate institutional levers to boost pay and productivity, such as support of trade unions and worker voice in pay determination. This trend started already in 2002 and is the longest period of falling real wages since wage records began in 1964 (Taylor et al. 2014). By contrast the rise in median income before the crisis was high in some central and eastern European countries that experienced rapid growth, such as the three Baltic states, Poland, the Czech Republic and Slovakia. The declining wage share that is affecting many European countries has thus had an impact on the middle class. The wage share decline observed in most European countries (ILO 2014) precisely reflects the increasing gap between rising productivity – to which the middle class has contributed a lot – and wage increases that have failed to catch up. Mechanisms such as collective bargaining but also pay systems – for instance, related to productivity – can only help to restore that link.7 Figure 1.10b confirms that with the crisis, median income declined in almost all countries within all classes, including the middle class, although the lowest income classes were often hurt more. In Figure 1.10b all classes in 15 countries experienced a fall in median income. Middle-­ income groups were thus not protected from this decline; their median income also fell in more than two-­thirds of the European countries under

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 24

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­25

Lithuania

Estonia

Portugal

Norway

Latvia

Poland

Slovakia

Czech Rep.

Iceland

Hungary

Norway

Slovenia

Italy

Cyprus

Sweden

UK

Lower middle-income class Upper income class

Austria

Lower income class Upper middle-income class

Denmark

Spain

Belgium

Portugal

Luxembourg

Greece

Netherlands

Austria

France

Ireland

Europe

a) 2004–06

Germany

30 25 20 15 10 5 0 –5 –10 –15

Core middle class

b) 2008–11

5 0 –5 –10 –15

Lower income class Upper middle-income class

UK

Sweden

Slovakia

Czech Rep.

Greece

Italy

Lower middle-income class Upper income class

Belgium

Netherlands

Denmark

Cyprus

Germany

Ireland

Europe

Spain

France

Finland

Luxembourg

Poland

Slovenia

Hungary

Estonia

Lithuania

Latvia

–25

Iceland

–20

Core middle class

Figure 1.10 Annualized percentage change in (equivalized) median (net) income per adult by class, Europe, 2004–11 examination (Figure  1.10b). The most severe income fall was observed in Iceland (–15  per cent for all classes following the collapse of their economy during the financial crisis), but also in the Baltic states. The high percentage decrease in income can be explained by jobs lost in the crisis, especially among the bottom income groups. But the middle-­ income groups have also been affected. The chapter on the United Kingdom in this volume shows how polarization seems to have been accompanied by a collapse of middle wages and incomes. Moreover, relative to the top and bottom of the wage distribution, middle earners have fared the worst both before and since the crisis. High earners benefited most from employer practices to reward staff with bonuses, as well as other non-­pay rewards, such as stock options. Low earners, on the other hand, benefited from

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 25

28/09/2016 08:24

26

Europe’s disappearing middle class?

the upward pressure at the bottom exerted by the statutory minimum wage introduced in 1999 and especially the period 2003–07, when the Low Pay Commission acted purposefully to raise the relative value of the minimum wage. While Austria and Belgium managed to stabilize median income for nearly all classes (confirmed by the Belgian chapter in this volume), and to a lesser extent France, the Netherlands and Norway (where, however, lower-­income groups seem to be experiencing a decline), other countries – especially from central and eastern Europe, such as the Czech Republic and Poland – seem already to have experienced some catch-­up in median income alongside economic recovery. Income seems to have contributed to general middle class trends. Not only has the size of the middle class been eroded, but the median income around which we define middle class groups has also been declining, which highlights a general loss for middle-­income groups (see Figures 1.10a and 1.10b). This also explains our previous results indicating a declining share in total household income. The outcome of these downward wage trends is not the same in a context of high-­income inequality (as in the United Kingdom) compared with a context of low-­income inequality (as in Belgium). It is also important to take into account the low-­pay threshold and the development of the low-­pay segment that has extended to the middle class. Recent studies have shown that the middle class in Spain is progressively falling into poverty; this involves 1.7 million wage earners, 10 per cent more than in 2012. According to the IFRC (2013: 20), ‘in Romania 20 per cent of the population was classified as middle class in 2008. Today the number is about 10 per cent, the same as in Croatia and Serbia’. Not only low-­ income families but also the middle class have been impoverished in Italy as a result of increased cuts in social investment and social expenditure, privatization and higher co-­payments for social services, as well as (direct and indirect) tax increases. A study by the Bertelsmann Foundation found that in Germany 5.5 million middle class people became low earners between 1997 and 2012, while half a million people joined the high-­earner ranks.8 In Denmark as well, the middle class decreased by 111 000 people between 2002 and 2009. In France we also witness increased polarization. In Portugal, middle-­income brackets were found to lack upward mobility to higher-­income brackets during the crisis. Middle deciles 4, 5, 6 and 7 experienced the greatest increase in immobility between 2008 and 2011: 12.1, 11.1 and 10.3 percentage points for deciles 4, 7 and 6, respectively. This translates into fewer opportunities for career progression among employees belonging to middle-­income groups.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 26

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­27

The middle class in southern European countries may be particularly affected as young people move from the south to countries with better employment and social investment prospects. The most mobile tend to be among the best skilled, leading to a depletion of skills in sending countries and more skills in the receiving countries; this process can be halted only through better social investment and labour market developments in the home countries. Falling wages and incomes have pushed many middle-­class households to take on more debt. In the Netherlands, for instance, as shown by Wiemer Salverda in this volume, households’ average mortgage debt relative to their average net income more than doubled between 1993 and 2013, and explains the lack of growth in household consumption. Moreover, the fact that the dual or multiple earners model has developed shows that while the middle class has remained large and stable, household members have to work hard and get top and multiple earnings if they want to maintain their middle-­class life. This concentration of debt may put such households – even with a high level of assets – at risk if there are sudden changes in asset prices, as happened in the 2007–08 financial and economic crisis. 4.5  Collective Bargaining and Social Dialogue The growth of non-­standard forms of employment has also brought lower coverage of collective bargaining and social dialogue. Temporary workers, agency workers and the self-­employed are usually not covered by collective bargaining, and the crisis further aggravated the coverage rate of collective bargaining. The middle class is traditionally well covered by collective bargaining because of their concentration in sectors and occupations that are generally well unionized – as in education and health. As explained in the chapter on Spain, trust in trade unions is higher among the middle class than among unskilled workers, and the affiliation rate is high among middle-­income employees. This income category also traditionally enjoys collective agreements in the public sector. Table 1A.1 in the appendix shows, however, that collective bargaining has been weakened through a number of reforms included in fiscal consolidation packages: removal of extension mechanisms, derogation at enterprise level from upper-­level agreements, no systematic renewal of enterprise collective agreements, lower application of collective agreements and priority given to enterprise bargaining. Collective bargaining on wages and other working conditions was also weakened in the public sector, thus making way for more radical

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 27

28/09/2016 08:24

28

Europe’s disappearing middle class?

adjustments that certainly had an impact on the middle class. In the United Kingdom, weakly embedded institutions for collective bargaining are likely to have played a role in the decline in real wages. The share of employees whose pay is agreed through negotiations between employer(s) and union(s) fell from 23 per cent in 1995 to just 17 per cent in 2013. In the public sector, too, despite strong joint regulatory mechanisms to set pay, restrictive government rules since 2010 have restrained pay growth in the public sector. Declining union coverage has had a disequalizing effect on the wage distribution, as shown in various chapters and reported by the OECD (2015). The weakening of collective bargaining provides a direct explanation of the wage moderation and falling median income reported in the previous sub-­section. By contrast those countries that managed to stabilize collective bargaining and industrial relations are also the ones that managed to maintain some progression of wages and incomes. For instance Belgium, France and the Netherlands have managed to stabilize their low and middle classes and have maintained strong collective bargaining coverage (Visser et al. 2015). In this volume, the case of Belgium shows that consultation mechanisms have contributed to maintaining wages and working conditions, and therefore represent a major factor behind middle-­class stability there. Industrial relations stability has also played a major role in limiting inequalities and stabilizing the middle class in France, where inequalities seem to have decreased slightly over the most recent period (INSEE 2015). Mechanisms of collective agreement extension have also greatly helped in those two countries. As stressed by Visser et al. (2015: 10), ‘in many cases the underlying reason for the observed stability of collective bargaining coverage is the continued willingness of employers’ organizations and trade unions to negotiate with each other – and to act as social partners in the regulation of labour markets’. This has allowed social partners to set framework conditions in sectoral and/or inter-­sectoral agreements that are relevant for enterprise negotiations, something which is key in a context of decentralization to enterprise-­level bargaining. The Netherlands offers another illustration in this volume of stable social dialogue – with 80 per cent coverage of collective agreements through extension mechanisms – which has enabled distribution and the middle class to remain important, with 80 per cent of Dutch households being members of the middle class (when measured by means of net equivalized incomes). Sectoral bargaining and the administrative extension of collective agreements in the industrial sector have also helped to maintain cohesion among different categories of workers and between sectors of

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 28

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­29

activity. In the Netherlands, collective bargaining since the crisis even expanded into new sectors, while collective agreements were also extended (cleaning, services, security and so on). By contrast, the sharpest decline in bargaining coverage occurred in the group of European countries that suffered the most from the economic implications of the recent crisis, many of which had to seek direct assistance from international financial organizations. The bargaining coverage rate for this group of ten countries (according to the typology of Visser et al. 2015 and including countries studied in this volume, such as Hungary, Greece, Ireland and Portugal) fell by an average of 21 per cent. These are also the countries in which the wages of both low-­income and middle-­income groups fell the most, as explained in individual chapters. The weakening of tripartite consultations in these countries also contributed to this (see appendix). In Romania, Greece and Slovenia, national pay agreements were ended in 2009 and 2010. At the same time, a rapid deterioration of the social climate seems to have been avoided in countries in which the government has managed to organize tripartite consultations, as in Estonia, as distinct from Latvia and Lithuania, and in Ireland over the second period after the crisis – the first period after the crisis saw a weakening of tripartite bargaining – with an agreement concluded in the public sector in 2010. 4.6  The Role of the Public Sector The public sector has traditionally been a haven for middle-­class employment because many of its jobs require a high level of education and skills. It has also represented a fairly secure source of employment and income. The public sector hosts many of the professions traditionally associated with the middle class – school teachers, nurses, doctors and civil servants, for example. Figure 1.11 shows that the proportion of employees in the public sector – proxied here by the sectors of public administration, health and education – per household increases between 2004 and 2006 among the five income groups and is most important among the core and upper middle class. We can see that public sector employment is most common among working-­age households defined as upper-­middle income. The share of public sector jobs among lower-­middle income and lower-­income households is below the average. Nevertheless, the key finding is that public sector jobs are distributed progressively along the income scale, with the lowest proportion at the bottom for those with net incomes below 60 per cent of the median. Any changes in the public sector may thus directly affect the middle class, represented by these middle-­income groups.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 29

28/09/2016 08:24

30

Europe’s disappearing middle class?

0.26

0.29

0.30

0.74

0.71

0.70

0.23 0.77

0.34 0.66

0.21

0.32 0.68

0.79

0.29 0.71

0.26 0.74

0.34 0.66

0.26

0.32 0.68

0.74

0.29 0.71

0.26 0.74

80

0.25

90

0.75

% 100

70 60 50 40 30 20 10 0

LIC LMIC MC UMC UC

LIC LMIC MC UMC UC

LIC LMIC MC UMC UC

2004

2006

2011

Private sector

Public sector

Note:  LIC = lower-income class; LMIC = lower middle-income class; MC = middle class; UMC = upper-middle class; UC = upper class.

Figure 1.11 Proportion of public sector employees by classes, Europe, 2004–11 2004–2006

2006–2011

Germany Denmark Cyprus Estonia Italy Portugal Austria Finland Norway Czech Rep. Spain Iceland Europe Poland Slovakia France Latvia Belgium Greece Sweden UK Slovenia Netherlands Luxembourg Lithuania Hungary

4% 3% 2% 1% 0% –1% –2%

Figure 1.12 Annual changes in public employment among the middle class, selected European countries, 2004–06 and 2006–11 (percentage) A rapid expansion of jobs in the public sector in the past two decades in Europe led to significant job growth in the middle class, as in Greece, Portugal, Spain, the Netherlands and many others. Figure 1.12 shows that public sector employment among the middle class had already fallen in a number of countries before the crisis, for instance in Germany, Denmark, Cyprus, Estonia, Italy, Portugal and Austria. It continued to grow among

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 30

28/09/2016 11:36



Is the world of work behind middle-­class erosion? ­31

the middle class in a majority of countries, especially in Hungary, Lithuania but also Luxembourg, thus confirming the importance of this sector for middle-­income groups. The role of the public sector has been reversed during the crisis and subsequent years. As documented in previous work (Vaughan-­Whitehead 2013, 2015), fuelled by the economic crisis, the decision to target budget deficits has plunged the public sector in Europe into an unprecedented wave of ‘adjustments’, with policy-­makers favouring quantitative adjustments, mainly spending cuts, but also cuts in public sector jobs and wages. Quantitative adjustments have also been accompanied – and often preceded – by more structural adjustments in the composition of employment, mainly due to the recourse to more fixed-­term contracts and the outsourcing of a number of functions previously carried out by public services. The number of temporary contracts in the public sector has increased rapidly throughout Europe. Employment security is thus no longer the norm in the public sector, where an increasing number of employees are hired on temporary work contracts and permanent employees can be laid off. Moreover, the public sector does not seem to represent a lever for employment growth for the middle class as witnessed in the past. Even more extreme, for Rafael Muñoz de Bustillo and José-Ignacio Antón in their chapter, ‘the Spanish public sector, from being an employment generating actor . . . became a major factor in the destruction of middle-­and high-­quality employment’. Women have been particularly affected by this process. The representation of women in the extended definition of the public sector here (public administration, education, health care and social work) in 14 selected EU member states was reported to be between 65 and 76 per cent. The public sector is not only a major source of employment for women but also provides many jobs for highly qualified women. In the context of employment cuts in the public sector, this may lead not only to job losses for women but also to a decline in quality employment for women that cannot be immediately replaced, if at all, by private sector quality employment (Rubery 2013: 44). This process is directly affecting the dual model and the growth of the middle class. The public sector has also experienced significant wage cuts, often due to a weakening of collective bargaining that was traditionally strong in this sector. In the United Kingdom, for instance, despite strong joint regulatory mechanisms to set pay, restrictive government rules since 2010 have restrained pay growth (in the form of a two-­year pay freeze followed by a 1 per cent cap on pay rises until 2020). Another way to gain flexibility and reduce spending is to replace jobs previously carried out in the public sector by new private sector jobs through outsourcing and privatizing public services. Governments’ use

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 31

28/09/2016 08:24

32

Europe’s disappearing middle class?

of outsourcing has also rapidly increased in OECD countries, from 8.7 to 10.3 per cent between 2000 and 2009. This process has been implemented on a large scale in the UK (including job search services and, most recently, the national post office, the Royal Mail) which explains the decrease in the United Kingdom in terms of public sector employment (Figure 1.12). A similar process is occurring in Germany and other European countries. Outright privatization has led to a reduction in the share of government expenditure and public sector employment. Alongside such cuts in expenditure, more difficult career progressions owing to fiscal consolidation programmes may also have an impact. Career development was frozen in Portugal and wage increments frozen in many parts of the UK public sector, where, moreover, public career services were even dismantled. In Greece, too, career progression has become more difficult owing to the obstacles hindering public sector employees from moving up the ranking scale, regardless of performance. Lower career prospects, combined with cuts in wages and benefits, risk affecting public sector employees’ ability to maintain living standards and may contribute to shifting them towards lower-­income groups. The magnitude of current public sector adjustment, together with decreasing public expenditure, notably on training, will lead to significant changes in the skill composition of public sector employees, which may also affect younger graduates’ prospects. All these changes – especially when resulting in a growing mismatch between increasing demand for public services and falling supply – cannot be neutral for the future quality of such services. In this context it has not been possible to achieve the professed aim of most of the reforms, as indicated by the OECD, namely to ‘improve efficiency . . . using fewer resources’ (OECD 2011). The examples so far show that resources have certainly fallen, but that this has also brought a deterioration in performance and service quality. This is already to be observed in education and health care – on such simple indicators as higher ratios of teachers and students in classes and longer hospital waiting lists – but also threatens public administration in a substantial number of European countries (Vaughan-­Whitehead 2013). Social investment has thus been affected. There are other instances of deteriorating public services: cuts in security services leading to increased insecurity; longer judicial delays, along with pay reductions, leading to increased corruption; lack of skills, including information technology (IT), in the public sector owing to reduced investment; and lower services also in health care, including the closure of emergency units. All these trends have also affected the middle class not only as public sector employees, but also as public service users.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 32

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­33

4.7  Intergenerational and Regional Gaps Factors in the world of work can also explain differences in inequalities and middle-­class outcomes between different generations. The currently high rates of youth unemployment, which are particularly dramatic in southern European countries (50 per cent youth unemployment in Spain and Portugal, 60 per cent in Greece), can explain the gap found between job and income prospects of young employees compared with those of their parents, as reported in Italy, but also in other countries in this volume. Labour market difficulties and different ‘world of work’ prospects in certain regions may explain why attaining middle-­class income levels might be more difficult for individual households, as highlighted in the chapters on Belgium, Germany and Italy. Table 1A.1 in the appendix presents the links found between elements in the world of work and the middle class. Throughout this volume national experts complement their information and statistical analysis with qualitative analyses and two case studies for each country – summarized in Table 1.3 – aimed at illustrating and providing further evidence on the national stories. Table 1.3 Case studies in selected European counties (see individual chapters) Country

First case study

Second case study

Baltic states   (Chapter 2)

Migration in Latvia and its effects on the middle class

Belgium   (Chapter 3)

The middle class and the welfare state: the Belgian service voucher system

France   (Chapter 4)

Working time regulations and heterogeneity across social groups and occupations Effects of the minimum wage on middle-­income classes Neutralizing ‘the public sector’, a traditional nurturer of the middle classes

Position of teachers and Estonian middle class The role of immigrants in the relationship between the labour market and the middle class Education and the middle class

Germany   (Chapter 5) Greece   (Chapter 6)

Upgrading of skills did not pay off The emergence of professionals and technicians as middle-­class groups

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 33

28/09/2016 08:24

34

Europe’s disappearing middle class?

Table 1.3 (continued) Country

First case study

Second case study

Hungary   (Chapter 7)

Emerging middle class and age/ education bias in the transition

Ireland   (Chapter 8)

Credit, debt and the housing market effects on income groups

Italy (Chapter 9)

Mobility along the income distribution

Netherlands   (Chapter 10)

Interaction between the world of work and the new dual earners model Structural transformation of employment and the middle classes: polarization or upgrading? The welfare state as ‘middle-­class’ job provider

Intergenerational mobility patterns of the middle class Middle class occupations in Ireland’s boom and bust Belonging to the middle class: intergenerational mobility Wealth and debt of the middle class

Portugal   (Chapter 11) Spain   (Chapter 12) Sweden   (Chapter 13)

Long-­term transformation in the jobs’ structure, and effects on the middle class

United Kingdom   (Chapter 14)

Patterns of pay among mid-­level jobs

The effects of public sector employment adjustments Education and middle-­ class growth at a turning point Changes in the ethnic composition of the labour force and impact on inequality The rise of precarious employment and self-­ employment and effects on income groups

5.  BEYOND THE WORLD OF WORK 5.1  Effects of Taxation Other policies outside the world of work (summarized in Table 1A.2 in the appendix) may also have influenced different income groups. This is typically the case of tax policy, whose purpose is not only to generate revenues for the public budget but also to redistribute – equally or equitably – the burden of such contributions among all income groups. The more

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 34

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­35

Lithuania Denmark Belgium Sweden Slovenia Czech Rep. Austria Norway Hungary Luxembourg Germany Iceland Lithuania Portugal Slovakia Italy Europe Latvia UK Netherlands Greece Poland Estonia France Spain Cyprus –10%

Lower income class Lower middle-income class Core middle class Upper middle-income class Upper income class

–5%

0%

5%

10%

Figure 1.13 Redistribution effect of income tax (difference in respective size of each class between gross and net incomes), European countries, 2004–11 progressive the taxation system, the more equalizing the effect it will have on different income groups, with a relatively higher contribution being asked of higher-­income groups, which should lead to lower inequalities. By contrast, regressive taxation may lead to higher inequalities. Similarly, indirect taxation, such as value added tax on goods, proportionally hurts those on lower incomes more than those with higher incomes. Figure 1.13 confirms the important effects of tax policy, showing how the size of the respective income classes is modified between gross and net income, that is, after the effect of income tax. It clearly highlights that, in general, income tax – which usually increases along the income scale – tends to reduce the extent of upper and upper-­middle class groups and of lower-­income groups, which instead join the lower-­middle class or core middle-­class groups, whose size increases after taxation. An effective tax policy is thus essential to avoid an increase in both extremes and to reduce ­inequalities, while at the same time inflating the number of people belonging to middle-­income groups. This ­balancing effect of reducing polarization is particularly strong (at the top of the inverse pyramid in Figure 1.13) in the countries that rely most on progressive income tax policy, such as the Scandinavian countries (Denmark, Sweden and Norway), but also

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 35

28/09/2016 08:24

36

Europe’s disappearing middle class?

Belgium, the Netherlands, Luxembourg, Germany and Slovenia, where the middle class is also increasing most between gross and net incomes. Income tax policy seems to have played a role – but to a much lesser extent – in countries such as Cyprus, Spain and Greece, with sometimes a negative effect on the size of the core middle-­class group. Even after the crisis, we can observe that income tax had clearly helped to increase the size of the middle class in almost all countries, by allowing many who belonged to the lower-­income group in terms of their gross income to rise to the middle-­class category. Although less distributive than many European countries, the British system of tax and benefits also has a significant impact on household income distribution. Since 2010 cuts in the real value of tax credits and other means-­tested benefits have affected many workers in low-­and middle-­income households reliant on top-­ups to wage income. The declining size of the middle class documented in this volume can thus also be traced back not only to the changes in the world of work, but also to the tax reforms that have been adopted in recent years. In order to increase tax revenues while not overburdening the lowest paid (because the most vulnerable are unable to pay) nor the highest incomes (because of the dogma that they are ‘job creators’) most governments generally decided to tax middle-­range incomes. In Spain, for instance, the reforms in 2011 concentrated the tax burden on the wage-­earning middle class. In Greece, a new series of regressive tax measures adopted in 2011, coupled with drastic increases in indirect taxes, have hit not only middle-­class but also lower-­class incomes the most. Income tax reform in 2013 exacerbated this. The main elements of the reform were: the reduction of income tax rate bands from eight to three; the elimination of selective tax credits; tax allowances for children; the tax-­ free personal income threshold for the self-­employed and professionals; the special tax regimes based on imputed income for farmers and seamen; and the restructuring of the tax regime for corporate profits, resulting in a reduction of the gross tax rate on distributed profits from 40 per cent to 33.4 per cent. This reinforced the tax squeeze on the lower and middle classes. class earners experienced cuts in previous tax In Portugal, middle-­ deductions. Moreover, public servants, a core group of the middle class, have lost disposable income owing to increased targeted taxes. Fiscal ­consolidation policies have also been concentrated on the public sector with – as described above – cuts in wages and pensions and a freeze on promotion in public careers, implemented since 2010. In Italy, tax and pension reforms have hurt both low-­income families and the middle class. In Hungary, the new government has favoured the upper-­middle class and upper-­income earners. The economic reasoning is to boost investment

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 36

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­37

and labour supply at the top. The 16 per cent flat-­rate income tax and the generous child tax-­credit for high-­income families are intended to serve this aim. The chapter on the United Kingdom also shows how tax changes since 2010 have further reduced the redistributional function of tax policy and hurt most low-­and middle-­income households. The middle class has also been hit by property tax changes. In Ireland, for instance, the ‘value based’ property tax hit middle-­class Dublin homeowners. In Italy, the new property tax for primary residents also hurt middle-­class homeowners and affected small businesses. 5.2  Returns to Education The various chapters in this volume highlight the importance of education for moving people into middle-­class or upper-­class groups. Figure 1.14, based on SILC data, shows that those belonging to the upper class and the middle class have a higher proportion of adults with higher education than lower-­income groups. The middle class also has a high proportion of % 100 90

12

13

13

19

51

21

20 10 0

51

23

16 16

10

6

52

13

35

11 10

8

55

5

55 48

23

16 6 2

60

56

21

11 14

37

56

51

36

19

23

60

57 50

16

36

56

56

40

16

21

56 52

55

50

14

33 55

70

30

14

19

32

80

60

13

6 2

12

21

48

33 17 11

9

21

7

4

6 2

11

20

9

32 16 11 6

3

6 1

LIC LMIC MC UMC UC

LIC LMIC MC UMC UC

LIC LMIC MC UMC UC

LIC LMIC MC UMC UC

2004

2006

2008

2011

Lowest

Basic

Secondary school

Higher education

Note:  LIC = lower-income class; LMIC = lower middle-income class; MC = middle class; UMC= upper-middle class; UC = upper class. In the case of Bulgaria and Romania there is no information for 2004.

Figure 1.14 Proportion of employees with primary, secondary and tertiary education, by classes, Europe, 2004–11

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 37

28/09/2016 08:24

38

Europe’s disappearing middle class?

7

16

13

23

26

37

30

20

14

24

14 29

32

58

54

13 1

14 1

19

27

16 32

24

12

19

48 49

56

55

72

47 45

58

64

67

40

74

66

64 64

46

48

82

76

29

Lowest

Basic

Secondary school

0

0

6 0

5 0

Slovakia

1

Denmark

28

Czech Rep.

23

Austria

1

18

Norway

1

9 1

Lithuania

2

UK

2

16 Latvia

5

12 1

Estonia

5

17 Iceland

6

13

Hungary

6

9 3

Slovenia

14

Sweden

8

35

22

Germany

16

5 7

Belgium

15

Europe

16

9

Netherlands

16

33

11

Italy

11

France

Portugal

Luxembourg

20

26

Poland

12

26

27 49

Spain

45

46

28

Greece

26

25

Cyprus

22 22

Higher education

Figure 1.15 Proportion of employees with primary, secondary and tertiary education, among core middle-class, European countries, 2011 (percentage) people with secondary qualifications (more than the upper class), but less than the lower-­income class. Clearly, the higher the level of educational attainment, the higher the income group to which individuals will belong. In the United Kingdom, for instance, among lower-­income households, two-­thirds of reference adults have only secondary school education, while in upper-­income households this is true of fewer than one in five adults. Reference adults with higher education account for a majority in the upper middle-­class and upper-­class households and only around one in five in the lower and lower middle-­income classes. Figure 1.15 shows the same distribution of education level within the core middle class for individual countries. The different endowment can help explain the size of the middle class in the respective countries. Tertiary and secondary education remain lower than average in Portugal, with tertiary education lower in Italy, Romania and the Czech Republic (although this is compensated by a higher proportion of secondary education). The proportion of tertiary and secondary education is, for instance, higher in Finland, Norway, the United Kingdom, Lithuania and Belgium. Education systems have also changed radically in recent decades, with notably an increase of individuals in tertiary education. The percentage of workers in tertiary education, for instance, doubled in Spain in a period of 10–15 years, but this was accompanied by a high rate of early school abandonment – 32 per cent in 2008, with Portugal registering the highest

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 38

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­39

in Europe – thus creating a V-­profile of educational attainment that can only further increase inequalities and exacerbate polarization in the labour market. As in Spain, the growth of tertiary education in Portugal has contributed greatly to developing the middle class thanks also to a move from an ‘elitist’ system’ to wider access to university. However, cuts in health care and education since 2009 have affected this process. In the United Kingdom and Ireland, a series of measures such as the abolition of education maintenance grants and the tripling of university fees have disproportionately affected disadvantaged groups and the middle class. A similar process is reported in Spain with an increase in university tuition and suppression of a number of grants. As a result, Spanish families have increased their share of expenditure on university education by nearly 50 per cent in a few years and especially since 2012–13. University loans (Latvia and Lithuania) and scholarships (Portugal) have been curtailed and school curricula have been narrowed (Hungary). Various chapters in this volume show how this is making it more difficult for young people to enter the labour market, as in Italy, either because of a lack of needed skills and/or because of over-­educated profiles, while the fact that older people tend to remain in employment for longer may exacerbate an intergenerational conflict. Cuts in expenditure on education, combined with lower spending by employers on training, have led to a rapid deterioration of human capital throughout Europe. This process can only further exacerbate the current mismatch that is observed and reported here in many countries between the increased but not always adequate human capital of the growing number of new graduates and the educational requirements of the economy. As a result, and as emphasized in the chapter on Spain, for ‘young workers, having a tertiary education is no longer a passport to a middle-class job with a middle-class wage’. Nevertheless, we can see that the countries with the highest level of education among the middle class also have the highest ­percentage of highly skilled within this middle class, as shown in Figure 1.16, for instance, in Denmark, the Netherlands, Norway, Sweden and Belgium. Those high skills not only contribute to the large size of the middle class in those countries but also increase the chances of households’ remaining in the middle class or even upper class. 5.3  Effects of Social Protection Social protection, by safeguarding people against health problems, loss of income or employment, contributes to well-­being, especially for those at the bottom of the wage scale. Social protection has also contributed to the growth of the middle class.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 39

28/09/2016 08:24

40

31

High skill

45

47

49

43

43

35

39

40

40

22

25

22

46

45

47

44

40

43

33

33

33

34

35

Medium skill

41

Denmark

31

Netherlands

31

Sweden

31

35

Norway

30

40

46

Iceland

Greece

29

43

44

Belgium

Spain

Hungary

42

Germany

Poland

41

16

Luxembourg

Cyprus

47

16

17

20

Lithuania

26

48

13

14

22

UK

24

51

46

11

18

21

Slovakia

23

27

Europe

23

28

26

Czech Rep.

14

21

23

Latvia

49

21

Estonia

50

57

58

19

France

54

27

Italy

55

27

Austria

23

Portugal

15

24

30

Slovenia

Europe’s disappearing middle class?

Low skill

Figure 1.16 Distribution of occupational skills among the middle class, Europe, 2011 (percentage) The increased female employment rates that we have documented previously in this chapter for countries such as Sweden, Belgium, France, Germany, Greece, Ireland, Italy and the United Kingdom were generally facilitated by an increase in welfare policies, such as the expansion of childcare, parental leave and long-­term care insurance for the elderly. The cuts in social protection since the crisis that intervened in many European countries have thus had a reverse impact. Low-­income workers are not the only ones to have been affected by the changes to and reforms of social protection. The middle class has also been affected, notably by the progressive shift away from a universal social protection system in which it had a stake and also by labour market reforms. The decreased quality of public services has also affected them directly. By contrast, in Sweden the middle class has continued to have access to generous and encompassing social protection systems, as well as high-­ quality public services. This is also why the middle class has remained supportive of the welfare state and the relatively high tax burden associated with it. Sweden thus provides an example of the importance of maintaining a universal system to keep the middle class benefiting from – and financing – the social model. 5.4  Family Composition The probability of belonging to the middle class is also very much influenced by the many different types of household structure. The formation and composition of families have rapidly changed and

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 40

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­41

are influencing the probability of belonging to the middle class. Countries in this volume – for instance, the Netherlands – have experienced a rapid increase in the number of single people and a decline of larger households as a result of declining or delayed fertility. Demographic factors are also influential with, for instance, an increase in the number of pensioners due to higher life expectancy. In several countries covered in this volume, the major share of upper middle-­income and upper-­income class households comprises non-­retired adults without children, with a much lower representation of single-­parent and pensioner households. Significantly in the United Kingdom, three-­ quarters of all single parent households are below the median income level compared with only half of dual-­parent households. 5.5 Migration Migration – notably through its impact on the world of work – has also influenced the middle-­income groups, first, by providing a labour force that has contributed to economic growth. Since 1997 immigrant employment has represented at least 40 per cent of total employment growth in Austria, Denmark, Italy and Spain (OECD 2009). Second, because migrants have often filled jobs that locals did not want – generally manual and low-­skilled jobs – and thus provided services that also benefited the middle class. The chapter on Spain, for instance, highlights how the arrival of female migrants has increased the supply of workers for domestic services, thus meeting the demand mainly from middle-­income households. This process stimulated the dual-­earner model by providing cheap domestic services for families and allowing women to join the workforce. The crisis also led to adverse developments. Worsening wages and working conditions in several occupations, including in the public sector, have also led to waves of migration. Some countries that were receiving immigrants have been confronted by growing emigration, especially by highly educated and skilled professionals, leading to a depletion of skills in sending countries and more skills in the receiving countries. Emigration of young graduates occurred throughout Southern Europe, particularly in Spain, Greece, Portugal and Italy, as well as in Ireland and the United Kingdom – generally recipient countries. The changes in the world of work that we describe in this volume for some southern countries – such as real wage falls, weakened collective bargaining, cuts in employment and wages in the public sector and reduced expenditure in health and education – have led many middle-­class employees to emigrate to seek better employment and wage conditions. Emigration of middle-­class public sector employees could also be observed from southern countries to northern countries. The

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 41

28/09/2016 08:24

42

Europe’s disappearing middle class?

ILO documented the illustrative case of a middle-­class employee from the Portuguese state administration who decided to emigrate to Switzerland to work as a cleaning lady to earn more income and ensure the coverage of her family needs.9 A high number of doctors have emigrated from Hungary, Romania, Czech Republic, Poland and Slovakia, but also Spain and Greece, to work in northern countries (Vaughan-­Whitehead 2013).

6. CONCLUSION This volume aims to provide evidence on the general trends with regard to inequalities and middle-­income groups on the basis of extensive research (both statistics and case studies) in individual countries as well as some comparative analysis. These results converge towards a number of conclusions. First, while the middle class in most European countries experienced a rapid increase in the 1980s and 1990s – notably owing to increased labour participation – the past decade seems to have brought some erosion of the middle class, exacerbated by the effects of the recent financial and economic crisis. The expansion of middle-­income households below median income levels is reported in several countries, highlighting a shift towards the left of the distribution. This suggests that several European countries are now moving from a pyramid-­shaped society, in which most people fill low-­income ranks, to a diamond shape, in which a majority enjoy incomes clustered around the mean. In particular this volume highlights a correlation between increasing income inequality and erosion of the middle class. At the same time, great diversity can be found between individual countries. Elements in the world of work were found to help explain these general trends, while also providing some explanations for national differences. The higher labour market participation of women and the prevalence of dual-­earner households, for instance, have clearly lifted a number of households up to middle-­class status. At the same time, the changing structure of jobs and occupations, the growth of non-­standard forms of employment and contracts, and general wage moderation seem to be common factors behind the middle-­ class erosion in Europe already observed in many ­countries before the crisis. As shown for the Netherlands, households, to cover their needs, may have to compensate for the lack of wage growth through multiple income earners within the household. Additional factors were brought by the crisis, such as further increases in unemployment, especially among young people, further real wage decline, the dismantling of some social dialogue institutions and cuts in both

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 42

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­43

employment and wages in the public sector, previously a secure employment and income oasis for the middle class. Some occupations that traditionally represented the middle class, such as teachers or doctors, may not systematically belong to the middle-­income groups anymore. This calls for further research on professional and individual trajectories. All these factors have played a depressing role in reducing both the size and the median income of the middle class in most European countries. Other drivers in the world of work help to explain national stories, such as the importance of social dialogue in the stability of the middle class in Belgium, the importance of the welfare system in Sweden and the education and skills composition in Germany. Industrial relations as a process for influencing the world of work were found to play a particular role. Even if industrial relations systems are very different between, for instance, Belgium, France, the Netherlands and Sweden, their resilience in those four countries seems to have represented one major reason for relatively lower inequalities and greater stability of the middle class. Conversely, the weakening of a number of mechanisms of social dialogue since the crisis seems to have had a direct impact on the world of work in countries such as Greece, Spain and Ireland, with significant effects on inequalities and the middle class. Rather than the crisis, long-­term transformations in industrial relations and labour markets seem to explain the growth of the low-­pay segment and the erosion of the middle class in countries such as Germany, the United Kingdom and Italy. Limited collective bargaining in countries such as the Baltic states and Hungary seems to render both low-­and middle-­class development directly dependent on the economic context and on government tax policy, with some therefore lacking a cushion to sustain the world of work and thus both aggregate demand and production in times of crisis. Mechanisms of wage fixing and wage bargaining were also found to play some role. While the removal of the wage indexation mechanism in Italy in the early 1990s was accompanied by an immediate increase in inequalities, the survival of the indexation system in Belgium seems to have contributed to limit inequalities and to stabilize the middle class there. Extension mechanisms and coordinated collective bargaining also seem to contribute to more coherence along the income scale and thus to less inequalities between the two extremes. Common factors also seem to explain the more insecure position of middle-­income groups across countries, such as greater vulnerability of jobs and incomes in the public sector, and greater exposure to factors such as involuntary part-­time and temporary jobs. In particular the decline in middle-­skill employment that went hand in hand with a decrease of standard work contracts led to job and income polarization and to a decline of

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 43

28/09/2016 08:24

44

Europe’s disappearing middle class?

the middle class. Basic features such as higher education no longer seem to represent a sufficient asset for middle-­income groups to avoid employment and income insecurity. Such an erosion of the middle class is worrying – especially when it seems to hurt young people most, thus leading to an intergenerational gap – and calls for policy initiatives. Increased inequalities and the progressive erosion of the middle class and their possible implications demands urgent policy action specifically aimed at stopping this trend. This volume highlights the policy programmes that could help to reduce inequalities and strengthen the middle class. Such initiatives should be launched in the world of work, but also in other related areas, such as taxation, education and social protection. This requires a new and comprehensive policy agenda extended to different and complementary fields that would finally establish the middle class as a target group and lead to both sustained economic growth and improved living standards.

NOTES 1. For recent articles in the United States, see for instance ‘America’s incredible shrinking middle class’, CBS News, 23 March 2015; ‘Data point to poorer global middle class’, Shawn Donnan and Sam Fleming, Financial Times, ft.com, 8 July 2015; for Canada, see ‘The erosion of middle-­class jobs and incomes in Canada is finally exposed’, Christia Freeland, Financial Post, 30 March 2015; for Europe, see ‘Explaining the struggles of the European middle classes’, by Evert-­Jan Quak, The Broker, 23 April 2015; and ‘Winter of discontent? Why Europe’s middle class is revolting’, CNBC, 9 October 2015. 2. Speech by Commissioner Marianne Thyssen at ‘Roundtable with civil society ­organisations: forging common action to achieve the social triple A for Europe’, Brussels, 1 October 2015. 3. Inequality is often measured using the ratio of the income of the top 10 per cent to the income of the bottom 10 per cent. 4. When the middle class is defined as the middle 60 per cent of the distribution there will always be an equally large middle class because every distribution has a middle ­one-­third. A middle class based on the median will also never cease to exist (there is always a median), but does reflect any polarization trends. 5. We defined here the middle class according to the gross median income – and not the net median income as in the previous section – to better highlight the role of income from labour before taxation. 6. The wage penalty for non-­regular workers was estimated at between 1 and 34 per cent in developed countries (ILO 2015: 26). 7. It is interesting to note that some candidates to the Presidency election in the US have recently proposed profit-­sharing schemes to better ensure wage and living standards progression for the middle class; see ‘Profit-­sharing is a win-­win’ by Hillary Clinton, Boston Globe, 28 July 2015, see also ‘Hillary Clinton focuses on middle class wages in sweeping economic policy speech’, by Samantha Lachman, Huffington Post, 13 July 2015. 8. See the article ‘The shrinking middle class’, 16 December 2012, on the basis of a study conducted by the Bertelsmann Foundation by the University of Bremen and the German Institute for Economic Research, accessed 26 April 2016 at http://www.dw.com/en/ the-­shrinking-­middle-­class/a-­16457570.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 44

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­45

9. ‘From public sector worker in Portugal to cleaner in Switzerland’, ILO News, 20 September 2012, ILO, Geneva, accessed 8 November 2015 at http://www.ilo.org/global/ about-­the-­ilo/newsroom/features/WCMS_189941/lang-­-­en/index.htm.

BIBLIOGRAPHY Alvaredo, F., A. Atkinson, T. Piketty and E. Saez (2013), ‘The top 1 percent in international and historical perspective’, Journal of Economic Perspectives, 27 (3), 3–20. Aranitou, V. (2014), ‘The collapse of the middle class in Greece during the era of the memoranda (2009–2014)’, draft paper for the conference on ‘The Decline of the Middle Classes Around the World?’, Madrid, 28–30 September. Atkinson, A.B. and A. Brandolini (2012), ‘On the identification of the middle class’, working paper ECINEQ WP 2011–217, Society for the Study of Economic Inequality, Verona. Atkinson, T. and A. Brandolini (2013), ‘On the identification of the middle-­class’, in J. Gornick and M. Jäntti (eds), Income Inequality: Economic Disparities and the Middle Class in Affluent Countries, Stanford, CA: Stanford University Press, pp. 77–100. Autor, D.H., W.R. Kerr and A.D. Kugler (2007), ‘Does employment protection reduce productivity? Evidence from US States’, The Economic Journal, 117 (June), F189–F217. Banerjee, A.V. and E. Duflo (2008), ‘What is middle class about the middle classes around the world?’, Journal of Economic Perspectives, 22 (2), 3–28. Beckett, A. (2010), ‘Is the British middle class an endangered species?’, Guardian, 24 July 2010, p. 28, accessed 27 April 2016 at http://www.guardian.co.uk/ uk/2010/jul/24/middle-­class-­in-­decline-­society. Berg, J. (ed.) (2015), Labour Markets, Institutions and Inequality – Building Just Societies in the 21st Century, Cheltenham, UK and Northampton, MA, USA: Edward Elgar and Geneva: ILO. Bettio, F., M. Corsi, C. D’Ippoliti, A. Lyberaki, M. Samek Lodovici and A.  Verashchagina (2013), ‘The impact of the economic crisis on the situation of women and men and on gender equality policies’, European Commission, Directorate-­General for Justice, Publications Office of the European Union, Luxembourg. Billitteri, T.J. (2009), ‘Middle-­class squeeze: is more government aid needed?’, CQ Researcher, 19 (9), 243–62. Cashell, B.W. (2008), ‘Who are the “middle class”?’, Congressional Research Service, Government and Finance Division, Washington, DC. Centre Européen de Formation Professionnelle (CEDEFOP) (2011), ‘Labour-­ market polarisation and elementary occupations in Europe: blip or long-­term trend?’, Publications Office of the European Union, Luxembourg. Centre de Recherche pour l’Étude et l’Observation des Conditions de Vie (CREDOC) (2011), Les classes moyennes en Europe (The Middle Classes in Europe), Cahiers de Recherche, No. 282, December, Paris: CREDOC. Chauvel, L. (2006), Les classes moyennes à la dérive, Paris: Seuil. Corlett, A. and M. Whitakker (2014), ‘Low pay Britain 2014’, Resolution Foundation Report, accessed 5 May 2016 at http://www.resolutionfoundation. org/wp-­content/uploads/2014/10/Low-­Pay-­Britain-­20141.pdf.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 45

28/09/2016 08:24

46

Europe’s disappearing middle class?

Dabla-­ Norris, E., K. Kochhar, N. Suphaphiphat, F. Ricka and E. Tsounta (2015), ‘Causes and consequences of income inequality: a global perspective’, International Monetary Fund (IMF) staff discussion note, June, SDN/15/1, pp. 1–39. Dallinger, D. (2013), ‘The endangered middle class? A comparative analysis of the role played by income redistribution’, Journal of European Social Policy, 23 (1), 83–101. Davis, J.C. and J.H. Huston (1992), ‘The shrinking middle-­income class: a multivariate analysis’, Eastern Economic Journal, 18 (3), 277–85. Easterly, W. (2001), The Middle Class Consensus and Economic Development, Washington, DC: World Bank. Eisenhauer, J.G. (2008), ‘An economic definition of the middle class’, Forum for Social Economics, 37 (2), 103–13. Eisenhauer, J.G. (2011), ‘The rich, the poor, and the middle class: thresholds and intensity indices’, Research in Economics, 65 (4), 294–304 Economic Inequalities European Commission (EC) (2010), Why Are Socio-­ Increasing? Facts and Policy Responses in Europe, Research policy, Brussels: European Commission. European Commission (EC) (2013), Health inequalities in the EU, December, accessed 28 April 2016 at http://ec.europa.eu/health/social_determinants/docs/ healthinequalitiesineu_2013_en.pdf. European Commission (EC) (2015), Research Findings – Social Situation Monitor – Income Inequality in EU Countries, Employment, Social Affairs and Inclusion, Brussels: European Commission. European Parliament (2015), Wage and Income Inequality in the European Union, Study for the EMPL Committee, Directorate General for Internal Policies, Policy department A. Economic and Scientific Policy, European Parliament (IP/A/EMPL/2013-­05), January, Brussels, pp. 1–423. Foster, J.E. and M.C. Wolfson (2010), ‘Polarization and the decline of the middle class: Canada and the U.S’, Journal of Economic Inequality, 8 (2), 247–73. Gigliarano, C. and K. Mosler (2009), ‘Measuring middle-­class decline in one and many attributes’, Working Paper No. 333, Università Politecnica delle Marche (I), Dipartimento di Scienze Economiche e Sociali, Ancona. Goldthorpe, J. (2000), ‘Outline of a theory of social mobility’, in J. Goldthorpe, On Sociology, Oxford: Oxford University Press, ch. 1. Goldthorpe, J. (2010), ‘Analysing social inequality: a critique of two recent contributions from economics and epiemiology’, European Sociological Review, 26 (6), 731–44. Goos, M., A. Manning and A. Salomons (2009), ‘Job polarization in Europe’, American Economic Review Papers and Proceedings, 99 (2), 58–63. Goos, M., A. Manning and A. Salomons (2014), ‘Explaining job polarization: routine-­biased technological change and offshoring’, American Economic Review, 104 (8), 2509–26. Gornick, J.C. and M. Jänti (eds) (2013), Income Inequality – Economic Disparities and the Middle Class in Affluent Countries, Stanford, CA: Stanford University Press. International Federation of Red Cross and Red Crescent Societies (IFRC) (2013), ‘Think differently: humanitarian impacts of the economic crisis in Europe’, IFRC, Geneva. International Labour Office (ILO) (2014), Global Wage Report 2014/15: Wages and Income Inequality, December, Geneva: ILO.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 46

28/09/2016 08:24



Is the world of work behind middle-­class erosion? ­47

International Labour Office (ILO) (2015), Non-­standard Forms of Employment, Report for discussion at the Meeting of Experts on Non-­Standard Forms of Employment, Geneva, 16–19 February, Geneva: ILO. INSEE (National Institute of Statistics and Economic Studies) (2015), ‘Les niveaux de vie en 2013’, INSEE-­Première, No. 1566, September. Kumar, A., M. Rotik and K. Ussher (2014), ‘Pay progression: understanding the barriers for the lowest paid’, October, CIPD and John Lewis policy report, London, accessed 5 May 2016 at http://www.cipd.co.uk/binaries/pay-­ progression_2014-­understanding-­the-­barriersfor-­the-­lowest-­paid.pdf. Marginson, P. and C. Welz (2014), ‘Changes to wage-­ setting mechanisms in the context of the crisis and the EU’s new economic governance regime’, EurWORK, European Observatory of Working Life, June, accessed 5 May 2016 at http://www.eurofound.europa.eu/observatories/eurwork/comparative-­ information/changes-­to-­wage-­setting-­mechanisms-­in-­the-­context-­of-­the-­crisis-­ and-­the-­eus-­new-­economic-­governance. Milanovic, B. and S. Yitzhaki (2002), ‘Decomposing world income distribution: does the world have a middle class?’, Review of Income and Wealth, 48 (2), 155–78. Nolan, B., W. Salverda, D. Checchi, I. Marx, A. McKnight, I.G. Tóth and H. van de Werfhsot (2014), Changing Inequalities and Societal Impacts in Rich Countries  – Thirty Countries’ Experiences, Oxford: Oxford University Press. operation and Development (OECD) (2008), Organisation for Economic Co-­ Growing Unequal? Income Distribution and Poverty in OECD Countries, Paris: OECD. operation and Development (OECD) (2009), Organisation for Economic Co-­ ‘International migration and the economic crisis: understanding the links and shaping policy responses’, in International Migration Outlook 2009, Paris: OECD, accessed 26 April 2016 at http://dx.doi.org/10.1787/ migr_outlook-­2009-­3-­en. Organisation for Economic Co-­ operation and Development (OECD) (2011), Governance at a Glance, Paris: OECD. Organisation for Economic Co-­ operation and Development (OECD) (2014), OECD Employment Outlook, Paris: OECD. Organisation for Economic Co-­operation and Development (OECD) (2015), In It Together: Why Less Inequality Benefits All, Paris: OECD. Petmesidou, M. (2011), ‘What future for the middle classes and “inclusive solidarity” in South Europe?’, Global Social Policy, 11 (3), 225–7. Pressman, S. (2007), ‘The decline of the middle class: an international perspective’, Journal of Economic Issues, 41 (1), 181–200. Pressman, S. (2010), ‘Public policy and the middle class throughout the world in the mid 2000s’, Journal of Economic Issues, 44 (1), 243–62. Rubery, J. (2013), ‘Public sector adjustment and the threat to gender equality’, in D. Vaughan-­Whitehead (ed.), Public Sector Shock: The Impact of Policy Retrenchment in Europe, Cheltenham, UK and Northampton, MA, USA: Edward Elgar and Geneva: ILO, pp. 43–83. Scott, R. and S. Pressman (2011), ‘A double squeeze on the middle class’, Journal of Economic Issues, 45 (2), 333–41. Taylor, C., A. Jowett and M. Hardie (2014), ‘An examination of falling real wages, 2010–2013’, Office for National Statistics, accessed 17 October 2014 at http:// www.ons.gov.uk/ons/dcp171766_351467.pdf.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 47

28/09/2016 08:24

48

Europe’s disappearing middle class?

Thorson, G. (2014), ‘The rise of inequality, the decline of the middle class, and educational outcomes’, paper presented at the 2014 International Conference of the Associate for Public Policy Analysis and Management (APPAM), 29–30 September, Segovia. Toomse, M. (2010), ‘Looking for a middle class bias: salary and co-­operation in social surveys’, working paper, Institute for Social and Economic Research, University of Essex, Colchester. Vaughan-­Whitehead, D. (ed.) (2013), Public Sector Shock: The Impact of Policy Retrenchment in Europe, Cheltenham, UK and Northampton, MA, USA: Edward Elgar and Geneva: ILO. Vaughan-­Whitehead, D. (ed.) (2015), The European Social Model – Is Europe Losing its Soul?, Cheltenham, UK and Northampton, MA, USA: Edward Elgar and Geneva: ILO. Visser, J., S. Hayter and R. Gammarano (2015), ‘Trends in collective bargaining coverage: stability, erosion or decline?’, Labour Relations and Collective Bargaining, Issue Brief No. 1, Geneva: ILO.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 48

28/09/2016 08:24

49

VAUGHAN 9781786430595 PRINT (M4051).indd 49

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

Labour market participation Women 1980–90s: Rise in female labour market  participation rates in most countries, which allowed families to acquire or sustain middle-­class status. Women’s participation was facilitated by the improvement of welfare policies such as the expansion of childcare, parental leave and long-­ term care insurance for the elderly in Belgium, Sweden, Norway, Germany and France 2008–15: The EU-­wide gender gap in the  employment rate has fallen by 3.2 percentage points: from 14.1 points in 2008 to 10.9 in the first quarter of 2012 since men were most affected by the impact of the crisis on manufacturing. This affected the dual-­earner model and in some cases even brought a shift from dual to female breadwinner households: more pronounced in countries worst hit by the crisis, such as Estonia, Latvia and Lithuania

Positive effect on the middle class

1980–2000s: Over-­representation of women in the  service sector (including public sector jobs) and part-­time employment; under-­representation in manufacturing, construction 2008–15: Cuts in public spending in the second phase  of the crisis, and fewer legal restrictions on firing part-­timers, adversely affected female employment rates in Greece where it fell by 8 per cent, in Spain (−4 per cent), and in Ireland (−3 per cent), Portugal (−3 per cent) the Netherlands (−3 per cent) and Latvia (−3 per cent). The collapse of dual-­earner households has made it harder for families to sustain middle-­class status

Adverse effect on the middle class

Table 1A.1  Some effects of the world of work on the middle class, Europe, 1990–2015

APPENDIX 1A.1

50

VAUGHAN 9781786430595 PRINT (M4051).indd 50

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

Youth

Older workers (55–64 years)

2000–2015: An ageing population has led to  a lengthening of working life and a means of sustaining middle-­class status for most countries. Europe’s employment rates for older workers rose from 38.4 per cent in 2002 to 51.8 per cent in 2014. This was exacerbated by the 2008 financial crisis which caused many countries to implement a gradual increase in retirement age 1990–2007: In 2007 youth unemployment remained  below the EU28 average of 15.6 per cent in Germany, Estonia, Ireland, Latvia, Lithuania, the Netherlands and the United Kingdom owing to higher enrolment rates in tertiary education, stable labour market growth and in some cases youth emigration (notably the Baltic states). Vocational training and short-­time working schemes in Germany and Belgium helped to maintain low youth unemployment, thereby helping younger generations to achieve or maintain middle-­class status. These schemes were particularly effective during the 2008 crisis, notably in Germany – the only country to experience a decline in youth unemployment rates (3 per cent)

Positive effect on the middle class

Table 1A.1  (continued)

1990–2007: Higher youth unemployment rates in  countries such as Greece, Spain, Italy, Hungary and Portugal have created an intergenerational gap between different age cohorts 2008–15: Less stable job opportunities for young people  during the crisis added higher cyclical sensitivity to temporary employment. The average young worker in temporary employment in the EU experienced a wage penalty of 14.4 per cent in 2010 in comparison to a worker on a standard contract with similar characteristics. With the exception of Luxembourg, all EU member states have seen a rise in the number of those not in employment, education or training (NEETs) since the onset of the crisis; in 2010 in Italy and the United Kingdom, the size of the NEET population reached approximately 1.1 million of those aged 15–24 years. Youth unemployment rates rose in all

2000–2015: The postponement of older workers’ exit  from the labour market has implied a higher probability for them to remain within the middle class

Adverse effect on the middle class

51

VAUGHAN 9781786430595 PRINT (M4051).indd 51

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

1990–2000s: A change in the occupational  composition of the middle class led to growth in middle-­income jobs owing to an expansion of the public and service sectors, as in Sweden, Belgium and the Netherlands

New forms of work contracts 2008–15: The crisis had a negative impact on the 1990–2008: Countries such as France, Italy and the Self-­  creation of self-­employed workers and increased their employment  United Kingdom adopted specific social vulnerability owing to less demand for their products measures, such as tax incentives and employment and services, less financial support from lending protection legislation, to encourage development institutions and cuts to their social benefits. of the self-­employed, leading to increased middle-­ class well-­being

Job restructuring

 EU28 states between 2008 and 2014 (with the exception of Germany) and in 2014 remained above 50 per cent in Greece and Spain, above 40 per cent in Italy and above 30 per cent in Portugal 1990–2000s: During 1995–2008 countries began to  witness varying degrees of job polarization owing to a concomitant rising demand at the upper and lower ends of the occupational skill distribution, driven by technical change and globalization. This polarization can also be linked to institutional transformations such as the de-­standardization of employment regulation (most prominent in continental Europe in late 1990s to early 2000s) 2008–15: The crisis brought increasing evidence of job  polarization mainly through the destruction of middle-­wage jobs. This was evident in Greece, Spain, Austria, Latvia and Lithuania which changed from centripetal development to very intense job polarization. Jobs at the bottom of the employment structure either decreased less more markedly than the middle or even expanded significantly in a few cases (such as France, Greece and the UK)

52

VAUGHAN 9781786430595 PRINT (M4051).indd 52

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

2008–15 Self-­employed workers increased in UK  (18 per cent), France (9.9 per cent) and the Netherlands (23 per cent)

1990–2000s: Labour market reforms – aimed at  increasing flexibility – in countries such as France, Spain, the United Kingdom, Sweden and Italy that simultaneously led to the growth in part-­time employment, made it possible for many women to enter the labour market. These dual-­ earner households allowed many families to reach the middle class 2008–15: Short-­time working was used across  countries to cushion the fall in employment in the recent crisis in Sweden, Germany and Belgium. This helped to preserve both low-­class and middle-­class employment

2008–15: The rise in involuntary part-­time  employment in many countries was used as a practical way to reduce hours to avoid employment cuts in the recent economic crisis

Self-­ employment

Temporary and part-­time employment

Involuntary part-­time

Positive effect on the middle class

Table 1A.1  (continued)

 This caused a decline in self-­employed workers in Spain (−13 per cent), Italy (−5.7 per cent), Greece (−17 per cent) since 2008 and in Germany since 2010 (−2.1 per cent). At the same time, development of ‘dependent self-­employed’ with weak statute and weak protection 1990–2000s: Reforms aimed at increasing labour  market flexibility in countries such as France, Spain, the United Kingdom, Sweden and Italy led to a growth in temporary and part-­time employment, which put at risk a core feature of the middle class: a permanent job with an open-­ended contract. The growth of different types of non-­standard forms of employment (NSFEs) strengthened this trend 2008–15: The contemporary ‘buffers’ in the crisis were  young men and women on temporary and other atypical employment contracts, as well as migrants, many of whom were employed irregularly or temporarily. A last-­in-­first-­out principle also prevailed, as firms found it more difficult to fire tenured employees than to stop renewing temporary contracts 2008–15: Involuntary part-­time as a percentage of total   part-­time employment rose between 2007 and 2014 in France from 30 to 40 per cent, Spain from 33 to 63 per cent, Greece from 42 to 69 per cent and Ireland from 10 to 41 per cent, putting at risk the stable nature of employment, which many middle-­class people

Adverse effect on the middle class

53

VAUGHAN 9781786430595 PRINT (M4051).indd 53

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

The public sector Wages 1990–2000s: Growth of public sector wages in  most countries during this period, which sustained growth and maintained the living standards of the middle class It allowed women to have access to jobs and   to highly skilled jobs They set the standards in terms of work and family  reconciliation schemes that contributed to the growth of women’s participation rates and to the emergence of the dual-­earner model as an engine of middle class growth Employment 1990–2000s: Sustained public sector employment in  most countries during this period. The public sector acted as a major source of employment for the highly educated, especially for women, and was a major nurturer of the middle class 2008–15: In Ireland two public sector agreements  were negotiated during the crisis, thereby

2008–15: Large-­scale public sector wage cuts during the  crisis in most countries caused a substantial change in gross and net wages, impacting upward social mobility and middle class aspirations. Wages fell in Greece (−15–30 per cent in 2010), Hungary (−37 per cent for unskilled and −13 per cent for high skilled in 2008–10), Latvia and Lithuania (−25 per cent and −15 per cent, respectively, in 2009–10), Spain (−5 per cent in 2010), Portugal (−5 per cent average wages 2010–11), United Kingdom (wage freeze in 2010–12 led to −5 per cent fall in real wages) 1990–2000s: An increase in privatization programmes  led to a decline in public sector employment in in Greece (from 40 to 34.5 per cent between 1993 and 2008), Sweden (−19.6 per cent in same period) and Germany (6.74 million in 1991 to 4.65 million in 2014) 2008–15: Large-­scale public sector employment cuts  during the crisis in most countries caused a break

 are accustomed to. In 2010, when the first employment dip occurred in the EU28, the share of involuntary over total part-­time had risen to 32.4 per cent, up 5.5 points from its 2007 level. Underemployment also remains an issue: across the EU 9.8 million part-­time workers (or 22 per cent) worked fewer hours than they would have liked to in 2014. In Greece the proportion rose to 72 per cent, 66 per cent in Cyprus and 57 per cent in Spain. More than two-­thirds of all underemployed workers were women

54

VAUGHAN 9781786430595 PRINT (M4051).indd 54

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

Growth of public services helped middle class  families in terms of provision of childcare and so on

Services

Industrial relations Indexation 1990–2000s: In the period before the crisis,  indexation measures linking wages to the cost mechanisms of living were in place in Belgium, Cyprus, Luxembourg, Malta and Spain. These have historically ensured less inequality between lower and middle-­income classes through more egalitarian income distribution 2008–15: Maintaining indexation mechanisms  contribute to explain middle-­class stability as explained for Belgium in this volume In Italy a 2009 cross-­sectoral agreement introduced  a new indicator of expected inflation as the reference point for sectoral wage negotiations

 reinstating the middle-­class status of many public sector employees

Employment

Positive effect on the middle class

Table 1A.1  (continued)

1990–2000s: In 1992 the removal of the indexation  system ‘scale mobile’ directly contributed to decrease real wages and to increase wage inequality 2008–15: Changes to indexation arrangements have led  to greater polarization between lower-­and middle-­ income classes as the wage gap widened in, for example, Cyprus, Luxembourg and Spain

 in the middle class ‘safety net’. Strict hiring rules in Greece since 2010 have caused precarious public sector employment with reduced rights, as in Italy and Portugal. The launch of extensive privatization programmes in Greece, Portugal, Spain and, to a lesser extent, in Ireland, Italy and the United Kingdom has impacted job quality and security 2008–15: Cuts in expenditure in the public sector  affected the middle class through lower supply and quality of public services; it also rendered more difficult women’s participation in the labour market, hurting the dual model and the middle class

Adverse effect on the middle class

55

VAUGHAN 9781786430595 PRINT (M4051).indd 55

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

Trade union density

Minimum wage setting

1990–2015: Higher affiliation rates of middle-­  income classes compared with low-­and top-­ income classes in Belgium, Spain and Estonia; in the United Kingdom middle earners are more likely to join a union. In Belgium and Sweden relatively low wage inequality can be attributed to high trade union density for all cohorts. Higher trade union density of middle classes has enabled them to have an impact on collective bargaining, thereby preserving wages and welfare

1990–2008: Twenty-­one of the EU28 set a universal,  statutory minimum wage. In most of central and eastern Europe, minimum wages are subject to attempted tripartite agreements. Wage increases in Ireland were negotiated under the national wage agreement. In the UK the government decides following the recommendations of a tripartite commission. In Germany, since January 2015, the new government introduced a national statutory minimum wage

2008–15: Changes to statutory minimum wage-­setting  procedures occurred in Croatia, Cyprus, Greece, Hungary, Ireland (cut in 2010 but restored by new government in 2012), Poland, Portugal, Spain, Slovakia and Slovenia. In Greece the minimum wage is no longer defined by the social partners through the National General Collective Agreement (NGCA), but by the government in consultation with social partners. In Spain since 2011 the government no longer consults the social partners before setting the minimum wage. In Portugal the Memorandum of Understanding (MoU) imposes restrictions on criteria for any increase in the minimum wage (Marginson and Welz 2014). This has increased inequality and affected both bottom and middle-­income groups 1990–2015: Long-­term decline in trade union density in  Greece, the Baltic states, Hungary, the Netherlands, the United Kingdom and France 2008–15: The crisis caused an accelerated decline in  density in the worst affected countries; and a loss of trade union trust and dampened reputations in Spain and also Ireland. In the United Kingdom the proposed implementation of a trade union bill will make it harder to strike

56

VAUGHAN 9781786430595 PRINT (M4051).indd 56

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

Extension procedures and automatic continuation of collective agreement

Adverse effect on the middle class 2008–15: The crisis led to changes to extension  procedures, making them more difficult or impossible in Greece, Slovakia, Portugal and Romania; in Ireland extension orders were suspended after a ruling of the Supreme Court. There have also been changes in their use in Bulgaria, Germany and Italy where judicial decisions have underpinned de facto extension of the wage provisions of sector agreements. Changes to clauses providing for the continuation of agreements beyond expiry have featured in Spain where the validity of expired agreements was limited to one year, in Portugal to 18 months and in Greece and Croatia it is limited to three months. In Estonia, legislative changes required parties to agree to the continuation of agreements

Positive effect on the middle class

1990–2000s: In France, Austria and Belgium  extensions of collective agreements are guaranteed by the intervention of a public body, and in Portugal by ministerial decree. In Ireland agreements must be registered with the Labour Court. Extension is ensured by criteria related to representation in Greece and Spain, or like the presence of a substantial proportion of workers in the sector already covered by the agreement, or the representativeness of the signatory organizations. These mechanisms are used as a tool to ensure a minimum standard of working conditions, thus contributing to the stability of the middle class. Some countries have mechanisms of automatic continuation of collective agreements on expiry – for example, Spain (ultraactividad), Greece (metenergeia) and Portugal (sobrevigencia)

Table 1A.1  (continued)

57

VAUGHAN 9781786430595 PRINT (M4051).indd 57

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

Coverage of collective bargaining/ coordinated or multilateral collective bargaining

1990–2000s: Coordination and multi-­level  bargaining make it possible to cover a higher number of employees by collective agreements, as in Belgium (96 per cent of total employment), France (93 per cent), Sweden (90 per cent), the Netherlands (80 per cent) and in contrast with UK and the Baltic states where single-­employer bargaining predominates. Single-­employer bargaining can lead to unilateral employer decisions, that can cause greater polarization between income classes Multi-­employer bargaining involving sector and/  or cross-­sector arrangements also predominated in Portugal, Greece and Spain 2008–15: Coverage rates remained stable in France,  Italy, Austria and Belgium. In Belgium and France the extension of collective agreements played an important role. In Finland and the Netherlands, coverage increased (Visser et al. 2015)

1990–2000s: Collective bargaining agreements in  Germany in the 1990s led to a decline in collective bargaining coverage and a rise of low wage workers. Similarly in Sweden the breakdown of centralized bargaining in 1983 led to wage setting based on individual performance 2008–15: The crisis caused a sharp decline in coverage  rates in countries that underwent extensive changes to their collective bargaining mechanisms, such as Cyprus, Greece, Ireland, Latvia, Portugal and Romania

58

VAUGHAN 9781786430595 PRINT (M4051).indd 58

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

1980–2000s: Countries such as Belgium, Portugal,  Sweden, Ireland and the Netherlands have historically used social dialogue that helped to stabilize the middle class. These consultation mechanisms have maintained wages and working conditions, and therefore middle-­class well-­being. Spain and Greece are also marked by a succession of bilateral or trilateral social agreements 2008–15: During the economic crisis a rapid  deterioration of the social climate was avoided in countries in which the government has managed to organize tripartite consultations, as in Estonia, Latvia, Lithuania, Portugal and Ireland (second phase of the crisis)

1980–2000s: In Italy, there has been a trend towards  diminishing the role of social dialogue institutions, and in the Baltic states unions and social dialogue have declined in importance 2008–15: A lack of effective social dialogue prevailed   in many countries during the implementation of labour market reforms. This led to increased strikes and in the case of Italy, Spain, Portugal and Greece, recourse to the courts. In Spain in 2010 and 2012 bipartite negotiations aimed at modernizing collective bargaining were considered unsatisfactory and led to the unilateral reform of the legal framework. In Hungary institutions of tripartite social dialogue were modified by the government in 2010

Adverse effect on the middle class

Source:  National studies and variety of other sources including Bettio et al. (2013), Marginson and Welz (2014) and Visser et al. (2015).

Tripartite consultations

Positive effect on the middle class

Table 1A.1  (continued)

59

VAUGHAN 9781786430595 PRINT (M4051).indd 59

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

Tax and benefits

1990s–2000s: Public policies have helped to preserve the  middle class in a number of countries prior to the 2008 crisis, having a mostly positive effect on household income. These redistributive policies have particularly

Trends beyond the world of work Higher 1990s –2000s: This period saw a general rise of the share education  of individuals with tertiary education. Extended stints in education have also contributed to declining youth unemployment. Those with just a primary education have generally been over-­represented in the bottom and lower middle classes. In Spain the number of those with a tertiary education doubled over the past three decades; in Estonia it tripled. Hungary experienced a rise from 6.5 per cent in 1980 to 18.9 per cent in 2011. Such increased education was beneficial for integrating young generations into the middle class Most governments also invested in education, which   favoured the growth of the middle class

Positive effect on the middle class

2000s–2015: Investing in education is a strategy often used  by the middle class to guarantee a better future for their offspring. However in labour markets characterized by over-­supply of skilled labour and insufficient demand for skills owing to high unemployment rates, new higher-­skilled entrants are experiencing a more severe wage penalty, as in Italy and Spain. Moreover new graduates do not find a job at their level of education/skills. While this mismatch between education and labour market requirements is a long-­term trend that was already emerging before the crisis, the reduction of job opportunities as a result of the crisis has aggravated this mismatch and caused young people to spend longer in higher education (as a long-­term strategy to achieve middle class status or above) 2008–15: The financing of public universities, as in Spain,  has been reduced, causing students to rely on bank loans. The Baltic states after having increased education expenditure during 2006–09, started to cut it in recent years. Expenditure also declined in Portugal and Romania between 2007 and 2011 (by 0.12 per cent and 0.27 per cent respectively) 2008–15: Crisis-­induced welfare reforms, coupled with  government measures to increase revenue through higher taxation, without overburdening the low paid has had a negative effect on household incomes in the middle class.

Adverse effect on the middle class

Table 1A.2  Effects of other factors (outside the world of work) on the middle class, Europe, 1990–2015

60

VAUGHAN 9781786430595 PRINT (M4051).indd 60

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

Adverse effect on the middle class  While the middle class in those countries with relatively more inclusive welfare states has been less affected by these cuts, the opposite holds true for Southern Europe and Ireland (property taxes) and Hungary where the introduction of a 16 per cent flat tax rate favoured upper and middle classes at the expense of the poor and has not led to an intended boost of the economy. In Greece the welfare state acted as a shock absorber but only for the middle and upper classes. In the United Kingdom, policy post-­2010 reduced the earnings threshold for 40 per cent income tax and raised the personal allowance by 30 per cent, with a negative effect on middle-­class household incomes. In Portugal a new surtax on public wages impacted middle-­class aspirations 2008–15: In Spain regional disparities in terms of university  fees occurred during the crisis. University tuition was raised, increasing the average cost from the school year 2009–10 to the school year 2013–14 by an average of 47 per cent (37 per cent in real terms). Regions such as Madrid and Catalonia have applied larger increases of 108 per cent or 138 per cent, respectively (in real terms). This has led to an increase in the cost differences of university tuition depending on region of residence, resulting in an increase in the maximum/minimum ratio from 1.4 in 2009–10 to 2.8 in 2013–14

Positive effect on the middle class

 benefited the middle class in Sweden, France, the Netherlands and Belgium (where the middle class have one of the highest tax burdens in the OECD, at around 50 per cent but they are also among the main beneficiaries of cash and non-­cash benefits). In the United Kingdom, Southern Europe (Greece and Italy owing also to tax avoidance), Hungary and the Baltic states, welfare policies are less redistributive in comparison, leading to larger inequalities and polarization between classes. In Estonia a gradual reduction of the income tax rate from 26 per cent in 2000 to 21 per cent in 2008 benefited people in higher-­income deciles more and increased inequality in net incomes

1990s–2015: Regional disparities in the composition of  the middle class can be identified in Belgium (35 per cent of Flemish and Walloon households are middle class in contrast to 23.4 per cent in Brussels); France (the Paris region is over-­represented by middle/upper-­middle class); Germany (fewer eastern than western Germans assign themselves to the upper-­middle and upper class). In Italy the Gini index is higher in the south (0.34) than it is in the North (0.3), which may signify lower middle-­class attainment in the south

Tax and benefits

Regional disparities

Table 1A.2  (continued)

61

VAUGHAN 9781786430595 PRINT (M4051).indd 61

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

Family changes

Migration

1990s–2000s: Inflow of migrants to western and parts  of southern Europe in 2004 following enlargement of the EU. Many took manual and low-­skilled jobs, which nationals often did not want. Between 2000 and 2008 the non-­national component of elementary workers increased by 85.6 per cent, while the rise in the national component was only 13.7 per cent (CEDEFOP 2011). Immigrants from non-­Western countries tend to reside at the bottom of the income distribution in contrast to those from western Europe who are rather situated in top classes. This has resulted in a polarization of households with foreign backgrounds. Since 1997 immigrant employment has represented at least 40 per cent of total employment growth in Austria, Denmark, Italy and Spain (OECD 2009). Immigration also stimulates the dual-­earner model by providing cheap domestic services for families, and allowing women to join the workforce 1990s–2000s: Rise of dual/multiple earner households  of the middle class in most countries, in parallel with the increase in female employment rates. Middle-­class households are generally characterized by a higher share of better-­educated, non-­retired adults without children. Lone parents are more represented in the lower class, with divorce rates generally cited as an explanation for declining middle classes

2008–15: Changing composition of households during the  crisis; for example, in Greece between 2008 and 2014 the share of dual earner households collapsed in the middle class (−18.9 percentage points against 9 and 10 percentage points among the lower and upper classes, respectively)

2008–15: Immigration countries during 2001–07 have  become net exporters of migrants. Emigration of young graduates occurred throughout Southern Europe, particularly in Spain, Greece, Portugal and Italy, as well as in Ireland, the United Kingdom (generally recipient countries). This marked a defining feature of the recession for many middle and upper-­income households. The most mobile tended to be among the best skilled, leading to a depletion of skills in sending countries and more skills in the receiving countries. Emigration of middle-­class public sector employees could also be observed from southern European countries to northern countries. Female migrants emigrated to work in the care sector, where demographic ageing continues to sustain demand for personal services (Bettio et al. 2013)

2. Is the world of work stimulating middle-­class growth in the Baltic states? Jaan Masso, Inta Mierina and Kerly Espenberg 1. INTRODUCTION The transition from a command to a market economy and the ensuing convergence processes have brought fundamental changes to post-­communist countries’ labour markets in a relatively short period. These have been analysed from different angles, but at least in the economics literature not much has been said (at least not explicitly) about the evolution of the middle class during and after the transition. In the case of the former communist countries, aggregate developments have been well documented, but these do not reveal the impact on the middle class (Atkinson and Brandolini 2011). It has been argued that the middle class was a loser in the transformation process. This claim merits careful consideration because of its importance for the growth and economic development of these ­countries. Since the Industrial Revolution the middle class has been thought of as the source of various elements needed for growth: new ideas, physical capital accumulation, human capital accumulation, entrepreneurship and innovation, and small businesses (Acemoglu and Zilibotti 1997; Kharas 2010). The middle class also contributes to economic growth through its substantial contribution to the development of democracy and active political participation (Banerjee and Duflo 2007). In this chapter we present the development of the middle class in the Baltic states in line with long-­term labour market developments. The case of the Baltic states as post-­communist economies is particularly interesting for that purpose owing to the major changes in the world of work that have occurred in a relatively short time. The chapter is structured as follows. Section 2 gives some idea of the size of the middle class in the Baltic states in terms of income, occupation (sociological approach) and people’s self-­perception. Section 3 seeks to establish long-­term labour market developments, especially in terms of 62

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 62

28/09/2016 08:24



Is the world of work stimulating middle-­class growth in the Baltic states? ­63

what the structure of employment, labour supply, wages and labour market institutions mean for the situation and development of the middle class in Estonia. The two cases studies in section 4 take a close look at the effects of migration in Latvia on middle-­class formation and the situation of one occupation commonly included in the middle class, teachers, in Estonia.

2. EVOLUTION AND POSITION OF THE MIDDLE CLASS IN THE BALTIC STATES 2.1  Size of the Middle Class When the Baltic states were part of the Soviet Union, household income inequality was fairly modest, as it was in other Soviet republics (Masso et al. 2014b). That was owing to various reasons, such as low returns to education, full employment and practically non-­existent private entrepreneurship. At the end of the Soviet period income inequality was at a similar level in the three Baltic states (see Table 2.1). The start of the transition brought a sharp increase in all these countries, although they were far from uniform (Masso et al. 2014b). In Estonia the Gini coefficient increased over just three years from 0.23 to 0.38, reaching its highest point by around 2000, after which it decreased to around 0.3. In Lithuania and, especially, Latvia, the increase was more gradual. In Latvia, the gradual increase in inequality did not bring favourable developments in the long run; the gradual increase of inequality made Latvia the country with the most unequal income distribution by 2003. These developments are to some extent surprising as the Baltic states have had well-­correlated business cycles and institutions. The differences in their respective transition processes provide some explanations, however. Compared with its neighbours, Estonia introduced radical economic reforms and privatization earlier and moved more rapidly to a liberal market economy, while the development of private entrepreneurship was more successful in Estonia owing to more rapid development of the financial sector and the ability to attract foreign direct investment. These developments supported growth, but may also have led to a temporary rise in inequality. A different way of looking at developments with regard to inequalities would be to analyse the dynamics of the income shares of different income groups, such as deciles or quintiles. The share of income going to the highest decile increased substantially in Estonia, somewhat less in Lithuania and least of all in Latvia. Developments concerning the income of the lowest 10 per cent differed even more across the three countries: a sharp drop in Estonia (from 4.2 per cent in 1988 to 2.5 per cent in 2011),

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 63

28/09/2016 08:24

64

VAUGHAN 9781786430595 PRINT (M4051).indd 64

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

Estonia Latvia Lithuania Estonia Latvia Lithuania Estonia Latvia Lithuania Estonia Latvia Lithuania Estonia Latvia Lithuania Estonia Latvia Lithuania

Gini coefficient

23.0 22.5 22.5 4.2 4.5 4.5 18.4 18.7 19.0 57.4 56.8 56.6 1.7 1.7 1.8

1988 39.5 27.0 33.6 2.5 4.3 3.1 31.4 22.1 27.8 47.3 53.7 50.1 3.1 2.1 2.6

1993 30.1 31.7 32.3 3.4 2.6 3.1 22.7 25.3 25.6 53.7 52.8 51.8 2.2 2.3 2.4 11 245 8408

1996 37.6 33.5 32.1 3.0 2.8 3.2 29.8 26.1 25.2 47.9 51.6 52.0 2.9 2.5 2.4 13 705 9912

1998 34.9 35.1 35.5 2.6 2.8 2.9 26.8 27.5 27.9 51.0 50.3 49.6 2.6 2.6 2.7 18 359 14 020

2003

Source:  Standardized World Income Inequality database, World Bank database, Eurostat, authors’ calculations.

The gap between the  top and the middle class: ratio Q5/Q3 GDP per capita,  PPP (constant 2011 international USD)

Income share by   middle 60%

Income share held   by highest 10%

Income share held   by lowest 10%

Country

Indicator

Table 2.1  Income inequality indicators, Baltic states, 1988–2011

32.0 35.8 37.1 2.9 2.7 2.7 25.0 27.4 29.0 52.2 50.2 49.0 2.4 2.6 2.8 24 743 20 977 23 223

2008

32.7 36.0 32.6 2.5 2.1 2.6 24.8 27.4 24.6 52.5 50.8 53.0 2.4 2.6 2.3 23 576 19 405 22 530

2011



Is the world of work stimulating middle-­class growth in the Baltic states? ­65

a bit less in Lithuania (4.5 to 3.1 per cent) and Latvia (from 4.5 per cent to 3.1 per cent). However, once again, developments in Latvia and Lithuania lagged behind those in Estonia, and by 2011 Latvia was clearly the country with the most unequal income distribution. For the current study the indicator of particular interest is the income share going to the 60 per cent in the middle (Solow’s indicator of the middle class). As we can see, meaningful developments can be spotted in the longer term, not so much year by year. In all countries the share decreased substantially from the pre-­transition level of 56–57 per cent by 1993, though again most in Estonia and least in Lithuania. Although in Estonia the share decreased to 47.9 per cent by 1998, it increased thereafter to 52.5 per cent by 2011. The developments with regard to inequalities can be compared with general labour market developments. Generally, the following periodization can be applied to developments in the Baltic economies and labour markets (based on Masso et al. 2012b). The years 1995–99 can be described as the period of stabilization that ended with the Russian crisis in 1998. The years 2000–2007 can be described as a period of rapid economic development and welfare growth in which many social indicators improved. Especially since 2004, tight labour markets (with unemployment close to what mainstream economics calls ‘natural’ levels) caused high upward pressure on wages that in some periods was ahead of productivity growth and led to an increasing labour share of gross domestic product (GDP). Similar to several other European countries, a housing bubble developed. The opening of European Union (EU) labour markets provided migration possibilities which were exercised in particular by the low skilled. The two latter developments may explain declining wage inequalities and decreasing returns to education. The period of the Great Recession (since 2008) hit the Baltic states hard, as did the preceding overheating of the economy (for example, in real estate) and the ‘austerity’ measures applied at the beginning of the crisis. The whole period after regaining independence can be characterized, despite setbacks owing to the economic cycle, by strong convergence of income levels: during 1996–2011 GDP per capita adjusted for purchasing power increased by 2.1 times in Estonia and by 2.3 times in Latvia (Table 2.1). In addition to successful economic reforms at the beginning of the transition processes, joining the EU has had a positive effect (as shown, for example, by Campos and Dabušinskas 2009). Thus, with regard to the performance of the middle class in the Baltic states, its relatively better performance over time and across countries is probably related to convergence rather than other factors. While the above indicators assume a fixed 60 per cent share of the middle class, Table 2.2 presents the size of the different income c­ ategories (defined relative to the median), with 60–80 per cent corresponding to

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 65

28/09/2016 08:24

66

Europe’s disappearing middle class?

Table 2.2 Size of different income categories, Baltic states, 2004–11 (percentage) Income class (% of median) 0–60 60–80 80–120 120–200 60–200 >200

Estonia

Latvia

Lithuania

2004

2006

2011

2004

2006

2011

2004

2006

2011

25 12 22 25 60 15

21 14 28 26 69 11

23 14 25 25 67 12

23 14 24 26 65 14

24 13 23 26 64 14

23 14 25 28 63 10

24 14 24 25 62 14

22 14 27 26 66 12

25 13 24 26 63 12

Source:  Authors’ calculations from EU-­SILC database with the assistance of Nicolas MaÎtre from the International Labour Office (ILO).

the lower-­ middle class, 80–120 per cent to the core middle class and 120–200 per cent to the upper-­middle class. As we can see, the size of the middle class varies within 60–69 per cent, being larger in Estonia, probably because of its lower level of inequality. Masso et al. (2015) – using the EU’s Statistics on Living and Income Conditions (SILC) data and definitions by Atkinson and Brandolini (2011), first those with equivalized incomes between 75 per cent and 125 per cent of the median, and second those whose income falls between 75 per cent and 200 per cent of the median – found that the middle class in Estonia was similar to that in Poland as presented in Atkinson and Brandolini (2011) for 2004 (respectively 30 and 60 per cent in 2012), while both indicators show a slightly growing share of the middle class during 2004–12 in line with decreasing inequality. Zickute (2013) proxied the middle class as inhabitants whose income exceeds the absolute poverty threshold (subsistence minimum) by 2–5 times. Her calculations for 2007 and 2011 showed that the share of the middle class was largest in Estonia (it increased from 50 to 60 per cent, while our numbers based on micro data show an increasing trend from 33 per cent in 2004 to 52 per cent in 2011), while in Latvia it remained at a stable low level of 27 per cent and in Lithuania it plummeted from 50 to 9 per cent. One explanation was that while the minimum wage significantly exceeds the poverty threshold in Estonia, the opposite was the case in Lithuania, which contributed to the rapid growth of the working poor. It is common for researchers in Latvia to argue that a ‘middle class’ has not yet been established there (for example, Krastiņš 2011; Meņšikovs and Lavrinoviča 2011). Another relevant aspect of the middle class concerns its position relative to the top. This could be a key indicator in the analysis; while the relative

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 66

28/09/2016 08:24



Is the world of work stimulating middle-­class growth in the Baltic states? ­67

income position of the middle class in developed countries has remained stable, the distance between the middle and the top, defined as the Q5/Q3 income share ratio, has been growing (Dallinger 2013). Table 2.2 shows that while in Estonia and Lithuania some decrease in the distance from the top can be seen in recent years, the figure is highest in Latvia. In any case, there is a consistent pattern of a higher gap in the Baltics compared with the rest of the EU. Another way of defining the middle class (or social classes in general) is based on occupations (for example, the International Standard Classification of Occupations, that is, ISCO classification), although of course it leaves out the unemployed and inactive people or households. The question in the literature concerns the polarization of employment; in other words, whether middle-­class jobs are vanishing due to processes such as skill-­biased technological change, globalization and shifts to the service economy (Dallinger 2013). One particular aspect of the shift to a service economy is the replacement of mid-­level jobs (skilled manual workers) with jobs requiring lower or higher skills (Dallinger 2013). While some classifications have simply distinguished between middle and working classes, Oesch (2006) argued that the western European employment structure is not usefully summarized by a class scheme that includes a monolithic middle class and working class that are separated along the manual/non-­manual boundary. We have thus considered the classification by Leiulfsrud et al. (2010). Accordingly, we distinguished between the top and upper-­middle class (ISCO 1–2), the middle class (3–5) and the lower class (6–9). On this basis, in our opinion, occupations in ISCO 2 such as teachers and doctors should be classified in the middle class in the case of the Baltic states.1 We modified this approach by leaving only ISCO 1 occupations in the upper class (Table 2.3). Naturally, this classification could be criticized for some countries or time periods; for example, it has been argued that in post-­communist ­countries – for example, the Russian Federation in transition – occupation is not a good indicator of a person’s economic position (Harwin and Fajth 1998; Domański 2011). In the case of Estonia, we note that such a classification is not entirely correlated with the ranking of 1-­digit occupations by wages (Campos and Dabušinskas 2009). Generally, in all three Baltic states (Table 2.3) a strong upward trend can be observed in the size of the middle class accompanied by a downward trend in the size of the lower class. Nevertheless, even by 2012 the gap with the old EU member states remained substantial. Thus in our view the development of a strong middle class in the Baltic states requires further structural changes in the economy. Apart from the objective indicators, Eurobarometer data enable us to analyse people’s subjective placement in terms of social class. Looking at

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 67

28/09/2016 08:24

68

Europe’s disappearing middle class?

Table 2.3 Size of the middle class based on the occupational approach, Baltic states, 1989–2012 Class by ISCO

Country

1989

1996

2000

2005

2008

2012

Upper (top) and upper middle: ISCO 1

EE LV LT EU15 EE LV LT EU15 EE LV LT EU15

11.3

11.5 7.5

12.4 10.0 8.7 8.5 42.4 42.5 39.3 55.2 45.2 47.4 52.0 36.3

12.7 9.8 8.1 8.9 44.8 45.0 41.7 56.5 42.5 45.2 50.2 34.6

12.5 9.2 10.5 9.0 44.8 47.8 45.6 57.4 42.7 43.1 43.9 33.7

9.3 9.2 8.9 6.2 52.4 52.0 51.9 64.2 38.3 38.8 39.2 29.7

Middle: ISCO 2–5

Lower (working class): ISCO 6–9

36.1

8.4 42.9 38.6

52.4

53.7 45.5 53.9 38.0

Source:  ILO Laborsta, Eurostat, national statistical offices databases.

self-­placement in the social structure,2 we see that the number of people that place themselves in the middle layer of society was stable in 2009–12 in all the Baltic states and at a similar level. In 2009, for example, self-­ placement in the middle was highest in Estonia (54 per cent), followed closely by Latvia (51 per cent) and Lithuania (50 per cent), while in 2012 the numbers were 47 per cent in Estonia, 52 per cent in Latvia and 55 per cent in Lithuania. Interestingly, if we look at the ‘Marketing House’ survey ‘Class self-­identification in Latvia’ data, these numbers have not changed since 2005. The upper class in the Baltic countries made up, on average, 2 per cent and the working class 52.6 per cent of the population. 2.2  Characteristics of Members of the Middle Class In order to obtain more insight into who the middle class (or middle-­ income groups) are, Table 2.4 presents the frequencies of households by the socio-­demographic dimensions of the household head in various income brackets. Many patterns can be explained by the labour market performance of the different socio-­demographic groups. Migrants – ­primarily the Estonian and Latvian Russian-­speaking population, though not all of them are included in that group – are clearly much less represented in the top income group, especially in Estonia and less so in Latvia (­consistent with their higher level of human capital). In Estonia, in particular, households with male heads were more frequent in top and upper middle-­income

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 68

28/09/2016 08:24



Is the world of work stimulating middle-­class growth in the Baltic states? ­69

Table 2.4 Distribution of persons in various income brackets by socio-­ demographic dimensions of the household head, Baltic states, 2004–11 (percentage) Variable

Income class (% of median)

Age 15–34

Education college or above Employee

Male

Migrant

0–60 60–200 > 200 Total 0–60 60–200 > 200 Total 0–60 60–200 > 200 Total 0–60 60–200 > 200 Total 0–60 60–200 > 200 Total

Estonia

Latvia

Lithuania

2004

2011

2004

2011

2004

2011

31 23 33 27 17 29 59 31 42 84 97 76 44 53 63 52 14 16 3 13

29 27 40 29 21 37 74 37 36 80 92 71 46 45 55 46 11 8 3 8

18 21 28 21 7 21 55 22 36 71 85 65 41 38 47 40 15 14 8 13

20 22 25 22 8 26 61 27 44 66 80 62 46 43 45 44 12 13 8 12

34 36 42 36 6 22 61 24 32 64 80 59 45 44 50 45 0 1 1 1

39 31 36 34 11 28 72 29 36 65 85 60 46 48 47 48 1 1 1 1

Source:  Authors’ calculations from EU-­SILC database with the assistance of Nicolas MaÎtre of the ILO.

groups, especially in 2004, probably partly related to the highest gender pay gap in Europe. The link with the tertiary education of the household head is, as expected, strong and increasing over time (in Estonia it rose from 59 per cent in 2004 to 74 per cent in 2011); the link is weaker in Latvia but it may be that the same development is occurring there but with a time lag. In terms of age, the patterns are different across countries: people aged 35 or below (used as a broader definition of young people) have increased their share in the four upper brackets in Estonia, while their share has decreased in Latvia and Lithuania. Perhaps part of the explanation is higher emigration of young people in Latvia and Lithuania.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 69

28/09/2016 08:24

70

Europe’s disappearing middle class?

Table 2.5 Labour market dynamics of different social classes defined by occupations, Estonia, 2008–11 Indicator Middle and year (ISCO 1–4)

Upper-­ Middle Lower Upper-­ Middle Lower Work (ISCO (ISCO (ISCO 1) (ISCO (ISCO middle (ISCO 6–9) 2–5) 6–9) 5–9) (ISCO 1–2) 3–5)

Average hourly wages, EUR 2008 3.78 2.92 2009 3.76 2.69 2010 3.72 2.70 2011 3.93 2.77

4.21 4.16 4.06 4.25

2.88 2.83 2.80 2.98

3.06 2.83 2.82 2.90

Probability of becoming unemployed over the year 2008 1.4 2.5 0.9 2.3 2.6 2009 5.6 8.8 4.6 7.1 9.2 2010 4.9 10.3 5.0 7.5 10.0 2011 3.3 4.3 2.4 5.5 3.7

4.50 4.36 4.37 4.47

3.22 3.21 3.21 3.42

1.2 5.3 5.6 2.3

1.7 6.1 6.6 4.3

3.06 2.83 2.82 2.90 2.6 9.2 10.0 3.7

Source:  Authors’ calculations based on Estonian Labour Force Survey (hereafter LFS) database.

2.3 Labour Market Performance, Subjective Well-­being and Material Conditions of the Middle Class In the Baltic states no labour market segment or income group was unaffected by the crisis, but certain segments were harder hit than others, such as males, young people and non-­natives (Masso and Krillo 2011). The boom preceding the crisis accompanied by increased wages and easily available loans in the Baltic countries allowed more people to enjoy a middle-­class lifestyle. Looking at job and wage dynamics during the crisis using Estonian Labour Force Survey data (Table 2.5) and classes based on occupations, the middle class clearly suffered less during the crisis: their average wages decreased less (in 2009 −0.6 per cent for the middle class and −7.7 per cent for the working class) and the probability of becoming unemployed increased less. In 2010, the average wages of the middle class continued to decrease, but in 2011 they already exceeded their 2008 level. Their better labour market performance is probably related to the fact that many of them work in the public sector. Concerning subjective self-­evaluations, using Eurobarometer data, the impact of the crisis on those in the middle of the social structure has been stronger than on those at the top and less profound than on those at the bottom of the social ladder (Table 2.6). People at the middle of the social structure suffered comparatively less in Latvia than in Estonia and

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 70

28/09/2016 08:24



Is the world of work stimulating middle-­class growth in the Baltic states? ­71

Table 2.6 Impact of the crisis on different social groups, Baltic states, 2012 (percentage) Indicator

Estonia

Latvia

Lithuania

Lowest Middle Highest Lowest Middle Highest Lowest Middle Highest Crisis impact on personal situation Very 30.9 18.9 42.0  significant impact Fairly 43.8 44.0 44.0  significant impact Not really 27.1 10.1 17.7   any impact No impact 7.5 9.9 3.9   at all You/partner lost job Yes, as 35.4  a direct consequence of the crisis Yes but not 11.5  as a direct consequence of the crisis No, it has not 53.1  happened at all

44.6

20.8

16.2

42.9

29.4

14.7

36.9

47.5

32.2

36.8

42.4

44.7

14.1

27.0

38.0

14.2

21.1

29.9

4.4

4.7

13.6

6.0

7.1

10.6

20.6

12.3

38.8

28.2

23.4

33.4

27.0

16.4

10.2

5.0

10.3

8.8

6.2

7.1

6.2

13.0

69.1

82.7

50.9

63.0

70.4

59.5

66.8

70.6

Note:  The lowest level of society are those with score of 1–4, the middle level those with a score of 5–6 and the highest level those with a score of 7–10. Source:  Eurobarometer 2012/9.

Lithuania. This can partly be explained by the fact that part of the ‘middle class’ are pensioners, and pensioners’ income was relatively unaffected in Latvia.3 At the same time, in Estonia the middle class lost jobs less often as a consequence of the crisis than in Latvia or Lithuania. Looking at social class in terms of occupation, the conclusions are similar. People whose situation was affected by the crisis the most – according to their own ­judgement – were unskilled manual workers, followed by skilled manual workers, whereas general managers, directors or top managers, or supervisors, as well as those working in agriculture, were least affected. In Latvia the incomes of people working in the private sector have since recovered and now even exceed pre-­crisis levels. This is not the case,

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 71

28/09/2016 08:24

72

Europe’s disappearing middle class?

however, for those working in the public sector, such as teachers, doctors and policemen. Spending, cut as the crisis began, has not returned to previous levels, which means that, unlike in western European countries, in Latvia many public sector employees cannot be considered part of the middle class in economic terms. Another inequality emerging in the Latvian labour market is that while before the crisis the rapid economic growth diminished the differences in at-­work poverty risk between men and women, since the crisis inequalities have been increasing again. This means that working women have become more likely to lose their middle-­ class position. The slight increase in self-­placement in the middle class has coincided with a slight improvement in the self-­evaluation of households’ financial situation and their personal job situation. Compared with 2009, in 2012 people in the middle of the social structure had a much more positive outlook for the future. At the same time, their overall life satisfaction had not improved, which might partly be related to the fact that the middle class found it increasingly difficult to pay their bills. Importantly, in 2014 even more middle-­class people in the Baltic countries were dissatisfied with their lives and struggling to pay their bills. As noted by Bourdieu (1979), belonging to a higher social class requires a certain lifestyle, taste and style. Cultural capital indicates a person’s status and distinguishes them from the lower classes. However, formation of cultural capital requires a lot of resources, which may put a strain on people’s budgets, limiting life-­choices and raising stress levels. Currently, in the Baltic countries, 17 per cent of people are very satisfied, 66 per cent fairly satisfied, 15 per cent not very satisfied and 2 per cent not at all satisfied with their life; 7 per cent found it difficult to pay their bills last year most of the time, and more than 31 per cent did so from time to time. There are large differences between countries, however. In Estonia only 21 per cent of the middle class struggle to pay their bills, while in Latvia and Lithuania this number is 34 per cent and 48 per cent, respectively. Some information on the material conditions of the middle class can also be found in the statistical data (Table 2.7). For example, if we look at household disposable income per household member in Latvia, we can see that the incomes of the third quintile were affected little by the crisis compared with those of the fourth and fifth quintiles. However, the effect is much larger at the household level. Here we can see that incomes in all top three quintiles significantly fell during the Great Recession and have been slowly recovering since. Overall, the data indicate a slight decrease in the level of inequality in Latvia, which is also supported by other indicators. In terms of severe material deprivation, in Estonia the middle quintiles differ significantly from the first quintile, while in Lithuania and especially in

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 72

28/09/2016 08:24



Is the world of work stimulating middle-­class growth in the Baltic states? ­73

Table 2.7 Evolution of net disposable income and severe material deprivation by income quintiles, Baltic states, 2005–13 Indicator

Net disposable income Severe material deprivation

Quantile

1 2 3 4 5 1 2 3 4 5

Estonia

Latvia

Lithuania

2005

2008

2013

2005

2008

2013

2005

2008

2013

134 224 304 415 737 32 18 9 3 1

235 379 516 697 1176 16 5 2 1 0

234 421 599 857 1542 20 8 6 3 0

56 102 136 198 407 67 54 39 25 11

112 201 281 411 797 48 24 14 8 3

125 228 295 413 780 49 31 25 10 5

53 93 122 163 307 65 42 30 20 6

109 177 231 304 562 30 19 9 4 1

119 207 270 367 666 39 21 14 5 3

Note:  For Lithuania average income and for Estonia and Latvia median income is reported. Source:  Eurostat, national statistical offices databases.

Latvia the second and third quintiles have non-­negligible material deprivation. Thus these numbers perhaps add some additional evidence that the Estonian middle-­income groups are more likely to satisfy the other conditions than those in Latvia and Lithuania. Thus perhaps the middle class is also probably the most vulnerable in Latvia. Concerning the additional evidence for the position of middle-­income groups in terms of consumption in Latvia, data from the country’s Central Statistical Bureau show that the crisis has led to a substantial increase in the number of people who cannot afford to pay arrears on mortgage or rent, utility bills or hire purchase instalments (from 10 per cent in 2007 to 21 per cent in 2011) and the increase was especially acute among the first and second quintiles (from 10 to 27 per cent). When the crisis began, many households tried to maintain their living standards by taking out loans or postponing payments of utility bills. The accumulated debt will be a significant burden for these households and limits their chances of (re)joining the middle class in the near future. The high percentage of people not able to face unexpected financial expenses is also worrying, especially in Latvia (72 per cent in 2013, Statistics Latvia). It means that many households in Latvia are just one serious accident or mishap away from falling into poverty. The middle class – if we can argue that it exists in Latvia – is very fragile and vulnerable.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 73

28/09/2016 08:24

74

Europe’s disappearing middle class?

3. LONG-­TERM TRENDS IN THE WORLD OF WORK IN THE BALTIC STATES 3.1  Industrial Structure of Employment Developments in inequality and the relative position of the various social classes can be related to labour market developments, which may apply particularly to the middle class owing to its higher share of labour income. Transition processes brought fundamental changes to the sectoral and industrial composition of employment in the Baltic states (Eamets 2001). Serious employment losses were experienced especially in the industrial sector (in particular in Latvia and Lithuania) and in agriculture. The decline of agriculture was most dramatic in Estonia, while it remained relatively stable in Lithuania for a long period (no change during 1992–97), possibly as a kind of subsistence farming. Lithuania also suffered heavily from employment fall in the industrial sector, 64 per cent during 1992–97. While the employment decline in agriculture – especially in the early transition – could be attributed to falling output, especially in the case of Estonia where it was also due to increasing efficiency. Estonia experienced the biggest employment increase in the services sector: during 1991–98 the share of services in employment increased from 43 per cent to 58 per cent (Eamets 2001). By 2008 Estonia had the lowest share of agricultural employment (1.6 per cent) and the highest share of manufacturing employment (21.1 per cent), while the service sector shares are similar across the three countries (Table 2.8). Despite these developments, the tertiary sector share is below that of the old EU member states (73 per cent in 2010 compared with 65–67 per cent in the Baltic states; Eurostat). Looking at employment dynamics at a more disaggregated level – for example, in the 1-­digit industries – fairly substantial developments can also be spotted. The permanent employment decline in primary sector industries was attributed to the unreasonably high level of initial employment (Eamets 2001). However, a rapid increase can be seen in certain service sectors, such as financial ‘intermediation’, trade, real estate and business, which are part of a market economy, but were underdeveloped in the command economy; for example, in the command economy there was virtually no real estate market (Eamets 2001). Also, some of these have benefited more from the development of private business owing to lower capital requirements. Concerning the public sector, building structures needed for an independent country may have contributed to the growth of jobs in public administration. Financial ‘intermediation’ started from a fairly low level (0.5 per cent in 1989) and showed the highest growth rates. During market the period 2001–07 employment increased heavily in domestic-­

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 74

28/09/2016 08:24



Is the world of work stimulating middle-­class growth in the Baltic states? ­75

Table 2.8 Changes in employment levels by industries, 1989–2008, Baltic states Industry

Estonia

Primary sector A Agriculture,  hunting and forestry B Fishing C Mining and  quarrying Secondary sector D Manufacturing E Electricity, gas   and water supply F Construction Tertiary sector G Wholesale and   retail trade H Hotels and  restaurants I Transport, storage  and communications J Financial  intermediation K Real estate,  renting and business activities L Public  administration and defence; compulsory social security M Education N Health and   social work O Other community,  social and personal service activities

Latvia

Lithuania

1989

1998

2008

1998

2008

1998

2008

7.5 18.0

3.4 8.1

1.6 3.7

9.5 18.6

2.7 7.8

6.5 19.1

2.7 7.7

3.2 1.5

0.8 1.2

0.2 0.9

0.5

0.2 0.3

0.1 0.2

0.3 0.3

11.9 25.6 2.2

10.6 21.7 2.8

11.5 21.1 1.4

8.8 18.6 2.4

9.4 15.2 1.9

9.5 19.4 2.5

10.1 17.5 1.8

7.7 4.6 7.4

7.3 6.4 14.1

12.2 6.7 14.2

5.5 6.0 14.7

11.2 7.0 16.6

6.7 5.8 14.3

10.9 6.8 18.1

2.2

2.3

3.7

1.8

2.7

1.6

2.6

7.8

9.1

8.5

8.0

9.4

6.7

6.9

0.5

1.3

1.6

1.2

1.7

1.2

1.3

4.0

6.2

7.9

3.4

7.0

2.8

6.7

3.9

5.7

5.8

6.8

7.7

4.9

5.5

6.1 6.0

8.9 5.6

9.1 4.8

8.5 5.3

8.0 4.9

9.6 6.8

9.8 6.3

3.6

4.8

4.9

4.5

5.1

4.1

4.3

Source:  ILO Laborsta database.

oriented sectors, such as construction, wholesale and retail trade, and business services (Roosaar et al. 2012). Later, many of the jobs created during that period were lost again because, in addition to hitting the open sector of the economy (manufacturing), the Great Recession hit the domestic

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 75

28/09/2016 08:24

76

Europe’s disappearing middle class?

market-­oriented sectors of the economy due to the reduction of lending from Scandinavian banks (Roosaar et al. 2014). Thus, during the real estate boom construction was the biggest job creator and during the recession it experienced the biggest employment decline. 3.2  Occupational Structure The Baltic states have experienced profound changes in occupational structure over the past 20 years. The biggest changes occurred in the early transition – for example, in Estonia, during 1989–95 between 35 per cent (occupation at 1-­digit level) and 50 per cent (occupations at 4-­digit levels) of Estonian workers changed their occupation (Campos and Dabušinskas 2009, based on LFS data). About 10–30 per cent (depending on the level of occupation and the year) of these were due to net flows, that is, changes in the structure of occupations. For the later period, calculations by Roosaar et al. (2014) from the CV-­Keskus data indicated smaller net occupational flows similar to western European countries: the occupational mobility rate declined from 6 per cent in 1997 to 3.2 per cent in 2001, increased again to 4.3 per cent in 2006 and then decreased to 2.7 per cent in 2010. Thus, occupational mobility was higher during the recovery and boom periods than during the recession. Among different occupations, low-­skilled white-­ collar employees had a higher chance of changing their occupation than high-­skilled white-­collar employees (Roosaar et al. 2014). During 1991–98 the odds were very low for young adults in blue-­collar jobs who wanted to move to higher positions in the Baltic states, while skilled workers had a substantial probability of moving into the stratum of unskilled workers, and the opening up of business opportunities allowed young adults to compete for managerial positions on an equal footing with adults. Our calculations presented in Table 2.9 using the data on people’s employment histories from the job search portal CV-­Keskus show that there is no evidence of reduced mobility across the broad occupational groups corresponding to different social classes; rather we see lower probabilities of remaining in the same group. There are increased probabilities, for example, for professionals to become managers, and for ‘middle’ professions (ISCO 3–5) to become professionals or managers. The changes in the occupational structure (Table 2.10) reflect the changes in the industrial structure of the economy: a decreasing share of skilled agricultural and fishery workers (during 1989–2008 from 5.1  per  cent to 1.9 per cent), an increasing share of service workers (6.7  per cent to 12.4 per cent), an increasing share of technicians and associate professionals (10.5 per cent to 13.1 per cent), and decreasing craft and related workers (21.9 per cent to 16.9 per cent). The share of elementary occupations

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 76

28/09/2016 08:24



Is the world of work stimulating middle-­class growth in the Baltic states? ­77

Table 2.9 Mobility between broad occupational categories over four years, Baltic states 1996–2009 (percentage) Starting year

End year

Occupation

1996

2000

2001

2005

2005

2009

ISCO 1 ISCO 2 ISCO 3–5 ISCO 6–9 ISCO 1 ISCO 2 ISCO 3–5 ISCO 6–9 ISCO 1 ISCO 2 ISCO 3–5 ISCO 6–9

ISCO1

ISCO2

71.1 3.1 4.1 4.1 66.1 3.9 3.7 3.9 59.2 5.0 4.9 5.1

2.2 80.6 1.2 0.7 2.6 76.8 1.5 0.8 3.3 69.3 1.4 0.9

ISCO 3–5

ISCO 6–9

16.5 13.2 87.6 10.1 20.1 15.1 87.2 11.8 23.5 20.2 82.7 15.0

10.2 3.2 7.1 85.1 11.1 4.2 7.7 83.6 13.9 5.5 11.1 79.0

Source:  Authors’ calculations based on CV-­Keskus database.

Table 2.10 Occupational structure among the employed, Baltic states, 1989–2013 (percentage) Country Year EE EE EE EE LV LV LV LT LT LT

1989 1998 2008 2013 1998 2008 2013 1998 2008 2013

1-­digit ISCO code 1

2

3

4

5

6

7

8

9

11.3 14.1 12.5 9.7 7.7 9.2 9.9 10.4 10.5 8.9

13.2 11.1 13.9 19.2 11.9 14.2 17.1 13.4 18.0 23.1

10.5 13.4 13.0 13.1 11.6 16.4 12.6 7.6 11.0 11.1

5.7 4.5 5.4 5.9 5.2 4.6 5.7 5.7 4.6 3.8

6.7 11.1 12.4 13.1 12.0 12.5 15.7 8.6 12.0 14.1

5.1 4.9 1.9 1.7 12.2 4.1 3.5 13.8 5.4 5.9

21.9 16.9 16.9 14.3 15.0 15.0 12.8 18.7 18.9 14.3

17.6 12.8 14.3 13.7 10.8 10.1 9.9 10.5 10.4 11.0

7.8 11.2 9.6 9.3 13.5 13.9 12.6 11.4 9.3 7.9

Note:  1 Legislators, senior officials and managers; 2 Professionals; 3 Technicians and associate professionals; 4 Clerks; 5 Service workers and shop and market sales workers; 6 Skilled agricultural and fishery workers; 7 Craft and related trade workers; 8 Plant and machine operators and assemblers; 9 Elementary occupations. Source:  Eurostat, national statistical offices databases.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 77

28/09/2016 08:24

78

Europe’s disappearing middle class?

increased from 7.8 per cent to 11.8 per cent in 2001 and then decreased to 9.6 per cent. The changes were most rapid during the transitional crisis period in the early 1990s, while across the last business cycle (2000–2013) changes were more modest. These trends have also continued since 2008, the most noticeable development being the increasing share of professionals, from 13.8 per cent in 2008 to around 19 per cent in later years, probably related to the growing number of jobs in knowledge intensive services (Espenberg 2013). The dynamics and higher number of managers in Estonia compared with some other countries can be explained by ‘bottom-­up’ privatization (Titma and Roots 2008). Similar tendencies in the occupational structure can be observed in Latvia: the number of highly skilled professionals has increased, as well as the percentage of service and sales workers, whereas very few workers nowadays are employed as skilled agricultural, forestry or fishery workers. Also note that the number of craft and related trades workers decreased significantly during the crisis in Latvia, while the number of people working in elementary occupations has remained comparatively stable over time. As a result of these developments, the share of white-­collar workers increased both in the private and public sectors from 40 per cent in 1989 to 48 per cent by 2013. Thus, in terms of the middle class defined in terms of occupations the shifts are fairly favourable. However, the structure is lagging behind more developed western European countries. The size of the middle class could, of course, also be related to the extent of self-­employment and entrepreneurship. Especially in conditions of high wage inequalities, high labour and relatively low capital taxation in relation to the middle class (not just the middle-­income group) could depend on whether one is in business or not. The developments in the occupational structure are naturally related to the development of entrepreneurship and self-­employment. In all the Baltic states, but especially in Estonia, self-­employment rates are fairly low: in 2013, these were 8 per cent in Estonia and 10 per cent in Latvia and Lithuania (compared with 14 per cent in the old EU member states). The self-­employed are a fairly heterogeneous group; the percentage of blue-­ collar occupations is quite high among the self-­employed without employees (77 per cent, average during 2005–12, according to Estonian LFS data), but among self-­employed salaried entrepreneurs around 82 per cent have white-­collar occupations (mostly managers) (Masso and Paes 2015). There are, however, further heterogeneities here as entrepreneurs with only a few employees are similar to individual self-­employed and there are also ‘bogus self-­employed’; for example, the large growth in self-­employment in the construction sector during 2002–04 may have reflected the growth

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 78

28/09/2016 08:24



Is the world of work stimulating middle-­class growth in the Baltic states? ­79

in forced self-­employment. The Global Entrepreneurship Monitor (GEM) indicates fairly high levels of entrepreneurship in Estonia compared with other European countries and entrepreneurs there are at the top of the income distribution (Global Entrepreneurship Monitor 2012, the Estonian Report 2013). The same is revealed by the European Working Conditions Survey in 2010; among the self-­employed and salaried employees 19 and 16 per cent, respectively, were among the upper 20 per cent in terms of earnings. 3.3  Developments in the Labour Supply Besides changes in labour demand, major changes have also occurred in labour supply during the transition. Similar to other European countries, but more profoundly, the population has been falling since the 1990s. The Baltic states have experienced a tremendous population decline since regaining independence (Schlitte and Stiller 2007). The declining of the population has been the result of both the high emigration and low fertility rates. The latter had been caused by economic decline, uncertainty about the future, new individual challenges and opportunities, such as for higher education, and high labour force participation among women (Schlitte and Stiller 2007). Especially among males, life expectancy decreased and mortality increased at the beginning of the transition. The period since 2000 has been marked especially in Estonia by increasing fertility and declining mortality. As a result, after the transition, life expectancy has increased considerably in the Baltic states, related to improved health care, working and housing conditions, and economic development. In addition to demographics the labour supply has also been shaped by labour migration. The level of emigration has been particularly high for Latvia and Lithuania, possibly because of the higher wages in Estonia, but also because many Estonians commute between Estonia and Finland due to their proximity. With regard to the impact on the middle class, it is important to note who the migrants are and what the impact of temporary employment abroad is on performance in the home country labour market after they return. This issue is analysed in a case study in section 4. The labour force participation rates decreased in the Baltic states from the beginning of the 1990s and have stabilized since 2000 at levels close to those of the old EU member states, at around 70 per cent (Eamets et al. 2007). An educated citizenry forms the basis of a stable middle class. In the literature it is controversial whether education may be used to construct skill variables (Oesch 2006). However, it is linked with commonly used occupational data and the occupational classification ISCO-­88 has been designed

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 79

28/09/2016 08:24

80

Europe’s disappearing middle class?

around the skills (education and training) required to perform certain jobs. During the past three decades enrolment in education has been stable in Estonia. However, a clear shift towards increasing participation in higher education has also been observable (see Table 2.11) and enrolment in tertiary education has almost tripled (Masso et al. 2013). Participation in education has been higher for females compared with males in the Baltic states and has increased considerably in all three Baltic states overall. As the result, the educational level of the population is higher than the EU average in the Baltic states. The issue thus seems to be whether the economy is able to make use of the relatively high potential of the labour force, for example, by offering jobs at appropriate skill levels. Failure to do so may limit the growth of the middle class. During the past ten years the proportion of those with a lower-­ secondary level of education has decreased in all the Baltic states. In terms of educational inequality Paulus (2004) found that during 1959–2000 the educational Gini decreased by almost half, from 0.330 to 0.183. There are differences in average study time by place of residence, human capital is concentrated in the cities, and gender and ethnic differences are smaller. The educational Gini, however, is higher for those living in rural areas, for women, for Estonians and for the elderly. 3.4 Wages Developments in wages and wage inequalities to a large extent determine the size and position of the middle class. In the Soviet command economy, wages were set centrally and were not directly linked to the supply and demand of skills, including education. Thus there were virtually no returns to education during that time. However, overall wage inequality was not particularly low at the beginning of the transition owing to the presence of other wage inequalities, although it increased thereafter owing to the increasing inequality in the lower part of the wage distribution. Wage inequalities decreased during the Russian crisis (1998–99) and the following recovery (until 2005). Wage inequality dynamics in the Baltics have been affected by the high flexibility of wages; that is, there were wage cuts in Estonia both in response to the Russian crisis (Masso and Eamets 2007) and during the Great Recession (Masso and Krillo 2011). In 2012 Estonia had one of the highest levels of wage inequality among the Organisation for Economic Co-­operation and Development (OECD) countries in terms of the D9/D1 ratio (4.1), as well as one of the highest incidences of low pay (24.5), although these are even higher in Latvia and Lithuania (4.6 and 4.3 in 2010, respectively; Eurostat). Owing to high unemployment rates and very weak trade unions, there

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 80

28/09/2016 08:24

81

VAUGHAN 9781786430595 PRINT (M4051).indd 81

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

16.5 29.5 16.7 39.8

Females EU27 EE LV LT

16.6 30.3 17.7 22.3

17.6 19.2 12.5 15.7

2001

Source:  Eurostat database.

17.8 18.2 13.4 30.5

Males EU27 EE LV LT

2000

17.1 30.3 18.3 22.0

17.7 19.2 13.9 15.7

2002

17.9 30.5 17.6 22.8

18.3 19.4 12.4 16.6

2003

19.1 31.0 19.8 23.7

19.1 19.9 13.3 18.5

2004

19.7 32.4 20.7 25.2

19.5 22.0 13.3 18.7

2005

20.4 32.6 21.5 25.8

19.7 21.4 13.4 18.6

2006

21.1 33.2 22.6 27.8

20.1 20.8 14.8 20.2

2007

21.9 34.5 26.0 29.6

20.5 21.1 15.7 20.8

2008

22.9 37.6 27.4 30.0

21.1 21.4 15.6 20.7

2009

23.8 37.3 28.1 32.4

21.7 21.5 16.6 21.3

2010

24.8 38.4 29.9 33.7

22.4 23.1 17.0 22.5

2011

26.0 39.7 31.7 34.4

23.3 24.3 18.0 22.4

2012

27.0 40.1 33.8 35.5

23.9 24.0 19.6 23.7

2013

27.5 40.7 33.9 36.9

24.5 24.3 19.3 25.6

2014

Table 2.11 Tertiary education by gender, age group 15–64, Baltic states, 2000–2014 (percentage of corresponding age group)

82

Europe’s disappearing middle class?

Table 2.12 Indicators of wages and wage inequalities, Baltic states, 1988–2013 Indicator Country 1988 Average  wage (EUR) Gender  pay gap Labour  share (%)

EE LV LT EE LV LT EE LV LT

– – – 34 – – – – –

1993

1996

1998

2003

2008

2011

2013

68 67 – 29 – – – – –

191 140 179 27 – 22 – 44.5 –

257 190 269.3 26 20 22 – 41.9 –

430 274 310.6 24 16 17 44.4 38.2 –

825 682 623.2 26.6 – 20.5 49.9 47 43.9

839 949 660 716 592.5 646.3 25.8 28.2 11.4 12.4 11.6 12.5 45.5 46.5 39.3 41.2 39.4 39.2

Source:  Eurostat, national statistical offices databases. Gender pay gap data for Estonia in 1988 are from Noorkõiv et al. (1997).

was little pressure to increase wages at the beginning of the 2000s (Masso et al. 2014b). The pressure on wages increased with EU accession in 2004, and especially during 2006–07 when unemployment had fallen to extremely low levels (Table 2.12). As people found it much easier to find a job and move abroad, employers were forced to increase wages in order to keep their most qualified employees. However, this situation did little to decrease inequality. On the contrary, the most qualified workers were both the most mobile and the most highly valued by their employers. Therefore, free movement of labour benefited them the most (‘Human Development Report Latvia 2006/2007’), further increasing the gap between them and other society groups. At the beginning of the transition there was a tremendous increase in returns to education: in only five years the wage premium for university-­ educated workers (compared with primary-­educated workers) rose from 11 per cent to 69 per cent. The results of the Mincerian wage equation either from earlier (Hazans 2003) or later (Anspal et al. 2011) show that returns are especially high for those with tertiary education, and relatively less for secondary education, the latter being relatively low compared with developed economies and other central European countries (Hazans 2003). Among the different wage disparities, the gender pay gap has received the most attention in the Estonian context, also because it is at the highest level among the EU member states: around 25–30 per cent (although varying somewhat across years and surveys). To some extent this was inherited from the Soviet system, related to gender segregation and promotion policies. The downward trend in the gender pay gap up to 2005

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 82

28/09/2016 08:24



Is the world of work stimulating middle-­class growth in the Baltic states? ­83

reflected the increasing value attributed to higher education; in general, females have had higher levels of education compared with males (Rõõm and Kallaste 2004). The gender-­pay gap also decreased during the Great Recession owing to its severe effect on male-­dominated industries (Masso and Krillo 2011). While inward migration after regaining independence has been limited, the large-­scale immigration of Russian-­speakers during the Soviet period has resulted in a large non-­native population in Estonia and Latvia, whose labour-­market performance tends to be below that of the native population 3.5 Implications of Labour Market Institutions for the Middle Class in the Baltic States While social protection spending in the Baltic states is generally at a fairly modest level, it cannot be said the middle class are disadvantaged by this. The share of means-­tested benefits has been very low in all the Baltic states: in 2010 the figure was 0.9 per cent in Estonia, 5.3 per cent in Latvia and 4.1 per cent in Lithuania (despite occasional proposals to convert some universal benefits into means-­tested benefits). These levels are similar to the Nordic social model, though the Baltic states have been described as more similar to the liberal social model, albeit with strong elements of the continental social model. Employers’ contributions form a high proportion of total receipts through employers’ earmarked social contributions used to finance pension and health insurance (Masso et al. 2014b). Thus social benefits are not biased towards low-­income deciles and against middle-­income groups. When comparing Gini coefficients net and gross of income transfers or at-­risk-­of-­poverty rates before and after social transfers, it is clear that in Latvia and Lithuania the state has done a lot to smooth market-­related inequalities (Masso et al. 2014b). Estonia has a relatively small public sector, owing to the dominance of right-­wing parties in government and relatively low revenues. However, it cannot be said that, as a result, the provision of public services is also insufficient. For instance, Estonia has been able to establish one of the most efficient health-­care systems in the Baltic states. On the other hand, it is also clear that many austerity measures during the crises critically lowered the provision of certain services, such as health care, childcare and rescue services (Masso and Espenberg 2013). The other important element is the tax system. The low level of social protection expenditures is also due to the lower level of tax revenues. All the Baltic states, especially Latvia and Lithuania, but to an increasing extent also Estonia, rely relatively strongly on indirect taxes. The high share of wage taxes in the Baltic states reflects the important role played

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 83

28/09/2016 08:24

84

Europe’s disappearing middle class?

by wage-­ based contributions in financing the social security system. Relatively high labour taxes are partly inherited from the Soviet system but also reflect the lower taxes on capital introduced to attract foreign direct investment (Võrk et al. 2007). This tax system also means that most entrepreneurs prefer to provide additional payments in the form of dividends rather than wages: the former are taxed at 21 per cent, the latter at 61 per cent (Tartu Ülikool and Praxis 2010). The Baltic states have the lowest taxes on capital among the EU27 countries. Another feature of the tax systems of the Baltic states is the flat income tax (subject to tax-­free allowances). Estonia was the first to introduce a flat tax in 1994, followed by Latvia in 1994 and Lithuania in 1997. With the exception of Latvia, the introduction of a flat tax was associated with an increase in tax-­free allowances to limit the tax increase on low-­paid individuals. For the current study this could imply that the self-­employed or people with capital incomes are more likely to be able to enter the middle class, less so households dependent solely on labour income, especially if the middle class is not defined based on the middle-­income deciles. Various taxes have naturally affected income deciles differently. Võrk et al. (2007) analysed the impact of taxation policy on the distribution of household tax burdens (measured either as the share of taxes in income, share of taxes in expenditure or the Kakwani index), using the Estonian household budget survey data from 2000 to 2007 and a micro-­simulation model. They concluded that the gradual reduction in the income tax rate in Estonia from 26 per cent in 2000 to 21 per cent in 2008 benefited people in higher income deciles more, while the increase in the tax threshold (and also the total effect) favoured those in the middle deciles, although in absolute terms the gains were greatest for people in the upper deciles, and thus this move has increased inequality in net incomes. Income tax deductions (household loan and student loan interest payments, educational expenditure, payments to the third pension pillar) benefit people in the higher deciles more. Another important feature of the Baltic social model or its labour market institutions is the low (and declining) importance of unions and social dialogue in general. Union density in all the Baltic states is currently the lowest in the EU27. In 2009 only 7 per cent of workers in Estonia, 10 per  cent in Lithuania and 15 per cent in Latvia were members of a trade union (Armingeon et al. 2011). Concerning different occupations, in Estonia the unionization rates were somewhat higher among middle-­class professions: 9.5 per cent among ISCO 2–5 occupations and 5.5 per cent among ISCO 6–9 occupations (authors’ calculations from LFS data). The low unionization is one factor behind the high wage inequality, potentially reducing the size and income position of the middle-­income groups. However, the question is whether that has any special implications for the Baltic states

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 84

28/09/2016 08:24



Is the world of work stimulating middle-­class growth in the Baltic states? ­85

if middle classes generally have a lower level of unionization. Oesch (2006) showed that in all analysed countries (the United Kingdom, Germany, Sweden and Switzerland) union membership was high among sociocultural professions and semi-­professionals, and below average among managers, office clerks and routine service workers. Thus, it can be said that, given this definition, middle-­class professions have the highest rate of unionization in Estonia, even though the numbers are still rather low. The relatively low minimum wages and the low coverage of collective agreements may mean that wage floors do not affect the position of the middle class much, but possibly only employees at the bottom of the wage distribution. However, while generally social dialogue, unionization and collective agreements are limited in the Baltic states, they tend to be higher in the public sector than in the private sector. In the private sector in Latvia the sectors in which employers are comparatively active in the development of social dialogue are metalworking, the wood industry and food production, whereas the problematic sectors include the chemical industry, production of electrical and optical equipment, construction, transport and logistics (Ozola 2010). Collective agreements are more common in areas where there are large privatized companies, such as the energy sector, transport and food production (LDDK 2014) and – in the public sector – ­education and health care. The most active social dialogue concerns effective governance, education and employment, while social dialogue has so far been weakest in the areas of national economy and social and health care. In Latvia, social dialogue is underdeveloped and could definitely be improved in all areas and sectors. According to the Latvijas Darba Devēju Konfederācija (Latvian Employers’ Confederation, LDDK) the social dialogue between social partners and state is especially insufficient in the annual state budgeting process. The following problems hinder unionization and social dialogue: unwillingness of employees to join unions; membership fees; difficulty of finding volunteers to lead a new non-­governmental organization (NGO); a lack of understanding of the meaning and importance of social dialogue among some employers, employees and municipal representatives; concentration of industries in Riga; and uneven economic development (LDDK 2014). Situations in which agreements with social partners are not respected are common in Latvia, thus the challenge is how to move from a formal consultation process to effective involvement of the social partners in the decision-­making process (Ozola 2010). However, some successful cases of social dialogue can be mentioned. Entrepreneurs’ advisory councils operate in several larger cities (Jelgava, Valmiera and Liepaja). Among the oldest is the Liepaja city Tripartite Cooperation Council, established in 2009, which aims at cooperation

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 85

28/09/2016 08:24

86

Europe’s disappearing middle class?

between Liepaja municipality, employers and the Liepaja trade union in order to ensure coherent problem solving, consistent with the interests of the city as regards social and economic issues that would guarantee social stability and increase prosperity in the city, enhancing the co-­responsibility of social partners for decisions and their implementation (Ozola 2010). Besides reaching its main goals, the council also ensures that social partners have an opportunity to voice their concerns and opinions about the planned decisions of the municipality and their impact on their work, thereby preventing the adoption of decisions that could have a negative impact not only on employers and workers but also the overall development of the municipality. It allows the employers, too, to defend their interests and, as partners in the social dialogue, to participate in the preparation and adoption of certain municipal decisions, preventing decisions that would be harmful to entrepreneurs and making it possible to find a solution that would be acceptable to all three parties. One of the most common forms of cooperation with social partners at the sectoral level in Latvia is the advisory councils. Different advisory councils function in different sectors. One successful example is the State Employment Agency Advisory Board established in 2004. Its aim is to analyse problems and develop proposals to reduce unemployment and to support the unemployed, job seekers and people at risk of unemployment in close cooperation with the representatives of employers, other sectoral bodies, local authorities and NGOs. Among its various activities, the following can be mentioned: information activities for partners to strengthen social partnerships; seminars to inform employers and other partners about cooperation possibilities with the State Employment Agency (SEA); improvement of cooperation with employers at local level; support for strengthening the advisory boards established in the regional branches of the SEA; and development of action plans for promoting employment in cooperation with municipalities (Ozola 2010).

4.  CASE STUDIES 4.1  Case Study 1: Migration in Latvia and its Effects on the Middle Class Latvia is one of the European countries most affected by the Great Recession. Official unemployment rose from 5.3 per cent in late 2007 to 20.5 per cent in early 2010 and the average disposable income per person decreased by 20  per cent. Median real earnings from work fell by nearly 25  per cent during 2009 and stayed at this level at least until June 2012 (Hazans 2013). Many more middle-­class families found themselves facing

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 86

28/09/2016 08:24



Is the world of work stimulating middle-­class growth in the Baltic states? ­87

poverty. Having lost hope that the situation at home could improve (Mierina 2014; Hazans 2011b), many chose to leave. During the years of the crisis (2009–10) the Latvian annual net emigration rate more than doubled, and between 2008 and 2011 the stock of Latvian citizens aged 15–64 who arrived in another EU member state within the past three years increased by 47 per cent (EU-­LFS data, Hazans 2013). Economic factors – desire to earn a decent salary, to improve one’s quality of life and to afford what they see as a middle-­class lifestyle – were among the main drivers of emigration. Since 2000 Latvia has lost 9.1 per cent of its population (Hazans 2013: 66), and even though emigration rates have slowed down since 2012, they remain well above the pre-­crisis level. A worrying characteristic of the latest crisis-­driven emigration is the number of young people as well as whole families emigrating with a view towards permanent employment and residence abroad (Ribakova 2009; Hazans 2016). This raises concerns about the sustainability of the social security system in a country with an ageing society, and about the future potential and competitiveness of the economy, considering the loss of human capital. This could lead to long-­term consequences for the development of the middle class in Latvia. Thus, in our first case study we analyse how migration has affected, or might affect, the situation of the Latvian middle class, both at home and abroad. Study design The main source of information for this case study is the recent large-­scale survey of Latvian emigrants conducted as part of the European Social Fund (ESF) supported research grant ‘The emigrant communities of Latvia: National identity, transnational relations, and diaspora politics’. A total of 14 068 emigrants from Latvia aged 15 and over were quantitatively surveyed during August–October 2014 in 118 different countries. In addition, 150 in-­depth interviews were conducted in the countries in which the Latvian diaspora is the largest: the United Kingdom, Ireland, the United States, Germany, Sweden and Norway, as well as with return migrants and diaspora policy experts in Latvia. In addition, we used secondary analysis of statistical data (Eurostat, Central Statistical Bureau of Latvia and OECD) and previously conducted surveys in Latvia. How many have emigrated and where? According to Central Statistical Bureau (CSB) data, joining the EU in 2004 did not immediately lead to a massive increase in emigration rates from Latvia, but all the statistical accounts and expert estimates show a large increase in long-­term emigration4 from Latvia after 2008, with the beginning of the crisis. The net outflow of people from Latvia during 2000–2010 was approximately 200 000 (Hazans 2011a; Krasnopjorovs

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 87

28/09/2016 08:24

88

Europe’s disappearing middle class?

50 000

All

45 000

EU-28

40 000

EU-15

35 000

EFTA

30 000

NVS

25 000 20 000 15 000 10 000 5000 0

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Source:  CSB Latvia, 2014, database.

Figure 2.1 International long-term emigration by country group, Latvia, 2000–2014 2012), and 70 000–90 000 were lost to migration during 2009–10 alone. Gross outflows of migrants from Latvia during 2009 to 2013 accounted for 9.1 per cent of the population (Hazans 2016). About 70 per cent of long-­term emigrants moved to a western European country, offering much higher living standards and wage levels. Net emigration to Ireland – which was also hard hit by the crisis – decreased during 2009–11, while the United Kingdom became by far the most common destination country for Latvian migrants. Emigration to European Free Trade Association (EFTA) countries (particularly Norway) has increased since 2011, too, reaching about 7 per cent of total emigration (Figure 2.1). The choice of destination countries once more illustrates the point that, by migrating, people aim to improve their income and living standards. Who emigrates? During 2000–2014, 37 per cent of the people who emigrated were 15–29 years old, and this proportion increased to 42 per cent during 2009–11. The most mobile group are young people 20–29 years of age – those who just finished school or university, but have not yet started a family (CSB Latvia data). Approximately three-­quarters of adult emigrants were younger than 35 at the time of their departure (Hazans 2013). One of the reasons for that is that young people were one of the groups that suffered most during the crisis; their employment opportunities were greatly reduced.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 88

28/09/2016 08:24



Is the world of work stimulating middle-­class growth in the Baltic states? ­89

A detailed analysis of emigrant selectivity in terms of education is provided in Hazans (2016). He notes that the share of university graduates among migrants was highest in the pre-­EU accession period. Post-­accession emigrants are, overall, less educated than pre-­accession emigrants, because expected gains were higher for those with a lower level of education. Conversely, by mid-­2014 the share of tertiary educated Latvian emigrants abroad had reached 45 per cent, a much higher share than among the population in Latvia, pointing to a significant ‘brain drain’. When asked about their reasons for emigrating, 19.7 per cent of those who left in 2000 or later mentioned financial difficulties or an inability to make ends meet as the main reason. For 5 per cent, the main reason was their inability to find a job in Latvia. Another 10.4 per cent left mainly because they could earn more abroad and, similarly, 9.7 per cent wanted to improve their quality of life. One can conclude that roughly half of Latvian emigrants left mainly for economic reasons, seeking sufficient income and a middle-­class lifestyle, which is often difficult to achieve for an average worker in Latvia. According to the Labour Force Survey (LFS) data in 2003–13 one-­fifth to one-­third of Latvian guest workers had experienced unemployment or economic inactivity in Latvia during the year prior to their departure, which was especially common during the years of crisis (Hazans 2016). In the 2014 emigrant survey 23 per cent mention their inability to find a job as one of the reasons for leaving Latvia, but for only 5 per cent was it the main reason. We can conclude that low wages and a perceived lack of attractive opportunities, not unemployment as such, are the core reasons for people choosing a job abroad. Social partners in the Baltic states have argued that minimum wage increases could reduce labour emigration (Masso and Krillo 2010). Formation of a stable middle class enabling people to support their families and to ensure their own and their children’s futures would certainly lower emigration rates. Situation of migrants abroad Compared with the general population, the financial situation of emigrants before leaving Latvia was considerably worse (Figure 2.2). If we compare the data on emigrants who left in 2000 or later with the general population aged 16–64 in 2013 (Eurostat data), the most overrepresented group among migrants are those who found it very hard to make ends meet at home (43 per cent among migrants compared with 24 per cent among the general population).5 Migration allowed household income to increase for all income groups (Figure 2.2). For instance, about 85–90 per cent of those who struggled to make ends meet at home now make ends meet fairly easily or even better. Currently

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 89

28/09/2016 08:24

90

Europe’s disappearing middle class?

100 90

24.0

80

1.6 4.3

0.4 4.1

9.5

15.8

16.0

70 60

31.6

2.2 10.2

0.1 7.1 15.8

Easily

29.7 35.8

25.3

50 40 30

24.8

26.1 43.4

10

10.1

0

2.5 1.1 Abroad, 2014

25.3

In Latvia, before leaving, 2000–2014

Inhabitants of Latvia (16–64), 2014

Inhabitants of Estonia (16–64), 2014

With difficulty With great difficulty

15.5 9.7

Fairly easily With some difficulty

35.5

30.7

20

Very easily

26.9

15.1 Inhabitants of Lithuania (16–64), 2014

Note:  Emigrants who left Latvia in 2000 or later. Source:  Eurostat data and ‘The emigrant communities of Latvia’ database.

Figure 2.2 Is your household able to make ends meet? (Pay for the necessities of everyday life), Latvia, 2014 (percentage) only 14 per cent of post-­2000 emigrants find it difficult to make ends meet. This means that owing to emigration more people have been lifted out of poverty and have joined the middle class in terms of income. We define the middle class in two ways: in terms of occupation (profession in Latvia was ISCO 1–4) and in terms of income (made ends meet with some difficulty, fairly easily or easily). As we can see in Table 2.13, a large majority of middle-­class emigrants who left Latvia in 2000 or later now feel more satisfied with their standard of living. They are also more satisfied with their job, working conditions and life as a whole. Only about 6 per cent are less satisfied with their life than they used to be in Latvia. As regards dwelling and family life, at least half of those who left Latvia in 2000 or later are more satisfied than they used to be at home, and just 12–15 per cent are less satisfied. The evaluation of satisfaction with relationships with people outside the family differs; some are more satisfied, but some (about 20 per cent) admit that they miss their relationships with people at home. We can conclude that overall the quality of life of middle-­ class people who emigrate increases, especially in terms of income, job and working conditions, and that the aspect of life that they sometimes sacrifice is the network of people in whose company they felt more comfortable.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 90

28/09/2016 08:24

91

VAUGHAN 9781786430595 PRINT (M4051).indd 91

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

22.4 23.5 27.8 39.5 31.3 16.8 20.0

66.7 67.0 59.3 41.1 56.6 79.9 74.1

Equally satisfied

Source:  ‘The emigrant communities of Latvia’ database.

Note:  Emigrants who left Latvia in 2000 or later.

Job Working conditions Family life Relationships with people   outside your family Your home/dwelling Your standard of living Your life as a whole

More satisfied

12.1 3.4 5.9

11.0 9.5 12.9 19.4

Less satisfied

Middle class (ISCO 1–4)

51.6 74.0 65.8

66.5 67.8 52.6 34.8

More satisfied

34.3 21.7 28.3

24.5 24.8 32.8 44.4

Equally satisfied

14.1 4.3 5.9

9.0 7.4 14.6 20.8

Less satisfied

Middle class (income)

Table 2.13 Satisfaction with different aspects of life abroad compared with Latvia, 2014

61.9 80.0 72.7

74.5 72.5 58.3 47.1

More satisfied

26.3 16.1 21.5

18.7 21.0 25.8 37.3

Equally satisfied

All

11.8 3.9 5.8

6.8 6.5 15.9 15.7

Less satisfied

92

Europe’s disappearing middle class?

If we compare the improvements of the situation of the middle class with other migrant groups, we can see that the gains of the middle class in terms of job and working conditions are smaller than those of lower income/professional groups, and the same is true with regard to housing and relationships with people outside their family, which in some cases even worsened. Moreover, those who used to have middle-­class income in Latvia (did not struggle to get by) evidence smaller improvements in terms of their standard of living and life as a whole. With regard to institutions, middle class Latvian nationals who emigrated in 2000 or later evaluate all institutions in the receiving country higher than those in Latvia (sig. 0. Source:  ERF (1996, 2001); ERFS (2007, 2011), authors’ calculation.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 180

28/09/2016 08:24



Transformation in the world of work and the middle class: France ­181

Table 4.7  Annual wage inequalities (Gini coefficient), France, 1996–2011

Total population Top-­income class (TIC) Middle-­income class (MIC)   Top middle-­income class (TMIC)   Core middle-­class (CMIC)   Bottom middle-­income class (BMIC) Bottom-­income class (BIC)

1996

2001

2007

2011

0.37 0.34 0.30 0.25 0.26 0.33 0.45

0.36 0.38 0.28 0.24 0.24 0.31 0.43

0.37 0.39 0.29 0.25 0.24 0.33 0.48

0.38 0.38 0.30 0.26 0.25 0.33 0.48

Note:  Calculation of the Gini coefficient only includes people with an annual wage > 0. Source:  ERF (1996, 2001); ERFS (2007, 2011), authors’ calculation.

shows great stability in annual wage inequality before the crisis and a small rise from 0.37 to 0.38 between 2007 and 2011. When we disentangle wage inequality trends by income groups, the Gini coefficients for annual wages are less stable than those for living standards: by nature they are more dependent on the economic cycle. However, wage inequality for the middle-­income class stands at the same level in 1996 and 2011 and is also stable across the different income groups that compose the middle class. Again, the main rise in wage inequality between 1996 and 2001 concerns the top-­income class, and the BIC. Stability before the crisis has received several potential explanations in the literature focusing on the French case (Charnoz et al. 2013; Verdugo 2014): the minimum wage is relatively high and has been increasing more rapidly than the consumption price index, thus compressing wage careers; the growth in the highly educated workforce has occurred later in France than in other countries and high and persistent unemployment decreases workers’ bargaining power. Table 4.7 shows that between 1996 and 2001 – a boom period – annual wage inequality decreased for all income categories except top incomes. The rise in wage inequality within each category generally took place in the 2000s, and started later for the core middle-­income class, whose inequality increased only after 2007. This may be related to general trends on the labour market (towards more flexibility) and to changes in human resource management practices, including the development of more flexible pay arrangements (including bonuses, premiums). In order to explain trends in income inequalities, it is also interesting to decompose the sources of income for the different classes and to study their evolution over time (Figure 4.4). In general, for the active age population, the proportions of the different

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 181

28/09/2016 08:24

182

Europe’s disappearing middle class?

100

5 4 13

20

77

73

Middle income class

Top income class

90 80

39

1

7

70 60

2

50

17

40 30 20

42

10 0

Bottom income class Direct labour income (a)

Indirect labour market income (b)

Other income (c)

Welfare benefits (d)

Note:  Individuals are composed of the head of the household and partner for couples; (a) direct labour market income includes net wages and self-employment incomes, (b) indirect labour market income includes unemployment insurance and retirement pension, (c) other income includes capital income, (d) welfare benefit includes minimum income and minimum pension. Source:  ERFS (2011), authors’ calculation.

Figure 4.4 Sources of income by income classes, France, 2011 (percentage) sources of income vary across income classes but remained stable during the 1996–2011 period. In relative terms, the main characteristic of the MIC is the importance of income stemming directly or indirectly from work. First, the main difference between the middle-­and bottom-­ income classes is the share of income stemming directly from work: 76 per cent for the MIC and only 41 per cent for the BIC. This characteristic remained remarkably stable during the period: the share of direct labour market income in 1996 was 77 per cent for the MIC and 43 per cent for the BIC. When one considers replacement revenues, the main changes do not concern MIC but rather BIC. For this last category the share of income stemming from welfare has increased by 4 percentage points since 1996 and the share of indirect labour market incomes stemming from unemployment insurance or retirement pension (excluding minimum pension) has decreased by 2 percentage points. Second, the main difference between the middle and the top-­income

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 182

28/09/2016 08:24



Transformation in the world of work and the middle class: France ­183

classes is the share of income stemming indirectly from work (unemployment insurance). This source of income represents 14 per cent of total income for the MIC, but only 7 per cent for the TIC. In this view, the MIC is more sensitive to the risk coverage of income stemming from work than the TIC. For this last category, one could also note an increase in the share of ‘other income’ (which includes capital income) and a decrease in direct work income.12 However, in these data sources, this is partly due to a change in the survey methodology and a better appraisal of capital income for the last years of the period. Given these parallel trends in income and wage inequalities, it seems that the middle class has not only been protected by social protection and fiscal redistribution, but also by the characteristics and trends of their employment. In section 4.3 we clarify these trends by considering the growing issue of job polarization: has it happened in France and what are the links with trends in inequalities? 4.3 Changes in Occupational Structure: Limited Polarization, without a Rise in Wage Inequalities? According to Goos and Manning (2007) and Autor et al. (2003), technological progress in information and communications technology (ICT) has made it possible to replace routine tasks with machines, thereby decreasing labour demand for middle occupations in which these tasks appear to be the most important. At the same time, ICT increases demand for non-­ routine cognitive tasks at the top of the educational/wage level distribution. Because occupations at the bottom of the wage distribution are less affected by ICT, they are likely to absorb the residual supply of labour, a process called ‘job polarization’ in the literature. In the French case, there are some signs of job polarization, although  they still have to be confirmed. First, in parallel with the increase in the level of education, the structure of the working population by occupation has also evolved towards increasing demand for better educated workers, with rising employment for managers and professionals, as well as middle occupations. At the same time, the number of skilled clerks and blue-­collar workers has decreased, whereas the number of unskilled white-­collars has risen, which is consistent with the hypothesis of labour market polarization. The crisis has not impeded these trends and has even accelerated them, Figure 4.5 shows that the number of managers and professionals increased by 600 000 between 2007 and 2012, while the number of skilled clerks and blue-­ collar workers fell by 0.7 million. A detailed analysis of the Labour Force Survey also shows that the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 183

28/09/2016 08:24

184

Europe’s disappearing middle class?

7000 Craftsmen, shopkeepers

6000

Managers and professionals

5000

Middle occupations

4000

Skilled clerks

3000

Unskilled clerks

2000

Skilled blue-collar Unskilled blue-collar

1000

2012

2010

2008

2006

2004

2002

2000

1998

1996

1994

1992

1990

1988

1986

1984

1982

0

Source:  French Labour Force Survey (INSEE).

Figure 4.5  Employment by occupation, France, 1982–2012 (000s) share of several high-­skilled occupations has increased radically (corporate managers, physics, mathematics and engineering science professionals, teaching professionals, technicians and associate professionals, life science and health associate professionals), whereas the share of medium and lower-­skilled occupations has been declining (clerks, office clerks, craft and related trades workers, metal machinery and related trades workers, plant and machine operators and assemblers) (Askénazy and Erhel 2015). Second, a recent study shows that, as measured by occupational hours shares (see Harrigan et al. 2015), rapidly expanding occupations between 1994 and 2007 included upper managers, technical managers and technicians. The share of hours worked by middle managers, foremen/­supervisors and office workers fell, while retail and personal services grew. These trends might be related to the growth of technology-­related occupations and are consistent with technology-­driven hypotheses. Third, a polarization trend also appears when taking a wage distribution perspective. According to Eurofound’s analyses of employment growth by wage quintiles, France was characterized by polarization between 1995 and 2010. However, between 2011 and 2012 there seems to have been some upgrading, because low-­ wage jobs were destroyed (Eurofound 2013), although this needs to be confirmed over a longer time period. Taking an occupational perspective, data and the literature show that middle-­class occupations have been stable or expanding during the past

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 184

28/09/2016 08:24



Transformation in the world of work and the middle class: France ­185

20 years, but their content has been changing, with an increasing number of technicians, while the number of mid-­level managers, supervisors and foremen has been decreasing. The impact on middle-­class working and employment conditions requires further exploration. As shown in this section, middle-­class workers have been affected by the important structural trends in France since the 1980s: the average quality of their jobs has decreased through the development of precarious and part-­time employment, and through work intensification. However, inequalities have remained stable, with the exception of the years following the 2008 crisis.

5.  CASE STUDIES 5.1 Case Study 1: Working Time Regulations in France and Heterogeneity across Social Groups and Occupations Working time policy has been one of the most original but also most debated issues in French labour market policy over the past 20 years. The most important reform took place in 1998 with the Aubry law that provided for the progressive implementation of a new legal weekly working time of 35 hours (in 2000 for firms over 20 employees, and in 2002 for smaller firms) and incentives to reduce overtime hours (that became more expensive for firms). The goal of this policy was first to create jobs through a work-­sharing effect, but also (to a lesser extent) to build the grounds of a new society with increased leisure time and better conciliation between work and private life. The reduction in working time was not compensated by a decrease in monthly wages, but it was accompanied by other measures to maintain firms’ profitability. In particular, social contributions cuts that had been implemented in 1993 for low wages just above the minimum wage level were extended up to 1.8 times the minimum wage, and firms were also encouraged to increase working time flexibility through social bargaining (with annualized hours or working time accounts). Although this policy met its goals in terms of job creation (estimated at 300 000 according to most evaluations), it did not enjoy political success and was accused of reducing both household income (by cutting the number of overtime hours) and firms’ profits. Therefore, the right-­wing government elected in 2001 reformed the Aubry law by limiting its focus (35 hours a week became non-­compulsory for smaller firms) and reducing the cost of overtime hours (Fillon law, 2003). In 2007, the ‘purchasing power law’ (TEPA law, also named ‘working more to earn more’) introduced incentives for overtime, which became exempt from firms’ social contributions and

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 185

28/09/2016 08:24

186

Europe’s disappearing middle class?

income tax on the employee side. This last measure, which turned out to be quite paradoxical in a time of crisis, was removed in 2012 by the socialist government. In 2013, in response to the crisis, the possibility of reducing working time and wages through social bargaining was introduced in the law (accords compétitivité-­emploi). After nearly 20 years, the working time debate is still very important in France: despite its popularity among the higher and middle classes, it is regularly criticized by the right wing, which uses both supply-­side arguments (firms’ productive capacity and labour costs) and references to household income (especially for lower-­income groups). In practice, reductions in average weekly working time have remained limited: average effective weekly working hours decreased from 39.5 hours in 1996 to 37.9 in 2001. Considering full-­time work only, weekly working hours fell by two hours, on average, between these two dates (from 42 to 40 hours), and then stabilized in the 2000s. Therefore, actual adjustments remained below the four-­hour cut in legal working time, which is explained by the exemption of small firms and the possibility to use overtime hours. In relation to these legal changes, but also to bargaining processes on working time within firms, working time appears fairly heterogeneous in France. Indeed, it is usually higher in smaller firms that did not implement the 35 hours law than in firms with 20 employees or more. It also varies across occupations: for managers, reference working time is usually not defined weekly, but as a number of days worked annually. As a result, average weekly working time for the top-­income class and especially for managers appears stable, even after the implementation of the Aubry laws (Table 4.8). Working time adjustments take place over the year for these social groups, resulting in additional days off. For the middle-­income group, as well as the bottom-­income group, weekly working time for full-­ timers decreased (Table 4.8). Adopting an occupational approach leads to similar conclusions, with decreasing working time for full-­timers in almost all middle occupations, with the exception of some specific occupations that were not affected by the 35 hours law. Work sharing and flexibilization through the development of part-­time employment was also part of French labour market policy at the beginning of the 1990s. From 1992 to 1998, part-­time jobs were partially exempted from social contributions, a fiscal incentive that was suppressed in the Aubry laws in 1998. As a result the share of part-­time employment in France grew over the 1990s and stabilized at an average level (in comparison with other EU or OECD countries) in the years following the 35 hours laws. Part-­time employment started growing again slightly in the second half of the 2000s. In 2011, 16.8 per cent of French employees were working

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 186

28/09/2016 08:24



Transformation in the world of work and the middle class: France ­187

Table 4.8 Average effective weekly working hours (full-­time and part-­time) by income class, France, 1996–2011 1996

2001

2007

2011

Working hours full time All employees BIC MIC TIC

42.2 45.4 41.5 45.2

40.5 43.4 39.5 45.3

39.7 41.9 38.8 44.5

39.4 40.9 38.6 44.7

Working hours part time All employees BIC MIC TIC

26.1 23.1 26.5 26.6

26.0 23.7 26.1 27.9

23.9 20.2 24.5 25.5

23.4 19.0 24.0 26.2

Note:  People in employment (individuals are composed of the head of the household and their spouse/partner for couples). Owing to change in the LFS in 2003, there is a break in the definition of effective working hours between 2001 and 2007. Source:  ERF (1996, 2001); ERFS (2007, 2011), authors’ calculation.

part-­time (compared with an EU average of 19.2 per cent). In comparison with other countries, the duration of part-­time employment appears quite long in France: 23.4 hours in 2011 as against 20.1 hours on average in the EU (Eurostat). However, part-­time employment is more often involuntary in France (31.4 per cent of part-­time employment in 2012 as against 9.1 per cent in, for instance, the Netherlands). The incidence and proportion of involuntary part-­time employment appear related to income class and occupation (Figure 4.6 and Table 4.9). time employment is 30 per cent for the bottom-­ The share of part-­ income group, as against 16 per cent for the middle-­income category and 12 per cent for the top-­income group. Involuntary part-­time is limited for the middle class, almost non-­existent for the top-­income group and very high for the bottom-­income category. As for occupations, the incidence of part-­time is highest among white-­collar workers (31 per cent), and remains limited for all occupations that constitute the middle class. When individuals in middle occupations work part-­time, it is less often involuntary than for lower occupations and working time is longer (24.5 hours a week in 2011). However, there is some heterogeneity depending on employment status (craftsmen and shopkeepers generally do not work part-­time) and on sector (in the public sector, involuntary part-­time is more limited, except in the case of teachers). In general, part-­time employment appears

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 187

28/09/2016 08:24

188

Europe’s disappearing middle class?

35 30 25 20 15 10 5 0

1996

2001 All employees

2007 BIC

MIC

2011 TIC

Note:  People in employment (individuals are composed of the head of the household and partner for couples); TIC: top-income class, MIC: middle-income class, BIC: bottomincome class. Source:  ERF (1996, 2001); ERFS (2007, 2011), authors’ calculation.

Figure 4.6 Part-time rates by income class, France, 1996–2011 (percentage) strongly related to occupational hierarchy in France and is concentrated in lower qualified jobs. For these disadvantaged groups the situation worsened at the end of the 2000s, with a marked decrease in average working hours, which also affects middle occupations, although the trend is more limited. In the context of the crisis and facing a general trend towards increasing atypical employment (short temporary contracts and short part-­time), the social partners and the government introduced a threshold for part-­time jobs of 24 hours a week in 2013 (Accord National Interprofessionnel, 11 January 2013). That measure was part of a more general flexicurity agreement, exchanging more flexibility (easier redundancies, wage flexibility in case of economic difficulties, and increased mobility for employees) for new rights for workers (health insurance transferability, increased contribution to unemployment insurance for temporary jobs, and part-­time duration threshold). However, there are many limits to its actual implementation, as the employer can depart from that 24-­hour minimum if the employee agrees. Besides, branch-­level social bargaining can also modify the 24-­hour threshold and introduce a lower minimum duration. However, in that case

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 188

28/09/2016 08:24



Transformation in the world of work and the middle class: France ­189

Table 4.9 Characteristics of working time by occupational group, France, 2011 Occupational group

Working time characteristic

Managers

Part-­time rate (per cent) Involuntary part-­time (as a share of total employees, per cent) Total average effective weekly working hours Average effective weekly working hours full-­time Average effective weekly working hours part-­time Part-­time rate (per cent) Involuntary part-­time (as a share of total employees, per cent) Total average effective weekly working hours Average effective weekly working hours full-­time Average effective weekly working hours part-­time Part-­time rate (per cent) Involuntary part-­time (as a share of total employees, per cent) Total average effective weekly working hours Average effective weekly working hours full-­time Average effective weekly working hours part-­time Part-­time rate (per cent) Involuntary part-­time (as a share of total employees, per cent) Total average effective weekly working hours Average effective weekly working hours full-­time Average effective weekly working hours part-­time

Middle occupations

Blue-­collar workers

White-­collar workers

9.7 1.3 40.9 42.3 26.9 13.2 2.3 37.9 39.9 24.5 10.1 4.5 35.1 36.7 21.1 31.6 11.2 32.3 36.7 22.7

Note:  People in employment (individuals are composed of the head of the household and partner for couples), detailed results by occupation are available on request. Source:  ERFS (2011), authors’ calculation.

working hours should be regrouped as much as possible to avoid dispersed working time. An evaluation of that 2013 national agreement and law was conducted in 2015.13 In March 2015, 48 industry-­level agreements were signed (and bargaining was still going on in summer 2015) in 20 industries. As a result, 44 per cent of part-­time employees are now covered by an industry-­level agreement on part-­ time working. In general, these agreements set a minimum part-­time duration below the 24-­hour threshold, either for all employees, or according to occupation or type of firm. Minimum durations are heterogeneous across industries but might be very low (two hours for social workers, clothes shops or some activities related to social housing). Regulation of working time dispersion is generally fairly close to the law (no more than one break during the day), but there are some exceptions that introduce more flexibility. As a consequence of this bargaining

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 189

28/09/2016 08:24

190

Europe’s disappearing middle class?

process the 24-­hour threshold adopted in 2013 can be expected to have only a limited impact on part-­time duration in France. Working time regulation changes have been very important in France since the beginning of the 1990s, and followed diverse and sometimes contradictory goals: employment and work sharing, internal flexibility, incentives to overtime as an income component, development of part-­time and limitation of short part-­time. Middle occupations have been impacted by these reforms and policies, but far less than have the lower occupations. 5.2  Case Study 2: Education and the Middle Class In France, the overall level of education has increased and now half of a generation has access to tertiary education (Afsa 2009). This development did not have a particular impact on the French financial arrangements for education. France is characterized by education services that are free or largely subsidized, whatever the education level concerned, and even the private sector is largely subsidized. For secondary education, the share of secondary private schooling is about 20 per cent (Brizard et al. 2013). The subsidies for the private sector cover teachers’ wages, but also some operating costs. The difference in the proportions of pupils from more advantaged socio-­economic backgrounds who attend public and private secondary schools is 16 percentage points, on average, but is characterized by high regional heterogeneity (DEPP 2014). In this perspective, on average, access to private education in France is not a privilege for those with high socio-­economic status or high income and is also largely used by the MIC. After secondary school, three main education tracks are available: two-­ year vocational training,14 university degrees and preparatory classes for elite schools (grandes écoles). Public expenditure per student is closely related to the tertiary education track. For 2001, Zuber (2004) shows that, for instance, a preparatory class to elite school costs the state €77 281 a year per student, whereas a scientific degree at university for the same education level amounts to €33 889. State subsidies for a master’s degree in an engineering school amount to €67 868 and are closer to the subsidy for a scientific degree at university, which amounts to €66 882. However, for the subgroup of highly ranked grandes écoles the subsidy per student amounts to €207 246. In the French system, the distribution of education resources is mediated by the state. The level of individual subsidies is strongly linked to the educational performance of students over their initial education curriculum. A student with very good academic performance generally accumulates a number of years in preparatory classes and some years in top-­ranked engineering schools, whereas a poorly performing student could leave the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 190

28/09/2016 08:24



Transformation in the world of work and the middle class: France ­191

education system without particular compensation in terms of public education spending.15 In other cases, students have to rely on their family and/ or housing allocation (if they leave home). The French financial arrangement for higher education is characterized by a very low level of public transfers (scholarship or loans) for living costs (Charles 2014). According to Afsa (2009) the number of students leaving basic education without qualifications has decreased, but has been stabilized since the mid-­ 1990s: this status applied to one-­third of students in 1965, one-­fifth in 1975, one-­eighth in 1985 and is one-­tenth now. However, 8 per cent of young people have only a very low general mid-­secondary diploma (Brevet des collèges). In this context it is interesting to estimate how the benefits of education could be considered to explain the willingness of the middle class to pay high taxes and how this could account for the stability of the French social model during the crisis. Table 4.10 shows that total public education services represent on average about 12 per cent of household income (€2600 per consumption unit within the household). Public education services taken as a whole are highly redistributive, even if they are not narrowly targeted. When we focus on families with children in education, the public education in-­kind transfers represent a more important share of income (30 per cent). This share is very high for the first (110 per cent) and the second quintile (49 per cent) Table 4.10 Education public spending as a percentage of net income by quintile of income (before tax and subsidies), France, 2007 and 2009 1st quintile

2nd 3rd 4th 5th Total quintile quintile quintile quintile population

Whole education services (total population)

57.9

19.3

12.5

Education services (total population)   Primary education   Secondary education   Tertiary education Family with children in education

59.3

20.7

13.8

22.4 31.6 5.3 110.1

7.3 11.1 2.3 43.3

5.0 7.0 1.8 29.3

2009a

8.7

4.4

11.8

9.3

4.6

12.6

3.3 4.4 1.6 21.7

1.6 2.2 0.8 12.0

4.6 6.4 1.6 30.0

2007b

Source:  (a) Bonnefoy et al. (2010); (b) Amar et al. (2008).

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 191

28/09/2016 08:24

192

Europe’s disappearing middle class?

and shows that the redistributive role of public education policies is particularly important for low-­income families. Decomposition by education level is available for 2007 (Amar et al. 2008). It shows that tertiary education has a particular status in that view: whereas the subsidy tends to decrease with the quintile for primary and secondary education, this is not the case for tertiary education. The amount decreases between the first and second quintiles and then increases and the decrease in the share of net income is not as sharp as for other education services (Table 4.10). These results indicate that the spending on higher education is not progressive. When we focus on families with children in education, the lack of progressivity of the higher education transfers is confirmed from the second to the fifth quintile. Moreover, public secondary education transfers are no more progressive at the top of the distribution: the amount of transfers increases between the fourth and the fifth decile. According to Amar et al. (2008), these results correspond to differences within the secondary education level. The first level – common to every pupil – (collège) is progressive with low living standards. After this first level there is a distinction between vocational curriculum and the more general curriculum that leads pupils to a baccalauréat degree and then higher education. In contrast to the other education tracks of this second stage, the general curriculum does not correspond to progressive transfers. The income analysis indicates that the free education services that characterize France are highly progressive when they are taken as a whole. However, when one looks in more detail, these results are not verified for the last years of secondary education leading on to higher education or for higher education in general. The highest transfer levels for tertiary education are for the fourth and fifth quintiles. In this view, as their children access this level more often, we could argue that TIC and TMIC receive some money back from the taxes they pay at a higher rate than their fellow citizens. Notably, these results do not take into account the whole heterogeneity of higher education (grandes écoles versus university) in terms of subsidy level, as mentioned earlier. Taking into account such differences may worsen the redistributive impact of public education services if those on higher incomes tend to have children who access the more prestigious and subsidized segment of the higher education system. More generally there are some limits to this kind of analysis that are directly linked to the static appraisal of redistribution. A given household may have a single earner when the children are in primary school and then two earners when the children become autonomous. To avoid this kind of problem it is possible to take a generational point of view and look at the highest initial degree obtained by parents’ socio-­economic status. In that perspective Figure 4.7 shows that between the 1970s’ generation

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 192

28/09/2016 08:24



Transformation in the world of work and the middle class: France ­193

25

a) Individual born between 1968 and 1972

20 15 10 5 0

Whole generation

Low SES

Intermediate SES

High SES

Two-year degree

Three-year degree

Four-year degree

Five-year degree (university)

Five-year degree (grande école)

More than five-year degree

b) Individual born between 1978 and 1982 25 20 15 10 5 0 Whole generation

Low SES

Intermediate SES

High SES

Two-year degree

Three-year degree

Four-year degree

Five-year degree (university)

Five-year degree (grande école)

More than five-year degree

Note:  SES = socio-economic status of parents. Source:  French Labour Force Survey 1982–2013 (INSEE) and Courtioux (2011).

Figure 4.7 Highest tertiary degree obtained by social economic status as a percentage of the whole cohort, France, 2011 (percentage)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 193

28/09/2016 08:24

194

Europe’s disappearing middle class?

and the 1980s’ generation the share of two-­year tertiary degrees increased from 15 per cent to 19 per cent, but mainly concerned former students with low and intermediary socio-­economic backgrounds. For these two cohorts the students from higher socio-­economic status families are overrepresented in five-­year university degrees and the prestigious grande école degrees. In this view, there is no fundamental change in the way the middle classes and top-­income classes benefit from free public education.

6.  CONCLUSION AND POLICY ISSUES Studying the French middle class in relation to changes in the world of work confirms a number of fairly well-­known results: middle-­class size did not decrease in France between 1996 and 2011, and inequalities did not increase notably over that period. The impact of recessions and especially of the 2008 crisis tends to concentrate on the lower-­income group and on the occupations that are at the bottom of the occupational hierarchy, who are particularly concerned by rising unemployment and bad job quality. However, this study also highlights some factors of vulnerability for the French middle class. First, it provides evidence of the importance of heterogeneity inside the middle class: the lower-­income levels inside the middle class were also hit fairly severely by the ‘Great Recession’. More generally it shows that the jobs held by the middle class changed between 1996 and 2011, as a consequence of general trends towards labour market flexibility. Finally, the importance of public policies (including social protection, wage and education policies) in sustaining middle-­class income and social position appears quite clearly. In times of substantial budget constraints, these policies might become more limited and the rise in inequalities between 2007 and 2011 might be a permanent trend.

NOTES   1. With computation stemming from the Luxembourg Income Study and EU-­SILC, Bigot et al. (2011) retain an income-­based definition of the middle class that differs from the definition we use in this study (see Chapter 1). By definition, our income-­based definition, which is larger, tends to produce a MIC with a larger size than the results of Bigot et al. (2011).   2. Detailed results are available on request.   3. In 2011, source: OECD Family database.   4. Source: OECD Family database.   5. Source: Observatoire National de la Petite Enfance (2012).   6. Further details on this law and its effects are provided in section 5.   7. Source: Askénazy and Erhel (2015).

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 194

28/09/2016 08:24



Transformation in the world of work and the middle class: France ­195

 8. The Couverture maladie universelle (CMU) is health insurance targeted at those on minimum income.   9. Sources for this paragraph: French Ministry of Labour and Eurofound, accessed 1 April 2016 at http://www.eurofound.europa.eu/observatories/eurwork/comparative-­informa tion/national-­contributions/france/france-­industrial-­relations-­profile. 10. Source: OECD Income Distribution database, Gini at disposable income post-­tax and transfer. 11. This trend, which is not specific to France, has received attention in the recent literature (see especially Piketty 2013). 12. The rise in the importance of capital income for the top revenues has been documented in the literature by Piketty (2013). 13. Ministère du Travail, de l’Emploi et de la Formation Professionnelle: ‘Bilan de la loi de sécurisation de l’emploi du 14 juin 2013 au 03 avril 2015’ (‘A follow-­up of the employment security law from 14th June 2013 to 3rd April 2015’), accessed 6 May 2016 at http://travail-­emploi.gouv.fr/IMG/pdf/CONFERENCE_THEMATIQUE_DU_3_ AVRIL_2015_-­_Bilan_de_la_loi_de_securisation_de_l_emploi.pdf. 14. Brevet de technicien supérieur (BTS) are operated at Lycée (secondary school), whereas Diplôme Universitaire Technique (DUT) are operated at university. 15. In a life-­course perspective, access to continuous vocational training does not compensate for these large differences in individual public spending on education (Lignon 2014).

REFERENCES Afsa, C. (2009), ‘La moitié d’une génération accède à l’enseignement supérieur’ (‘Half of a generation has access to tertiary education’), in INSEE, France Portait social édition 2009, Paris: La documentation française, pp. 29–39. Amar, E., M. Beffy, F. Marical and E. Raynaud (2008), ‘Les services publics de santé, d’éducation et logement contribuent deux fois plus que les transferts monétaires à la réduction des inégalités de niveau de vie’ (‘Public services in health, education or housing contribute two times more to the reduction of living standards inequalities than monetary transfers’), in INSEE, France Portrait social édition 2008, Paris: La documentation française, pp. 85–101. Askénazy, P. (2004), Les désordres du travail (The Disorders of Work), Paris: Le seuil, République des Idées. Askénazy, P. and Erhel, C. (2015), ‘Productivity puzzles in France’, IZA Discussion Paper No. 9188, Institute for the Study of Labour, Bonn. Askénazy, P., A. Bozio and C. Garcia-­Penalosa (2013), ‘Dynamique des salaires par temps de crise’ (‘Wage dynamics in times of recession’), Notes du Conseil d’Analyse Economique, no. 5, April. Autor, D.H., F. Levy and R.J. Murnane (2003), ‘The skill content of recent technological change: an empirical exploration’, Quarterly Journal of Economics, 118 (4), 1279–333. Barruel, F., S. Thomas, O. Filatriau and H. Mariotte (2014), ‘Créateurs d’entreprises: avec l’auto-­entreprenariat de nouveaux profils’ (‘Firms creations: new profiles with a new self-­employed status’), INSEE première, no. 1487, Paris. Bigot, R., P. Croutte, J. Müller and G. Osier (2011), ‘Les classes moyennes en Europe’ (‘The middle class in Europe’), Cahier de Recherche, no. 282, Centre de Recherche pour l’Etude et l’Observation des Conditions de vie (CREDOC), Paris. Bonnefoy, V., M.-­C. Cazenave, A. Eidelman and T. Razafindranovona (2010), ‘La

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 195

28/09/2016 08:24

196

Europe’s disappearing middle class?

redistribution en 2009’ (‘Redistribution in 2009’), in INSEE, France Portrait sociale édition 2010, Paris: La documentation française, pp. 79–95. Brizard, A., E. Blanché and P.-­O. Gasq (2013), ‘Les élèves du second degré à la rentrée 2013 dans les établissements publics et privés’ (‘Secondary education students in public and private schools for the academic year 2013–2014’), Note d’information Direction de l’Evaluation de la Persepective et de la Performance (DEPP), no. 13.34, Paris. Charles, N. (2014), ‘France: a low-­fee, low aid system challenged from the margins’, in B. Ertl and C. Dupuy (eds), Between Marketization and Social Justice. A Comparative Approach to Tuition Fee Policies in Western Europe and Beyond, Oxford: Oxford University Press, pp. 67–83. Charnoz, P., E. Coudin and M. Gaini (2013), ‘Une diminution des disparités salariales en France entre 1967 et 2009’ (‘Reducing wage disparities in France between 1967 and 2009’), in INSEE, Emplois et salaires edition 2013, Paris: INSEE. Chauvel, L. (2006), Les classes moyennes à la derive (Middle-­classes’ Drift), Paris: Seuil, République des Idées. Courtioux, P. (2011), ‘L’origine sociale joue-­t-­elle sur le rendement des études supérieures?’ (‘Does social background have an impact on higher education returns?’), Ecole des Hautes Etudes Commerciales du Nord (EDHEC) Position Paper, November, Nice. Direction de l’Evaluation de la Persepective et de la Performance (DEPP) (2014), Géographie de l’école (School’s Geography), no. 11, Paris: Ministère de l’Education Nationale. Enquête revenus fiscaux (Taxable Income Survey) (ERF) (1996), INSEE, ADISP-­CMH. Enquête revenus fiscaux (Taxable Income Survey) (ERF) (2001), INSEE, ADISP-­CMH. Enquête revenus fiscaux et sociaux (Tax and Social Income Survey) (ERFS) (2007), INSEE, ADISP-­CMH. Enquête revenus fiscaux et sociaux (Tax and Social Income Survey) (ERFS) (2011), INSEE, ADISP-CMH. Eurofound (2013), Employment Polarisation and Job Quality in the Crisis: European Jobs Monitor 2013, Dublin: Eurofound. Fontagné, L., C. Mitaritonna and A. Tritah (2014), ‘Les inégalités salariales ont diminué en France’ (‘Wage inequalities have decreased in France’), La Lettre du CEPII, no. 343, April. French Labour Force Survey (FLFS) 1982–2013 – Institut National de la Statistique et des Etudes Economiques (INSEE) (producer), Archives des données issues de la statistique publique statistiques-­Centre Maurice Halbwachs (ADISP-­CHM) (distributor). Gollac, M. (2005), ‘L’intensité du travail: formes et effets’ (‘Work intensity: forms and effects’), Revue économique, 56 (2), 95–216. Goos, M. and A. Manning (2007), ‘Lousy and lovely jobs: the rising polarization of work in Britain’, Review of Economics and Statistics, 89 (1), 118–33. Goux, D. and E. Maurin (2012), Les nouvelles classes moyennes (The New Middle Classes), Paris: Seuil, République des Idées. Harrigan, J., A. Reshef and F. Toubal (2015), ‘The march of the techies: technology, trade, and job polarization in France, 1994–2007’, mimeo, University of Virginia, Charlottesville, VA, accessed 30 September 2015 at http://people.virginia.edu/~ar7kf/papers/Harrigan_Reshef_Toubal_September_2015.pdf.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 196

28/09/2016 08:24



Transformation in the world of work and the middle class: France ­197

Lignon, V. (2014), ‘Transitions familiales, professionnelles et investissements éducatifs. Une analyse par microsimulation dynamique’ (‘Family transitions, professional transitions and investments in education a dynamic microsimulation analysis’), PhD dissertation, Université Paris 1 Panthéon-­Sorbonne. Observatoire National de la Petite Enfance (2012), L’accueil du jeune enfant en 2011 (Childcare Policy in 2011), Paris: CNAF. operation and Development (OECD) (2014), Organisation for Economic Co-­ Divided We Stand, Why Inequality Keeps Rising, Paris: OECD. Organisation for Economic Co-­operation and Development (OECD) Family database, OECD (distributor), 7 September 2015 update, accessed 1 April 2016 at http://www.oecd.org/els/family/database.htm. operation and Development (OECD) Income Organisation for Economic Co-­ Distribution database, OECD (distributor), accessed 4 April at https://stats. oecd.org/Index.aspx?DataSetCode=IDD. Piketty, T. (2014), Capital in the Twenty-­First Century, trans. A. Goldhammer, Cambridge, MA: Harvard University Press. Verdugo, G. (2014), ‘The great compression of the French wage structure, 1969– 2008’, Labour Economics, 28 (April), 131–44. Zuber, S. (2004), ‘Evolution de la concentration de la dépense publique d’éducation en France 1900–2000’, Education et Formation, 79 (December), 97–108.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 197

28/09/2016 08:24

5. The erosion of the German middle class: The end of the ‘levelled-­out, middle-­class society’? Gerhard Bosch and Thorsten Kalina 1. INTRODUCTION At the end of the Second World War, Germany lay in ruins. As a result of the physical destruction of much of the country, the large number of refugees who had left their possessions behind, and the 1949 currency reform, which devalued financial assets, sections of the former middle and upper classes found themselves ‘déclassé’. At the same time, the country’s rapid reconstruction offered other social groups unique opportunities for advancement. These simultaneous processes of upward and downward mobility made the traditional boundaries between the social classes temporarily more permeable. On top of this, hitherto disadvantaged blue-­and white-­collar workers made known their demands for a fair share of the fruits of economic growth and participation in corporate decision-­making, giving rise in some cases to bitter industrial disputes. The rivalry between competing economic systems in the Cold War made it easier to enforce these demands in West Germany and to develop the welfare state. In comparative research, the German social model is often regarded as based on the prototype of the conservative Bismarckian welfare state, in which differences in labour market status are perpetuated by income-­related benefits (Esping-­Andersen 1990). This status-­perpetuating effect certainly cannot be dismissed; nevertheless, this categorization underestimates the levelling effect of the German welfare state in the post-­war period, for two reasons. First, virtually all dependent employees were (and still are) members of the major social insurance schemes. Among dependent employees, only tenured civil servants (Beamte) have their own special system. The universal nature of these insurance schemes means that the vast majority of the population up to the upper-­middle strata are protected against life’s main contingencies. Second, earnings differentials were reduced by industry-­level collective agreements covering virtually all workers in the relevant industry and geographical area. 198

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 198

28/09/2016 08:24



The erosion of the German middle class ­199

In combination with an inclusive wage-­ setting system, these social i­nsurance schemes perpetuate only slight differences in status. By virtue of these complementarities – and to depart somewhat from Andersen’s typology – the German social model in the post-­ ­ war period can be described as an inclusive Bismarckian welfare state (Bosch 2015). As such, it provided the economic underpinning for a broadly based middle class that enjoyed a high level of stability over the various phases of the life course, including old age. Because of the high level of upward mobility in the post-­war years and the equalization of living conditions and consumption patterns, the well-­known German sociologist Schelsky, writing at the beginning of the 1950s, heralded the end of the traditional class-­ based society, which, he argued, had been replaced by a ‘levelled-­out, middle-­class society’ (Schelsky 1953: 218). Social stratification in post-­war Germany was usually compared to an onion, whose bulbous middle represented the broad middle class (Geißler 2011). Since the mid-­1990s, however, income inequality has increased more sharply in Germany than in many other European countries. However, the question of whether this has led to the erosion of the incomes policy that provided a firm foundation for the middle classes is hotly debated (ISG 2011; Arndt 2012). In fact, it does not necessarily follow that increasing inequality in individual incomes inevitably leads to inequalities in the household disposable income that is a decisive factor in determining class affiliation. Household members can compensate for losses of income by increased economic activity, for example, and the welfare state can also make good such losses. In what follows, we first outline the growing inequality in individual incomes between 1995 and 2013 and investigate whether and to what extent this has altered the earnings position of the middle-­income groups and their various sub-­ groups (section 2). Because class affiliation is determined multi-­dimensionally and we are examining only one of these dimensions, we refer in what follows to the middle-­income groups. We then go on to examine household structure and composition in the various income groups and their evolution over time (section 3). This is followed by an analysis of household composition in the various income groups by demographic characteristics and qualifications (section 4). The data analysis concludes with a section on income group composition by self-­ definition and occupational classification and well-­being by income group (section  5). After this analysis of the evolution of the middle-­income groups, we summarize the investigations on the redistributive effects of the taxation system and welfare state (section 6). This is followed by a lengthy section in which the causes of the decline and stabilization of the middle classes in Germany are investigated (section 7). The chapter ends with

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 199

28/09/2016 08:24

200

Europe’s disappearing middle class?

two case studies of the role of the minimum wage and vocational training (section 8) and a conclusion (section 9).

2. EVOLUTION OF THE MIDDLE INDIVIDUAL AND HOUSEHOLD INCOME GROUPS BETWEEN 1995 AND 2013 The size and evolution of the middle-­income groups are very strongly influenced by how the middle is defined statistically. As the lower limit, we adopt the internationally acknowledged poverty threshold of 60 per cent of median income and as the upper limit, double that median income. Such a broad definition, which puts more than two-­thirds of all households in the middle-­income groups, brings with it the risk that major shifts within this large group will be overlooked; these shifts may be of similar or even greater significance than those between the three income groups. For this reason, we divide the middle-­income group into three sub-­groups, defining the lower and upper income groups as follows: 1. 2. 3. 4. 5.

Less than 60 per cent of the median (bottom-­income group). 60 to 80 per cent of the median (lower middle). 80 to 120 per cent of the median (intermediate middle). 120 to 200 per cent of the median (upper middle). 200 per cent of the median and above (upper-­income group).

The measures of income adopted are just as important as the boundaries between the income groups. We make a distinction between, on one hand, individual and household income, which because of the different numbers of household members can be compared only on an equivalence weighted basis.1 This distinction is important in order to ascertain whether the increasing wage inequality is also reflected in inequalities in household incomes or whether it is offset by, for example, increases in hours worked or in the numbers of earners per household. On the other hand, we differentiate between the primary distribution – that is, the distribution of income generated directly by the market process – and the secondary distribution, which relates to disposable income after taxation and transfer payments. This will enable us to ascertain whether, and to what extent, the state (including the social insurance funds) compensates for possible increasing inequalities in primary incomes in the middle-­ income groups via the revenue side (taxes and social insurance contributions) or the expenditure side (transfers) and whether redistribution is stronger or weaker than two decades ago. The secondary distribution can

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 200

28/09/2016 08:24



The erosion of the German middle class ­201

BOX 5.1 DATA ANALYSIS BASED ON THE SOEP, GERMANY The SOEP is a household survey that has been carried out annually in Germany since 1984 on behalf of DIW Berlin (the German Institute for Economic Research) by TNS Infratest Sozialforschung. A representative sample of households and all people aged 17 and over living in them are surveyed. The SOEP’s basic population is the resident population living in private households within the current boundaries of the Federal Republic of Germany. The SOEP data are rendered representative by means of a differentiated extrapolation procedure. The SOEP was established as a means of investigating social and economic issues and includes a large number of household and individual-­related variables (Wagner et al. 2007). In 2011, more than 12 000 households and a total of more than 21 000 individuals were surveyed (Bohlender et al. 2011). For our analysis we use data distribution version 30 with data up to 2013, including all available samples. Data on various components of individual and household income are gathered in the SOEP. In the individual questionnaires, data on income from work are gathered by means of questions on gross and net earned income in the previous month. Data on earned income and special payments/bonuses are also collected elsewhere. Information on actual and contractual weekly working time are also recorded, from which hourly pay can be calculated. For the purposes of international comparison, the publishers make available the Cross-­National Equivalent File, which contains a considerable volume of information on incomes generated from the original data and tested for plausibility (Frick et al. 2008). These data are usually used to calculate poverty rates and also form the basis for our income calculations.

be analysed only at household level, because married couples in Germany are taxed jointly and transfer payments are made to a large extent on a household basis. This applies even to the contribution-­related social insurance benefits which, it is true, are linked to the individual contributions but also contain many family-­related elements, such as the derived social protection for family members or family-­related supplements, such as those applied to unemployment benefit. The different income measures enable us to analyse more precisely the observed shifts in the income distribution. 2.1 Evolution of the Primary Distribution of Individual and Household Incomes from 1995 to 2013 Figure 5.1 shows the evolution of the distribution of hourly pay between 1995 and 2013. It provides evidence of clear increases in the shares of the upper-­and lower-­income groups. The share of the lowest-­income group

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 201

28/09/2016 08:24

202

8.9

8.2

7.6

7.5

7.3

7.6

7.0

6.7

6.6

6.2

6.3

6.6

6.0

6.2

7.0

6.2

6.0

6.5

6.9

100

Europe’s disappearing middle class?

27.2

27.2

27.6

26.5

26.9

26.6

28.0

28.0

27.2

27.4

27.8

27.7

26.8

26.9

27.0

27.2

26.4

26.4

80

26.5

90

70

30.0

31.0

30.5

32.0

31.9

32.5

31.9

31.7

32.9

34.0

33.2

33.8

36.1

35.4

35.1

37.1

38.4

37.6

50

37.5

60

18.7

18.8

17.9

17.8

10

11

12

13

16.1

19.3

09

15.7

18.0

08

15.6

18.9

07

15.3

14.6

18.3

06

15.3

18.5

05

14.2

17.3

04

15.3

17.3

03

15.0

16.4

02

15.5

16.1

01

15.5

15.7

00

15.0

15.8

16.1 14.9

99

15.8 13.7

98

13.8

15.4

13.6

97

15.9

13.4

96

10

95

20

15.6

30

14.7

40

Bottom

Lower middle

Intermediate middle

Upper middle

20

20

20

20

20

20

20

20

20

20

20

20

20

20

19

19

19

19

19

0

Top

Source:  SOEP v30, authors’ calculations.

Figure 5.1 Distribution of hourly gross wages by income brackets, Germany, 1995–2013 (percentage; dependent employees) increased between 1995 and 2013 from 13.4 to 17.8 per cent, while the share of the upper-­income groups rose from 6.9 per cent to 8.9  per cent. Because the shares of the upper and lower middle-­income groups remained virtually unchanged, the losses were suffered almost exclusively by the intermediate middle-­income group, whose share fell from 37.5 per cent to 30.0 per cent. In contrast to the wage income shown in Figure 5.1, the equivalence-­ weighted data on income at household level include all market income from dependent employment and self-­employment, as well as income from assets, maintenance payments and private pension plans (Figure 5.2). Thus the figures also include transfers, albeit – in contrast to the secondary distribution – only private transfers. According to these figures, the share of households in the middle-­income groups in all households fell between 1992 and 2013 by 8 percentage points, from 56.4 per cent to 48 per cent. Here, too, it was the intermediate middle-­income group that suffered the greatest loss of share (−5.5 percentage points). The shares of the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 202

28/09/2016 08:24



The erosion of the German middle class ­203

17.2

17.1

17.3

17.0

17.3

17.1

17.1

17.9

16.9

17.7

16.7

15.1

15.6

15.3

15.7

15.2

14.5

14.8

12.6

13.6

13.5

90

13.9

100

24.0

25.0

24.6

24.0

24.1

24.5

23.8

24.8

24.5

24.3

24.3

25.8

24.9

24.9

24.8

26.0

26.0

26.2

26.8

26.2

25.2

70

26.1

80

15.7

16.8

16.5

17.5

16.6

16.9

16.0

16.6

5.8

6.4

6.7

6.7

6.3

6.9

7.4

36.5

35.5

34.8

35.3

34.8

35.2

34.7

35.4

6.2

16.1

15.9

7.2

6.7

17.4 7.0 34.6

35.4

18.3

35.3

18.2

7.2

33.1

34.0

18.4

7.8

18.7

8.6 32.8

18.3 8.3 32.4

17.7

19.2 8.3 32.0

8.5

19.3 8.4 32.4

8.4

20.7 9.2 30.7

30

32.5

20.9 9.0

40

32.6

22.1 9.1 29.7

50

30.5

60

20 10

19 92 19 93 19 94 19 95 19 96 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12 20 13

0

Bottom

Lower middle

Intermediate middle

Upper middle

Top

Source:  SOEP v30, authors’ calculations.

Figure 5.2 Distribution of equivalized household gross income before taxes and transfers in the previous year by income brackets, Germany, 1992–2013 (percentage) upper-­and lower-­income groups in all households are considerably higher than for individual incomes from dependent work, for which there may be several reasons. The most likely are the growing inequality in income from assets, the high share of high-­earning self-­employed workers, who are not included in the analyses of individual pay, and increasing differences in the number of earners and hours of paid work in households.2 The evolution of the number of earners and hours worked at household level will be investigated in greater detail in section 3. Taken as a whole, the data show a clear trend towards an increasingly unequal primary distribution. For all three indicators of income distribution, the trend over the past 20 years has been relatively continuous and thus cannot be explained by one-­off events, such as the Hartz Acts of 2003. In contrast to other countries (Vaughan-­Whitehead 2011), even the financial crisis has not had any detectable influence on the primary income distribution in Germany. Despite a sharp decline in gross national

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 203

28/09/2016 08:24

204

Europe’s disappearing middle class?

product (GNP) of 4.9 per cent in 2009, employment remained stable, because the drop in orders was cushioned by a redistribution of working time and, above all, by state-­subsidized short-­time working (Bosch 2011). The main beneficiaries of this protection were the middle-­income groups in the high-­earning German export sector. In our cause analysis, therefore, we will have to focus rather on the long-­term erosion of the German wage system. 2.2  Evolution of Household Disposable Incomes between 1992 and 2013

8.0

8.0

7.1

7.6

7.6

7.8

7.9

7.2

8.0

6.7

6.2

7.0

6.0

6.2

6.5

6.1

6.2

6.0

6.3

6.4

6.0

100

6.1

The share of all the middle-­income groups in disposable income in 2013, at 77.8 per cent, was 30 percentage points higher than their share in market income (Figure 5.3).3 Compared with the primary distribution, the German welfare state halves the highest income groups’ share in all households, mainly via the progressive income tax system. At the other end

26.5

27.3

26.7

26.8

26.7

26.3

26.8

26.8

25.8

26.9

26.5

26.7

27.3

26.7

27.7

26.4

26.8

27.7

26.8

26.7

26.8

80

27.9

90

70

33.6

33.4

33.8

34.2

33.9

34.0

34.8

35.0

35.5

36.0

36.3

38.6

38.4

39.4

37.8

39.0

39.6

37.8

36.7

37.4

36.6

50

35.2

60

40

17.6

17.2

14.2

14.6

17.7

16.5 14.0

14.2

17.2 14.0

16.9

16.5 14.0

14.3

17.3 13.0

17.2

17.0 13.0

17.1

16.6 12.2

15.2

16.8 11.5

14.3

16.8 10.5

17.8 10.7

17.9

16.6 12.1

10.3

17.4 12.9

16.8

17.9 11.6

10.6

17.8

18.1

10

13.1

20

12.5

30

19 92 19 93 19 94 19 95 19 96 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12 20 13

0

Bottom

Lower middle

Intermediate middle

Upper middle

Top

Source:  SOEP v30, authors’ calculations.

Figure 5.3 Equivalized disposable household income after taxes, social insurance contributions and transfers in the previous year by income brackets, Germany, 1992–2013 (percentage)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 204

28/09/2016 08:24



The erosion of the German middle class ­205

of the income distribution, the combined effect of a reduced tax burden and transfer payments reduces the share of the lowest-­income group by a third, because many households are ‘promoted’ to the lower-­middle or intermediate-­middle group when classified by disposable income. Thus, to a large extent, though not fully, the welfare state has been able to compensate for the growing inequality in the primary income distribution. Particularly since the beginning of the new millennium, the shares of the upper-­and lower-­income groups in all households have been growing, admittedly not dramatically but nevertheless by 3.7 percentage points (bottom) and by 1.5 percentage points (top), so the upward trend is quite clear, while the share of the middle-­income groups declined from 83 per cent in 2000 to less than 78 per cent in 2013. This development has been interpreted in very different ways in the extremely vigorous German debate on the changes in the middle-­income groups. According to Goebel et al. (2010), incomes are becoming increasingly polarized and the middle-­income groups are shrinking. Enste et al. (2011), on the other hand, do not detect any clear trend but rather fluctuations in the shares of the various income groups and point to what is, by international standards, the high share of the middle-­income groups in disposable income. According to Enste et al., the slight increase at the lower margin is above all a consequence of the growth in one-­person households: when couples split up, the partner with a low income could be relegated to the lowest income group. Our more up-­to-­date data support the position adopted by Goebel et al. (2010). However, what is surprising about the debate, for several reasons, is the one-­sided focus on disposable household income after tax and transfers and the consequent neglect of the primary distribution. First, for people of working age, it makes a big difference whether you earn your income yourself or are reliant on transfers. Achieving your income position ‘under your own steam’ imparts self-­confidence and a feeling of greater social independence than being dependent on transfers, which can be restricted at any time by political decisions. Second, the growing inequality in primary incomes means that the welfare state is increasingly being required to compensate for growing income deficiencies in the middle and lower-­income groups. The increasing demands on the welfare state were offset by the spurts of strong economic growth between 2004 and 2008 and from 2009 to 2015 and the associated growth in employment. However, it is becoming increasingly uncertain whether this will continue to be possible in the event of a sustained economic crisis and with an ageing population. Finally, it takes decades for the social costs of an unequal primary distribution to show up fully in the social protection system. This applies in particular to the effects of low pay on future pensions. Thus growing inequality in

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 205

28/09/2016 08:24

206

Europe’s disappearing middle class?

primary incomes is a reliable indicator of increasing distribution problems in fiscal and social policy.

3. GAINFUL EMPLOYMENT, WORKING TIMES AND WAGES BY HOUSEHOLD FROM 1995–97 TO 2011–13 The distribution of equivalence-­ weighted household incomes can be influenced by changes in employment patterns. Thus households can slide down the income distribution if the total number of paid hours worked by household members is reduced and, conversely, rise up the distribution if the number of paid hours worked increases. Increasing the household labour supply by increasing individual working times or increasing the number of earners is a familiar compensation strategy for maintaining living standards in the event of wage reductions. However, such compensation strategies are not always successful if household members are offered only part-­time work because of inadequate skills or qualifications or if they are involuntarily unemployed. This is why we talk about employment patterns rather than preferred employment behaviour. In what follows, we describe changes in employment patterns. In order to obtain reliable results despite the inadequate number of cases in several cells, we pooled the data from three years to form two periods for comparison (1995–97 and 2011–13) (see Table 5.1).4 Households in Germany have become smaller over the past two decades. The share of the total population living in one-­person households has increased by a little under 4 percentage points.5 The biggest increase is in multi-­person households with low levels of gainful employment, that is, at most one part-­time job (up 5 percentage points). The share of households with several earners has evolved in different ways since the 1990s. The share of multiple-­earner and single-­breadwinner households has declined, while the share of households with secondary earners and two part-­time workers has increased. Thus, the increase in women’s employment in Germany has not been reflected in an increase in two-­earner households with two full-­time jobs but rather in an increase in the number of secondary earners working part-­time.6 Figure 5.4 shows the distribution of the population among the various types of household by equivalence-­weighted disposable income. In multi-­ person households with, for example, four people, all four individuals are included in the analysis. The high proportion of single-­person households and households in the lower-­income groups with no or only marginal levels of gainful employment is clearly evident. As household income rises, so the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 206

28/09/2016 08:24



The erosion of the German middle class ­207

Table 5.1 Evolution of household employment patterns from 1995–97 to 2011–13, Germany (share in total population) Employment pattern

All households

Multi-­person households with substantial levels of employment

1995–97

2011–13

1995–97

2011–13

17.6 19.8 21.0 3.5 21.5

12.3 21.8 15.0 4.0 26.5

28.4 32.0 34.0 5.6

23.2 41.0 28.2 7.6

16.6 100.0

20.3 100.0

100.0

100.0

Multiple earners Secondary earner Single breadwinners Dual part-­time Low level of gainful  employment Single Total

Source:  SOEP v30, authors’ calculations.

share of people in multiple-­earner households also rises, which is indicative of the close association between the presence of several gainfully employed individuals and membership of the higher-­earning households. The upper-­income groups have not only more earners but also significantly longer working times. Figure 5.5 shows average gross paid hours worked per household by income group. Gross paid hours include the hours actually worked together with paid holiday, paid sick days and statutory public holidays. The working times of households in the highest-­ income bracket are on average 2000 hours per year higher than those of households in the lowest-­income group. Since 1995–97, the number of paid hours worked has declined in all income groups, but the reductions are much greater in the lower-­income groups than in the middle and upper groups. Thus the increasing differentiation of household working times has contributed to the growing inequality among households in primary income. The third variable besides the number of earners and working time that determines the level of household market income is the hourly rate of pay. Average hourly pay in the upper-­income class, at 38.62 euros in 2011–13, was around five times greater than the hourly rate in the bottom-­income group. Hourly rates, just as working time, rise with income level, such that the differences in euros between the groups widen and are at their greatest (more than 17 euros) between the upper middle-­and top-­income groups. These average figures conceal large differences between employment

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 207

28/09/2016 08:24

208

Europe’s disappearing middle class?

100 90 80

43.8

33.4

22.9

50 40

36.0

5.1

17.8

2.9

4.2

16.0

16.1

20

5.2

10

9.9

0

4.4 0.9 Bottom

15.1 14.1 2.6

No or marginal employment

14.3

Single Dual part-time

25.0 27.2

15.7

30

17.4 15.4

70 60

25.8

Secondary earner

25.6 18.9 4.0 Lower middle

10.5 Intermediate middle

Single breadwinner

Multiple earners 21.3

Upper middle

29.0

Top

Source:  SOEP v30, authors’ calculations.

Figure 5.4 Employment patterns by income brackets (net disposable, equivalence-weighted household income in the previous year), Germany, 2011–13 (percentage share in total population) patterns. By far the highest hourly rate, about 52 euros, is found among households with two part-­ timers in the top-­ income bracket. In each income group, the hourly rates of households with a single breadwinner or with two part-­time workers are higher than those of multiple-­earner and secondary earner households (Table 5.2). This is evidence of a clear trade-­ off between pay and working time. As hourly rates of pay rise, so do the working-­time options and households are better able to manage with two part-­time earners or a single breadwinner. Thus high hourly rates of pay encourage households to reduce their working time, while low pay has the opposite effect, encouraging households to increase their labour supply.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 208

28/09/2016 08:24



The erosion of the German middle class ­209

Working time (per year)

3500

Top

Total

1995–97 2011–13 Change in % 2381 2083

2500 1500

Lower Intermediate Upper middle middle middle

2818 2539

3215 2902

3342 3243

150 130

110 2846 2539 90 70

1667 1329

50 30

500 –500

Change in %

4500

Bottom

10

–20.3%

–12.5%

–9.9%

–9.7%

–3.0%

–10.8%

–10 –30 –50

–1500

Note:  Also includes paid holidays, sick days and statutory public holidays; the basis for calculation is the working times of all economically active individuals. The inactive population is not included. Source:  SOEP v30, authors’ calculations.

Figure 5.5 Gross working times per household by income brackets, Germany, 1995–97 and 2011–13

Table 5.2 Average hourly wage of households by income brackets (equivalized disposable household income) and household structure, Germany, 2011–13 (euros) Income brackets Bottom Lower middle Intermediate middle Upper middle Top Total

Multiple earners 4.76 7.72 10.81 15.72 25.97 16.32

Secondary Single Dual earner breadwinner part-­time 7.11 10.87 14.46 20.35 40.79 18.80

7.70 11.94 18.06 27.48 51.35 23.50

7.98 12.55 17.77 24.10 52.07 19.21

Total 7.55 11.23 15.29 21.43 38.62 19.88

Source:  SOEP v30, authors’ calculations.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 209

28/09/2016 08:24

210

Europe’s disappearing middle class?

4. HOUSEHOLD DEMOGRAPHIC COMPOSITION AND QUALIFICATION STRUCTURE BY INCOME GROUP In this section we investigate which social categories are particularly strongly represented in the various income groups. Differences in individual groups’ relative shares in the various income groups point to possible causes of lower or higher incomes. Our data analysis reveals obvious differences in the composition of the various income classes by gender, age and qualifications. Taking into account all adult household members, the share of women in the bottom-­ income class is around 10 percentage points greater than in the top-­income class (Table 5.3). If the analysis is restricted to heads of household only, the gap widens still further to around 22 percentage points. This pronounced gender bias results from several factors. They include the high share of single mothers among low earners (Jaehrling et al. 2015), the marked gender pay gap in Germany, the low pensions women receive because of their discontinuous employment histories and the strong tax incentives that favour the single breadwinner model, particularly in the middle-­and high-­income classes (BMFSFJ 2011). It has already been noted that the share of single households is particularly high in the lower-­income classes (Figure 5.4). These households include many young people in education or training, as well as pensioners, particularly women. In multi-­person households, allocation to a particular income class can also depend on the number of household members to be provided for. Households with several earners but without any non-­earning members are in a particularly favourable Table 5.3 Household composition by gender and income brackets, Germany, 2011–13 (percentage) Income brackets

All adult household members

Bottom Lower middle Intermediate middle Upper middle Top Total

Head of household

Male

Female

Male

Female

44.8 44.3 48.6 50.6 54.6 48.4

55.2 55.7 51.4 49.4 45.5 51.6

46.9 48.9 56.6 60.5 68.5 55.7

53.1 51.1 43.4 39.5 31.5 44.4

Source:  SOEP version 30, authors’ analysis.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 210

28/09/2016 08:24



The erosion of the German middle class ­211

Table 5.4 Number of persons in multi-­person households by income brackets (equivalized disposable household income), Germany, 2011–13 Income brackets

Multiple earners

Bottom Lower middle Intermediate middle Upper middle Top Total

4.0 3.7 3.1 2.6 2.4 2.8

Secondary Single earner breadwinner 3.6 3.6 3.2 3.0 2.7 3.2

3.4 3.1 2.9 2.5 2.4 2.8

Dual part-­time

Total

3.1 3.3 3.0 2.7 2.3 2.9

3.4 3.3 3.1 2.7 2.5 2.9

Source:  SOEP version 30, authors’ calculations.

position. They can share the fixed living costs without having to provide for anyone else. In multi-­person households with more than marginal employment, the number of household members declines as household income rises. In the lower-­income class, there are 3.4 people per household to be ­provided for, compared with only 2.5 in the top-­income class (Table 5.4). These differences are even greater in households with multiple earners. To what extent and in what way changes in household structure have contributed to the growing inequality in the primary and secondary distribution is an issue that is hotly debated. Peichl et al. (2012) argue, in a descriptive analysis, that between 1991 and 2007 changes in the primary distribution were attributable mainly to the sharp reduction in household size in the lower-­income classes, while changes in the secondary distribution could be explained by household-­linked transfer payments. In a multivariate analysis of the risk of falling into poverty, however, Haupt and Nollmann (2014: 624) show that between 1992 and 2011 the risks of poverty were driven upwards mainly by unemployment, unstable employment and part-­time and low-­wage work. In the two lower-­income classes, the share of individuals with only low-­level qualifications is significantly higher than in the intermediate income class and the two upper-­ income classes. This did not middle-­ change between 1995–97 and 2011–13. However, the share of people with low-­level qualifications only in Germany fell sharply during this period, so that by 2011–13 the share of poorly qualified household heads was only around 32 per cent even in the lowest-­income class (Figures 5.6a and 5.6b). This means, conversely, that around two-­thirds of people in the lower-­income group have intermediate and higher-­level qualifications and

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 211

28/09/2016 08:24

212

Europe’s disappearing middle class? 100 90

7.8

5.3

9.9

16.2

26.9

80 70

50.7 50.8

60

60.8 67.5

60.6

50 59.8

40 30 20

41.4

45.5 33.8

10 0

Bottom

Lower middle

22.6 Intermediate middle Low

23.2

13.3 Upper middle

Middle

3.8 Top

Total

High

Source:  SOEP v30, authors’ calculations.

Figure 5.6a Educational level of the head of the household by income brackets (equivalized post-government income of household in the previous year), Germany, 1995–97 100 90

9.6

10.4

17.8

23.5 36.8

80

58.5

70 60

58.1

50

67.5 68.8

61.2

40

56.7

30 20 10 0

32.3

Bottom

38.4 22.0 Lower middle

13.5

6.5

Intermediate middle Low

Upper middle

Middle

3.1 Top

15.3 Total

High

Source:  SOEP v30, authors’ calculations.

Figure 5.6b Educational level of the head of the household (2011–13) by income brackets (equivalized post government income of household in the previous year), Germany

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 212

28/09/2016 08:24



The erosion of the German middle class ­213

that these good qualifications are not sufficient to ensure inclusion in the middle-­income classes. In section 8.2 we see that this can be explained by the expansion of the low-­wage sector.

5. GROUP/CLASS ANALYSIS BY SELF-­DEFINITION AND OCCUPATIONAL CLASSIFICATION Income is just one characteristic of class affiliation. Other classifications are based on self-­definition or occupational position. There can be inconsistencies between these various characteristics. It should also be noted that classifications based on self-­definition and occupational position take the individual rather than the household as their starting point; as a result, members of the same household may be allocated to different categories. Figure 5.7 shows that only a very small share of Germans belong to the lower class although – as we have seen – the economic foundations of many households are crumbling. The largest and growing segment of the population regards itself as middle class. This reflects above all the growing share of white-­collar workers in the economically active population. The vast majority of blue-­collar workers classify themselves as working class, unless they have moved up and become white-­ collar workers or self-­ employed (Figure 5.7). The extraordinarily high level of assignment to the

Upper middle and upper class

11 13 60 64

2 5

37

Middle class

27 22

57

Working class

2 2

1990 2012

53

Lower class

West

39 3 4

1990 2012 East

Source:  Destatis and WZB (2013: 186).

Figure 5.7  Self-definition of middle class (1990 and 2012), Germany

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 213

28/09/2016 08:24

214

Europe’s disappearing middle class?

manual worker category in East Germany in 1990 can be explained by the politics of the German Democratic Republic (GDR), which defined itself as a workers’ and peasants’ state. Even today many white-­collar workers in eastern Germany clearly still regard themselves as belonging to the working class. The Erikson-­Goldthorpe-­Portocarero class schema (EGP) plays a key role in sociological research on inequality. It is a multi-­dimensional classification schema that assigns individuals to classes according to their position in the employment system (Ganzeboom and Treiman 2003). Three groups of workers are identified: employers, self-­employed workers and employees. Employees are further differentiated by reference to the mode of regulation of their employment. ‘Blue-­collar wage relationships’, in which performance is evaluated on the basis of the number of units produced and time taken and workers are under the proprietor’s direct control, are differentiated from ‘service relationships’, in which the execution of tasks is delegated to the workforce. In addition to these two basic types there are other activities that are informed in various ways by these two basic types. From the combination of these characteristics, ten and later 11 categories were formed in such a way that the cells were adequately populated. For the analysis based on the SOEP, data on the International Standard Classification of Occupations (ISCO) code, income, level of education and position in occupation are used. Data on authority to give directives (managerial/supervisory authority) have been collected in the SOEP only since 2007 and cannot therefore be used to differentiate EGP status (Lengfeld and Hirschle 2009: 385; SOEP Group 2014). For our own analysis, the EGP classes are derived virtually unchanged from the SOEP. We have simply allocated agricultural workers to the unskilled and semi-­skilled manual worker class. Manual workers with managerial/supervisory responsibilities are categorized in the SOEP with skilled workers (Lengfeld and Hirschle 2009: 385). We exclude self-­employed workers and the economically inactive. Lengfeld and Hirschle (2009: 385) allocate the groups shown in Figure 5.8 to the lower, middle and upper class. In their schema, the upper and middle service classes constitute the upper class. The lower service class constitutes the upper middle class, while the intermediate middle class is made up of routine non-­manual employees. Routine workers in services and sales are allocated together with skilled manual workers to the lower middle class and unskilled and semi-­skilled manual workers constitute the lower class. In our view, the allocation of skilled manual workers to the lower middle class is the least comprehensible aspect of this classification. However, the EPG classification does facilitate more detailed investigation

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 214

28/09/2016 08:24



The erosion of the German middle class ­215

100 12.8

12.6

20.6

22.8

60

11.0

10.9

50

12.5

90 80 70

13.4

40 30

14.6

26.0

9.7 12.7

22.4

18.4

15.5

20.7

21.9

21.7

1995

2004

2013

20 10 0

Semi- and unskilled manual

Skilled manual

Routine service-sales

Routine non-manual

Low service

High/middle service

Source:  SOEP v30, authors’ calculations.

Figure 5.8  Occupational structure (EGP), 1995, 2004 and 2013 of which groups of employees are particularly affected by the increase in low-­wage work. The risk of low pay increased between 1995 and 2013 for unskilled and semi-­skilled workers in particular (Figure 5.9). Virtually one in every two workers in this group is now low paid, the next largest group being employees in routine jobs in services and sales. The risk of low pay has increased for skilled manual workers but is close to the average value for the economy as a whole, which suggests that their allocation to the lower middle class, alongside routine workers in services and sales, is mistaken. Otherwise, the risk of low pay declines as class affiliation rises. There has been a significant rise in the risk of low pay for skilled manual workers and lower classes who in the past were protected against low pay mainly by collective agreements. The higher classes are at low risk of low pay, a situation that scarcely changed over the entire observation period. For Lengfeld and Hirschle (2009), one advantage of using the EGP classification rather than an income-­based classification is that it is more

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 215

28/09/2016 08:24

216

Europe’s disappearing middle class?

50 45 35

Semi- and unskilled manual Routine service-sales

30

Skilled manual

25

Total

20

Routine non-manual

15

Low service

40

10

High/middle service

5 0

1995

2004

2013

Source:  SOEP v30, authors’ calculations.

Figure 5.9 Risk of low pay (hourly pay less than 2/3 of the median wage) by occupational groups (EGP classification) 1995, 2004 and 2013 directly linked to the labour market, because income-­based classifications also include investment and rental income. A further advantage, in their view, is that in an occupation-­based classification the partner’s employment position has no influence on class affiliation and that allocation to a class on the basis of occupation is more stable than allocation on the basis of income. However, these points can also be adduced as arguments against an occupation-­based classification and in favour of an income-­based classification. If someone has assets at their disposal, they should be included in a classification of social status. Changes in the requirements or employment conditions in occupations are also disregarded. If employment conditions are being eroded – for example, as a result of the expansion of atypical employment or low-­wage jobs in individual ­occupations – this will not be evident from an occupation-­based classification.

6. THE ROLE OF TAXES, SOCIAL SECURITY CONTRIBUTIONS AND BENEFITS IN REDISTRIBUTION It was noted in section 2.2 above that disposable household income was significantly less unevenly distributed than market incomes. In this section,

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 216

28/09/2016 08:24



The erosion of the German middle class ­217

the object of investigation is the role played by private transfers, taxes, social insurance contributions, insurance benefits and means-­tested social transfers in redistribution. No account is taken of the fact that the tax and social protection system not only influences disposable income ex post through redistribution but also has far-­reaching effects ex ante on primary distribution, for example by offering incentives to enter the labour market or increasing productivity through investment in education or training. Table 5.5 shows that the tenth decile in particular benefits substantially from private pension plans and transfers. Social insurance cash benefits are lowest in the lowest decile, where low incomes or an employment history with many spells of inactivity or unemployment make it difficult to build up entitlements. For all the other deciles, social insurance cash benefits play a nearly equally important role. This is not surprising, because contributions and benefit levels are linked to previous income. The lower deciles receive a somewhat higher share of ‘other transfers’, because some tested benefits are paid social insurances benefits are capped. Means-­ ­predominantly to the lower deciles. As a result of the extension of social insurance benefits in the post-­ war period a large share of the social budget in the German Bismarckian welfare state is funded by social insurance contributions. Since most employees are covered by social insurance schemes and have to contribute a certain percentage of their wages, the middle and higher deciles contribute more in absolute terms than the poor but as a result also acquire higher entitlements, so that the redistribution effect is limited. The progressive tax system, however, raises most taxes from the higher deciles. The tenth decile contributes around 45 per cent of the total revenue from income tax, while the lowest decile pays virtually no income tax at all. The overall redistribution effect can be measured by the Gini coefficient. In 2012, the Gini coefficient for market incomes was 0.50. It was reduced by welfare transfers to 0.35. Social insurance contributions do not reduce the Gini coefficient any further because they are paid proportionally to income, but the progressive tax system brought the Gini coefficient down to 0.29. An important element of developed welfare states is benefits in kind, whose distributive effects are not taken into account in Table 5.5. In Germany, benefits in kind account for 17 per cent of the net income of all households. This is significantly below the level of the Scandinavian welfare states, which invest considerably more in childcare and education in particular. There is no doubt that benefits in kind have to be taken into account in international comparisons of living standards, because such benefits certainly relieve households from a good deal of otherwise essential expenditure, for private education or health care, for example. Many attempts have

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 217

28/09/2016 08:24

218

VAUGHAN 9781786430595 PRINT (M4051).indd 218

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

Billion euros 1st decile 2nd decile 3rd decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 10th decile Total

Equivalized household income

19.9 43.8 63.3 87.1 103.5 142.5 170.6 203.2 269.4 447.0 1550.3

2.7 2.6 2.6 3.1 3.3 3.0 4.3 4.6 5.2 13.2 44.6

16.7 29.1 35.9 32.8 34.0 26.4 27.7 30.0 29.9 33.2 295.9

5.5 6.3 5.2 5.6 4.9 5.6 5.0 4.9 3.9 4.1 50.9

13.0 6.6 4.0 2.2 2.1 1.9 0.7 1.2 0.5 1.0 33.4

57.0 86.8 110.0 129.7 149.2 178.8 207.7 245.2 303.9 495.2 1963.4

Market Private Cash benefits Gross income1 pensions and income Social insurance Other Means-­tested transfers2 benefits3 transfers4 benefits5 7.5 17.5 25.0 32.5 37.3 47.6 54.7 62.6 73.4 76.4 434.5

Social contributions

Table 5.5 Private household income and redistribution by taxes and transfers, Germany

0.1 1.2 3.4 6.5 9.7 16.0 21.9 31.1 48.1 113.2 251.2

Income tax

49.4 68.0 81.6 90.7 102.2 115.2 131.1 151.5 182.5 305.5 1277.8

Net income

219

VAUGHAN 9781786430595 PRINT (M4051).indd 219

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

1.3 2.8 4.1 5.6 6.7 9.2 11.0 13.1 17.4 28.8 100.0 0.50

6.1 5.9 5.8 7.0 7.4 6.6 9.6 10.4 11.7 29.6 100.0 0.49

5.7 9.8 12.1 11.1 11.5 8.9 9.4 10.2 10.1 11.2 100.0 0.38

10.8 12.3 10.2 11.0 9.7 10.9 9.7 9.6 7.6 8.1 100.0 0.37

39.0 19.8 12.1 6.7 6.4 5.8 2.2 3.6 1.5 3.0 100.0

Source:  Bach et al. (2015: 151).

Notes: 1 Income from employment, asset income including rental value of residences. 2 Private pensions, company pensions, alimony payments, other private transfers. 3 Social security pensions, civil service pensions, unemployment benefits, care allowance. 4 Child allowance, maternity benefit, grants, tax allowance for housing. 5  Unemployment allowance, welfare aid, housing subsidy.

Structure in % 1st decile 2nd decile 3rd decile 4th decile 5th decile 6th decile 7th decile 8th decile 9th decile 10th decile Total Gini 2.9 4.4 5.6 6.6 7.6 9.1 10.6 12.5 15.5 25.2 100.0 0.35

1.7 4.0 5.8 7.5 8.6 11.0 12.6 14.4 16.9 17.6 100.0 0.35

0.0 0.5 1.3 2.6 3.8 6.4 8.7 12.4 19.1 45.1 100.0

3.9 5.3 6.4 7.1 8.0 9.0 10.3 11.9 14.3 23.9 100.0 0.29

220

Europe’s disappearing middle class?

been made in recent years to quantify the distributive effects of benefits in kind. A counterfactual calculation shows that inequality would be considerably higher if benefits in kind had to be financed privately. In Germany, the Gini coefficient for 2007 would rise from 0.2995 to 0.3508 (Verbist and Matsanganis 2014: 203). This is hardly surprising, because social benefits in kind are usually very labour intensive which, if privately financed, would give rise to high costs that would in many cases be prohibitive for the lower-­ income groups. Funding via the welfare state is a mechanism that prevents the lower-­income groups from being excluded from basic social benefits, such as education and health care (Bosch and Wagner 2005). However, calculations of the distributive effects of individual benefits in kind (health care, compulsory schooling, continuing education and training, and old-­age care) are not convincing. In Germany, for example, they show a high level of redistribution in favour of poorer households. However, this is largely attributable to the fact that students in Germany usually live alone and are included among the low-­earning households (Verbist and Matsanganis 2014: 200). The high shares of school drop-­outs and the strong link between higher education and social background in Germany suggest, on the contrary, that the redistributive effect of education expenditure is somewhat less strong (Schwahn and Schwarz 2015: 36). earning households contribute disproportionately to the While high-­ revenue raised through income tax, their contribution to the revenue from indirect taxes is small because of their lower propensity to consume. In 2011 the savings rate in the bottom decile of the income distribution was just 1.8 per cent, while in the top decile it was 17 per cent (Wagner and Brenke 2013: table 2). Consequently, indirect taxes are regressive and low-­ income households contribute a disproportionate share of the revenue raised in this way (Figure 5.10). Since the share of indirect taxes has risen sharply in the past two decades7 and the top rate of income tax was reduced in several stages from 1999 onwards from 53 per cent to 42 per cent, the tax system as a whole is less redistributive than in the 1990s. The increase of the top rate for incomes in excess of 250 000 euros to 45 per cent in 2007 has not changed the situation in any way.

7. EMPLOYMENT-­RELATED DRIVERS OF FALLING OR IMPROVING MIDDLE-­CLASS STANDARDS The pronounced shrinking of the middle-­income classes in the primary distribution, like their only modest decline in the secondary distribution, has been explained by the interaction of various, usually politically

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 220

28/09/2016 08:24



The erosion of the German middle class ­221

16

16

14

Percentage contributions to revenue, indirect taxes (%)

12

Percentage contributions to revenue, social insurance contributions of employees (%)

10

Percentage contributions to revenue, income tax (%)

14 12 10

200≤

190–199

180–189

170–179

160–169

150–159

140–149

130–139

120–129

110–119

90–99

100–109

0 80–89

0 70–79

2

60–69

2 50–59

4

40–49

4

30–39

6

20–29

6

= 2 full-­time; secondary earner: 1 full-­time + > = 1 part-­time; single breadwinner: 1 full-­time; dual part-­time: >= 2 part-­ time; low level of gainful employment: maximum 1 part-­time; single: only one person in household. 5. If the number of households were considered rather than the number of people, one-­ person households would have a much higher share because multi-­person households are included several times at individual level in the analysis. 6. This increase could constitute a compensation strategy for a reduction in the volume of working time by increasing the number of earners. In order to be able to assess this accurately, however, we would have to investigate the evolution of individual households over time in order to ascertain whether a second-­earner household, for example, was in the past a two-­earner or single-­breadwinner household. 7. The share of value added tax (VAT) alone rose between 1992 and 2012 from 27 per cent to 32.4 per cent; available at http://www.bpb.de/nachschlagen/zahlen-­und-­fakten/soziale-­ situation-­in-­deutschland/61874/steuereinnahmen (accessed 20 July 2015). 8. The authors here do not adopt the standard low-­wage threshold of two-­thirds of the median wage, because their aim is to calculate the effects of a European minimum wage set at 60 per cent of the median wage. 9. Available at http://www.sozialpolitik-­aktuell.de/tl_files/sozialpolitik-­aktuell/_Politikfelder/ Alter-­Rente/Datensammlung/PDF-­Dateien/abbVIII37.pdf (accessed 21 July 2015).

REFERENCES Arndt, C. (2012), Zwischen Stabilität und Fragilität: Was wissen wir über die Mittelschicht in Deutschland? Berlin: Konrad Adenauer Stiftung. Autorengruppe Bildungsberichterstattung (2014), ‘Bildung in Deutschland 2014. Ein indikatorengestützter Bericht mit einer Analyse zur Bildung von Menschen mit Behinderungen’, Bielefeld.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 237

28/09/2016 08:24

238

Europe’s disappearing middle class?

Bach, S., M. Grabka and E. Tomasch (2015), ‘Steuer-­und Transfersystem: hohe Umverteilung vor allem über die Sozialversicherung’, in DIW Wochenbericht, no. 8, pp. 147–56. Beimann, B., R. Kambeck, T. Kasten and L.-­H. Siemers (2011), ‘Wer trägt den Staat? Eine Analyse von Steuer-­und Abgabenlasten’, RWI Position, 43 (April), 1–33. Bundesministeriums für Familie, Senioren, Frauen und Jugend (Federal Ministry for Family Affairs, Senior Citizens, Women and Youth) (BMFSFJ) (2011), Erster Gleichstellungsbericht der Bundesregierung, Neue Wege – Gleiche Chancen – Gleichstellung von Frauen und Männern im Lebensverlauf, Berlin. Bohlender, A., S. Huber and N.A. Siegel (2011), ‘SOEP 2011 Methodenbericht oekonomischen Panels in zum Befragungsjahr 2011 (version 28), des Sozio-­ den Fortführungsstichproben und dem Aufwuchssample’, TNS Infratest Sozialforschung. Bosch, G. (2011), ‘The German labour market after the financial crisis: miracle or just a good policy mix?’, in D. Vaughan-­Whitehead (ed.), Work Inequalities in the Crisis: Evidence from Europe, Cheltenham, UK and Northampton, MA, USA: Edward Elgar and Geneva: ILO, pp. 243–77. Bosch, G. (2014), ‘Facharbeit, Berufe und berufliche Arbeitsmärkte’, WSI-­ Mitteilungen, 67 (1), 5–13. Bosch, G. (2015), ‘The German welfare state: from an inclusive to an exclusive Whitehead (ed.), The European Social Bismarckian model’, in D. Vaughan-­ Model in Crisis: Is Europe Losing its Soul?, Cheltenham, UK and Northampton, MA, USA: Edward Elgar and Geneva: ILO, pp. 175–229. Bosch, G. and Wagner, A. (2005), ‘Why do countries have such different service-­ ­ sector employment rates?’, in G. Bosch and S. Lehndorff (eds), Working in the Service Sector: A Tale from Different Worlds, London: Routledge, pp. 74–102. Bosch, G., T. Kalina and C. Weinkopf (2014), ‘25 Jahre nach dem Mauerfall – Ostlöhne holen nur schleppend auf’, in Deutsches Institut für Wirtschaftsforschung (DIW) SOEP Papers on Multidisciplinary Panel Data Research Berlin, no. 711. Bundesagentur für Arbeit (2014), ‘Arbeitsmarkt in Deutschland. Zeitreihen bis 2013’, Nürnberg. Castells, M. (1996), The Rise of the Network Society. The Information Age, vol. 1, Oxford: Blackwell. Destatis (2014), Finanzen und Steuern. Personal des öffentlichen Dienstes, Fachserie 14, Reihe 6, Ergebnisse der Personalstandstatistik 2014, Wiesbaden. Destatis and WZB (Statistisches Bundesamt und Wissenschaftszentrum Berlin für Sozialforschung) (eds), (2013), Datenreport 2013, Ein Sozialbericht für die Bundesrepublik Deutschland, Bonn. Enste, D.H., V. Erdmann and T. Kleineberg (2011), Mythen über die Mittelschicht. Wie schlecht steht es wirklich um die gesellschaftliche Mitte?, Munich: Roman Herzog Institut e.V. Esping-­Andersen, G. (1990), Three Worlds of Welfare Capitalism. Princeton, NJ: Princeton University Press. Fernández-­Macías, E. and C. Vacas-­Soriano (2013), A Coordinated EU Minimum Wage Policy? Dublin: Eurofound. Frick, J.R., S.P. Jenkins, D.R. Lillard, O. Lipps and M. Wooden (2008), ‘Die internationale Einbettung des Sozio-­ oekonomischen Panels (SOEP), im

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 238

28/09/2016 08:24



The erosion of the German middle class ­239

Rahmen des Cross-­ National Equivalent File (CNEF)’, Vierteljahrshefte zur Wirtschaftsforschung, 77 (3), 110–29. Fritsch, M., A.S. Kritikos and A. Sorgner (2015), ‘Verdienen Selbständige tatsächlich weniger als Angestellte?’, DIW-­Wochenbericht, no.7, pp. 134–47. Ganzeboom, H.B.G. and D.J. Treiman (2003), ‘Three internationally standardised measures for comparative research on occupational status’, in J.H.P. Hoffmeyer-­ Zlotnik (ed.), Advances in Cross-­National Comparison, New York: Kluwer and Plenum, pp. 159–93. Geißler, R. (2002), Die Sozialstruktur Deutschlands, Wiesbaden: VS-­Verlag für Sozialwissenschaften. Geißler, R. (2011), Die Sozialstruktur Deutschlands. Zur gesellschaftlichen Entwicklung mit einer Bilanz zur Vereinigung, 6th edn, Wiesbaden: VS-­Verlag für Sozialwissenschaften. Geyer, J. (2011), ‘Riester-­Rente: Rezept gegen Altersarmut?’, DIW-­Wochenbericht, no. 47. Goebel, J., M. Gornig and H. Häußermann (2010), ‘Polarisierung der Einkommen. Die Mittelschicht verliert’, DIW-­Wochenbericht, no. 24, 2–9. Grabka, M.M. and C. Westermeier (2014), Anhaltend hohe Vermögensungleichheit in Deutschland, Wochenbericht DIW, no. 9, 164. Grimshaw, D. and G. Bosch (2013), ‘The intersections between minimum wage and collective bargaining institutions’, in D. Grimshaw (ed.), Minimum Wages, Pay Equity, and Comparative Industrial Relations, New York and London: Routledge, pp. 50–80. Haipeter, T. (2011), ‘“Unbound” employers’ associations and derogations: erosion and renewal of collective bargaining in the German metalworking industry’, Industrial Relations Journal, 42 (2), 174–94. Haupt, A. and G. Nollmann (2014), ‘Warum werden immer mehr Haushalte von Armut gefährdet? Zur Erklärung erhöhter Armutsrisikoquoten mit unbedingten Quantilregressionen’, Kölner Zeitschrift für Soziologie und Sozialpsychologie, 66 (4), 603–27. Institut für Sozialforschung und Gesellschaftspolitik (ISG), (2011), ‘Überprüfung der These einer “schrumpfenden Mittelschicht” in Deutschland’, Cologne. Jaehrling, K., T. Kalina and L. Mesaros (2015), ‘A paradox of activation strategies: why increasing labour market participation among single mothers failed to bring down poverty rates’, Social Politics, 22 (1), 86–110. Jaehrling, K. and P. Méhaut (2013), ‘Varieties of institutional avoidance: employers’ strategies in low-­waged service sector occupations in France and Germany’, Socio-­Economic Review, 11 (4), 687–710. Kalina, T. and C. Weinkopf (2014), ‘Niedriglohnbeschäftigung 2012 und was ein gesetzlicher Mindestlohn von 8,50 € verändern könnte’, IAQ-­Report 2014-­02, Duisburg: Institut Arbeit und Qualifikation. Kalina, T. and C. Weinkopf (2015), ‘Niedriglohnbeschäftigung 2013: Stagnation auf hohem Niveau’, IAQ-­Report 2015-­03, Internet-­Dokument, Institut Arbeit und Qualifikation, Duisburg, accessed 11 April 2016 at http://www.insee.fr/fr/ffc/ docs_ffc/ref/SALFRA07ae.PDF. Koubi, M. and B. Lhommeau (2007), Les effets de diffusion de court terme des hausses du Smic dans les grilles salariales des enterprises de dix salariés ou plus sur la période 2000–2005, French National Institute of Statistics and Economic Studies, accessed 11 April 2016 at http://www.insee.fr/fr/ffc/docs_ffc/ ref/SALFRA07ae.PDF.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 239

28/09/2016 08:24

240

Europe’s disappearing middle class?

Lengfeld, H. and J. Hirschle (2009), ‘Die Angst der Mittelschicht vor dem sozialen Abstieg. Eine Längsschnittanalyse 1984–2007’, Zeitschrift für Soziologie, 5, 379–98. Organisation for Economic Co-­ operation and Development (OECD) (1996), Employment Outlook, Paris: OECD. Peichl, A., N. Pestel and H. Schneider (2012), ‘The impact of changes in household structure on income distribution in Germany’, Review of Income and Wealth, 58 (1), 118–41. Schelsky, H. (1953), Wandlungen der deutschen Familie in der Gegenwart, Dortmund: Ardey Verlag. Schulten, T. (2014), ‘WSI-­Mindestlohnbericht 2014 – stagnierende Mindestlöhne’, WSI-­Mitteilungen, 67 (2), 132–9. Schwahn, F. and N. Schwarz (2015), ‘Einkommenskonzepte zur Wohlfahrtsmessung: Soziale Sachleistungen – Ein Einkommensbestandteil?’, WISTA – Wirtschaft und Statistik, no. 3, 25–40. SOEP Group (2014), SOEP 2013 – Documentation of Person-­related Status and Generated Variables in PGEN for SOEP v30. SOEP Survey Papers 250: Series D, Berlin: DIW and SOEP. Steiner, V. and J. Geyer (2010), Erwerbsbiografien und Alterseinkommen im demografischen Wandel – eine Mikrosimulationsstudie für Deutschland, Berlin: Politikberatung kompakt DIW 55. level wage changes and spillover effects of Stewart, M.B. (2011), ‘Individual-­ minimum wage increases’, Warwick Economic Research Papers No. 965, University of Warwick. Vaughan-­Whitehead, D. (2011), Work Inequalities in the Crisis: Evidence from Europe, Cheltenham, UK and Northampton, MA, USA: Edward Elgar and Geneva: ILO. Verbist, G. and M. Matsaganis (2014), ‘The redistributive capacity of services in the EU’, in B. Cantillon and F. Vandenbroucke (eds), Reconciling Work and Poverty Reduction: How Successful Are European Welfare States?, Oxford: Oxford University Press, pp. 185–211. Visser, J. (2015), ‘Institutional Characteristics of Trade Unions, Wage Setting, State Intervention and Social Pacts (ICTWSS), an international database’, Amsterdam Institute for Advanced Labour Studies (AIAS), Amsterdam. Voss, D. and C. Weinkopf (2012), ‘Niedriglohnfalle Minijob’, WSI-­Mitteilungen, 1, 5–12. Wagner, G.G. and K. Brenke (2013), ‘Ungleiche Verteilung der Einkommen bremst das Wirtschaftswachstum’, Wirtschaftsdienst, 93 (2), 110–16. Wagner, G.G., J.R. Frick and J. Schupp, J.(2007), ‘The German Socio-­Economic Panel Study (SOEP) – scope, evolution and enhancements’, Schmollers Jahrbuch, 127 (1), 139–69. Wrohlich, K., E. Berger, J. Geyer, P. Haan, D. Sengül, K.C. Spieß and A. Thiemann (2012), Elterngeld Monitor, DIW Politikberatung kompakt, no. 61. WSI Tarifarchiv (2003), ‘Tarifpolitischer Jahresbericht 2003: Konflikt um Tarifautonomie–Kampf um Lohnprozente’, Düsseldorf, accessed 17 July 2015 at http://www.boeckler.de/pdf/p_ta_jb_2003.pdf. WSI Tarifarchiv (2015), ‘Statistik Tarifbindung’, Düsseldorf, accessed 17 March 2015 at http://www.boeckler.de/wsi-­tarifarchiv_2257.htm.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 240

28/09/2016 08:24



The erosion of the German middle class ­241

APPENDIX 5A.1 Table 5A.1 Summary of world of work-­related drivers of falling middle-­ class standards in Germany since the mid-­1990s High  unemployment Low productivity  in eastern Germany Change of  strategy by employers and employers’ associations

Product market  deregulation

Fragmenting  firms Derogation  clauses in collective agreements Deregulation of   labour market

Traditional  family model – women directed into precarious employment

Weakening of employees’ bargaining position as a result of the rise in unemployment in the united Germany from 1.9 million in 1990 to 4.9 million in 2003, an increase of 3 million Transfer of the western German collective bargaining system and rapid pay convergence came up against low productivity in eastern German firms, mostly newly founded, which withdrew from the employers’ associations and did not implement the collective agreements Instead of the traditional corporatist support for an inclusive pay system before 1990, increasing withdrawals from the employers’ associations and use of opportunities for ‘institutional avoidance’, particularly through outsourcing and use of non-­standard employment forms. Refusal by employers’ associations to agree to applications to make collective agreements generally binding and membership offered without obligation to adhere to collective agreements Transfer of previously publicly operated services into the hands of private providers; increasing wage competition, because collectively agreed rates not generally applicable; increased pressure on public and privatized providers to engage in concession bargaining Reduction in internal value creation and cost-­driven outsourcing of activities to sectors with lower rates of pay or with no collective bargaining coverage Economic and political pressure leads to acceptance of derogation clauses in collective agreements; agreements on lower pay and longer working hours in exchange for promises on job security Deregulation of temporary agency work, expansion of mini-­jobs, abolition of unemployment assistance, reduction in period of entitlement to unemployment benefit for older workers and tightening of restrictions on jobseekers’ rights to refuse a job offer, thereby increasing pressure on the unemployed to accept low-­paid work Splitting system for assessing married couples’ income tax liability, derived social insurance benefits and tax-­free mini-­jobs create strong financial incentives for married women either not to work at all or to take marginal part-­time jobs

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 241

28/09/2016 08:24

242

Europe’s disappearing middle class?

Table 5A.1  (continued) Job cuts in   public sector Reduction in   pension levels

Reduction in number of public-­sector workers by more than 2 million since 1990 as a result of privatization, outsourcing and cuts in services; well-­paid middle-­income activities replaced by less well paid jobs Repeated reductions in pension levels; introduction of an optional, subsidized, private pension scheme (‘Riester’ pension), take-­up of which is very low among low earners; new pensions for men already lower than existing entitlements; old-­age poverty expected to rise over the long term

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 242

28/09/2016 08:24



The erosion of the German middle class ­243

Table 5A.2 Summary of world of work-­related drivers of improved middle-­class standards since the mid-­1990s Employment growth   since 2005 ‘Employment  miracle’ in the financial crisis

Increasing share of   skilled workers Introduction of  long-­term care insurance in 1995 Development of a   new family model

Act on the  Strengthening of Free Collective Bargaining

From 2005 to 2013, employment rose by 3 million and unemployment declined by 2 million, which improved employees’ bargaining position No mass redundancies in the financial crisis; instead, reduction in working time through short-­time working and other measures; well-­paid middle-­income jobs safeguarded, particularly in manufacturing sector, and avoidance of youth unemployment through the recruitment of 500 000 new trainees in the crisis Expansion of higher education and vocational training; German companies increasingly concerned to retain their skilled workers Many households freed of burden of care costs; by 2013 already 2.73 million benefit recipients Expansion of childcare; 1996 legal entitlement to a nursery place (children aged 3–6), because 2013 children aged 1 and 2 legally entitled to nursery place as well; expansion of all-­day schooling and introduction of 14 months’ paid parental leave; increase in number of mothers in insurable employment Introduction of a minimum wage set at 8.50 euros on 1 January 2015 and easing of process for declaring collective agreements legally binding: sharp rise in low wages; no negative employment effects observed to date; possible positive secondary effects through collective agreements based on the minimum wage

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 243

28/09/2016 08:24

6. The middle classes in the Greek Great Depression: Dissolution or resilience? Maria Karamessini and Stefanos Giakoumatos 1. INTRODUCTION The size, composition and growth of the middle classes in Greece has historically depended on the opportunities provided by the state to the lower classes for upward social mobility through education, through employment in the public sector and access to land and real estate property (Tsoukalas 1980, 1987; Elefantis 1991), a favourable context for petty entrepreneurship outside agriculture (Karabelias 1982; Karamessini 1992; Lyberaki and Mouriki 1996; Moschonas 1986) and, since the 1980s, through democratic industrial relations and employment opportunities for women (Karamessini 2012). Although social expenditure grew rapidly over the past 40 years, and almost reached the European Union (EU) 15 average before the current crisis, the Greek welfare state has remained insufficiently redistributive (Papatheodorou and Petmesidou 2004), reproducing rather than reducing income inequalities. This explains to a great extent why the latter have remained among the highest in the EU over recent decades and why, despite the significant expansion of the middle classes during 1993–2008, their relative size was smaller than the EU average at the onset of the 2007–08 global financial crisis (Bigot et al. 2011). In 2008, Greece entered a recessionary period during which the country lost 26 per cent of its gross domestic product (GDP), disposable income contracted by about 30 per cent and private consumption by 35 per cent, while unemployment reached 27 per cent of the labour force, the risk of poverty or social exclusion 36 per cent of the population, and ‘anchored’ poverty, according to the 2008 poverty line, 48 per cent. Since the 2009 sovereign debt crisis, recession has been fuelled by the harsh ‘austerity’ imposed by the Troika (the European Commission, the European Central Bank and the International Monetary Fund) and implemented by Greek governments as a prerequisite for two loans granted to Greece by its 244

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 244

28/09/2016 08:24



The middle classes in the Greek Great Depression ­245

euro-­zone partners and the International Monetary Fund (IMF) in 2010 and 2012. Unsurprisingly, the magnitude and severity of the economic crisis caused a rupture in the reproduction of the middle classes, whose expansion has been interrupted. At the same time, the dramatic social consequences of the deep recession engendered by sustained austerity have raised concern about the socially disruptive impact of the rapid expansion of poverty and deprivation not only among the traditionally vulnerable population groups but also the ‘new poor’ coming from ruined middle-­class strata. Predictably, the argument that the middle classes are financially converging towards the lower classes, leading to higher social inequality and polarization, has become widespread. At the subjective level, the fear or actual loss of savings and property beside income and jobs, exacerbated by seizures of money in bank accounts and real estate for outstanding debts to fiscal authorities, banks and social security agencies, has heightened financial insecurity among the middle classes. Given these circumstances, our research has focused on the following crucial questions. Have the middle classes in Greece been decimated during the economic crisis or are they manifesting resilience to the attack they are under? Losing social status is a constant fear of the middle classes, but to what extent is a generalized fear grounded? Has their financial situation deteriorated more than that of the lower and upper classes? How is their relative position linked to the different effects of the crisis on the various sectors, occupations, kinds of jobs and labour force groups? Recent literature on the middle classes often discusses their shrinking or squeezing as part of an increased polarization of the income distribution in countries that had witnessed rising inequalities well before the 2008 global financial crisis. However, it has failed to establish a universal trend, especially in Europe, because in many countries the middle classes remained constant or even grew through the 1990s and 2000s (Bigot et al. 2011). Furthermore, the long-­term trends in their relative position have been attributed to various determinants. For instance, to explain the ­40-­year stagnation in middle-­class incomes in the United States, the annual report for 2015 of the Council of Economic Advisers of President Obama has identified three key factors: productivity growth, labour force participation and income inequality (Council of Economic Advisers 2015). To summarize the findings, recent literature on the middle classes has established that structural changes and processes in the labour market are crucial for changes in income inequality and middle-­class incomes. Nevertheless, the final position of the middle class in the income distribution is also determined by the redistributive function of the welfare state (Pressman 2007, 2009; Dallinger 2013).

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 245

28/09/2016 08:24

246

Europe’s disappearing middle class?

This chapter focuses on the impact of changes in the world of work on the size, income and evolution of the middle classes in Greece since the first half of the 1990s, after estimating and briefly commenting on the effect of redistributive policies. We argue that the main reasons for the expansion of the middle classes during the growth period (1993–2008) were an increasing share of dual-­ earner families, rising educational attainment and the shift of employment dynamics towards medium-­and high-­paid occupations. Since 2008, the Greek Great Depression has ruined large parts of the middle classes and entailed great financial losses for the remainder, thus generalizing insecurity about the future and the fear of social downgrading, especially for their offspring. The relative size of the middle classes has shrunk during the crisis as a result of the downward social mobility of large numbers of middle-­class households. However, in a period of enormous income losses by all classes, the remaining middle classes have, on average, increased their income differences with the lower classes and retained these differences with the upper ones. The main drivers of this development are the much higher incidence of unemployment among the lower classes and comparable wage reductions among upper and middle classes during the crisis. In addition to the introduction, the chapter has four sections. Section 2 describes the evolution of the middle classes over the period of economic growth (1993–2008) and the first four years of the current economic crisis (2008–12) and assesses the role of redistributive policies in this evolution. Section 3 sketches the profile of the middle classes at the beginning of the period under scrutiny, in accordance with the distinctive labour market characteristics of their members, and identifies the major transformations in the world of work that account for the trends in their size and composition during the growth and crisis periods. Section 4 provides detailed evidence on the aforementioned causal relationship by focusing on two groups of workers traditionally thought of as belonging to the middle class – public sector employees and professionals and technicians – while the last section summarizes the main findings of our research.

2. 

 HE GREEK MIDDLE CLASSES THROUGH T GOOD AND BAD TIMES: FROM EXPANSION TO DECLINE

The definition of the middle class(es) constitutes a major challenge in the economics and sociology literature. The former identifies class through the position of individuals in the income and wealth distributions, the latter

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 246

28/09/2016 08:24



The middle classes in the Greek Great Depression ­247

through their position in the social relations of production and division of labour, on which depends control over resources, power and social status (Atkinson and Brandolini 2011; Goldthorpe 2012). In this chapter – as in the project as a whole – we define the middle classes as society’s middle-­ income categories (economic approach) and try to link changes in their relative size and income with developments in the world of work (sociological approach). In this section we describe the trends in the relative size and income of the middle class in Greece from 1993 to date, in the context of prevailing income inequalities. To do so, we have used official data on self-­reported income from ECHP and EU-­SILC surveys. Given widespread tax evasion in Greece, estimated at 27.5 per cent of GDP in the period 1999–2007 (Schneider 2012), which has been shown to result in a more unequal income distribution (Matsaganis et al. 2012), the estimates of indices and trends in inequality and class structure in the present study should be treated with caution. This holds for all empirical research on income inequalities and the distributional impact of public policies in Greece. 2.1  Income Inequalities In 2012, Greece was the fourth most unequal country in the EU28, surpassed only by Bulgaria, Latvia and Lithuania and closely followed by Portugal, Romania and Spain. Persistently high inequality is not only the outcome of the insufficiently redistributive welfare state (see above) and social transfers in particular, but it is also accounted for by the tax system and widespread tax avoidance and evasion practised by most self-­ employed people and firms but benefiting the upper and upper-­middle classes in particular. Using data from Household Budget Surveys, Tsakloglou and Mitrakos temporal trends of inequality from 1974 (2012) recently studied inter-­ to 2008 and found that all inequality indices report a non-­monotonic decline in inequality that is far more pronounced in the case of indices that are relatively more sensitive to changes in the tails of the distribution (Atkinson 0.5 and 2.0). For the period 1994–2009, these trends are largely confirmed using the data of the European Household Panel (ECHP) and EU Statistics on Income and Living Conditions (EU-­SILC) (Figure 6.1). From 2009 to 2013 – a period that coincides with the first four years of the austerity plan implemented in Greece – EU-­SILC data reveal an increase in inequality measured by the Gini coefficient. We may thus conclude that the period of economic growth (1993–2008) and the first two years of the economic crisis (2008–09) were accompanied by decreasing – albeit high – income inequality, while the first four years

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 247

28/09/2016 08:24

248

Europe’s disappearing middle class?

36 35 34 33 32 31 30 29

Greece European Union (27 countries)

28

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

27

Source:  Eurostat data online.

Figure 6.1 Long-term trend of income inequality (Gini coefficient of equivalized disposable income), Greece, 1995–2014 of austerity reversed this long-­term trend and the value of Gini coefficient has increased from to 0.329 to 0.345. 2.2 Long-­term Trends in the Relative Size and Income of the Middle Classes To study the long-­term trends in the relative size and income of the middle classes in Greece we have used middle-­income categories as a proxy for the concept of middle classes and defined them – according to the common definition of our project – as the share of the population having an equivalized disposable income between 60 per cent and 200 per cent of the median equivalized disposable income. The share of the population with an equivalized disposable income up to 60 per cent of the median belongs to the lower-­ income class while the one with an equivalized disposable income higher than 200 per cent of the median belongs to the upper class. We then constructed three middle-­income categories with reference to the median equivalized disposable income: the lower middle-­income class (60–80 per cent of the median), the middle middle-­income class (80–120 per cent of the median) and the upper middle-­income class (120–200 per cent of the median). To describe the long-­term trends in the relative size and income of the middle classes in Greece we have applied the project definitions of the middle-­income categories on (ECHP) data for 1994 (reference year 1993) and EU-­SILC data for 2009 and 2013 (reference years 2008 and 2012 respectively).

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 248

28/09/2016 08:24



The middle classes in the Greek Great Depression ­249 100

11.3

10.9

9.1

8.3

65.1

67.7

70.8

68.6

10 23.4

21.5

20.1

23.1

0 1994

2003

2009

90 80 POP share

70 60 50 40 30 20

Lower

Middle

2013

Upper

Source:  Authors’ elaboration of ECHP 1994 and EU-SILC 2003, 2009 and 2013 microdata.

Figure 6.2 Population shares by income class, Greece, 1994–2013 (percentage) Figure 6.2 shows that the middle classes expanded significantly during 1993–2008, which was a period of sustained economic growth and decreasing income inequalities. Their share rose from 65.1 per cent to 70.8 per cent of total population while at the same time the share of both the lower and the upper classes shrank. Consequently, the expansion of the middle classes is accounted for by both the upward mobility of the lower class and the downward mobility from the upper class. Conversely, the crisis years (2008–12) saw downward mobility from both the upper and the middle classes. The picture changes completely, however, if we calculate the population shares of the different classes in 2012 choosing either the income thresholds of 2012 (Figure 6.2) or those of 2008 (Figure 6.3). In the first case, we take into account that the median income decreased by 27.2 per cent between 2008 and 2012 and reconstruct the lower, middle and upper classes in relative terms around the lower median. According to this approach, the middle class shrank by 2.2 percentage points between 2008 and 2012 (Figure 6.2). In the second case, we keep the median income of 2008 as the reference point and measure what part of the population in 2012 had a disposable equivalized income that would be considered middle-­ class income according to the middle-­ class income thresholds

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 249

28/09/2016 08:24

250 100 90 80 70

Europe’s disappearing middle class? 10.9

9.1

26.3

27.5

26.2

29.5

13.1

15.2

13.8

23.5

21.5

20.1

2002

2008

11.5 26.0

3.0 13.1 26.1

60 50 26.0 40 30 20 10

0 1993 Lower

Lower middle

Core middle

Upper middle

17.9

39.9

2012 Upper

Note:  * Figures for 2012 are calculated on the basis of 2008 income thresholds. Source:  Authors’ elaboration of ECHP 1994 and EU-SILC 2003, 2009 and 2013 microdata.

Figure 6.3 Population shares by income class* – decomposition of the middle class in three income classes, Greece, 1993–2012 (percentage) of 2008. This approach captures the magnitude of impoverishment of the middle classes since the beginning of the crisis, by measuring the downward mobility of the middle classes towards the lower class (which is defined as the population with equivalized disposable income below the poverty line equal to 60 per cent of the median income). Based on the second approach, Figure 6.3 shows that the middle classes were decimated during the first four years of the crisis. While at the beginning of the crisis 70.8 per cent of the population enjoyed middle-­class incomes, four years later only 57.1 per cent of total population enjoyed incomes that would be considered middle class according to the income thresholds prevailing at the beginning of the crisis. Although not necessarily impoverished, it seems that the upper middle-­income class has shrunk the most among the different middle-­income categories. At the same time, in 2012, only 3 per cent of total population enjoyed an income corresponding to the upper-­class incomes of 2008 against 9.1 per cent in 2008. Given that the share of the lower class rose from 20.1 per cent to 39.9 per cent across the same period, it follows that about 20 per cent of the population that belonged to the middle or upper class at the beginning of the crisis

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 250

28/09/2016 08:24



The middle classes in the Greek Great Depression ­251

300 250

1993–08

243.7 203.8

200

196.4

193.4

195

2008–12 198.4

150 100 50 0 –50

–32.9 Lower

–28.5 Lower middle

–25.2 Core middle

–27.4 Upper middle

–27.6 Middle

–28.1 Upper

Source:  Authors’ elaboration of ECHP 1994 and EU-SILC 2009 and 2013 microdata.

Figure 6.4 Average income by class – changes over the pre-crisis and crisis periods, Greece, 1993–2012 (percentage) had fallen below the poverty line of 2008 four years later. The great bulk of the impoverished population has certainly stemmed from the middle classes. Although they constituted a period of expansion in the relative size of the middle class, the growth years saw a smaller nominal increase in the average middle-­class income (195 per cent) than that of the lower class (244 per cent) and similar to that of the upper class (198 per cent) (Figure 6.4). Interestingly, in the first four years of the crisis the average middle-­class income fell less than that of the lower class (28 per cent against 33 per cent) and as much as that of the upper class (28 per cent). These developments point to stability in the income position of the middle class relative to the upper class over the whole period under examination as well as to a deterioration of its position relative to the lower class during the period of growth and an improvement during the crisis. They are consistent with the decrease of income inequalities during the growth period and their increase during the crisis period. However, it should be remembered that median income collapsed during the crisis and membership of the middle class changed immensely: in 2012, a large part of the 2008 middle class had joined the lower class owing to impoverishment, while a large part of the 2008 upper class had joined the middle class because of major income losses.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 251

28/09/2016 08:24

252

Europe’s disappearing middle class?

2.3  The Redistributive Effect of the Tax and Benefit Systems The role of the welfare state as social shock absorber has been reinforced during the crisis, but increased protection has benefited only the middle and upper classes (Figure 6.5). In 2012, 37.6 per cent of middle-­ class and 28  per cent of upper-­class disposable income was due to net social ­transfers  – social benefits minus taxes – against 27.5 per cent and 16.2  per  cent, respectively, in 2008. Conversely, the share of lower-­class income stemming from net social transfers fell from 47.7 per cent to 46.6 per cent across the same period. Comparing the redistributive effect of the tax and benefit system by including and excluding pensions from benefits, we find that the increased protection enjoyed by the middle and upper classes during the crisis was due exclusively to pensions. Given that high and medium pensions were reduced relatively more under the austerity plan, the phenomenon can be attributed only to higher exit rates to retirement among the medium and higher classes. When pensions are excluded from benefits, the redistributive effect of the tax and benefit system on the disposable income of all classes reduces considerably. Between 2008 and 2012, this effect remained stable for the lower class, diminished for the middle class and turned from positive into negative for the upper class (Figure 6.5).

50.0

2002

40.0

2008

30.0

2012

20.0 10.0 0.0 –10.0

Lower (D-M)

Lower (D-M-P)

Middle (D-M)

Middle (D-M-P)

Upper (D-M)

Upper (D-M-P)

Note:  * All incomes are equivalized. Source:  Authors’ elaboration of ECHP 1994 and EU-SILC 2009 and 2013 microdata.

Figure 6.5 Gap between average disposable income and average market income (D-M) and plus pensions (D-M-P) by income class* (percentage points), Greece, 2002–12

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 252

28/09/2016 08:24



The middle classes in the Greek Great Depression ­253

3. MAJOR TRANSFORMATIONS IN THE WORLD OF WORK AND THEIR IMPACT ON THE MIDDLE CLASSES In this section we first describe the main trends in the world of work since 1993 and then study their impact on the evolution of the middle class during both the growth period (up to 2008) and the crisis (after 2008). To analyse their impact on the middle classes, we divide long-­term trends in the world of work into two groups: the first includes trends in the employment system that affect the labour market profile of the middle classes relative to those of the lower and upper classes; the second comprises changes in industrial relations and conditions that affect wage inequalities and the relative labour income of employees and the self-­employed. The recent global financial crisis represents a moment of rupture; it has led to a complete reversal or an acceleration of the trends observed during the growth period. 3.1 Changes in the Employment System and Impact on the Profile of the Middle Classes Table 6.1 portrays the major trends in the employment system from 1993 to 2008: ●●

●●

●●

●● ●●

The employment rate of 25–64 year olds increased considerably, mainly because of the spectacular rise in the female employment rate. The youth employment rate decreased, time spent in education and enrolments in higher education increased and the share of the highly educated in all employed rose significantly. The self-­employment rate declined because of the great decrease in the numbers of self-­employed in agriculture, not compensated by growth outside agriculture, and the huge increase in the numbers of wage earners in construction and services. The incidence of part time work and temporary employment showed an insignificant increase. The share of public sector workers in all employees fell due to a more substantial rise in the number of private sector employees while the share of professionals and technicians in all employed sharply increased. Both occupational categories are highly represented among the middle classes (see case studies, sections 4 and 5).

Last but not least, the 1990s and 2000s saw a massive inflow of non-­EU migrants who took over manual and low-­skilled jobs. According

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 253

28/09/2016 08:24

254

Europe’s disappearing middle class?

Table 6.1 Long-­term trends in the world of work – employment indicators, Greece, 1993–2014

Employment rate (%) 25–64 years Male employment rate (%) 25–64 years Female employment rate (%) 25–64 years Youth employment rate (%) 15–24 years Highly educated (% of all employed) Self-­employment rate (% all employed) Wage employment rate (% of all employed) Part time workers (% of all employed) Temporary workers (% of all employees) Professionals and technicians (% of all employed) Public sector employees (% of all employees)

1993

2008

2014

60.5 81.6 40.6 27.5 15.9 33.4 54.6 4.0 10.4 15.7 39.9

69.1 83.8 54.7 23.5 26.4 28.4 65.9 5.4 11.6 23.5 34.5

56.0 65.7 46.6 13.3 33.7 30.7 64.7 9.3 11.6 27.0 36.1

Note:  Figures refer to the second quarter of the year. Source:  European Labour Force Survey (ELFS), Eurostat data online; National Labour Force Survey data for public sector employees.

to population census data, foreign workers represented 2 per cent of all employed in 1991, 9.5 per cent in 2001 and 10.5 per cent in 2011. Unfortunately, Labour Force Survey (LFS) and ECHP/EU-­SILC data are not reliable enough – especially those before 2002 – to establish detailed trends and cover the whole 1993–2008 period. The deep and prolonged economic crisis that started in 2008 produced a tremendous shock for the Greek economy and society (Figure 6.6). With the state-­ led fall of economic activity as a result of the austerity-­ ­ recession spiral that started in 2010, employment collapsed and  unemployment skyrocketed. Between 2008 and 2014, the employment rate of both young people and those 25–64 years of age plummeted, the male rate of 25–64-­year-­olds incurring the greatest decline: 18 percentage points. Having suffered proportionally the same net job losses as permanent employees, the incidence of temporary work remained stable through the crisis, while the part-­time employment rate almost doubled, still from a very low starting point, working time flexibility represented a practical way for firms to adjust to the vicissitudes of the crisis. To assess the impact of long-­term trends in the employment system on the middle classes we have to take as the starting point their profile at the beginning of the period.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 254

28/09/2016 08:24

The middle classes in the Greek Great Depression ­255 Employment (in thousands)

4600

30

4400

25

4200

Unemployment rate (%)

20

4000

15

3800

10

3600

5

3200

0

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

3400

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014



Figure 6.6  The shock of the crisis, Greece, 2001–14 Income-­class disparities are directly associated with different shares of labour income households by income class. In 1993, 50 per cent of lower-­ class, 71 per cent of middle-­class and 82 per cent of upper-­class households were labour income households (Table 6.2). In this year, a second important class differentiation was related to work intensity in labour income households: the share of dual/multiple earner in all households was 34 per cent among the lower, 37 per cent among the middle and 54 per cent among the upper class. The proximity between lower-­and middle-­ class shares is due to a very small difference in female employment rates between these same classes to the advantage of the middle class. At the same time, the huge difference in the female employment rate between the middle and upper classes (11.5 percentage points) was the main reason for the considerably higher share of dual-­earner households among the latter. Interestingly, in 1993, the youth employment rate was highest in middle-­ class households. Table 6.3 illustrates the profile of the different classes according to the employment status and characteristics of their members aged 15 years or over. In 1993, middle-­class households had a higher employment rate than lower-­class households and a much lower employment rate than upper class households; the same unemployment rate as in lower-­class households and a much higher unemployment rate than in upper-­class households; a much lower propensity to have retired members than the lower class and similar to that of the upper class; a propensity to be inactive (other than retired) similar to that of the lower class and much higher than that of the upper class. In the same year, the self-­employment and part-­time rates among the employed members of the middle class were

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 255

28/09/2016 08:24

256

Europe’s disappearing middle class?

Table 6.2 Labour income and dual/multiple earner households by income class, Greece, 1993–2012 (percentage) Lower class

Middle class Upper class

Labour income households (% of all) Dual-­or multiple-­earner households   (% of labour income households) Female employment rate (25–64) Youth employment rate (15–24)

50.0 40.1

1993 70.8 42.3

82.0 53.8

34.2 24.4

37.4 28.4

52.6 25.4

Labour income households (% of all) Dual-­or multiple-­earner households   (% of labour income households) Female employment rate (25–64) Youth employment rate (15–24)

55.0 31.3

2008 68.0 54.0

86 64.9

33.1 12.7

58.6 20.9

75.2 16.8

Labour income households (% of all) Dual-­or multiple-­earner households   (% of labour income households) Female employment rate (25–64) Youth employment rate (15–24)

41.0 22.4

2012 57.0 38.8

75.0 54.7

25.0 6.4

47.2 10.3

64.1 11.1

Source:  Authors’ elaboration of ECHP 1994 and EU-­SILC 2009 and 2013 data.

lower than among their lower-­class and higher than among their upper-­ class counterparts, while the share of high-­educated labour earners was 22  per  cent among the middle class against 5 per cent among the lower class and 56 per cent among the upper class. class households were overwhelmingly populated by In 1993, lower-­ retired persons with low pensions, self-­employed people with low income and inactive women and young people. Compared with lower-­class households, middle-­ class households had fewer retired and self-­ employed members and more employed members with higher educational attainment and enjoyed better employment conditions than their lower-­class counterparts. At the same time, they had far fewer women in employment, dual-­earner couples and highly educated wage earners than upper-­class households, while their employed members enjoyed worse employment conditions than their upper-­class counterparts. Three important changes took place in the composition of the middle classes during 1993–2008. First, the share of labour income in all

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 256

28/09/2016 08:24



The middle classes in the Greek Great Depression ­257

Table 6.3 Employment status and characteristics of the population (15 and over), by income class, Greece, 1993–2012 (percentage) Lower class Middle class Upper class Employed Unemployed Retired Other inactive Part-­timers (% of all employed) Self-­employed (% of all employed) Temporary (% of all employees) High educated (% of labour income earners)

27.8 7.9 27.8 36.5 21.3 77.6 – 5.2

1993 38.1 7.5 16.1 38.4 13.6 40.2 – 21.9

56.2 3.5 15.1 25.2 18.8 37.4 – 55.9

Employed Unemployed Retired Other inactive Part-­timers (% of all employed) Self-­employed (% of all employed) Temporary (% of all employees) High educated (% of labour income earners)

35.6 9.1 21.8 33.5 16.7 52.3 46.0 14.2

2008 49.1 4.8 23.1 23.0 7.9 26.1 23.3 36.4

65.3 2.2 18.2 14.3 3.8 30.7 8.8 73.7

Employed Unemployed Retired Other inactive Part-­timers (% of all employed) Self-­employed (% of all employed) Temporary (% of all employees) High educated (% of labour income earners)

23.5 32.0 15.1 29.4 25.8 46.3 27.0 18.0

2012 38.2 11.8 29.3 20.7 10.5 31.3 15.4 43.1

56.4 3.7 29.4 10.4 4.0 30.7 9.0 81.9

Source:  Authors’ elaboration of ECHP 1994 and EU-­SILC 2009 and 2013 microdata.

middle-­income households dropped from 71 per cent in 1993 to 68 per cent in 2008 (Table 6.2) owing to the increased incidence of retired people in middle-­class households. The share of retired people in the population aged 15 years and over living in middle-­class households saw an upsurge of 7 percentage points, against a fall of 6 percentage points in lower-­class households and an increase of 3 percentage points in upper-­class households (Table 6.3). Second, a spectacular rise in the share of dual/multiple-­ earner households from 42 per cent to 54 per cent took place, which went in parallel with the remarkable increase in the female employment rate

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 257

28/09/2016 08:24

258

Europe’s disappearing middle class?

from 37 per cent to 59 per cent. The causal link between the rises in the female employment rate and the share of dual/multiple-­earner households is more than obvious, given the fall in the youth employment rate from 28 per cent to 21 per cent between 1993 and 2008 (Table 6.2). Public policy supporting low-­ income pensioners through a non-­ contributory means-­ tested benefit (Allowance of Social Solidarity for Pensioners – EKAS) introduced in 1996 and the increasing participation of female spouses/partners in paid work, thus contributing a second income to couple households, are the main developments that determined changes in the composition of the middle classes in favour of retired and dual-­earner households over the fifteen-­year growth period. The mobilization of women as second earners may also partly explain the high incidence of temporary employees in these households in 2008, but the fall in the part-­time rate between 1993 and 2008 indicates that women who joined the labour market have been mainly absorbed into full-­time jobs (Table 6.3). Moreover, the employed members of the lower class displayed a higher propensity to work part-­time or on a temporary contract than those of the middle class in 2008. Contrary to what occurred to the middle class, the share of labour income households in all households increased among both the lower and upper classes between 1993 and 2008 (Table 6.2). This trend has been coupled, in the case of the lower class, with a fall, and in the case of the upper class, with a rise in the share of dual/multiple-­earner households in all households of each class. Focusing on the lower classes, the contradictory trends of an increasing share of labour income and a decreasing share of dual/multiple-­earner households can be explained by two reasons in particular. The first is changes in the composition of lower-­class households brought about by the heavy immigration flows of the 1990s and 2000s. According to ECHP and EU-­SILC data, 12 per cent of the population living in such households was foreign-­born in 2008 as against 4.5 per cent in 1993; the respective shares for the middle classes were 6.6 per cent and 3.7 per cent. Migrant households usually receive labour income and have no retired members. The second reason is the increasing concentration of the unemployed in lower-­ income class households, especially among young people and women: 33 per cent of the unemployed were living in lower-­income class households in 2008 as against 26 per cent in 1993. Besides, in these households, the female employment rate fell from 34.2 per cent to 33.1 per cent while the youth employment rate fell from 24.4 per cent to 12.7 per cent between 1993 and 2008 (Table 6.2). The third change in the composition of the middle class during ­1993–2008 is the large fall (14 percentage points) in the self-­employment

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 258

28/09/2016 08:24



The middle classes in the Greek Great Depression ­259 50.0 0.0 –50.0

–100.0

1993–2008 2008–14

–150.0 –200.0

Employers

Self-employed without employees

Unpaid family workers

Employees

Source:  Labour Force Survey 1993, 2008 quarter 2, 2014 quarter 2.

Figure 6.7 Variation of employment in agriculture by occupational status, Greece, 1993–2014 (000s) rate among its employed members in favour of the wage employment rate. The lower and upper classes also saw falls in the self-­employment rate of 25 and 9 percentage points, respectively (Table 6.3). This is the outcome of a huge contraction of self-­employment in agriculture, which has mainly affected the lower class, on one hand, and the quasi-­stagnation of self-­ employment and the sharp rise in wage employment outside agriculture, which have mainly affected the middle and upper classes, on the other (Figures 6.7 and 6.8). Another important trend in the world of work that must have affected the expansion and average income of the middle classes during the period of economic growth (1993–2008) is the significant rise in the share of the highly educated in all employed. As a result, the share of highly educated people among wage earners picked up among all income classes, with the highest increase among the upper class (18 percentage points), the second highest among the middle class (14.5 percentage points) and the smallest among the lower class (9 percentage points). Class differentiation in the size of increase has thus contributed to the class divergence of average income. By reversing or accelerating prior long-­term trends in employment, the 2008–14 crisis has had a major impact on the profile of the different income classes. First, the share of labour income households in all households has significantly retreated among all income classes but more so among the lower class (−14 percentage points as against −11 percentage

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 259

28/09/2016 08:24

260 1200.0 1000.0 800.0 600.0 400.0 200.0 0.0 –200.0 –400.0 –600.0 –800.0 –1000.0

Europe’s disappearing middle class? 1993–2008 2008–14

Employers

Self-employed without employees

Unpaid family workers

Employees

Source:  Labour Force Survey 1993, 2008 quarter 2, 2014 quarter 2.

Figure 6.8 Variation of non-agricultural employment by occupational status, Greece, 1993–2014 (000s) points among the middle and upper classes). This class-­ differentiated employment effect of the crisis has increased the concentration of unemployment in lower-­income households (45 per cent of all unemployed in 2012, up from 33 per cent in 2008). Besides, having better resisted the collapse of employment, the highly educated increased their share of labour income earners during the crisis, more so among the upper and middle classes (8.2 and 5.3 percentage points respectively) than among the lower class (3.8  ­percentage points). Second, the share of dual/multiple-­earner households in all labour income households has declined among all income classes, but has collapsed among the middle class (−18.9 ­percentage points as against −9 and −10 percentage points among the lower and upper classes respectively). The class-­differentiated employment effects of the crisis can be thus summarized as follows. The collapse of employment during the crisis has raised the share of households with no one employed mainly among the lower class and the share of single income-­earner households mainly among the middle class. Finally, the crisis has increased the concentration of the retired in middle-­income households (78 per cent of all retired in 2012, up from 73.5 per cent in 2008) at the expense of lower-­ income households. This may be accounted for by the relatively lower reductions in pensions than in the income of those employed becoming unemployed.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 260

28/09/2016 08:24



The middle classes in the Greek Great Depression ­261

3.2 Changes in the Wage-­setting System, Social Dialogue and the Employment Structure: Impact on Labour Income by Income Class From 1974 through the 1980s industrial relations in Greece were extremely adversarial and highly politicized. The 1990s and 2000s saw a decrease in strike activity and union density, a move towards more consensual industrial relations and a decline in state intervention in wage setting, although wage regulation remained strong (Karamessini 2015). In 1990 the system of automatic indexation of wages to inflation, in place since 1982, was abolished while in the same year a new law on ‘free’ collective bargaining replaced compulsory arbitration with independent mediation and arbitration. Despite these changes, the Ministry of Labour kept its prerogative to extend collective agreements to non-­unionized employees and employers, which accounted for the high coverage of employees by collective agreements until 2011 when extension was suspended under the austerity and reform programme imposed by the Troika. setting system of the 1990s and 2000s combined The Greek wage-­ high coverage of employees by collective agreements with a system of articulated collective bargaining that mingled national bargaining on the national minimum wage and sectoral or occupational minima with company-­level bargaining improving over the latter. Alongside constant and robust productivity growth, the wage-­setting system allowed for an uninterrupted increase in real compensation per employee between 1993 and 2009, amounting to 52.5 per cent in cumulative terms (Figure 6.9). However, it is not average wage growth but wage inequality that is relevant for examining the impact of the wage-­setting system on middle-­class labour income. The aforementioned collective bargaining system of articulated bargaining led to substantial wage inequalities, especially between the public and private sectors, but at the same time reinforced the middle of the wage distribution through national bargaining by sector and occupational minima, assisted by the Mediation and Arbitration Agency, which intervened in cases of industrial conflict and when called upon by weak unions. In the late 1990s, the bargaining rounds between management and the strong unions in public utilities and banking were decoupled from those on the national minimum wage between the National Confederation of Greek Labour (GSEE) and peak employers’ organizations. A basic mechanism of articulated bargaining and wage drift was thus broken (Ioannou 2000).1 However, public utilities and banking remained in the 2000s, as had been the case in the 1990s, the strongholds of the union movement; they dominated the leadership of GSEE and reinforced its bargaining power with regard to determination of the national minimum wage through collective negotiations.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 261

28/09/2016 08:24

262

Europe’s disappearing middle class?

120 100 80 60 40 20

1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

0

Source:  Annual macro-economic (AMECO) database.

Figure 6.9 Real compensation per employee, total economy, Greece, 1960– 2014 (deflator of private consumption, 2010 = 100) At the onset of the current crisis, overall wage inequality in Greece was medium compared with other Organisation for Economic Co-­operation and Development (OECD) member states. In 2007, Greece ranked twelfth among 25 OECD countries according to top/bottom inequality as measured by the decile 9/decile 1 ratio.2 However, the national minimum wage was above the poverty line and its gap with the average wage was small by international comparison. This means that overall wage inequality before the crisis was accounted for mainly by top-­centre inequality. Indeed, in 2007, Greece ranked eleventh among 25 OECD countries as regards the decile 9/decile 5 ratio and fifteenth among the same countries as regards the decile 5/decile 1 ratio.3 Since mid-­2010 the wage-­setting system has seen unprecedented changes, meant to bring about substantial wage cuts in both public companies and agencies in which wages were set at firm level through collective bargaining and by employers. Between May 2010 and February 2012, the first Economic Adjustment Programme (EAP) gave priority to dismantling of collective bargaining in state-­owned enterprises and agencies with the legislated reduction of nominal wages and, in the private sector, to the decentralization of collective bargaining to firm level where the employers’ position is stronger than that of the unions. In late 2011 and early 2012 the government launched a tripartite dialogue in order to discuss with the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 262

28/09/2016 08:24



The middle classes in the Greek Great Depression ­263

0.7 0.6 0.5 0.4 0.3 0.2

Minimum/mean wage

0.1

Minimum/median wage 2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

1994

1993

0

Source:  OECD.stat (data extracted on 25 August 2015).

Figure 6.10  Minimum to mean/median wage ratio, Greece, 1993–2013 social partners national minimum wage developments compatible with boosting competitiveness and preserving employment. The results of this dialogue were considered by the Troika to be unsatisfactory (European Commission 2012: 38) and led in February 2012 to a statutory reduction of the national minimum wage for the first time and the dismantling of the collective bargaining system through a series of measures included in the second EAP. The decision was taken in order to allow a substantial reduction in nominal wages in a context of mass unemployment, in which individual employees have very weak bargaining power, if any. In 2009, the national minimum wage was 45.2 per cent of the average wage, the third highest rate among the EU countries that had a national minimum wage (Institute of Labour-­GSEE/ADEDY 2011), but Greece’s rank according to its absolute level was average and much lower than in most EU15 countries. The trends in the ratios of minimum to mean/ median wage (Figure 6.10) point to an increase in wage inequality at the centre-­to-­bottom part of the wage distribution between 1993 and 2004, a decrease between 2004 and 2011 and a steep increase in 2012. The latter is due to the reduction of the minimum wage by 22 per cent and the establishment of special minimum wage for young people below 25 years of age at 68 per cent of the previous unitary minimum wage. Both measures, introduced under the reform plan imposed by the Troika, were explicitly meant to reverse the trend of narrowing wage inequalities at the bottom end of the wage distribution, which is thought to price the low-­skilled, whose pay is at or above the minimum wage, out of employment. In the medium term, however, the downward adjustment of wage floors alongside the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 263

28/09/2016 08:24

264

Europe’s disappearing middle class?

Table 6.4  Low-­and high-­pay incidence, Greece, 2004–12 (percentage)

Low-­pay incidence* High-­pay incidence**

2004

2007

2008

2010

2011

2012

20.1 22.1

17.6 22.3

13.5 21.6

13.3 20.8

12.3 18.1

11.8 16.2

Notes: All measures calculated from EU-­SILC data on gross earnings of full-­time dependent employees. * Share of full-­time workers earning less than two-­thirds of gross median earnings of all full-­time workers. ** Share of full-­time workers earning more than one-­and-­half time gross median earnings of all full-­time workers. Source:  OECD.Stat (data extracted on 12 August 2015).

deregulation of the wage-­setting system by dismantling collective bargaining led to the individualization of wage bargaining and, ultimately, under conditions of mass unemployment, to a compression of wage differentials towards the minima. Other measures of wage inequality do not cover the whole period under scrutiny but only the latest period. According to the OECD database, low-­ and high-­pay incidence both fell between 2004 and 2012, pointing to a continuous decrease in wage inequality not only at the bottom but also at the top end of the wage distribution (Table 6.4). Both trends must have exerted a favourable impact on the expansion of the middle classes observed in the 2000s before the eruption of the crisis in 2008. In order to jointly examine the impact of changes in the wage distribution and the employment structure on the expansion of the middle classes over the whole period 1993–2014, and inspired by relevant methodologies developed in recent international literature (Eurofound 2008, 2013; Fernández-­ Macías et al. 2012), we have used unpublished Labour Force Survey data to calculate the variation of wage employment in low, medium and highly paid occupations. These were defined on the basis of the wage distributions of 2000 and 2011, the first used to study employment trends in low-­, medium-­ and high-­paid occupations between 1993 and 2010 and the second between 2011 and 2014. The results of this exercise are presented in Figure 6.11. The figure shows that the thrust of employment over the growth period (1993–2008) was based primarily on medium paid occupations that were responsible for 71 per cent of net job creation between 1993 and 2000 and 61 per cent between 2000 and 2008. However, medium paid occupations were also those that incurred the biggest job losses during the economic crisis. The above developments greatly influenced the expansion of the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 264

28/09/2016 08:24

Wage employment



The middle classes in the Greek Great Depression ­265

200 000

–200 000

Highly paid

Medium paid

1993–2000

12 242

231 200

Low paid 83 003

2000–2008

58 649

436 753

216 503

2008–10

–5661

–143 285

–9430

2011–14

–10 873

–191 338

–148 454

Source:  Authors’ elaboration of Labour Force Survey data.

Figure 6.11 Variation of wage employment in low, medium and highly paid occupations, Greece, 1993–2014 (number of persons) middle classes during the growth period and its contraction during the crisis. To assess the impact on the relative income of the middle classes of the above-­described trends in average wages, wage inequality and the structure of wage employment by high-­, medium-­and low-­paid occupations we have estimated average wages and labour income from self-­employment by income class using ECHP and EU-­SILC microdata. The first observation, based on Table 6.6, is that, in 1993, the average employment was higher than average annual labour income from self-­ annual wages in all income classes. Fifteen years later this held only for the upper-­income class while the opposite held for lower-­and middle-­income classes: average wages were higher than the average labour income from self-­employment. These developments are explained by relative changes in these two kinds of labour income (Table 6.5). Changes in annual wages were higher than those in annual labour income from self-­employment in all income classes between 1993 and 2008, but the increase in the income of self-­employed people belonging to the upper class was sufficiently high to sustain the large gap between labour income from self-­employment and wages in this class. The second observation based on Table 6.6 is that, between 1993 and

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 265

28/09/2016 08:24

266

Europe’s disappearing middle class?

Table 6.5 Annual labour income in lower, middle and upper class households, Greece, 1993–2012 (euros) Lower class Middle class Upper class Average wages Average labour income from self-­employment

2181.70 2664.60

1993 5770.00 8153.93

10 322.50 21 019.90

Average wages Average labour income from self-­employment

7880.04 3942.71

2008 13 857.58 11 884.57

25 532.85 45 024.36

Average wages Average labour income from self-­employment

5807.66 4101.57

2012 11 767.58 10 046.12

20 996.29 39 365.01

Source:  Authors’ elaboration of ECHP 1994 and EU-­SILC 2009 and 2013 microdata.

Table 6.6 Average net annual labour income by income class, Greece, 1993–2012 (base to end year changes – percentage) 1993–2008

2008–12

Annual wages Lower class Middle lower Core middle Middle upper Middle class Upper class

261.2 132.0 125.3 152.4 140.2 147.4

−26.3 −8.3 −11.1 −19.0 −15.1 −17.8

Annual labour income from self-­employment Lower class Middle lower Core middle Middle upper Middle class Upper class

48.0 58.5 36.3 45.3 45.8 114.2

4.0 −19.6 −11.8 −15.7 −15.5 −12.6

Source:  Authors’ elaboration of ECHP 1994 and EU-­SILC 2009 and 2013 microdata.

2008, the average annual wages of wage earners in middle-­income households increased less, while between 2008 and 2012 they decreased less than those of their counterparts in lower-­and upper-­income households. As regards the average annual labour income of the self-­employed, this

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 266

28/09/2016 08:24



The middle classes in the Greek Great Depression ­267

increased less in middle-­class than in lower-­and upper-­class households between 1993 and 2008 and decreased more in middle-­than in upper-­class households between 2008 and 2012, while the labour income of the self-­ employed members of lower-­class households increased during the crisis period. We can thus conclude that relative wage developments were unfavourable for the middle classes during the growth period and favourable during the crisis period, while relative changes in labour income from self-­ employment proved unfavourable for the middle classes during both the growth and the crisis periods.

4.  CASE STUDIES In this section we focus on the cases of public sector employees, on the one hand, and professionals and technicians, on the other, in order to study the impact of long-­term trends in the world of work on the middle classes. We have selected these two cases because the occupational categories under scrutiny have been key to the creation and reproduction of the middle classes in Greece and elsewhere, historically or more recently. Public sector employment is currently under attack in Greece, because a ‘lean public sector’ is one of the main objectives of the fiscal consolidation and structural reform programmes imposed by the Troika since mid-­2010. As for professionals and technicians, these have proved more resilient to the crisis than other occupations and provide future prospects for the middle classes currently facing an uncertain future. 4.1 Case Study 1: Public Sector: Neutralizing a Traditional Producer of the Middle Classes Since the creation of the modern Greek state in 1830, the public sector as employer has been a major producer of white-­collar elites and the middle class (Tsoukalas 1980, 1987). It has invariably absorbed a very important share of the highly educated labour force and has also been a good employer for women, offering employment stability, promotion opportunities, good working conditions and relatively high pay, especially after the Second World War, when the principle of equal treatment of men and women in the public sector was established in the Greek Constitution of 1952 (Karamessini 2012). The public sector is important for the expansion and reproduction of the middle classes in Greece, since the basic split in employment and working conditions during the post-­war growth period (1960–74) was between the private and public sector (Karamessini and

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 267

28/09/2016 08:24

268

Europe’s disappearing middle class?

Table 6.7 Public/private sector employees and self-­employed as a percentage of labour income earners in lower-­, middle-­and upper-­class households, Greece, 1993–2012 (percentage) Lower Middle class class

Lower middle

Middle Upper middle Middle

Upper class

Public sector employees Private sector employees Self-­employed

4.4 18.0 77.6

24.1 35.7 40.2

10.7 27.9 61.4

1993 19.6 37.9 42.6

32.5 36.7 30.9

30.9 31.8 37.4

Public sector employees Private sector employees Self-­employed

7.2 40.5 52.3

25.1 48.8 26.1

11.0 50.0 39.0

2008 18.5 54.2 27.3

34.3 44.3 21.4

31.9 37.4 30.7

Public sector employees Private sector employees Self-­employed

7.1 46.6 46.3

24.9 43.8 31.3

10.7 42.9 46.4

2012 21.9 44.6 33.5

31.5 43.6 24.9

27.9 41.4 30.7

Source:  Authors’ elaboration based on ECHP 1994 and EU-­SILC 2009 and 2013 microdata.

Kaminioti 1999). Nationalizations, the expansion of the welfare state and direct job creation intended to absorb rising unemployment boosted public sector employment in the second half of the 1970s and the 1980s. In 1993, after the recession of the early 1990s, the public sector’s share reached 39.9 per cent of all employees – the peak of the upward trend – and 22.5 per cent of total employment. In that same year, 75 per cent of all public sector employees belonged to the middle class, as defined in this project, and mainly to the upper-­ middle and middle-­ middle classes.4 Although the majority of public sector employees belonged to the middle class, public sector employees represented a larger share of labour income earners of the upper class (30.8 per cent) than the middle class (24.1 per cent) as can be attested by Table 6.7. After 1993, the public sector’s share of all employees went into a 15-­year decline owing to a relatively greater net growth of wage employment in the private sector between 1993 and 2008. In 2008, 79 per cent of public sector employees belonged to the middle class,5 but only 25.1 per cent of all labour income earners of the middle class were public sector employees against 31.9 per cent of all labour income earners of the upper class (Table 6.7). The impact of the economic crisis on employment in the public sector

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 268

28/09/2016 08:24



The middle classes in the Greek Great Depression ­269 500 000

Public sector Private sector

400 000 300 000 200 000 100 000 0 –100 000 –200 000 –300 000 –400 000 –500 000 –600 000

1994–2000

2000–2008

Public sector

35 325

170 123

–217 114

2008–14

Private sector

259 313

421 445

–480 180

Source:  ELSTAT, Labour Force Survey, unpublished data.

Figure 6.12 Variation in public and private sector employment, Greece, 1994–2014 (absolute change in the number of employees) has differed in different phases of the crisis. In the first two years, the public sector not only protected jobs but also played an anticyclical role by employing people on temporary contracts. It was only after the sovereign debt crisis and the adoption in 2010 of the fiscal consolidation and economic adjustment programme, supervised by the Troika, that employment freezes and strict hiring rules leading to non-­full replacement of those who left were imposed on the public sector. In the same programme, targets were set for the reduction in the sector’s total number of employees and commitments were made by the Greek government for redundancies among certain categories of personnel. However, the decline of employment in the public sector was less important than in the private sector, which has shed labour in great numbers since the beginning of the crisis (Figure 6.12). Between 2008 and 2014 the number of employees contracted by 24.4 per cent (against 23 per cent of all employed). Because the decline was proportionally greater in the private sector, the share of the public sector in all employees increased slightly, from 35 per cent to 36 per cent. The strict hiring rules in the public sector since 2010 (currently one hire for ten exits) have created substantial labour shortages in many crucial state services. Besides, these have not only stalled the access of women – especially the highly educated – to good jobs (Karamessini 2014), but also the intergenerational reproduction of the middle classes and the upward social mobility of the lower classes through employment in the sector.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 269

28/09/2016 08:24

270

Europe’s disappearing middle class?

Table 6.8 Variation of net annual earnings of employees, Greece, 1993–2012 (base to end year changes – percentage) Private sector 1993–2008 2008–12

91.0 −10.5

Public sector 151.4 −21.1

Source:  Authors’ elaboration of ECHP 1994 and EU-­SILC 2009 and 2013 microdata.

There is a rich literature on public–private sector wage differentials in Greece. Kioulafas et al. (1991) found a 20–25 per cent pay gap in favour of the public sector between 1975 and 1985, and proved that the public sector rewarded education and experience more than the private sector; while Kanellopoulos (1997) reported a difference in average wages of 19 per cent for men and 42 per cent for women in 1988 but attributed the observed pay advantage of male public sector employees entirely to their higher qualifications and part of the pay advantage of female public sector employees to the same reason. Similarly, in a more recent article, Papapetrou (2006) established that 59.6 per cent of wage differentials between the two sectors in 1999 were accounted for by differences in education, experience and occupation. In the latest study available, Christopoulou and Monastiriotis (2014) estimated the net public sector premium (after controlling for observable characteristics) in 2005 at 11.3 per cent for monthly and 15.2 per cent for hourly wages and found that it amounts to about 35–40 per cent of the corresponding public–private wage differential observed in the raw data. Although important, this literature does not provide any trend in public–private sector wage differentials over the whole period examined in this chapter, on the basis of the same dataset. We have thus used ECHP and EU-­SILC microdata on net annual earnings to identify such a trend. Our results show that the net annual earnings of public-­sector employees increased during the period of economic growth and decreased during the first four years of the economic crisis much more than those of their private-­sector counterparts (Table 6.8). This means that the public–private sector wage differentials increased between 1993 and 2008 and shrank between 2008 and 2012. Bank of Greece data cover changes in gross earnings over a longer period (2009–2014) and provide a different picture for the austerity period of the crisis, which saw a roughly equivalent reduction in the earnings of private and public sector employees (Table 6.9). In conclusion, during the growth period, the public sector expanded

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 270

28/09/2016 08:24



The middle classes in the Greek Great Depression ­271

Table 6.9 Average gross nominal earnings, Greece, 2009–14 (base to end year changes – percentage) 2009–14 Public administration Public utilities and companies Private sector: banking Private sector: non-­banking National minimum wage Average wage

−26.0 −35.4 −30.0 −27.4 −24.8 −26.3

Source:  Bank of Greece (2015: 94, table IV.8).

less than the private sector but public sector employees saw their wages increase more than their private sector counterparts. During the crisis period, employment in the public sector contracted less than private sector employment while the wage reductions in the public sector were roughly equivalent to these in the private sector. This means that, although the role of the public sector as traditional producer of the middle classes was neutralized, the sector has continued to offer better employment and working conditions and middle-­class status to its employees. Last but not least, the contraction of private sector employment has mainly fuelled unemployment while the reduction in permanent public sector employment corresponds to exits to retirement, which have given access to decent or even high pensions. They may also have contributed to the further concentration of the retired among middle-­class households during the crisis. 4.2 Case Study 2: The Emergence of Professionals and Technicians as a Middle-­class Group Professionals and technicians are a heterogeneous occupational group. Professionals graduate from universities and are employed equally in the private and public sectors, while technicians are trained in higher technological and initial vocational training institutes and are employed mainly in the private sector. However, before the 2008–14 economic crisis, professionals and technicians had converging prospects as regards job opportunities and class positioning on the basis of income levels. Professionals and technicians is the occupational group with the largest increase in employment during the growth period (Figure 6.13), which fuelled the expansion of the middle class and changed its occupational composition. In 1993, 56.8 per cent of professionals and 71.7 per cent

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 271

28/09/2016 08:24

272

Europe’s disappearing middle class?

600.0

1993–2008

400.0 200.0 0.0

110.2

–200.0

–345.0

2008–14

486.6

421.4

–116.2

–108.1

164.3 –484.9

–400.0 –600.0

Legislators, senior officials, managers

Professionals and technicians

Clerks and service workers

Skilled and unskilled manual workers

–281.3 –46.7 Skilled agricultural and fishery workers

Source:  Labour Force Survey.

Figure 6.13 Variation of employment by large occupational groups, Greece, 1993–2014 (000s) of technicians had middle-­class incomes; the respective rates had risen to 64.7 per cent and 75.4 per cent in 2008. In parallel, the share of professionals belonging to the upper class decreased from 40.2 per cent to 31.6 per cent and that of technicians from 24.1 per cent to 21.1 per cent. In 2012 the majority of professionals and technicians belonged to the middle class and were mainly concentrated in the middle-­ middle and upper-­ middle categories. Not only do the majority of professionals and technicians belong to the middle classes but their share of all labour income earners in middle-­ class households constantly increased during the growth period, passing from 14.7 per cent in 1993 to 24.2 per cent in 2008 (Table 6.10). Moreover, the labour income gained by professionals increased by 122.3 per cent on average, while that earned by technicians rose by 150 per cent; managers and the remaining occupational groups obtained much smaller increases (Table 6.11). Although the majority of professionals and technicians belong to the middle class, it is mostly among the upper-­middle and the upper classes that these occupational groups are overrepresented. In 2008, they represented 45 per cent of all labour income earners in upper middle-­class households and 56 per cent in upper-­class households (Table 6.10). This means that upper middle-­class income is directly and largely affected by the job opportunities and income prospects of professionals and technicians. The number of employed professionals and technicians diminished

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 272

28/09/2016 08:24



The middle classes in the Greek Great Depression ­273

Table 6.10 Professionals and technicians as a percentage of labour income earners in lower-­, middle-­and upper-­income class households, Greece, 1993–2012 (percentage) Lower class

Middle class

Lower middle

Upper Middle

Upper class

4.5 3.0 7.5

16.9 7.6 24.5

30.9 7.6 38.5

9.6 6.3 15.9

23.1 11.9 45.0

42.6 13.4 56.0

15.2 6.5 21.7

23.5 12.4 35.9

48.6 15.5 64.1

Core middle 1993

Professionals Technicians Total

1.8 1.1 2.9

9.7 5.0 14.7

1.2 2.4 3.6

Professionals Technicians Total

4.9 2.3 7.2

15.6 8.6 24.2

5.5 3.4 8.9

2008

2012 Professionals Technicians Total

6.5 2.4 8.9

18.0 9.3 27.3

7.0 5.5 12.5

Source:  Authors’ elaboration of ECHP 1994 and EU-­SILC 2009 and 2013 microdata.

Table 6.11 Average net annual labour income, Greece, 1993–2012 (base end year changes – percentage)

Managers Professionals Technicians Remaining occupations

1993

1993–2008

2008–12

12 671.18 9729.51 7063.01 5666.10

67.1 122.3 150.0 118.0

−4.7 −17.0 −17.9 −15.6

Source:  Authors’ elaboration of ECHP 1994 and EU-­SILC 2009 and 2013 microdata.

during the crisis (Figure 6.13). The reduction was however proportionally smaller than that of total employment. As a result, the share of professionals and technicians in labour income earners of each income class increased, passing from 24.2 per cent in 2008 to 27.3 per cent in 2012 of labour income earners living in middle-­class households in particular. Over the same period, the average disposable income of professionals and technicians declined much more than that of managers (−17 per cent and

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 273

28/09/2016 08:24

274

Europe’s disappearing middle class?

−17.9 per cent against −4.7 per cent) and slightly more than that of the remaining occupations, thus negatively affecting middle-­class income, and especially average upper middle-­class market and disposable income. Notwithstanding slightly higher than average income losses, their relatively lower vulnerability to job loss than other categories of workers has maintained a large proportion of these occupational groups among the middle classes; the remainder have lost this status owing to redundancy (employees) or interruption of activity (self-­employed) that have led to impoverishment and downward social mobility for many members of these groups.

5.  CONCLUSION AND POLICY ISSUES The middle classes expanded substantially during 1993–2008 and contracted during the recent economic crisis in Greece, which evolved into a Great Depression. Taking as a reference point the income distribution at the beginning of the crisis, we have shown that 20 per cent of the population that belonged to the middle and upper classes in 2008 received an income below the 2008 poverty line four years later. Since the majority of the ‘new poor’ came from the middle classes, the hypothesis on the polarization of the class structure has been confirmed. Conversely, if we accept a relative approach to the middle classes, defined with reference to a substantially lower median income, their share of the population fell by 2.2 percentage points while their income distance from the lower class widened between 2008 and 2012. It follows that what could be described, according to the first approach, as decimation of the middle classes during the Greek Great Depression, could be seen as resilience according to the second. Whatever approach one may endorse, the middle classes in Greece, alongside the rest of the society, have experienced a terrible shock in recent years and are undoubtedly facing great uncertainty about the future, both their own and that of their offspring. During the period of growth (1993–2008) the composition of middle-­ class households changed as regards both the employment status of their members aged 15 years and over and work intensity. The share of retired persons among their adult members increased while the share of dual/ multiple earners in all middle-­class households rose substantially. The spectacular growth in the participation of women in paid work is the major change in the world of work responsible for the expansion of the middle class during the growth period. The other important determinants are the rise in the share of highly educated people in all employed; the overwhelming contribution of medium-­paid occupations in net job creation; the narrowing of wage inequalities after 2004; the ever-­increasing share of

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 274

28/09/2016 08:24



The middle classes in the Greek Great Depression ­275

professionals and technicians in the occupational structure and, last but not least, the widening of public–private sector wage differentials. The major change in the world of work that determined the retreat of the middle classes during the first four years of the current economic crisis (2008–12) is the collapse of employment and the explosion of unemployment. These developments have impinged relatively more upon the lower class but have also led to a greater decrease in the share of dual/multiple-­ earner households among the middle classes than among the other classes. The fall in this share has contributed to a retreat in the average income of the middle class, which was proportionally greater than that in the average income of the lower and upper classes. The greater decline in the income of professionals and technicians than in the remaining occupations has also influenced this outcome. However, several other developments have contributed to the resilience of the middle classes. Professionals and technicians incurred fewer job losses than the other occupational groups. Moreover, although the role of the public sector as producer of the middle classes was neutralized during the crisis, because strict hiring rules have been implemented under the austerity and reform programmes, employment in the public sector has retreated less than in the private sector, while most permanent employees who left the public sector in practical terms retired and received pensions. This may explain, to a greater or lesser extent, the further concentration of pensioners among the population of middle-­class households between 2008 and 2012. Finally, the public sector has continued to offer, on average, better employment and working conditions than the private sector and middle-­class status to its employees. Industrial relations, collective bargaining and social dialogue contributed to the expansion of the middle class during the growth period. Strong unions in the civil service, public utilities and enterprises and banking, national bargaining on sectoral and occupational minima assisted by arbitration and administrative extension of collective agreements to non-­ unionized employees and employers, as well as articulated collective bargaining, successfully regulated wage floors, reinforced the middle of the wage distribution, but also widened wage differentials. By dismantling collective bargaining and undermining social dialogue with a massive attack on the trade unions, the two Economic Adjustment Programmes that have accompanied the two loan agreements since 2010 between Greece, its euro-­ zone partners and the IMF have produced a further compression of wage inequalities, following the trend that started in 2014. However, compression is this time taking place around a falling median wage while, before the crisis, the median wage was rising. The compression is thus contributing to the impoverishment of the middle classes, although the major role in this process is being played by unemployment and company closures.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 275

28/09/2016 08:24

276

Europe’s disappearing middle class?

NOTES 1. This is so because, in bargaining with employers, all private sector union federations had systematically used national minimum wage increases as the floor and the best rate achieved by public utilities and banking federations as the target. 2. OECD.Stat online, ‘Decile ratios of gross earnings’, accessed 17 April 2016 at http://stats. oecd.org/Index.aspx?DataSetCode=IDD. 3. OECD.Stat online, ‘Decile ratios of gross earnings’, accessed 17 April 2016 at http://stats. oecd.org/Index.aspx?DataSetCode=IDD. 4. According to authors’ calculations based on ECHP microdata. 5. According to authors’ calculations based on EU-­SILC microdata.

REFERENCES Atkinson, B.A. and A. Brandolini (2011), ‘On the identification of the middle class’, ECINEQ WP 2011-­ 217, accessed 17 April 2016 at www.ecineq.org/ milano/WP/ECINEQ2011-­217.pdf. Bank of Greece (2015), Monetary Policy 2014–2015, Athens: Bank of Greece, accessed 17 April 2016 at http://www.bankofgreece.gr/Pages/el/Publications/ MonPolicy.aspx?Filter_By=11. Bigot, R., P. Croutte, J. Mller and G. Osier (2011), ‘Les classes moyennes en Europe’, Cahiers de Recherche du Crédoc, no. 282, accessed 17 April 2016 at www.credoc.fr/pdf/Rech/C282.pdf. Christopoulou, R. and V. Monastiriotis (2014), ‘The Greek public sector wage premium before the crisis: size, selection and relative valuation of characteristics’, British Journal of Industrial Relations, 52 (3), 579–602. Council of Economic Advisers (2015), 2015 Economic Report of the President, Executive Office of the President of the United States, accessed 17 April 2016 at https://www.whitehouse.gov/administration/eop/cea/ economic-­report-­of-­the-­President/2015. Dallinger, U. (2013), ‘The endangered middle class? A comparative analysis of the role played by income redistribution’, Journal of European Social Policy, 23 (1), 83–101. Elefantis, A. (1991), In the Constellation of Populism, Athens: Politis (in Greek). Eurofound (2008), ERM REPORT 2008 More and Better Jobs: Patterns of Employment Expansion in Europe, Luxembourg: Office for Official Publications of the European Communities. Eurofound (2013), Employment Polarisation and Job Quality in the Crisis: European Jobs Monitor 2013, Dublin: Eurofound. European Commission (2012), The Second Economic Adjustment Programme for Greece, European Economy Occasional Papers No. 94, Brussels: Directorate-­ General for Economic and Financial Affairs Publications, accessed 17 April 2016 at ec.europa.eu/economy_finance/publications/occasional_paper/2012/pdf/ ocp94_en.pdf Fernández-­ Macías, E., T. Ward and R. Stehrer (2012), ‘Methodology’, in E. Fernández-­ Macías, J. Hurley and D. Storrie (eds), Transformation of the Employment Structure in the EU and USA, Basingstoke: Palgrave Macmillan, pp. 16–25.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 276

28/09/2016 08:24



The middle classes in the Greek Great Depression ­277

Goldthorpe, J.H. (2012), ‘Back to class and status: or why a sociological view of social inequality should be reasserted’, Reis, 137 (January–March), 201–16. Ioannou, C. (2000), ‘Social pacts in Hellenic industrial relations: Odysseus or Sisyphus?’, in G. Fajertag and P. Pochet (eds), Social Pacts in Europe – New Dynamics, 2nd edn, Brussels: ETUI, pp. 219–36. Kanellopoulos, C.N. (1997), ‘Public–private wage differentials in Greece’, Applied Economics, 29 (8), 1023–32. Karabelias, G. (1982), The Small and Medium Democracy, Athens: Kommouna (in Greek). Karamessini, M. (1992), ‘Flexibilité du travail et restructuration du capital en Grèce dans les années 1990’, doctoral thesis, Université de Paris VII. Karamessini, M. (2012), ‘Female activity and employment trends and patterns in Greece. Women’s difficult road to economic independence’, in A. Buğra and Y. Özkan (eds), Trajectories of Female Employment in the Mediterranean, Basingstoke: Palgrave Macmillan, pp. 65–90. Karamessini, M. (2014), ‘Structural crisis and adjustment in Greece: social regression and the challenge to gender equality’, in M. Karamessini and J. Rubery (eds), Women and Austerity. The Economic Crisis and the Future for Gender Equality, London and New York: Routledge, pp. 165–85. Karamessini, M. (2015), ‘The Greek social model: towards a deregulated labour market and residual social protection’, in D. Vaughan-­Whitehead (ed.), The European Social Model in Crisis. Is Europe Losing its Soul?, Cheltenham, UK and Northampton, MA, USA and Geneva: Edward Elgar and the International Labour Office. Karamessini, M. and O. Kaminioti (1999), ‘Labour market segmentation in Greece: historical perspective and recent trends’, Discussion Papers No. 69, Athens: K.E.P.E.-­Centre for Planning and Economic Research (in Greek). Kioulafas, K., G. Donatos and G. Michailidis (1991), ‘Public and private sector wage differentials in Greece’, International Journal of Manpower, 12 (3), 9–14. Lyberaki, A. and A. Mouriki (1996), The Silent Revolution: New Forms of Production and Work Organisation, Athens: Gutenberg (in Greek). Matsaganis, M., C. Leventi and M. Flevotomou (2012), ‘The crisis and tax evasion in Greece: what are the distributional implications?’, CESifo Forum, 13(2), 26–32. Moschonas, A. (1986), The Traditional Strata of the Petty Bourgeoisie: The Greek Case, Athens: IMM-­Foundation for Mediterranean Studies (in Greek). Papapetrou, E. (2006), ‘The public–private sector pay differential in Greece’, Public Finance Review, 34 (4), 450–73. Papatheodorou, C. and M. Petmesidou (2004), ‘Inequality, poverty and redistribution through social transfers: Greece in a comparative perspective’, in M.  Petmesidou and C. Papatheodorou (eds), Poverty and Social Exclusion, Athens: Exantas (in Greek), pp. 307–66. Pressman, S. (2007), ‘The decline of the middle class: an international perspective’, Journal of Economic Issues, 41 (1), 181–200. Pressman, S. (2009), ‘Public policies and the middle class throughout the world in the mid 2000s’, Luxembourg Income Study (LIS) Working Paper 517, LIS Cross-­national Data Centre, Luxembourg, accessed 17 April 2016 at http://www. lisdatacenter.org/wps/liswps/517.pdf. Schneider, F. (2012), ‘The shadow economy and work in the shadow: what do we

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 277

28/09/2016 08:24

278

Europe’s disappearing middle class?

(not) know?’, IZA Discussion Paper 6423, available at http://ftp.iza.org/dp6423. pdf. Tsakloglou, P. and T. Mitrakos (2012), ‘Inequality and poverty in Greece: myths, reality and the crisis’, in O. Anastasakis and D. Singh (eds), Reforming Greece, Sisyphean Task or Herculean Challenge? Oxford: Sessox, European Studies Centre, St Antony’s College, University of Oxford, pp. 90–99. Tsoukalas, K. (1980), Social Development and the State. The Formation of Public Space in Greece, Athens: Themelio (in Greek). Tsoukalas, K. (1987), State, Society and Work in Post-­ war Greece, Athens: Themelio (in Greek).

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 278

28/09/2016 08:24

7. Is Hungary still in search of its middle class? István György Tóth1 1. INTRODUCTION The simplest, purely relativistic economic definition of the middle class covers people who belong to the middle of the income distribution. As the distribution of income – like any chosen continuous measure – always has a middle range, the task of identifying the middle class at first glance looks easy. By these standards, the width of the Hungarian middle-­income class is not outstanding by international comparison. In 2014, roughly 76  per  cent of all Hungarians belonged to the income bracket ranging from 60 to 200 per cent of the median equivalent income (henceforth defined as the ‘broad middle-­income class’), with an internal distribution of 16  per  cent belonging to the ‘lower middle-­income class’ (from 60 to 80 per cent of the median), 32 per cent to the ‘core middle-­income class’ (from 80 to 120 per cent of the median) and some 28 per cent belonging to the ‘upper middle-­income class’ (from 120 to 200 per cent of the median).2 Income inequality in Hungary belongs to the middle range of European Union (EU) countries and also of the Organisation for Economic Co-­ operation and Development (OECD) (OECD 2008, 2011, 2015c; Tóth 2008, 2014, 2015; Ward et al. 2009) and by these definitions, the size of the Hungarian middle class looks wide by cross-­country comparison. However, the meaning of ‘middle class’ has a much broader connotation, implying the possession of adequate housing, the possibility of geographical mobility (for example, own car), adequate resources (or insurance) to cover periods with weaker health or lower working capacities in old age, as well as the ability to provide for regular recreation for the individual and family.3 Furthermore, the middle class is not simply about those things that we have or can have, but also about their security. In other words, if a middle-­class family suffers an unexpected shock, their ability to overcome the shock without a major risk of falling out of the middle class can also be important. For example, a middle-­class person can become unemployed or may lose their home owing to loss of employment or a failing business. To 279

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 279

28/09/2016 08:24

280

Europe’s disappearing middle class?

stay in the middle class, they have to have a good chance of finding another job or at least sufficient unemployment insurance to help them through the hard times.4 By these broader standards, the picture is much less favourable in Hungary. The same source survey of Hungarian households (from which the above-­mentioned distribution was drawn)5 also shows that roughly half of the ‘core middle-­income’ households are materially deprived (lacking three of the nine listed deprivation items) and around a quarter of them are reported as being affected by ‘severe material deprivation’ (lacking four of the nine).6 What we see, therefore, is a modestly wide but seriously weak middle-­ income class in Hungary. The aim of this chapter is to present the evolution of developments leading to this characterization. In what follows I present the historical destruction and reconstruction of the (traditionally weak) Hungarian middle classes (section 2.1), followed by a presentation of the economic transition, the consequent changes in the world of work (section 2.2) and the development of the crisis (section 2.3). Section 3 is devoted to the socio-­economic composition of the middle-­income classes (sections 3.1 and 3.2) and the trends of what we call the broader middle classes (defined by occupational and consumption patterns) in sections 3.3 and 3.4. Section 4 is devoted to case studies that briefly present the skill and age bias of the transition (case study 1) and mobility trends (case study 2). Section 5 concludes.

2. WEAK MIDDLE CLASSES: A HISTORICAL ACCOUNT In pre-­transition Hungary, as in other central and eastern European countries, the official ideology was to build a classless society. Concentrated efforts to industrialize the economy led to the large-­scale social mobility of previously poor and mostly agricultural masses (Andorka 1982), while serious constraints on economic freedom and the overwhelming presence of regulation in the shortage economy (Kornai 1980, 1992b) prevented the creation of a wealth-­based upper middle class. Inequalities (at least of measured incomes and wealth) appeared relatively low (Atkinson and Micklewright 1992) in Hungary, as throughout the region. This is not to say that there was no strong social stratification, however (for Hungary, see Ferge 1969; Andorka 1982; Kolosi 1987). One of the major questions of post-­socialist socio-­economic development is how the economic transition and political transformation would affect the class structure. It was also an open question how the inherited

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 280

28/09/2016 08:24



Is Hungary still in search of its middle class? ­281

(from both pre-­socialist and socialist times) social structures would facilitate, block or alter transformation processes in post-­socialist societies. 2.1  Origin of the Historical Weakness of the Middle Classes in Hungary The fact that Hungary did not have a strong middle class after the socialist period is not a historical novelty. The social and economic structure of the country prior to the Second World War was characterized by the dominance of agriculture, with only embryonic industrialization; strong concentration in both agriculture and industry; a high degree of inequality in wealth and earnings; and, as a consequence of all this, a dualistic social structure. Back at the beginning of the twentieth century, approximately 0.5 per cent of the landowners controlled 44 per cent of the land, while most independent peasants operated tiny farms. Meanwhile, some two-­fifths of those working in agriculture did not own any land at all (Andorka 1982: 33–4). The financial and industrial sector was controlled by about 50 families, and medium-­ sized enterprises were largely missing. Large public bureaucracies lost their functions after the treaties after the First World War reduced the country to one-­third of its pre-­war territory (and population). Owing to the lack of comprehensive social and institutional reforms, inequality of wealth and income remained very high in Hungary until the Second World War. In the interwar period, about half of the Hungarian population lived in extreme poverty. In 1930, approximately 20 per cent of total incomes were concentrated in the hands of 0.6 per cent of the population, while 81 per cent shared 44 per cent of total incomes. The average income of the rich was around 50 times higher than that of the  poor (Matolcsy, quoted by Ferge 1986: 40). Stagnation was also reflected in social mobility, which may even have declined, especially among the agricultural population (Andorka 1982: 249–50). The high inequality in agriculture as well as industry and the cleavage between a quasi-­feudal class of landowners, public bureaucracies and the church, on the one hand, and a small class of modern capitalists, intelligentsia and industrial workers, on the other, created a dualistic social structure (Erdei 1980; see also Gyáni and Kövér 2003). This structure was a major obstacle to social reform (Andorka and Harcsa 1988). As a result, Hungary was a relatively backward country in the interwar period. Just before the Second World War, some 37 per cent of the national product was produced in agriculture and only 36 per cent in industry. National income per capita was less than one-­third of the most developed countries in Europe (Great Britain and Sweden), but exceeded that of most Mediterranean and eastern European countries, with the exception of Italy (Pető and Szakács 1985: 11).

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 281

28/09/2016 08:24

282

Europe’s disappearing middle class?

The public sector was relatively large in terms of public administration, employment and industrial enterprises. Hungary arrived at the Second World War with approximately a 10 per cent public sector employment share in industry (especially mining and public utilities) plus a reasonable share in agriculture (state farms and state owned forests) and an extensive public administration (Pető and Szakács 1985: 77). The post-­war land reform covering more than 35 per cent of the land could not contribute to the emergence of a strong middle class, given that 40 per cent of the expropriated land was immediately nationalized and the rest was soon taken back from the new owners in the process of forced collectivization, despite suggestions for compensatory strategies by the Smallholders’ Party (Donath 1977). Nationalization of economic activities7 went in parallel with the raising of political, economic, technical and legal obstacles to individual wealth accumulation and thus stymied the development of a middle class. The widespread regional mobilization of workers (because extensive industrialization required cheap labour) led to the break-­up of work patterns. The world of work soon became over-­politicized: party membership and loyalty to the regime became the most important criteria for getting ahead, especially at the higher levels of the hierarchy. A large share of incomes were redistributed through the state budget, again leaving only a constrained role for market forces in middle-­class formation. While aggressive forces were at work to build a socialist society, their success was undermined by opposing factors rooted in the functioning and failures of the regime itself (Kornai 1992a, 1994). The inadequate performance of the state sector pushed the government to gradually allow private activities and this led to the emergence of a ‘second economy’. This term is broader than ‘illegal economy’ because it includes all activities outside state control. Many people started living a ‘double life’: working in the socialist sector during the day and otherwise operating private or quasi-­private businesses. Approximately two-­thirds of households in the 1970s became involved in small-­scale agricultural production. At the beginning of the 1980s, small, semi-­private (or sometimes purely private) firms were established, first in trade and services, then in industry. Relaxation of controls on labour (starting at the end of the 1960s) allowed people to manage their working time more freely and to share their energies between first and second employment (Szalai 1992). Many households thus developed a diversified portfolio of jobs and earnings in order to supplement their inadequate incomes from the formal sector. This extension of the second economy later provided the nucleus of a fully private sector, and to some extent, changes on which a middle class could be built.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 282

28/09/2016 08:24



Is Hungary still in search of its middle class? ­283

The dualistic nature of Hungarian society was also characteristic – to some extent and in a different form – of the period of socialist rule. In addition to the fact that different kinds and levels of involvement in the ‘second economy’ led to further social differentiation, new, horizontal dimensions of inequality and stratification appeared. Neither people’s jobs nor their property ownership nor their incomes were enough to define their social position: material differentiation, cultural inequalities, housing conditions and power relationships all had a role in defining social strata. As Kolosi (1987: 196) has shown, some two-­thirds of Hungarian society could be described as ‘status inconsistent’ (meaning that they occupy different positions in the various dimensions of stratification). It is important to highlight that status inconsistency prevailed primarily in the middle of society, while consistently good and consistently bad combinations could be found both at the top and the bottom. The size of the consistent elite was estimated to be small (around 6 per cent), while the set of people who were deprived at least to some extent was estimated to be much larger (around 20 per cent), while status groups in the middle showed different types of status inconsistency, alongside cultural, material and regional (settlement) dimensions. In addition, the wide range of redistribution (the provision of education, health services, various in-­kind benefits, as well as the rationing of many goods, most importantly, housing) has also shaped the social structure. To summarize, the middle classes were traditionally weak in Hungary and while the socialist system allowed a rise in the living standards of the masses, no consistent middle class emerged before 1990. Status inconsistency across various dimensions of social stratification prevailed, and large classes with decent incomes and sufficient reserves could not develop. When the socialist economic and political system turned out to be unsustainable, the socio-­economic and political transition brought the promise of stabilization, accumulation and development of a middle class. The reasons for the only partial success lie in the management of wealth accumulation (privatization) and the development of labour markets (and their interaction with the social protection system). 2.2 Factors Affecting Middle-­class Formation since the Transition until Recently class formation in This section reports on various aspects of middle-­ Hungary during the transition8 and on their implications for the formation of the middle classes. To understand the political context of the systemic changes in Hungary, it must be underlined that it was not like a ‘revolution’, in which economic

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 283

28/09/2016 08:24

284

Europe’s disappearing middle class?

and political actors organize themselves to break the power of the old elite. Rather, driven by geopolitical processes such as the end of the Cold War and the collapse of the Soviet Union, the changes in the rules of the game and of the political and economic institutional settings were under the control of various elite groups, competing with each other. Because capitalization was in general very low and the class of capitalists and the middle classes were weak, the ‘intelligentsia’ working in various redistributive sectors of the economy were strong enough to be able to set the tone and shape the process of transition. State enterprises were broken up and privatized, but service sectors such as education and health care proved difficult to reform, given the many cross-­cutting vested interests and lobby groups.9 2.2.1 Privatization The privatization process in Hungary can be divided into several phases (Mihályi 2010: 29; Kolosi 2000). Starting in the second half of the 1980s (accelerating between 1988 and 1990) in a sizeable segment of the economy, a management-­initiated ‘spontaneous’ process of partial or total ownership conversion took place. This affected mostly small and medium-­sized enterprises (SMEs), in which control over usable resources ended up in the hands of the first-­or second-­tier managers of the old socialist enterprises. This was possible mainly in sectors and enterprises where the general lack of capital prevented transformation. For the privatization of larger companies, with more sophisticated institutional forms and in greater need of capital, special privatization techniques were developed. Between 1990 and 1994, the sale of marketable firms started. In some cases the breaking up of middle-­sized firms was followed by sale of assets; in other cases companies were transformed into shareholding companies and their shares were sold. The first freely elected government (led by József Antall) after the introduction of the multiparty system prioritized the creation of a national bourgeoisie. Compensation vouchers were introduced for those who lost their lands and property during nationalization after the Soviet occupation. Compensation vouchers were issued to 2 million Hungarians plus some 250 000 who were not citizens at the time; 760 000 new owners in agriculture and one-­third of total land for cultivation were affected. In addition, 300 000 people received increased pensions (Mihályi 2010). Most, however, were not interested in privatization, so newly emerging entrepreneurs started buying the compensation vouchers at discount prices, thus boosting demand for assets to be privatized. Some 50 per cent of assets sold in the main phases of privatization were paid for in vouchers (mostly bought in secondary voucher markets). In addition, the government introduced (in 1993–94)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 284

28/09/2016 08:24



Is Hungary still in search of its middle class? ­285

special, very favourable loan schemes for those wishing to buy stakes in companies for sale. This further contributed to the process of wealth accumulation of the new Hungarian upper middle classes (Kolosi 2000). However, the general undercapitalization of the economy prevailed and when in 1994 the Socialist Party, in coalition with the Free Democrats, was elected, the government (led by Gyula Horn) opted for a large-­scale sale of major firms in energy, utility and banking. This occurred on a massive scale: the largest revenue flow arrived in 1995, amounting to over 8 per cent of gross domestic product (GDP) from privatization (but 1994 and 1996 were also significant) (Mihályi 2010). As Mihályi (2010) concludes, the penetration of transnational corporations was fast and smooth in the Hungarian privatization process. However, this did not really contribute to the slow and hesitant development of home-­grown private firms,10 as a result of which the new market economy was unable to create jobs for hundreds of thousands of under-­ qualified Hungarians. In addition, various assessments of the privatization process and of the emergence of the Hungarian ‘new rich’ agree that (with a few exceptions) the old communist nomenclature was not able to convert its power and links into economic power after the transition, as they were, by and large, outmanoeuvred by the emerging young generation of Hungarian businessmen (Laki and Szalai 2004, 2013; Kolosi and Szelényi 2010; Mihályi 2010). In summary, foreign direct investment (FDI) had a major role in the initiation and acceleration of economic growth, but also in the creation of a new economic structure. The massive influx of FDI brought about significant technological modernization. Technological change increased demand for young educated labour, while employment prospects worsened for the low-­educated and older cohorts with obsolete human capital. The employment rate of the low-­educated remained the lowest in the EU owing also to the underdeveloped SME sector. 2.2.2  Transition to a lower employment equilibrium The first years of the transition witnessed a large-­scale change in employment equilibrium. This led to a situation in which a society initially characterized by seemingly high (but, owing to unemployment within factory walls, very inefficient) level of employment during the course of transition moved to a regime which was characterized by very low employment and a high rate of exit from the labour force.11 As Figure 7.1 shows, employment fell dramatically between 1990 and 1993, resulting in long-­prevailing lower employment rates. Total annual average number of employed fell from 4.9 million (4.5  million of active age) in 1990 to 3.7 million (3.6 million of active

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 285

28/09/2016 08:24

286

Europe’s disappearing middle class? Per cent

In thousands 5000

60 Employed

4500

58

Employment ratio

2013

2011

2012

2010

2008

2009

2007

40 2006

0 2005

42

2003

500

2004

44

2002

1000

2001

46

1999

1500

2000

48

1998

2000

1997

50

1996

2500

1994

52

1995

3000

1993

54

1992

3500

1991

56

1990

4000

Source:  Fazekas and Neumann (2014: fig. 4.1).

Figure 7.1 Employment (000s, left scale) and employment ratio (percentage, right scale), Hungary, 1990–2013 age) by 1995. This massive drop in employment was not fully taken up by unemployment. The two major inactive groups absorbing the difference were pensioners and other inactives, in addition to the relatively smaller increases in students and those on maternity leave. As Figure 7.2 shows, the period between 1990 and 1995 is of special interest here. This was a period of large structural change, which is discussed in the first case study (section 4.1). To better understand the long-­term freeze of overall employment levels, it is worth analysing employment rates by education categories for both sexes separately. As shown in Figure 7.3, levels of employment are relatively high for those with at least a secondary education. The long-­term low employment rates for the lowest educated, however, indicate the core of the problem. The high level of inactivity is blocked by the mutual inability of supply (low skills of the people involved, and no attempts to provide

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 286

28/09/2016 08:24



Is Hungary still in search of its middle class? ­287

6500

6000

5500

5000

4500

4000

3500

3000 1980

1990

1995

2000

2005

Other inactive

Pensioner

On childcare leave

Unemployed

Student

Employed

2010

2013

Source:  Fazekas and Neumann (2014).

Figure 7.2 Number of people (male 15–59 and female 15–54) belonging to the various labour market categories (000s), Hungary, 1980–2013 (selected years) adequate second schooling) as well as of demand (no proper jobs offered for the low skilled by the perhaps technologically ‘over-­developed’ labour market). We can conclude that the transition to a market economy and the emergence of a much more efficient economy as a ‘by-­product’ has led to the development of a large, marginalized segment. From the perspective of the middle classes this had far-­reaching consequences: it increased the clientele of the welfare state (primarily to be paid for by middle-­class taxes),

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 287

28/09/2016 08:24

288

Europe’s disappearing middle class? Males

100 90 80 70 60 50 40 30 20 10 0

1993 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 College, university

Secondary school

8 grades of primary school or less

Total

Vocational school

Females 90 80 70 60 50 40 30 20 10 0

1993 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 College, university

Secondary school

8 grades of primary school or less

Total

Vocational school

Notes: Data based on KSH MEF (CSO LFS, Central Statistical Office Labour Force Survey). Up to 2000, data are weighted on the basis of the 1990 Population Census. Source:  Fazekas and Neumann (2014).

Figure 7.3 Employment rates of population aged 15–64 by level of education and gender, Hungary, 1993–2013 (percentage)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 288

28/09/2016 08:24



Is Hungary still in search of its middle class? ­289

decreased the tax base (given that massive parts of society did not have taxable incomes) and, in purely distributional terms, created a large class of people who had low incomes (‘below’ the middle classes), without any positive effect on the absolute position of the middle. 2.2.3  Sectoral changes in employment The distribution of employment by broad economic sectors shows that massive re-­composition of the labour force took place (after a long period without change) before our reference period started (between 1949 and 1980, shaded area in Figure 7.4). While the service sector has shown a 70

60 Agriculture

50

Industry Tertiary sector

40

30

20

10

0

1900 1910 1920 1930 1949 1960 1970 1980 1989 2000 2005 2010 2014

Note:  Employment: active earners, excluding people on paternity leave and working students. Source:  1900–1989, Andorka-Harcsa (1988: 91); 1992–2005, Harcsa (2008); 2006–14, KSH (2014).

Figure 7.4 Employment by broad sector, Hungary, 1900–2014 (percentage of total employment)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 289

28/09/2016 08:24

290

Europe’s disappearing middle class?

gradual increase in this period, the decline in agricultural employment was very large (from 54 per cent in 1950 to 24 per cent in 1970), and the pace of industrialization was also remarkable (its employment share rose from 22 to 44 per cent). The period following the systemic change saw acceleration of the services sector growth and a further fall in agriculture. It should also be noted that many of the people active in the labour markets in the 1980s were mobile, moving from agriculture to industry or from industry to services, for example. This vast intergenerational (and, in many cases, intra-­generational) mobility involved different socialization experiences and employment histories for a great proportion of the population. This process also contributed to widespread status inconsistencies across the social spectrum, including the middle-­income class. The transition affected these broad trends by accelerating the decline of agricultural employment, on one hand, and the expansion of tertiary employment, on the other. Figure 7.5 shows the distribution of the active earners by socio-­ economic strata for the years between 1949 and 2014. The share of agricultural workers experienced the biggest fall, shrinking by over 30 percentage points between 1949 and 1990 and declining further thereafter to reach roughly 2 per cent of all active earners. During 1949–90, by contrast, the size of the non-­agricultural manual worker population increased by more than two and a half times, while subsequent increases meant that they accounted for roughly one-­third of all earners by 2009–14. There was also an increase in average skill levels: while, in 1949, 42 per cent of manual workers did not have any skills, that applied to only 13 per cent by 1990. However, at the end of the period the share of skilled workers was still just 51 per cent among manual workers and this share fell further by 2014, when the share of semi-­skilled and unskilled workers amounted to around 17–18 per cent of total active earners and roughly one-­third of the total manual workforce. As a result of socialist industrialization and efforts to eliminate economic units outside the state sector, the share of artisans and small shop owners fell to fewer than 80 000 people by 1970 and remained unchanged in the next decade. Even at the end of the 1980s (after years of gradual opening up) the share of this population was less than half the 1949 level, falling back to just under 10 per cent by 2009. The already mentioned sectoral re-­composition of employment was driven partially by the fact that the ‘designers’ and ‘engineers’ of the transition were mostly the high skilled intelligentsia and those working in the public sector (Kolosi and Tóth 2012a). This left its footprint on employment and wages developments in the public sector. Public sector employment is estimated to be around 30 per cent of all employees and roughly a quarter of total employment (Köllő 2014). The

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 290

28/09/2016 08:24



Is Hungary still in search of its middle class? ­291

60 50 40 30 20 10 0

1949

1960

1970

1980

1990

2003

2009

2014

Managers, intellectuals

Skilled worker

Clerical workers

Semi-skilled workers and unskilled workers

Artisan, small shop owner

Agricultural workers

Note:  Active earners: employed population of working age, excluding people on paternity leave and working students. Source:  Andorka and Harcsa (1988: 110) based on census data, between 1949 and 1990; Kolosi and Keller (2010) for 1982–2009; and Fábián (2015) for 2014 (based on Tárki Household survey). Regroupings by the author.

Figure 7.5 Active earners by social group, Hungary, 1949–2014 (percentage of all active earners) time trend of public sector employment has been volatile. The size of public sector employment increased during the political transition (from 650 000 to approximately 800 000 employees), due mainly to reclassification and the shift of previous political positions into public sector positions (Köllő 2014). This was followed by a 60 000 drop after the so-­called ‘Bokros [austerity] package’ in 1995, with similar increases at the beginning of the 2000s, followed by a major 150 000 fall in employment in the next four years.12 Despite this volatility, public sector employment is very high by OECD comparison (close to that of Sweden, though well below both Norway and Denmark; see OECD 2015a: 85). General government employment (including state-­owned enterprises) in Hungary is also reported to be among the highest in the OECD (OECD 2015b). Public sector employment is heavily gender-­biased: in 2011 the share of public sector workers was 19 per cent among men and 45 per cent

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 291

28/09/2016 08:24

292

P90/P10

P90/P50

P50/P10

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0

Europe’s disappearing middle class?

Note:  In May of each year. Source:  Fazekas and Neumann (2014: fig. 6.3).

Figure 7.6 Inequality in gross monthly earnings of full-time employees (men and women), Hungary, 1986–2013 among women (Köllő 2014: 45). Also, the public sector share is very high among the higher educated (54 per cent, primarily because of the many administration jobs, health care and education personnel) and among the lowest educated (82 per cent among those who did not finish their primary schools). This latter figure is obviously a result of the generally very low economic activity of the lower educated in general, but also of the fact that for this group public work schemes provide the bulk of their employment. 2.2.4  Wage development The rise of earnings inequality started before the political transition: the P90/P10 ratio increased from 2.6 in 1986 to 3.07 in 1989, followed by a continued increase to peak at 4.66 in 2000 (Figure 7.6). After 2000 we see a fluctuation in earnings inequality, with the P90/P10 index varying between 4 and 4.5 until 2011. Then it dropped to around 3.7 in consecutive years. The overall decline was driven mainly by a catch-­up of the lower tail (with P50/P10 dropping from above 1.9 between 2003 and 2006 to below 1.6 by 2012–2013). Another indicator of earnings inequality is the proportion of employees with low earnings (defined as the share of full-­time workers earning less than two-­thirds of the gross median earnings of all full-­time workers). This proportion rose from 13.1 per cent in 1986 to 19.4 per cent in 1992, and then 23.4 per cent in 2000. During the crisis years this ratio fell gradually from 23.1 per cent in 2006 to below 17 per cent in 2013.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 292

28/09/2016 08:24



Is Hungary still in search of its middle class? ­293

11 500 10 500 9500 8500

Czech Republic Hungary Poland Slovak Republic

7500 6500 5500 3500

1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

4500

Source:  OECD Earnings database, accessed 1 September 2015 at http://stats.oecd.org/.

Figure 7.7 Annual real minimum wages in Visegrad countries (constant 2014 USD PPS), 1988–2014 The path of the minimum wage has had two abrupt turning points. In 2001–02, the minimum wage doubled, in two stages. While average wages also grew, the relative level of the minimum wage to the average wage increased from 0.29 to 0.41 within this short period. After a stagnation during the 2000s, the government took another leap forward in 2012, at a stroke increasing the minimum wage significantly and thus raising its ratio to the average wage by 5 percentage points (to 0.42) in that year. The long-­term evolution of minimum wages (by constant 2014 USD purchasing power standards) is shown by Figure 7.7, in comparison with similar trends in Poland, the Czech Republic and Slovakia, the three important benchmark countries for Hungary. As shown, the two leaps are large in comparison as well, although similar one-­off increases also occurred in Poland for example.13 The minimum wage increasingly effected the distribution from the bottom in consecutive periods. This effect was larger for women’s than for men’s wages (Fazekas and Neumann 2014: fig. 6.5). Also, it seems that the minimum wage increase did not have a marked effect on the other segments of the distribution (rather than pushing all categories upwards, it seemed to cause ‘congestion’ in the bottom tail). This is also important for the evolution of the middle classes. While minimums have risen, it did not contribute to stronger wages for those in the middle. Rather, the inequality between the middle and the bottom has declined, as already shown by the percentile ratios. The earnings differentials between the public and private sectors were,

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 293

28/09/2016 08:24

294

Europe’s disappearing middle class?

similar to employment in these sectors, also very volatile after the transition. As Köllő (2014) shows, instability of pay differentials was more prevalent in Hungary than in any other EU member state, at least before the economic crisis. Taking the period 1993–2008, the relative pay level of the public sector fluctuated between a 22 per cent regression adjusted (by age, education and gender) disadvantage in 1996 (following the 1995 austerity package) and an 18 per cent premium in 2004 (following the massive public sector wage increase in 2003). This was followed by further erosion (partly due to the crisis, partly because of the austerity packages). All of this sent very mixed signals to a large part of the workforce (mainly to those in traditionally core middle-­class occupations). In addition to the volatility, this also indicated vulnerability to electoral (political) cycles.14 Furthermore, when relative wages of relatively large groups of typical middle-­ class occupations are examined, very low levels are found for Hungary. Two examples may illustrate this, namely, teachers’ and doctors’ wages. Comparing teachers’ salaries with the earnings of tertiary educated workers overall, 25–64 years of age, the Hungarian figure of around 50 per cent is the second lowest in the OECD (see OECD 2014: 455, indicator D3). Doctors’ wages are similar: relative remuneration (gross wages in relation to the national average) of Hungarian doctors (both specialists and general practitioners) is the lowest in the OECD (see OECD 2013: 75). While this latter figure has to be treated with great care, given the widespread system of informal payments in the Hungarian health-­care system, it clearly illustrates the problem. Typical middle-­class job categories (to which doctors certainly belong, whether public or private employees, working in hospitals or otherwise) earn just 1.5 times the national average wage (including all categories of jobs and education levels). This is a very low ratio (in most OECD countries this ratio is at least double the national average). The highly unequal distribution of their incomes (inclusive of formal salaries and informal gratuities) means that some (but only a few) specialists belong to the upper-­income class, while the majority are only average earners. For others in health care, the situation is even worse. While, for example, nurses earning 1.4 times the national average in Luxembourg (OECD 2013: 81) can reasonably be expected to belong to the middle class there, Hungarian nurses, with wages at 0.8 times the national average (OECD 2013: 81), are nowhere near it. 2.2.5 Trade unions and the institutional background to wage-­setting mechanisms Trade unions, deeply rooted in the old communist system, lost ground after the systemic changes. This is partly because structural adjustment and

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 294

28/09/2016 08:24



Is Hungary still in search of its middle class? ­295

rapid technological development in the economy drove their traditional host firms out of existence, and partly because they could not establish sufficient distance from political parties. Among other factors, these trends contributed to a serious decline in the unionization rate. Institutions of tripartite social dialogue, in existence for the first 20 years of the transition, gradually lost their strength and were dismantled by the government in 2010. With no powerful tripartite bodies in operation, the amendments of the Labour Code enabled employers to impose more ‘flexibility’ and thus much less ‘security’ for many middle-­class jobs. Trade unionism has declined in several stages in Hungary; unions can now reach and represent only a relatively small proportion of workers. Although data on industrial relations are relatively scarce and fragmented, a comparison of various sources shows very low levels since the traditional communist trade unions broke up and lost their memberships after the political changes in and around 1990. As the Institutional Characteristics of Trade Unions, Wage Setting, State Intervention and Social Pacts ICTWSS)15 database shows, unions had slightly below 4 million members in 1990, falling rapidly below 1.5 million by 1995, halving (to 700 000) by 2000, and, as the latest available estimate shows, standing at around 560 000 by 2008. This may even be an overestimation as surveys have shown lower unionization figures.16 The OECD online employment d ­ atabase17 shows that trade union density fell from around 24.5 per cent in 1999 to 10.6 per cent in 2012. The weakness of the trade unions in Hungary is partly due to their loss of credibility during the socialist period. In addition, the social-­liberal Horn government (1994–98) gave trade union leaders prominent places on their party lists, but when these leaders became part of the government, they did not effectively represent workers’ interests in the period of the first austerity packages (the above-­mentioned ‘Bokros package’). In addition, the legal successor of the ‘old’ National Confederation of Hungarian Trade Unions (MSZOSZ) trade union confederation was close to the Hungarian Socialist Party (MSZP) and signed another electoral agreement with it in 2005. Other (smaller) confederations, such as the National Federation of Workers’ Councils (MOSZ), having identified themselves as ‘Christian’ unions, maintained alliances with right-­wing parties. Others have tried to become independent from political parties, but trade unionism in general has remained ‘political’. Another contributor to the decline of union density is the nature and timing of privatization in some sectors. During the 1990s, driven mostly by privatization and structural adjustment policies, manufacturing and private services underwent many changes, resulting in a massive decline in unionization (especially for MSZOSZ). In later periods other sectors

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 295

28/09/2016 08:24

296

Europe’s disappearing middle class?

(including tertiary sectors, such as information technology, education and communications) were more affected. The public sector and public utilities, including transport and energy supply, were less affected and these areas are still unionized at a higher level. Trade unionism is very fragmented in Hungary, with six competing confederations, especially in those industries in which union membership is still relatively high (mostly large state-­owned companies). However, given that some confederations are sector-­specific (one covering manufacturing industry and private services and another utilities and transport) competition appears only in some segments of – strangely enough – public services. With other fundamental changes in the economy, the legal system was also adjusted to the changing macro environment and to the altered operational principles of the market economy at the beginning of the transition. The new Labour Code, redrafted in 1992, established a system of collective agreements at enterprise and sectoral level. National sectoral agreements were of secondary importance (except in the public sector where wage negotiations were national). The Act also stipulated that the government should discuss issues of national significance pertaining to labour relations and employment with the representative organizations of employees and employers, through the National Labour Council, which was (until the second FIDESZ government dissolved it after its landslide victory in 2010) a tripartite forum for conciliation of (government, employee and employer) interests. Although unionization in Hungary fell to very low levels after the systemic change, collective bargaining coverage was still about 35–40 per cent of total employment. For long after the transition, trade unions were able to influence bargaining developments through the tripartite Országos Érdekegyeztető Tanács (OÉT  –National Interest Reconciliation Council), although the extent of their influence depended on the government. The Interest Reconciliation Council existed until 2011 and provided a forum in which the three parties could agree on the national minimum wage and set a minimum rate for skilled workers. It also made (­non-­binding) recommendations to lower-­level negotiators on proposed pay increases. However, in 2011 the government replaced the OÉT with a new body, the Nemzeti Gazdasági és Társadalmi Tanács (NGTT – National Council on Economy and Social Issues) with more participants (including chambers of commerce and churches), but fewer competences (it was no longer able to set the minimum wage). Since then, Hungary’s national minimum wage has been laid down by government decree. The new Labour Code of 2012 has resulted in a significant increase in ‘flexibility’ and thus far less security of employment (more room for firing at shorter notice, cuts in severance pay, options for firing without giving detailed reasons, and longer probation periods). The labour market effects

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 296

28/09/2016 08:24



Is Hungary still in search of its middle class? ­297

of these changes are not yet known, but early assessments conclude that job security has indeed suffered serious damage (Laki et al. 2012); furthermore, as most of these changes have affected public sector jobs, middle-­ class job security has been seriously and negatively affected. This has been exacerbated by adverse changes in unemployment provisions (lower coverage, extremely short eligibility). 2.2.6  Inequality and polarization An analytic overview of changes in income dispersion between 1987 and 2014 in Hungary shows substantial shifts between critical moments of the transition: 1987 (baseline), 1992 (deep recession), 1996 (austerity), 2001 (high GDP growth), 2003 (massive welfare expansion), 2006 (austerity and tax rises), 2010 (austerity and benefit cuts) and afterwards (benefit cuts and shift towards workfare) (Tóth 2005, 2015). Inequality increased the most during 1987–92. After slowing down, there was a levelling off, then another increase due to the 2008 crisis (Tóth and Medgyesi 2011). The evolution of the major inequality indices can be followed in Table 7.1, presenting various measures sensitive to various parts of the distribution. Income inequality began to increase as early as the 1980s, with the Gini index reaching 0.24 in 1987 (KSH 1990). There was a marked increase in inequality during the first years of transition and the Gini index reached 0.30 in 1995 (Andorka et al. 1997). Changes in inequality were smaller during the period 1995–2005, with stagnation overall (or even a fall in inequality in 2003–07) until the economic crisis, followed by a significant increase afterwards. Redistribution policies during the 2000s have played a significant role, Table 7.1 Distribution of equivalent incomes, Hungary, 1987–2014 (as measured by indicators sensitive to different segments of the income distribution) 1987

1992

1996

2000

2003

2005

2007

2009

2012

2014

P90/P50 1.69 S10/S1 4.55 P90/P10 2.8 Gini 0.236 P10/P50 0.60

1.86 5.52 3.1 0.263 0.59

1.90 6.62 3.6 0.290 0.54

1.92 6.63 3.5 0.292 0.55

1.92 7.30 3.58 0.302 0.54

1.91 6.68 3.42 0.291 0.56

1.74 6.00 3.16 0.271 0.55

1.81 6.35 3.53 0.272 0.51

1.84 7.3 4.03 0.293 0.46

1.71 7.1 3.58 0.288 0.48

Note:  PX = Xth percentile value. SX = share of the xth decile from total incomes. Author’s calculations. Source:  Own computations based on various surveys: 1987, KSH Income Survey; 1992, 1996, Hungarian Household Panel (HHP); 2001–14, Tárki Household Monitor.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 297

28/09/2016 08:24

298

Europe’s disappearing middle class?

first by increasing transfers to the lower middle-­income class, and then increasing the tax burden on the upper middle-­income class (Szivós and Tóth 2008). Most indicators of income distribution show growth of inequality between 2007 and 2012 and then stagnation between 2012 and 2014. As the gap between the two ends of the distribution widened, households at both the low end and the top experienced losses, but the sharpest decrease in real incomes was suffered by the lowest decile (Table 7.1; also Szivós and Tóth 2015). The size of the middle groups shrank over time (with the exception of a small broadening in 2009). The share of the well-­off seemed to increase between 1987 and 2003 (from 5 per cent to 9 per cent), but afterwards the circle of those fortunate enough to have incomes at least two times higher than the median shrank (to 6 per cent by 2014). The poverty rate (below 60 per cent of the median) rose from around 10 per cent in 1987 to around 18 per cent by 2014 (panel A of Figure 7.8). As for the relative share of the various income classes in total incomes (panel B, Figure 7.8), there was an overall gain in the two upper-­income groups (with a slight interruption in 2009). Panel A: Share of people in various income classes

100

100

90

90

80

80

70

70

60

60

50

50

40

40

30

30

20

20

10

10

0

1987 1992 1996 2003 2009 2012 2014

–60

60–80

80–120

120–200

200+

0 –60

Panel B: Share of income classes in total incomes

1987 1992 1996 2003 2009 2012 2014 60–80

80–120

120–200

200+

Note:  Income classes defined relative to the median of equivalent household income distribution. Source:  1987: HCSO Income Survey; 1992, 1996: Hungarian Household Panel; 2001–14: Tárki Household Monitor.

Figure 7.8 Relative size of various income classes (percentage of median equivalent household disposable incomes), Hungary, 1987–2014

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 298

28/09/2016 08:24



Is Hungary still in search of its middle class? ­299

A combined reading of the two panels of Figure 7.8 shows a concentration of incomes. The income-­poor in 1987 (who constituted 10 per  cent of society) had 4 per cent of total incomes, while the income-­poor in 2014 (accounting for 18 per cent of society) controlled 7 per cent of total incomes. At the other end, the well-­off (5 per cent of society) controlled around 13 per cent of total incomes in 1987, while in 2014 the well-­off (now 6 per cent) were reported as having 16 per cent of all incomes in 2014. The core middle class shrank from 40 to 32 per cent (number of people) and from 35 to 29 per cent (income share) between 1987 and 2014. 2.2.7  GDP convergence We have seen how the chances of forming a strong middle class have been constrained by government policies and developments in the world of work. However, a middle class could also have developed without any relative change if wages, incomes and living standards had risen across the board. It is therefore worth looking at GDP and household income convergence trends. Oblath (2014) underlines that the Hungarian economy seriously underperformed (in terms of GDP convergence with other EU countries) between 1991 and 2013. In terms of growth in GDP per capita, out of the 22 years observed, only in six years did the Hungarian economy achieve significant convergence (exceeding a 1.5 per cent annual change) with the EU15. Per capita GDP convergence was low, whether calculated in current or real purchasing power parity (PPP) terms. Constant PPP time series (fixed at 2013 levels), adjusted for population composition change (see Figure 7.9, based on Oblath 2014), even show a setback between 2006 and 2013. At the micro level, median equivalent household disposable incomes (European Union Statistics on Income and Living Conditions – EU-­SILC) also show much slower growth (even halted at certain points) than those of Czech, Polish and Slovak households (Figure 7.10, based on Szivós 2014). The pace of growth of household incomes was highest in Slovakia for most of the period, while in Hungary it lagged behind. Income and consumption (overall and food) figures from national accounts also show very similar trends (Szivós 2014). A cross-­section comparison of middle incomes in purchasing parity standards is proxied in Figure 7.11. The width of income bands is roughly similar in Hungary, Croatia and Lithuania, meaning that the absolute material situation of the middle-­income classes is roughly comparable in these countries. However, when the income situation of the Hungarian upper middle class is compared to neighbouring Austria’s lower middle class, we see very serious disparities. Part of the Hungarian upper middle-­income class has smaller incomes

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 299

28/09/2016 08:24

300

Europe’s disappearing middle class?

75 70 65 60 55 50 45

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

30

1994

Slovakia

1993

Hungary

Poland 1992

Czech Republic

35 1991

40

Source:  Oblath (2014: fig. 3b).

Figure 7.9 Relative GDP change in Hungary, Czech Republic, Poland and Slovakia, 1991–2013 (2013 PPP) 12 000 11 000 10 000 9000 8000 7000 6000 5000 4000 3000

2005

2006

2007

2008

2009

Czech Republic

Hungary

Poland

Slovakia

2010

2011

2012

2013

Source:  Szivós (2014), based on Eurostat data.

Figure 7.10 Median equivalent household disposable incomes in euros, Hungary, Czech Republic, Poland and Slovakia, 2005–13 than people in Austria’s lower middle class (in other words, some Hungarians who qualify as ‘upper middle’ at home have incomes that would qualify them as poor in Austria (and in many EU15 and even some Central and Eastern Europe – CEE – countries).

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 300

28/09/2016 08:24



Is Hungary still in search of its middle class? ­301

70 000 60 000 50 000

Lower middle Middle middle Upper middle

40 000 30 000 20 000 10 000

HU EE PL SK EL CZ PT ES SI MT IT IE BE FI DK UK SE NL DE CY FR AT LU

RO BG LV LT HR

0

Note:  Definition of ranges: lower middle: from average of second decile to average of fourth decile, middle: from average of fourth decile to average of seventh decile, upper middle: from average of seventh decile to average of tenth decile. Source:  Calculations based on cross-sectional EU SILC 2014, UDB August 2014. Social Situation Monitor, accessed 7 May 2016 at http://ec.europa.eu/social/main. jsp?catId=1049&langId=en.

Figure 7.11 Average incomes of various income bands (proxying income classes) in the EU28 (in euros at PPS), 2011 2.2.8 Income security of middle-­income class households: effects of crisis and policies The financial and economic crisis has provoked policy changes that, together with the evolution of the crisis itself, have had a long-­term effect on households, adversely impacting large segments of society. In the second half of 2008, the crisis was primarily of a financial nature and affected those with substantial savings. Some of the losses were ‘virtual’ in the sense that those who could afford to hang on to their securities or savings avoided having to realize a loss. Those, however, who for some reason had no choice but to ‘cash out’ suffered an effective loss of assets. In the next phase of the crisis, the volatility of the Hungarian forint led to an increase in interest payments on housing and car loans originally denominated in foreign currencies. As a result, the living standards and consumption potential of affected households were substantially narrowed, although the effects were not directly reflected in income distribution statistics. After stagnating in the late 2000s, the Hungarian economy was hit by

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 301

28/09/2016 08:24

302

Europe’s disappearing middle class?

severe recession in 2009. Gross domestic product fell by 7 per cent, with both employment and real wages declining. Employment fell by 100 000 and unemployment increased from 7.8 per cent in the second half of 2008 to 10 per cent in first half of 2009 (Köllő 2011). Gross real wages declined by 3.4 per cent, net wages a bit more moderately. Adjustment was completely different in the public and in the private sector, however. The public sector showed no decline (even an increase) in employment, while real wages declined by more than 10 per cent. In the private sector real wages remained at their pre-­crisis level, while the number of employees declined by 7 per cent (Köllő 2011). The decline was achieved by halting recruitment rather than massive layoffs. In the third phase, the effects brought on by the crisis were mediated by the austerity programmes announced in spring 2009: cuts in government spending with public sector consequences and contraction of the welfare system. The measures included cuts in pensions (the thirteenth month pension was abolished), family support (a freeze on child benefits) and social transfers (an upper limit introduced on household entitlements). Households responded to these measures by showing considerable – perhaps overcautious – restraint in spending, which of course only exacerbated the economic downturn. After the major macro shocks, the fourth phase of the crisis saw strong interventionism by the government after the 2010 elections. A supply-­side economic policy gained ground and renationalization occurred in various sectors. There was a concerted effort to redirect social assistance recipients into the secondary labour market via public works programmes. At the same time, the government endeavoured to protect households from further austerity. Since 2010, various policy measures18 have affected the position and behavioural constraints of households, in response to public disquiet. Here we consider only a few. As already mentioned, benefit cuts included a reduction of the ceiling on unemployment and other benefits and a shorter eligibility period. An extension of conditionality was introduced (child allowance was tied to school attendance and a premium for kindergarten attendance was introduced). On the contribution side a flat-­rate income tax (with substantial tax breaks for – in particular, large – families with children) was introduced. In the labour market a minimum wage increase was announced and firms were forced to compensate employees’ losses due to tax changes. With regard to living costs, the government intervened in the financial sector and the banks were forced to accept that over 160 000 households pay back foreign currency mortgages in forints. There were some layoffs in the public sector and unemployment in general climbed to levels comparable with those of the early 1990s in the first period of

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 302

28/09/2016 08:24



Is Hungary still in search of its middle class? ­303

the crisis, but then fell back slightly, partly as a result of public works programmes and partly as a result of job creation in the primary labour market. The decline in incomes owing to the crisis caused financial difficulties for households – primarily in the middle-­income class – with a loan (especially in foreign currency) repayment obligation (Bernát and Kőszeghy 2011). After a decade of stagnation, the last two years of the past decade saw an increase in the percentage of households experiencing material deprivation or financial difficulties. The proportion of households having difficulties paying their rent or utility bills also increased (from 12 to 18 per cent as early as 2012). After 2012, this repayment burden was eased somewhat (Szivós and Tóth 2015). The rising frequency of households with loan repayment obligations characterized the whole 2000s, both for bank debts and non-­bank debts. The various bank loans also reached low-­income groups at that time; the incidence of loan debts among them roughly corresponds to the average. The proportion of households with regular loan repayment obligations among low-­income groups continued to grow between 2010 and 2012, but seemed to halt between 2012 and 2014 (Tóth 2015). income households have suffered most from the As expected, low-­ burden of loan repayments. Indebted households in the lowest income quintile inevitably pay a higher share of their income towards their debts and are also more likely to be in arrears because of financial difficulties. Our analysis of differences between various demographic groups also indicates that the incidence of repayment difficulties is higher for typically low-­income population groups (those with low educational attainment, the unemployed, large households and households in disadvantaged geographical regions).

3. MIDDLE CLASS COMPOSITION IN PERSPECTIVE (1980–2014) 3.1  Socio-­demographic Composition of the Middle-­income Classes Changes in the socio-­ economic composition of the various levels of middle (and other) income groups can be followed in Figure 7.12. The middle-­ income group – together with Hungarian society as a whole  – is clearly more highly educated in 2014 than it was in 1987. However, the social gradient also became steeper over time: higher education degrees provided a more efficient passport to the middle classes and to the higher social strata in 2014 than in 1987. With regard to various labour

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 303

28/09/2016 08:24

304

Europe’s disappearing middle class? 1987

2014

200+

200+

120–200

120–200

80–120

80–120

60–80

60–80

–60

–60

0%

50%

100%

Employed, other also Pensioner, not employed

Employed Inactive/unemployed

0%

50%

100%

Employed, other also Pensioner, not employed

Employed Inactive/unemployed

Employment situation of household head 2014

1987 200+

200+

120–200

120–200

80–120

80–120

60–80

60–80

–60

–60

0%

50%

Max primary Tertiary

0%

100%

50%

Max primary

Vocational Higher education

100%

Vocational Higher education

Tertiary

Education of household head 1987

2014

200+

200+

120–200

120–200

80–120

80–120

60–80

60–80 –60

–60 0%

50% –34

0%

100% 60+

35–59

50% –34

100%

35–59

60+

Age of household head 1987

2014

200+

200+

120–200

120–200

80–120

80–120

60–80

60–80

–60

–60

0%

50% 0

1

100% 2

3+

0%

50% 0

1

100% 2

3+

Number of children in household

Source:  Own computations based on KSH Income Survey for 1987 and TÁRKI Household Monitor for 2014.

Figure 7.12 Distribution of people in various income brackets by socio-demographic dimensions, Hungary, 1987 and 2014

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 304

28/09/2016 08:24



Is Hungary still in search of its middle class? ­305

market participation categories, however, the picture looks less clear-­cut. The share of pensioners (partly owing to demographic ageing and partly to expansion of the pension system) has increased in the whole spectrum, but mostly in the middle. The increase of inactivity appeared almost exclusively in the group of the income-­poor. This may have been an important reason behind the fact that households with an employed head were much less often found in poverty in 1987 than in 2014. Correspondingly, there is a characteristic shift in the composition of the various income brackets by age groups. The share of younger households increases as we move down the income ladder in both 1987 and 2014, but poverty risk (the relative concentration of the younger cohorts in poverty compared with their share in other income brackets) increases significantly during the period. The case of older households is peculiar. There seems to be a non-­linear relationship here: while the share of older age households was highest in lower-­income groups in 1987, parallel to the moves in the concentration of the younger cohort, aged households tended to shift towards the middle-­income groups. Finally, the share of those living in childless households increased from 41 to 58 per cent from 1987 to 2014. The majority of this overall increase appeared in the upper, upper-­middle and middle-­income groups, while the share of households with dependent children (especially those with three or more) increased among the income-­poor. Poverty rates for childless households decreased, while poverty rates for households with children decreased in all categories of households with children. It is important to note that the poverty risk of single parent households is high and has increased. In summary, the composition of the middle-­income class has tended to shift (as in society as a whole) towards the highly educated, and towards the childless; it has also become older, more than society as a whole. 3.2 How Strong are Middle-­income Groups? Deprivation and Middle-­ income Classes In addition to socio-­demographic variables, vulnerability to various material shocks also differs along the social and income hierarchy. An absolute proxy of the vulnerability to material shocks of an income group is the prevalence of material deprivation. According to Eurostat, material deprivation is defined as the inability to pay for at least three of the below-­mentioned items and is expressed as a headcount or as a percentage of the total population: (1) rent, mortgage or utility bills; (2) keeping the home adequately warm; (3) facing unexpected expenses; (4) eating meat or proteins regularly; (5) going on holiday; (6) having a television set; (7) having a washing machine; (8) having a car; (9) having a

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 305

28/09/2016 08:24

306

Europe’s disappearing middle class?

Table 7.2 Prevalence of various deprivation items in various income brackets, Hungary, 2014 (percentage of people in the various brackets reporting the given problem)

Inability to meet   unexpected expenses Problem going on holiday Problem paying utility bills Problem buying meat Problem paying for heating No washing machine No colour television No telephone No car use Lack of at least three items Lack of at least four items Lack of three ‘middle   class essential’ items

−60

60–80

80–120

120–200

200+

Total

93

92

75

56

31

73

89 34 69 49 22 2 12 57 81 69 47

81 20 51 26 9 1 2 36 64 42 24

68 8 35 12 1 0 1 19 42 20 12

46 6 18 7 0 0 1 9 22 10 6

21 3 4 0 0 0 0 2 5 0 0

65 14 37 19 6 0 3 25 44 28 18

Source:  Author’s calculations based on Tarki Household Monitor 2014.

telephone. Inability to pay for at least four of the above items defines severe material deprivation. Three out of four of Hungarian core middle-­income class households declare it difficult or impossible to meet unexpected expenses, while almost 70 per cent reports difficulties going on holiday for a week (Table 7.2). Two out of five in this group says they are unable to pay for three out of the nine items, a fifth of them report four out of the nine items difficult or impossible to meet. Even a third of the highest income-­bracket members declares it difficult to meet unexpected expenses and one in five reports difficulties paying for a week’s holiday. These rates are certainly too high. For a person or a household to qualify as ‘middle class’, as the minimum they must be free of material deprivation. The figures presented here highlight that the middle-­income category cannot be really called ‘middle class’ (in sociological terms). 3.3  Beyond Middle-­income Classes: The Middle by Broader Definition A broader definition of the middle class (stemming from a broader perspective on social structures, see Kolosi 1987, 2000; Goldthorpe 2012; Atkinson and Brandolini 2013) may include a number of

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 306

28/09/2016 08:24



Is Hungary still in search of its middle class? ­307

Table 7.3  Class structure of Hungarian society, 1982–2009 (percentage) Class position Elite Upper middle class Middle class Working class Deprived N

1982

1992

2003

2009

2.3 4.4 42.3 36.3 14.7 14 011

1.8 8.8 40.3 34.7 14.4 3213

2.6 6.8 32.8 38.3 19.6 4211

4.6 8.2 29.5 39.3 18.4 3575

Source:  Kolosi and Keller (2010: 121). Data from stratification model survey of 1982, Hungarian Household Panel Study of 1992 and Tárki Household Monitor surveys of 2003 and 2009.

other s­ tratification dimensions (housing conditions, savings and wealth, lifestyle). Earlier studies based on neo-­Weberian accounts of social stratification in Hungary have made attempts to map the relative sizes of the various social classes defined in a multidimensional framework (Fábián et al. 1998, 2000; Kolosi and Keller 2010; Kolosi and Pósch 2014). These approaches combine dimensions other than incomes to define middle classes. They vary in terms of the dimensions they apply (housing, wealth, savings, possession of various durable goods, cultural consumption and lifestyles). However, the general message – that classifications based purely on income will miss important points about the middle classes – is the same. The approach to social stratification of Kolosi and Keller (2010) is based on a social status index (a composite measure of income, wealth and housing conditions), combined with occupation (current for actives and last occupation for inactives). Their social structure time series is shown in Table 7.3. A recent study by Fábián (2015), following the methodology of earlier studies by Tamás Kolosi (see Kolosi 1987, 2000; Fábián et al. 2000; Kolosi and Keller 2010; Kolosi and Pósch 2014) combines incomes (household, equivalized), wealth (mostly housing, but also some durables) and consumption to define social classes. Aggregated (via z-­scoring – ­normalizing to the mean) from a number of elementary indicators, a composite is being and people are classified created reflecting overall material well-­ based on this. The final typology shows that there are five basic categories of social structure in Hungary (Table 7.4). First, the ‘lower status group’ contains roughly 36 per cent of the population. Another 30 per cent belongs to the ‘lower middle’ category. This means that, at a rough estimate, some

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 307

28/09/2016 08:24

308

Europe’s disappearing middle class?

Table 7.4 Size, income and savings of the various social classes, Hungary, 2014 Class position

Population share, households, %

Household net disposable income as % of average

Upper Upper middle Middle Lower middle Lower

4 11 20 29 36

267 158 124 88 57

Per capita net disposable income, as % of average

Current savings,* as % of average

236 135 119 90 70

400 207 122 67 44

Notes: N = 1739. * Incomes minus current consumption expenditures on food, housing and mortgages. Source:  Author’s calculations, based on Fábián (2015).

two-­thirds of Hungarian society belongs to classes that are (in terms of consumption and wealth but also cultural activities) lower than the middle classes. Roughly 20 per cent can be categorized as ‘middle’, another 10  per  cent ‘upper-­middle’ and around 4 per cent ‘upper’. It should be emphasized that this categorization is by and large consistent with the results of similar earlier studies referred to above. The composition of the various social strata follows expected differentiation. The uppermost three status groups (the middle classes in a broader sense) can be found mainly in Budapest and other cities and their housing conditions (judged by the survey) are significantly better. Lower-­status groups are older, while household size and consumption status seems to correlate positively. What is important is work intensity (WI – measured as proposed by Eurostat’s WI indicator), which is significantly lower for the lowest two groups than for the middle and upper classes. Also, education correlates highly with social status – the higher educated are extremely underrepresented in the group of the deprived and overrepresented in the upper three social strata. This also draws attention to perhaps one of the most serious problems of the Hungarian social structure – that is, the presence of a large, lower educated, marginalized and deprived group in society. This is another argument for not simply sticking to income-based and ‘symmetric’ measures of inequality.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 308

28/09/2016 08:24



Is Hungary still in search of its middle class? ­309

4.  CASE STUDIES In two case studies we seek to illustrate key points of the argument so far. In the first, the selection (or clearing) process of the labour market adjustment at the beginning of the 1990s favoured particular age and education categories, thus providing opportunities for different labour market segments to move towards middle-­class status. The second adds to our understanding of middle-­class mobility from an intergenerational perspective and for a longer time span. 4.1 Case Study 1: The Age/Education Bias of the Transition and Consequences for the Middle Class Section 2 highlighted that the shift to a lower employment equilibrium was highly selective by skills and by age. This case study is about how this happened and what consequences it might have for middle-­class formation in Hungary. Labour market transition initially led to increasing inactivity and unemployment. A special feature of the process was that structural adjustment in labour markets involved a much larger share of job-­to-­job shifts than previously expected. The dominant routes led from employment to employment (perhaps with sectoral change in between), together with paths from the labour market to inactivity. A small proportion only changed sector through paths from (public sector) employment via unemployment to (private sector) employment. The causes and consequences of this process are complex, but education policy failure (inadequate skills portfolio of the low educated), badly calibrated economic policies (mostly failed assumptions about incentive structures and behavioural reactions in terms of labour supply) and a (perhaps too) rapid modernization of the economy (creating high-­skilled jobs but offering almost nothing to the marginalized low skilled) all played their role (Köllő 2009). Figures on stocks in Table 7.5 show no real differences in the average years of schooling of people working in the public or private sector, while the unemployed were younger and much less educated. The share of women – and this seems to be a peculiarly Hungarian phenomenon – was much lower among the unemployed (35 per cent) than that of men and more concentrated in the public sector, which showed a high rate of feminization (almost 60 per cent). Turning to the flows between labour market categories, we see that people changing from public to private firms were younger and more educated, with women being underrepresented. On the other hand, people

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 309

28/09/2016 08:24

310

Europe’s disappearing middle class?

Table 7.5 Social characteristics of labour market actors, Hungary, 1992–93 Observed population

Total active age population Public sector Private sector Unemployed From public to private of which: ‘privatized’ ‘mobile’ From unemployed and inactive to private From private to public From private to unemployed and inactive

Share of women

50.3 59.1 46.4 34.7

Average age Completed (years) school years (average) Stock (March 1993) 37.70 37.72 37.32 36.56

10.57 11.32 11.24 9.90

Flows (between April 92 and March 93) 52.8 38.0 11.0 51.8 56.7 56.4 58.0 39.9

38.4 36.4 30.5 38.2 40.7

10.9 11.4 10.9 11.0 10.0

Notes: N = 2610. Italic/bold: under/overrepresented in the relevant group, respectively. Source:  Hungarian Household Panel.

working in privatized firms were clearly older and less educated, with women even more underrepresented. Flows from unemployment to private sector employment were characterized by a larger share of women (56 per cent), by the better educated and by younger people. For example, the average age of these ‘lucky’ unemployed was some six years lower than the average age of the unemployed in general. Generally, the road from private sector employment led mostly to unemployment and inactivity. Mostly male, the lower educated and older persons travelled that path. Structural adjustment, therefore, seemed to favour the higher educated, younger people and women, at every possible point of potential labour market shifts. This one-­year transition perspective cannot, of course, offer a particularly broad picture. In a broader context, Kertesi and Köllő (2002), when describing the process of skills obsolescence in the post-­socialist economy, differentiate two early phases of the transition. In the first period there was a massive collapse of demand for unskilled labour, while technology

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 310

28/09/2016 08:24



Is Hungary still in search of its middle class? ­311

import and high-­technology FDI was at a very early stage and increased returns to education were still only relative: there were no particularly substantial gains for the higher educated, but their relative position improved compared with the low skilled, who lost out dramatically. At the next stage, when market institutions were operating at full swing and modern technological FDI was coming in on a massive scale, the gains of younger educated people grew, while returns to experience continued to decline. In other words, the returns to education increased only for those just coming into the labour market. Returns to older higher-­educated people started to decline, together with a general devaluation of experience (Kertesi and Köllő 2002). To illustrate the implications for relative position of households (characterized by age and education of household heads), a ‘quasi-­ ­ longitudinal’ cohort analysis can be presented. The first cohort we differentiate was born before 1931, of which the youngest member was around 60 in 1990; we may call them ‘retired in 1990’. Members of the second group (born between 1932 and 1951) were (being 39–59 years old in 1990) supposed to be ‘on top in 1990’. The third group, born between 1952 and 1965, reached the age of 25–38 in 1990 and are termed the ‘transition core’ generation. Finally, the systemic change found the fourth group at 14–25 years of age in 1990, hence we term them ‘career starters in 1990’. The story is as follows. The within-­cohort variance is smallest among the ‘retired in 1990’ group. This is not surprising as their income differentials were accumulated in the socialist period and most were in retirement after 1990. The within-­cohort variance (by education group) is largest among the ‘career starters in 1990’ and ‘transition core’ groups. In these groups, which were already on their education track when the systemic change came, status depended on how far they got up the educational ladder. Those attaining higher qualifications received by far the highest premium compared with their lower-­educated peers. The time-­change of the relative position (compared with annual national average) of the higher and the lower educated in various cohorts clearly shows (Figure 7.13) that the lowest educated are paid the least in each cohort. Moreover, this group’s relative position seems to have deteriorated over time. The relative position of the higher educated in the ‘retired in 1990’ cohort also seems to have been deteriorating, probably because of this ageing group’s reliance on pensions. The relative position of the higher educated within the ‘on top in 1990’ and the ‘transition core’ groups seems to show a ‘normal’ (Mincerian) inverted U-­shape over time. Their relative position first grows to a high level (especially for the higher educated of the ‘on the top in 1990’ cohort), then starts to fall. The fact that the career path of the higher educated in the youngest cohort does not really show

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 311

28/09/2016 08:24

312

Europe’s disappearing middle class?

210 190 170 150 130 110 90 70 50

1987

1992

1996

2005

2010

2014

Primary, born 66–77

Primary, born 32–51

Secondary, born 66–77

Secondary, born 32–51

Tertiary, born 66–77

Tertiary, born 32–51

Primary, born 52–65

Primary, born before 31

Secondary, born 52–65

Secondary, born before 31

Tertiary, born 52–65

Notes:  Age and education of household head. Author’s calculations. Some categories (secondary educated and those born after 1977) were excluded owing to lack of space or the small number of cases. Sources:  Own computations. 1987: KSH Income Survey, 1992–96: Hungarian Household Panel, 2005–14: Tárki Household Monitor Survey.

Figure 7.13 Person equivalent household incomes of the various age/ education categories, as a percentage of annual national average, Hungary, 1987–2014

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 312

28/09/2016 08:24



Is Hungary still in search of its middle class? ­313

this inverted U-­shape (with a little exaggeration: it starts declining before it could have increased) is a new and remarkable phenomenon. It seems to indicate the emergence of a lost generation, perhaps owing to congestion among older workers just ahead of them on the career ladder, perhaps because the crisis hit them at a more vulnerable phase of the life cycle. To sum up, the economic transition led to a drastic revaluation of relative positions between households with different initial asset portfolios in Hungary. Institutional restructuring, structural adjustment and (social) policy reforms all played their part in reshaping the relative economic prospects of households in various social segments. In this process, much depended on how households were able to combine their members’ skills, educational levels and employment activities. The process was deeply selective, starting with a strong disregard of low skills at the beginning, followed by increased appreciation of high skills with the massive influx of high-­technology FDI. Long experience – even if in highly skilled jobs – did not pay enough for the older generations to keep up with younger entrants. A potential ‘lost generation’ of the higher educated was also found, with clear implications for the weaker supply of new entrants to the middle classes. 4.2 Case Study 2: Groups and Factors behind Middle-­class Mobility in Hungary, 1992–2007 This case study provides background information on intergenerational up research (Lifepath Survey of Hungarian mobility based on follow-­ Households – LSHH) of the 1992 wave of the Hungarian Household Panel (HHP) survey (covering 2000 households) in 2007. Although a small survey, it is fortunate that this information is available for a relatively long, 15-­year time-­span real panel and for some variables, through retrospective questions, for much longer periods. The completed survey covered 2682 individuals of the original sample.19 Results show that one-­tenth of the respondents (about 8 per cent of Hungarians in 2007) belong to the underclass, originating in the early shocks of the systemic change, when a significant decline in employment resulted in their permanent and/or periodic exclusion from the labour market. Later on they were attracted/forced to retire for economic reasons. These people can unequivocally be regarded as the real losers of regime change; by 2007 most were living on welfare. For some others, the upsurge in entrepreneurship in and around the systemic change has provided prospects (however risky). In the 25-­year period after 1980 approximately 9 per cent of all respondents had, at some stage, been involved, at least for some time, in running a business (cumulative

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 313

28/09/2016 08:24

314

Europe’s disappearing middle class?

Table 7.6 Proxy for income mobility: transition matrix, quantiles of individuals, based on a combined material – financial index, Hungary, 1992–2007 (percentage) 2007 Bottom 25% 1992

Bottom 25% Middle 50% Upper 25% Total

57.4 18.4 9.4 26.0

Middle 50%

Upper 25%

Total

36.1 59.4 42.7 49.3

6.5 22.1 47.9 24.7

25.2 49.6 25.2 100.0

Source:  Kolosi and Tóth (2008a).

figure). Roughly half of them (4.4 per cent) were in business on their own account in 2007. The overwhelming majority of those in this position in 2007 (90 per cent) set up their ventures after the change of regime, but more than a third of them were either employees or not working by 2007. A relatively small proportion of those who remained in the labour market for 15 years were upward or downwardly mobile. On the whole, the employment hierarchy proved to be relatively stable during the regime change. The first half of the 1990s was characterized by significant employment turbulence (Róbert and Balogh 2008). This was prompted by internal shifts (politically inspired economic processes, such as privatization), but globalization, further strengthening the post-­industrial processes, the expansion of the service sector and the influx of multinational capital also had an effect. In terms of material conditions and income status, both a decline and a rise could be observed (Table 7.6). Almost 60 per cent of those who, in 1992, belonged in the bottom quartile were there in 2007, while almost half of those in the top quartile were still there 15 years later. That is, in the wake of the systemic change, examples of both upward and downward movements were to be found in society, but there was a larger movement at the top than at the bottom. Five categories of people are of particular interest here. The first is the stable group; 60 per cent of the 1992 middle belonged to the middle in 2007 as well. The second and third groups were able to improve their relative positions: 36.1 per cent of the 1992 bottom quantile rose to the middle and somewhat more than one-­fifth of the middle in 1992 rose to the top by 2007. The fourth and fifth groups slipped back in this 15-­year period. Some one-­fifth (18.4 per cent) of those in the middle ended up in

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 314

28/09/2016 08:24



Is Hungary still in search of its middle class? ­315

the bottom quantile and over two-­fifths (42.7 per cent) of the top fourth in 1992 had fallen back to the middle by 2007. In order to present the differential chances of the various categories moving up or down we first calculate the composition of ‘stayers’ and ‘movers’ from a certain category.20 Ratios for the various category outflows and inflows (Table 7.7) tell the following story. For upward mobility from the bottom quartile (getting into the middle class), (1) those with relatively high cultural capital (measured by parental education); (2) those running their own business; and (3) those who have completed some education (the higher, the better) had the highest chances. Speaking at least one foreign language, belonging to the majority ethnicity and living in Budapest was also advantageous. To leave the middle-­income category to move upwards is also driven by cultural variables and by education (but only at higher levels). Running one’s own business also matters but less so than in the movements from the bottom to the middle. Having a lower initial education or acquiring less education in the period observed means a higher risk of falling out of the middle category. Also, being Roma represents a higher risk than belonging to the majority ethnic group. Finally, the risk of falling out of the top in 1992 was higher for: (1) lower educated; (2) people with unemployment spells in the period; and (3) those who were already over 55 years of age in 1992. The same factors (education and cultural capital) played the biggest role in shaping material-­income status in 1992 and 2007. In addition, the movement between various categories was also driven to a very large extent by initial education levels and further studies. In the period since regime change, property-­income status has become more and more connected to the most important socio-­demographic variables. This shows a decline in the inconsistency in status that was to be found in the 1980s, and the emergence of a stronger hierarchical inequality system. Level of education plays a larger role than ever in explaining property and income differences.

5.  CONCLUSION AND POLICY ISSUES The middle classes in Hungary have traditionally been weak. This was partly a cause, partly a consequence of major socio-­structural and political developments (the authoritarian trend prior to the Second World War, the communist regime and the transition after 1989). We have looked at the various factors that lie behind the historical weaknesses of the Hungarian middle classes, as well as the social, economic and political developments as a result of which the middle classes remain weak even 25 years after the political changes in 1989–90.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 315

28/09/2016 08:24

316

Europe’s disappearing middle class?

Table 7.7 Factors of upward and downward mobility: odds ratios for category outflows and category inflows between quantiles of individuals, based on a combined material/financial index, Hungary, 1992–2007 Upward mobility Up from the bottom 25% Vocational school Secondary school Tertiary education County capital Budapest Not Roma Medium level cultural capital* High level cultural capital* Entrepreneurship** Finished school after 1992   (2–5 pts) Finished school after 1992 (6+) No. of workplaces, 1992–2007  (2–6) Language (yes) Downward mobility Down from the middle 50%

Primary school Roma Finished school after 1992 (1pt)

Up from the middle 50% 2.2 4.2 4.2 3.9 3.7 1.5 4.3 11.6 7.2 2.2

Secondary school Tertiary education

1.5 2.5

Budapest

2.7

High level cultural capital* Entrepreneurship **

2.5 1.5

4.2 Finished school after 1992 (6+) 3.0 No. workplaces 1992–2007  (2–6) 2.3 Language (yes)

1.8 1.5 1.7

Down from the top 25% Age 55+ in 1992 Not (yet) in school in 1992 1.7 Primary school 5.1 Village 1.5 Finished school after 1992   (1 pt)*** Finished school after 1992  (2–5 pts)*** Unemployment spells 92+   (at least 1) Unemployment spells 92+   (at least 2) Number of workplaces 92+ (0)

1.7 1.6 2.2 1.8 2.3 1.7 1.5 2.9 1.9

Notes: Odds ratios with smaller than 1.5 values are not shown. All variables as of 1992. * Index created from combined parental education. ** Index based on personal entrepreneurial record and willingness to become entrepreneur in 1992. For more see Kolosi and Tóth (2008a).

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 316

28/09/2016 08:24



Is Hungary still in search of its middle class? ­317

Highly unequal and fairly backward agriculture-­dominated social structures prior to the Second World War and the ‘decapitation’ of various segments of the elite (by the Nazis and the Communists) in the middle of the last century are the historical reasons. Socialism brought a high level of social mobility, resulting in a very high level of status inconsistency (measured by economic, housing, cultural and political dimensions) in the middle classes. Most developments in the transition tended towards weaker rather than stronger middle classes. Attempts (via privatization) to develop a Hungarian bourgeoisie in this seriously undercapitalized country were weak, while social and economic policies resulted in a low labour market equilibrium (low employment, high inactivity, low job supply). Public sector employment (mostly due to postponed or never initiated reforms in health care and education) remained sizeable, but wages in traditional middle-­class jobs remained very low by international standards. After the collapse of the socialist system the trade unions were unable to strengthen their position and play a significant role in the world of work. Their capacity to defend the interests of the labour market insiders (a major potential source for the middle classes) remained weak. Polarization of income distribution continued, with the additional effect of hollowing out the middle. In addition, because the country was unable to fully utilize its growth potential (very promising at the beginning of the transition but largely unrealized), GDP convergence with richer EU countries was slow and even negative in some periods. Altogether this means that it is not simply relative social status that explains the weak middle classes; Hungary’s overall low level of economic development makes even the ‘real’ middle class weak in terms of economic security, compared with their peers in not only the EU15, but also in neighbouring countries. The two case studies analysed how labour market selection processes can change relative social status, with the most drastic changes being due to age and education. The transition has not only introduced a skills bias via the massive high-­technology FDI influx during the 1990s, but the selection process was very much age biased. Upward intra-­generational mobility was driven by the human and cultural capital of people at the right age to be able to adjust their life strategies to cope with the challenges of major labour market adjustments. Categorization based on income classes creates an image of society in which there are symmetrical relationships between the bottom, the middle and the top. However, deeper analysis has shown that the bottom thirds of the income distribution cannot really be classified as two-­ middle. Rather, deprivation analyses, consumption categorizations and multidimensional social stratification analyses have shown that what is

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 317

28/09/2016 08:24

318

Europe’s disappearing middle class?

traditionally called the ‘middle class’ should be sought somewhere in the upper third of Hungarian society. This share of the ‘real’ middle class is certainly higher than it was 100 years ago, but much lower than it should be, especially with a view to establishing a stable and balanced social structure in Hungary.

NOTES  1. Thanks for comments to earlier versions are due to Anna B. Kis, Réka Branyiczki, János Köllő, Mihály Laki, Péter Róbert, Annamaria Simonazzi and Daniel Vaughan-­ Whitehead. All errors are my own.   2. Below 60 per cent of the median can (corresponding to the Eurostat definition and academic conventions) be named ‘income poor’, while people living in households with above 200 per cent of the median can be considered ‘well-­off’.   3. The meaning of ‘middle classes’ in sociological terms is broader than this, of course. Atkinson and Brandolini (2013) call for the integration of occupation and also wealth into the definition of the middle class. Sociological accounts take a broader look at class and social status (Goldthorpe 2012). Policy-­making focuses more on living standards issues and on consumption prospects (see, for example, White House Task Force 2010).   4. An even broader definition would classify assets that are necessary for membership of the middle class into three categories: in addition to decent incomes, a stable middle-­ class position assumes sufficient resources originating from human capital (skills, education and occupation), from network capital (which can be mobilized in hard times) and physical capital (wealth, preferably in forms that can be mobilized if needed). This would, however, go beyond the scope of this chapter.   5. The Tárki Household Monitor Survey covers 2000 households in each wave. The series goes back to 1992 (the first six waves were an annual panel, then regular cross-­sections every second year; for more on this, see Szivós and Tóth 2015).   6. According to Eurostat, material deprivation is defined as the inability to pay for at least three items on a list of basic expenditure items. The inability to pay for at least four of the listed nine items defines severe material deprivation.   7. State employment in manufacturing, for example, reached 84 per cent as early as the beginning of 1948 (Berend 1979: 80).   8. For the full story of the Hungarian transition from the perspective of inequality, see Fábián et al. (2014).   9. Using the German terms for differentiating between ownership classes and skills and education classes, we can speak of ‘Besitzbürgertum’, busy in these years with accumulating capital in a low capital context, on one hand, and ‘Bildungsbürgertum’, active in setting the rules in politics and the economy but protecting their interests in science, education and health care (Kolosi and Tóth 2012a). 10. The steady growth of entrepreneurial activity (as measured by the number of newly formed enterprises) was, especially in the 1990s, to some extent an artefact. Many of these were ‘forced entrepreneurs’ (in reality, they became self-­employed to minimize tax and social security contributions). 11. In fact, the transition could also be described as a way of removing microeconomic inefficiencies (overemployment at firm level) by pushing the burden onto the state. 12. This volatility also highlights the political exposure and vulnerability of many occupations that otherwise would be considered typical middle-­class jobs. Given that the wage sensitivity of public sector employment is especially high in Hungary, electoral spending cycles may well be behind this employment rollercoaster. 13. According to a study by Köllő (2008), the immediate negative effect of doubling the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 318

28/09/2016 08:24



14. 15. 16. 17. 18. 19. 20.

Is Hungary still in search of its middle class? ­319 minimum wage contributed to reducing wage inequality, but also created some employment disincentives in the small-­firm sector and adversely affected the probability of low-­wage workers losing their jobs and of finding another job. These effects appear to be stronger in low-­wage segments of the market and in depressed regions, where the minimum wage bites deeper into the wage distribution. However, these ups and downs cannot be linked systematically to particular parties. In fact, the extreme low in 1996 and the extreme high in 2004 happened under socialist governments and the 2012 extreme low was under the right-­wing FIDESZ government. Database on the Institutional Characteristics of Trade Unions, Wage Setting, State Intervention and Social pacts in 51 countries between 1960 and 2014, accessed 1 March 2016 at http://www.uva-­aias.net/208. For a further, concise analysis on the development of the trade union movement, see participation.eu/ the ETUI website: accessed 1 March 2016 at http://www.worker-­ National-­Industrial-­Relations/Countries/Hungary/Trade-­Unions. Accessed 1 March 2016 at https://stats.oecd.org/. An analysis of these in the context of the European Social Model can be found in Scharle and Szikra (2014). For more on the survey: TARKI (2008), Kolosi and Tóth (2008a, 2008b). To illustrate, we present how the value 2.2 in the upper left cells of Table 7.7 was calculated. The share of those with vocational training among those who remained in the bottom 25 per cent throughout the period (‘stayers’) is 7.8 per cent (not shown in the table); their share among those who were able to rise from the bottom quartile during the period (‘movers’) is 17.4 per cent. Dividing the movers by the stayers results in 2.2.

REFERENCES Andorka, R. (1982), A társadalmi mobilitás változásai Magyarországon, Budapest: Gondolat. Andorka, R. and I. Harcsa (1988), ‘Modernization in Hungary in the long and short run as measured by social indicators’, Sociological Working Papers, (former) Karl Marx University of Economic Sciences, Department of Sociology, Budapest. Andorka, R., Z. Ferge and I.G. Tóth (1997), ‘Is Hungary really the least unequal? (A discussion of data on income inequalities and poverty in Central and East European Countries)’, Journal of Russian and East European Finance and Trade, 33 (6), 67–94. Atkinson, A.B. and A. Brandolini (2013), ‘On the identification of the “middle class”’, in J. Gornick and M. Jantti (eds), Income Inequality: Economic Disparities and the Middle Class in Affluent Countries, Palo Alto, CA: Stanford University Press, pp. 77–100. Atkinson, A. and J. Micklewright (1992), Economic Transformation in Eastern Europe and the Distribution of Income, Cambridge: Cambridge University Press. Berend T. I. (1979), A szocialista gazdaság fejlődése Magyarországon 1945–1975, Budapest: Kossuth. Bernát, A. and L. Kőszeghy (2011), Managing Household Debts: Hungarian Country Report, prepared for Eurofound, Dublin, accessed 1 March 2016 at http://issuu.com/ tarki_research/docs/managing_household_debt_hungarian_country_report/3 Donáth, F. (1977), Reform és forradalom. A magyar mezőgazdaság strukturális átalakulása 1945–1975, Budapest: Akadémiai. Erdei, F. (1980), A magyar társadalomról, Budapest: Akadémiai.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 319

28/09/2016 08:24

320

Europe’s disappearing middle class?

Fábián Z. (2015), ‘Társadalmi rétegek, fogyasztási státuscsoportok magyarországon’, in P. Szivós and I.G. Tóth (eds), Jól nézünk ki (. . .?!), Háztartások helyzete a válság után. Tárki Monitor jelentések 2014, Budapest: TÁRKI, pp. 128–42. Fábián, Z., A. Gábos, M. Kopasz, M. Medgyesi, P. Szivós and I.G. Tóth (2014), ‘Hungary: a country caught in its own trap’, in B. Nolan et al. (eds), Changing Inequalities and Societal Impacts in Rich Countries: Thirty Countries’ Experiences, Oxford: Oxford University Press, pp. 322–45. Fábián, Z., T. Kolosi and P. Róbert (2000), ‘Fogyasztás és életstílus’, in T. Kolosi, I.G. Tóth and G. Vukovich (eds), Társadalmi riport 2000, Budapest: TÁRKI, pp. 225–59. Fábián, Z., P. Róbert and P. Szívós (1998), ‘Anyagi-­jóléti státuscsoportok társadalmi miliői’, in T. Kolosi, I.G. Tóth and G. Vukovich (eds), Társadalmi riport 1998, Budapest: TÁRKI, pp. 72–91. Fazekas, K. and L. Neumann (eds) (2014), The Hungarian Labour Market 2014, Budapest: Research Centre for Economic and Regional Studies, Hungarian Academy of Sciences & National Employment Non-­profit Public Company Ltd. Ferge, Z. (1969), Társadalmunk rétegződése, Budapest: KJK. Ferge, Z. (1986), Fejezetek a magyar szegénypolitika, Budapest: Magvető Kiadó. Goldthorpe, J.H. (2012), ‘Back to class and status: or why a sociological view of social inequality should be reasserted’, Revista Espanola de Investigaciones Sociologicas, 137 (1), 201–16. Gyáni G. and G. Kövér (2003), Magyarország társadalomtörténete, Budapest: Osiris. Harcsa, I. (2008), ‘Magyarország a jelzőszámok tükrében’, in T. Kolosi, I.G. Tóth and G. Vukovich (eds), Társadalmi riport 2008, Budapest: TÁRKI, pp. 471–98. Kertesi, G. and J. Köllő (2002), ‘Economic transformation and the revaluation of human capital – Hungary, 1986–1999’, in A. de Grip, J. van Loo and K. Mayhew (eds), The Economics of Skills Obsolescence, vol. 21, Oxford: Elsevier Science, pp. 235–73. Köllő, J. (2008), ‘Hungary – the consequences of doubling the minimum wage’, in D. Vaughan-­Whitehead (ed.), The Minimum Wage Revisited in the Enlarged EU, Cheltenham, UK and Northampton, MA, USA: Edward Elgar and Geneva: ILO, pp. 239–66. Köllő J. (2009), A pálya szélén. Iskolázatlan munkanélküliek a posztszocialista gazdaságban, Budapest: Osiris. Köllő, J. (2011), ‘Employment, unemployment and wages in the first year of the crisis’, in K. Fazekas and G. Molnár (eds), The Hungarian Labour Market 2011. Review and Analysis, Budapest: Institute of Economics, IE HAS – National Employment Foundation, pp. 43–68. Köllő, J. (2014), ‘What do we know about public sector employment?’, in K. Fazekas and L. Neumann (eds), The Hungarian Labour Market 2014, Budapest: Research Centre for Economic and Regional Studies, Hungarian Academy of Sciences & National Employment Non-­profit Public Company Ltd, pp. 43–55. Kolosi, T. (1987), Tagolt társadalom, Budapest: Gondolat. Kolosi, T. (2000), A terhes babapiskóta. A rendszerváltás társadalomszerkezete, Budapest: Osiris. Kolosi, T. and T. Keller (2010), ‘Kikristályosodó társadalomszerkezet’, in T. Kolosi and I.G. Tóth (eds), Társadalmi riport 2010, Budapest: TÁRKI, pp. 105–38. Kolosi, T. and K. Pósch (2014), ‘Osztályok és társadalomkép’, in T. Kolosi and I. G. Tóth (eds), Társadalmi riport 2014, Budapest: TÁRKI, pp. 139–56.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 320

28/09/2016 08:24



Is Hungary still in search of its middle class? ­321

Kolosi, T. and I. Szelényi (2010), Hogyan legyünk milliárdosok? – avagy A neoliberális etika és a posztkommunista kapitalizmus szelleme, Budapest: Corvina. Kolosi, T. and I.G. Tóth (2008a), ‘A rendszerváltás nyertesei és vesztesei – generációs oldalnézetből (10 állítás a gazdasági átalakulás társadalmi hatásairól)’, in T. Kolosi and I.G. Tóth (eds), Újratervezés. Életutak és alkalmazkodás a rendszerváltozás évtizedeiben, Kutatási jelentés a Háztartások életút vizsgálata alapján, Budapest: TÁRKI, pp. 15–45. Kolosi, T. and I.G. Tóth (eds), (2008b), Újratervezés. Életutak és alkalmazkodás a rendszerváltozás évtizedeiben, Kutatási jelentés a Háztartások életút vizsgálata alapján, Budapest: TÁRKI. Kolosi, T. and I.G. Tóth (2012a), ‘Előszó’, in T. Kolosi and I.G. Tóth (eds), Társadalmi riport 2012, Budapest: TÁRKI, pp. 7–15. Kolosi, T. and I.G. Tóth (eds) (2012b), Társadalmi riport 2012, Budapest: TÁRKI. Kolosi, T. and I.G. Tóth (eds) (2014), Társadalmi riport 2014, Budapest: TÁRKI. Kornai, J. (1980), The Economics of Shortage, Amsterdam: North Holland. Kornai, J. (1992a), ‘The post socialist transition and the state: reflections in the light of Hungarian fiscal problems’, American Economic Review, 82 (2), 1–21. Kornai, J. (1992b), The Socialist System: The Political Economy of Communism, Princeton, NJ: Princeton University Press. Kornai J. (1994), ‘Transformational recession. The main causes’, Journal of Comparative Economics, 19 (1), 39–63. Központi Statisztikai Hivatal (Central Statistical Office) (KSH) (1990), Jövedelemeloszlás Magyarországon. Az 1988. évi felmérés adatai, Budapest: KSH. Központi Statisztikai Hivatal (Central Statistical Office) (KSH) (2014), Magyar Statisztikai Évkönyv, Budapest: KSH. Laki, M. and J. Szalai (2004), Vállalkozók vagy polgárok? A nagyvállalkozók gazdasági és társadalmi helyzetének ambivalenciái az ezredforduló Magyarországán, Budapest: Osiris. Laki, M. and J. Szalai (2013), Tíz évvel később: magyar nagyvállalkozók európai környezetben, Budapest: Közgazdasági Szemle Alapítvány. Laki, M., B. Nacsa and L. Neumann (2012), Az átalakuló munka világa. Az új Munka Törvénykönyve bevezetésének kezdeti tapasztalatai. Szakszervezeti Kiskönyvtár 33, Budapest: Liga Szakszervezetek 2012 December. Mihályi, P. (2010), A magyar privatizáció enciklopédiája, 1–2. kötet, Pannon Egyetemi Kiadó – MTA Közgazdságtudományi Intézet. Oblath, G. (2014), ‘Gazdasági átalakuklás, nekilendülés és elakadás. Magyarország makrogazdasági konvergenciája az európai unió fejlett térségéhez az 1990-­es évek elejétől 2013-­ig’, in T. Kolosi and I.G. Tóth (eds), Társadalmi riport 2014, Budapest: TÁRKI, pp. 21–50. Organisation for Economic Co-­ operation and Development (OECD) (2008), Growing Unequal? Income Distribution and Poverty in OECD Countries, Paris: OECD. Organisation for Economic Co-­ operation and Development (OECD) (2011), Divided We Stand. Why Inequality Keeps Rising? Paris: OECD. operation and Development (OECD) (2013), Organisation for Economic Co-­ Health at a Glance. OECD Indicators. Paris: OECD. Organisation for Economic Co-­ operation and Development (OECD) (2014), Education at a Glance 2014, Paris: OECD. Organisation for Economic Co-­ operation and Development (OECD) (2015a), Government at a Glance 2015, Paris: OECD.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 321

28/09/2016 08:24

322

Europe’s disappearing middle class?

Organisation for Economic Co-­ operation and Development (OECD) (2015b), Government at a Glance: How Hungary Compares, Paris: OECD. Organisation for Economic Co-­operation and Development (OECD) (2015c), In It Together. Why Less Inequality Benefits All, Paris: OECD Pető, I. and S. Szakács (1985), A hazai gazdaság négy évtizedének története ­1945–1985, Budapest: Közgazdasági és Jogi Könyvkiadó. Róbert, P. and A. Balogh (2008), ‘Foglalkozási turbulencia Magyarországon, ­1992–2007’, in T. Kolosi and I.G. Tóth (eds), Újratervezés – Életutak és alkalmazkodás a rendszerváltás évtizedeiben, kutatási jelentés ‘A háztartások életút vizsgálata (HÉV)’ alapján, Budapest: TÁRKI. Scharle, Á. and D. Szikra (2014), ‘Recent changes moving Hungary away from the European Social Model’, in D. Vaughan-­Whitehead (ed.), The European Social Model in Times of Economic Crisis and Austerity Policies, Geneva: ILO, pp. 219–54. Szalai, J. (1992), ‘Az érdekviszonyok történeti és jelenkori alakulása a társadalombiztosításban’, Szociológiai Szemle, no. 3. Szivós P. and I.G. Tóth (eds) (2008), Köz, teher, eloszlás, TÁRKI Monitor jelentések, Budapest: TÁRKI. Szivós, P. and I. G. Tóth (eds), (2015), Jól nézünk ki (. . .?!), Háztartások helyzete a válság után. Tárki Monitor jelentések 2014, Budapest: TÁRKI. Szivós, P. (2014), ‘A magyar háztartások jövedelme és fogyasztása nemzetközi összehasonlításban: Felzárkózás vagy leszakadás?’, in T. Kolosi and I.G. Tóth (eds), Társadalmi riport 2014, Budapest: TÁRKI, pp. 51–62. TÁRKI (2008), Lifepath Analysis of Hungarian Households, final project report, Budapest: TÁRKI. Tóth, I.G. (2005), A gazdasági rendszerváltástól az uniós csatlakozásig, Budapest: Andorka Rudolf Társadalomtudományi Társaság – Századvég Kiadó, Budapest. Tóth, I.G. (2008), ‘Economic transition and income distribution in Hungary, 1987–2001’, in J.M. Fanelli and L. Squire (eds), Economic Reform in Developing Countries: Reach, Range, Reason, Cheltenham, UK and Northampton, MA, USA: Edward Elgar, pp. 3–34. Tóth, I.G. (2014), ‘Revisiting grand narratives of growing income inequalities: lessons from 30 country studies’, in B. Nolan et al. (eds), Changing Inequalities and Societal Impacts in Rich Countries: Thirty Countries’ Experiences, Oxford: Oxford University Press, pp. 11–47. Tóth, I.G. (2015), ‘Jövedelem-­eloszlás 2009–2014 között’, in P. Szivós and I.G. Tóth (eds), Jól nézünk ki (. . .?!), Háztartások helyzete a válság után, Tárki Monitor jelentések 2014, Budapest: TÁRKI, pp. 11–44. Tóth, I.G. and M. Medgyesi (2011), ‘Income distribution and living difficulties in the midst of consolidation programmes and crises in Hungary’, in K. Fazekas and G. Molnár (eds), The Hungarian Labour Market 2011. Review and Analysis, Budapest: Institute of Economics, IE HAS – National Employment Foundation, pp. 170–86. Ward, T., O. Lelkes, H. Sutherland and I.G. Tóth (eds), (2009), European Inequalities. Social Inclusion and Income Distribution in the European Union, Budapest: TÁRKI. White House Task Force (2010), Annual Report of the White House Task Force on the Middle Class, February, Washington, DC.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 322

28/09/2016 08:24

8. Middle incomes in boom and bust: The Irish experience Bertrand Maître and Brian Nolan 1. INTRODUCTION In investigating the extent and nature of a ‘squeezed middle’, Ireland’s remarkable macroeconomic fluctuations over the past two decades make it a particularly interesting case study. Ireland had some of the fastest rates of economic growth in the Organisation for Economic Co-­operation and Development (OECD) from the mid-­1990s, followed by the most precipitous decline in national output of any OECD country at the onset of the Great Recession. The scale and nature of the economic boom and subsequent ‘bust’, accompanied by sharp swings in taxation and public spending, make Ireland a clear outlier in terms of macroeconomic fluctuations, providing a remarkable context in which to explore what happened to the middle-­income groups that are the focus of this book. The chapter begins by describing key trends and developments in the macro-­economy and in the labour market over the period from 1995 onwards, with particular attention to both the institutional context and key changes in the world of work. We then turn to how middle-­income groups have fared in terms of the evolution of average disposable household incomes. To help understand the patterns observed, we then look at the profile of the different income groups in terms of age, education and employment status, and how this changed over the period. The role of redistribution via cash social transfers and direct taxes, particularly over the period of recession and ‘austerity’, is then examined. We next go beyond income to investigate what other indicators of living ­standards – namely, measures of deprivation and financial stress – reveal. That analysis points to the importance of debt and the housing market in boom and bust, which constitutes the first of our case studies. Changes in the occupational structure and the relative fortunes of different occupational groups are at the core of trends in the world of work and how the middle has fared, and this is the focus of our second case study. Finally, the conclusions are brought together and their implications explored. 323

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 323

28/09/2016 08:24

324

Europe’s disappearing middle class?

2.  THE MACROECONOMIC CONTEXT After substantial economic growth in the 1970s, followed by stagnation in the 1980s, Ireland experienced the so-­called ‘Celtic Tiger’ boom from 1994, with average annual increases in real gross national product (GNP) as high as 7 per cent up to 2000.1 Growth continued mostly at 4–6 per cent per annum up to 2007, at which point Ireland’s GNP per capita was among the highest in the European Union (EU). However, the latter part of Ireland’s economic boom was accompanied by an unbridled property boom and a very rapid increase in levels of household debt. Aggregate borrowing of the household sector rose very rapidly, fuelled by low interest rates as Ireland, like other ‘periphery’ members of the Eurozone, saw substantial inflows of capital from its centre. The tax base became highly unbalanced, with income tax being cut on the back of revenue from stamp duties and other taxes on property development and sales. The construction sector grew very rapidly and accounted for 14 per cent of total employment at the peak, while domestic demand replaced exports as the core driver of economic growth. The global financial crisis and the bursting of the domestic property bubble then led to an unprecedented contraction in national output and income from 2008–09. The economic crisis had a more negative immediate impact on national output in Ireland than in any other OECD country: by 2010 GNP per head in nominal terms had fallen by close to one-­fifth compared with 2007, and in real terms was back to levels seen a decade earlier. This went together with a bursting of the property bubble, a collapse in asset values, a banking crisis of unprecedented proportions and a ballooning fiscal deficit. The general government balance went from a position of surplus to a deficit of 7 per cent of national income in 2008 and close to 14 per cent in 2009, while support for the banking system drove the deficit to over 30 per cent in 2010. The Irish state rescued the banking system by introducing a blanket guarantee for (virtually) all creditors in late 2008, and subsequently injected very substantial new capital as the scale of the banks’ losses on property-­related loans emerged. The government debt to national income ratio, having fallen to 25 per cent by 2007, soared to 90 per cent by 2010 and reached 120 per cent by 2012, with support for the banks accounting for about one-­third of that total. By 2010, despite substantial increases in taxation and expenditure cuts, the Irish government was no longer able to borrow on international financial markets at acceptable rates and had to avail itself of a ‘bailout’ by the EU and the International Monetary Fund (IMF). The extremely demanding objective associated with the bailout was to bring the deficit down to 3 per cent, in line with the EU’s Stability and

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 324

28/09/2016 08:24



Middle incomes in boom and bust: The Irish experience ­325

Growth Pact. Fiscal adjustment involved both very substantial reductions in current and capital expenditure and increases in taxation, with cuts in expenditure accounting for about two-­thirds of the total, among the largest budgetary adjustments seen in the OECD in recent years. The effects of domestic ‘austerity’, combined with weak external demand meant that the economy ‘flat-­lined’ with no increase in GNP until 2012, with some export growth being offset by weak domestic demand. The beginnings of an economic recovery were seen by 2012–13 and GNP growth then picked up strongly in 2014 and 2015, driven by strong export growth as competitiveness improved, and with foreign direct investment from the United States playing a key role, as it had throughout the boom. The government’s finances were also substantially improved by then, with the fiscal deficit on target to meet EU Stability and Growth Pact limits.

3. 

 HANGES IN THE LABOUR MARKET AND THE C WORLD OF WORK

Against the backdrop of macroeconomic boom and bust, Ireland saw dramatic changes in the labour market, in terms of patterns of employment and unemployment, earnings, occupational structures and job ‘quality’, and institutional context. This can be seen most immediately in the number of people employed, which expanded in an unprecedented fashion from 1.2 million in 1994 to 1.7 million by 2000 and 2.1 million by 2008. Unemployment declined very rapidly, from 16 per cent in 1994 to 4 per cent by 2000, remaining at that level up to 2008, and the long-­term unemployment rate fell to just over 1 per cent, a level which would have been regarded as unachievable before the onset of the boom. The employment rate for women rose from 40 per cent to 60 per cent, and in-­migration was also important in the expansion of the workforce for the first time. A wave of return migration by those who had left Ireland in the 1980s was followed by substantial numbers arriving from other EU countries, particularly after the 2004 enlargement of the EU when a substantial number came from Poland and the Baltic countries in particular, with the annual net inflow reaching 70 000. Employment growth spanned most sectors of the economy, with both public sector numbers and construction seeing very marked increases in the latter part of the boom. Employment protection was enhanced over the period as a whole, and while part-­time employment continued to be attractive for some, full-­time work was available to most of those seeking it and the strength of labour demand also empowered workers in bargaining on contract type and working conditions. In terms of earnings, the boom was not accompanied by the pronounced

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 325

28/09/2016 08:24

326

Europe’s disappearing middle class?

increase in earnings inequality and widening gap between high versus low levels of education seen in some other OECD countries in recent decades, notably the United States and the United Kingdom. Strong demand for low-­skilled employees boosted their returns, while returns for the highly educated were limited by strong increases in supply via an increasing flow of new graduates, together with in-­migration including skilled and experienced Irish people returning from abroad (Barrett et al. 2002; McGuinness et al. 2008; Voitchovsky et al. 2013). This meant that, while mean/median hourly earnings rose by one-­third in real terms, there was a considerable decline in earnings dispersion from the mid-­1990s to the height of the boom, with the ratio of hourly gross earnings for the top decile to the bottom decile falling from 4.8 to 4. A major institutional change in the labour market during the boom period was the introduction of a national minimum wage in 2001, at a level that appeared to affect about 5 per cent of employees (see Voitchovsky et al. 2013). This followed the introduction of a similar national minimum in the United Kingdom, though with changes in the level of the minimum over time determined by the government following consultations with employer and union representative bodies, rather than relying on the recommendations of an independent low pay commission, as in the United Kingdom. Up to the crisis the level of the minimum was increased over time at irregular intervals broadly in line with median earnings and studies suggest that it effectively underpinned earnings towards the bottom wage and contributed to declining dispersion (Nolan et al. 2010). This was particularly important for women, who are employed disproportionately in low-­paid services, and being expressed in hourly terms the minimum also has particular salience for part-­time workers. An even more central institutional feature of the boom period was that wage bargaining operated at the national level through a ‘social partnership’ process, in which the government, employers, unions and the ­voluntary/community sector concluded agreements on wage levels in both the private and public sectors, and on a wide range of economic and social policies. This highly structured form of social dialogue had been in place in Ireland since the late 1980s, arising as a response to the demoralizing period of economic stagnation experienced from the early 1980s. The contribution of these agreements to economic growth and the extent to which they represented successful social corporatism has been a focus for significant debate, narrowly in terms of their impact on wages and more broadly in terms of the management of the boom. As far as wages are concerned, Baccaro and Simoni (2002), for example, argued that centralized wage bargaining contributed to the ‘Celtic Tiger’ phenomenon by linking wage increases in the dynamic multinational

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 326

28/09/2016 08:24



Middle incomes in boom and bust: The Irish experience ­327

companies sector to wage and productivity increases in the much more sluggish domestic sector of the economy and, in so doing, considerably increased the competitiveness of foreign multinational companies, a key driver of Irish growth. This was contested by econometric studies of wage trends, and Honohan and Walsh’s (2002: 33) authoritative review of the boom period up to the early 2000s is nuanced on this: ‘Despite the inconclusive econometric results, most observers regard the coincidence of timing of the reversal of the deteriorating trend in competitiveness with the new approach to pay bargaining as suggestive that the latter did pay dividends’. They certainly see wage restraint as a hallmark of the boom contributing to enhanced competitiveness, attributing that to high levels of unemployment at the outset as well as union restraint exercised in the centralized pay agreements. In that context they also note that union power was declining in much of the economy as the proportion of workers in unions fell markedly from the late 1970s, a decline that continued through the boom. As Regan (2012) also notes, employers welcomed the industrial peace and the saving of time and energy at the level of the firm achieved by the centralization and coordination of wage bargaining, with many multinational firms comfortable with the corporatist approach at the national level, while having a union-­free workplace. The centrally bargained increases generally set a floor, with more profitable firms – particularly but not only in the multinational sector – often giving higher increases. Honohan and Walsh (2002) also emphasize that a key feature of the collective agreements in this early part of the boom was the lowering of the tax burden on employees, crucial to the moderation of nominal wage claims. This was, as they put it, ‘a somewhat Faustian bargain’ (Honohan and Walsh 2002: 34): the time would come when the government would have little more to offer in this respect as demands for more spending on public services, notably health and education, became pressing, as reflected in the social partnership agreements themselves. This included demands for pay rises from public sector workers, who received substantial increases from 2002 onwards over and above those agreed for the private sector arising from a ‘benchmarking’ process that sought to compare wage levels in the public and private sectors. This was contested at the time, but proved particularly significant as the backdrop to the response to the crisis, as discussed shortly, as well as in assessments of the overall role of social dialogue in what soon came to be seen as the mismanagement of the boom. Social partnership was seen by some as important in increasing the capacity of the political system to problem-­solve (O’Donnell 2008) and in enabling the Irish state to adapt to globalization (O’Riain 2008); critics saw it as reinforcing the tendency of the political system to seek to satisfy as broad a range of interests as possible while

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 327

28/09/2016 08:24

328

Europe’s disappearing middle class?

16 14 12 10 8

13.9 12.3

14.714.7

12.0

11.7

13.1 11.3

9.9 7.5

6

5.7

4

6.4 4.3 3.9 4.5 4.6 4.5 4.4 4.5 4.7

2

19 95 19 96 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12 20 13 20 14

0

Figure 8.1  Unemployment rate, Ireland, 1995–2014 (percentage) avoiding hard choices, and also questioned the way it structured who was included in social dialogue. The economic crisis from 2008 had an immediate and deep impact on the labour market, in terms of employment, earnings and wage-­setting. As Figure 8.1 shows, the unemployment rate rose sharply to almost 15  per  cent. The decline in employment was very heavily concentrated among young men: unemployment rates for those aged 20–24 rose from 8 per cent to 32 per cent and for those aged 25–34 from 5 per cent to almost 20 per cent. Net emigration also returned, both of Irish citizens and recent arrivals from central and eastern Europe, with the net outflow reaching 60 000 in the year to April 2011. The proportion of working-­age individuals in ‘jobless’ households (employing the measure incorporated in the EU’s Poverty and Social Exclusion target) increased to 22 per cent, among the highest in the EU, with jobless adults less likely than elsewhere to live with other adults who are in work (Watson et al. 2012). Earnings in the hardest-­ hit sectors, such as construction, fell sharply, while cuts in public sector wage levels as well as numbers were imposed (as discussed in more detail below). The number in part-­time employment rose and the proportion of them involuntarily in that position rose significantly, and other forms of precarious employment such as zero-­hours contracts became much more prevalent. There is also evidence that work pressure increased for those remaining in employment, associated with staff reductions and company reorganization, as well as job insecurity (Russell and McGinnity 2014). The minimum wage was initially cut at the onset of the crisis, but subsequently restored to its previous level after a change in government in 2010.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 328

28/09/2016 08:24



Middle incomes in boom and bust: The Irish experience ­329

As far as the institutional setting for social dialogue is concerned, under that new government centralized wage bargaining also became a casualty of the crisis, with the social partnership process disintegrating as public sector pay and jobs were cut in the face of union opposition. This meant there was no collective negotiation to minimize lay-­offs in return for wage and/or working-­time concessions, as occurred in some other countries with strong social dialogue, although the scale of the fall-­off in economic activity would have militated against this in any case. Instead, reducing public sector pay became a target, with the perception that excessive increases had fuelled government spending in the latter part of the boom being reinforced by studies suggesting that pay levels were in excess of those in the private sector (Kelly et al. 2009). Substantial cuts in pay for public sector employees were implemented as part of the fiscal correction strategy described in detail below, with attempts to bring this about through the social partnership process being unsuccessful and that process collapsing as a consequence. Regan (2013) and Culpepper and Regan (2014) argue that a social pact could not be negotiated at this point because Irish trade unions lacked sufficient deterrent power in the labour market, and the voluntarist nature of the process (with no legal requirement to implement collectively negotiated wage increases or extend them to non-­union employees) enabled employers and government to walk away from social partnership with limited repercussions. Nonetheless, the government did then negotiate a sectoral agreement with public sector unions (the ‘Croke Park Agreement’) in 2010 guaranteeing no further pay cuts and that reductions in employee numbers would be through voluntary redundancy/retirement, in return for employee participation in efficiency-­enhancing public sector reform. A further such agreement was subsequently agreed covering 2013–16 (the ‘Haddington Road Agreement’). This limited degree of social dialogue succeeded in maintaining industrial peace through the period of fiscal adjustment/‘austerity’. As economic growth returned in 2013 unemployment also began to fall, at a pace that accelerated in 2014, and the unemployment rate had fallen below 10 per cent by mid-­2015. Average weekly earnings began to recover by 2015, and some restoration of the cuts in public sector pay rates and perhaps an increase in the level of the minimum wage were in prospect at that point, although no restoration of a centralized wage-­bargaining process is in view. Even with sustained growth in employment and earnings, though, dealing with the scarring effects of the recession will be a central challenge for labour market policy for years to come.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 329

28/09/2016 08:24

330

Europe’s disappearing middle class?

4. EVOLUTION OF HOUSEHOLD INCOMES: WAS THE MIDDLE ‘SQUEEZED’? The remarkable increase in national output and employment over the course of Ireland’s Celtic Tiger boom had a dramatic impact on household incomes. As shown in Figure 8.2, median disposable household income (adjusted for household size) in real terms increased very rapidly over the period from 1995 to 2008, more than doubling over that relatively short period. This brought Ireland towards the top of the EU ranking in terms of that indicator of living standards: in 2008 only Austria, Luxembourg, the Netherlands and the United Kingdom were above it in terms of median equivalized disposable income expressed in purchasing power parity (PPP) terms. The crisis then led to major reverses. The initial impact on household incomes was less than that on national output (as also found elsewhere – see for example Jenkins et al. 2013), as automatic stabilizers in the form of increasing social transfers and declining tax revenues buffered disposable income. Self-­employment incomes fell immediately, however, while direct taxes were increased successively and social transfers for those of working age were cut in order to address the fiscal deficit. By 2012 median disposable income had fallen by about 12 per cent from its 2008 peak (see Figure 8.2), at which point it stabilized. While median or average income fluctuated so significantly, the distribution of disposable income among households remained fairly stable. 260 240

228

220 200 180

165

160

138 137

140

170

175

185

234

223

205 194

217 215

207 206

145

115

120

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

100

Figure 8.2 Median household disposable income (equivalized) in PPP terms, Ireland, 1997–2013

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 330

28/09/2016 08:24



Middle incomes in boom and bust: The Irish experience ­331

29.9

28.5 29.1 29 27.3 28.1

24.1

2008 2012

6.2

10

8.3 8.4 8.6

15

2000

10.9

20

1995 2005

16.1 13.2 14.9 16.2 17

25

20.4

30

23.8 22 18.8 16.3

35

27.7 25.8 29.2

The Gini coefficient, the most widely used summary indicator of income inequality, moved within a relatively narrow band, rising modestly from 1995 to 2005, then falling so that by 2012 it was back at close to its initial level. In terms of income shares across the deciles of the distribution, the top 10 per cent saw an increase in its share in the boom years, falling back in the bust. The bottom two deciles, by contrast, saw some decline in share up to 2008 and from then to 2012. The share of total income going to deciles 4, 5 and 6 rose by 1 per cent overall between 1995 and 2008, and changed little between then and 2012; the combined share of deciles 3–7 rose by 1 per cent of total income between 1995 and 2008, and by slightly more over the entire period from 1995 to 2012. Thus, if seen in terms of the share of disposable income going to deciles around the middle, there was no ‘squeeze’ of the middle in the Irish case over this period, either in the boom or the bust. An alternative and helpful way to look at what happened to the middle, as outlined in earlier chapters, is to define income categories with reference to the median of the equivalized household disposable income distribution. Employing the categories described there, Figure 8.3 shows how the distribution of persons in terms of their household equivalized disposable income changed over the period. We see that the proportion falling into the middle category (between 80 per cent and 120 per cent of the median) rose from 24 per cent in 1995 to 29 per cent by 2008, while the shares in the adjoining income groups, the lower and upper-­middle categories, were more stable, at about 16 and 28 per cent respectively. The increasing share in the middle was associated with a decline in the proportion in both the top and bottom groups; this reflected the broad-­based nature of the increases in employment and incomes from work throughout the

5 0

y < 60%

60% ≤ y < 80%

80% ≤ y < 120%

120% ≤ y < 200%

y ≥ 200%

Figure 8.3 Distribution of persons by income category, Ireland, 1995–2012

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 331

28/09/2016 08:24

332

Europe’s disappearing middle class?

boom period, reducing the number of working-­age households relying on social welfare and substantially increasing the number of multi-­earner households across the middle ranges. From 2008 to 2012, the proportion in the middle categories rose modestly as the numbers below 60 per cent of the median fell. This decline reflects the sharp fall in the level of this median-­based poverty threshold in the crisis, so despite soaring unemployment and increasing numbers relying on social transfers, more recipients (notably pensioners) were now above that threshold. (Assessed against a more ‘absolute’ income poverty standard held fixed in purchasing power terms at 60 per cent of the pre-­crisis median, income poverty had risen substantially by 2012.) As well as the numbers falling into each of these groups, what happened to their average incomes is also key. Figure 8.4 shows the evolution of mean incomes in real terms (deflated by the Consumer Price Index) for each of these groups over the boom, in the crisis and over the period from 1995 to 2012 as a whole. Very substantial increases for all groups were seen from 1995 up to 2008, slightly greater for the core middle (between 80 per cent and 120 per cent of the median) than for other groups. In the recession incomes fell sharply for all groups, with the largest declines for the bottom income group. Over the entire period from 1995 to 2012, then, middle-­income groups – whether narrowly or broadly defined – fared quite well relative to the top and bottom of the distribution, seeing a greater 160 140

132

126.7

134.1

128.3

121.9

120 100 80

90.7 72.8

88.4

87.8

119.1

80

75.7

60 40 20 0 –20 –40

–23.8 y < 60%

–17.8 60% ≤ y < 80%

% change 1995–2008

–19.5

–17.7

80% ≤ y < 120% 120% ≤ y < 200%

% change 2008–12

–20.8

–17.8

y ≥ 200%

Total

% change 1995–2012

Figure 8.4 Percentage change in mean equivalized household disposable income (in constant prices) by income category, Ireland, 1995–2012

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 332

28/09/2016 08:24



Middle incomes in boom and bust: The Irish experience ­333

overall increase in mean income. Once again this primarily reflects the developments in the labour market already described, namely, the broad-­ based increase in income from work accruing across the broad middle of the income distribution in the boom period, followed by the pervasive impact of the crisis and accompanying ‘austerity’, which are discussed in more detail below.

5.  CHANGING PROFILE OF INCOME GROUPS In comparing these different income groups and how their incomes evolved over time, it helps to look at the way their composition changed in order to bring out the forces driving income change and the centrality of developments in the world of work, as well as in social transfers in that respect. The characteristics and situation of the household reference person (HRP)  – defined as the person responsible for meeting the household’s housing costs or the older of those jointly responsible – are particularly instructive in that regard. The shifts in the age composition of the HRP for middle-­income groups versus others are shown in Figures 8.5a and 8.5b. Over the course of the economic boom from 1995 to 2008, Figure 8.5a shows that the proportion of households in the middle-­income groups with an HRP aged under 35 rose, while Figure 8.5b shows that the proportion aged 65 or over fell. This reflects the strength of growth in incomes from employment over the boom, with pensions also increasing substantially, but not (quite) keeping pace. In the crisis, by contrast, the proportion of households in the bottom two income categories with HRP aged 65 or over fell sharply, with a corresponding increase in that proportion for the middle-­, upper middle-­and upper-­income groups, as those of working age bore the brunt of the effects of falling employment and earnings while pensioners were more protected and so their households moved up the income categories, as discussed in more detail below. The educational profile of the different income groups also changed significantly over the period. A general improvement in educational attainment levels was reflected in declining proportions of HRPs with only intermediate levels of secondary education in all income categories, as shown in Figure 8.6a. For the middle-­income category, for example, the proportion of HRPs with only that level of attainment fell from 60 per cent to 40 per cent over the course of the boom, a striking upgrading in education levels. At the other end of the attainment spectrum there was a marked increase in the proportion of HRPs with third-­level attainment across all the income categories, as Figure 8.6b brings out. For the middle-­income group the proportion with third-­level attainment rose from 8 per cent to 25 per cent, while

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 333

28/09/2016 08:24

334 30 25

Europe’s disappearing middle class? 25 25

27

24 24

20 15 30 10 25 5 20 0 15

14

15 25 25

27

y < 60%

20

17

24 24

15 60% ≤ y < 80%

1995

24 16

1995

24 17 14 80% ≤ y < 120%

17

20 16

17

120% ≤ y < 200%

2008

2012

16 16 2008

14 2012

16 16

14 y ≥ 200%

10 5 30 Figure 8.5a Percentage of households in income class1995 with HRP 2008 aged 2012 24 23 0 25 under 35, Ireland, 1995–2012 20 15 30 10 25 5 20 0 15

y < 60% 18 19

60% ≤ y < 80% 19

24

19

19

y < 60%12

8

23

60% ≤ y < 80%

10

y < 60%

60% ≤ y < 80%

y ≥ 200%

10

13 19959 6

20089

7

2012

16 13 80% ≤ y < 120% 8

120% ≤ y < 200% 10 9

13 y ≥ 200% 9 7

80% ≤ y < 120%

120% ≤ y < 200%

y ≥ 200%

5 0

120% ≤ y < 200%

16

13

12

18

80% ≤ y < 120%

6

Figure 8.5b Percentage of households in income class with HRP aged 65+, Ireland, 1995–2012 for the upper middle group the increase was even greater, from 20 per cent to 45 per cent. This reflects both the sharply increasing attainment levels of young people coming through the Irish education system before and during the boom, and (to a much more limited extent) the increasing proportion of households with an HRP aged under 35 in the middle-­and upper middle-­ income categories in 2008 versus 1995 already noted above. The crisis saw a continuation of the increase in the proportion of HRPs with third-­level education across all the income categories from 2008 to 2012, but strikingly this was now most concentrated at the bottom, among households below 60 per cent of median income. By 2012, 28 per cent of HRPs below that threshold had third-­level qualifications, up from 11 per cent in 2008. This remarkably high figure for households below the conventional relative income poverty threshold reflects the wide-­ranging impact of the recession on employment levels across the distribution, the declining importance of pensioners (with lower education levels on average

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 334

28/09/2016 08:24



Middle incomes in boom and bust: The Irish experience ­335

100 80

81.3 62.2

60

60 80

55.5

38.2

40 20 100 0 80

1995

72.9

81.3 y < 60% 62.2

47.7

2008

2012

59.3 40

33.7

42.7 24.6 18.5 1995

72.9 60% ≤ y < 80% 55.5 47.7

80% ≤ y < 120% 59.3

60% ≤ y < 80% 23.7

36.5 80% ≤ y < 120% 24.9

40

120% ≤ y < 200%

21.9 2008

12.8 10.9 2012

y ≥ 200%

42.7

74.2

38.2 1995 2008 2012 Figure 8.6a  Percentage of households in33.7 income class with HRP having 40 66.5 24.6 70 21.9 18.5 no more than intermediate second level education, Ireland, 55.9 12.8 10.9 60 20 47.2 1995–2012 45.3 50 0 40 30 20 80 10 70 0 60 50 40 30 20 10 0

y < 60% 27.5 11.4 1995 1.1

2008

y < 60%

2.7

11.6 2012

60% ≤ y < 80%

27.5 11.4 1.1 y < 60%

2.7 60% ≤ y < 80%

y ≥ 200%

20.1

7.8

66.5

80% ≤ y < 120% 36.5

23.7 11.6

120% ≤ y < 200%

24.9

55.9 120% ≤45.3 y < 200%

74.2

y ≥ 200% 47.2

20.1

7.8 80% ≤ y < 120%

120% ≤ y < 200%

y ≥ 200%

Figure 8.6b Percentage of households in income class with HRP having third level education, Ireland, 1995–2012 than younger people) among those below the poverty threshold already mentioned, and the very poor employment prospects of young people, especially those coming on the labour market for the first time during the recession, even with third-­level qualifications. These poor employment prospects for graduates drove the resurgence of emigration and would be widely seen as a defining feature of the recession for middle-­and upper-­ income parents, over and above the direct impact on their household income. The labour force status of the household reference person for the various income groups also fluctuated markedly over the period, as one would expect given both the swings in the overall employment rate in boom and bust and the changing composition of the different income groups already noted. Figure 8.7 shows that the proportion of HRPs at work for middle-­income groups was fairly stable over the boom, rising for the group below the relative income poverty threshold. In the crisis, however, the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 335

28/09/2016 08:24

336 90 80

Europe’s disappearing middle class? 77 79 75

1995 %

70

2008 %

60

2012 %

55 45 43 35

40

20

73

64 63

50

30

82 84

22

26 19

10 0 y < 60%

60% ≤ y < 80%

80% ≤ y < 120%

120% ≤ y < 200%

y ≥ 200%

Figure 8.7 Percentage of households in income class with HRP at work, Ireland, 1995–2012 percentage of HRPs at work declined across all the income groups with the exception of the upper middle one. By 2012, only 55 per cent of HRPs in the middle-­income group were at work, while the corresponding figure was about one-­third for households in the lower middle group, and only 20 per cent for those below 60 per cent of median income. The broad-­based nature of the reduction in employment in the recession becomes even clearer when we focus on the total number of adults at work in the household rather than the labour force status of the HRP. Table 8.1 shows the extent to which the average number of persons at work in the household fell for all income groups, with a particularly sharp decline in the proportion of working-­age adults in work in poor and lower middle-­ income households. These changes in numbers in work were central to the declines in income from the market registered over the 2008–12 period by the different income groups. Average income from the market – before the impact of social transfers and direct taxes – fell by about one-­third for the bottom-­and middle-­ income groups, but by even more – 40 per cent – for the lower-­middle group. The changing composition of the different income groups over time – in terms of age, education and labour force status of the HRP – as documented in this section is the result of changes in the overall profile of the population, as education levels rise and employment fluctuates, most importantly, but also, as brought out in the discussion, by movements up and down the income distribution, as some households benefit more than others from periods of strong economic growth or suffer particularly in periods of recession. This mobility is difficult to track at the level of the household in the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 336

28/09/2016 08:24



Middle incomes in boom and bust: The Irish experience ­337

Table 8.1  Number in work by income category, Ireland, 2008–12

Mean number in work < 60% 60% < = y < 80% 80% < = y < 120% 120% < = y < 200% y > = 200% Total

2008

2012

0.58 0.95 1.50 2.04 2.17 1.48

0.41 0.76 1.23 1.78 1.83 1.24

Proportion of working-­age adults in household in work y < 60% 0.30 60% < = y < 80% 0.51 80% < = y < 120% 0.65 120% < = y < 200% 0.80 y > = 200% 0.82 Total 0.63

0.21 0.39 0.59 0.76 0.79 0.57

absence of comprehensive longitudinal data,2 but it is clear from the changing profiles of the income groups and from the longitudinal data that both boom and bust were associated with considerable re-­ranking of households. Households directly affected by increasing employment rates in the boom and soaring unemployment (or collapsing business/self-­employment income) in the bust will have displayed particularly high levels of mobility, while the relative position of those relying primarily on social transfers (including many pensioners) vis-­à-­vis those relying on income from work will also have fluctuated a good deal as changes in social protection rates lagged behind or caught up on those in incomes from work. Households towards the middle of the distribution particularly hard hit by the crisis will have moved down the distribution, and the sharp increase in the proportion with a third-­level qualification in the bottom income category in particular suggests that the crisis impacted throughout the earnings and occupational distribution to an unusual extent (although the incidence of unemployment was certainly higher for the low-­skilled). Callan et al.’s (2014b) analysis of the limited year-­to-­year panel data in the European Union Statistics on Income and Living Conditions (EU-­SILC) concludes that declines in the income of those falling into the bottom decile at the onset of recession were central to the relatively marked decline in income share for that decile, and that most of those households came from the bottom one-­third rather than from the middle or upper parts of the income distribution.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 337

28/09/2016 08:24

338

Europe’s disappearing middle class?

6.  AUSTERITY AND THE MIDDLE While the immediate and direct effect of Ireland’s economic crisis was on employment and incomes from the market, buffered to some extent by social transfers and a reduced tax take as those incomes fell, the need to close the yawning fiscal deficit meant that a succession of ‘austerity’ measures were implemented that also had a major impact, in a manner which varied across the income distribution. With the numbers relying on social welfare payments rising rapidly – the total number of beneficiaries (including dependents) of weekly social welfare payments rose quickly from about 1.6 million (out of a total population of 4.2 million) to 2.2 million – social protection support rates became even more important. These were actually increased in 2009, before the scale of the crisis had become fully apparent, but subsequently cut several times for those of working age, particularly the young unemployed. Whereas these cuts would impact most on households in the lower parts of the distribution, substantial cuts in Universal Child Benefit will have affected families throughout the distribution. Social protection pensions, on the other hand, were largely protected, helping to explain the shift in the position of some older people up the income categories that we saw in the previous section. Successive governments also implemented substantial increases in direct taxes, including the introduction of a new levy which became the Universal Social Charge and increases in income tax via reducing credits and tax reliefs. The combined effect of these cuts in welfare rates and increases in direct taxation from 2008 to 2012 were quite progressive, in the sense that the consequent reductions in income in proportionate terms rose as one moves up the income distribution, as illustrated in Figure 8.8 (see Callan et al. 2014a). Overall, the redistributive impact of the direct tax and transfer system increased markedly over the period from 2008 to 2012. This is reflected in the percentage reduction in the Gini coefficient when going from market to disposable income, which is often used as a summary indicator of redistribution: whereas the impact of transfers and direct taxes meant that the Gini coefficient for (equivalized) disposable income was 20 per cent lower than that for market income in 2008, this figure had increased to 27 per cent by 2012. The relative importance of the direct tax versus the transfer system remained about the same, with three-­quarters of the total reduction in inequality coming from transfers and one-­quarter from the tax component (see Callan et al. 2015). The reduction in public sector numbers was a core component of the government’s response to the need to address the fiscal deficit in the crisis, and even with private sector employment falling rapidly this reduced the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 338

28/09/2016 08:24



Middle incomes in boom and bust: The Irish experience ­339 0 –2

Bottom

2

3

4

Top

–2.1

–4 –5.2

–6

–6.3 –8

–7.8

–10 –10.5 –12

Figure 8.8 Percentage impact of direct tax increases and social welfare reductions on household disposable income by quintile, Ireland, 2008–12 proportion of all employees who were in the public sector. This might be expected to have particularly affected middle-­ income groups: the largest decline was for households between 60 per cent and 80 per cent of the median (the ‘lower middle class’), where public sector employees accounted for 28 per cent of all employees in 2008 but only 21 per cent in 2012. The reductions in public sector pay already mentioned also played a major role in the fiscal correction. Public sector pay rose rapidly during the boom and a widening public/private sector premium was also used to justify these cuts (Kelly et al. 2009). A public sector pension levy was first introduced at a rate of 5–10 per cent, and then reductions in public service salaries of 5 per cent on the first €30 000, 7.5 per cent on the next €40 000, and 10 per cent on the next €55 000. Over the period from 2008 to 2012, Central Statistics Office (CSO) earnings series show average annual earnings in the public administration and defence, education and health sectors fell by 4–4.5 per cent in nominal terms, compared with a 2 per cent decline across all economic sectors; only the construction sector, which saw average annual earnings fall by over 9 per cent, did worse. The impact of the public sector pay cuts on the total disposable income of different decile groups has been simulated based on EU-­SILC survey data (see Callan et al. 2014b), and the results show that the net impact on the bottom four deciles will have been close to zero as this part of the overall household income distribution contains few public sector

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 339

28/09/2016 08:24

340

Europe’s disappearing middle class?

employees. The proportionate fall in disposable income then rises steadily as one moves up the distribution, from an average fall of 0.5 per cent for the fifth and sixth decile groups up to 3 per cent for the top decile group. This reflects both the position of public sector workers in the distribution and the fact that the pay cuts were structured to have greater impact on higher earners. Looking at the impact purely on those affected (rather than everyone in the decile), the losses in disposable household income were of the order of 2.5–4 per cent for affected workers in deciles 5 to 9, and close to 6 per cent for those in the top decile. If we looks at policy measures over the entire period to Budget 2015, and includes changes in direct taxes and social transfers, public service pay, indirect taxes and the introduction of a property tax, then Callan et al. (2015) show that greatest negative impact in proportionate terms was on the top decile, followed by the bottom decile, with the second and third decile seeing the least reduction in income. This reflected differences across family types as well as income levels, with the retired relatively favourably treated and the unemployed seeing particularly large reductions, helping to explain why the bottom decile fared worse than those above it. These analyses do not include the impact of the substantial reductions in the numbers employed in the public service and the embargo on new hires into public employment, which are difficult to model in this way but will have impacted particularly on young people coming onto the labour market over these years, including those with third-­level qualifications. So while those in ‘the middle’ did not fare worse in income terms than others in the recession, they experienced substantial declines on average and were particularly seriously affected by certain elements of the ­government’s (successful) efforts to address the fiscal deficit. Furthermore, trends in current income may not tell the whole story, either in terms of the benefits from the boom or the negative impacts of the recession. For that reason, we explore in the next section how other, non-­monetary indicators can be used to get a more comprehensive picture of the changing financial situation of households at different parts of the distribution and how the middle in particular fared.

7. 

LOOKING BEYOND INCOME

Income is a key, but by no means the sole, determinant of household living standards, and may thus not always be a reliable indicator of how living standards are changing in absolute or relative terms. In the first place, as well as income measured over a specific period – whether that be a week, a month, or a year – the income earned in earlier periods, and prospects

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 340

28/09/2016 08:24



Middle incomes in boom and bust: The Irish experience ­341

for the future, matter. Furthermore, the assets or debts accumulated by the household are also central to their financial resources. Needs also differ across households, in ways that go well beyond the differences in household size and composition taken into account by conventional equivalence scales – the most obvious example being the needs associated with disability. Finally, income itself will not be measured with full accuracy in household surveys, so other indicators may add to reliability in capturing living standards. For all these reasons, it is now common in assessing poverty and living standards to complement measures of income with non-­income information, including indicators of deprivation and measures of self-­assessed financial security or stress. Such an approach has been embedded in the EU’s social inclusion process since the adoption of the ‘Laeken’ set of indicators in 2001 (see Atkinson et al. 2002), and more recently in the framing of the EU’s poverty reduction target for 2020 in terms of a measure that includes both income and deprivation (as well as household ‘joblessness’). In a similar vein, it is helpful here in assessing how different income groups fared over Ireland’s boom and bust to look beyond income and see how both levels of deprivation and households’ own assessments of the extent of financial stress evolved. The measure of material deprivation we use for this purpose is constructed from responses to questions about whether the household had to do without the following items owing to lack of resources: ●● ●● ●● ●●

one week’s annual holiday away from home; a meal with meat, chicken, fish or vegetarian equivalent every second day; keeping the home adequately warm; and a personal computer.

In using these to create a deprivation index, we simply add the number of items the household reports it cannot afford. (Indices where the different items are weighted by their prevalence in the sample were also constructed, and show similar patterns to those described here). The measure of economic stress we use here is also constructed as an index with simple additive versus weighted and normalized variants, in this case using three items reflecting whether households reported that: ●● ●●

they had ‘great difficulty’ or ‘difficulty’ making ends meet; they had a problem with arrears (in the past 12 months) relating to mortgage or rent, or utility bills or hire purchase instalment payments; and

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 341

28/09/2016 08:24

342

Europe’s disappearing middle class?

Table 8.2  Deprivation by income category, Ireland, 1995–2012 Deprivation index (mean additive measure) y < 60% 60% < = y < 80% 80% < = y < 120% 120% < = y < 200% y > = 200% Total ●●

1995

2000

2005

2008

2012

1.26 0.96 0.65 0.32 0.11 0.67

0.81 0.53 0.25 0.14 0.00 0.37

0.97 0.68 0.29 0.12 0.03 0.42

0.89 0.78 0.43 0.13 0.02 0.46

1.24 1.06 0.77 0.31 0.10 0.71

total housing costs including mortgage repayment or rent, insurance and service charges were a ‘heavy financial burden’ or ‘somewhat of a burden’.

In capturing trends over the entire period of the boom from 1995 and subsequent recession, we are restricted to survey items included both in EU-­SILC and in the European Community Household Panel (ECHP), which is more limiting than either of these data sources would be if used separately. Unlike the measure of material deprivation included in the EU’s set of social inclusion indicators, we separate objective measures of material deprivation from subjective measures of economic stress since each may cast a distinct light on the nature of the trends observed. (For an in-­depth discussion see Nolan and Whelan 2011). Table 8.2 looks at the variation across income groups over time in scores on the deprivation index. Very marked differences in mean levels of deprivation across the income groups are to be seen throughout the period, but the varying trends over time are also striking. A pronounced decline in deprivation from 1995 to 2000 was registered for all groups, but with that decline being greater for those above 80 per cent of the median than those below that threshold. No further decline was seen from 2000 to 2008, which is surprising given the scale of the further increases in average income across all the income categories from 2000 to 2008 that were described earlier. The switch from ECHP in 2000 to EU-­SILC in 2005 complicates assessment of change between those two years, but this is not the case between 2005 and 2008 when again deprivation levels did not decline, indeed for lower middle-­and middle-­income groups they increased if anything. In focusing on how the middle-­income groups have fared, the very sharp increase in deprivation from 2008 to 2012 for each of the groups above 80 per cent of median income is even more noteworthy, being considerably greater in proportionate terms than the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 342

28/09/2016 08:24



Middle incomes in boom and bust: The Irish experience ­343

Table 8.3 Economic stress by income category, Ireland, 1995–2012 (mean additive measure)

y < 60% 60% < = y < 80% 80% < = y < 120% 120% < = y < 200% y > = 200% Total

1995

2000

2005

2008

2012

1.64 1.39 1.16 0.93 0.70 1.17

1.21 1.19 0.71 0.64 0.38 0.85

1.55 1.34 1.06 0.81 0.54 1.09

1.44 1.43 1.18 0.85 0.65 1.13

1.84 1.74 1.53 1.16 0.82 1.45

(still substantial) increase registered by the two income groups below that threshold. This is not the pattern we would have predicted on the basis of the way the mean or median incomes of these groups evolved, and suggests that a sole concentration on income would indeed miss important aspects of the way different groups were affected by the boom and then by the economic crisis. This is reinforced by the patterns in the self-­assessed reports by these households about their own financial situation. Scores on the financial stress indices we have constructed are shown in Table 8.3. Like deprivation, subjectively assessed economic stress declined for all income groups between 1995 and 2000, but then rose somewhat to 2005 and remained higher to 2008. The scale of the decline from 1995 to 2000 was again greatest for income groups over 80 per cent of the median, but from 2005 to 2008, the height of the economic boom, the group between 80 per cent and 120 per cent of the median actually saw a marked increase in financial stress. In the crisis these stress measures then rose substantially for all groups, but by most (in proportionate terms) for those between 80 per cent and 120 per cent of the median, though the variation across income groups is considerably less than was the case for deprivation. Patterns over the height of the boom and through the recession are particularly striking, so it is worth illustrating these by looking at individual items, first for deprivation and then for financial stress. Figure 8.9 shows how the percentage reporting that they could not afford an annual holiday away from home evolved from 2005, and brings out that this increased from 2005 to 2008 except at the top. The much more rapid increase from 2008 to 2012 was most pronounced for the middle-­income group, going from 29 per cent to 57 per cent for those between 80 per cent and 120 per cent of the median, while the bottom group saw an increase from 57 per cent to 79 per cent. In the case of financial stress, the indicator we show in Figure 8.10 is the percentage of households stating that they were having difficulty in

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 343

28/09/2016 08:24

344 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

Europe’s disappearing middle class?

2005

2008

2012

y < 60%

60% ≤ y < 80%

80% ≤ y < 120%

120% ≤ y < 200%

y ≥ 200%

Total

Figure 8.9 Percentage reporting they were not able to afford an annual holiday by income group, Ireland, 2005–12

0.6 0.5 0.4

0.3 0.2 0.1 0

2005

2008

2012

y < 60%

60% ≤ y < 80%

80% ≤ y < 120%

120% ≤ y < 200%

y ≥ 200%

Total

Figure 8.10 Percentage reporting great or very great difficulty ‘making ends meet’ by income group, Ireland, 2005–12

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 344

28/09/2016 08:24



Middle incomes in boom and bust: The Irish experience ­345

‘making ends meet’. We see first that from 2005 to 2008 the middle-­income group, between 80 per cent and 120 per cent of the median, was the only one to report an increase in this type of financial stress. From 2008 to 2012, by contrast, all groups except the top saw a considerable increase. This was pronounced for both bottom-­and middle-­income groups, whereas for the overall stress index we saw that the proportion seeing an increase in the recession was more marked for middle-­and upper-­income groups than for the bottom. This divergence reflects the fact that the overall stress index is significantly influenced by the other items relating to arrears and debt. Their importance is brought out by the analysis of trends in financial stress during the recession by Whelan et al. (2014) and MaÎtre et al. (2014), using a broader set of five indicators available in EU-­SILC only3 and incorporating life-­cycle stage as well as income and social class in the analytical framework. Their multivariate analysis brings out that the impact of the recession on levels of financial stress was distinctively high for households in the lower middle-­and middle-­income groups who were in the mid-­stage of the life course, headed by someone aged between 35 and 65. They also find that the pattern by age is critically related to housing: the addition of housing tenure adds substantially to the explanatory power of their model and shows that owner-­occupiers with a mortgage report the steepest rise in financial stress over the recession, even when one controls for income, social class and a range of other factors. Since what happened to credit, debt and the housing market during the economic boom and through the bust is among the most striking features of the Irish experience, underpinning the patterns in deprivation and subjective economic stress just described and playing a central role for many middle-­income households, we now focus on this as the first case study in the Irish experience from which broader lessons may be learned.

8. CASE STUDY 1: CREDIT, DEBT AND THE HOUSING MARKET IN BOOM AND BUST Ireland’s boom period saw a dramatic increase in credit, which left both the banking system and households highly vulnerable when the crash came from 2008. The level of mortgage credit per capita increased over tenfold between 1995 to 2008 (Russell et al. 2013), and other types of personal borrowing including via credit cards also increased very rapidly. The pace of expansion in credit accelerated strongly from 2002, following the introduction of the euro of which Ireland was a member from the outset, with bank lending to households and non-­financial corporates increasing

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 345

28/09/2016 08:24

346

Europe’s disappearing middle class?

450

House prices (new)

400

CPI

350 300 250 200 150 100 50

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

0

Figure 8.11 House prices and consumer prices (index base 1995, 100), Ireland, 1995–2013 from 62 per cent to 176 per cent of GDP by 2008. A stable low interest rate environment in the Eurozone, financial liberalization, a permissive regulatory framework and increased bank competition in the retail credit market were key drivers of the expansion in domestic credit supply in Ireland, and the leverage cycle of banks driven by fluctuations in global risk also contributed (see Everett 2015). Membership of the Eurozone was key: cross-­border funding inflows to Irish retail banks grew by a factor of five during the first ten years of the euro, and peaked at 205 per cent of GDP in September 2008, and interest rates, determined at the Eurozone level, remained low. This fed a property boom, with house prices rising much more rapidly than consumer prices generally, as illustrated in Figure 8.11. From 1995 to 2008 consumer prices rose by 50 per cent whereas new house prices rose by a factor of four. This placed significant strain not only on those trying to enter the housing market and buy their first house, but also on those ‘trading up’, with ready availability of credit but rapidly rising prices meaning that such households took on very substantial loans relative to their incomes. At the peak of the boom the ratio of house prices to average earnings and loan-­to-­value ratios among first-­time buyers were exceptionally high (Kelly 2009), and other households were also taking out substantial loans to trade up, to invest in buy-­to-­let properties which became significant for the first time in Ireland, and to fund consumption. This acceleration in debt is likely to have been a significant factor underpinning the levels of financial stress at household level towards the height of the boom, including the (modest) increase from 2005 to 2008 in the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 346

28/09/2016 08:24



Middle incomes in boom and bust: The Irish experience ­347 0.4

0.35 0.3 0.25 0.2 0.15 0.1 0.05 0

2005

2008

2012

y < 60%

60% ≤ y < 80%

80% ≤ y < 120%

120% ≤ y < 200%

y ≥ 200%

Total

Figure 8.12 Percentage reporting they were in arrears by income group, Ireland, 2005–2012 proportion of lower middle-­and middle-­income households falling into arrears on their mortgage, rent, utility bills or hire-­purchase ­payments – one of the items included in our overall stress index – shown in Figure 8.12. It also meant that the onset of the economic crisis was particularly damaging both for the broader economy and for heavily indebted households. The bail-­out required for the Irish banking system was among the greatest on record relative to the size of the economy, adding very significantly to the government’s fiscal challenges, but the combination of soaring unemployment, reduced earnings in some sectors, tax increases and cuts to working-­age benefits together with collapsing property values was equally toxic for households. This manifested itself in substantial numbers unable to service their borrowings and many properties worth less than the outstanding borrowings against them. From close to zero, mortgage arrears soared and by the depth of the recession about 13 per cent of mortgage holders were in arrears for 90 days or more on their principal dwelling, as were a further 20 per cent of buy-­to-­let mortgages holders (which had grown to account for almost one-­fifth of the stock of outstanding mortgage balances); one in four mortgage holders faced some form of difficulty meeting their repayments. While the position of those losing their jobs was particularly serious, a majority of those in such debt repayment difficulties were still in employment (McCarthy 2014); their household incomes may have been negatively affected by unemployment of other household

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 347

28/09/2016 08:24

348

Europe’s disappearing middle class?

members, by pay rate and overtime reductions, and by the increases in taxes described earlier. About 35 per cent of mortgage-­holders were in negative equity, and this was the case for over half of those in arrears for 90 or more days. The exceptionally high levels of debt of Irish households relative to other Eurozone area members at that point is clear from the harmonized Household Finance and Consumption Survey (HFCS) carried out in 2011–13. The percentage of Irish households with some debt was among the highest in the Eurozone, at 57 per cent, and their median values of outstanding debt were three times higher, principal residence mortgage debt was twice as high and other property mortgage amounts two and a half times higher than the Eurozone average. By their nature, the incidence of such debt-­related difficulties varied greatly by income level and by age/stage in the life cycle. The HFCS shows outstanding debt amounts to be highest for younger age groups, partly reflecting the shorter amount of time they have been repaying loans but also the fact that they bought property at peak prices with relatively large loans. The ratio of debt to household net income levels also shows a strong life-­cycle effect, peaking for the 34–45 age group then gradually declining. The median household in these age groups owed three times its income. As far as arrears are concerned, we can see how critical the timing of house purchase/borrowing was from the fact that over half of all the mortgage accounts in arrears were ‘originated’ in the years 2006–07. The differential impact of the recession by age, with debt and housing as central elements, is also clear from other sources of evidence. Gerlach-­ Kristen (2013) concluded from an analysis of household consumption patterns that the main burden of the Irish crisis has been borne by younger households. A special module on the effect on households of the economic downturn included in the CSO’s Quarterly National Household Survey in 2012 found that both households headed by someone aged under 35 and by someone aged 35–54 were more likely to have reduced their spending as a result of the economic downturn than those headed by someone aged 55 or over, with similar differentials with respect to experiencing difficulties in keeping up with their bills and debts, but those aged under 35 were the most likely to have failed to make a scheduled mortgage repayment. Returning to the EU-­SILC survey, Table 8.4 shows the percentage of households in each income group in owner-­occupied rather than rented housing over the period. We see that even for the lowest income group a majority were in owner-­occupied housing throughout the boom years, and for the highest income category that figure was over 90 per cent. Indeed, in 2000 the income groups in the middle of the distribution also had owner-­occupation rates of 90 per cent or above. This had fallen somewhat

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 348

28/09/2016 08:24



Middle incomes in boom and bust: The Irish experience ­349

Table 8.4 Percentage by income category in owner-­occupied housing, Ireland, 1995–2012

y < 60% 60% < = y < 80% 80% < = y < 120% 120% < = y < 200% y > = 200%

1995

2000

2005

2008

2012

56.1 76.4 87.0 92.8 96.1

62.6 71.0 90.4 97.1 96.2

56.8 68.3 85.1 89.2 92.4

60.1 66.1 81.1 87.3 92.1

49.4 54.2 67.5 87.2 88.0

Table 8.5 Percentage reporting housing costs a burden by income category, Ireland, 2005–12

y < 60% 60% < = y < 80% 80% < = y < 120% 120% < = y < 200% y > = 200% Total

2005

2008

2012

85.8 82.3 78.8 68.3 47.3 75.2

86.8 88.0 81.6 71.8 57.7 78.9

92.9 92.2 90.6 83.2 64.1 86.9

by 2005 as prices raced ahead of incomes, but then fell markedly in the recession so that by 2012 only two-­thirds of those between 80 per cent and 120 per cent of median income were in owner-­occupied housing. This primarily reflected the fact that new house building was minimal in the recession as the construction sector collapsed and mortgages were very difficult to obtain owing to the banking crisis. Table 8.5 then focuses on the percentage of households reporting that housing costs were a burden, another of the items in our financial stress index. We see that this increased between 2005 and 2008 especially for lower middle-­income groups as well as those at the top, but rose very much more rapidly between 2008 and 2012, with the increase then being greatest for the middle-­and upper middle-­income groups. This bears out the importance of housing, housing costs and housing-­related debt for the overall financial position and security of Irish households towards the height of the boom and especially in the recession.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 349

28/09/2016 08:24

350

Europe’s disappearing middle class?

9. CASE STUDY 2: MIDDLE OCCUPATIONS IN IRELAND’S BOOM AND BUST The occupation in which one works plays a central role in determining both access to economic resources and socio-­economic position more broadly, and changes in both the occupational structure and the relative fortunes of different occupational groups are at the core of trends in the world of work and how the middle has fared. Ireland saw some particularly striking developments in that regard over the course of the boom and bust, so this is the focus of our second case study. We look first at overall patterns and trends in terms of occupational and social-­class composition, and then hone in on two occupations that may be thought of as quintessentially middle class, namely, teachers and nurses. The very rapid expansion in the numbers at work in Ireland over the course of the economic boom was associated with major changes in the occupational structure. Table 8.6 draws on data from the Census of Population carried out in 1996, 2006 and 2011 which allows all adults at work to be categorized into a set of 25 occupational groups (based on those used by the UK Office for National Statistics). These show that between 1996 and 2006 – capturing most of the boom period – the proportion in farming halved, and the proportion in textiles and clothing, food and drink, and chemicals, wood and so on also fell, while the proportion in building and construction and in personal services (including childcare) rose sharply, and that in managers and executives, business and commerce, and computer software also rose. In the recession, by contrast, Table 8.6 shows that as the overall number at work fell sharply, it was building and construction together with the associated electrical and engineering occupations and ‘other manufacturing’ and clerical and office workers that declined significantly. In proportionate terms, this was associated with an increase in the share of managers and executives, sales occupations, health-­care occupations, teachers, and personal services. With the numbers at work and the occupational structure changing substantially over such a relatively short period, the profile in terms of social class also shifted. Table 8.7 also draws on the Censuses of Population and shows how the composition of the overall population changed from 1996 to 2006 and from 2006 to 2011 in terms of the six social classes used by the Irish Central Statistics Office. We see that in the central decade of the economic boom, there was a substantial increase in the proportion in the two classes at the top of the socio-­economic hierarchy, namely, the professional and managerial/technical classes, and a sharp fall in the proportions in the manual social classes. Importantly, in the recession this

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 350

28/09/2016 08:24



Middle incomes in boom and bust: The Irish experience ­351

Table 8.6 Population aged 15 years and over at work by occupational group and census year, Ireland, 1996–2011 1996

2006

2011

%

%

%

Farming, etc. 10.06 4.42 Electrical trades 1.89 1.91 Engineering etc. 4.55 3.91 Textile, clothing and leather 1.24 0.31 Food, drink and tobacco 1.87 1.22 Chemical, paper, wood, 1.55 0.93   rubber, plastics and printing Other manufacturing 3.90 3.60 Building and construction 5.77 8.84 Managers and executives 4.77 6.36 Communication, warehouse 5.86 5.67   and transport Clerical and office 10.08 9.11 Sales 10.44 10.06 Business and commerce 2.95 4.04 Computer software 1.43 2.29 Scientific and technical 2.58 2.99 Health and related 5.02 4.76 Social workers and related 0.51 0.89 Religious 0.49 0.20 Other professional 1.87 2.09 Personal service and childcare 8.37 10.05 Teachers 4.62 4.13 Central and local government 2.95 2.97 Police 0.81 0.64 Army 0.75 0.37 Other 5.66 8.27 All occupations 100.0 100.0

1996–2006

2006–11

Change in %

4.89 1.43 3.03 0.22 1.25 0.86

−5.64 0.02 −0.64 −0.94 −0.65 −0.62

0.47 −0.49 −0.87 −0.08 0.04 −0.07

1.83 4.17 7.63 5.58

−0.30 3.06 1.59 −0.19

−1.77 −4.66 1.26 −0.08

8.25 11.17 4.76 2.57 3.48 5.85 1.18 0.20 2.28 11.57 5.23 3.59 0.76 0.42 7.79 100.0

−0.97 −0.38 1.09 0.86 0.41 −0.26 0.38 −0.29 0.21 1.68 −0.49 0.02 −0.18 −0.38 2.60 0.0

−0.87 1.11 0.72 0.28 0.49 1.09 0.29 0.00 0.19 1.53 1.10 0.62 0.12 0.05 −0.47 0.0

was not reversed: indeed, there was a further more modest increase in the proportion in the top two classes as the proportion in the manual classes continued to decline, with a particularly marked fall in the size of the skilled manual class (which may be associated with the decline in building and construction and the emigration of skilled workers from that sector). It would be very informative to be able to link occupational groups and social classes to the income categories we are using here to distinguish ‘middle’, ‘lower middle’, ‘upper middle’ and so on income groups

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 351

28/09/2016 08:24

352

Europe’s disappearing middle class?

Table 8.7 Population aged 15 years and over by social class and census year, Ireland, 1996–2011

Professional workers Managerial and technical Non-­manual Skilled manual Semi-­skilled Unskilled All other gainfully occupied   and unknown All social classes

1996 %

2006 %

2011 %

5.35 21.97 18.44 20.45 12.81 8.53 12.44

6.53 26.37 17.02 17.12 11.07 4.25 17.63

7.34 27.28 17.46 15.42 10.62 3.71 18.17

100.00

100.00

100.00

1996–2006 2006–11 Change in % 1.18 4.40 −1.42 −3.33 −1.74 −4.28 5.19

0.80 0.91 0.44 −1.70 −0.45 −0.55 0.54

through Ireland’s boom and bust, but this would require consistent and reliable occupational coding/categorization across each of the household surveys on which we rely for income data, and this is not available.4 We can, however, look at this relationship at the height of the boom, in the EU-­SILC surveys for 2005 and 2008. Table 8.8 shows the distribution of the different social classes across the five income categories, focusing on all adults, categorized using previous occupation where not currently at work. We see that in both years the professional class is highly concentrated in the top two, ‘upper middle’ and ‘upper’, income categories, with over 80 per cent in those two categories and only about 12 per cent in the ‘middle’ income category. For all the other social classes, though, ­two-­thirds or more of their members are to be seen in the three middle-­ income categories taken together: even for the unskilled manual class, this figure is about 70 per cent. If one concentrates only on those currently at work, the proportions in the bottom two income categories are lower for all social classes, and the percentage in the three middle-­income categories is over 80 per cent for all three manual social classes. This brings out both the centrality and the diversity of the middle-­income groups which have been our primary focus here. It is also of interest to focus on two specific occupations that would be commonly considered to be ‘middle income’ or ‘middle class’ and which have featured in the discourse about the ‘squeezed middle’, namely, teachers and nurses. The distribution of individuals identified as being in those occupations in terms of their household equivalized income, using our five income categories, can be investigated using the household surveys employed earlier, though with some question-­ marks about the degree

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 352

28/09/2016 08:24

353

VAUGHAN 9781786430595 PRINT (M4051).indd 353

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

2.6 2.7 11.6 43.1 40.1 100

3.4 2.0 13.5 40.8 40.3 100

2005 y < 60% 60% < = y < 80% 80% < = y < 120% 120% < = y < 200% y > = 200% Total

2008 y < 60% 60% < = y < 80% 80% < = y < 120% 120% < = y < 200% y > = 200% Total

Professional workers %

6.1 7.1 20.6 42.2 24.1 100

8.4 6.0 20.0 41.5 24.2 100

Managerial and technical %

13.2 12.9 26.7 34.9 12.4 100

14.1 11.4 24.9 36.4 13.2 100

Non-­manual %

19.6 19.6 28.5 25.9 6.4 100

19.8 15.9 26.2 30.9 7.2 100

Skilled manual %

Table 8.8  Adults by social class and income group, Ireland, 2005 and 2008

21.5 16.1 33.2 25.3 3.9 100

23.4 17.0 27.7 27.0 4.9 100

Semi-­skilled manual %

23.6 23.3 30.5 19.2 3.4 100

30.5 18.6 24.7 23.0 3.1 100

Unskilled manual %

15.7 14.3 32.6 25.1 12.3 100

20.6 16.9 20.8 28.7 13 100

Other %

354

Europe’s disappearing middle class?

of consistency and precise categorization over time. Both teachers and nurses were found to be heavily concentrated in the top three income groups throughout the boom, with very few in the lower-­middle or poor categories. While the recession was associated with some increase in the proportion of nurses in households below the 60 per cent relative income threshold, there was also a marked increase for both nurses and teachers in the proportion to be seen in the top-­income group. While the reductions in pay for those working in the public sector, as well as the increases in direct taxation described earlier, would have affected most of those in these two occupational groups, the recession may have had a greater impact on some of those relying on income from self-­employment, and the incidence of job loss for other household members may also have been an important part of the picture.

10.  CONCLUSION AND POLICY ISSUES Ireland over the period from 1995 has seen a remarkable economic boom followed by deep recession, going together with dramatic fluctuations in employment levels, major changes in the composition of the labour force and in occupation and social class structures, a housing bubble followed by collapse, and important institutional changes affecting the world of work, with collective bargaining and social partnership central throughout the boom but collapsing at the onset of recession and a national minimum wage introduced and embedded. Our examination in this chapter of how the middle has fared through these turbulent times serves to highlight the complexity and ambiguity of the notion of ‘middle-­class squeeze’ and of ‘the middle’ itself. If seen in terms of the disposable income going to households around the middle of the income distribution, there was no ‘squeeze’ of the middle in the Irish case over this period, either in the boom or the bust. Distinguishing income categories defined vis-­à-­vis the median, the proportion in the middle rather than bottom or top categories rose over both the boom and the bust; having risen very rapidly in the boom, the average income of these groups declined sharply in the recession, but not by as much as the bottom-­income group. So over the entire period from 1995 to 2012, middle-­income groups – whether narrowly or broadly defined – fared quite well relative to the top and bottom of the distribution. While important, this does not provide a comprehensive picture and there are some important complications in interpretation. The first is that the type of households being categorized as ‘in the middle’ versus towards the bottom or top change over time; we have shown that there

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 354

28/09/2016 08:24



Middle incomes in boom and bust: The Irish experience ­355

were particularly marked shifts in the position of older people in the income distribution as the relative value of their pensions vis-­à-­vis other incomes fluctuated. In a similar dynamic rather than static vein, households towards the middle of the distribution but particularly those hard hit by the crisis will have moved down the distribution, and a longitudinal perspective would be required to capture that process. In the absence of suitable data, the sharp increase in the proportion with a third-­level qualification in the bottom-­income category suggests that the crisis impacted throughout the earnings and occupational distribution to an unusual extent (although the incidence of unemployment was certainly higher for the low paid and low skilled). It also reflects the very poor employment prospects of those coming on the labour market for the first time during the recession, even with those qualifications, which has driven graduate emigration. Those emigrants are among those most affected by the recession but are, of course, no longer in the data analysed, and their emigration is a defining feature of the recession for many parents, often in middle-­or upper-­income households. The distributional implications of the austerity measures used to address the yawning fiscal deficit have been a matter of some contention, in Ireland as elsewhere. The effects of discretionary cuts in welfare rates and increases in direct taxation were to reduce incomes substantially for middle-­income groups but by more for the top, with the redistributive impact of the direct tax and transfer system increasing markedly. Certain aspects of the fiscal adjustment, such as the reduction in public sector employment and pay, have seriously affected middle-­income groups, though the pay cuts were structured to have greatest impact on higher earners. The impact of such reductions, and of public expenditure cuts more broadly, on the availability and quality of public services will have had wide-­ranging effects, representing an important aspect of living standards not captured by household disposable income. Another aspect of living standards which is not reflected in household incomes relates to assets and debt, with particular reference to housing, which is central to the wealth of middle-­income households and to the Irish experience over the boom and bust. This was brought out here first by going beyond income to incorporate indicators of deprivation and economic stress into the analysis of trends over time for different income groups. This highlighted the very sharp increase in deprivation in the recession for the income groups above 80 per cent of median income, as well as the marked increases in financial stress for those around the middle. We then showed how the evolution of credit, debt and the housing market over the peak of the boom and through the bust was a central feature of the Irish case, with the explosion in borrowing during

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 355

28/09/2016 08:24

356

Europe’s disappearing middle class?

the latter part of the boom leaving indebted households highly vulnerable. The onset of the economic crisis was then particularly damaging for these households; those who bought their first house at the peak of the boom – mostly younger households – or borrowed to fund buy-­to-­let properties are very badly affected and difficulties with servicing loans are widespread. This, as well as the evolution of household incomes, is key to understanding the impact of Ireland’s boom and bust on different income groups, and brings out the extent to which that varied by age and circumstances. From a policy perspective, then, the Irish experience brings out the multidimensional nature of living standards for ‘middle’ as for other households, and the multifaceted nature of the policies required to underpin them. Current income plays a central role, which for most working-­age households depends on there being adequately remunerated work, but the importance of security versus precarity in that income over time and of the capacity to accumulate some savings rather than taking on unsustainable debt is also illustrated by the Irish case. Uncertainty in earnings at the level of the individual arises both from macroeconomic shocks, of which the economic crisis in Ireland from 2008 is an extreme example, and the way wage bargaining is organized and employment contracts are framed. While the need for much better management of the macro-­economy, and of the financial sector in particular, has been the dominant lesson drawn domestically from recent Irish experience, concern about wage-­setting and about such emerging features as zero-­hours contracts is also emerging as the economy recovers and unemployment comes down. Income adequacy and security in retirement for middle-­ income households is also under increasing threat in a system that relies on employer-­based occupational schemes to provide earnings-­related pensions on top of the social insurance-­ based flat-­rate pension. Social transfers also play a key role for those of working age both as supplement to income from work for those on ‘lower middle’ incomes and as floor for those who become unemployed, and policy in that respect continues to struggle with easing the transition into work. Awareness of the importance of housing as a core aspect of living standards has also been reinforced by recent experience, and while the failure of current policies to generate adequate affordable housing is also recognized the appropriate mix is not clearly in view. Finally, the appropriate level and structure of taxation is at the forefront of political debate, in the light of the substantial increases imposed during the severe but successful post-­crisis fiscal adjustment: this echoes broader debates about the quality of public services and how best to generate the resources required to fund them, in a world where – as the Irish

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 356

28/09/2016 08:24



Middle incomes in boom and bust: The Irish experience ­357

experience certainly illustrates  – both labour and capital are mobile. Social dialogue, having been prominent throughout the boom in the form of structured social partnership – including but going well beyond centralized wage bargaining – helped to limit increased inequalities but was among the casualties of the crisis. However, it is unclear what form it will take and what role it will play as these debates evolve.

NOTES 1. The exceptionally high levels of profit repatriations to multinational companies mean that for Ireland, GNP (or gross national income – GNI) is a more appropriate indicator of national living standards than GDP. 2. The ECHP constituted a full-­scale panel for Ireland from 1994 to 2001, whereas the EU-­SILC includes only a relatively small rotating panel component where small numbers make tracking the impact of the recession difficult. 3. In addition to the three indicators used here, items capturing whether the household had to go into debt within the last year to meet ordinary living expenses and whether it was able to save some of its income regularly were included in that study. 4. The 2012 results deviate from the distribution in 2005 and 2008, the change of schema employed (from what the CSO term SOC90 to SOC2010) from EU-­SILC for 2011.

REFERENCES Atkinson, A.B., B. Cantillon, E. Marlier and B. Nolan (2002), Social Indicators: The EU and Social Inclusion, Oxford: Oxford University Press. Baccaro, L. and M. Simoni (2002), ‘Centralized wage bargaining and the “Celtic Tiger” phenomenon’, Industrial Relations, 46 (3), 426–55. Barrett, A., J. FitzGerald and B. Nolan (2002), ‘Earnings inequality, returns to education and immigration into Ireland’, Labour Economics, 9 (5), 665–80. Callan, T., B. Colgan, C. Logue, M. Savage, and J. Walsh (2015), ‘Distributional Impact of tax, welfare and public service pay policies: budget 2016 and budgets 2009–2016’, Quarterly Economic Commentary, special article, Economic and Social Research Institute, Dublin. Callan, T., C. Keane, M. Savage, J.R. Walsh and B. Colgan (2014a), ‘Ireland’s economic adjustment: from crisis to recovery’, Intereconomics, 49 (6), 318–22. Callan, T., B. Nolan, C. Keane, M. Savage and J.R. Walsh (2014b), ‘Crisis, response and distributional impact: the case of Ireland’, IZA Journal of European Labor Studies, 3 (9), DOI: 10.1186/2193-­9012-­3-­9. Culpepper, P. and A. Regan (2014), ‘Why don’t governments need trade unions anymore? The death of social pacts in Ireland and Italy’, Socio-­Economic Review, 12 (4), 723–45. Everett, M. (2015), ‘Blowing the bubble: the global funding of the Irish credit boom’, Economic and Social Review, 46 (3), 339–65. Gerlach-­ Kristen, P. (2013), ‘Younger and older households in the crisis’, Research Notes RN2013/1/4, Economic and Social Research Institute (ESRI), Dublin.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 357

28/09/2016 08:24

358

Europe’s disappearing middle class?

Honohan, P. and B. Walsh, (2002), ‘Catching up with the leaders: the Irish hare’, Brookings Papers on Economic Activity, 33 (1), 1–78. Jenkins, S., A. Brandolini, J. Micklewright and B. Nolan (2013), The Great Recession and the Distribution of Household Income, Oxford: Oxford University Press. Kelly, M. (2009), ‘The Irish credit bubble’, UCD Centre For Economic Research Working Paper Series WP09/32, Dublin. Kelly, E., S. McGuinness and P. O’Connell (2009), ‘Benchmarking, social partnership and higher remuneration: wage settling institutions and the public-­ private sector wage gap in Ireland’, Economic and Social Review, 40 (3), 339–70. MaÎtre, B., H. Russell and C.T. Whelan (2014), Trends in Economic Stress and the Great Recession in Ireland: An Analysis of the CSO Survey in Income and Living Conditions (SILC), Dublin: Department of Social Protection. McCarthy, Y. (2014), ‘Disentangling the mortgage arrears crisis: the role of the labour market, income volatility and negative equity’, Barrington Lecture, Statistical and Social Inquiry Society, Dublin. McGuinness, S., F. McGinnity and P. O’Connell (2009), ‘Changing returns to education during a boom? The case of Ireland’, Labour, 23 (S1), 197–221. Nolan, B., J. Williams and S. Blackwell (2003), ‘New results on the impact of the minimum wage on Irish firms’, Quarterly Economic Commentary, 4 (Winter), 1–10. Nolan, B. and C.T. Whelan (2011), Poverty and Deprivation in Europe, Oxford: Oxford University Press. O’Donnell, R. (2008), ‘The partnership state: building the ship at sea’, in M. Adshead, P. Kirby and M. Millar (eds), Contesting the State: Lessons from the Irish Case, Manchester: Manchester University Press, pp. 73–99. O’Riain, S. (2008), ‘Competing state projects in the contemporary Irish political economy’, in M. Adshead, P. Kirby and M. Millar (eds), Contesting the State: Lessons from the Irish Case, Manchester: Manchester University Press, pp. 144–63. Regan, A. (2012), ‘The political economy of social pacts in the EMU: Irish liberal market corporatism in crisis’, New Political Economy, 17 (4), 465–91. Regan, A. (2013), ‘The impact of the Eurozone crisis on Irish social partnership: a political economy analysis’, Dialogue Working Paper 49, ILO, Geneva. Russell, H. and F. McGinnity (2014), ‘Work-­life balance, working conditions and the great recession’, in S.Ó. Riain, F. Behling, R. Ciccia and E. Flaherty (eds), The Changing Worlds and Workplaces of Capitalism, Basingstoke: Palgrave Macmillan, pp. 201–20. Voitchovsky, S., B. Maître and B. Nolan (2013), ‘Wage inequality in Ireland’s “Celtic Tiger” boom’, Economic and Social Review, 43 (1), 99–133. Watson, D., B. Maître and C.T. Whelan (2012), ‘Work and poverty in Ireland: an analysis of CSO Survey on Income and Living Conditions 2004–2010’, Social Inclusion Report No. 3, Department of Social Protection and Economic and Social Research Institute, Dublin. Watson, D., B. Maître and C.T. Whelan (2013), ‘Economic vulnerability and severity of debt problems: an analysis of the Irish EU-­SILC 2008’, European Sociological Review, 29 (4), 695–706 Whelan, C.T., H. Russell and B. Maître (2014), ‘Trends in economic stress and the great recession in Ireland: an analysis of the CSO Survey on Income and Living

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 358

28/09/2016 08:24



Middle incomes in boom and bust: The Irish experience ­359

Conditions (SILC)’, Social Inclusion Technical Paper No. 5, Department of Social Protection, Dublin. Whelan, C.T., H. Russell and B. Maître (2015), ‘Economic stress and the Great Recession in Ireland: polarization, individualization or middle class squeeze?’, Social Indicators Research, DOI: 10.1007/s11205-­015-­0905.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 359

28/09/2016 08:24

9. The middle class in Italy: Reshuffling, erosion, polarization Annamaria Simonazzi and Teresa Barbieri 1. FROM THE ‘ECONOMIC MIRACLE’ TO THE ‘YEARS OF DISCONTENT’ According to data provided by Sylos Labini (1974), in 1951 the urban middle classes – defined on an occupational basis as including small entrepreneurs, public employees and private clerical workers, artisans and traders – made up 26 per cent of total population. In 1971 they were up to 39 per cent and stood at 54 per cent in 2009 (Table 9.1; Bison 2013). The number of self-­employed – small entrepreneurs, shopkeepers and professional workers – soared: in the decade between the end of the 1960s and the end of the 1970s, the number of firms doubled, from 490 000 to 1 million. Table 9.1  Evolution of class composition, Italy, 1951–2009 1951

1971

1983

1998

2003

2009

1.9 26.5

2.5 38.5

3.3 46.4

8.3 47.2

10.2 53.5

10.3 53.7

5.2 8.0 6.0 7.3

8.7 11.0 5.3 13.5

10.2 15.8 5.8 14.6

28.2 3.1 15.9

16.1 18.3 3.6 15.5

18.8 17.8 3.4 13.7

Farmers Working class

30.2 41.2

11.9 47.1

7.6 42.7

4.3 40.2

1.9 34.4

1.7 34.3

Of which: Agriculture Industry Services

11.8 22.9 6.5

6.1 31.1 9.9

4.0 26.1 12.6

4.6 21.2 14.4

1.8 16.5 16.1

1.7 15.6 17.0

Bourgeoisie Middle classes Of which: Private employees Public employees Artisans Others

}

Source:  Bison (2013) based on data of (1) Sylos Labini (1974) and (2) ISTAT survey, Famiglia e soggetti sociali, years 1998, 2003 and 2009.

360

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 360

28/09/2016 08:24



The middle class in Italy ­361

The structure of society changed radically, from the shape of a pyramid to that of an onion (Bagnasco 2004: 280): Italy became a country of former poor people, with the illusion of having all become ‘bourgeois’.1 The difficult decades of the 1970s and 1980s led to the breakdown of the social bloc that had dominated up to that time: rents, public employees and financial interests on the one hand, small and large business and their employees on the other. The new social bloc ruling in the 1980s was an alliance between big and small enterprises, classes depending on public transfers and the public sector, which was made stable by the continuous growth of the public debt (Amable and Palombarini 2014). Since the mid-­1990s Italy has gone through a prolonged period of stagnation. It has recorded one of the lowest rates of growth in the European Union (EU) and has not undergone the booms and busts experienced by the other peripheral Eurozone countries. Moreover, its historically high debt to gross domestic product (GDP) ratio has entailed long-­standing fiscal austerity. The transition from rapid growth to quasi-­stagnation, coupled with a tighter public balance constraint, formed the basis for middle-­class discontent, dwindling political support and reform fatigue, opening the way to two colliding trends: a populist drive in the political arena and a neoliberal strategy in the economic policy field. Faced with the impossibility of implementing a ‘conservative plan’, the social forces favouring the neoliberal agenda attempted to create a new ‘bourgeois bloc’, putting together segments of social forces originally leaning towards the centre-­left on a compromise of ‘modernization’, focused on reducing ‘generous’ social transfers, increasing labour and product market competition and pushing forward European integration (Amable and Palombarini 2014). The labour market and welfare reforms (with their contradictions) reflect as much the confrontation of these two fronts as the difficulty of rebuilding a social bloc capable of supporting a viable social democratic alternative. After a sudden jump in the early 1990s, inequality and poverty measures show a substantial stagnation that contrasts with the widely shared sentiment of deteriorating economic conditions, increasing insecurity and vulnerability pervading large parts of the middle class (Massari et al. 2009). The aim of this chapter is to investigate whether and to what extent the hypothesis of a shrinking middle class and the transformation of the world of work can explain the sentiment of increased insecurity and vulnerability of the middle classes. Section 2 describes the evolution of income distribution over recent decades (1977–2012) and looks at the changes in the relative position of the middle classes in terms of income shares. Section 3 explores the issue of income polarization; that is, whether there has been a tendency for the lower and higher middle classes to move out towards

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 361

28/09/2016 08:24

362

Europe’s disappearing middle class?

the two extremes of the distribution. Sections 4 and 5 look at the main features of the middle classes according to the occupational status of their members and identify the major transformations in the world of work, while section 6 describes the importance of cohort effects in explaining the widening perception of insecurity and vulnerability pervading the middle classes. Section 7 provides an analysis of the twin effects produced by the fiscal crisis on the middle classes. The case studies take up the issues of insecurity and vulnerability by looking at mobility along the income distribution and intergenerational mobility.

2. THE EVOLUTION OF INCOME AND INEQUALITY 2.1 The ‘Economic Miracle’ (1950–60) and the ‘Cetomedizzazione’2 of Italian Society Middle-­ income groups are internally highly heterogeneous. Their size depends on how they are defined, whether by income, wealth, occupation, education, environment, social network, manners or values. In this section, we use the historical archive of the Bank of Italy’s Survey of Household Income and Wealth (SHIW), which gathers homogeneous and harmonized information on the income, wealth, savings and consumption of Italian households for the period 1977–2012, to assess the changes in income and inequality over the past 35 years, focusing on the middle class. The data cover a period of more than three decades, during which time substantial economic and political changes have occurred. We shall adopt the commonly agreed definition of the middle-­income class as formed by those families with equivalent disposable income between 60 per cent and 200 per cent of median income. In Figure 9.1 we present the trends of households’ disposable income in constant prices3 for three definitions of income.4 Disposable income increased over the period 1993–2006 and has fallen dramatically thereafter: the Organisation for Economic Co-­operation and Development (OECD) equivalent total disposable income fell by 11 per cent and total household disposable income by more than 13 per cent. Over the period 1977–2012, the Gini index exhibits two main cycles (Figure 9.2): a sharp decline from the end of the 1970s to the early 1980s and again at the end of the 1980s. After 1991, a sharp increase in the index brought inequality back to the level of the beginning of the period, where it stayed until the mid-­2000s, rising again after 2008. The reduction of the Gini coefficient in the late 1970s can be explained by labour market

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 362

28/09/2016 08:24



The middle class in Italy ­363

150

OECD equiv. tot. disp income Tot disp. household income

140

Sq. root equiv. tot. disp. income

130 120 110

2004

2006

2008

2010

2012

2006

2008

2010

2012

2002

2004

2000

1998

1995

1993

1991

1989

1987

1986

1984

1983

1982

1981

1980

1979

1978

90

1977

100

Source:  Elaboration on SHIW database.

Figure 9.1  Average disposable income, Italy, 1977–2012 0.37 0.35 0.33 0.31 0.29

OECD equiv. tot. disp. income

2002

2000

1998

1995

1993

1991

1989

1987

1986

1984

1983

1982

1981

1980

1979

1978

0.25

1977

0.27

Sq. root equiv. tot. disp. income

Total disp. household income

Source:  Elaboration on SHIW database.

Figure 9.2  Gini index, Italy, 1977–2012 and earnings dynamics (Brandolini et al. 2001): a period of great social conflict (which started at the end of the 1960s) had strengthened workers’ bargaining power. In 1975, their demand for a more egalitarian income distribution led to the reform of the indexation mechanism (scala mobile), ensuring to all workers the same absolute wage increase for each percentage point of inflation. In a context of high inflation, this mechanism caused a compression of wage differentials (Manacorda 2004). Indeed, in the same

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 363

28/09/2016 08:24

364

Europe’s disappearing middle class?

140 120 100 80 60 40

p90p10 p90p50

20

p50p10

12

10

20

08

20

06

20

04

20

02

20

00

20

98

20

95

19

93

19

91

19

89

19

87

19

86

19

84

19

83

19

82

19

81

19

80

19

79

19

78

19

19

19

77

0

Source:  Elaboration on SHIW database.

Figure 9.3  Selected income ratios, Italy, 1977–2012 period, the 90/10 ratio (Figure 9.3) shows a reduction of almost 14 per cent (compared with a reduction of 13 per cent of the Gini index, from 0.32 to 0.278), giving support to the idea that the fall in inequalities was due to the increase in the income share of the lowest deciles of the population. Similarly, the rise in inequalities in the mid-­1980s can be ascribed to the reduced power of the trade unions and the reduced redistributive effectiveness of the indexation mechanism (Pisano and Tedeschi 2007). The equalizing effect of the scala mobile declined both because inflation declined and because contingent wage increases were reduced by policy reforms. The disinflationary policy and the consequent extremely high real interest rates of the 1980s may have also played a role in increasing inequality. In the following decade, the abolition of the scala mobile in 1992, wage compression and wage ‘moderation’ (for example, the wage freeze agreement in 1993, just after a huge devaluation of the lira), labour and welfare reforms and industrial restructuring all led to the widening of the distributive gap. From 2006, the economic crisis affected the income distribution, widening inequalities still further. Looking at the evolution of the income share across deciles of the distribution (Table 9.2) the early 1990s seem to mark a watershed. The income shares of all the income classes up to the sixth decile peaked in 1991 and declined thereafter. Symmetrically, the income shares of the top four income classes experienced a setback in 1991 and regained their initial share by 2012. In particular, the two extremes of the distribution exhibit opposite trends, with the income shares of the first three deciles steadily

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 364

28/09/2016 08:24

365

VAUGHAN 9781786430595 PRINT (M4051).indd 365

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

5643.29 7231.48 8719.69 10 027.65 11 456.42 13 414.91 15 883.63 18 856.87 23 755.58 –

Quantile

49.26 63.12 76.11 87.53 100.00 117.10 138.64 164.60 207.36 –

% of median

1977

1991

3.05 4.63 5.91 6.70 7.77 9.01 10.52 12.55 15.25 24.61

7689.47 9586.41 11 309.80 13 026.02 14 900.22 16 904.87 19 273.34 22 405.57 27 426.77 –

51.61 64.34 75.90 87.42 100.00 113.45 129.35 150.37 184.07 –

3.49 5.22 6.36 7.21 8.41 9.53 10.87 12.49 14.81 21.62

8001.18 10 631.32 12 962.46 15 282.53 17 641.35 19 954.77 22 923.79 26 994.26 33 809.82 –

Share of Quantile % of Share of Quantile income, % median income, %

Source:  Elaboration on SHIW database.

 1  2  3  4  5  6  7  8  9 10

Quantile group

2012

45.36 60.26 73.48 86.63 100.00 113.11 129.94 153.02 191.65 –

2.90 4.68 5.92 7.04 8.16 9.35 10.70 12.39 14.99 23.88

6553.33 9100.00 11 266.67 13 300.00 15 615.39 18 100.00 20 900.00 24 200.00 30 533.33 –

41.97 58.28 72.15 85.17 100.00 115.91 133.84 154.98 195.53 –

2.20 4.42 5.72 6.90 8.12 9.50 10.91 12.65 15.15 24.43

% of Share of Quantile % of Share of median income, % median income, %

2006

Table 9.2 Distributional summary statistics, 10 quantile groups, Italy, 1977, 1991, 2006, 2012

366

Europe’s disappearing middle class?

declining after 1991, to account for a lower share in 2012 compared with 1997, and the top 10 per cent steadily increasing their share, though without reverting to their 1977 level. The share of the deciles around the middle of the distribution (deciles 4, 5 and 6) declined from the peak reached in 1991, though ending up with a higher share relative to 1977. The remaining three deciles (7, 8 and 9) improved their position relative to 1991, going back to (or increasing) their 1977 share. These preliminary findings seem to indicate a cleavage around the middle of the distribution, dating back to the early 1990s, which the economic crisis may have accentuated. In the next sections, we use the income classes defined with reference to the median of the equivalized household disposable income to investigate the evolution of the income in the middle of the distribution and the issue of a ‘shrinking’ middle class in Italy. 2.2  Middle Classes: Population and Income Shares Considering OECD equivalent disposable income, over the whole period all income classes – demarcated according to our definition – recorded income growth, although at very different rates (Figure 9.4): the ‘core income middle’ gained the most (40 per cent), whereas the bottom-­ class gained almost nothing (2 per cent). After 2006 all income classes 160 150 140 130 120 110

Bottom income

Lower middle income

Core middle income

Upper middle income

2012

2010

2008

2006

2004

2002

2000

1998

1995

1993

1991

1987

1989

1986

1984

1983

1982

1981

1980

1979

1978

90

1977

100

Top income Source:  Elaboration on SHIW database.

Figure 9.4 OECD modified equivalent total disposable income (mean) by income class, Italy, 1977–2012

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 366

28/09/2016 08:24



The middle class in Italy ­367

0.35

1977 1991 2006 2012

28.3

28.7

27.1

27.2

26.3

0.00

Bottom income

Lower middle income

Core middle income

Upper middle income

9.3

8.9

11.1

0.05

7.2

14.9

17.0

15.0

21.2

19.8

16.2

0.10

17.5

0.15

16.5

0.20

30.6

28.0

0.25

29.2

0.30

Top income

Source:  Elaboration on SHIW database.

Figure 9.5 Population share by income class, Italy, 1977, 1991, 2006, 2012 (percentage) experienced a decrease in their income, once again to different degrees: the top-­income class has been less affected by the recession (its income decreased by 10 per cent), whereas the bottom-­income class has been harder hit (–20 per cent). The three middle-­income groups experienced similar deteriorations of income (around 11 per cent). When considering the population and income shares accruing to each income class, again 1991 seems to mark a watershed (Figures 9.5 and 9.6). The shares of population falling in the two extreme classes fall in 1991 and rise thereafter. The middle classes show a symmetrical pattern, with their shares of population declining from the peak reached in 1991 (with the exception of the upper-­middle class in the period of the crisis). The income shares follow a roughly similar pattern: the bottom’s income share increases only marginally and declines in the crisis, while the top’s income share increased substantially after the trough of 1991. The middle classes exhibit a peak in 1991, but show a different pattern thereafter. The lower and the core middle-­income classes worsen their position, while, since the crisis, the upper-­middle class seems to have moved in line with the top-­income class. Since the beginning of the crisis, we observe some reshuffling, with the extreme classes increasing their share of population and the upper-­middle class becoming the most populated. The point is to understand whether the core-­middle class upgraded towards the upper classes or downgraded towards the low-­income classes.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 367

28/09/2016 08:24

368

23.2

16.9 9.3

9.5 10.7

7.4

7.2

6.7

6.4

0.15

9.1

0.20 0.10

22.0

2012

0.25

35.9 37.5

1991 2006

26.5

0.30

34.7 38.7

0.35

1977

25.4 22.9

0.40

22.9 27.1

0.45

Europe’s disappearing middle class?

0.05 0

Bottom income

Lower middle income

Core middle income

Upper middle income

Top income

Source:  Elaboration on SHIW database.

Figure 9.6 Income share by income class, Italy, 1977, 1991, 2006, 2012 (percentage)

3.  POLARIZATION OF THE MIDDLE CLASSES? The ‘shrinkage’ in the middle class could be due to an increase in both the lower and the upper part of the distribution, leading to a spreading away from the median and consequently to a rise in polarization. Analysing income polarization in several European and extra-­European countries, Atkinson and Brandolini (2011) do not find signs of income polarization in Italy up to 2004. Using a relative distribution approach, Petrarca and Ricciuti (2015) compare polarization in six European countries (the United Kingdom, Germany, Poland, Italy, France and Spain), and find that polarization increased in all the countries except Italy (data available until 2010), with the effect of the middle class getting poorer exceeding that of the middle class getting richer. Because the results presented above depend on the selected income range, to test the polarization hypothesis we use a methodology that does not depend on arbitrary income boundaries. First, we use Kernel density estimation to see how the income distribution changed.5 We compare two different periods: 1991–2006, characterized, as we saw previously, by a high but stable level of inequalities and 2006–12, in which the Italian economy had to deal with the 2008 economic crisis, the recession and growing inequalities. In Figure 9.7 we present the income density functions for 1991, 2006 and 2012.6 From 1991 to 2006 the income distribution shifted to the right along the income scale, pointing to an increase in income levels (location effect). This finding is consistent with

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 368

28/09/2016 08:24



The middle class in Italy ­369 0.8

1991 2006

Density

0.6

2012 0.4 0.2 0 0

5

10

15

Figure 9.7 Kernel density estimation, log of household equivalent disposable income, Italy, 1991, 2006, 2012 (2012 prices) the increase in the median income that occurred in this period. From 2006 to 2012 we have instead a small shift to the left, probably explained by the economic downturn (the median income decreased in this period). From 1991 to 2006 we notice an increase in the dispersion of income: a lower mode and a larger right tail. In 2012 the middle of the distribution was squashed down, a small bump appears on the left of the secondary mode and the left tail gets thicker, pointing to a greater concentration in the lower part of the distribution. Thus, we can speculate that the central part of the distribution, after the economic crisis, moved toward lower rather than higher income ranges, while the increased distance between the two modes could be a sign of polarization. The shrinkage of the middle class is related to the concept of bipolarization when its reduction occurs because of an increase in both the lower and the upper class. As argued above, there is no consensus on which income ranges represent better the middle class, thus making it difficult to compare studies using different cut-­off points. Foster and Wolfson (2010) derive a method to indicate when one distribution is unambiguously more polarized than another, irrespective of which income range is chosen (see Appendix 9A.1 for analytical details). In order to compare different income distributions, the first step is to measure the middle class, represented by the proportion of individuals belonging to a specific income range. We obtain the M-­curve, which measures the concentration of mass around the median of the income distribution and does not depend on an arbitrary definition of the middle class. If the M-­curve of period 1 is located above the M-­curve of period 2, we can affirm that in period 1 the middle class was larger, irrespective of the income range chosen to define income classes.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 369

28/09/2016 08:24

370

Europe’s disappearing middle class?

To derive their polarization measures, Foster and Wolfson divide the income distribution in two parts, one above and one below the median. The first-­degree polarization curve (the S-­curve) is related to the concept of ‘increasing spread’: we have a growth in polarization when there is an increase in the income range (spread) required to quantify any defined population range. Therefore, a wider income spread means a lower proportion of the population around the middle. We can state that the income distribution of period 1 has a larger middle-­income class with respect to the income distribution of period 2, irrespective of the chosen population range, when its S-­curve is located below that of period 2. The second-­degree polarization curve is a representation of both concepts of ‘increased spread’ and ‘increased bipolarization’, that is, when we observe a higher mass in the tails of the income distribution. This curve (the B-­curve) ranks one distribution above the other in terms of polarization when the average distance from the median is higher for every range of families around the median. The distribution of period 1 is more polarized than the distribution of period 2 if its B-­curve is located above the curve from period 2. We estimate the curves for income distribution in 1991, 2006 and 2012. We first compare the 1991 curves with the 2006 curves and then compare them with the 2012 curves (Figure 9.8). Since the 1991 M-­curve and the 2006 M-­curve cross each other in several parts, especially around the median, we are not able to state that one of the two distributions has a larger middle class. The first and second-­degree curves also cross each other, making it impossible to derive an unambiguous conclusion on polarization. The 2006 M-­curve is above the 2012 M-­curve, meaning that, no matter what cut-­off points are chosen, the 2006 income distribution has more mass around the median than the 2012 distribution. Therefore, the middle class in 2006 was unambiguously larger than in 2012. If we observe the first-­degree polarization curves for the second period, the two curves cross each other, but only at one point away from the fiftieth percentile and in all the other parts the 2012 curve is above the 2006 curve, meaning that in 2012 polarization is higher. Thus, the 2012 curve has a greater income spread and consequently a smaller middle class, because the proportion of population around the middle is lower. Looking at the second-­degree curve, the 2012 curve is clearly above the 2006 curve. In other words, the 2012 income distribution has a greater spread and bipolarity than income distribution in 2006. In Figure 9.9 we present the Foster–Wolfson polarization index, a synthetic index of bipolarization derived by Foster and Wolfson consistently with the polarization curves and similar to the Gini index.7 The index confirms our previous findings of a similar degree of polarization

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 370

28/09/2016 08:24



The middle class in Italy ­371

0.5 0.4 0.3 0.2 0.1 0 0

0.5 1 1.5 Normalized income 1991

2

population share

Population share

Evolution of the middle class. M - curve. 0.5 0.4 0.3 0.2 0.1 0 0

0.5

1

1.5

2

Normalized income

2006

2006

2012

Spread

Spread

Evolution of polarization. S - curve. 1.5 1.0 0.5 0 0

0.2

0.4

0.6

0.8

1.5 1.0 0.5 0 0

1

0.2

Percentile 1991

0.4

0.6

0.8

1

Percentile 2006

2006

2012

0.3 0.2 0.1 0 0

0.2

0.4

0.6

0.8

Percentile 1991

1

Cumulative spread

Cumulative spread

Evolution of polarization. B - second degree curve.

0.3 0.2 0.1 0 0

0.2

0.4

0.6

0.8

1

Percentile 2006

2006

2012

Figure 9.8  Middle class and polarization, Italy, 1991, 2006, 2012 between 1991 and 2006 (from 0.25 to 0.248) (although it shows an increase in-­between) and an increase thereafter (in 2012 the index reaches the value of 0.274). In conclusion, it seems that, after the start of the economic crisis, we can detect the first signs of polarization and thus of a declining middle class. If the economic and fiscal crisis persists, the reshuffling is likely to continue, with households at the top of the distribution upgrading their position and households at the bottom losing ground. A range of not mutually exclusive causes have been mentioned: technical change leading to the polarization of skills and jobs (as described in Chapter 1), the effects of the fiscal crisis on public employment (wage freeze and precarious jobs) and welfare, the more severe effects of the economic crisis on the lower-­middle class, institutional and economic factors favouring an increase in top-­labour income, and labour reforms penalizing young cohorts independent of their class of

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 371

28/09/2016 08:24

372

Europe’s disappearing middle class?

0.35 0.30 0.25 0.20 0.15 0.10

2012

2010

2008

2006

2004

2002

2000

1998

1995

1993

1991

1989

1987

1986

1984

1983

1982

1981

1980

1979

1978

0 1977

0.05

Source:  Elaboration on SHIW database.

Figure 9.9  Foster and Wolfson bipolarization index, Italy, 1977–2012 origin. In the next section we focus on the effects of the changes that have occurred in the world of work.

4. MIDDLE CLASSES: WEALTH, EDUCATION AND EMPLOYMENT As argued in the introduction, middle-­class status is defined also by features other than income, such as education, occupation, wealth and social networks. Wealth plays a crucial role in defining middle-­class living standards and status, and thus their economic stability and resilience to shocks. The historical archive provides data only since 1987: over the entire period net total wealth – real and financial assets minus liabilities – increased by 80 per cent. Net wealth fell by 7 per cent between 2010 and 2012, reflecting mainly falling real estate prices.8 Wealth is much more concentrated than income and this unequal distribution of resources has increased over time: from 2004, the Gini index increased steeply, to reach 64 per cent in 2012 (it was 60 per cent in 2006, before the financial crisis). Indeed, while net wealth has increased for all income groups, its rate of growth has differed widely: the top-­income class more than doubled its assets (Figure  9.10), while the lower middle-­income and the bottom-­income classes showed a more modest increase (36 per cent). It is worth noticing that the upper-­middle class shows a steep decrease since 2008 (–20 per cent). Not surprisingly, the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 372

28/09/2016 08:24



The middle class in Italy ­373

250 200 150 100 50 0

Bottom income

Lower middle income

Upper middle income

Top income

Core middle income 1987 1989 1991 1993 1995 1998 2000 2002 2004 2006 2008 2010 2012

Source:  Elaboration on SHIW database.

Figure 9.10  Total net wealth by income groups, Italy, 1987–2012 trends of net worth follow those of real estate. In fact, real assets account for the greater part of total assets (Figure 9.11). Despite a general improvement since 1977, as expected, educational attainment increases with income (Figure 9.12), as does the employment rate (Figure 9.13). The proportion of people aged between 25 and 64 with at least upper secondary qualifications has grown substantially over the past 35 years across all income classes, but its increase was the highest among the middle classes (the two upper classes if we consider only tertiary graduates). Since 1977, the employment rate has increased for all income classes with the notable exception of the bottom (Figure 9.13). The female employment rate has also increased sharply in all income classes: although the two bottom classes have almost tripled the proportion of women employed (Figure 9.14), the gap with the middle and upper classes remains massive: women belonging to the higher-­income classes are still much more likely to be employed. The share of wage earners and the share of self-­employed (Figures 9.15 and 9.16) increase with income. However, after 2006, the proportion of self-­employed increased only in the bottom-­and lower middle-­income classes, whereas it decreased in all the other classes. As argued in the following section, the sharp increase in self-­employed – hitherto the prevailing form of non-­standard employment – conceals a high percentage of ‘false’ self-­employed. Thus, within the same category, we have both true, high-­income ‘professional’ workers, and low-­paid, precarious workers.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 373

28/09/2016 08:24

374

Europe’s disappearing middle class?

130 125 120 115 110 105 100 95

1987 1989 1991 1993 1995 1998 2000 2002 2004 2006 2008 2010 2012 Bottom income

Lower middle income

Core middle income

Upper middle income

Top income Source:  Elaboration on SHIW database.

0

82.42

28.67

29.3 13.95

20.55

26.42

Bottom income

8

10

12.35

20

4.19

30

25.55

40

38.2 38.29

1977

50

50.21

1991

60

44.14

70

50.14

2006

73.26 84.59

2012

80

53.17

90

65.55 60.99

Figure 9.11 Ratio between real assets and total assets by income level, Italy, 1987–2012

Lower middle income

Core middle income

Upper middle income

Top income

Source:  Elaboration on SHIW database.

Figure 9.12 Educational attainment (share of population 25–64 with at least an upper secondary education), Italy, 1977, 1991, 2006, 2012 (by income groups)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 374

28/09/2016 08:24



66.8

58.3

61.3

61.3

52.5

50.4

46.4

45.5

48.3

60

43.4

80

96.3

95.2

90.2

83.7

92.2

90.7

2012

82.5

1991

2006

80.0

100

1977

74.4

120

79.5

The middle class in Italy ­375

40 20 0 Bottom income

Core middle income

Lower middle income

Upper middle income

Top income

Source:  Elaboration on SHIW database.

0

94.1

91.0

81.3 68.3

83.8

Bottom income

7.6

10

11.7

20

8.0

30

17.3

20.1 27.8

40

37.1

50

70.2

2012

50.8

2006

60

65.4

70

65.6

1991

42.4

80

1977

25.4

90

36.8

100

86.6

Figure 9.13 Employment rate, Italy, 1977, 1991, 2006, 2012 (by income groups)

Lower middle income

Core middle income

Upper middle income

Top income

Source:  Elaboration on SHIW database.

Figure 9.14  Female employment rate, Italy, 1977, 1991, 2006, 2012

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 375

28/09/2016 08:24

376

40

69.8

62.9

65.3

60.1

76.8

74.1

66.9

63.0

68.3

67.6 49.1

48.1

51.1

50.8

42.1

40.9

50

2006 2012 37.6

60

1991

35.8

70

1977

36.1

80

34.3

90

Europe’s disappearing middle class?

30 20 10 0

Bottom income

Lower middle income

Core middle income

Upper middle income

Top income

Note:  Percentage of household heads and spouses in the age range 25–64. Source:  Elaboration on SHIW database.

10

18.4

11.5 15.6 16.6 15.5

18.7 12.4 11.1

15

9.2

20

11.5 10.2 10.5

25

1977 1991 2006 2012

8.3

30

12.2 9.2 8.9 9.6

35

30.1 32.4 26.5

Figure 9.15  Percentage of wage earners, Italy, 1977, 1991, 2006, 2012

5 0

Bottom income

Lower middle income

Core middle income

Upper middle income

Top income

Note:  Percentage of household heads and spouses in the age range 25–64. Source:  Elaboration on SHIW database.

Figure 9.16  Percentage of self-employed, Italy, 1977, 1991, 2006, 2012

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 376

28/09/2016 08:24



The middle class in Italy ­377

5.  THE CHANGING WORLD OF WORK In the 1980s, the persistence of inflation and the difficulties faced by the Italian economy were blamed on the labour market, more particularly, wage indexation and labour market regulation. In the early 1990s, a period of cooperative industrial relations – so-­called concertazione – brought wage ‘moderation’ and disinflation. In the following decades, social dialogue, industrial relations, collective bargaining and labour market regulation all underwent substantial changes (Simonazzi 2015). In 1997 the Italian government approved the so-­called ‘Treu law’ (Law 196/1997) that introduced temporary contracts and incentives for part-­time work. This law also gave legislative recognition to temporary work agencies (staff leasing). In 2003 the ‘Biagi law’ introduced new forms of atypical employment contract: on-­call jobs, job sharing and occasional work (lavoro a progetto), and further deregulated staff leasing and part-­time work. In 2012, in response to the adverse macroeconomic situation and pressure from the European institutions, the Italian government approved an extensive labour market reform, the so-­called ‘riforma Fornero’ (Law 92/2012), which relaxed the conditions regulating temporary contracts, while attempting to reduce the number of contractual arrangements. Finally, the so-­called ‘Jobs Act’ (Law 78/2014), passed by the Renzi government in February 2014, replaced the sanction of reintegration of employees after wrongful dismissal with financial compensation alone, thus reducing employers’ perceived costs of firing. The reform de facto abolished the much contested Article 18 of the Code of Labour Rights, regulating the conditions of dismissal, and introduced a three-­year progressive maturing of entitlements to stabilization, supported by substantial fiscal subsidies for firms. The stated aim of these reforms was to enhance the ‘flexibility’ of the labour market, which supposedly would reduce unemployment and make the labour force more adaptable to meet rapidly changing production demands. However, to avert the risk of conflict, until the Jobs Act flexibility was increased ‘at the margin’ by changing the rules on hiring but leaving unchanged the rules on firing and income maintenance schemes for core workers. The result was a new segmentation among successive cohorts entering the labour market: the latest cohorts experienced both lower employment protection and lower income. A large number of atypical contractual arrangements (including apprenticeships, fixed-­term contracts, collaborators, agency work and project work) came to coexist with standard employment contracts, characterized by high social security protection. Up until the end of the 1980s, in a context of an even balance of power in industrial relations, labour reforms tended to proceed by consensus between government and social partners. Thereafter, labour

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 377

28/09/2016 08:24

378

Europe’s disappearing middle class?

market deregulation progressed erratically, with no formal agreement with employers and unions. Since the 1990s, there has been a trend towards diminishing the role of social dialogue institutions (with the government withdrawing from negotiations), weakening trade unions by fostering their divisions, altering the procedures governing social legislation, and calling for amendments to labour codes/labour regulations in order to simplify hiring and dismissal rules and reduce the cost of layoffs (Simonazzi 2015). These labour reforms have not been accompanied by any reform of the so-­called ‘shock absorber’ systems. A truly universal unemployment benefit scheme and income support policies are still lacking. Social expenditure, mainly in the form of monetary transfers (pensions and monetary benefits), has been drastically curtailed because of the fiscal crisis. Social investment (education, research and development, health care, training and active labour policies) has been reduced to keep within the Fiscal Compact. Reforms of the product market have been limited to privatization, but not accompanied by true liberalization, and the restructuring of the economy has been left to ‘market mechanisms’, without any guidance provided by industrial policy. Because of the way in which labour reforms were designed and implemented, the substantial growth of atypical employment is strongly concentrated in the younger cohorts. In 2012 the share of fixed-­term contracts (including ‘collaborations’) was 11.6 per cent of total employment, but it was 25.4 per cent for people aged 15–34 (and 31.7 per cent for those with a college degree) (ISTAT 2014). Although the share of fixed-­term contracts decreases with age, due to the long stagnation of the Italian economy an increasing proportion of young people face an increasing risk of being trapped in repeated atypical employment contracts: they experience discontinuous careers, low pay, inadequate social protection and low future pension benefits (Rosolia and Torrini 2013; Berloffa et al. 2014). This is especially true for young women: 18.2 per cent of women between 30–34 years of age are in non-­standard employment, compared with 12.7 per cent of men in the same age bracket (ISTAT 2014). An analysis of the flows out of non-­standard jobs for young people between 15 and 34 years of age confirms this discouraging picture. Between 2007 and 2013, the percentage of exits from non-­standard to standard employment decreased from 31 to 22 per cent; the rate of permanence within the area of precariousness has remained more or less stable at very high levels (from 54 to 55 per cent), while the rate of those exiting precarious jobs to enter into unemployment increased substantially (from 10.5 to 16.4 per cent) (Figure 9.17). The suggestion that entering the labour market by accepting a precarious job can provide a ‘stepping stone’ to a more stable position needs qualification: its probability differs

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 378

28/09/2016 08:24



The middle class in Italy ­379

2012–13

55.20

22.30

2011–12

57.20

2010–11

58.00

22.60

2009–10

59.10

20.30

2008–09

21.10

50.60

2007–08 0

10

20

30

6.00

15.40

6.30

13.50 13.70

24.90

53.60

16.40

16.10 30.80

40

50

Non-standard employment Unemployed and potential labor force

60

70

6.90 8.50

10.50 80

5.90

90

5.20 100

Standard employment Not seeking a job

Source:  ISTAT 2014 based on Labor Force Survey.

Figure 9.17 Flows out of non-standard contracts for young people, 15–34 years of age, Italy, first quarter 2007–first quarter 2013 (percentage) substantially across workers, and it has been made more and more difficult by the long crisis and labour market deregulation, which have progressively reduced firms’ constraints in new hiring. It remains to be seen whether the Jobs Act will be able to increase the creation of new stable jobs. If the deep economic crisis persists, the heavily subsidized attempt to favour a progressive stabilization of newly hired workers over a period of three years is unlikely to produce any concrete result. As the president of the young entrepreneurs association was quick to point out, this reform does not create more new jobs, which only an increase in the demand for products can elicit, but may encourage a propensity on the part of firms to transform temporary contracts into more stable ones, which is made more likely, for the time being, by the fact that the increased costs are almost entirely taken up by the state through the waiving of social contributions.

6.  COHORT EFFECTS Labour market reforms, the long-­term stagnation in incomes and the current crisis have affected the middle classes mostly by souring their

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 379

28/09/2016 08:24

380

Europe’s disappearing middle class?

expectations concerning the future of their offspring. In fact, the difference between the average incomes of young and of old people has increased, mainly because of the increased instability of young workers’ incomes: changes in the labour market and especially the diffusion of temporary contracts have affected older and young workers differently. Thus, middle-­ class discontent is certainly related to the increased inequality and loss of absolute income (analysed in section 2), but it has equally to do with the perceived reduced opportunities for younger generations and a decreasing intergenerational social mobility. Some periods are more open and effective for the socialization of the younger population, while in others lack of opportunities at entry into adulthood can generate long-­term ‘scarring’ effects. These unequal dynamics could produce ‘social generations’ (Mannheim 1952): some benefit from fortunate circumstances, while others suffer long-­lasting hardship. The transformation of the set of opportunities that has occurred over the past three or four decades has had a profound impact on different birth cohorts. The cohorts of young people entering the labour market in the present crisis risk being a ‘lost generation’: they will be too old to compete with the next cohorts following hard on their heels. The structure of birth cohorts may differ in terms of size, family structure, and level of education, but individuals with the same background will face divergent and fluctuating outcomes because of their specific period of socialization. Berloffa et al. (2015) compare the transition between the occupational status of two different educational cohorts: those individuals who finished education between 1971 and 1985, and those who finished after 1992. The results show that for the second cohort (the cohort hit by the effects of the labour market reforms enacted since the end of the 1990s) there is a significant increase in the probability of being trapped in a non-­standard, and thus insecure, form of employment. Countries differ in their capacity for inclusion (if only in terms of market participation). According to Chauvel (2013), status societies, such as those of the conservative and Mediterranean welfare regimes, but also Japan, provide an institutional context that favours the emergence of deeply contrasted cohorts with diverging social paths and life chances. More resilient societies, where second chances exist (in terms of education or work opportunities), have less cohort-­based inequality: this is the case of the Nordic countries and English-­speaking ‘liberal’ countries, where new social or market opportunities over the life course can lessen, at least partially, difficulties in early adulthood. In a comparative analysis of four countries between 1985 and 2005, Chauvel shows the degree to which different birth cohorts faced divergent life chances, and how the responses of middle-­class dynamics to the economic slowdown of the post-­1970 period

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 380

28/09/2016 08:24



The middle class in Italy ­381

varied across welfare regimes. The Nordic welfare regime, exemplified by Norway, seems fairly stable in terms of how economic difficulties are distributed across cohorts, and the transformation of both the upper-­and the lower-­middle class is not strong. At the opposite end of the inequality spectrum, the United States shows less between-­cohort inequality and a much smaller negative effect for the cohorts born in the 1950s compared with the French and Italian cohorts. The economic slowdown experienced by young Italians in the 1980s (those born in the 1960s) still has consequences for the contemporary middle-­aged population. In France, the cohort scars in terms of lower incomes and greater difficulties in reaching upper-­middle-­class positions are visible over the long run: those born too late, in the 1960s or later, do not equally share the economic position and affluence of earlier cohorts. As for the effect of changes in the welfare state, Chauvel finds that in Italy and France there has been a substantial shift in relative disposable income in favour of older cohorts. The French and Italian responses to new challenges – a mix of affluent seniors and deepening difficulties for the younger cohorts (lower relative income, economic dependence, unemployment, plus deeper inequality in Italy) – creates a paradoxical situation in which the older cohorts’ income position improves (to more middle-­or upper-­class positions) while the young of the 1980s (today’s middle-­aged adults) are durably destabilized, with lower opportunities to climb into the upper classes and higher risks of lower incomes (Chauvel 2013: 20). Thus, in France and Italy, the early baby-­boom generations (the young adults of the late 1960s) benefited from the expansion of the middle class, while the cohorts born in the 1960s are the victims of a reversal in its development. These cohorts were the first to experience the sharp economic slowdown and expansion of youth unemployment in the 1980s, with its deep long-­ term consequences. Here lies the problem of sustainability for the current welfare regime: the basis of the present welfare state will mechanically erode with cohort replacement, and the security it offers to seniors will not be extended to the young cohorts facing insecure employment and radical uncertainty. 6.1  Decreasing Intergenerational Mobility In such stern labour market conditions, the channels by which young people enter the labour market have become an important issue for equality of opportunities. Among young people aged 15–34 who started work in 2013, more than one-­third acknowledged the importance of the informal network of relatives and friends, 26.3 per cent submitted an application to the employer and 11.8 per cent had previous experience of internship or

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 381

28/09/2016 08:24

382

Europe’s disappearing middle class?

apprenticeship with the firm. The percentage of young people who found a job through the public employment services (1.4 per cent) or other employment agencies (5.4 per cent) is discouragingly low, given that it is the task of these institutions to match demand and supply of jobs. Examining the mechanisms behind career progression, a stream of literature has focused on ‘nature versus nurture’, that is, the relative importance of an individual’s supposedly innate qualities (nature) versus the role of personal experiences, such as family upbringing, peer groups and other environmental factors (nurture). Career paths are linked also to economic advantages transmitted in adulthood through, for example, family ties, social networks, and the other benefits deriving from parents’ positional advantage. While these connections may convey a wealth of reliable information, act as a source of peer monitoring and accentuate the role of trust, they may also hide nepotism and corporate practices, thus representing a source of both inefficiency and persistent inequality (Mocetti 2014). Thus, family background can influence earnings in various ways: (1)  indirectly, through the probability of higher educational attainment and the likelihood of getting a good job; and (2) directly, through a residual effect that may be due either to unobservable abilities or to networks of social contacts. While investing in education has been a strategy followed mostly by the middle class to ensure a better future for their offspring, education does not seem to be enough any longer to ensure a good quality job. In a labour market characterized by over-­supply of skilled labour and an insufficient demand for skills, which reflects the characteristics of the economic structure and the incentives provided by labour laws, Italy’s peculiarity is over-­education. Although new entrants’ jobs are less and less stable and low-­paid for all the younger cohorts, also the more skilled young workers seem to experience a severe wage penalty (Naticchioni et al. 2014). 6.2  ‘Closed Shops’: Enforced by Regulation? When the crisis reduces the availability of positions within a segment, class self-­defence gets tougher and the ‘closed shop’ effects become more evident. Regulation of the ‘professions’ provide positional rents that may help to explain the lower degree of social fluidity in Italy. Bison (2013) has estimated children’s probability of getting a job in the same sub-­class of origin as their parents when first entering the labour market by cohort of entry, distinguishing between bourgeoisie, lower urban bourgeoisie, clerical workers and urban blue-­collar workers. In the case of the first two categories, he finds clear evidence of self-­defensive closure when growth ebbed and opportunities declined. At the same time, upward inter-­class mobility became more difficult for the lower-­middle

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 382

28/09/2016 08:24



The middle class in Italy ­383

classes. Because the relative power of one’s network of social contacts depends on how income is distributed, when distribution is more unequal, the relative power of the more privileged social networks can be reinforced, to the detriment of economic mobility. Both glass ceiling and parachute effects can be amplified, with the undesirable result that reductions in educational inequality may not be particularly effective in enhancing economic mobility (Franzini and Raitano 2013: 334). Summing up, families support their children through all the transitions they must make on the way to adulthood, including the transition to active and full-­time engagement in the labour market. A more polarized and unequal labour market makes this transition more of a challenge for some than for others and implies that family connections will matter all the more. The interactions between families, labour markets, and public policies structure a child’s opportunities and determine the extent to which adult earnings are related to family background (Corak 2013). The data on intergenerational persistence within occupations provides an indication of the low degree of social fluidity in Italian society. Indeed, there is a striking contradiction between the sweeping process of labour market deregulation and the resistance that has been put up to any attempt at liberalization of the professions. The substantial rise in the income share of the top 1 per cent, their access to sources of high-­quality human capital investment for their children and the intergenerational transmission of employment and wealth will imply a much higher rate of transmission of economic advantage at the very top, in a way that many, even – or especially – in the middle class, will perceive as evidence of inequality of opportunity. In our second case study we use the European Union Survey on Income and Living Conditions ­(EU-­SILC) database to assess the degree of intergenerational mobility among ­occupational groups.

7.  WHO BEARS THE COSTS OF THE CRISIS? As in most other peripheral countries (with the exception of Portugal), the consequences of the economic and fiscal crisis – in terms of reduction in disposable income – have fallen more heavily on the lower income classes (OECD 2014). How has the fiscal crisis affected middle-­class incomes? Government transfers – cash benefits and direct taxation – are only one part of government redistribution. Although it is usually assumed that the middle classes are far less dependent on public redistribution than the lowest income group, transfers, social services and indirect taxation have a considerable impact on the middle classes’ overall position. Thus, we need

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 383

28/09/2016 08:24

384

Europe’s disappearing middle class?

Table 9.3 Households’ market income, transfers, taxes and social contributions, and net disposable income as a share of gross income by quintiles, Italy, 2012 Quintiles of equivalent Market Monetary Gross Social Disposable market disposable incomea transfers incomeb contributions incomec income and taxes First quintile (poorest) Second Third Fourth Fifth (richest)

20.7 54.5 80.2 88.6 94.8

79.3 45.5 19.8 11.4 5.2

100 100 100 100 100

11.9 18.5 26.8 31.2 36.6

88.1 81.5 73.2 68.8 63.4

Notes: a Market income: income before taxes, contributions and transfers. b Gross income: market income + transfers. c Disposable income: gross income – direct taxation and social contribution. Source:  ISTAT (2014).

to assess how fiscal ‘austerity’ has changed net transfers and costs, as well as the availability and quality of services. Long-­term series on gross and net incomes are not available for Italy. The Bank of Italy SHIW survey provides only net incomes, while the EU-­ SILC data are only available since 2002. However, it is possible to derive the overall effect of fiscal redistribution on the various classes by putting together results on the various aspects of government activity. When considering the total redistribution carried out by the Italian system – taxes, social contributions and monetary transfers, including ­pensions – we can see (Table 9.3) that it benefits primarily the bottom and lower middle-­ income classes (two bottom quintiles), whose disposable incomes increase in relation to their market income. Conversely, it reduces the disposable income of the top quintile and the upper and core middle classes (ISTAT 2014). Thus, the scale of government income redistribution has improved the position of the lowest income groups, at the ‘cost’ of the largest part of the middle classes, though it failed to fully compensate for the growing gap between the top and the bottom. If, also because of the crisis, the inequality produced by ‘the market’ becomes too great, the task for redistribution may become too demanding and will increasingly be perceived as ‘oppressive’. In order to assess the overall redistributive effect of the welfare state on the middle classes, we should also consider the services in kind provided by the state, as well as the employment opportunities that the service provision offered to the middle classes, especially to women. While services such

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 384

28/09/2016 08:24



The middle class in Italy ­385

as higher education, provided free or at low cost, have disproportionally favoured the middle classes, others, such as health care, have had a more egalitarian role. Moreover, with the increase in female labour supply, these sectors have provided relatively good employment opportunities for the middle classes. In recent decades, female labour force participation has steadily increased, although the female employment rate is still among the lowest in the EU. The participation rate has increased largely for educated women, and for women of high-­income families, while it did not change substantially for less educated women from families in the lower deciles of the income distribution. The lack of reconciliation measures was somehow compensated by the fairly ‘flexible’ rules governing public employment: short and flexible working times, secure employment and generous leaves. The implication is that the increase in the female participation rate might have allowed several households to reach and stay in the middle classes, but it also tended to increase household income inequality. The conditions making for this arrangement have now all but disappeared: even long before the crisis, the decision to implement fiscal ‘austerity’ blocked the increase in public employment, while responding to the increased demand for services with the creation of a huge amount of precarious employment (Simonazzi 2015). The quality of public employment has dramatically worsened, in terms of pay, rights and status: primary (and higher) schoolteachers are the most obvious example. The quality of services has correspondingly deteriorated, with increasingly widespread ‘exit’ effects (Hirschman 1970) by the upper-­middle classes. For the rest of the middle classes, the provision of otherwise indispensable services – such as child and elderly care – has been increasingly carried out in a burgeoning black market, filled by female immigrants active in the care sector, while the state has turned a blind eye.

8.  CASE STUDIES 8.1  Case Study 1: Mobility Along the Income Distribution Mobility can be considered an essential feature of a meritocratic and efficient society in which economic outcomes depend on skills and equality of opportunity is guaranteed. However, mobility can also mean vulnerability and insecurity when individuals are exposed to shocks and risks. If, following Ferreira et al. (2012), we define the middle class in terms of its high level of economic stability and low level of vulnerability, a key point is to understand the capacity of individuals belonging to the middle class to remain in the same class in the face of economic shocks.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 385

28/09/2016 08:24

386

Europe’s disappearing middle class?

We use the longitudinal component of the SHIW survey to investigate the pattern of (upward and downward) mobility along the income distribution.9 The transition matrices represent the conditional probability of belonging to an income class given the position occupied in the income distribution in the previous period. We estimate the probability of transition for our income classes in three different time intervals: 2000–2002, 2006–08 and 2010–12 (Table 9.4). The elements on the main diagonal identify the share of population that stayed in the same income class. The probability of a member of the bottom-­income class remaining in the same income class has increased (from 59.8 per cent in 2000–2002 to 62.3 in 2010–12) and their probability of climbing the income distribution has correspondingly decreased. At the opposite end of the distribution, the probability of remaining in the top-­income class has increased even more (from 58.8 per cent in 2000–2002 to 73 per cent in 2010–12), and the probability of moving down has decreased. This significant improvement has occurred mainly since 2008, thus confirming our previous findings that the top-­income class has been strong enough to keep its position during the recession, and has even strengthened it. Focusing on the middle class, the upper-­middle class shows an increase in the probability of upgrading (from 11.4 per cent in 2000–2002 to 15.5 per cent in 2010–12), a decrease in the downward mobility towards the core middle class (from 23.8 per cent in the first time period to 14.3 per cent in the third time period), and a sharp increase in the level of persistence (from 60 per cent to 67 per cent). The core-­middle class shows an increased probability of remaining in the class between 2000–2002 and 2006–08 (from 52 per cent to 60 per cent), which it loses completely in the following period, when it goes back to the initial level. However, it acquires a higher probability of upgrading to the upper-­middle class and a lower probability income class. The lower middle-­ of downgrading to the lower middle-­ income class seems to share the pattern of the core-­middle class: while the level of persistence has slightly increased, the probability of downgrading has decreased and upward mobility has increased. The lower level of persistence in the last period reflects the fluidity of the chances of the people falling within this income bracket, with a higher probability of upgrading than regressing to the lower class, which surprisingly increases in the crisis. Thus, if we consider the middle class as the one that shows a high degree of stability and resilience to shocks, we can see that from the beginning of the 2000s the level of persistence has remained fairly stable for the lower-­and the core-­middle class and has increased for the upper-­middle class. Moreover, all the income classes exhibit a higher probability to move upward to the next higher class than down to the bottom of the income distribution. Comparing the 2006–08 matrix with the 2010–12 matrix we

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 386

28/09/2016 08:24



The middle class in Italy ­387

Table 9.4 Mobility among income classes, Italy, 2000–2002, 2006–08 and 2010–12 2000

2002

Bottom Lower middle Core middle Upper middle Top Total

Bottom

Lower middle

Core middle

Upper middle

Top

Total

59.83 20.21 7.38 1.35 0.55 15.28

23.33 37.94 17.34 3.51 1.10 15.20

14.00 32.16 51.82 23.83 7.42 29.76

2.00 8.66 20.98 59.89 32.14 29.29

0.83 1.03 2.49 11.42 58.79 10.46

100.00 100.00 100.00 100.00 100.00 100.00

2006

2008

Bottom Lower middle Core middle Upper middle Top Total

Bottom

Lower middle

Core middle

Upper middle

Top

Total

64.31 20.22 3.43 0.62 0.00 14.48

20.81 44.17 14.07 3.03 0.26 15.12

12.28 30.17 60.24 18.04 3.94 30.72

1.88 4.98 20.25 68.12 35.96 30.61

0.72 0.47 2.01 10.19 59.84 9.07

100.00 100.00 100.00 100.00 100.00 100.00

2010

2012

Bottom Lower middle Core middle Upper middle Top Total

Bottom

Lower middle

Core middle

Upper middle

Top

Total

62.31 14.92 5.33 1.09 1.06 14.14

22.75 40.92 11.61 1.57 0.21 13.36

12.92 37.52 52.62 14.30 3.82 26.87

1.88 6.35 29.01 67.51 21.87 32.79

0.13 0.30 1.43 15.53 73.04 12.84

100.00 100.00 100.00 100.00 100.00 100.00

Source:  Authors’ elaboration on SHIW.

can also derive some hints on how the middle class reacted after the crisis. Contrary to the previous period, when all the middle classes increased their level of persistence, from 2006–08 to 2010–12 the upper-­middle class was the only one able to maintain its high level of persistence. However, the other middle classes traded a lower level of persistence with a higher level of upward mobility. Our analysis differs from a recent work on middle class mobility in Italy (Ricci 2015),10 which portrays a general

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 387

28/09/2016 08:24

388

Europe’s disappearing middle class?

impoverishment and increased immobility for all classes. Looking at the determinants of upward and downward mobility, it finds that after 2008 the middle class’s probability of moving downward in case of negative events has significantly increased. 8.2  Case Study 2: Intergenerational Mobility The 2011 wave of the EU-­SILC includes a specific module about the intergenerational transmission of disadvantages. This section presents information on family background – for example, parents’ educational attainments and occupations – referring to the period when the interviewee was about 14 years old. Following Raitano and Vona (2014), we use the highest occupation of the two parents as a proxy for family background and analyse its effect on the occupational attainments of their offspring by means of a mobility table of parents’ and children’s occupations. For both children and parents, we rank the ISCO categories into five groups: (1) managers (excluding hospitality, retail and services managers, who, differently from the other managers, have an average income below the twenty-­fifth ­percentile); (2) professionals (whom we keep separate from managers even if their average income is above the twenty-­fifth percentile, in order to take into account the peculiar role the professions have in Italy); (3) medium jobs, such as managers in hospitality and retail, associate profession­ lue-­collar als, technicians and clerks; (4) services and sales workers; (5) b workers (craft, elementary occupations and skilled agricultural). We restrict our sample to individuals aged 35–50 since, according to the literature, the process of transition should be completed for these workers. Table 9.5 presents the composition of our income classes according to occupational groups. The top-­income class has the highest proportion of individuals in top occupations (managers and professionals), confirming that occupation is a proxy for an individual’s social and economic status. The percentage of people employed in middle occupations, such as technicians and clerks, is higher in higher-­income classes. The opposite is true of services and blue-­collar workers, with the shares decreasing with income. Table 9.6 presents the mobility matrix, whose elements represent the conditional probability of being in an occupation given the parental occupation when the individual was 14 years of age. Looking at the two marginal distributions (last column and last row, respectively) we notice that the proportion of blue collars has decreased, whereas the proportion of services workers and middle occupations has increased. If we consider managers and professionals together, the proportion of top occupations has remained almost constant, although the internal composition

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 388

28/09/2016 08:24



The middle class in Italy ­389

Table 9.5 Income classes by occupational group, Italy, 2011 (percentage of total) Income class

Occupation Managers Professionals

Bottom Lower middle Core middle Upper middle Top Total

2.02 1.93 2.37 3.13 10.16 3.29

Associate professionals, technicians and clerks

Service and sales workers

Blue-­collar workers

13.21 16.38 24.26 34.99 34.49 25.96

21.53 18.59 16.91 14.02 9.20 16.13

59.89 57.68 47.98 31.23 16.62 42.61

3.35 5.41 8.49 16.63 29.52 12.00

Source:  Authors’ elaboration on EU-­SILC 2011.

of the top occupations has changed: the percentage of professionals has increased whereas the proportion of managers has decreased. The elements along the main diagonal identify the share of population that ended up in the same occupational group of its parents. The values on the diagonal are distant from those we would have in the case of perfect mobility, hinting at a high level of intergenerational persistence of occupation. For all the occupational groups except the managers’ group, a higher proportion remains in the same occupational group as their parents. The persistence is particularly high for blue-­collar workers, whose children have a 50 per cent of probability of staying in the same occupation of their parents. Occupational persistence is also high among professionals and associate professionals. This could be linked to the entry barriers that still limit access to certain professions, as argued in section 6. It is also interesting to note the low probability of managers’ children doing the same job as their parents, with a 30 per cent probability of downgrading towards middle-­income professions (which might be explained by age).

9.  CONCLUSION AND POLICY ISSUES Our analysis of the Italian middle classes seems to conform, with some qualifications, to the international results that have found evidence of a decrease in the rate of growth of median family income and an impoverished middle class (with the possible exception of the upper-­middle class),

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 389

28/09/2016 08:24

390

VAUGHAN 9781786430595 PRINT (M4051).indd 390

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:24

Professionals 22.84 40.04 18.03 10.78 7.43 13.22

Managers 9.44 3.82 2.98 2.97 2.27 3.1

Source:  Authors’ elaboration of EU-­SILC 2011.

Managers Professionals Associate  professionals, technicians, clerks Service workers Blue-­collar workers Offspring’s  distribution

Parental occupation

28.63 23.3 28.92

31.78 32.02 45.01

Associate professionals, technicians, clerks

21.2 14.71 21.59 27.49 50 38.02

30.13 17 16.75

Blue-­collar workers

14.74 9.41 12.4

Service and sales workers

Offspring’s occupation

Table 9.6 Intergenerational mobility among occupational groups, Italy, 2011

9.9 57.71 100.00

7.24 7.61 17.55

Parents’ distribution



The middle class in Italy ­391

to the benefit of the richest fifth (Pressman 2007; Atkinson and Brandolini 2011). Over the whole period under examination, all income classes recorded a growth in equivalent disposable income, but the ‘core-­middle class’ gained the most and the bottom-­income class virtually nothing. The current crisis has hit all classes, but once again unevenly. The remarkable increase in inequality seems to be driven by the trend at the bottom of the distribution. Our findings suggest that, in the wake of the economic crisis, the increasing income inequality in Italy has been accompanied by a declining middle-­ income class and an increase in polarization. Labour market reforms and increasing precariousness of jobs, long-­term stagnation in real income, fiscal ‘austerity’, with its effects on public employment and welfare, and, finally, the economic crisis have hurt the lower-­middle class more heavily. Middle-­class discontent is certainly related to these factors, but our analysis of cohort effects suggests that these changes have affected the middle classes also by souring their expectations about their offspring’s future: perceived reduced opportunities for younger generations and a decreased intergenerational social mobility. More reshuffling is to be expected in the future; decreasing intergenerational mobility may lead to greater inequality within the middle classes, which the persistence of the fiscal crisis and the austerity measures will only worsen.

NOTES   1. Pasolini had warned that it was not the birth of a new bourgeoisie, but the bourgeoisification of Italian society. Giuseppe de Rita, the influential director of CENSIS, an Italian think tank, invented the term ‘cetomedizzazione’, which can be translated as ‘bourgeoisification’ and has a derogatory meaning. It refers to the huge process of convergence towards the middle of the entire society, obtained by the drifting down of the upper class and rising up of the lower classes, the glue being provided by mass media and mass consumption (interview with Lietta Tornabuoni, La Stampa, 14 January 1995, p. 15).   2. See note 1.   3. All the results presentedchapter in this chapter 3 are expressed in 2012 prices.   4. We compare three different definitions of households’ disposable income: total disposable income and equivalent total disposable income, equivalized using two different scales, the OECD modified equivalence scale and the square-­root of family size equivam lence scale. Because inequality measures can be quite sensitive to extreme values, we also P 5 (T 2G) m , performed a top/bottom coding procedure: we set all values that were less than zero to zero and all values that were greater than ten times the median to ten times the median. This analysis is of course sensitive to the treatment of outliers.   5. While the Gini index summarizes chapter 9 the income distribution with only one value, Kernel density estimation provides a picture of the entire (log) equivalent disposable income distribution: its location, spread and modality.   6. We use an ‘adaptive’ Kernel approach with varying bandwidths. m   7. The index is defined as: P 5 (G B 2G w ) m where GB is the between groups Gini index, Gw

| F

R 5 [z , z ] File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 391

28/09/2016 08:24

392

Europe’s disappearing middle class?

is the within groups Gini index, μ is the overall mean and m is the median. An increase in inequality between the two groups (the one above and the one below the median) raises polarization, whereas the increase in inequality in each group decreases polarization, since each group is less homogeneous. 0 means no polarization and 1 indicates the maximum level of polarization.   8. Caution should be used with estimates of wealth, because they may reflect also subjective family opinions on the dynamics of real estate prices: the increase from 2008 to 2010 may be due to families’ delayed adjustment of their perceptions to market price dynamics.   9. As in the previous section, we consider total household disposable income, equivalized with the OECD modified equivalence scale. 10. In this study lower, middle and upper classes are identified following the procedure of Esteban et al. (2007).

REFERENCES Amable, B. and S. Palombarini (2014), ‘The bloc bourgeois in France and Italy’, in H. Magara (ed.), Economic Crises and Policy Regimes, Cheltenham, UK and Northampton, MA, USA: Edward Elgar, pp. 177–216. Atkinson, A.B. and A. Brandolini (2011), ‘On the identification of the “middle class”’, Working Paper WP 217, ECINEQ, Society for the Study of Economic Inequality, Verona. Bagnasco, A. (2004), ‘Quasi poveri e vulnerabili’, Il Mulino, 53 (2), 278–89. Berloffa, G., F. Modena and P. Villa (2014), ‘Changing labour market opportunities for young people in Italy and the role of the family of origin’, Rivista Italiana degli Economisti, 19 (2), 227–51. Bison, I. (2013), ‘Le classi medie: definizione, mobilità e declino nel caso italiano’, Società, mutamento, politica, 4 (7), 155–83. Brandolini, A., P. Cipollone and P. Sestito (2001), ‘Earnings dispersion, low pay and household poverty in Italy, 1977–1998’, in D. Cohen, T. Piketty and G. Saint-­ Paul (eds), The Economics of Rising Inequalities, Oxford; Oxford University Press, pp. 225–64. Chauvel, L. (2013), ‘Welfare regimes, cohorts, and the middle classes’, in J. Gornick and M. Jantti, (eds), Income Inequality: Economic Disparities and the Middle Class in Affluent Countries, Palo Alto, CA: Stanford University Press, pp. 115–41. Corak, M (2013), ‘Income inequality, equality of opportunity, and intergenerational mobility’, Journal of Economic Perspectives, 27 (3), 79–102. Esteban, J., C. Gradín and D. Ray (2007), ‘An extension of a measure of polarization, with an application to the income distribution of five OECD countries’, Journal of Economic Inequality, 5 (1), 1–19. Ferreira, F.H., J. Messina, J. Rigolini, L. López-­Calva, M.A. Lugo and R. Vakis (2012), Economic Mobility and the Middle Class: Concepts and Measurement in Economic Mobility and the Rise of the Latin American Middle Class, Washington, DC: World Bank. Foster, J.E., and M.C. Wolfson (2010), ‘Polarization and the decline of the middle class: Canada and the US’, Journal of Economic Inequality, 8 (2), 247–73. Franzini, M. and M. Raitano (2013), ‘Economic inequality and its impact on intergenerational mobility’, Intereconomics, 48 (6), 328–35.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 392

28/09/2016 08:24



The middle class in Italy ­393

Hirschman, A.O. (1970), Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States, Cambridge, MA: Harvard University Press. Istituto Nazionale di Statistica (ISTAT) (2014), Rapporto Annuale 2014, La Situazione del Paese, Rome: Istituto Nazionale di Statistica. Istituto Nazionale di Statistica (ISTAT) (various years), Famiglia e soggetti sociali (Families and Social Subjects) Rome: ISTAT, accessed 28 April 2016 at http:// www.istat.it/it/archivio/81546. Manacorda, M. (2004), ‘Can the Scala Mobile explain the fall and rise of earnings inequality in Italy? A semiparametric analysis, 1977–1993’, Journal of Labor Economics, 22 (3), 585–613. Mannheim, K. (1952), ‘The problem of generations’, in P. Kecskemeti (ed.), Essays on the Sociology of Knowledge by Karl Mannheim, New York: Routledge and Kegan Paul, pp. 279–322. Massari, R., M.G. Pittau and R. Zelli (2009), ‘A dwindling middle class? Italian evidence in the 2000s’, Journal of Economic Inequality, 7 (4), 333–50. Mocetti, S. (2014), ‘Dynasties in professions: the role of rents’, Temi di Discussione, Working Paper 995, November. Naticchioni, P., M. Raitano and C. Vittori (2014), ‘La meglio gioventù: earnings gaps across generations and skills in Italy’, IZA Discussion Paper 8140, IZA, Bonn. operation and Development (OECD) (2014), Organisation for Economic Co-­ ‘Rising inequality: youth and poor fall further behind’, Income Inequality Update, Paris, June. Petrarca, I. and R. Ricciuti (2015), ‘Relative income distribution in six European countries: market and disposable income’, Luxembourg Income Study (LIS) Working Paper Series No 629, Luxembourg. Pisano, E. and S. Tedeschi (2007), ‘Tendenze della distribuzione dei redditi in Italia e impoverimento della classe media: percezione o realtà?’, Meridiana, (59–60), 131–55. Pressman, S. (2007), ‘The decline of the middle class: an international perspective’, Journal of Economic Issues, 41 (1), 181–200. Raitano, M. and F. Vona (2014), ‘Direct and indirect influences of parental background on children’s earnings: a comparison across countries and genders’, The Manchester School, online, 8 May, DOI: 10.1111/manc.12064. Ricci, C.A. (2015), ‘Middle class income dynamics and mobility: an Italian ­perspective’, paper presented at the sixth ECINEQ meeting, Luxembourg, 13–14 July. Rosolia, A. and R. Torrini (2013), ‘The generation gap: a cohort-­level analysis of earnings levels, dispersion and the role of initial labor market conditions in Italy, 1974–2010’, ECINEQ Conference, Bari, 22–24 July, accessed 28 April 2016 at http://www.ecineq.org/ecineq_bari13/FILESxBari13/CR2/p163.pdf. Simonazzi, A. (2015), ‘Italy: continuity and change in welfare state retrenchment’, in D. Vaughan-­Whitehead (ed.), The European Social Model in Crisis – Is Europe Losing its Soul?, Cheltenham, UK and Northampton, MA, USA and Geneva: Edward Elgar and ILO, pp. 339–85. Sylos Labini, P. (1974), Saggio sulle classi sociali, Bari: Laterza.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 393

28/09/2016 08:24

P 5 (G 2G ) m

chapter 3 chapter 3 | chapter 3 F

P 5 (G B 2G w ) m m

w m P 5 (G B 2G P 5 ()Tm 2G) m ,

| F

chapter 3 | F

m

P 5 (T 2G) mP, 5 (T 2G) m , mm P 5 (T 2G) m ,

394

disappearing middle class? R 5 [z , z ] Europe’s [z , z ] R5

chapter 9 m

P 5 (T 2G) m , m P 5 (G B 2G w ) m

chapter 9 chapter 9 [z , z ] R 5MIDDLE APPENDIX 9A.1 MEASURING THE chapter 9 CLASS: THE FOSTER AND chapter 9 0 # z # 1 # z. 0 # z # 1 # z. | WOLFSON METHODOLOGY m F m P 5 (G 2G ) B

w

m (G B 2G w ) | |(| )] | P5 w mm ( z) 1M (z ) 5 [F (1) 2F (0 [F| (|z1()12F ) 5M Mf # # # z.1 ( z )]m1[F| ( z ) 2F|(1)] )z5 (z1 [F ) 2F (G(BR )m f Mff (R) 5Mf ( z) 1Mzf)] P5 2G ( P 5 G B 2G w ) m the M-­curve

Deriving | | | Mf (R) 5Mf ( z) 1Mf (z ) 5 [F (1) 2F ( z )]1[F ( z ) | chapter 3 | , z ]first norF R 5 [zwe F Since we want Mf ( z ) distribution functions, |Mf ( zto) compare different F chapter 3 | malize the functions with regard to the median. Let F be the median-­ M (z) normalized income distribution function of thef distribution F andm let P 5 (T 2G) m , ,zq] ] ]Q [ q,interval R 5 [z , z R be whose [ q , qo]ff points satisfy 0 # z # 1 # z. The Q 5 cut-­ [z 55an ] )m , R 5size middle class therefore represented by the proportion of P [z 5 (,is Tz2G ,individuals z ] ( z) 1M (z ) 5 [F| (1) 2F| ( z )] q # 0.5 # qm qR] (5 ,Mf )[z q #define 5 [ qindex Rfor 5M 0.5 # aq middleQclass f belonging to this range. We now F that givesf chapter 9 chapter 3 the population the range R: q # 0.5 # q z # 1of 0 #share zz. # 1 # zin 0##chapter . 9 z 0 # # 1 # z. | ( z )z # 1 # z | |)] [ | (| ) |( |)] M 0f# . | 5 ) 1M [F (1)(z2F ( zR F |z( z2F Mf (R) 5M f (z()z5 ) 1M ) 5([z| )]1[1| )](G B 2G w ) mm (11)m 2F ( z ) 2F f ( ) 5M F F Mf |((P1)] f P5 f ) ( 2G , T ) ) ( )] ) ( [ ( ) [ ( ( ) ( 1M 5 2F z 1 2F 1 z F 1 F z m Mf R 5Mf z | | f m Mf (R) 5Mf ( z) 1Mf (z ) 5 [F (1) 2F ( z ) 5)(G B 2G w ) m MPf (z Mf (z ) Q 5 [ q ,|q ] Mf ( z ) M (z) F ) (z Where Mff( z ) is the ‘lower middle class’ and is the ‘upper middle chapter 9 Mf qM z) # q #f (0.5 | class’. M F(R) , function Mf (R) , we obtain the M-­curve that measures If [we q , qgraph ]Q 5f [ qthe Q5 ,q ] w m B )m P 5 (G 2G the concentration the median of (the income ] mass around Q 5 [ q , q of R distribution 5 [z , z ] M f R) , Q 5 [ q , q ] q # does 0.5 #not q and depend on an arbitrary definition of the middle class. q # 0.5 # q | be5the z] #[zq,median-­ | Now letq # in G R0.5 q # 0.5function Gnormalized income distribution #q ) (z M period 2 and compare it with F| in period 1. We can state that F has an f | 0 # z # 1 # z. G unambiguously larger middle class than G, irrespective of which cut-­off z # 1 # ifz.and | points are actually = 1,. . ., n and ( ) [ |( ) (R) 5M ( ) | only if Mf(Ri) $ Mg(Ri) 4 i Mf F 0 # chosen, f z 1Mf z 5 F 1 2 F Mff(R ) > M (R ) for some i. It is clear that the requirements that permit us (z )i ( ), | i) R 5( )[z , [zF|] (1) 2F||( z )]1M Mg (z (1)] [F|f( zR) 2F ) 5Mf ( z) 1M Mf ) (R to state that one distribution hasf zan5unambiguously larger) middle class Mff(z F Mf (zM ( z ) than another are quite strong. Therefore it is not always possible to make f an unambiguous ranking between two distributions. | ) ( z Mf (R) , M (M G 0 # z # 1 # z. R)f , (Rc)urve MffS-­ , Deriving the M|(R) ,5 | [ q(,zq)]]1[F| ( z ) 2F|(1)] Mf (R) 5Mf ( z) 1Mf (z ) 5 [Ff (1Q) 2F | | q # 0.5 # q ), Let | Q 5 [ q , q ] for a given population range, (with G us define F G | the first degree polarization curve S (q) for distribution F G q # 0.5 # q | measures the M (z) f width of the income range – thatf is, the income spreadG– given a population | range. If we compare two income distributions – for example, F and F – F hasF|a greater income spread than G, hence greater polarization, if G | F [ q , q ]n and S (Q ) > F|S (Q ) for some i. and only if Sf(Qi) ≥ Sg(Qi) 4 iQ=51,. . ., f i g M i f (z ) If this statement holds, since a qgreater income spreads means less propor# 0.5 # q Mf (z )around the middle, G has a larger middle-­income class, tion of population irrespective of the chosen population range. The first degree polarization Mf (R) ,  

Mf (R) , Mf (z )

| G

Mf (R) ,

| F

| G File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

|

VAUGHAN 9781786430595 PRINT (M4051).indd F 394

28/09/2016 08:25



The middle class in Italy ­395

curve of one distribution is of course higher exactly when its M-­curve is lower than the other distribution’s M-­curve. Deriving the B-­curve The second degree curve B(Qi) cumulates income spreads from the middle to the top and from the middle to the bottom of the income distribution, and more weight is placed on changes around the middle of the income distribution. The income distribution F is characterized by more polarization than G (and thus by a smaller middle class) if, and only if, B(Qi) ≥ Bg(Qi) 4 i = 1,. . ., n and FBf(Qi) > Bfg(Qi) for some i.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 395

28/09/2016 08:25

10. Stagnating incomes and the middle class in the Netherlands: Running to stand still? Wiemer Salverda1 1. INTRODUCTION In the Netherlands, the public debate about the possible decline of the middle class is inspired primarily by international debates, which started in the United States. Domestically, there is some complacency concerning relatively low poverty and income inequality. At first sight, this may seem justified in relation to the present issue, because (as we will see) the majority – over 80 per cent – of Dutch households belong to the middle class when it is measured as between 60 and 200 per cent of net-­equivalized median household income (as chosen for this collective work). In addition, this level shows very little change over time. However, this focus on net-­equivalized incomes – the common currency of the European poverty rate as well as the Organisation for Economic Co-­operation and Development (OECD) Income Distribution Database – ignores the important roles played by the acquisition of market incomes by households, on the one hand, and the redistribution and equivalization of those incomes, on the other hand. The personal, independent acquisition of market incomes should be at the core of any analysis as that income lends the middle class its independent basis in society and builds an element of autonomy into its definition. Evidently, this is essential also when linking the middle class to the world of work, a central aim of this volume. For example, it is this footing that is at stake in the current discussion about the polarization of the job structure and in the fears about the weakening of the middle class as the political backbone of society. Accordingly, this chapter starts by considering the distribution of gross incomes, which encompasses all possible sources of income. Starting here makes it possible to address the question how both the redistribution and the equivalization of household incomes affect the size and evolution of the middle class; it also examines the role of combining individual incomes into household incomes. In this way, first, earnings from labour as a source of 396

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 396

28/09/2016 08:25



Stagnating incomes and the middle class in the Netherlands ­397

income and its potential links to class can be examined. The public debate about the nature of top incomes underlines the importance of this – and gross incomes are the currency of that debate. Second, it is then also possible to study the effects of the demise of the single-­breadwinner model and the ascent of the household with more than one earner in the labour market. Owing to this alteration, economic efforts and incomes are sorted rather differently now in terms of households than they were previously, concentrating on dual-­and multiple-­earners and potentially ousting some (persons and households) from the world of work which in the single-­ earner world might still have been part of it. These issues seem vital to an up-­to-­date consideration of the middle class, and they certainly are when focusing on the world of work. Third, proceeding from gross to disposable incomes allows us to analyse the role of income redistribution as a support of and/or a drain on the middle class, that is, the extent to which redistribution acts as a prop for the middle class. Fourth, the equivalizing adjustment of disposable income for household size and composition enables us to consider the family composition of the middle class and the effects of household formation. This formation has changed markedly in recent decades as the number of singles has multiplied and larger households have thinned as a result of declining or postponed fertility.2 Section 2 provides some general background with a brief sketch of aggregate developments in the Dutch economy, the labour market and household demographics, and the income distribution since 1990. Section 3 presents important aspects of the evolution of the middle class vis-­à-­vis the other classes since 1990. Section 4 details the effects of transiting from gross incomes in two steps, first via taxation to disposable incomes and then via equivalization to net-­equivalized incomes: how do households move either away from or into the middle class as a whole or one of its ­segments? Section 5 discusses the sources of income that link to the world of work. Section 6 presents two case studies, one about the role of two-­earner households, which is highly relevant for the world of work, and the other about the wealth and debt of the middle class, which has a wider economic significance. Section 7 concludes with a brief discussion and summary of the findings and suggestions for further research. Appended to the chapter is a short note explaining the data used and the corresponding caveats.

2. GENERAL BACKGROUND: ECONOMY, EMPLOYMENT AND INCOME The evolution of the middle class since 1990 is best understood against the backdrop of trends in areas of general relevance, such as production,

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 397

28/09/2016 08:25

398

Europe’s disappearing middle class?

110 105 100 95 90

Private consumption

85

Gross capital formation

GDP

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

1994

1993

1992

1991

1990

80

Note:  Volume figures. Unweighted average, excluding Estonia as extreme economic growth distorts the average. Source:  OECD (2015) with some additions from the EU Labour Force Survey (LFS), http://ec.europa.eu/eurostat/web/lfs/data/database.

Figure 10.1 Dutch economic performance relative to average of other countries, 1990–2013 (1991 = 100) consumption and investment, and employment. The evolution of households and the distribution of their incomes are also addressed as an equally important background. Figure 10.1 compares three Dutch macroeconomic outcomes with the average of the other countries covered in this volume. This is done on a per capita basis to control for diverging population growth over time and across countries. Relative fluctuations appear to be considerable and to differ significantly between the variables. Gross domestic product (GDP) growth is a bit better in the early 1990s, when the downward economic cycle hit the country less. It does much worse, however, after the dot.com crisis but makes a comeback after that. The most recent trend looks similar to the rest. Investments follow roughly the same path of ups and down followed by GDP, but with much larger fluctuations. The dot.com crisis hits investments very strongly and, as a result, they lag more than 15 per cent behind those of other countries. Household consumption lags behind during the first few years but expands during the second half of the 1990s. It is also hit by the dot.com crisis, in fact more strongly than GDP; after that, consumption does not follow the upswing of GDP between 2004 and 2009 but remains flat at a much lower level, from where it develops largely in parallel with the other countries. Households form the fundamental unit in the present analysis and it is

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 398

28/09/2016 08:25



Stagnating incomes and the middle class in the Netherlands ­399

85 80 75 70 65 60

Persons

55

2013

2012

2011

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

1994

1993

1992

1991

1990

2010

Households

50

Note:  Persons and households ages 15–74 years. Source:  Statistics Netherlands, Income statistics for Household rates; Statistics Netherlands, Population statistics; and Eurostat, Labour force survey for personal rate.

Figure 10.2 Employment rate: households versus persons, the Netherlands, 1990–2013 (percentage) important to see how they have developed. Their absolute number increases between 1990 and 2013 by more than 20 per cent, from 6.1  million to 7.5 million. From a comparison with the number of persons a considerable shrinking of household size can be deduced: the average number of persons per household declines by 8 per cent, from 2.39 to 2.20. This is caused by the growing share of singles, which increases from 32 per cent to 36 per cent of all households, and a 6 per cent decline in the average number of persons in the other, non-­single households, from 3.05 to 2.88. The upshot of the recent literature on household joblessness is that the level and evolution of the employment rate on a household basis may differ from the commonly used rate on a personal basis, owing to increasing numbers of earners in a household. Both rates are compared in Figure  10.2. Clearly, the personal rate increases strongly in one long shot up to 2002, from 55 per cent to 69 per cent, followed after 2008 by a decline of 4 per cent. The employment growth is overwhelmingly in part-­time jobs. In comparison, the household rate remains basically flat, fluctuating between 81 and 85 per cent, and ends roughly where it started. Apparently, from the perspective of household participation in employment these are largely years of stagnation. The growing gap between the two rates implies an increasing concentration of persons with an income from the world of work in households where there is already such a person. The long-­run rise in part-­time employment runs parallel to this increasing concentration

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 399

28/09/2016 08:25

400

Europe’s disappearing middle class?

200 180 160 140

(000s)

120 100 80 60 40 20

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

0

1

2

3

4

5

Median

6

7

8

9

10

MC

Note:  MC = middle class. Source:  Statistics Netherlands, income statistics and Consumer Price Index (CPI).

Figure 10.3 Average incomes of households in deciles and middle class range, the Netherlands, 1990–2013 (€ 2013) of employment in households. Note that the household rate concerns any income from labour or enterprise, as does the personal rate. Figure 10.3 presents the income distribution with the help of average gross real incomes across all ten deciles.3 This is quite literally the backdrop to the sketch of the middle class that will follow, as the income range over which this class extends is shown as the shaded grey area. The space below that area indicates the bottom class, while the incomes above belong to the top class. Clearly, the income distribution is characterized by a massive long-­term stagnation of real incomes in the bottom half of the distribution and rapid growth at the very top. It also shows a large and growing distance in absolute and relative terms between the highest decile and the ninth, and

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 400

28/09/2016 08:25



Stagnating incomes and the middle class in the Netherlands ­401

a large relative, but not absolute, distance between the first and the second decile (50 per cent) at the bottom of the distribution. However, mutual differences between the other deciles seem small and steady by comparison. The figure stimulates curiosity about a possible increase in the aggregate level of inequality. No Gini coefficient is available for gross incomes over the period as a whole. However, the S10/S1 ratio between the average incomes (or income shares) of the first and tenth deciles, which measures the inequality across the two sensitive tails of the distribution, grows between 1990 and 2013 from 12.8 to 16.6, or by 23 per cent (excluding 2000–2001). The median income or fifth-­decile upper cut-­off, which provides the basis for delineating the middle class here, is also shown, as the dashed line. Over the period it declines by 4 per cent (again excluding 2000–2001). Note that the other two medians develop somewhat more favourably: the net median income remains unchanged, while net-­equivalized median income increases by almost 9 per cent. Apparently, there are positive effects of income redistribution and household formation. In summary, we find no outstanding economic performance for the Netherlands relative to the other countries, while household consumption and investments lag behind. The employment rate for persons, which does much better than elsewhere, reflects the growth of part-­time jobs, while the household employment rate stagnates, pointing to an important redistribution of employment over households. In real terms, gross incomes remain virtually unchanged in the lower half of the distribution, while the top runs away and inequality grows. This provides a discomforting starting point for examining the situation of the middle class.

3. COMPARATIVE EVOLUTION OF THE MIDDLE CLASS This section examines the aggregate evolution of the middle class since 1990, discusses characteristics of household size and composition, and elaborates on the segments of the middle class. Characteristics pertaining to the world of work will be discussed in section 5. 3.1  Size and Income Before starting out from the distribution of gross household incomes it is important to realize that these include market incomes, as well as benefits and transfers. That is, one essential aspect of the redistribution of incomes is already included, while transiting from gross to net incomes brings in the other, equally essential side: the taxation of income and payment of

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 401

28/09/2016 08:25

402

Europe’s disappearing middle class?

90 80 70 60 50 40 30 20 10

200%

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

1994

1993

1992

1991

1990

0

Note:  ex P = benefits and transfers excluding old age public pensions and occupational pensions. Source:  Statistics Netherlands/Statline, income statistics, author’s calculations.

Figure 10.4 Benefit and transfer gross income shares within classes, the Netherlands, 1990–2013 (percentage) social contributions (‘taxation’ for short), which serve to finance those benefits and transfers (and, naturally, also government and public services). Figure 10.4 shows the wide variation in the role of redistribution across the classes of gross income. In the bottom class redistribution encompasses the large majority of all gross incomes, deviating strongly from the other two classes. Within the bottom the share declines from 82 per cent to 75 per cent over the 1990s and remains stable with cyclical variation afterwards. Within the middle the share is roughly stable over the 1990s but increases considerably – from 21 to 32 per cent between 2001 and 2013. At the top the share declines slightly from 10 per cent to 8 per cent until 2000, only to rise to the same extent during the second period. Notably, this picture

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 402

28/09/2016 08:25



Stagnating incomes and the middle class in the Netherlands ­403

lumps social benefits and assistance together with old-­age pensions, which combine the basic public pension (Algemene Ouderdomswet – AOW) with the occupational pension a household may receive. For the latter the Netherlands has built the most extensive capital-­ funded system worldwide (which we come back to later), which bears responsibility for the improving outcomes. It seems incorrect to treat these pensions as social benefits as they result from individual savings from earnings and the returns from their investment. The figure also shows the role of the social benefits when excluding pensions – it is a tentative picture as, unfortunately, AOW and the occupational pensions could not be differentiated. In all three classes, the importance of benefits has more than halved. Interestingly, their share in the middle class is unchanged while apparently pensions are responsible for the growth over the 2000s. We return to this side of the redistribution in section 4 to draw the full picture in conjunction with taxation, while in the rest of this section the focus is on the trajectory from gross to net to net-­equivalized incomes. Of particular importance for examining the middle class is the perspective taken with regard to the types of income – gross, net or net-­equivalized – which define the ‘income space’ within which the classes are measured. The income type determines the median, as well as the density of the distribution around it. Figure 10.5, panel A, depicts the size of the middle class, ranging between 60 and 200 per cent of the median. This is done with the help of its shares in total numbers and incomes of households. Three radically different outcomes are found depending on the type of income, both over time and in cross-­section – the two go together but have to be addressed separately. For ease of comprehension, ‘change’ and ‘evolution’ will refer here to developments over time, while ‘shift’, ‘flow’ and ‘gap’ refer to the cross-­section comparisons. The differences are large and persistent, and diverge over time. In gross-­income space a relatively small and strongly declining size is found for the number of households, as well as incomes. The middle class shrinks from 60–65 per cent at the start in 1990 to only 50–55 per cent at the end in 2013. In net-­income space the middle shows, again, a decline, but this time it is smaller – from around 70 to around 65 per cent. In net-­equivalized income space, by contrast, the income and number levels are much higher, at about 80 per cent, and they remain also virtually unchanged over the entire period. Thus, in cross-­section the size of the middle class grows larger with each of the two subsequent steps taken from gross to net (taxation) and from net to net-­equivalized income. Over time, the middle class declines significantly for gross income, and also, but less, for net income, but not at all for net-­ equivalized income. These substantial divergences clearly demonstrate the importance of the mechanisms that underlie the income types. Most of

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 403

28/09/2016 08:25

404

Europe’s disappearing middle class? a) Middle class 90

75

45

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

60

Gross HH

Disposable HH

Net-equivalized HH

Gross INC

Disposable INC

Net-equivalized INC

b) Bottom class 45

30

0

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

15

Gross HH

Disposable HH

Net-equivalized HH

Gross INC

Disposable INC

Net-equivalized INC

c) Top class 45

30

0

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

15

Gross HH

Disposable HH

Net-equivalized HH

Gross INC

Disposable INC

Net-equivalized INC

Note:  HH – household frequency or numbers share; INC – income share. Source:  Statistics Netherlands/Statline, income statistics, author’s calculations.

Figure 10.5 Class shares of numbers and incomes of households, by type of income, the Netherlands, 1990–2013 (percentage)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 404

28/09/2016 08:25



Stagnating incomes and the middle class in the Netherlands ­405

the action seems to occur after 2001, while the 1990s show much flatter trends. This is reflected – albeit in different directions – by the evolution of the two complementary classes, the poor and the rich, shown in panels B and C. The size of the bottom class is substantially reduced when moving from gross to net-­equivalized incomes, but it remains stable over time in all respects. In contrast, the top class shows strong growth over time for gross incomes (from 25 to 40 per cent of incomes), less growth for net income (21 to 29 per cent) and stability for net-­equivalized incomes (15 to 16 per cent); top-­class household numbers show a similar pattern, shifting from growth to stability but at much lower levels than for incomes. The complete picture suggests very important shifts of households in cross-­section between the three classes as a result of taxation and household formation – the subject of section 4. Reconciling the different outcomes is not simply a matter of the same households found in the middle class in gross-­income space (55 per cent – the lower line in Figure 10.5, panel A) obtaining these higher shares when the taxation and equivalization of their incomes are accounted for. This ‘travel’ from one income space to another of given households, steadily ranked by their gross incomes, has hardly any effect on the size of the middle class. It extends its income share by a few percentage points only (2 to 4 percentage points), and the outcome is nowhere near bridging the almost 30 percentage points gap to the upper lines in panel A in 2013. Evidently, most of the action must reside in shifts of households between the top and middle classes, trading places owing to taxation and equivalization. 3.2  Household Characteristics We have already emphasized the importance of household size as a driving force behind the effects of income equivalization. Indeed, a significant gradient is present across the classes. In cross-­section, the average middle-­ class household currently comprises 2.3 persons, as against more than 3.1 in the top class and only 1.4 persons in the bottom class, a difference of 0.8 and 0.9, respectively, with regard to the middle. Though over time these numbers are declining slowly owing to the shrinking size of households, the mutual differences between the classes are mostly stable. This stability goes together with considerable changes in the aggregate shares of the classes in the total number of persons (Figure 10.6). A sizeable decrease over time occurs in the share of the gross-­income middle among the total number of persons (69 per cent to 57 per cent). At the same time, the corresponding share of the top increases (15 per cent to 26 per cent) and the bottom share remains largely stable (18 per cent). The population of persons has clearly become more concentrated at the top of the gross-­income distribution,

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 405

28/09/2016 08:25

406

Europe’s disappearing middle class?

100 60–200% Net-eq 80

60–200% Net 60–200% Gross >200% Gross

60

200% Net

40

200% Net-eq

20

2012

2010

2008

2006

2004

2002

2000

1998

1996

1994

1992

1990

0

11%

–40

1 5 –7

53% > 67% > 83%

Coming on taxation

Coming on equivalization

Going on taxation

Going on equivalization

17% > 11% > 6% Staying

b) Income of households 100

Percentage of all households

80 60

66 50

40 20

14

14

19 0 –19 –20

–15

1

2

–4

–3

17 –2

15 1 –14

52% > 67% > 81%

40% > 27% > 16%

–11 8% > 6% > 3%

–40 Coming on taxation

Coming on equivalization

Going on taxation

Going on equivalization

Staying

Source:  Eurostat SILC, available at http://ec.europa.eu/eurostat/web/income-and-livingconditions/overview, author’s calculations on microdata made available by Eurostat.

Figure 10.10 Class shifts from gross to net to net-equivalized incomes, the Netherlands, 2010 (percentage of all households)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 411

28/09/2016 08:25

412

Europe’s disappearing middle class?

the top. If only because of the sheer size of the middle class, there are basically no households that manage to jump over it from the rich to the poor or vice versa. Conversely, among the poor and the rich, large swathes in gross-­income space leave their classes to the advantage of the middle class, while very few join them. Panel B presents similar detail for the corresponding flows of income. The income picture is different from the numbers picture, primarily at the two tails, but shows little difference in the middle. Understandably, the bottom has a much lower initial share (8 per cent) and the income movements in and out are also much smaller. The share and the movements are much larger at the top: a combined total of 25 per cent of all incomes is lost by the top to the benefit of the middle class. Again, only a tiny 3–4 per cent is lost out of the initial total of 52 per cent of income of the gross middle class. Note that the two panels may be read equally in the opposite direction, from net-­equivalized income space towards gross-­income space. Though a proportion of people remaining in the middle class of 50 percentage points is a substantial share it is telling that more than 30 percentage points of the net-­equivalized middle class cannot be attributed directly to households’ own income generation in the market alone, but depend on the redistribution and equivalization of gross incomes that are initially obtained in other classes. The most helpful message of the panels is not the (large) size of the inward flows into the middle class, but the very limited extent of the outward flows. This can be interpreted as a stylized fact: the class position of the households found in the middle class for gross incomes remains basically unchanged over the three income spaces. The implication is that on the basis of these three classes there is little re-­ranking of households over the income spaces; virtually all traffic is downwards between the top and the middle, and upwards between the bottom and the middle. The fact that this observation is found for 2010, at a time when the gaps between the classes had widened considerably compared with earlier years, adds to its significance as a stylized fact. On this assumption, the net shifts from the poor and the rich to the middle class can be equated to the gross shifts for the entire period since 1990. Note, however, that the assumption also rests on the large size of the middle class.4 On this supposition we can now in practice use the prima facie gaps of Figure 10.5 to examine the size and nature of the flows, a reading of the data which we questioned above for reasons of principle. Figure 10.11 shows the importance of the shifts and includes a breakdown by class of origin, rich or poor (panel A), and also by the effects of taxation and equivalization (panel B). The breakdowns are presented cumulatively, so

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 412

28/09/2016 08:25



Stagnating incomes and the middle class in the Netherlands ­413

a) Class of origin 35

30

30

25

25

20

20

15

15

10

10

5

0

0

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

5

Bottom HH

Top HH

Bottom INC

Top INC

b) Taxation and equivalization 35

30

30

25

25

20

20

15

15

10

10

5

0

0

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

5

Taxation HH

Equivalization HH

Taxation INC

Equivalization INC

Note:  INC = income. Source:  Statistics Netherlands/Statline, Income statistics, author’s calculations.

Figure 10.11 Gross-to-net-equivalized shifts from bottom and top to middle class, the Netherlands, 1990–2013 (percentage of the total)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 413

28/09/2016 08:25

414

Europe’s disappearing middle class?

the upper line (incomes) and the upper area boundary (numbers) indicate the entire shifts. The outcomes fully mirror the large and increasing differences found in Figure 10.5, panel A. The numbers shift grows by almost half, from 19 per cent to 27 per cent, and notably the incomes shift almost doubles over time, from 15 per cent to 28 per cent. Absolute differences between the two, numbers and incomes, seem modest, however. Nevertheless, the breakdown by class of origin: bottom or top (panel A) reveals important differences. Large numbers appear to move up from the bottom, but they bring very little income; in contrast, relatively small numbers move down from the top but take a large amount of income with them. The flows from the bottom are virtually constant over time, and therefore the growing cross-­sectional expansion of the middle class owing to taxation and equivalization can be attributed primarily to the increasing downward flows from the top to the middle. Close to 25 per cent of total income tumble down together with 11 per cent of the total number of households. Interestingly, much of the increase is concentrated in the recent period. The common approach to income distribution is based on net-­ equivalized incomes and usually shows little awareness of the significant effect on income distribution exerted by income equivalization. Panel B, however, does make this distinction. In cross-­section the picture broadly resembles that of the cross-­section breakdown of Figure 10.10: an almost 50/50 role for equivalization and taxation and very similar outcomes for numbers and incomes. Over time, the increase in the equivalization effect seems more structural and slightly more contributory to the overall growth of the shifts, while that of taxation suggests a cyclical sensitivity. Between the 1990s and the 2000s the importance of equivalization declines for the numbers moving up from the poor to the middle class but increases for numbers as well as incomes moving down from the rich (not shown). eminent Clearly, the equivalization of household incomes is of pre-­ importance for understanding the evolution of the three classes. The effect of taxation technically depends on the household’s level of income, which by definition increases with class; materially, it rests on policy-­ making. Equivalization, by contrast, depends on the size of the household, and results from an underlying, more structural process of household formation. The structural and cyclical effects alluded to with regard to Figure 10.11, panel B, seem to reflect the difference. The larger the household the greater the downward effect of equivalization on the level of income. As with income, household size increases also with a clear gradient by class in gross-­income space, as we saw in the preceding section. The cross-­sectional shifts between classes are substantial: in 2013, out of the initial 2.09 million households at the bottom, 1.14 million move up

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 414

28/09/2016 08:25



Stagnating incomes and the middle class in the Netherlands ­415

Couple, ≥1 child ≥18 5+-persons

16 12

4-persons

30 24

29

Couple, children 200%

200% ex P

2013

2012

2010

200% Pension/benefits

60–200% Enterprise

>200% Enterprise

>200% Labour

60–200% Labour

200% Inc

60–200% Inc

60–200% Nr

< 60% Inc

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

1994

1993

1992

1991

1990

0

< 60% Nr

Note:  Nr – household numbers share; Inc – income share. Source:  Statistics Netherlands/Statline, income statistics, author’s calculations.

Figure 10.15 Within-shares in gross-income classes of numbers and incomes for households with earnings, the Netherlands, 1990–2013 household earnings at the top than for enterprise incomes over the period. Starting from identical levels in 1990 (70 per cent) earnings are now considerably more important among the top (85 per cent) than among the middle (69 per cent), where the contribution of earnings first rose but has been under pressure since 2001 (when its share was still 76 per cent). The shares at the bottom are much smaller and their evolution is basically flat, as the changes are minute – in the order of magnitude of 0.5 percentage points: moving downwards from the middle to the bottom for labour and enterprise taken together, and upwards for benefits and pensions (separately). Consequently, the middle shares for earnings are hardly affected by any traffic with the bottom and can be directly compared with the top shares.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 420

28/09/2016 08:25



Stagnating incomes and the middle class in the Netherlands ­421

45

Pensions

40

Benefits

35

Earnings

30

8

6

5

3

1

–1

2010

2011

2012

2013

8

2009

8

2008

10

Total 16 15 13 11 9

2006

15

Enterprise

2001

28 27 26 24 25 27 23 24 24 24 23 22 20

2007

2005

2004

2003

2002

2000

1999

1998

1997

1996

1995

1994

1993

1992

1991

0

1990

5

Source:  Statistics Netherlands/Statline, income statistics, author’s calculations.

Figure 10.16 Gap – middle minus top gross-income shares, by source of income, the Netherlands, 1990–2013 (percentage) Figure 10.16 situates the top/middle-­earnings gap squarely in their total income gap together with the other income sources. Indeed, earnings make by far the most important contribution to the size of the gap and are also responsible for the entire decrease. Income from enterprise always has a small gap (1 per cent), notably to the advantage of the top over the middle. The importance of benefits for the gap between the two classes decreases,7 while the contribution of pensions remains unchanged.8 The gap is a phenomenon of household incomes and can come about in two different ways: (1) higher earnings for individuals already in work (intensive margin), or (2) increased earnings when the household’s number of individuals in work expands (extensive margin). The growing gap between the stagnating household employment rate and the increasing personal employment rate (Figure 10.2) gives credence to the extensive margin. Figure 10.17 spells out the class distribution of persons with an income who are members of a household receiving its main income from earnings (unfortunately only since 2001 due to lack of preceding data). In the course of a few years, the top gains significantly as its person share grows from 24 per cent to 35 per cent, while the middle share declines ­correspondingly.9 Understandably, the top share among persons is larger than among the corresponding households (18 per cent to 27 per cent) as these bring more earning persons with them to the top.10 Evidently, individuals with earnings play a large role in the shifting size of classes when transiting from gross to net-­equivalized income. They make up around 90 per cent of all persons shifting from the top to the middle

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 421

28/09/2016 08:25

422

Europe’s disappearing middle class?

100 90 24 80

25

27

29

30

29

29

30

31

32

32

34

35

69

68

65

65

65

65

64

62

62

61

59

58

70 60 50 40 70 30 20 10

5.8 6.0 7.1 7.0 6.6 6.5 6.5 6.4 6.2 5.9 5.7 5.6 5.5 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 200

Source:  Statistics Netherlands/Statline, income statistics, author’s calculations.

Figure 10.17 Persons with income in main-earnings households by grossincome class, the Netherlands, 2001–13 (percentage) (not shown). As a result, the importance of all shifts taken together among all persons with earnings has gradually increased, from 20 per cent to 30 per cent, since 2001. We can conclude that the linkage between the world of work and the income of the middle class is highly significant. Household numbers and incomes in this domain are shifting structurally and strongly towards the top and the shift seems to be reinforced by the financial crisis. The increasing concentration of employees in top-­class households almost entirely accounts for the aggregate decline of the gross-­income middle class. At the same time, these employees bring larger households with them, which may explain the large and growing role of income equivalization. These households are earning more to provide for their households’ needs. Consequently, the large and almost immutable middle class found for net-­equivalized income is to a large and increasing degree based on top incomes obtained in the labour market. 5.2  The Role of Social Dialogue and Industrial Relations Evidently, wage bargaining between unions and employers is the most important institutional determinant from the world of work to consider.11 The Netherlands has one of the most extensive systems of collective labour

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 422

28/09/2016 08:25



Stagnating incomes and the middle class in the Netherlands ­423

agreements, providing very wide coverage of more than 80 per cent of all employees, after mandatory extension by the Minister of Social Affairs and Employment. It is primarily not union density but the ‘association density’ of employers that delivers the legal requirement for the extension (SER 2013). Many agreements are characterized by detailed pay scales, which provide experience-­related gradients but commonly for a limited number of years only. The scales are an important instrument for sorting out a person’s pay when he or she joins a company or changes jobs within it. The majority of employees depend for the increase of their pay, over time, not on the pay scale but on the general wage increases agreed between unions and employers. Unfortunately, next to nothing is known about the distribution of the population of employees over these scales. The institutional machinery may look imposing but, ultimately, what counts are the effects on actual wage earnings. Salverda (2009) shows that collective bargaining is responsible for only 40 per cent of actual wage formation, as a result of incomplete coverage, individual negotiations (wage drift) and important inputs (social contributions) coming from outside wage bargaining. Effectively, collectively negotiated wages after deflation stagnated over the 1990s and subsequently increased by 5 per cent up to 2009, only to fall back to less than 2 per cent above the initial level. This contrasts with actual earnings which, due to wage drift, increased by 10 per cent over the 1990s and by a further 12 per cent until 2008, followed by a 2 per cent decline since.12 In other words, unions have been unable to secure any significant advance in purchasing power in recent decades. This seems a drastic effect of the Wassenaar Accord concluded between employers and unions in 1982, which aimed to let wages lag behind productivity growth. At the same time wage inequality has grown and only the top 10 per cent of wages have developed on a par with productivity. These outcomes of secular wage moderation put on the table the question of the extent to which this institutional set-­up may in fact reflect a collective monopsony of Dutch employers in wage formation. They also help to understand the long stagnation of the bottom 60 per cent of the household income distribution which includes a significant share of the middle class (Figure 10.3). Wage bargaining outcomes are taken into account by the government in the social dialogue with employers and unions. The statutory minimum wage in the Netherlands is the legal responsibility of the government. The statutory uprating mechanism makes it dependent on the evolution of collectively negotiated wages and thus its level equally lags actual earnings growth. In addition, the level of the minimum wage lags behind collective negotiations owing to several freezes of the uprating mechanism during the 1990s and 2000s. The full-­time annual minimum wage of

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 423

28/09/2016 08:25

424

Europe’s disappearing middle class?

2013 (19 099 euros) is at a level well below the middle class −70 per cent of the lower threshold of 60 per cent of median gross household income (29 400 euros).13 Earning a full-­time, full-­year minimum wage has become much less common owing to the rise of low-­wage part-­time employment (Salverda 2010). This goes hand in a hand with a growing presence of individuals who are low paid (by the hour) as members of higher income households (Salverda 2015). As a result, augmenting the minimum wage is still highly suitable as an instrument for putting a bottom in the earnings distribution and reducing poverty among single employees, but at the same time it carries an increasing deadweight loss from the point of view of reducing income inequality as the augmentation also benefits higher incomes. Second, social security mechanisms are part of the Dutch tradition of industrial relations. Occupational pensions – which are postponed ­earnings – are by far the most important part of this. The vast majority of workers and their employers make obligatory payments into the capital-­ funded system on the basis of obligations agreed with the unions. These pensions serve to top up the public pension (AOW) to which any resident is entitled, but that on its own is about equal to the (net) minimum wage and remains also well below the lower-­income threshold of the middle class. The top-­up helps many pensioners over the threshold into the middle class, as implied in Figures 10.4 and 10.13. The governance of the pension system – that is, the industry-­and company-­based pension funds – is the exclusive legal and material responsibility of unions and employers, aided by the tax exemption of contributions at the time they are made, in exchange for taxation of the pension income when this is received later in life.14 The tax treatment gives government a lever for change, however. Occupational pensions have developed at a stunning pace owing to the extremely well-­developed system of capita funding. The total savings of the funds grow from 72 per cent per cent of GDP in 1990 to 161 per cent in 2013, the highest level among OECD countries. This governance responsibility contrasts with all types of employee insurance, such as for disability, unemployment and sickness, where the social partners were ousted entirely from governance in the early 1990s. Even if those benefits are no longer administered by the social partners, they remain an essential part of the welfare state for workers. However, employee insurance regarding sickness leave and disability has been significantly modified by the government in several stages since the early 1990s, while unemployment insurance was changed very recently and is still being changed. Figure 10.18 shows the outcomes for the three classes, indicating the percentage of relevant households within the three classes that receive such benefits within a given year – not necessarily their main income

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 424

28/09/2016 08:25



Stagnating incomes and the middle class in the Netherlands ­425 a) Health and disability

45 40 35 30 25 20 15 10

>200%: INC

2012

60–200%: INC

2013

200%: HH

2011

60–200%: HH

2012

2010

200 1-income

>200 2-incomes

>200 3+-incomes

Note:  Areas concern the middle class, lines the top class. Numbers above the lines relate to the lines, those below to the areas. Source:  Statistics Netherlands/Statline, income statistics, author’s calculations.

Figure 10.19 Gross-income shares of households by number of incomes in top and middle, the Netherlands, 2001–13 (percentage)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 427

28/09/2016 08:25

428

Europe’s disappearing middle class?

available only for the recent period).15 Clearly, single-­earner households play a much larger role in the middle, where they make up some 30 per cent, than in the top class, where their fraction is less than half as large and also decreasing by 4 percentage points. The smaller share is largely compensated by the doubling of the share of multiple-­earners (with three or more incomes), nonetheless this category also shows a 4 percentage point decline. Both declines taken together are mirrored in the 8 percentage point increase of the two-­earner households’ share, from 52 to 60 per cent, which secures them a substantial majority at the top. The picture relates to incomes, which evidently show developments more sharply, as the single-­earner has only one income, while the other two types combine more than one. The figure covers all incomes in all households and not only earnings. The present picture can be made more precise for 2010 with the help of the same SILC data that were used above. It also helps to scrutinize the shifts between the classes with respect to dual earning. The Netherlands has 2 million two-­earner households, who make up 40 per cent of all households with earners and comprise about half of all individual workers, as against 2.5 million single-­earners (50 per cent of the total) and 0.5 million multiple-­earner households (10 per cent). In the top class, dual earners comprise 58 per cent of all earner households, while single earners have only 17 per cent and multiple earners the rest. Conversely, of all earner households 24 per cent are members of the top, while for two-­earner households that percentage, 36 per cent, is significantly higher. This top incidence is very low among single earners (8 per cent), in sharp contrast to more than half (57 per cent) for multiple-­earner households. As we have already seen, a very large portion of the top shifts down to the middle class, amounting to 68 per cent on balance. Two-­earner households are overrepresented in the shift, albeit not by very much: 72 per cent of them equivalized income. Viewed from move down between gross and net-­ another angle, 36 per cent of two-­earners are part of the gross top, but only 10 per cent are also a member of the net-­equivalized top. The shift is even stronger for multiple-­earners, among whom the share falls from 57 per cent to 5 per  cent, but among single earners the decline is much less – from 8 per cent to 6 per cent. Figure 10.20 compares the shifts for two-­earner households with single-­ and multiple-­earner households together with a breakdown by the two mechanisms of taxation and equivalization. Going beyond the net shifts, it also details the gross flows up and down. The gross flows show an interesting difference. Upward flows do remain small compared with the total, as already shown (Figure 10.10). Indeed, they are negligible for dual and multiple earners. However, they have a remarkable effect for the few single earners found at the top. Their downward flows (39 per cent and

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 428

28/09/2016 08:25



Stagnating incomes and the middle class in the Netherlands ­429 60 40 20 0

20 7

–20

–39

–40 –60

–32

–50

–17 –25

–44

–53

–30

3

Total

–80 –100

1

2 No of earners Gross > Net-Down Net > Net-equivalized-Down

Gross > Net-Up Net > Net-equivalized Up

Source:  Eurostat SILC, available at http://ec.europa.eu/eurostat/web/income-and-livingconditions/overview, author’s calculations on microdata made available by Eurostat.

Figure 10.20 Household numbers shifting between top and middle, by number of earners in the household, the Netherlands, 2010 (percentage) 17 per cent respectively), which are already smaller than for the other two categories, are to a large extent counterweighed by upward flows from the middle to the top that, taken together, amount to 27 per cent of the initial number of single-­earner households at the top, including 20 per cent because of equivalization. This suggests that their households are smaller than for dual earners. The larger effect of equivalization (25 per cent) for dual earners compared with single earners (17 per cent) supports that view directly. Finally, the average earnings of these three different types of households are roughly equal (around 125 000 euros). The important implication is that single earners attain a level of pay individually that on average is twice to three times higher than for the individuals in dual-­and multiple-­earner households.16 Consequently, single-­earners are found at the very top of the distribution of individual earnings, while the individual pay of dual and multiple earners is much further down that distribution. It highlights an important tension in defining classes solely by the level of household income, as many of the individuals in the households that attain the top class because of their combined earnings will definitely occupy what would be considered middle-­class jobs.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 429

28/09/2016 08:25

430

Europe’s disappearing middle class?

6.2  Case Study 2: Wealth and Debt of the Middle Class Thus far the focus has been on the position of the middle class in the household-­income distribution. Normally, the income distribution gets most of the attention in research and policy-­making, especially since the widespread rise in income inequality has been established as a stylized fact. One important effect of Thomas Piketty’s 2014 book, Capital in the Twenty-­first Century, is that inequalities of wealth have joined the public debate. Generally, the wealth distribution is much more unequal than the income distribution.17 However, besides these two distributions – income over income and wealth over wealth – a third relevant distribution can be distinguished that ranks household wealth by household level of income. The level of inequality of this distribution generally lies between the lower level of the income distribution and the higher one of the wealth distribution (Maestri et al. 2014: s. 4.5). Typically, in the Netherlands in 2013 the top wealth decile owns 66 per cent of total net wealth and the top income decile owns 34 per cent, while the same top income decile receives ‘only’ 25 per cent of total income. Evidently, the wealth-­over-­ wealth distribution ranks those with the largest net debts at the bottom of the distribution, and as a result the lower 60 per cent of Dutch households have no meaningful positive wealth. However, the wealth-­over-­ income distribution knows no such concentration but instead spreads those with high or low wealth according to their income. Consequently, households in the lowest income decile may (and actually do) possess wealth that belongs to the highest wealth-­over-­wealth decile and, vice versa, many in the highest income decile may have low wealth. The bottom 60 per cent of incomes own about one-­third of all net wealth. In 2013 the lowest decile of wealth-­over-­wealth owned minus 6 per cent of total net wealth while the lowest decile of wealth-­over-­income owned plus 6 per cent. As a result the position in the wealth-­over-­income distribution of ‘our’ middle class, defined on the income distribution, is not obvious a priori as the wealth they hold may be found in any part of the wealth distribution. Statistics Netherlands provides tabulated data also for the wealth distribution, once more based on tax data. The data comprise gross and net wealth and debt, and a breakdown by components of wealth, such as owner-­occupied houses, and they also contain a ranking of wealth by income. Not all years since 1990 are covered and there is a gap halfway, from 2001 to 2005, when no data are available – this is particularly deplorable because during those years important changes occurred in the tax system and the dot.com crisis hit the wealth of many. Therefore, we compare for the two periods separately. Unfortunately, the data are available on

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 430

28/09/2016 08:25



Stagnating incomes and the middle class in the Netherlands ­431

the basis of disposable incomes only – consequently, they refer to the middle class which we found before as situated between gross income and net-­equivalized income (Figure 10.5, panel A), which registers a modest decline over time. We approximate the position of the middle class in this distribution by means of linear interpolation within income deciles, using the same number fractions of classes and segments found earlier for net income. This distribution is more relevant here because of the extreme differences that occur in wealth. Large fortunes can and do appear in any of the income deciles and they make up a large chunk of the decile’s wealth. Therefore, it would make a difference, however accidental, to the result if a large fortune happened to land at one side of a class boundary or another. In practice this appears to be no problem for the broad trends that we are interested in as there is no apparent evidence of greater annual volatility for the middle class than 3 percentage points at the maximum. Figure 10.21 presents the outcome of the approximation for the three classes. The wealth share of the middle class is more or less stable at between 63 and 66 per cent for the 1990s; it seems equally stable after 2006 for most years with the clear exception of the latest year, 2013. Things do not look too different from the results found previously for incomes in

38

37

34

54

54

55

2011 8

2011 INT 9

2012 INT 11

39

37

38 55 2010 7

52

40

56

66 2000 10

53

66 1999 9

2009 8

66 1998 9

2008 7

66 1997 10

36

64 1996 10

37

65 1995 7

57

63

56

64

1994 11

0.4

1993 8

0.6

2007 7

200

2006 7

25

25

24

25

26

27

0.8

27

28

1

2013 INT 9

2005

2004

2003

2002

2001

1992

1991

0

1990

0.2

Note:  See Appendix 10A.1. Source:  Statistics Netherlands/Statline, income statistics, accessed August 2015 at http:// statline.cbs.nl/Statweb/selection/?DM=SLNL&PA=80056NED&VW=T (for 2006–2013) and http://statline.cbs.nl/Statweb/selection/?DM=SLNL&PA=03725SVH&VW=T, author’s calculations.

Figure 10.21 Shares of net-income classes and segments in total net wealth, the Netherlands, 1990–2013 (percentage)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 431

28/09/2016 08:25

432

Europe’s disappearing middle class?

Figure 10.7: stable over the 1990s and also since 2007 after a considerable decline in 2006. The most significant part of wealth for most households concerns the ownership of their own home, the frequency of which expands from 42 per cent to 57 per cent of all households, stimulated strongly by full tax deductibility of mortgage interest payments at the marginal rate of income tax.18 The expansion of ownership is skewed by class and segment, being stronger at the bottom and the lower-­most segment (Figure 10.22, panel a), likely because initial ownership rates were lower than for the rest. Expansion for the middle class was below average. Predictably, home ownership expansion went hand in hand with an expansion of house prices, fired up by mortgage debt, which increased sixfold between 1993 and 2013. Obviously, this has led to very high levels of household indebtedness. The ratio of households’ average mortgage debt to average net income more than doubled, from 2.2 to 5.2. Though there is very little difference in the speed of this increase between the classes (Figure 10.22, panel b), the ratios differ enormously between the bottom, where they rise from 7 to no less than 15, and the much lower and surprisingly comparable middle (2.2 to 5.2) and top (1.7 to 3.8). Thus the middle class is facing largely the same wealth developments and debt problems as the rest and suffers equally from the effects of declining house prices on its net wealth. The debt burden may have contributed to the muted trend of household consumption since the mid-­2000s (Figure 10.1).

7.  CONCLUSION AND POLICY ISSUES An approximation using 25 years of national tabulated income statistics, in combination with the pivotal use of SILC microdata for a cross-­sectional study of 2010, aimed at establishing the rough trend in the evolution of the size of the middle class, leads us to the following conclusions. The relationship between the world of work and the middle class – and both other classes, the poor and the rich, for that matter – is highly relevant but has also become much more complicated in recent decades. If measured for gross incomes, the size of the middle in the Netherlands strongly declines between 1990 and 2013, by between 5 and 10 percentage points in terms of the number of households and total income, respectively. Notably, the change seems stronger in recent years. This is fully matched by an expansion of the top class, while the bottom class remains unchanged. Currently, the middle class covers a bit more than half the household population and gross incomes, while the top comprises one-­fifth of households and two-­fifths of total income. This evolution is due entirely to earnings from

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 432

28/09/2016 08:25



Stagnating incomes and the middle class in the Netherlands ­433

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2011 INT 2012 INT 2013 INT

a) Frequency of home ownership (class total of households = 100) 100 90 80 70 60 50 40 30 20 10 0

200

Total

b) Housing debt-to-income ratios (average housing debt per household with such debt divided by average income)

1400 1200 1000 800 600 400 0

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2011 INT 2012 INT 2013 INT

200

200

Total

Note:  See Appendix 10A.1. Source:  Statistics Netherlands/Statline, income statistics, accessed August 2015 at http:// statline.cbs.nl/Statweb/selection/?DM=SLNL&PA=80056NED&VW=T (for 2006–2013) and http://statline.cbs.nl/Statweb/selection/?DM=SLNL&PA=03725SVH&VW=T, author’s calculations.

Figure 10.22 Home ownership and debt within classes, 1990–2013 (percentage)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 433

28/09/2016 08:25

434

Europe’s disappearing middle class?

labour, which in turn largely depend on the number of earners in the household. Individual earners – and also individuals in general – have become much more concentrated in households at the top. Today, the top’s share of total earnings (49 per cent) equals that of the middle class and only 2 per cent is left for the bottom class. Understandably, as average earnings are lower, the middle-­class share of the number of households with earnings (60 per cent) far exceeds that of the top (25 per cent), but the gap is smaller for individuals with earnings – 58 per cent for the middle as against 35 per cent for the top. Redistributive policy, through income taxation and social contributions, shifts a portion of these upward moving households ­ down to the middle class if measured for net incomes. So redistribution does enlarge the middle, but still a clear decline in the size of the middle class remains, around half the size of the gap above. Factoring in also the equivalization of incomes – which relates the income of a household to its needs and economies of scale depending on the number of persons in the household – shifts another portion down. This enhances the middle class  further  and fully neutralizes its decline over time: an invariable 80 per cent of Dutch households are members of the middle class if measured for net-­equivalized incomes. Concomitantly, the equivalization reduces the size of the top class and at the same time stabilizes its level over time. This counteracting effect of equivalization suggests that, increasingly, households need to obtain top earnings to lead a life with a middle-­class consumption style. From that perspective a rapidly growing share of wage earners are running hard in the labour market only to stand still in middle-­ class life. Indeed, the great majority of the downward-­shifting households are larger dual-­and multiple-­earner households and equivalization makes a major contribution to their shift. On average, those households with more than one earner obtain combined earnings equal to those of the (few) single earners at the top. Evidently, industrial relations and social dialogue have not prevented this outcome, although they may have limited it. Collectively bargained wages are almost unchanged in real terms over the period. Social insurance for employees has been taken out of the social partners’ hands and is structurally declining as a result of government action. This may have incited households to increase labour supply aimed at increasing household income at the extensive margin. By contrast, occupational pensions can be deemed a success over the period. The social partners have built an impressive capital funded system, with the long-­run commitment of the government that exempted the earnings spent on pension contributions from income taxation by postponing taxation until the moment pensions

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 434

28/09/2016 08:25



Stagnating incomes and the middle class in the Netherlands ­435

are paid. However, that cornerstone of the system is under increasing threat from policymakers. It should be added that these outcomes depend importantly on our broad definition of the middle class, ranging from 60 to 200 per cent of the median income (be it gross, net or net-­equivalized). This definition enables the pivotal use of SILC outcomes which finds major flows towards the middle class, but hardly any outward flows in the transitions from gross to net and to net-­equivalized incomes. This allows the conclusion that households are not re-­ranked when transiting from gross to net or net-­equivalized incomes. However, a brief look at the segments of the middle class shows that the SILC outcome would not necessarily hold if the definition of the middle class were narrower. In addition, the segments indicate the existence of clear gradients within the middle class for important variables, such as size of household, number of earners and sources of income. In particular they suggest the possibility of a similarity between the top class and the upper segment of the middle class. Finally, the extremely high (80 per cent) and unchanged size of the middle class for Dutch net-­equivalized incomes also casts doubt on the validity of the present boundary chosen at 200 per cent of the median as the middle covers almost all. Therefore, technical scrutiny of the precise boundary qualifies as an obvious issue warranting further research. The results raise three other interesting material questions for future research. The mutual interdependence of the three classes, which results from the fact that they are based on the income distribution, can put a different light on the result that seemingly one kind of household is outgrowing the middle class. Perhaps, against the background of secular stagnation of real gross incomes in more than the lower half of the Dutch income distribution (Figure 10.3), the real issue is that the incomes of other households have not grown equally but have lagged behind instead. In a situation of massive long-­run stagnation the extensive growth of household income by supplying more earners, instead of the intensive growth of augmented individual pay, may be the best option left to the household for covering enhanced household needs. It would make family formation increasingly a matter of enhanced work effort and obtaining earnings in the top class – note that children are increasingly more concentrated in the top class.19 A corollary of this is that the level of earnings of individuals in dual-­and multiple-­earner households locates these individuals lower down the individual earnings distribution, far below the top where the individuals of single-­earner top households are found. This suggests that the former individuals’ working lives may actually qualify as being middle class, creating a tension with the fact that the household has left the middle class by attaining earnings at the top. It provides good reason to consider

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 435

28/09/2016 08:25

436

Europe’s disappearing middle class?

this category of households in depth and examine whether their attitudes to work and their political and cultural views have been influenced by their elevated income position or not.

NOTES   1. I am very grateful to Monique Verhoeven for her research assistance with the SILC data and comments on an earlier draft.   2. Note that, in contrast to gross and net incomes, which are directly observed, equivalized incomes result from treatment by the data user, aimed at estimating what a given income means for households of different sizes and compositions, depending on needs and economies of scale. This lends a certain arbitrariness to net-­equivalized incomes as equivalization can be undertaken in different ways, leading to different results. Mostly equivalization (the OECD modified method and its popular, simpler brother, the square root) proceeds regardless of the level of income, and the variation of results is fairly modest – including between the method of Statistics Netherlands, which is behind the data used here, and the SILC data, which use the OECD modified scale. It can be argued, however, that needs and economics of scale may depend on income – with potentially very different outcomes as a result (see Aaberge and Melby 1998).   3. Decile averages are used, not cut-­offs, to enable including the tenth decile.   4. Similar movements between the segments of the middle class do not add up to the class totals of stayers and flows. This should be no surprise as the more the distribution is split into smaller parts the higher becomes the likelihood of movements between them and the lower the resulting level of stability. This observation does not do away with the findings for the middle class as a whole as the flows from the top mostly go to the upper segment and from the bottom to the lower-­most segment, leading to a cascading of flows between neighbouring segments.   5. Compare Salverda and Checchi (2015) and Salverda and Haas (2014) for an extensive treatment of role of individuals in the household income distribution.   6. The share is 1.5 per cent lower in 2001 compared with 2000.   7. The odd benefit reduction for 2007 is due to the temporary tax measure mentioned before, which benefited the very top (see Appendix 10A.1).   8. This seems to imply that leaving the older population out would not make much difference to the comparative evolution of top and middle.  9. Note that only the numbers distribution differs, while the income distribution over classes is unchanged as this depends on households that aggregate over persons. 10. The top-­class share grows less for households with any earnings, not always their employed and benefit main income. The difference is 50/50 divided between self-­ households. 11. For more details and other aspects see Salverda (2008). 12. Author’s calculations from the Central Planning Bureau (CPB 2015). 13. Including employer contributions (23 410 euros) it amounts to 85 per cent. 14. Employees may profit from tax postponement when their pension income is lower than their earnings resulting in a lower marginal tax rate. 15. Two-­earner households are virtually non-­existent at the bottom, which is therefore left out of the picture. 16. The presence of part-­time jobs among earners in dual-­and multiple-­earner households may increase the spread over the wage distribution. 17. Salverda (2015) finds a record-­high Gini coefficient of 0.886 for net wealth inequality, including negative values, in the Netherlands in 2013. This matches US wealth inequality measured on the same basis (Wolff 2014). 18. In addition, mortgages in 2013 could still exceed the value of the house, and tax

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 436

28/09/2016 08:25



Stagnating incomes and the middle class in the Netherlands ­437

deductibility can be optimized over partners’ incomes in the household since 2001. Only recently have policies been implemented to reduce deductibility. 19. A rough and ready estimation suggests that the top’s share of children has increased from 21 per cent to 32 per cent in recent years, while the middle-­class share declined correspondingly.

REFERENCES Aaberge, R., and I. Melby (1998), ‘The sensitivity of income inequality to choice of equivalence scales’, Review of Income and Wealth, 44 (4), 565–9. Central Planning Bureau (CPB) (2015), Centraal Economisch Plan 2015 (Central Economc Plan 2015), The Hague: CPB, annex table 6. Maestri, V., F. Bogliacino and W. Salverda (2014), ‘Wealth inequality and the accumulation of debt’, in W. Salverda, B. Nolan, D. Checchi, I. Marx, A. McKnight, I.G. Tóth and H. van de Werfhorst (eds), Changing Inequalities in Rich Countries: Analytical and Comparative Perspectives, Oxford; Oxford University Press, pp. 82–120. Organisation for Economic Co-­ operation and Development (OECD) (2015), Economic Outlook, No. 97, Paris: OECD. Piketty, T. (2014), Capital in the Twenty-­first Century, Cambridge, MA: Harvard University Press. Salverda, W. (2008), ‘Labor market institutions, low-­wage work, and job quality’, in W. Salverda, M. van Klaveren and M. van der Meer (eds), Low-­Wage Work in the Netherlands, New York: Russell Sage, pp. 63–131. Salverda, W. (2009), ‘The bite and effects of wage bargaining in the Netherlands 1995–2005’, in M. Keune and B. Galgóczi (eds), Wages and Wage Bargaining in Europe: Developments since the Mid-­1990s, Brussels: ETUI, pp. 225–54. Salverda, W. (2010), ‘The Dutch minimum wage: minimum wage fall shifts focus to part-­time jobs’, in D. Vaughan-­Whitehead (ed.), The Minimum Wage Revisited in the Enlarged EU: Issues and Challenges, Cheltenham, UK and Northampton, MA, USA and Geneva: Edward Elgar and International Labour Organization, pp. 299–339. Salverda, W. (2013), ‘Inkomen, herverdeling en huishoudvorming 1977–2011: 35 jaar ongelijkheidsgroei in Nederland’ (‘Income redistribution and household lining 1977–2011: 35 years of inequality growth in the Netherlands’), TPEdigitaal, 7 (1), 66–94. Salverda, W. (2015), ‘Individual earnings and household incomes: mutually reinforcing inequalities?’, European Journal of Economic and Economic Policies, 12 (2), 190–203. Salverda, W. and D. Checchi (2015), ‘Labour-­market institutions and the dispersion of wage earnings’, in A. Atkinson and F. Bourguignon (eds), Handbook of Income Distribution, vol. 2B, Amsterdam: Elsevier/North Holland, pp. 1535–727. Salverda, W. and C. Haas (2014), ‘Earnings, employment and income inequality’, in W. Salverda, B. Nolan, D. Checchi, I. Marx, A. McKnight, I.G. Tóth and H. van de Werfhorst (eds), Changing Inequalities in Rich Countries: Analytical and Comparative Perspectives, Oxford; Oxford University Press, pp. 49–81. Social and Economic Council (SER) (2013), Verbreding draagvlak cao-­afspraken (Broadening Support for Collective Agreements), The Hague: SER.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 437

28/09/2016 08:25

438

Europe’s disappearing middle class?

Statistics Netherlands/Statline, income statistics database, accessed August 2015 at http://statline.cbs.nl/Statweb/selection/?DM=SLNL&PA=71013NED&VW=T (for 2001–2013) and http://statline.cbs.nl/Statweb/selection/?DM=SLNL&PA= 70191NED&VW=T (for 1990–2000). Wolff, E. (2014), Household Wealth Trends in the United States, 1962–2013: What Happened over the Great Recession? National Bureau of Economic Research (NBER) Working Paper 20733, accessed 6 May 2016 at http://www.nber.org/ papers/w20733.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 438

28/09/2016 08:25



Stagnating incomes and the middle class in the Netherlands ­439

APPENDIX 10A.1 INCOME AND WEALTH DATA: SOME CAVEATS The Dutch Labour Force Survey contains no income data and I have had to make do with the tabulated data of the income and wealth statistics of Statistics Netherlands as the main data source, although they neglect important labour-­market characteristics, such as educational attainment, hours of work and labour contracts. The data are sampled from the tax administration. They cover the long run since 1990 for numbers, incomes and various demographic and income characteristics of households, and they do this for gross, net and net-­ equivalized incomes. Households are ranked by the three different types of income, but also by gross income for the other two types. Results presented here are found through linear interpolation of the relevant fractions of the median within either ten deciles or 23 classes of the income distribution, depending on what works most precisely for the interpolations or is best for connecting to other variables. Unfortunately, the data are plagued by a serious series break halfway through the period, between 2000 and 2001; both the coverage of incomes and the contents of variables change. The break extends the coverage of gross incomes by incorporating, among other things, employer contributions to the occupational pension system (see Salverda 2013). It implies that figures since 2001 provide a more adequate representation of the income distribution than before. These pension contributions are significantly concentrated in the higher deciles and thus help to explain the much larger break for average income (+11 percentage points) than for median income (+6 p ­ ercentage points). In addition, imputed rent is lowered, household statistics are revised and some definitions change – in particular, there is an important shift of earnings towards self-­employed households, particularly at the top, as households are now characterized as self-­employed when they receive any income from that source, irrespective of the level, even if negative and which may be well below the earnings received. Unfortunately, it is impossible to quantify the changes between 2000 and 2001, which is regrettable as in 2001 fundamental changes were made to the income tax system (for example, replacing tax exemptions at the marginal rate with tax credits of equal value for all households, and abolishing the separate wealth tax). This serves as a caveat for comparing the periods before and after 2001. The break is indicated in the graphs, but not detailed in the text. I can, however, live with the approximate interpolation as well as the series break because the primary aim here is to uncover the main trends and not to make the most exacting comparison in cross-­section or over time. For several variables data are available for the period since 2001 only.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 439

28/09/2016 08:25

440

Europe’s disappearing middle class?

A specific ‘health warning’ about the income data concerns the peak reached by the highest decile in 2007. It gives a distorted impression of the evolution of incomes and income inequality just before and at the start of the financial crisis. The peak results from a temporary tax measure which involved only the very top (1 per cent), but is large enough to affect the average income and broad aggregate measures of income inequality, such as the Gini coefficient. It also affects the income shares of the middle and the top (Figure 10.4). In addition, it temporarily reduces the share of labour earnings (for example, Figures 10.14 and 10.15) as the advantage relates entirely to income from enterprise. The one-­time effect will be ignored in the text, however telling it may be about tax optimization by high-­income households and the underestimation of enterprise income in other years than 2007 The wealth data show an important gap between 2000 and 2006. This reflects another series break, but potentially the quantitative effect is less (see the text on the home ownership of classes). For example, the mode of valuation of owner-­occupied houses has changed, while house prices also rose drastically. In addition, the statistic shifts in 2011 from a sample of taxpayers to full coverage of the population; these results are indicated with INT for the years 2011 to 2013 in Figures 10.21 and 10.22. The second source of (micro)data on income is European Union (EU)-­ SILC, used here only for the purpose of a crucial cross-­sectional analysis for a recent year (2010). A fuller use of these data is outside the scope of this contribution. The SILC data are explained in the introduction to this volume.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 440

28/09/2016 08:25

11. Still holding on? Inequality, labour market and middle-­income groups in Portugal Pilar González, António Figueiredo, Hugo Figueiredo and Luis Delfim Santos 1. INTRODUCTION The middle classes play a crucial role in instigating and promoting processes of development and social transformation in market-­driven economies. The relative backwardness of the Portuguese economy, as well as the increasing signs that the country’s existing growth model is exhausted (González and Figueiredo 2015) – already evident since the beginning of the new millennium – make it a useful basis on which to revisit the role of the middle classes in such processes of change. Our approach is predominantly economic. We use middle-­ income groups as a proxy for the middle classes in general. As in other chapters of this volume, middle-­income groups include households whose income  is between 60 and 200 per cent of the overall median equivalized income, which allows us to compare their relative position vis-­à-­vis low-­income groups (defined as those with 60 per cent or less of the median income) and top-­income groups (with more than 200 per cent of the median income). We are fully aware of the limitations and constraints of such a quantitative identification of the middle classes, as well as of the fact that the notion ‘middle class’ remains controversial (for example, Milanovic and Yitzhaki 2002; Ravallion 2009; US Department of Commerce 2010; Atkinson and Brandolini 2011). Ideally, the analysis of the relative position of middle-­ income groups in Portugal should be combined with other important aspects, such as the material aspirations of such income groups. Ours is in no way intended to be a global view of the evolution of the middle classes in Portugal. We merely concentrate on the economic determinants of income inequality and, in particular, on the way changes in labour markets or broadly associated with the world of work have affected the relative position of Portugal’s middle-­income groups. 441

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 441

28/09/2016 08:25

442

Europe’s disappearing middle class?

In addition to this introduction the chapter has four sections. Section 2 presents an overview of overall inequality levels, as well as of the relative and absolute evolution of middle-­income groups in Portugal compared with other European Union (EU) economies. Section 3 contextualizes such data by discussing the effects experienced by the Portuguese middle classes, which will help us to understand some of the dissonance between the narrative of the relative resilience of middle classes in Portugal presented in section 2 and the widely disseminated conviction in the political debate that the middle classes have been heavily penalized by the bailout of the Portuguese economy. Section 4 focuses on changes in the labour market in both the private and public sectors, and their effects on the relative position of middle-­income groups. Taking into account the relevance of the issue of polarization in the debate on the middle classes in advanced countries, in section 5 we present a case study on the issue, together with another case study on some issues concerning public employment. Section 6 summarizes the main conclusions.

2. A BIRD’S EYE VIEW OF INEQUALITY TRENDS AND THE RELATIVE POSITION OF MIDDLE-­ INCOME GROUPS IN PORTUGAL: A SUCCESS STORY? In light of the predominantly economic notion of the middle classes assumed in this volume and before investigating their possible shrinkage, we need to look at the recent evolution of middle-­income groups in tandem with the overall trend in income inequality in Portugal. Portugal still is one of the most unequal countries in the EU with a Gini coefficient of equivalized disposable income of 0.342, which in 20121 was only surpassed by Bulgaria, Latvia, Lithuania and Greece, and well above the EU15 average (0.304) (Figure 11.1). The comparative evolution of this indicator from 1995 to 2012 for  Portugal and the EU15 is presented in Figure 11.2. In Portugal a clearly descending trajectory is observed after 2003, while a visible oscillation prevails in the period 1995–2001. In the EU15, it appears that the EU is currently re-­approaching the Gini coefficient values observed in 1995. Portugal has thus consistently been a very unequal country according to EU standards, although it has recently displayed a clear downward trend that was interrupted by the adjustment process initiated due to the country’s bailout. In any case, most of the 2003–12 trajectory has led to greater overall equality. That means that only very recently, and strongly

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 442

28/09/2016 08:25



Inequality, labour market and middle-­income groups in Portugal ­443

40 35 30 25 20 15 10 0

EU15 Belgium Bulgaria Czech Republic Denmark Germany Estonia Ireland Greece Spain France Croatia Italy Cyprus Latvia Lithuania Luxembourg Hungary Malta Netherlands Austria Poland Portugal Romania Slovenia Slovakia Finland Sweden United Kingdom Iceland Norway Switzerland

5

Source:  Eurostat, Statistics on Income and Living Conditions (SILC) data.

Figure 11.1 Gini coefficient of equivalized disposable income, Portugal and EU, 2012

40

35

European Union (15 countries)

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

25

1995

30

Portugal

Source:  Eurostat, SILC data.

Figure 11.2 Gini coefficient of equivalized disposable income, Portugal and EU15, 1995–2012

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 443

28/09/2016 08:25

444

Europe’s disappearing middle class?

determined by the bailout process, has the evolution of inequality in Portugal tended to follow the rise of inequality in the great majority of EU countries. These inequality figures (its relative incidence and recent trends) cannot be fully understood without disaggregating households by income groups. In recent years, high inequality in Portugal has been strongly determined by the comparative high share of total income captured by the top-­income groups. We analysed the dynamic behaviour of the middle-­income groups within an overall distribution comprising 14 income groups.2 We looked at both the share of households in each income group and the share of income they absorbed. The resulting Portuguese distribution is very distinct, principally when compared with a relatively egalitarian country such as Sweden,3 or even with Germany (Figure 11.3). While the Swedish distribution is bell shaped, in Portugal we observe a much higher share of total income for top-­income groups. The first part of the distribution is centred on the fifth and sixth groups (80–100 per cent and 100–120 per  cent of the median income), but the second part has lower values in groups 10 and 12 (180–200 per cent and 240–400 per cent of the median income), but very high shares in the thirteenth and fourteenth groups (280 per cent, maximum percentage of the median income). This suggests that the pattern of income concentration at the top is far more pronounced than what might be suggested by the analysis of the five income groups used throughout this book. In both 2004 and 2010 the influence of progressive taxation is also clear (Figures 11.4a and 11.4b). In gross terms, group 14 absorbs a significantly higher percentage of income than any of the other groups. However, when we look at net income, group 14 absorbs a lower percentage than that absorbed by groups 5 and 6. The effect is particularly visible in 2010. Taxation thus plays a very important role in decreasing income concentration from very high levels. Figure 11.5 then compares the percentages of households and income absorbed by the 14 income groups under consideration. It is evident that the former distribution is clearly more symmetrical than the latter and that a very low percentage of top-­income households appropriate a large fraction of total net income. In any case, ‘being in the middle’ can represent very diverse situations in different European countries. Furthermore, the sense of belonging to the middle class is expected to be stronger in countries such as Sweden, where there are much smaller groups at the extremes. The situation in Portugal is much more fluid and less contrasted. Looking in greater detail at the diversity of the relative position of middle-­income groups in EU income distributions and at the singularity of Portugal in this context, Table 11.1 uses EU-­SILC data on the share of

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 444

28/09/2016 08:25

445

VAUGHAN 9781786430595 PRINT (M4051).indd 445

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:25

0

0

Portugal 2004 2010

1 2 3 4 5 6 7 8 9 10 11 12 13 14

0

5

10

15

20

25

Sweden 2004 2010

1 2 3 4 5 6 7 8 9 10 11 12 13 14

Figure 11.3 Share of net income by income group, Portugal, Germany and Sweden, 2004 and 2010

Source:  EU-SILC (project calculations, see end note 2 for details).

5

5

1 2 3 4 5 6 7 8 9 10 11 12 13 14

10

20

10

2004 2010

25

15

Germany

15

20

25

446

Europe’s disappearing middle class?

15

2004 2010

15 10

2004 2010

10 5 5 0

1

2

3

4

5

6

7

8

9

10

11

12

13

14

0

Figure1 11.4a  Share 2 3 4of total 5 income 6 7by income 8 9 group 10 (gross 11 income), 12 13 14 Portugal, 2004 and 2010 15

2004 2010

15 10

2004 2010

10 5 5 0 0

1

2

3

4

5

6

7

8

9

10

11

12

13

14

1

2

3

4

5

6

7

8

9

10

11

12

13

14

Source:  EU-SILC data (project calculations).

Figure 11.4b Share of total income by income group (net income), Portugal, 2004 and 2010

income and of households by income groups to provide evidence of the diversity of the position of middle-­income groups across Europe. The Portuguese situation is determined not by the share of middle-­ income households in the total, but on the contrary by the fact that Portugal’s middle-­income groups appropriate a lower share of income than the majority of middle-­income groups in Europe (see also Figure 11.6). On the one hand, this points to the country’s backwardness, but, on the other hand, it reminds us that this group may have strong upward economic aspirations.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 446

28/09/2016 08:25



Inequality, labour market and middle-­income groups in Portugal ­447

20 15 10 5 0

1

2

3

4

5

6

7

Share of households

8

9

10

11

12

13

14

Share of income

Source:  EU-SILC data (project calculations).

Figure 11.5 Share of households and share of income by income groups (net income), Portugal, 2010

Table 11.1 Middle-­income groups’ share of income and households, Europe, 2010 Share of households in the middle-­income groups Share of income of middle­income groups

< 65%

< 65%

65 to 75%

Latvia Lithuania

Portugal UK

65 to 75%

> 75%

> 75%

Estonia Italy Spain Germany Greece Ireland France Netherlands Hungary Belgium Sweden

Source:  Eurostat, SILC (project calculations).

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 447

28/09/2016 08:25

448

Sweden

Belgium

Spain

Germany

Estonia

Lithuania

Italy

Holland

Greece

Hungary

Ireland

Latvia

Portugal

France

UK

8 6 4 2 0 –2 –4 –6 –8

Europe’s disappearing middle class?

Source:  EU-SILC (project calculations).

Figure 11.6 Middle-income groups: difference in shares of income and households (percentage points), Portugal, and selected EU, 2010 2.1  Trends in the Representativeness of Middle-­income Groups In order to look at the relative position of middle-­income groups over time we also used widely available data on the distribution of income by deciles.4 We started by calculating the average annual growth rate of the top cut-­off points (in real terms) of each decile of the distribution of the median equivalized disposable income for three different sub-­periods (Figure 11.7): ●●

●●

●●

1995–2001, a period of economic growth (3.16 per cent average annual growth rate of real gross domestic product (GDP) per capita in Portugal compared with 2.36 per cent in the Euro19); 2004–09, a period of slow economic growth or even stagnation (0.48 per cent average annual growth rate of real GDP per capita in Portugal compared with 0.55 per cent in the Euro19); 2009–12, a period that combines the impact of the international crisis in 2007–2008 and the sovereign debt crisis and bailout adjustment (–1.13 per cent average annual growth rate of real GDP per capita in the period 2010–12, compared with 0.7 per cent in the Euro19).

The first period shows a relatively similar pattern in Portugal and the EU despite the noticeable differences in the intensity of growth of such cut-­off

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 448

28/09/2016 08:25

449

2nd decile

1st decile

1995–2001

3rd decile

2004–2009

5th decile

Portugal

95th percentile

9th decile 2009–2012

0.0

2.0

4.0

6.0

8.0

2nd decile

1st decile 1995–2001

EU15

5th decile

2004–2009

6th decile

10.0

9th decile

2009–2012

95th percentile

8th decile

7th decile 4th decile

3rd decile

8th decile

7th decile

4th decile

Figure 11.7 Median equivalized net income, Portugal and EU15, 1995–2012 (average annual growth rate of top cut-off points) (percentage)

Source:  EU-SILC data.

10.0 8.0 6.0 4.0 2.0 0.0 –2.0 –4.0 –6.0 6th decile

VAUGHAN 9781786430595 PRINT (M4051).indd 449

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:25

450

Europe’s disappearing middle class?

values and despite being less redistributive than the EU15. This suggests that growth in Portugal had effective even if limited redistribution effects. In the last two periods, however, a very different pattern of evolution appears. In 2004–09, the median equivalized net income of the poorest 20 per cent had positive growth rates in Portugal and the top cut-­off points of the upper groups had very low income growth rates. Intermediate deciles succeeded in maintaining rather constant positive income growth rates. This means that, despite the fact that 2004–09 is seen as a period of slow growth – often designated a lost decade in terms of growth – it continued to be a pro-­equality period. As demonstrated in González and Figueiredo (2015), this was mainly the outcome of redistributive and social protection public policies and a more redistributive pattern than in the EU (Alves 2009, 2012; Rodrigues and Andrade 2013). This redistributive trend was reversed in 2009–12, however. All deciles experienced negative income growth rates but the upper and especially the bottom groups had the highest decreases. Intermediate deciles also lost income in real terms but in relative terms they lost less than both the poorer and the top income groups. The crisis and the austerity-­driven adjustments appear, therefore, to have inverted the previous pro-­equality trend in the Portuguese economy, as well as the convergence of the income levels of the poorest deciles, despite Portugal’s still wide development gap relative to other EU economies. The evidence presented thus far seems to suggest a globally positive income groups. Both in good and bad picture for Portuguese middle-­ growth performance periods the intermediate deciles appear to have had high growth rates in terms of income or a greater degree of resistance to income losses in the bailout period. This evidence should also be seen in tandem with the Portuguese economy’s capacity to reduce the proportion of the poor since the mid-­1990s, which makes the recent reversal of the trend for the very lowest deciles even more significant. In any case, it appears that such redistribution has continued for the middle classes. A number of contradictory facts have emerged in recent years in Portugal. Indicators of overall inequality deteriorated very recently but the top income groups have not been exempted from the effects of the crisis and ‘austerity’ policies. The very progressive fiscal system (including the progressive surtax introduced) and the significant cuts in top public wages and pensions that characterize this period seem to have been effective in reducing the gap between the income shares of the high-­and middle-­ income groups, even if the disposable income of all population groups appears to have decreased. In any case, increasing inequality cannot then be explained solely by referring to the top percentiles (Figure 11.8).

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 450

28/09/2016 08:25



Inequality, labour market and middle-­income groups in Portugal ­451

7

2003 2007 2010 2012

6 5 4 3 2 1

9 en 5th til e pe rc 96 en th til e pe rc 97 en th til e pe rc 98 en th til e pe rc 99 en th til e pe 1 rc 00 en th til e

rc pe

pe

rc

en 1s til t e pe rc 2 en nd til e pe rc 3 en rd til e pe rc 4r en th til e pe rc 5 en th til e

0

Source:  Eurostat, SILC data.

Figure 11.8 Share of national equivalized income for the bottom and top percentiles, Portugal, 2003–12 2.2  Income Mobility The sense of belonging to the middle classes entails not only a comparison (in each period) with those at the extremes, but also a view of the chances that each individual (household) has of moving up in the income (or, more generally, in the social) hierarchy. The analysis of inter-­decile mobility patterns adds, therefore, relevant information to the debate. With that purpose, we accessed data on the percentage of individuals remaining in the same decile for three consecutive years and the percentage of those moving to neighbouring deciles (either up or down). We looked at two different periods: 2006–09, a period still largely unaffected by the effects of the crisis and the ensuing austerity; and 2009–12, the period in which these effects were largely felt. The resulting information on upward and downward mobility in these two periods is synthesized in Figures 11.9 and 11.10.5 Figure 11.9 allows us to conclude, as expected, that upward mobility decreases as we move up the income distribution. There is, however, a positive but less linear relationship between income and downward mobility (Figure 11.10). There are, in particular, quite high rates of downward mobility in the upper mid-­deciles (sixth to eighth). Also, we do not see massively different patterns before and after the crisis and the implementation of ‘austerity’ policies. Looking at both figures, one could argue that most of the impact of the crisis in terms of income mobility was felt at this upper mid-­segment of the distribution, namely the sixth and seventh

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 451

28/09/2016 08:25

452

Europe’s disappearing middle class?

60

Upward mobility 2006–09 Upward mobility 2009–12

50 40 30 20 10 0

D1

D2

D3

D4

D5

D6

D7

D8

D9

D10

Source:  EU-SILC data.

Figure 11.9 Share of households who experienced upward mobility, Portugal, 2006–12 60 50 40 30 20 Downward mobility 2006–09

10 0

Downward mobility 2009–12 D1

D2

D3

D4

D5

D6

D7

D8

D9

D10

Source:  EU-SILC data.

Figure 11.10 Share of households who experienced downward mobility, Portugal, 2006–12 deciles, for which downward mobility increased and upward mobility decreased, and perhaps at the very bottom of the distribution (deciles 1 and 2). In the next section, we suggest that part of the explanation for this is the fall in real wages and rising unemployment.6 By contrast, upward

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 452

28/09/2016 08:25



Inequality, labour market and middle-­income groups in Portugal ­453

mobility increased and downward mobility decreased mainly for the fourth and fifth deciles. Comparing these results with those obtained for the EU15 (see Appendix 11A.3) we notice that, in average terms, the crisis brought less mobility in other European economies, particularly downward.

3. CONTEXTUALIZING THE EVOLUTION OF PORTUGAL’S MIDDLE CLASSES The narrative on the evolution of Portugal’s middle classes that we have presented thus far does not seem perfectly aligned with that resulting from the debate on the shrinking middle classes in more advanced economies (the United States in particular). Some European countries have also witnessed the hollowing-­out of their income structures but the pattern is generally less marked and there is significant variety. Our own analysis of the Portuguese middle-­income groups also alerts us to a more fuzzy pattern. While these groups have not been immune to the economic and social consequences of the crisis and of the austerity-­led fiscal consolidation, the income losses of the middle-­income groups are less striking than those of the bottom-­and top-­income groups. In the context of the low level of development of the Portuguese economy, this may stem, in part, from the limitations of relying on an exclusively economic and mainly quantitative approach to the identification of middle-­income groups. A wider sociological approach, for example, could lead us to focus on variables such as professional status, educational qualifications, position in professional hierarchies, social mobility or household property or other forms of wealth. Estanque (2012), for example, stresses two key factors that explain the comparative difficulties of the Portuguese middle classes in increasing their overall size and social recognition: the late industrialization of the country and the recent emergence of Portuguese democracy (which only dates back to the mid-­1970s). The author also argues that the emergence of the middle classes in Portugal has happened relatively fast and that this has contributed to their fragility. 3.1 Backwardness We start this section by arguing that Portugal’s relative backwardness still has a strong influence on the social configuration of its middle-­income groups. Furthermore, this is not only a historical matter. It is still reflected in the country’s relatively low level of development vis-­à-­vis other EU member states.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 453

28/09/2016 08:25

454

Europe’s disappearing middle class?

20 000

European Union (15 countries) Portugal

18 000 16 000 14 000 12 000 10 000 8000

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2001

2000

1999

1998

1997

1996

4000

1995

6000

Source:  Eurostat, SILC data.

Figure 11.11 Median equivalized disposable income (euros), Portugal and EU, 1995–2012 The process of European integration played a key role in the formation of expectations of upward social mobility, namely, through awareness of European consumption patterns. Portugal’s accession to the EU in the mid-­1980s has been presented mainly as a modernization and equalizing process potentially able to eliminate the structural constraints associated with low development levels, as well as to increase household income. Little progress has been achieved in this, however. Figures 11.11 and 11.12 depict the evolution of the median equivalized disposable income in Portugal, the EU15 and some other European economies in recent years. Both figures illustrate the prevailing development gap between Portugal and other relevant European economies. After almost 30 years of EU membership, Portugal still belongs among the poorest countries (Figure 11.13). Furthermore, the crisis, the bailout and associated austerity trajectory have increased the divergence between the EU and Portugal, aggravating what was already a significant development gap. This gap has an important effect on the aspirations of the middle classes. Just as the Portuguese middle classes changed their expectations in line with the political emphasis given to the positive effects of joining the EU, the recent evidence of effective divergence and of its dramatic effects is increasingly calling into question EU integration itself and the chance of improvement within the present context of EU policies. This is relevant to contextualize the relative resilience of middle-­income groups in Portugal as their income gap in relation to similar groups in other European economies has not been consistently reduced. Thus we

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 454

28/09/2016 08:25



Inequality, labour market and middle-­income groups in Portugal ­455

21 000 19 000 17 000 15 000 13 000 11 000 9000

Germany

Greece

Spain

France

Italy

Portugal

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2001

2000

1999

1998

1997

1996

5000

1995

7000

Source:  Eurostat, SILC data.

Figure 11.12 Median equivalized disposable income (PPS), Portugal and selected EU countries, 1995–2012 have to keep in mind that even if middle-­income groups have lost less than bottom and top income groups in very recent years (see section 2), their net income has been reduced, worsening an income gap in relation to other EU economies that was already high. 3.2  Structural Change, Modernization and the Labour Market As already mentioned, EU membership in the late 1980s and 1990s was represented as a pathway to greater modernization opportunities, based on two axes: (1) building the foundations of the European Social Model in Portugal, allowing generalized access to social protection at several levels, thereby creating a new set of aspirations for individuals and families; and (2) promoting modernization by stimulating structural change in the economy and the labour market, resulting in a more intensive use of knowledge, skilled labour and educational attainment. Both represent critical factors in developing and empowering the middle classes as they are supposed to increase the share of households in middle-­income groups. Those two processes were also expected to reverse the strong dependence on public employment that the Portuguese middle classes historically enjoyed from the early years of democracy. The concrete results of these two processes have been mixed. Social protection has progressed, but

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 455

28/09/2016 08:25

456

Europe’s disappearing middle class?

Luxembourg Austria Sweden Belgium Germany Denmark Netherlands France Finland Cyprus United Kingdom Ireland Italy Malta Spain Slovenia Czech Republic Portugal Slovakia Poland Greece Estonia Croatia Hungary Lithuania Latvia Bulgaria Romania 0

5000

10 000

15 000

20 000

25 000

30 000

Source:  Eurostat, SILC data.

Figure 11.13 Median equivalized disposable income (PPS), Portugal and EU, 2012 sustainability has not been fully achieved, while the modernization of the economy and the labour market have taken place mainly within the existing pattern of specialization. A slow process of structural change is under way, as demonstrated, for example, by the growth of high-­technology exports. Nevertheless, the massive increase in unemployment associated with the exhaustion of the low-­skilled growth model and the increase in the unemployment rates of young educated individuals appear as the other face of an incipient and vulnerable modernization process.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 456

28/09/2016 08:25



Inequality, labour market and middle-­income groups in Portugal ­457

100 90

3.10 9.12

3.42 12.84

2.50 9.34

4.16 14.69

4.13 11.46

7.31 20.77

80

10.52

19.24

19.58

70

45.18 24.03

49.57

60 50 40

87.8

83.7

88.2

81.2

84.4

30

23..48 71.9

69.9

56.7

20 10 0

2004

2013

< 60% median

2004

2013

2004

60%–80%

Tertiary education

2013

80%–120% Upper secondary

21.99

2004

2013

31.3

28.4

2004

2013

120%–200%

>200%

Lower secondary

Source:  EU-SILC data (project calculations).

Figure 11.14 Educational attainment levels by income group, Portugal, 2004–13 (percentage) 3.3  Educational Attainment Gaps Despite the very rapid increase in the educational attainment of young people in recent years, Portugal’s workforce – and population in general – still have very low levels of educational attainment in comparison with other European economies. Figure 11.14 depicts the composition of the various income groups according to their education level. It is clear that high income groups are the only ones in which high levels of educational attainment are the norm. About half of high income individuals have a tertiary education and this share increased in the past decade. Among middle-­ income groups, however, the highly educated remain a marginal group. Only the high-­middle-­income groups (120–200 per cent of the median equivalized disposable income) had 19.2 per cent of tertiary educated individuals in 2010. To put it in perspective, however, close to 57 per cent of all individuals in that group had a lower secondary education or less. The process of educational expansion should, in any case, slowly transform the educational structure of the middle-­income groups. Figure 11.15 makes this clear by depicting the distribution of all university graduates in 2003 and 2012 by equivalized disposable income decile. The figure suggests that, over time, a higher share of university educated individuals will

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 457

28/09/2016 08:25

458

Europe’s disappearing middle class?

60

2003 tertiary

50

2012 tertiary

40 30 20 10 0

D1

D2

D3

D4

D5

D6

D7

D8

D9

D10

Source:  EU-SILC data (project calculations).

Figure 11.15 Distribution of university graduates by equivalized disposable income decile, Portugal, 2003–12 (percentage) end up in progressively lower deciles and increasingly in middle-­income groups. 3.4  Public versus Private Employment Portugal’s backwardness is also the main factor explaining the middle classes’ historical strong dependence on public (central and local) administration employment. Figure 11.16 shows the distribution of public sector jobs and compares this with the distribution of total employees, again using detailed income groups.7 It is evident that the distribution of public sector jobs is biased towards the top-­income groups, affecting both the middle-­income groups but also the very top groups (for example, the ­eleventh and thirteenth income groups). This implicitly reveals what we call the inertia effect of the low-­wage based development model that still prevails in the private sector (which, to an extent, has been reinforced by the bailout’s focus on wage cuts). Under the recent adjustment and bailout process, however, the public sector has been at the epicentre of the ‘austerity’ and budget consolidation process. A mix of voluntary early retirements (anticipating wage cuts), old-­ age retirements, reorganization of services normally leading to early retirement agreements, wage cuts, suspension of recruitment rounds, freezes in career progression, a new surtax on public wages and above all an undermining (often politically driven) of trust in public services have created a very unfavourable work context for public servants with a major impact on

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 458

28/09/2016 08:25



Inequality, labour market and middle-­income groups in Portugal ­459

0.18 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0

1

2

3

4

5

6

7

Share of employees

8

9

10

11

12

13

14

Share of public servants

Source:  EU-SILC data (project calculations).

Figure 11.16 Distribution of employees and public servants by income group, Portugal, 2010 middle class aspirations. Not only has the representation of public employment as ‘safe’ been broken (Table 11.2), but also gross and net wages have substantially changed, affecting expectations of upward social mobility. It is clear now that, without the full modernization of private labour markets that was expected after adhesion to the EU and the Eurozone, employment in the private sector did not generate new and more qualified jobs in sufficient magnitude to maintain the labour market status of more educated Portuguese youngsters at levels similar to those of the previous generation. In sum, therefore, the destruction of public sector jobs and the creation of private sector jobs, while clearly relevant for the analysis of middle-­income groups, is likely to have differentiated impacts within the middle classes, by destroying, on average, relatively better jobs and substituting these by lower quality (lower paid) ones. Such changes are, therefore, very likely to adversely affect middle class aspirations in Portugal. 3.5  Progressive Taxation and Social Policy Social transfers and taxes play an important role in determining disposable income. Figures 11.17 and 11.18 depict the relevant characteristics of the Portuguese redistribution system. The contrast between the EU15 and Portugal is marked. In the EU15, the correction of market income by social transfers is impressive throughout the period. In Portugal, the influence of social transfers is also very

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 459

28/09/2016 08:25

460

Europe’s disappearing middle class?

Table 11.2 Average annual growth rate of employment in the public sector, Portugal, 1996–2014 Period

Average annual growth rates (%)

1996–2005 2005–09 2009–11 2011–14

1.76 −2.53 −4.67 −3.39

Source:  Direção-­Geral da Administração e do Emprego Público (DGAEP), public employment dataset, several years, available at http://www.dgaep.gov.pt/index. cfm?OBJID=3b2e32ef-­9ea6-­4e93-­9417-­73f3a8b74500.

important to correct market income levels but its scale has not been sufficient to prevent impoverishment since 2009. In any case, the Organisation for Economic Co-­operation and Development (OECD) refers to Portugal as an example of a country in which – in comparative terms – taxes and transfers had a relevant equalizing effect in the period 2007–11 (Figure 11.18). This recognition is particularly significant considering that before the 2000s, social policy in Portugal was often characterized by low redistributive effects. Alves (2012) used the EU-­SILC microdata to analyse the impact of taxes and monetary social transfers (pensions excluded) in Portugal and the EU. He stresses that the high income inequality in Portugal necessarily introduces substantial diversity between those who bear the highest share of income taxes and those who benefit from social transfers. The author concludes that social transfers in Portugal have a fairly low redistributive impact compared with other EU economies. This low impact is the outcome of the low value of those transfers as in Portugal social transfers are more efficiently targeted to low-­income groups. Income taxes have a more important redistributive effect in Portugal resulting from the higher progressivity of the tax system. Furthermore, the high share of total income taxes paid by the top deciles in Portugal (one of the highest in the EU) is mainly the outcome of the country’s high income inequality as average tax rates in those same deciles are similar to the EU average. 3.6  The Crisis Finally, the relative resilience of middle-­income groups in Portugal has to be seen in light of the effects of the crisis. These include the relatively moderate consequences of the 2007–08 financial crisis, but also the aftermath

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 460

28/09/2016 08:25

461

VAUGHAN 9781786430595 PRINT (M4051).indd 461

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:25

Median equivalized net income Median income before social transfers (inc. pensions) Median income before social transfers (ex. pensions)

Median income before social transfers (inc. pensions)

Median income before social transfers (ex. pensions)

4000

Portugal (€)

Median equivalized net income

1995 1996 1997 1998 1999 2000 2001 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

Figure 11.17 Median equivalized disposable income – the role of social transfers, Portugal and EU, 1995–2012

Source:  Eurostat, SILC data, extracted 5 June 2015.

4000

9000

9000

19000

14000

EU15 (€)

14000

19000

1995 1996 1997 1998 1999 2000 2001 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

462

Europe’s disappearing middle class?

6

6

4

4

2

2

0

0

–2

–2

–4

–4

–6

–6

–12

Total market income

–8

Social transfers

Disposable income ( )

–10

Greece Iceland Ireland Spain Mexico Portugal Estonia Italy Netherlands Hungary Japan Luxembourg New Zealand Slovenia Czech Republic United States France Denmark Belgium Germany Israel Australia Canada Norway Finland Turkey Korea, Rep. of Austria Sweden Chile Poland Slovakia

–10

Income taxes

OECD-29

–8

–12

Source:  OECD (2014).

Figure 11.18 Taxes and social transfers and annual percentage change in household disposable income by income component, OECD countries, 2007–11 (percentage) of those effects exacerbated by the sovereign debt turbulence and the bailout. Our main argument is that the implementation of austerity measures resulted in the collapse of the expectations generated by almost three decades of EU membership. At the time of writing, the full effects of the austerity period (mainly the years 2011 to 2013) are not yet fully represented in the SILC data. From the available evidence, however, it seems that the bottom-­and top-­ income groups were harder hit than middle-­income groups in a context that resulted in the impoverishment of all income groups. Add to this qualitative evidence on middle-­ class perceptions and reactions to this impoverishment trajectory and we may argue that the generalized lack of trust in the future that has emerged follows the sudden break in expectations induced by the EU-­led modernization and education efforts. Perhaps the most representative sign of this distrust is the emigration of highly educated youngsters that increased in response to rising unemployment and the impact of the austerity policies. Emigration flows have indeed risen since 2010 as a result of such policies and now account for around 110 000 people each year, a level that resembles only those witnessed in the 1960s and 1970s. This emigration is traditionally young but is now increasingly well qualified, too. What could be seen in another context as an indicator of modernity and cosmopolitism – temporary migration has

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 462

28/09/2016 08:25



Inequality, labour market and middle-­income groups in Portugal ­463

indeed also risen – has also to be interpreted in the present situation as a sign of blocked aspirations and of the destruction of the hopes created by the EU-­led modernization.

4. CHANGES IN WAGES AND WORKING CONDITIONS The main trends in the Portuguese labour market in the past two decades are the small increase in total employment (together with a much higher increase in the total labour force) and the huge increase in unemployment. Diversities among sub-­periods emerge, defining three different phases (Figure 11.19). From the mid-­1990s to 2002 there was an increase in both labour force and employment; from 2002 to 2008 the labour force increased (at a slower rate) and employment stabilized; after 2008 both the labour force and employment decreased at very different paces – ­employment decreased sharply (to a level in 2013 similar to the 1995 level), while the labour force decrease has been much smoother. The huge rise in unemployment is, then, the most striking novelty of the Portuguese labour market in recent years (Figure 11.20). Despite increasing since the early 2000s, Portuguese unemployment rose sharply after the emergence of the recent crisis, as one of its major outcomes. Traditionally low unemployment rates became high by EU standards and recent values contrast with previous cyclical maximums. 5600.0 5400.0 5200.0 5000.0 4800.0 4600.0

Labour force (15 and +)

4400.0

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

4200.0

Total employment

Source:  PORDATA, available at http://www.pordata.pt/Home.

Figure 11.19  Labour force and employment, Portugal, 1995–2014 (000s)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 463

28/09/2016 08:25

464 20 15

Europe’s disappearing middle class?

European Union (15 countries) Portugal

10

0

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

5

Note:  There is a break in the series of labour market indicators in 2010. Data from 2010 and 2011 are thus not strictly compatible. Source:  Eurostat, Labour Force Survey.

Figure 11.20 Unemployment rate, Portugal and EU, 1990–2014 (percentage) Also, long-­(and very long-­) term unemployment rates8 became historically high in terms of both national and EU standards (Figure 11.21). Unemployment increased transversely according to all relevant criteria: in all age groups, at all schooling levels and for men and women. Another unfavourable dimension of Portuguese employment relies on the high share of temporary contracts in total employment (21.4 per cent of total employment in 2013 compared with 13.7 per cent in the EU15; in that same year only Poland and Spain had a higher share of temporary contracts9). 4.1  Employment, Earnings and Household Income While labour market inequality is arguably a fundamental determinant of the relative position of the middle classes in the income distribution, it is by no means the only determinant. Middle classes should be defined, precisely, by their greater reliance on earnings and employment as sources of income. This also means, however, that social transfers, on the one hand, and wealth-­(or pension-­) related income components, on the other hand, may have a more important role to play in determining the evolution of relative income, respectively, at the bottom and top of the wage distribution. Figure 11.22 uses EU-­SILC data to plot the share of employees among all

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 464

28/09/2016 08:25

Very long-term unemployment rate

Long-term unemployment rate 10 9 8 7 6 5 4 3 2 1 0

1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

10 9 8 7 6 5 4 3 2 1 0

Inequality, labour market and middle-­income groups in Portugal ­465

1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013



European Union (15 countries)

European Union (15 countries)

Portugal

Portugal

Source:  Eurostat, Labour Force Survey.

Figure 11.21 Long-term and very long-term unemployment rate, Portugal and EU, 1992–2012 (percentage) individuals in each equivalized disposable income decile in 2003 and 2012. It also shows the share of (self-­reported) unemployed, retired and inactive individuals (other than retirees or those in education or training) across these same deciles.10 It is clear that only a minority of those at the bottom of the income distribution are employees (and can therefore rely on earnings as a major source of their income). The 2003 to 2012 period also appears to be characterized by important flows out of employment (a point to which we return in detail below) with the share of employees among all individuals in each decile decreasing, particularly among middle-­income groups (from the third to the eighth decile). Conversely, there are important increases in the share of unemployed within the deciles and particularly in the bottom deciles and decreasing as we move up the income distribution, as well as in the share of retired people, particularly above the fifth decile. By contrast, the share of retired people in the bottom two deciles actually decreased over this period. The data appear to indicate, therefore, a very differentiated impact of the flows into unemployment and retirement across the income distribution. While the former appears to be increasingly relevant to explain the relative situation of the lower (poorer) income groups, the latter appears to be an important factor in the relative position of middle-­ income groups. Looking now at the differences in individuals’ self-­reported activity status across the income distribution in 2003 and 2012 in terms of the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 465

28/09/2016 08:25

466

Europe’s disappearing middle class? Share of employees

Share of unemployed

0.6

0.35 0.3

0.5

0.25

0.4

0.2

0.3

0.15

0.2

D9

D8

D7

D6

D10

Share of retirees

0.4

D5

2004

D10

2013

D4

D3

D1

D9

D10

D8

D7

2009

D9

2004

D6

D5

D4

D3

0 D2

0 D1

0.05 D2

0.1

0.1

2013

Share of inactive

0.25

0.35 0.2

0.3 0.25

0.15

0.2

0.1

0.15 0.1

0.05

0.05 0

2004

2013

2004

D8

D7

D6

D5

D4

D3

D2

D1

D10

D9

D8

D7

D6

D5

D4

D3

D2

D1

0

2013

Source:  EU-SILC (project calculations).

Figure 11.22 Activity status by decile of equivalized disposable income, Portugal, 2004–13 groups addressed in this volume, more detailed information on middle-­ income groups emerges (Figure 11.23). It is clear, for example, that among those in poorer households or in lower middle-­income groups (60–80 per cent of the median), only a minority (close to 30 per cent in 2012) were in either part-­time or full-­time employment. Inactivity and unemployment rates, therefore, are much higher, particularly among those in poorer households. A particularly striking finding, however, relates to the significant increase in the risk of poverty associated with unemployment during the 2003–12 period. It is true that the share of employed individuals has decreased and the share of retired people grew during this period throughout the whole distribution. Among

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 466

28/09/2016 08:25



8.60

27.28

0

2004

2013

< 60% median

34.80

7.82

2.39

26.80

4.67

4.88

11.91

8.10 15.73

2004

53.91 26.28

36.86

23.14

29.21

10

7.7

6.28

30 20

7.0

4.6 5.3

57.27

40

6.8

4.9 8.8

5.4 5.6

62.61

50

28.21

18.06

60

27.56

70

7.7

19.03

7.2

6.9

7.4

24.22

8.8

9.2

54.40

7.3

12.3

14.3

13.94

17.7

62.09

80

18.5

45.40

90

25.34

100

18.22

Inequality, labour market and middle-­income groups in Portugal ­467

2013

60%–80%

2004

2013

80%–120%

2004

2013

2004

120%–200%

2013

>200% median

Equivalized disposable income groups Employed FT

Employed PT

Unemployed

Retired

Education or training

Other inactive

Source:  EU-SILC (project calculations).

Figure 11.23 Activity status by individuals’ relative equivalized disposable income, Portugal, 2003–12 (percentage) the poor, however, the share of those in unemployment increased very significantly while the share of those in retirement has actually decreased, contrary to the trend across the remaining groups in the distribution. 4.2  Wages and Earnings in the Private Sector mentioned importance of employment in household Given the above-­ income it is important to look in more detail at recent changes with regard to wages, presented in Figure 11.24: an increase of monthly real wages from the mid-­1990s to the early 2000s (slower for minimum wages), stability of real wages in the years before the crisis, big increase from 2008 to 2010 and decrease after 2010. The existence in Portugal of a dataset that registers information for all private sector employees (Quadros de Pessoal, an administrative dataset collected annually by the Portuguese Ministry of Employment) allows us to access information on workers’ attributes (such as gender, age, education, occupation, qualification level, years with the firm, hours worked and earnings) and job-related attributes (such as type of industry, geographical

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 467

28/09/2016 08:25

468

Europe’s disappearing middle class?

1150 1050 950 850 750 650 550

Average monthly wage

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

350

1995

450

Average monthly gain

Monthly minimum wage Source:  PORDATA, available at http://www.pordata.pt/Home.

Figure 11.24  Monthly real wages, Portugal, 1995–2013 (2011 prices) location and plant size). Response to the questionnaire is mandatory for all private sector firms with at least one employee.11 A detailed analysis of this database allows us to draw some additional insights on recent dynamics of wages and earnings. We first gathered information on the general evolution of wages in the period 2005–12 (Table 11.3).12 We started by analysing the situation of workers employed both in 2005 and 2012 (a total of 1 374 057 employees). As expected, wages differ with age (lower for young people and increasing with age), with education (much higher for the highly educated) and gender (with a gender wage gap). Median wages for those who were employed in both 2005 and 2012 increased slowly over the period. However, this increase has been diverse across different population groups: higher for young people (below 30 years of age in 2005) and for the highly educated and stagnating or decreasing for older people. Expanding the analysis to all workers in both years we can see that substantial flows of workers did occur over the period (Table 11.4). A reduction of total employment in the Portuguese private sector (305 159 fewer employees in 2012 compared with 2005) is the outcome of three diverse situations: workers who remained employed (2005 and 2012), workers who were not employed in 2005 but were in 2012 (inflow) and workers who were employed in 2005 but not in 2012 (outflow). Three main changes deserve to be stressed: (1) the strong reduction of employment among young people;

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 468

28/09/2016 08:25



Inequality, labour market and middle-­income groups in Portugal ­469

Table 11.3  Median real hourly wage, Portugal, 2005 and 2012 (euros)

Total Age =  45

2005 and 2012

Flows 2005–12 Flow in

Flow out

Net flow

2 596 231

2 291 072

1 374 057

917 015

1 222 174

−305 159

742 444

463 813

420 786

43 027

321 658

−278 631

1 153 470

1 080 011

685 171

394 840

468 299

−73 459

700 317

745 082

268 100

476 982

432 217

44 765

1 890 366

1 202 016

688 350

1 097 811

−409 461

400 706

172 041

228 665

124 363

104 302

1 212 014 1 079 058

758 889 615 168

453 125 463 890

710 343 511 831

−257 218 −47 941

Education 2 299 827 Non-­  tertiary education Tertiary 296 404  education Gender Male Female

2012

1 469 232 1 126 999

Source:  QP dataset (Figueiredo et al. 2015).

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 469

28/09/2016 08:25

470

Europe’s disappearing middle class?

(2) the heavy concentration of job destruction on men’s jobs; and (3) the noticeable increase in employment of highly qualified workers. As regards the impact of these changes on wages, the most noticeable aspect is that the evolution of the median wage between 2005 and 2012 became negative for the highly educated, meaning that the increase reported in Table 11.4 applies only to highly educated insiders. For highly qualified newcomers, wages are lower than they were for those who entered the labour market in 2005 or earlier. In a related analysis using the same dataset, Figueiredo et al. (2015) also show that inter-­decile wage mobility decreased from 2005 to 2012, particularly for the middle-­wage groups. This provides further evidence of relative lethargy in the Portuguese labour market, affecting both the relative situation of the young and the upward mobility of the middle classes.

5.  CASE STUDIES We conducted two case studies to highlight particular dimensions of the Portuguese labour market, aimed at establishing a closer link between changes in the labour market situation and the income distribution. Case study 1 discusses job polarization, while case study 2 highlights some issues related to public employment. 5.1 Case Study 1: The Structural Transformation of Employment and the Middle Classes: Polarization or Upgrading? The possible polarization or hollowing-­out of the labour markets of the most developed economies (Autor et al. 2008; Autor and Dorn 2013) is crucial to understanding the relative position of middle-­income groups. If jobs at the bottom and the top of the wage and skill distributions grow at the expense of medium-­skilled occupations, we may expect middle-­income groups to be disproportionally affected because their employment-­and earnings-­related components of income, in particular, could be significantly reduced (in relative terms) through the effect of employers’ reduced demand for their traditional jobs and skills. To date, however, the evidence on this polarization hypothesis is somewhat contradictory. Goos et al. (2014), for example, show that between 1993 and 2010 both high-­paying managerial, professional and associate professional jobs, as well as low-­ paying service and elementary jobs, increased their employment share in Portugal at the expense of middle-­level clerical, craft-­related and industrial (machine operators and assemblers) jobs. Fernández-­Macías (2012) argues, however, that the pervasiveness of job polarization in Europe – as

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 470

28/09/2016 08:25



Inequality, labour market and middle-­income groups in Portugal ­471

documented by those authors – is strongly dependent on the consideration of a common ranking of jobs (both by skill and wages) across countries and disregards differences in the relative importance (employment share or even average skill levels) of specific occupations in different economies. Using instead national-­level job rankings as well as wage and skill quintile measures, the author finds that the dynamics of the specific occupations used by Goos et al. can hide, in specific economies, significant structural upgrading. The author shows that this is the case for Portugal where, from 1995 to 2007, there was a very strong destruction of low-­paid (first wage quintile) jobs and a stronger creation of highly skilled (fifth skill quintile) jobs. The authors’ results further show that, when jobs are ranked according to their initial wage level, it appears to have been middling jobs (third and fourth quintiles) that experienced the strongest growth during this period. This tension between polarization and upgrading is, we believe, important for understanding the way in which recent changes in the structure of the Portuguese labour market can impact the relative position of middle-­ income groups and why we may expect that the existence of such polarization dynamics might not necessarily result in a very visible deterioration of their relative position. In this section and following Figueiredo et al. (2015), we look at changes in the composition of private sector jobs during 1995–2009, ranked according to their initial skill and wage levels. We again use the Quadros de Pessoal dataset, which collects the staff logs of practically all private sector firms with at least one employee. Jobs are here defined as specific four-­digit Classificação Nacional de Profissões (CNP, equivalent to International Standard Classification of Occupations, ISCO-­88) occupations. The period 1995 to 2009 is the longest available for which there are no breaks in the occupational classification available in this dataset. The non-­inclusion of public sector workers is an important limitation if we take into account the importance of this sector among highly skilled workers. We do, however, refer to the relative position of public sector versus private sector workers in the following section.13 We start by looking at the change in the employment share of specific jobs (in percentage points) ranked both by initial mean (median) hourly wage and skill (share of university graduates) levels. Figures 11.25 and 11.26 provide an initial visualization of the transformation of the Portuguese private sector labour market over this period, with the area of each bubble referring to the share of employment of each occupation in 1995. First, Figures 11.25 and 11.26 show that a relatively small number of occupations have the potential to account for a relatively large share of

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 471

28/09/2016 08:25

Change (in pp) in employment share, 1995–2009

472

Europe’s disappearing middle class? 2

0

–2

–4

–6 0

100

200

300

400

Rank of log hourly wage, 1995

Source:  QP dataset (Figueiredo et al. 2015).

Figure 11.25 Change in employment share of occupations ranked by initial (1995) hourly wage, Portugal, 1995–2009 employment in 1995. Changes in the representativeness of these jobs in employment over this period can then help us to understand a large part of the overall story of polarization or upgrading over this period. Second, they also make it clear that private sector employees were overly clustered in low-­wage and low/middle-­skilled occupations in 1995. Third, they do not display a very clear pattern of polarization. When jobs are ranked according to their initial earnings level, for example, there are both sizeable downward and upward shifts in the employment share of important low-­paid jobs. Similarly, when ranking jobs by their initial skill level, we can spot both large decreases and increases in the employment share of significant middle-­skilled occupations, as well as a number of relevant low-­ skilled occupations. In order to add detail to this story, Tables 11.5 and 11.6 provide a list of the top ten and bottom ten jobs when these are ranked by changes in their employment share over this period. These 20 jobs accounted for slightly more than 40 per cent of overall employment in 2009. In addition, Table  11.7 lists similar details about the ten biggest jobs (in terms of employment share) in 2009, which accounted for 35 per cent of overall

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 472

28/09/2016 08:25

Change (in pp) in employment share, 1995–2009



Inequality, labour market and middle-­income groups in Portugal ­473 2

0

–2

–4

–6 0

100

200

300

400

Rank of occupational skill level, 1995

Source:  QP dataset (Figueiredo et al. 2015).

Figure 11.26 Change in employment share of occupations ranked by initial (1995) skill level, Portugal, 1995–2009

employment. In addition to the information on changes in their employment share, these three tables also provide details about the skill intensity of these occupations (measured by the share of university educated individuals among these workers), their median hourly wage levels in 2009, as well as how they compare with other jobs in terms of the age structure of their workforces. The story that these three tables convey appears, at first glance, to show strong points of contact with Goos et al. (2014), as well as the broader ten growing jobs includes, for polarization narrative. The list of top-­ example, relatively low-­skill personal and customer-­related services jobs (for example, 5123, 5122, 5220, 4211), as well as relatively high-­skill jobs, such as business professionals. The list of fast-­decreasing jobs, in turn, also includes many of the occupations usually associated with middle-­level jobs in the international literature, namely office clerks (4122), craft and related trade workers (7436, 7442), machine operators and assemblers (8262, 8264, 8282). A crucial difference regarding the Portuguese productive structure, however, is that many of these latter jobs – which are considered middle-­skilled and middle-­wage occupations in the context of other

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 473

28/09/2016 08:25

474

VAUGHAN 9781786430595 PRINT (M4051).indd 474

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:25

5123 4121

4190 5122 7129

5133

2419 5220

4211 Total

 2  3

 4  5  6

 7

 8  9

10

Helpers and cleaners (offices, hotels, . . .) Waiters Accounting and bookkeeping clerks Other office clerks Cooks Other building frame and related trades workers Home-­based personal care workers Other business professionals Shop sales persons and demonstrators Cashier and ticket clerks 0.01 25.5%

0.01 0.07

0.76 0.67 0.64 9.42

0.01

0.01 0.02 0.01

0.04 0.01

0.06

Employment share 09

0.79

1.00 0.94 0.88

1.27 1.13

1.34

Change in emp share (pp)

3.63

9.10 4.10

3.77

4.69 3.43 3.85

3.27 5.03

3.23

Median hourly wage 09

0.05

0.59 0.05

0.01

0.18 0.02 0.02

0.03 0.20

0.02

Share of graduates

Source:  QP dataset (Figueiredo et al. 2015).

Note:  Age ratios refer to the ratio of the occupation workers’ average age and the average of workers in all occupations.

9132

 1

Job

Table 11.5 List of top ten fastest increasing jobs (in employment share), Portugal, 1995–2009

0.94

0.96 0.88

1.09

1.05 1.05 1.01

0.87 1.04

1.11

Age ratio 95

0.91

0.93 0.86

1.12

0.96 1.09 1.02

0.89 0.98

1.14

Age ratio 09

475

VAUGHAN 9781786430595 PRINT (M4051).indd 475

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:25

7436

7442

8282

8320 8264

8262

3415

7122 Total

 3

 4

 5

 6  7

 8

 9

10

Statistical and finance clerks Hand packers and other manufacturing labourers Sewers, embroiderers and related workers Shoe-­makers and related workers Electrical-­equipment assemblers Motor vehicle drivers Bleaching, dyeing and cleaning-machine operators Weaving-­and knitting-­ machine operators Technical and commercial sales representatives Bricklayers and stonemasons −0.54 −15.19

−0.55

−0.66

−0.86 −0.66

−0.96

−1.10

−2.59

−4.53 −2.73

0.020 15.2%

0.019

0.003

0.000 0.003

0.003

0.010

0.024

0.054 0.017

3.93

8.52

3.75

2.81 3.44

4.86

3.15

3.00

6.03 3.63

Change in emp Employment Median hourly share (pp) share 09 wage 09

0.02

0.24

0.00

0.27 0.01

0.01

0.00

0.00

0.22 0.02

Share of graduates

Source:  QP dataset (Figueiredo et al. 2015).

Note:  Age ratios refer to the ratio of the occupation workers’ average age and the average of workers in all occupations.

4122 9322

 1  2

Job

Table 11.6  List of ten fastest decreasing jobs (in employment share), Portugal, 1995–2009

1.02

1.02

1.02

1.10 0.97

0.82

0.86

0.84

0.97 0.92

Age ratio 95

1.09

1.00

1.07

1.05 1.04

0.98

0.99

0.99

0.98 0.96

Age ratio 09

476 −3.85

0.94 1.27 1.34

0.67

0.48

−0.32

−0.54

−0.55

−2.59

−4.53

3.43 3.27 3.23

4.10

5.48

3.52

3.93

8.52

3.00

6.03

35.1%

0.023 0.042 0.055

0.072

0.021

0.020

0.020

0.019

0.024

0.054

0.02 0.03 0.02

0.05

0.01

0.03

0.02

0.24

0.00

0.22

Source:  QP dataset (Figueiredo et al. 2015).

1.05 0.87 1.11

0.88

1.14

0.93

1.02

1.02

0.84

0.97

1.09 0.89 1.14

0.86

1.11

0.96

1.09

1.00

0.99

0.98

Change in emp Median hourly Employment Share of Age ratio Age ratio share (pp) wage 09 share 09 graduates 95 09

Note:  Age ratios refer to the ratio of the occupation workers’ average age and the average of workers in all occupations.

Total

5122 5123 9132

6 4 2

8324

7

5220

9312

9

1

7122

8

7436

5

3415

Statistical and finance clerks Sewers, embroiderers and related workers Technical and commercial sales representatives Bricklayers and stonemasons Construction and maintenance labourers (roads, . . .) Heavy-­truck and lorry drivers Shop sales persons and demonstrators Cooks Waiters Helpers and cleaners (offices, hotels, . . .)

4122

3

10

Job

Emp share rank

Table 11.7 List of ten most representative jobs (in employment share), Portugal, 2009

Decreasing

Increasing

VAUGHAN 9781786430595 PRINT (M4051).indd 476

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:25



Inequality, labour market and middle-­income groups in Portugal ­477

economies – cannot be classified as such here. It is informative to consider, for example, that the type of low-­skilled personal-­and customer-­related services we mentioned above do not pay necessarily less on an hourly basis than the manufacturing-­related jobs that have been destroyed to a greater extent.14 That is, while the shift from manufacturing to service-­related jobs appears clearly, this does not necessarily affect very different groups across the wage distribution, which may mean that there will be much less severe consequences for the fate of the Portuguese middle classes than might be expected in more developed economies.15 By contrast, the decrease in the relative importance of clerical or technical/sales representatives jobs (for example, 4122 and 3415) may have more substantial consequences for the middle classes; in particular, to groups located relatively high up in the earnings distribution. In order to provide a more systematic account of such changes in the employment structure of the Portuguese economy, Figures 11.27 and 11.28 look at the changes in the employment share of wage-­ranked and skill-­ranked job quintiles during this period. In order to construct these quintiles, we initially ranked occupations based on their 1995 hourly wage and skill levels and then allocated these to specific quintiles, weighting each occupation by its number of employees. Each quintile should, therefore,

Change in employment share, 1995–2009

10

5

0

–5 1

2

3

4

5

Source:  QP dataset (Figueiredo et al. 2015).

Figure 11.27 Changes in the employment share by wage-ranked 1995 quintile, Portugal, 1995–2009

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 477

28/09/2016 08:25

478

Europe’s disappearing middle class?

Change in employment share, 1995–2009

15

10

5

0

–5

1

2

3

4

5

Source:  QP dataset (Figueiredo et al. 2015).

Figure 11.28 Changes in employment share by skill-rank 1995 quintiles, Portugal, 1995–2009

correspond to close to 20 per cent16 of the overall number of employees in 1995. In both figures, it is possible to see that the most important changes during this period were the increase in the representation of the top-­skilled and top-­paying occupations in the initial year, as well as the important shift away from the bottom-­paid and bottom-­skilled occupations. This is largely consistent with the modernization and upgrading of the Portuguese productive structure. Also, when we look at skill quintiles, the pattern of upgrading is clear. Occupations that had already a higher share of university graduates in 1995 appear to have been the ones that increased the most in terms of their employment share. In order to contextualize this finding it is important to remember that the educational attainment levels of the Portuguese workforce also increased significantly during this period. We do not see, however, an unambiguous pattern of upgrading, namely when we look at the occupations’ initial hourly wage level. There is, in particular, a strong decrease in the employment share of fourth quintile occupations which can perhaps be better understood by referring to some of the occupations we looked at in Tables 11.5–11.7. In order to shed further light on this issue, however, we decompose the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 478

28/09/2016 08:25

Change in employment share 1995–2009(pp)



Inequality, labour market and middle-­income groups in Portugal ­479 10 8 6

Other jobs

Middling

Low-paying

High-paying

4 2 0 –2 –4 –6 –8

1

2

3

4

5

Wage quintiles Source:  QP dataset (Figueiredo et al. 2015).

Figure 11.29 Decomposition of wage quintile changes in employment share by type of job, Portugal, 1995–2009 changes in the employment share of each quintile according to the contribution of four groups of jobs using Goos et al.’s (2014) classification of ISCO two-­digit jobs into low-­paying (51, 52, 91 and 93), middling (41, 42, 71–74, 81–83) and high-­paying (12, 13, 21, 22, 24, 31, 32 and 34) jobs.17 Figures 11.29 and 11.30 present the results of this exercise. It is striking that the revealed pattern is now clearly compatible with the polarization hypothesis, with most job destruction happening at the level of so-­called middling jobs. A large portion of these middling jobs are, however, in the case of the Portuguese economy the lowest paid jobs, making this pattern of polarization compatible with some degree of apparent structural upgrading. In turn, jobs which in the international literature are considered low-­pay are, in the context of the Portuguese economy, relatively skilled (in the sense of absorbing a higher percentage of university graduates), thus reinforcing the possibility of simultaneous upgrading and polarization. Finally, we take a very quick look at changes (log differences) in all jobs’ nominal hourly wages across this period (Figures 11.31 and 11.32). If the decline of employment opportunities in the lowest-­paid jobs or quintiles of the distribution resulted in the displacement of a large number of workers and if these workers remained actively looking for similar jobs, we could expect strong downward pressure on wages. Conversely, if the creation of high-­paying and high-­skilled occupations greatly exceeded

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 479

28/09/2016 08:25

Changes in employment share, 1995–2009(pp)

480

Europe’s disappearing middle class? 15 10

Other jobs

Low-paying

Middling

High-paying

5 0 –5 –10

1

3

2

4

5

Skill quintiles Source:  QP dataset (Figueiredo et al. 2015).

Figure 11.30 Decomposition of skill quintile changes in employment share by type of job, Portugal, 1995–2009

Log difference of hourly wages, 1995–2009

1.5

1.0

0.5

0

–0.5 0

100

200 300 Rank of log hourly wages 1995

400

Source:  QP dataset (Figueiredo et al. 2015).

Figure 11.31 Change in hourly wages (log difference) of occupations ranked by 1995 wage level, Portugal, 1995–2009

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 480

28/09/2016 08:25



Log difference of hourly wages, 1995–2009

Inequality, labour market and middle-­income groups in Portugal ­481 1.5

1

0.5

0

–0.5 0

100

200

300

400

Skill level rank, 1995

Source:  QP dataset (Figueiredo et al. 2015).

Figure 11.32 Change in hourly wages (log difference) of occupations ranked by 1995 skill level, Portugal, 1995–2009

increases in the supply of workers with the necessary skills for such jobs we could anticipate increases in the relative wages of these jobs, with wage inequality growing as a result. The figures we present here, however, convey a different story because the lowest-­skill and lowest-­paid appear to be those with larger relative increases in their hourly wages. There are two possible – rather speculative – interpretations. On the one hand, given the very low levels of pay in the Portuguese economy we would expect that minimum wage legislation may have played a particularly important role in avoiding further decreases in the hourly wages of the lowest-­paid jobs. Furthermore, the increases in the minimum wage levels witnessed during this period may have contributed to the larger relative increases in these jobs’ hourly earnings. Second, this period is also characterized by important shifts in the educational attainment of the workforce. In 1995, for example, there were only slightly more than 4 per cent of university graduates in the whole dataset we used for these calculations. In 2009, the respective value was 15.5 per cent, with much higher values among the young. This means that the young and higher skilled are unlikely to search for the same jobs as the previous, less-­qualified generations of workers, thereby increasing the supply of workers competing for jobs in the top quintiles. In

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 481

28/09/2016 08:25

482

Europe’s disappearing middle class?

the opposite direction, flows into inactivity of the lowest skilled may also have played a role in curbing downward pressure on the earnings of the lowest-­paid occupations. At this stage these are fairly speculative hypotheses rather than well-­grounded conclusions. In sum, we can – tentatively – conclude that the type of structural upgrading we have just described – and therefore the greater availability of better paid jobs – may prevent, for the moment, a visible deterioration of the relative situation of the groups located in the middle of the skill or wage distributions. At the same time, the fact that the polarization story – as it is consistently described for other more advanced economies – also seems to lurk behind this upgrading suggests that we may be entering a period in which the level of wage inequality at the top of the wage distribution is likely to grow significantly. In other words, as the skill levels of the workforce grow and as relatively well-­paid middling jobs disappear, a growing gap is likely to emerge between those able to access what in the future will be called bad and good jobs. 5.2  Case Study 2: Public Employment and the Middle Class Recent changes in employment in Portugal have strongly affected the public sector. Early increases in employment between the mid-­1990s and mid-­2000s were higher in the public sector, but job destruction during the period 2005–14 has had a much higher relative impact in the public sector (Figure 11.33). Four main sectors comprise public employment: public administration (including security forces), education, health care and regional and local administration. In 2014 they accounted, respectively, for 22.9 per cent, 31.2 per cent, 18.7 per cent and 22.6 per cent of total public employment, and these shares have remained relatively stable since 2011. Comparisons with previous periods are impossible given the change of classification criteria in 2010. The overall story of job reduction in the public sector, however, includes a number of other important stories regarding particular job groups. We will concentrate on the education sub-­sector, given its magnitude and potential impact in creating upward mobility (into or out) of the middle-­income groups. The reduction of public sector employment in the period described in Table 11.8 was mainly the result of two dynamics: the resilience of education jobs before the crisis and the huge decrease in employment in the sector after 2011. Looking in detail at the professions more affected by public employment cuts, teachers appear to be the most affected group (37.8 per cent of the total reduction in the period 2011–14). Reference to the education sector is also particularly important as Portugal’s labour force18 has the highest education deficit in the EU

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 482

28/09/2016 08:25



Inequality, labour market and middle-­income groups in Portugal ­483 6.00 4.00

Public sector 3.88

Total

3.48

2.00

0.72 0.41

0.00 –0.39 –2.00

–3.39

–4.00 –6.00

–1.72

–2.33

–2.53 –4.67 1996–1999

1999–2005

2005–2009

2009–2011

2011–2014

Source:  PORDATA, available at http://www.pordata.pt/Home; DGAEP, several years, available at http://www.dgaep.gov.pt.

Figure 11.33 Annual average growth rate of employment, Portugal, 1996– 2014 (percentage)

Table 11.8  Public employment, Portugal, 1996–2014 Change in public employment

1996–99 1999–2005 Basic public  administration (including security forces and army) Education Health care Regional and local  administration Others Total

Share in public employment (%)

2005–09

2011–14

(2014)

24 906

4795

−13 145

−18 081

22.9

24 780 5869 11 361

16 976 −610 19 603

−14 232 −17 945 −34 561

−27 039 −5 430 −15 100

31.2 18.7 22.6

10 458 77 374

−9302 31 462

−10 160 −90 043

−5 824 −71 474

4.6 100.0

Source:  DGAEP, several years.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 483

28/09/2016 08:25

484

Europe’s disappearing middle class?

100 80 60 40

Tertiary

Secondary

UK

Sweden

Finland

Portugal

Austria

Netherlands

Luxembourg

Italy

France

Spain

Greece

Ireland

Germany

Denmark

Belgium

0

EU15

20

Primary

Source:  Eurostat, Labour Force Survey [lfsa_egaed].

Figure 11.34 Share of employment by level of education, Portugal, 2014 (percentage) (Figure  11.34). Investment in education is thus particularly important there. The public sector is also distinct because it is a major employer of specific groups. Public employment has an overrepresentation of women (in 2014 women’s share in public employment was 59 per cent compared with 48.2  per cent in total employment) and of the highly educated (50.3 per cent compared with 23.9 per cent). Also, the Portuguese public sector is ageing: 54.5 per cent of public employees were 45 years old or more in 2014 (45.9 per cent in total employment). Again, certain groups deserve particular analysis: in the case of teachers of pre-­primary, primary and secondary education the share of that age group was, in that same year, 61.2 per cent. The sustainability of the investment in education in Portugal may be threatened by the inability to renew the teaching staff. As regards pay and, as expected based on the abovementioned average age and education of public employees, pay is higher in average terms than in the private sector: in 2013, public average monthly pay was 1586.15 euros compared with average private monthly pay of 1093.82 euros.19 However, again as expected, the situation varies between different occupations (Table 11.9). The relevant sub-­sectors illustrate the diversity of pay in the public

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 484

28/09/2016 08:25



Inequality, labour market and middle-­income groups in Portugal ­485

Table 11.9  Monthly average wage in public sector, Portugal, 2013 Average monthly pay Variation 2014 (€) 2014/2011 (%) Public sector Public administration Foreign affairs, defence, security and justice Education Elementary, primary and secondary Tertiary Health care

1605.53 1273.36 1809.73 1754.20 1664.92 2326.26 1662.58

0.04 −0.65 3.47 0.36 0.04 −0.71 −3.12

Source:  Direção-­Geral da Administração e do Emprego Público (DGAEP), public employment dataset, several years, available at http://www.dgaep.gov.pt/index. cfm?OBJID=3b2e32ef-­9ea6-­4e93-­9417-­73f3a8b74500.

sector. Nevertheless, comparing the values of the public and private sectors the ‘average public worker’ would qualify for the eighth decile in the income distribution of the private sector. The relatively high wages, the stability and the prestige associated with most public sector careers have long been incentives to highly qualified workers. Furthermore, the public sector has generated a lot of qualified jobs taken by various generations of better qualified young people.20 The public sector has thus been an important conduit of the upward social mobility made possible by education. The recent closing of job opportunities, wage cuts and a (often political) campaign to depict public sector employees as a ‘privileged’ group have been changing this scenario, posing new challenges to the country. This organized vilification of public careers creates obstacles for retaining the best workers and demotivates those who stay. The ageing of the current labour force also poses new sustainability questions. Training of new entrants by the older ones is a challenge and the absence of public job opportunities for better qualified young people is closing the door to the renewal and modernization of public services.

6.  CONCLUSION AND POLICY ISSUES In recent times, particularly since the crisis, Portuguese middle-­income groups have showed substantial resilience in terms of their capacity to access an increasing (only slightly interrupted by a more stable trend since the crisis) share of national income. This appears to contradict the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 485

28/09/2016 08:25

486

Europe’s disappearing middle class?

widespread view that the current crisis is undermining the middle classes more than other groups and that this is generating higher inequality through polarization. Since the beginning of the 2000s (again, slightly interrupted by a more stable trend since the crisis) the Portuguese income distribution also shows a clear reduction in the income share of the top deciles. The main question posed in this chapter is thus the following: can Portugal be presented as a successful story of income redistribution? The global answer is that there have been successes, but only partial ones, related to the social redistribution mechanisms put in place at the establishment of the Portuguese welfare state. This process has historical particularities in Portugal that must be kept in mind, chiefly that we are looking at a recent (41 years old) democracy boosted by an EU integration process dating back to the mid-­1980s. Portugal’s backwardness compared with the core countries of the EU persists. However, the progressivity of the tax system, together with social transfers, has played a positive role. Despite important changes and drawbacks in social policy, the redistribution mechanisms of the Portuguese welfare state have been sufficiently resilient that high-­income groups have been relatively harder hit than middle-­income groups. Nevertheless, we emphasize that success is only partial. Why so? First, Portugal has long been a very unequal country. The decrease in inequality since the mid-­2000s was interrupted by the crisis and subsequent ‘austerity’ policy, which halted the ongoing convergence with EU patterns. Second, the income distribution in Portugal is much more concentrated in the top-­income groups than in the core EU countries. Both the share of households having a disposable income above 200 per cent of the median and the share of income they retain are high by EU standards. Third, and in line with the previous point, the middle class (the middle-­ income groups as defined by the present project) is smaller in Portugal than in the core EU countries. Conversely, high-­income groups are larger and more diverse according to EU standards. The fact that around 25 per cent of public employees come into the high-­income group illustrates the specific nature of public sector employment and pay in Portugal. Fourth, the absolute level of household income is low by EU standards. Relative changes are thus only part of the story of a country that, 30 years after joining the European Union, remains in a low-­income trajectory. It has to be emphasized that the crisis introduced a clear change, contributing to the country’s further impoverishment. It has also become clear in the course of our analysis that the emergence of the 2007–08 crisis affected Portugal in several ways relevant to the relative position of middle-­income groups. The impoverishment process started with austerity and has been transversal, although it affects different

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 486

28/09/2016 08:25



Inequality, labour market and middle-­income groups in Portugal ­487

groups differently. The richer, but also the poorer, income groups were disproportionately affected with regard to income in comparison with the middle. However, the disposable income of middle-­income groups also decreased. The impoverishment process has also interrupted a trend towards more equality that has emerged since the mid-­2000s, reintroducing divergence as regards the rest of the EU. In our view, we should not ignore the impact of this dynamic of divergence on the expectations of middle-­income groups, defined by their aspiration to upward mobility and convergence with the consumption patterns of the more developed EU countries. The empirical data allow us to argue that these expectations are being hit in all dimensions: income is decreasing, upward mobility is less pervasive and the gap with the core EU member states is increasing. Furthermore, we have also argued that, while the structural upgrading of the Portuguese employment (occupational) structure appears to be contributing to the creation of better paying jobs, this is to some extent going hand in hand with a growing polarization of employment opportunities. Such polarization, by increasing both bad and good jobs, can contribute to further inequality and difficulties for middle-­income groups in the future. We also collected evidence on the role of education in upward income mobility. The highly educated represent the majority in high income groups, but only a small share of the middle-­income groups. Because the private sector’s ability to create qualified jobs remained low and the construction and improvement of the welfare state needed qualified workers, the main employer of a growing (but still small) group of highly educated workers has been the public sector. Again, the crisis and ‘austerity’ policies threaten this mechanism of upward mobility from a number of perspectives. The reduction of public employment has practically closed new job opportunities in the public sector and the private sector is only marginally providing for new qualified entrants. This leads to quite contradictory evidence: a country with a high education deficit has a high and increasing unemployment rate among the highly educated.21 Diverse dynamics with direct or indirect effects on middle-­income groups emerge from this contradiction. The upward path made possible by education is being called into question in several ways, potentially leading to less intensive private investment in education. Also, public investment in education is under pressure (falling number of employees in the sector, ageing teaching staff, decline of prestige fuelled by political expediency), perhaps threatening the renewal and sustainability of the education system. It has also become clear that the labour market situation is, as expected, crucial with regard to the possibility of integrating middle-­ income groups: being employed or, more surprisingly, being retired increased the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 487

28/09/2016 08:25

488

Europe’s disappearing middle class?

probability of being a member of the middle-­income groups, while being unemployed increased the probability of joining low-­income groups. High unemployment has reduced upward mobility and reduced trust in education as an elevator to a better future for young people, undermining the expectations of middle-­income groups (in fact, of all groups). In this way, it is threatening the future prospects of economic and social development.

NOTES   1. We refer to SILC data using the reference year, not the publication year. Thus data referring to 2012 are those reported by Eurostat–SILC as 2013 values.   2. The groups are constructed as follows: groups 1, 2 and 3 are low-­income groups and represent households with an equivalized median income of the following intervals: [0,20]%, [20,40]% and [40,60]% of the median equivalized income; groups 4 to 10 represent the core groups of our analysis, middle-­income groups. Group 4 [60,80]% of median equivalized income comprises the low middle-­income group; groups 5 and 6 [80,100]% and [100,120]% of median equivalized income comprise what can be called the middle middle-­income group; groups 7 to 10 – [120,140]% [140,160]% [160,180]% and [180,200]% – represents the high middle-­income group; and finally, groups 11 to 14  – [200,240]% [240,280]% [280,400]% and [above 400]% of equivalized medium income – represent the high income group.   3. The Portuguese distribution is similar to those of Greece, Italy, Hungary, Estonia and Lithuania.   4. See Appendix 11A.1 for a comparative analysis of deciles and income groups.   5. The detailed tables on these patterns of mobility can be found in Appendix 11A.2.   6. For a description of recent changes in unemployment benefits (with cuts, among other changes) in Portugal see González and Figueiredo (2015).   7. The share of employees refers to the fraction of individuals in a given class who are classified as wage employees. The sum of these fractions among the 14 classes equals 100 per cent. The share of public sector workers is the fraction of individuals in a given class who are classified as having worked in the public sector. The fraction refers to total public sector workers, so that the sum of these fractions among the 14 classes equals 100 per cent. For further details regarding these fourteen income groups refer to note 2.   8. Long-­term and very long-­term unemployment refer to the number of people who are out of work and have been actively seeking employment for at least 12 months and 24 months, respectively.   9. Eurostat, Labour Force Survey, accessed on 9 March 2015. 10. We use self-­reported information on activity status available on the EU-­SILC to define those in unemployment, retirement or in other inactivity status other than retirement or education and training. Employees, however, are defined as those who have worked at least one hour in the reference week and that are directly classified as employees. 11. Information about employees in the public administration, the self-­employed and military personnel is not included in the dataset. In 2005, out of a total working population in Portugal of about 5 million, this dataset provides information on about 3 million. 12. The last year for which Quadros de Pessoal are available. 13. Goos et al. (2014) findings for Portugal are, however, calculated using European Community Household Panel (ECHP, Eurostat dataset) and EU-­SILC data which include public sector workers. 14. Interestingly, a number of these relatively low-­skilled services jobs appear to employ a disproportionate number of young people, in contrast to manufacturing-­related jobs over this period.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 488

28/09/2016 08:25



Inequality, labour market and middle-­income groups in Portugal ­489

15. Fernández-­Macias (2012) shows, for example, that ‘less-­knowledge intensive private services’ – which includes activities in retail, hotels and restaurants – were an important source of employment creation in Portugal between 1995 and 2007, affecting the second, third and fourth earnings quintiles. 16. Because workers are allocated to quintiles based on their occupation and some occupations are relatively large, each quintile does not necessarily represent exactly 20 per cent of all employees. 17. The fourth group corresponds to a residual category of all other jobs not explicitly considered by Goos et al. (2014). 18. We present the countries of the EU15 but the same applies for EU28. 19. DGAEP (2015) and Instituto Nacional de Estatística (Portuguese National Statistics Office) (INE), available at https://www.ine.pt/. 20. See González and Figueiredo (2015) for a detailed analysis of the recent evolution of education in Portugal. 21. High emigration, particularly of the highly educated, has also re-­emerged in Portugal, especially since 2011.

REFERENCES Alves, N. (2009), ‘Novos Factos sobre a Pobreza em Portugal’, Banco de Portugal, Economic Bulletin, Spring, 125–54. Alves, N. (2012), ‘Uma Perspetiva Sobre a Distribuição de Rendimento em Portugal e na União Europeia’, Banco de Portugal, Economic Bulletin, Winter, 41–58. Atkinson, A.B. and A. Brandolini (2011), ‘On the identification of middle classes’, ECINEQ – Society for the Study of Economic Inequality, Working Paper 2011-­ 217, Verona. Autor, D.H., L.F. Katz and M.S. Kearney (2008), ‘Trends in US wage inequality: revising the revisionists’, Review of Economics and Statistics, 90 (2), 300–323. David, H. and D. Dorn (2013), ‘The growth of low-­skill service jobs and the polarization of the US labor market’, American Economic Review, 103 (5), 1553–97. Direção-­Geral da Administração e do Emprego Público (Directorate-­General for Administration and Public Employment) (DGAEP) (2015), Síntese estatística do emprego público 4. trimestre 2014’ (‘Public employment summary statistics 4 trimester 2014’), DGAEP, Lisbon. Estanque, E. (2012), A Classe Média: Ascensão e Declínio, Lisbon: Fundação Francisco Manuel dos Santos. González, P. and A. Figueiredo (2015), ‘The European Social Model in a context of crisis and austerity in Portugal’, in D. Vaughan-­Whitehead (ed.), The European Social Model in Crisis: Is Europe Losing its Soul?, Cheltenham, UK and Northampton, MA, USA: Edward Elgar and Geneva: ILO, pp. 386–450. Goos, M., A. Manning and A. Salomons (2014), ‘Explaining job polarization: routine-­biased technological change and offshoring’, American Economic Review, 104 (8), 2509–26. Fernández-­Macías, E. (2012), ‘Patterns of employment expansion in Europe, 1995– 2007’, in E. Fernández-­Macías, J. Hurley and D. Storry (eds), Transformation of the Employment Structure in the EU and USA, 1995–2007, New York: Palgrave Macmillan, pp. 26–51. Figueiredo, H., L.D. Santos, P. Gonzalez and A. Figueiredo (2015), ‘Job

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 489

28/09/2016 08:25

490

Europe’s disappearing middle class?

polarisation and wage inequality in Portugal’, CEF.UP Working Papers, WP 2015-­04, Faculdade de Economia do Porto. Milanovic, B. and Yitzhaki, S. (2002), ‘Decomposing world income distribution: does the world have a middle class?’, Review of Income and Wealth, 48 (2), 155–78. Organisation for Economic Co-­ operation and Development (OECD) (2014), ‘Income inequality update – June 2014’, OECD Directorate on Employment Labour and Social Affairs, Paris. Rodrigues, C.F. and I. Andrade (2013), ‘Growing inequalities and their impacts in Portugal’, Project GINI – Growing Inequalities’ Impacts, Country Report for Portugal, accessed 6 Mat 2016 at http://gini-­research.org/system/uploads/452/ original/Portugal.pdf ?1370092031. US Department Of Commerce (2010), ‘Middle class in America’, report prepared for the Office of the Vice President of the United States Middle Class Task Force, Economics and Statistics Administration, Washington, DC.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 490

28/09/2016 08:25



Inequality, labour market and middle-­income groups in Portugal ­491

APPENDIX 11A.1 As the notion of middle-­income groups used in the project does not correspond directly to separate deciles, we present in Table 11A.1 the correspondence between the two groups in Portugal and in the EU15. This is, in its own right, informative of the relative position of middle-­income groups in both contexts. For that purpose, we compare the cut-­off points of the different deciles with the median income. According to these correspondence tables, the middle-­income groups in Portugal lie between the third and the eighth deciles (deciles 2 and 9 being transition deciles). The more visible difference between Portugal and the EU15 average lies in the Table 11A.1 Ratio of the cut-­off value of each decile and median income (percentage) Portugal First decile Second decile Third decile Fourth decile Fifth decile Sixth decile Seventh decile Eighth decile Ninth decile

1995

2000

2005

2009

2012

40.6 54.6 70.5 86.0 100.0 116.1 136.0 167.5 230.4

44.8 58.5 73.6 86.6 100.0 115.0 134.6 163.0 225.9

47.4 62.1 75.7 87.3 100.0 115.0 134.3 170.3 243.7

48.2 63.1 74.2 86.4 100.0 115.7 133.9 159.9 218.7

44.9 62.6 75.7 88.0 100.0 115.4 135.4 160.3 214.7

EU15 First decile Second decile Third decile Fourth decile Fifth decile Sixth decile Seventh decile Eighth decile Ninth decile

1995

2000

2005

2009

2012

49.6 64.6 76.1 88.1 100.0 113.8 130.1 152.0 189.1

53.0 67.0 78.1 89.3 100.0 112.2 127.7 147.8 182.3

51.4 65.5 77.2 88.4 100.0 112.6 127.6 148.4 185.1

51.1 65.1 76.9 88.3 100.0 113.3 129.4 151.0 189.6

50.8 65.2 76.7 88.0 100.0 113.2 128.9 150.2 188.1

Notes:  Low below 60% of the median; middle lower –> 60%–80% of the median; middle middle –> 80% to 120% of the median; middle upper –> 120% to 200% of the median. Source:  Eurostat, SILC data.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 491

28/09/2016 08:25

492

Europe’s disappearing middle class?

top deciles: in Portugal the ninth and tenth deciles correspond to high income households; in the EU15, however, only the tenth decile combines middle-­high and high income households. This suggests that concepts of middle-­income groups based on thresholds defined relative to the median may be hardly comparable across countries due to very different patterns of inequality across the various European economies.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 492

28/09/2016 08:25



Inequality, labour market and middle-­income groups in Portugal ­493

APPENDIX 11A.2 Table 11A.2 Mobility matrix of Portuguese households by income deciles, 2006–09 2009

2006

>1 dec down 1 dec down No change Decile   1 Decile   2 Decile   3 Decile   4 Decile   5 Decile   6 Decile   7 Decile   8 Decile   9 Decile 10

Source: 

14.3 17.2 28.5 28.8 20.9 30.7 19.3 9.1

16.7 12.7 18.8 15.4 18.7 16.5 20.8 25.0 24.4

46.2 31.6 26.8 23.9 24.4 19.9 32.8 25.0 43.4 66.4

1 dec up

>1 dec up

24.5 15.9 20.7 12.9 13.6 10.2 20.1 13.3 12.3

29.3 35.9 25.5 27.1 18.1 22.4 9.7 10.3

EU-­SILC data.

Table 11A.3 Mobility matrix of Portuguese households by income deciles, 2009–12 2012

2009

>1 dec down 1 dec down No change Decile   1 Decile   2 Decile   3 Decile   4 Decile   5 Decile   6 Decile   7 Decile   8 Decile   9 Decile 10

Source: 

13.9 15.8 18.3 37.8 31.2 18.8 26.8 7.1

20.1 18.2 15.1 15.7 16.5 16.1 21.4 18.9 25.7

49.9 25.2 24.7 25.8 25.8 23.1 25.9 37.5 33.9 67.2

1 dec up

>1 dec up

20.4 21.3 14.9 17.8 15.8 12.7 16.8 15.9 20.5

29.7 33.4 28.3 25.5 24.4 9.9 10.0 6.5

EU-­SILC data.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 493

28/09/2016 08:25

494

Europe’s disappearing middle class?

APPENDIX 11A.3 Table 11A.4 Mobility matrix of EU15 households by income deciles, 2006–09 2009

2006

>1 dec down 1 dec down Decile   1 Decile   2 Decile   3 Decile   4 Decile   5 Decile   6 Decile   7 Decile   8 Decile   9 Decile 10

Source: 

11.7 17.3 24.5 24.7 27.1 28.7 28.5 21.5

20.0 18.4 18.2 17.2 18.5 18.4 18.0 19.4 22.1

No change

1 dec up

>1 dec up

43.5 32.9 24.0 23.7 20.3 22.4 23.3 25.6 31.0 56.4

19.8 17.4 15.9 15.3 16.2 14.8 17.4 19.1 21.0

36.7 29.7 30.1 25.5 21.8 19.5 13.8 8.5

EU-­SILC data.

Table 11A.5 Mobility matrix of EU15 households by income deciles, 2009–12 2012

2009

>1dec down 1dec down No change Decile   1 Decile   2 Decile   3 Decile   4 Decile   5 Decile   6 Decile   7 Decile   8 Decile   9 Decile 10

Source: 

9.4 16.6 18.5 20.1 21.3 21.6 23.8 18.3

22.0 18.4 18.1 15.9 19.8 19.9 20.0 18.6 16.7

48.7 34.6 30.4 27.9 25.3 25.1 28.9 32.8 38.5 64.9

1 dec up

>1dec up

22.3 19.4 19.5 17.0 20.8 16.5 16.8 21.0 19.0

29.0 24.0 22.2 20.4 19.4 18.4 13.1 4.6

EU-­SILC data.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 494

28/09/2016 08:25

12. Knocking on heaven’s door: Changes in the world of work and the middle class in Spain Rafael Muñoz de Bustillo and José-­Ignacio Antón1 1. INTRODUCTION Spain has undergone enormous economic and social changes in the past 50 years. The generation now in their fifties and sixties grew up in a relatively closed and regulated economy, with a highly centralized labour market in which trade unions were illegal and an authoritarian regime in which political parties were outlawed. In less than a decade after the dictator Franco’s death in 1975 the country approved a new democratic constitution, held free elections, opened up the markets, experienced restructuring of the industrial sector and built a modern – if relatively stingy – welfare state. In the next two decades, the once backward country managed to converge – in terms of per capita income – with the core countries of the European Union (EU), turning from a country of emigration to a country of immigration. From the outside, Spain fulfilled all promises. It was the country in which, in the words of Carlos Solchaga, the Social Democratic minister of finance from 1985 to 1993, it was easier and quicker to get rich (Pérez 2014). A closer look leads us to different conclusions, however: unemployment remained high, social spending was among the lowest in the EU and the current account deficit was too high (around 10 per cent in 2007), among other things. The crisis that began in 2008 put an end to dreams of prosperity, with unprecedented losses in employment and gross domestic product (GDP) and rising inequality, poverty and social exclusion. From 2008 to 2013 employment fell by 3.3 million (16 per cent of total employment); the unemployment rate rose to 26 per cent; GDP fell by 8 per cent; and inequality, as measured by the Gini index, rose from 31.9 to 34.7, among the highest in the EU. Starting from 1975–84, this chapter will provide an account and analysis of the major changes produced in the labour market in the past four 495

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 495

28/09/2016 08:25

496

Europe’s disappearing middle class?

decades in Spain, as well as their implications for the different income categories, paying particular attention to the middle class. With that aim, in section 2, we present data on the evolution of the middle class since the early 1970s. In section 3, we analyse the major transformations in the world of work experienced in Spain in the recent past, focusing on those driven by changes in derived demand, changes in labour supply and the policy options followed. To complement the aggregate analysis of section 3, in section 4 we present two case studies, in which two of the above-­mentioned changes – the construction of a public education system and the development of the welfare state – are studied in detail, paying special attention to the implications of such changes for the middle class. Section 5 summarizes the major conclusions of the analysis.

2. SIZE AND EVOLUTION OF THE SPANISH MIDDLE CLASS As we have seen in the introductory chapter to this volume, defining the ‘middle class’ is not easy. In this section we use the indicator traditionally favoured by economists, based in income categories, as defined in Chapter 1. The middle-­income class (MiC) will be defined as those families with (equivalent) income between 60 per cent and 200 per cent of median income. This broad category will be divided into three subcategories: lower middle income (LMic, 60–80 per cent median family equivalent income), core middle income (CMic, 80–120 per cent median family equivalent income) and upper middle income (UMic, 120–200 per cent median family equivalent income). However, in order to obtain a broader view of the development and transformation of the middle class we complement the income approach with a neo-­Weberian perspective, according to which the middle class is defined in terms of individuals’ different opportunities for development and social status, as shown by their occupation in the labour market. In this section we present a long-­run analysis of the evolution of the Spanish middle class covering the past four decades, based on detailed reconstruction of income distribution using five different income and labour market databases all produced by the Spanish Instituto Nacional de Estadística (INE – National Statistics Institute). 2.1  The Income Approach How has the Spanish middle class fared in the past four decades? Table 12.1 and Figure 12.1 present the evolution of the middle class, with

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 496

28/09/2016 08:25

497

VAUGHAN 9781786430595 PRINT (M4051).indd 497

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:25

1980–81

1990–91

1998

1999

2000

69.2 15.2 28.1 25.9 20.0 10.8

69.9 14.5 29.4 26.0 19.5 10.6

73.4 16.7 29.4 27.3 17.2 9.4

77.1 16.7 33.3 27.1 13.8 9.09

77.4 17.4 34.4 25.6 13.8 8.8

76.5 16.2 33.9 26.3 14.5 9.0

77.1 17.7 32.6 26.8 14.0 8.9

2001

MiC: 60–200 LMic: 60–80 CMic: 80–120 UMic: 120–200 LiC: < 60 TiC: > 200

71.0 14.4 27.4 29.2 21.0 8.0

2003 69.1 14.7 26.6 27.9 21.0 9.7

2004 69.5 14.5 26.8 28.3 20.8 9.4

2005 70.4 15.4 26.5 28.5 20.4 9.13

2006 70.3 15.2 27.4 27.6 20.6 9.2

2007

70.1 15.0 26.9 28.3 20.5 9.5

2008

(b) Percentage of population by income brackets based on Statistics on Income and Living Conditions

MiC: 60–200 LMic: 60–80 CMic: 80–120 UMic: 120–200 LiC: < 60 TiC: > 200

(a) Percentage of population by income brackets based on Household Budget Surveys

1973–74

68.5 14.8 26.7 27.0 21.4 10.1

2009

77.5 17.8 34.1 25.5 14.22 8.29

2002

66.9 15.1 24.4 27.5 21.9 11.18

2010

77.7 17.5 34.4 25.8 14.8 7.5

2003

Table 12.1 Size and middle-­class participation in total income by subcategory, Spain, 1973–2004 and 2003–11

67.2 14.7 25.8 26.7 21.5 11.3

2011

78.2 17.6 33.7 26.9 14.5 7.3

2004

498

VAUGHAN 9781786430595 PRINT (M4051).indd 498

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:25

64.5 9.6 23.2 31.7 8.3 27.2

68.3 9.4 25.6 33.2 7.5 24.2

1980–81 71.6 10.3 25.9 35.4 7.1 21.3

1990–91 73.4 9.9 28.2 35.3 5.6 21.1

1998 74.4 10.7 30.3 33.4 5.7 20.0

1999 73.9 10.1 29.5 34.3 6.2 19.8

2000 74.1 10.8 28.2 35.2 5.9 19.9

2001

72.4 9.3 24.5 38.6 8.3 19.2

69.2 9.6 23.4 36.2 8.2 22.6

2004 70.6 9.3 23.8 37.5 8.3 21.2

2005 70.8 9.3 23.8 37.7 8.0 21.2

2006 71.9 9.8 24.7 37.4 8.1 20.0

2007

71.3 9.4 23.9 38.0 7.3 21.3

2008

68.1 9.0 23.2 35.9 7.3 24.5

2009

76.0 11.5 30.4 34.1 6.1 17.9

2002

66.7 8.9 21.9 36.0 7.3 26.0

2010

76.8 11.4 30.8 34.7 6.4 16.8

2003

66.1 9.0 22.1 35.0 6.8 27.1

2011

77.5 11.1 30.1 36.4 6.3 16.2

2004

Source:  Authors’ analysis from Household Budget Surveys 1973–74, 1980–81, 1990–91, Continuous Household Budget Survey 1998–2004 and Statistics of Income and Living Conditions, 2004–11 microdata.

MiC: 60–200 LMic: 60–80 CMic: 80–120 UMic: 120–200 LiC: < 60 TiC: > 200

2003

(d) Percentage of population by income brackets based on Statistics on Income and Living Conditions

MiC: 60–200 LMic: 60–80 CMic: 80–120 UMic: 120–200 LiC: < 60 TiC: > 200

1973–74

(c) Percentage of income by income brackets based on Household Budget Surveys

Table 12.1 (continued)



Changes in the world of work and the middle class in Spain ­499 100 % of total population

80 60 40 20 0

40 20

60-200 80-120

60-80 120-200

2011

2010

2009

2008

2007

2006

2005

60-200 80-120

2003

2011

2010

2009

2008

2007

2006

2005

2004

20

60–80 120–200

2001

40

1999

60

100 90 80 70 60 50 40 30 20 10 0

1990-91

% of total income

80

2003

2004

2003

2004

2002

2001

2000

1999

1998

1990–91

1980–81

1973–74

2003

60–200 80–120

60–80 120–200

100

% of income

60

0

60–200 80–120

0

80

1973-74

% of total population

100

60-80 120-200

Source:  Authors’ analysis from Household Budget Surveys 1973–74, 1980–1981, 1990– 1991, Continuous Household Budget Survey 1998–2004 and Statistics of Income and Living Conditions, 2004–2011 microdata.

Figure 12.1 Size and middle-class participation in total income by subcategory, Spain, 1973–2004 and 2004–2011 the subcategories defined above, from 1973 to 2004 (based on different family budget surveys) and from 2003 to 2011 (based on SILC). The different nature of the surveys makes it advisable to analyse the whole period in two different sub-­periods corresponding to the two sources used (family surveys and SILC). Taking the period as a whole and the broader definition of middle class (60–200 per cent of median income), we can observe a clear increase in the size of the middle-­income class by 13 per cent, from 69 per cent in 1973 to 78 per cent in 2004, followed by a period of stability during the years

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 499

28/09/2016 08:25

500

Europe’s disappearing middle class?

previous to the crisis and a reduction of 3 percentage points due to the crisis. By subcategories, from 1973 to 2004 the core middle class increased by 20 per cent, with one-­third of the population in that category, followed by the lower middle-­income class, with a total increase of 15 per cent. In contrast, the size of the upper-­income middle class remained roughly constant, with an overall increase of 4 per cent, which conceals much more instability than the other subcategories. In contrast, the crisis took a higher toll on the core middle-­class income class (more than half of the decrease in the middle-­income class comes from this segment), while the upper-­ income class improved its size by 2 percentage points. Summing up, in the three decades before the crisis, the Spanish middle-­ income class showed healthy development and grew in size. This was fed both from the top (by a reduction of the top-­income class from 10.8 to 7.3 per cent) and from the bottom (by a reduction of the low-­income class from 20 to 14.5 per cent). In this respect, it is interesting to note how the larger increases of the middle-­income class took place in the last two decades of the twentieth century, with almost no changes thereafter up to the crisis. In contrast, the crisis has shrunk, if not dramatically, the middle-­ income class by 3 percentage points: two-­thirds of that reduction is owing to the changing situation of people belonging to the upper middle-­income class due to a reduction of average income (and the corresponding lowering of the upper middle-­income class threshold), while the other third is the result of people sliding down from the middle-­income class into the lower-­income class. As expected, the increase in the size of the middle-­income class during the first sub-­period produced an increase in the participation of the middle class in total income from 64.5 per cent in 1973 to 77.5 per cent in 2004. As shown in Figure 12.1, the increase in the participation in total income was higher for the upper and core middle classes, while the participation of the lower middle class remained roughly constant throughout the period. As in terms of size, the participation of the middle-­income class of total income suffered a major reduction of almost 6 percentage points in the crisis. In this case, central and upper middle-­income groups were hit harder (85 per cent of the reduction). In contrast, the upper-­income group increased its share in total income by 7 per cent We can conclude that the last three decades of the twentieth century witnessed an increase in the size of the middle class in terms of both population and income, which came to a halt with the economic and financial crisis of 2008. With the crisis, it shrank both in terms of size (by 3 ­percentage points) and share of total income (almost 6 percentage points). Another complementary way to study the evolution of the middle class through time is to explore whether the income distribution during the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 500

28/09/2016 08:25



Changes in the world of work and the middle class in Spain ­501

Foster–Wolfson bipolarization index

0.500

HBS

EU-SILC

0.400

0.200

0.298

0.266

0.300 0.277

0.228

0.236

0.297

0.223

0.100 2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1990–91

1980–81

1973–74

0.000

Source:  Authors’ analysis from Household Budget Surveys 1973–74, 1980–81 and 1990– 91, Continuous Household Budget Survey 1998–2004 and Statistics of Income and Living Conditions, 2004–11 microdata.

Figure 12.2  Foster-Wolfson bipolarization index, Spain, 1973–2011

period has shown signs of polarization (hollowing out of the middle). Figure 12.2 reproduces the evolution of income polarization in Spain during 1973–2011 – with a break in 2004 resulting from a change in the data source from the Family Budget Survey to the Statistics on Income and Living Conditions – as measured by the Foster–Wolfson (Foster and Wolfson 2010) polarization index.2 The results confirm the pattern analysed above: (1) the reduction first and the consolidation later of the polarization index by nearly 20 per cent during the last quarter of the twentieth century, (2) an increase in the polarization index of 7 per cent due to the crisis. This can be interpreted in terms of the growth and consolidation of the middle class during the first period, with the corresponding reduction in the polarization index and decrease of the middle class owing to the crisis. Unfortunately the crisis is still too recent to see whether we are witnessing a permanent structural change or a temporary product of the economic conjuncture. 2.2 Characteristics In this section we answer two questions. The first, to what extent do middle-­ class families today have different demographic characteristics from the average family? Second, have the characteristics of middle-­class families changed during the period covered in the analysis? Regarding the first question (see Appendix 12A.1), and focusing on 2011,

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 501

28/09/2016 08:25

502

Europe’s disappearing middle class?

the following differences are worth mentioning: (1) middle-­class families have more senior citizens both as members and as head of household, (2) they have a lower proportion of unemployed or inactive as head of household, (3) there is a slightly higher proportion of female-­headed households, (4) there is a higher proportion of employers and self-­employed among household heads, (5) there is a lower proportion of mono-­parental households and two adults with children, (6) there is a higher number of working adults (although the differences are not very marked, the employment density of the household seems to be one of the differentiating elements of the MiC) (7) there is a lower share of economically dependent members and (8) there is no major difference in terms of access to consumer goods, with the exception of cars, holidays and heating. In relation to the second question, and focusing only on those items with a differential intensity of change during the observed period compared with the average household, we highlight the following long-­run changes: (1) an increase in the share of people living in female-­headed households (from 7 per cent in 1980 to 39 per cent in 2011), (2) an increase in the proportion of household heads with higher education (from 5 per cent in 1980 to 25 per cent in 2011), (3) an increase in the proportion of one-­person and mono-­ parental households, and (4) an increase in the average share of working age women employed, from 20 per cent in 1980 to 56 per cent in 2011. 2.3  Occupational Approach Following Goldthorpe’s (2012) criticism of the definition of social classes exclusively by means of income categories, in this section we explore the evolution of the middle class using a different perspective, centred on type of job. According to Goldthorpe, people in different social classes are different not only in terms of income, but also in terms of income security, stability of income in the short run and income prospects in the long run. Such differences are related to the type of job performed (difficulty/human capital requirements and autonomy/control over the working process). At the same time, class is related to status, and differences of status are also related to the type of job performed. Figure 12.3 reconstructs the evolution of social classes as defined from a labour-­market/occupational perspective. Occupations, adapted from Martínez García (2014), are presented in four categories: upper middle class (UMC, managers, firm owners and profesemployed excluding agriculture, sionals), core middle class (CMC, self-­ administrative and commercial, skilled workers), lower middle class/lower class (LMC/LC, unskilled service and industrial workers) and agriculture. In general we can observe an increase in UMC jobs, explained by the increase in professionals, which at the end of the period make up almost

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 502

28/09/2016 08:25



Changes in the world of work and the middle class in Spain ­503

100 80

9.1

7.4

19.1

18.9

21.9

22.7

23.4

49.9

48.4

46.1

44.8

41.5

16.5

21.9

25.3

26.9

28.3

31.1

1987

1994

1997

2004

2007

2012

14.2

18.3

16.2

15

16.3

15.7

54.6

52.7

53.6

12.1

14.8

1977

1984

60 40 20 0

Agriculture

LMC-LC

CMC

UMC

Source:  Authors’ analysis from Martínez García (2014: 6).

Figure 12.3 Distribution of jobs from a social class perspective, Spain, 1977–2012 (percentage) two-­thirds of the category (from less than half in 1977), and a substitution of CMC jobs by lower middle-­class/lower-­class jobs, explained by the reduction in self-­employment, but mainly by the reduction of the weight of skilled manufacturing jobs (from 31 per cent to 15 per cent). 2.4  New Risks: Temporality and Unemployment The rise of temporary employment is one of the major changes in the Spanish labour market in recent decades. The ubiquity of temporary contracts is such that it has altered one of the traditional attributes of the middle class: having a permanent job with an open-­ended contract and covered by good employment protection legislation. Together with the persistence of temporary employment, the rise of unemployment (from 1.8 million in 2007 to 6 million in 2013) and its duration (in 2008 less than 10 per cent of unemployed workers had been unemployed for two or more years, while in 2014 the figure was 42 per cent) has affected another of the certainties of middle class: the low risk of unemployment. Although the risk of unemployment is lower the higher the position in the social and wage ladder, the brutality of the crisis has taken the drama of unemployment also into middle-­class households, until now largely protected from it. One example of such change is the reduction, in the latest phase of the crisis, of public sector employment. Figure 12.4 clearly reflects how the fear of unemployment is higher among lower-­and lower middle-­class jobs, but is now also affecting middle class workers.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 503

28/09/2016 08:25

504 35.0

Europe’s disappearing middle class? 31.9 28.4

30.0 25.0 20.0

16.5

15.0

13.9

10.3

10.0 5.0 0.0

Unskilled workers

Skilled workers

New MC

HC-UMC

Old MC

Note:  New middle class (MC) = routine workers in administrative tasks and services + white-collar workers; Old middle class (MC) = small business and blue-collar selfemployed + part of agriculture; High class-upper middle class (HC-UMC) = traditional professional, managers in big firms, technical workers. Source:  Authors’ analysis from CIS (2014b: 42b).

Figure 12.4  Percentage of workers who think it very or quite probable they will lose their jobs in the next 12 months, Spain, December 2014 2.5  The Crisis and the Middle Class The severity of the crisis in terms of GDP and employment destruction across all segments of the labour market (see section 3.1) has been such that no income stratum, including the middle class, has been spared. In any case, the magnitude of the impact in terms of income lost has been highly unequal. In order to see the different impacts of boom and bust on the different income categories we have estimated the average growth rate of income in four periods constructed around the two different data sets used above. The first (1973–2004) is the period of growth and consolidation of the middle class, the second (2003–07) the period prior to the crisis, the third corresponds to the crisis (2007–11) and the last to the whole period covered by SILC (2003–11). As we are working with moving categories in terms of size, the data are presented in two different formats. Figure 12.5 reproduces the compound annual growth rate of total income by income category: MiC, above and below. Figure 12.6 presents the compound annual growth rate of average income of the above-­mentioned three categories and periods. The message presented by the former figure is clear and confirms the development and consolidation of the middle class in the last

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 504

28/09/2016 08:25



Changes in the world of work and the middle class in Spain ­505

8

Below

Above

Middle class

National 301.8

2

–434.8

–351.2

4

511.8

768.9

450.1

416.6

120.3

–2.1

532.8

25.3

234.2

61.9

–2

291.1

0 180.8

Growth (%)

6

–256.6

10

–4 –6

1973–2004

2003–11

2003–07

2007–11

Source:  Authors’ analysis from Household Budget Surveys 1973–74, 1980–81, 1990–91, Continuous Household Budget Survey 1998–2004 and Statistics of Income and Living Conditions, 2004–11 microdata.

Figure 12.5 Percentage compound annual growth rate of total income by income category, Spain, 1973–2011

–288.0

–259.4

–488.0 174.2

70.5

140.5

137.6

–145.0

–130.5

–247.1 161.1

121.1

176.2

0 216.2

Growth (%)

1

–1

–180.6

2

–358.0

3

–2 –3 –4

Below

Above

–5

Middle class

National

–6

1973–2004

2003–11

2003–07

2007–11

Source:  Authors’ analysis from Household Budget Surveys 1973–74, 1980–81, 1990–91, Continuous Household Budget Survey 1998–2004 and Statistics of Income and Living Conditions, 2004–11 microdata.

Figure 12.6 Percentage compound annual growth rate of average income by income category, Spain, 1973–2011

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 505

28/09/2016 08:25

506

Europe’s disappearing middle class?

three decades of the twentieth century, with the income allocated to the MiC growing at a higher rate than in the other two categories. In contrast, the first decade of the twenty-­first century shows a clear prevalence of the rate of growth of the top-­income class (TiC). Finally, the figure shows how during the crises the income of the TiC increases, while the opposite is true for the MiC and the lower-­income class (LiC), with a higher decrease in the former case (−4.3 per cent, compared with −3.5 per cent and +3 per cent). In terms of the evolution of the average income of our three income categories (Figure 12.6), we can see how in the period 1973–2004 MiC average income grows at a lower annual rate than the average income of LiC, but the situation changes with the coming of the crisis, with income decreasing in an inverse relationship with income: highest for low income: −4.9 per cent, −3.6 for MiC and only −2.6 for TiC. These results mean that, as a group, the middle class – and especially the upper middle class – has fared better than the lower middle and lower classes. Having said that, the middle class has also been affected by the crisis, with a reduction of income of around 4 per cent in the first part of the crisis. However, while Figure 12.6 can speak on behalf of the different income categories, it does not convey information about the specific members of the middle (or other classes) who might have slid down the income scale as a result of unemployment or some other life contingency. In order to shed some light on the short-­term transitions between income categories, Figure 12.7 presents the percentages of people belonging to three different income categories (low: deciles 1–2; middle: deciles 3–7; and high: deciles 8–10), according to whether they stay in the same decile, move up or move down in the period 2004–10. Focusing on middle income, this group shows a higher degree of mobility than the upper-­and lower-­income groups.3 Within the group, upward mobility was slightly dominant before the crisis, although the direction changed quickly with the recession: at the end of the period analysed, 44 per cent of individuals belonging to this category slid down to lower-­ income deciles, compared with 32 per cent who moved up. In contrast, the high-­income group is characterized by higher stability, with between 40 per cent and 50 per cent of the category remaining in their deciles throughout the period. The most noticeable change in this regard is the increase by 10 percentage points of the people experiencing downward mobility, as well as the decrease by 40 per cent of the people improving their position in terms of income deciles. Regarding the central deciles (3–7), the main feature to be highlighted is the lower stability in this bracket, with more people entering but also more people leaving this category each year. This issue is difficult to qualify as it means higher mobility, but also higher vulnerability of middle incomes during the crisis.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 506

28/09/2016 08:25



Changes in the world of work and the middle class in Spain ­507 Low income: deciles 1–2

60 50 40 30 20 10 0

2004/05

2005/06

2006/07 Up

2007/08 No change

2008/09

2009/10

Down

Middle income: deciles 3–7 60 50 40 30 20 10 0 2004/05

2005/06

2006/07 Up

2007/08 No change

2008/09

2009/10

Down

High income: deciles 9–10 60 50 40 30 20 10 0 2004/05

2005/06

2006/07 Up

2007/08 No change

2008/09

2009/10

Down

Source:  Adapted from Ayala et al. (2014: 96).

Figure 12.7 Transition between deciles by income groups, Spain, 2004–10 (percentage)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 507

28/09/2016 08:25

508

Europe’s disappearing middle class?

This result neatly summarizes the information gathered in this section: the development and consolidation of the middle class in the four or five decades before the crisis, in a time of economic growth, political transformation and development of the welfare state, and the uncertainty generated by a long and destructive crisis which has cast a shadow on the sustainability of the welfare state, generated massive unemployment and accelerated the deterioration of job quality.4

3. LONG-­TERM TRENDS IN THE WORLD OF WORK: ELEMENTS AND IMPLICATIONS 3.1  Transformation Driven by Changes in Demand The process of economic growth in the long run can be understood, in the famous words of the Austrian economist Joseph A. Schumpeter (1942), as a ‘perennial gale of creative destruction’, by which industries thrive and disappear with the passing of time. As different sectors use different types of labour – some are capital intensive, some low skilled, some high skilled, for example – this process of structural change can lead to major changes in the structure of employment. In order to study the evolution of the employment structure in Spain since the economic and political transition of the late 1970s, we will use the so-­called ‘jobs approach’, proposed by Joseph Stiglitz while he was Chair of the Council of Economic Advisers of the President of the United States in 1996, consolidated by the sociologists Erik O. Wright and Rachel Dwyer (2003) and further developed from a comparative perspective in a series of papers produced under the auspices of Eurofound (Fernández-­ Macías and Hurley 2008; Fernández-­ Macías 2012; Fernández-­ Macías et al. 2012). Briefly, the jobs approach applies the following methodology: (1) a ‘job’ is defined as the combination of sector (two-­digit Nomenclature statistique des Activités économiques dans la Communauté Européenne, NACE) and occupation (two-­digit International Standard Classification of Occupations, ISCO). There are, potentially 1680 jobs so defined (28 ISCO × 60 NACE), although the actual number obtained is much lower, as some of the combinations are not present in real life or host very small numbers of workers; (2) the different jobs are ranked in terms of ‘quality’ using the average wage; (3) once ranked, the jobs are ordered by quintiles; and (4) the comparison of the employment level in each quintile between any given years allows us to obtain information about the dynamics of job quality, creation and destruction in terms of polarization, upgrading and so on. Figure 12.8 shows the results obtained using this methodology in Spain

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 508

28/09/2016 08:25



Changes in the world of work and the middle class in Spain ­509 1977–85

1985–91

0

Thousands of workers

Thousands of workers

100 –100 –200 –300 –400 –500 –600 –700

I

II

II

IV

V

1000 800 600 400 200 0 –200 –400 –600 –800

I

II

II

Thousands of workers

Thousands of workers

2500.0

–50 –100 –150 –200 I

II

IV

V

–200 –300 –400 –500 –600 II

II

1500.0 1000.0 500.0 0.0

I

II

IV

V

II

IV

V

2011–13

0

–100

I

2000.0

–500.0

Thousands of workers

Thousands of workers

II 2008–10

0

–700

V

1994–2008

1992–93 0

–250

IV

–100 –200 –300 –400 –500 –600

I

II

II

IV

V

Source:  Muñoz de Bustillo and Antón (2014).

Figure 12.8  Changes in the employment structure, Spain, 1977–2013 for the period 1977–2013. Although the existence of several breaks in the statistics used in the analysis preclude presentation of a single figure for the whole period, the results clearly show that the nature of employment changed during the period, and follows two different patterns, an overall upgrading picture during the periods of employment creation and a certain polarization in recessions. The results are fairly interesting as they show how employment change

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 509

28/09/2016 08:25

510

Europe’s disappearing middle class?

does not follow a single pattern throughout the period: during the years of growth, employment grows across all categories, but especially in the middle (3–4) and upper quintiles. These quintiles are precisely those that concentrate on what is considered from a general point of view as typical ‘middle-­class’ jobs: public administration, technical and liberal professions, middle-­and high-­skilled jobs and so on. Thus, in Spain during the boom years we do not find trace of a ‘diminishing middle’, at least in terms of the number of middle-­class jobs. Nevertheless, this picture suffers a dramatic change when we look at the crisis years. Looking at the evolution of employment during the whole period, it can be concluded that ‘middle-­class jobs’ are more abundant in the Spanish employment landscape of the 2010s than three decades ago. Such a conclusion is consistent with the process of economic growth and structural transformation undergone by the Spanish economy in the past half century and, as we will see in one of the case studies in section 4, is also intimately related to the (late) development of the Spanish welfare state. Nevertheless, the intense destruction of employment in the middle of the jobs distribution experienced during the financial crisis could jeopardize this state of affairs. 3.2  Transformation Driven by Changes in the Labour Force From our perspective, the following changes are highly relevant when studying the changes in the labour market in the long run, while at the same time having important implications for the development and future of the Spanish middle class. These include: the normalization of female labour force participation rate experienced in the last two decades of the twentieth century; the huge increase in immigration in the decade and a half previous to the crisis; and finally, the educational transformation of the workforce experienced since the universalization of basic education in the 1970s. 3.2.1 The gender ‘normalization’ of the labour force of the 1980s and 1990s When the democratic constitution of 1978 was approved by referendum, women were greatly underrepresented in the labour force. The enormous power of the Catholic Church and the role of women as mothers blessed by the dictatorship that governed Spain from 1939 until 1977 (Manrique 2007; Morcillo 2012) led to the establishment of a dominant model of men as breadwinners for the main part of the twentieth century. As a result, in the late 1970s the female labour force participation rate was 28 per cent, among the lowest in Europe, compared with 75 per cent among men. The economic and social changes developed in the last quarter of the twentieth century led to an increase in women’s labour force participation rate, which rapidly converged with the male labour force participation rate (Figure 12.9).

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 510

28/09/2016 08:25



Changes in the world of work and the middle class in Spain ­511

90 80

70.4

70

65.7 59.5 53.7

60 50 49.1

40 30

Total

29.1

20

Male

10

Female 2014

2012

2010

2008

2006

2004

2002

2000

1998

1996

1994

1992

1990

1988

1986

1984

1982

1980

1978

1976

0

Source:  Authors’ analysis from Encuesta de Población Activa (Labour Force Survey – EPA).

Figure 12.9 Labour force participation rate by gender (16 and over), Spain, 1976–2014 According to the literature (Novales 1989; Bover and Arellano 1995), there are two types of factors behind the large increase in the female labour force participation rate in Spain: a cyclical factor, namely, the reactivation of the economy in the 1980s; and structural factors, mainly the decrease in birth rates and the large increase in university education. The increase in the overall labour force participation rate has, among other things, two main implications. The first is related to the appearance of new conflicting time demands on the individual, especially women. By breaking with the traditional division of labour within the family, where the woman attended to the domestic needs of the household (care of children and dependants, housework) and the man brought home the money, the rise of dual earner households creates the problem of ‘work–life balance’: the lack of time to perform their duties as workers and parents, at the personal or socially required levels. In second place, depending on the process of family formation, for example, whether there is assortative mating (homogamy) or mating across social classes, income categories and educational levels, the growth in dual earner households can have different implications in terms of income inequality and the income gap between low-­and middle-­income families and within the different middle-­income categories themselves. From a comparative perspective (OECD 2011), Spain has a level of

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 511

28/09/2016 08:25

512

Europe’s disappearing middle class?

% of pairings where both persons have the same educational attainment

100

80

78.6

74.6 64.6 57.6

60

54.4

40

20

0

1920–29

1930–39

1940–49

1950–59

1960–69

Source:  Esteve and Cortina (2006).

Figure 12.10 Educational homogamy by female birth cohort, Spain, 2006 (percentage of pairings) assortative mating (the coupling of people with different educational levels is not widespread) close to the Organisation for Economic Co-­ operation and Development (OECD) average, with nearly two-­thirds of couples having the same level of education.5 In any case, as we can see in Figure  12.10, there has been a significant reduction in educational homogamy over time. This reduction in the rate of assortative mating can be interpreted in terms of a trend towards greater equality and increasing size of the middle class over time. 3.2.2  The immigration ‘upsurge’ at the turn of the century Spain has historically been a country of emigrants; the long decay of the Spanish economy at the end of the nineteenth century led to a large outflow of the population. The economic development of the 1960s and the crisis of the 1970s put an end to the Spanish diaspora. By the early 1990s Spain was a relatively self-­contained country in terms of population dynamics, with only marginal emigration and immigration flows. The recovery of the economy and the huge increase in employment experienced until the start of the financial crisis in 2008 led to a rapid increase in immigration rates, mostly from Latin America, which changed the Spanish population landscape in little more than a decade.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 512

28/09/2016 08:25



Changes in the world of work and the middle class in Spain ­513

The increase in immigration rates had a profound impact in the world of work with the following implications: The large inflow of female immigrants6 increased the supply of  workers in the domestic help sector, in which around one-­third of female immigrants are employed. In 2009, 750 000 workers were employed in the domestic help sector, more than half of them immigrants. The impact of such an increase in supply is especially visible in the sector of home care for elderly people, where, according to a survey by IMSERSO (2005), this group of workers represented 40 per cent of total employees hired specifically to take care of older people. We might speculate that without such a profound increase in supply the cost of such services would be much higher that it is now. This increase in the supply of domestic workers has allowed families with two wage earners, especially middle-­class families (Parella 2003), to address the work/life conflict by subcontracting part of the family duties. According to the analysis of Farré et al. (2011), by increasing the local availability of household services and reducing their prices, immigration has had a positive impact on the labour supply of skilled native women by reducing time off work after childbirth and facilitating the care of elderly dependants (up to one-­third of the recent increase in the employment rate of college educated women is explained by the increase in availability of household services). ●● Concerning wages, the increase in immigration poses two interesting issues. The first is the impact of the labour supply shock on natives’ wages. The second is the behaviour of immigrant’s wages vis-­à-­vis nationals. Regarding the first question, contrary to the expected result, the few studies available on the impact of immigration on the wages of natives in Spain conclude that there has been no major negative impact (see Carrasco et al. 2008; Gonzalez and Ortega 2011).  Regarding the second question, all analyses performed on the differences between local and immigrant wages conclude that there is a serious wage gap between the two groups, even after controlling for differences in characteristics. In this respect, Antón et al. (2010), after estimating the raw and counterfactual (considering the differences in characteristics) wage gap between natives and immigrants along the income distribution, conclude that there is a sizeable gap that grows along the income distribution: relatively low (around 5 per cent) for the tenth percentile, reaching 14 per cent for the ninetieth percentile. ●● The lower wages of immigrants, and their higher unemployment rates (at its highest, first quarter of 2013, the unemployment rate of national ●●

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 513

28/09/2016 08:25

514

●●

Europe’s disappearing middle class?

workers was 25 per cent compared with 42 per cent among extra EU15 workers, 38 per cent and 26 per cent respectively among women), imply that immigration will show an increase in relative poverty rates and, therefore, in income inequality.7 For example (Muñoz de Bustillo and Antón, 2011), before the crisis, the poverty rate among immigrants reached 31 per cent, compared with 18.2 per cent among Spaniards. According to the authors, such differences are not driven by basic observable household characteristics and might well be explained by the very recent nature of Spanish immigration flows. These differences increased with the crisis, in 2013 reaching 18.4 per cent for nationals and a staggering 55.4 per cent among immigrants from non-­EU countries, while remaining roughly constant for EU immigrants. Unsurprisingly, according to EU-­SILC data, the foreign-­born population (over 16 years of age) is underrepresented among the middle-­class and overrepresented among the lowest income strata. For instance, in 2011, less than 60 per cent of the foreign-­born population belonged to the middle class and more than 35 per cent lived in a household with an income below 60 per cent of the median; these figures were roughly 70 per cent and 19 per cent, respectively, among natives. The sudden increase in the number of immigrants has had many implications in other areas than the labour market. From all of them, we would like to focus on the impact on the education system. In the school year 2011–12 there were nearly 800 000 foreign students in the pre-­tertiary Spanish education system. Most of them attended public (82.4 per cent) or privately run, public financed schools (13.7 per cent), while only a minority attended purely private schools. This increase in foreign students in Spanish classrooms has led to a process of switching of schools by middle-­class national students towards privately run public financed schools with lower immigration intensity, as well as to purely private schools. This behaviour is driven by the popular belief associating high concentration of immigrant students with higher levels of conflict (Ortiz 2012) and lower results (Garrido and Cebolla 2010). As result, middle-­class families have increased their expenditure on education and more educated families have switched to private schools in response to immigration (Farré et al. 2015). This dynamic may have important implications for the future of public education if the flight of middle-­class children from public schools leads to a reduction of the current strong popular backing of the role of the public sector in ensuring good-­ quality education. Finally, it is also worth mentioning that Jiménez and Jorgensen (2009) document a similar process with regard to health care, with some descriptive evidence suggesting that the higher

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 514

28/09/2016 08:25



Changes in the world of work and the middle class in Spain ­515

the presence of migrants, the higher the demand for private health care providers. These results should be interpreted with even more caution, as long as the evidence is more descriptive than causal and the effect is especially relevant only among public employees with the right to choose between publicly funded private insurance and access to health services through the National Health Service (NHS). 3.2.3  The educational transformation of the workforce: a dual landscape The last major change in the supply side of the Spanish labour market is related to the radical transformation of the human capital of the Spanish labour force. The education system was one of the casualties of the authoritarian regime that ruled from 1939 to 1977. For decades, education – which was considered a system for transmitting the ideology of the dictatorship – was largely in the hands of the Catholic Church, with the state playing a subsidiary role. It was not until 1970 with the approval of the General Law on Education that the state took decisive steps towards the creation of a modern education system and the universalization of education to all children from 6 to 14 years old. These trends can be observed by looking at the different levels of education of older and younger workers. Figure 12.11 presents the level of education in 2005 of workers 15 to 29 years old compared with that of workers from 55 to 59 years old. That way we can compare the education of today’s young workers with the education 50 45 40

From 25 to 29 From 55 to 59

44.2

41.2

35 30

26.3

25

24.9 13.7

15 10 5 0

22.1

18.8

20

6.0 0.2 0.8 Illiterate

Primary

Lower secondary

Upper secondary

Tertiary (vocational and university)

Source:  Authors’ analysis from Labour Force Survey data.

Figure 12.11  Educational attainment by age, Spain, 2005 (percentage)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 515

28/09/2016 08:25

516

Europe’s disappearing middle class?

of older workers, precisely the cohort that was 25–29 years of age years in 1975. The figure speaks for itself with regard to the gargantuan change in education undergone by the Spanish workforce in the past three decades, with a doubling of the percentage of workers with tertiary education. Nevertheless, there is also a silver lining to the success story represented in the figure. The profile of educational attainment of Spanish workers follows a V-­shape, with high levels of workers at the two extremes: 32 per cent with less than upper secondary education and 41 per cent with tertiary education, and relatively low levels of workers with upper secondary (25 per cent). In the EU context this distribution of educational attainment is quite specific to Spain, as most countries have an inverted V profile. This peculiarity points to one of the major failures of Spanish society in this area, the high levels of early school leavers, which reached 31.7 per cent before the economic crisis in 2008,8 the highest of the EU member states after Portugal. With the crisis and the reduction in employment opportunities for young workers, the early leaving rate came down to 23.6 per cent, still twice the EU (28) average. The above changes had three implications that we would like to highlight. The first is the polarization of the workforce in terms of educational attainment. The high level of workers who have not completed secondary education, largely hired in the construction sector during the years of the housing bubble, poses a serious problem in terms of future reinsertion in a labour market with a surplus of low-­middle and low-­wage jobs. The second is the role played by the process of educational ‘upgrading’ in facilitating educational mobility.9 The third and last implication is related to the existing mismatch between the intensity and timing of the development of the education system with the corresponding increase in workers with higher education and the considerably slower transformation of the Spanish productive system. Such a mismatch has resulted in the development of a gap between the educational requirements of firms and the human capital of (especially) new graduates. This gap, resulting from the relatively low development in Spain of the high-­skilled, knowledge-­intensive sectors, although already present before the crisis, has been made especially evident owing to the crisis and the stagnation of the labour market. As result, for many, mostly, but not only, young workers, having a tertiary education is no longer a passport to a middle-­class job with a middle-­class wage. 3.3  Transformation Driven by Policy Changes 3.3.1  The generalization of temporary employment The Spanish regulation of temporary employment in the Estatuto de los Trabajadores, the labour code approved in 1980 after the restoration of

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 516

28/09/2016 08:25



Changes in the world of work and the middle class in Spain ­517

40

34.6

35

34.5

30 24.6

25 20 15 10 5

1987QII 1988QII 1989QII 1990QII 1991QII 1992QII 1993QII 1994QII 1995QII 1996QII 1997QII 1998QII 1999QII 2000QII 2001QII 2002QII 2003QII 2004QII 2005QII 2006QII 2007QII 2008QII 2009QII 2010QII 2010QII 2011QII 2012QII 2013QII 2014QII

0

Source:  Authors’ analysis from Labour Force Survey data.

Figure 12.12 Temporary employment rate, Spain, 1987–2014 (percentage) democracy, considered open-­ended contracts to be the default contract, and temporary employment as an exception. Consistently, it required that all temporary contracts be justified. A few years later, in a context of high unemployment, it was considered that a major cause of the low employment creation of the Spanish economy, even in times of economic growth, was the rigidity of the labour market, which was partly a product of the narrow room for hiring workers on a temporary basis. In 1984 a major reform of the Labour Code (Law 32/1984) was approved by the first Social Democratic government, liberalizing temporary contracts. Although we do not have data on the level of temporary employment at the time of the reform, it is estimated that around 10 per cent of employees were on temporary contracts at the time. As we can see in Figure 12.12, a few years later, temporary employment reached one-­third of employees, a ­percentage unknown in the EU at the time. The economic crisis produced a sharp drop in the temporary employment rate, much deeper than the numerous legal labour regulation reforms aimed at reducing it. Unfortunately, this drop was not the product of a change in the dynamics of hiring, but the result of the over-­proportionate destruction of temporary jobs in contrast to open-­ended employment: 76  per cent of total employment lost from 2007 (third quarter) to 2013 (first quarter) belonged to this category (2.3 million out of 3.1 total). Summing up, the 1984 reform produced a long-­lasting transformation of the Spanish labour market, leading to the encroachment of a culture

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 517

28/09/2016 08:25

518

Europe’s disappearing middle class?

100 90 80

79.7

2007

85.0

2014 67.4

70

58.6

60 50

43.5 43.9

40

30.6

30

25.2 23.1

20

31.6 18.6

15.9 12.7 12.9

10 0

16–19

20–24

25–29

30–39

40–49

50–59

24.0

8.1

60–69

Total

Source:  Authors’ analysis from Labour Force Survey data.

Figure 12.13 Temporary employment rate by age group, Spain, 2007 and 2014 of temporary hiring by Spanish firms that has proven very difficult to change. To what extent does the increase in temporary employment affect the middle classes? The question is tricky because due to the impact that temporary employment has on income (more unemployment spells and lower wages), those with temporary employment will face a higher risk of slipping down the social ladder. One indirect way of addressing this issue is by looking at the extension of temporary employment by age (Figure 12.13). According to the data presented in the figure, temporary employment in Spain is generalized among young workers regardless of the social class of their parents. Thus, probably the lack of stable employment for their offspring will be a major concern for much of the Spanish middle class. Another indirect way of testing the extent to which employment insecurity is affecting the middle class is by looking at the rate of temporary employment by occupation. Although, as expected, the temporary employment rate is much higher among unskilled workers, white-­collar and other skilled occupations traditionally associated with middle-­class status are by no means safe from the temporary employment malaise. In this respect, in 2014, even after the strong destruction of temporary employment experienced during the crisis, 19 per cent of professionals and 17 per cent of technicians and associated professionals were on temporary contracts. Finally, if we look at the distribution of temporary employment

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 518

28/09/2016 08:25



Changes in the world of work and the middle class in Spain ­519

by wage decile, according to data from the National Statistics Institute, there is a clear decrease in the temporary employment rate from almost 50 per cent in the first earnings decile to 6 per cent in the tenth decile. Nevertheless, again, that does not mean that the middle class is protected from temporary employment. In fact, around 60 per cent of employees with fixed-­term contracts are in wage deciles 3 to 8, which we can associate with middle-­class wages. 3.3.2  Growth of involuntary part-­time working Together with the high rate of temporary employment, another ­statistical  anomaly of the Spanish economy was the very low rates of part-­time employment. The late modernization of the Spanish labour market from a gender perspective, the extended use of split shifts – in 2006 split shifts were the norm in 76 per cent of Spanish firms (Muñoz de Bustillo et al. 2008) – and a regulation that penalized part-­time working explains that at the beginning of the 1990s part-­time employment was a meagre 4.5 per cent of total employment. Since then, there has been a diligent drive from the regulatory authorities to make part-­time work more attractive to firms, resulting in an unprecedented increase in part-­ time employment that reached 16 per cent by 2014 (26 per cent among women). The most relevant item in relation to the structural policy-­driven change in the use of part-­time employment is the abnormally high proportion of involuntary part-­time employment in Spain. Although the involuntary part-­time employment rate has been always high in Spain, in the past decade the proportion increased to almost two-­thirds (Figure 12.14). The implications of involuntary part-­ time employment are ­straight­­forward: lower hours of work than desired and lower income and promotion opportunities. 3.3.3  Social dialogue and collective bargaining under fire At the end of the 1970s, during the process of democratic transformation, in a context of high social unrest, the Spanish Constitution recognized the major role of trade unions (and employers’ organizations) in the labour market and the development of labour market institutions. A few years later the Constitutional Court, the highest judicial institution in the country, stressed that trade unions are ‘a basic part of the political system’ and ‘essential economic and social instruments for the defence and promotion of the interests of the workers’ and ‘social formations with constitutional relevance’ (Mercader 2002). The socio-­economic history of modern democratic Spain is marked by a succession of bilateral or trilateral social agreements, with the exception of the period 1985–94, when no major

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 519

28/09/2016 08:25

520

Europe’s disappearing middle class?

70.0

63.4

60.0 50.0 40.0 30.0 20.0 10.0

1987QII 1988QII 1989QII 1990QII 1991QII 1992QII 1993QII 1994QII 1995QII 1996QII 1997QII 1998QII 1999QII 2000QII 2001QII 2002QII 2003QII 2004QII 2005QII 2006QII 2007QII 2008QII 2009QII 2010QII 2011QII 2012QII 2013QII 2014QII

0.0

Note:  2005 break in the series. Source:  Authors’ analysis from Labour Force Survey data.

Figure 12.14  Involuntary part-time employment rate, Spain, 1987–2014 agreements were signed. Social agreements contributed, to a large extent, to legitimize the system in a context of high risk of crisis of governance. The resulting industrial relations system was based on recognition of the special role played by trade unions with higher representation in accordance with the trade union elections every four years (in establishments with over five employees), regardless of the affiliation rate, and on legislation that facilitated their participation in various governing institutions. Collective agreements signed by the social partners are applied automatically to all employees within their scope, leading to a high rate of collective agreement coverage. The resulting structure of collective agreements, with multi-­company agreements predominating, has been criticized due to its excessive fragmentation and atomization and its lack of flexibility in adapting working conditions to the specific situations of firms. In 2010 the social partners started negotiations aimed at modernizing collective bargaining. After various delays and when the partners were close to an agreement (López 2011) new demands by one of the members of the employers’ associations brought the negotiations to a halt and led to the unilateral approval of major changes in collective agreements (see Molina and Miguélez 2013 for an overview of these changes), which was considered by most labour law specialists to be a radical change in the collective agreement system. The reform of collective agreements was adopted in a context of criticism and political and mass-­media attacks on trade unions, ranging from

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 520

28/09/2016 08:25



Changes in the world of work and the middle class in Spain ­521

% of interviewees

35 30

January 2005

25

April 2014

20 15 10 5 0

0 Not at all

1

2

3

4

5

6

7

8

9

10 Total

NS

Source:  Authors’ analysis from CIS (2005, 2014a).

Figure 12.15 Trust in trade unions in Spain (percentage distribution on a 0–10 scale, 2005 and 2014) their lack of representation owing to their relatively low affiliation rates, to their economic dependence on public subsidies, their involvement in corruption scandals and their defence of core labour market workers – the (in)famous insiders – at the expense of temporary workers or outsiders. Although many of these claims are dubious or specific to certain persons and regions, the result has been a significant loss of the prestige and reputation (Figure 12.15) that the trade unions had won over the years, as one of the few moral references in the country.10 In terms of social classes it is interesting that, according to the social classification used by the Centro de Investigaciones Sociológicas (Centre for Sociological Research, CIS), trust in trade unions is slightly higher among upper and new middle classes than among unskilled workers. This higher trust correlates with the also higher affiliation rates of middle-­income classes compared with low-­and top-­income classes.11 The specific ‘local’ problems mentioned above, together with the general problems faced by trade unions around the world (falling affiliation rates, new forms of employment less accessible to traditional trade union action and so on), might lead in the future to a reduction in the trade unions’ ability to continue to be a relevant actor in the shaping of labour market institutions, with the implications already known in terms of increase in wage inequality (Aidt and Tzannatos 2008) and a further decrease in the share of labour in GDP. In this respect, recent comparative research (Jaumotte and Osorio 2015) shows how de-­unionization restrains earnings

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 521

28/09/2016 08:25

522

Europe’s disappearing middle class?

at the middle and bottom of the income distribution, contributing to an increase in inequality. Similar conclusions are drawn by Madland et al. (2011: 1–2) for the United States: ‘Unions make the middle class strong by ensuring workers have a strong voice in both the market and in our democracy . . . Indeed, it is hard to imagine a middle-­class society without a strong union movement’. 3.4  Social Wage: A Low-­profile Welfare State Spain’s late social, economic and political modernization meant that in the 1970s and 1980s, when other countries already had well-­established welfare states, Spain was in the process of building one. Although in the first phase of constructing the welfare state the development of major social policy accrued at a fairly rapid speed, since the mid-­1980s the process of convergence in social protection was halted, leaving Spain at the lower end of EU countries in terms of relative social protection expenditure, at around 20 per cent of GDP. Paradoxically, the crisis, and the corresponding fall in GDP, gave an unprecedented boost to social protection as a percentage of GDP (25.4 per cent in 2012) despite the significant cuts in major social programmes, such as health care, or the old age pension reforms aimed at reducing pension rights (Muñoz de Bustillo and Antón 2015).12 Regardless of the lack of convergence, the construction of the welfare state had important implications in terms of income distribution that affected all income groups, including the middle classes. Figure 12.16 shows the impact of monetary social benefits on income distribution before and after the crisis, in Spain and the EU15, using the Gini index. The figure presents inequality in three different stages: market income, market income plus state pensions and disposable income (market income plus all social transfers). As we can see, the inequality in market income in Spain before the crisis was lower than in the EU15, and the larger income inequality in terms of disposable income was fully explained at that time by the less active redistribution of Spanish social policy vis-­à-­vis the EU15. Owing to the crisis, there has been an increase in market inequality of such intensity (15 per cent) that the weak and weakened Spanish welfare state was unable to fully compensate, leading to an increase in inequality in terms of disposable income of nearly 9 per cent.13 If we refer to in-­kind public services, such as education and health care, the available evidence shows that both low-­and middle-­income classes disproportionally benefit from this, with only the most well-­off avoiding using public services to some extent (Calero and Gil 2013).

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 522

28/09/2016 08:25



Changes in the world of work and the middle class in Spain ­523 55.0

52.8

51.8 50.0

2005 2012

48.9 46.0

Gini Index (%)

45.0 39.8 40.0 35.7

36.5 35.0

34.4

35.0

32.2 29.9

30.0

30.5

25.0 EU(15): market

Spain: market

EU(15): market + pensions

Spain: market + pensions

EU (15): Spain: disposable disposable

Source:  Authors’ analysis from EU-SILC.

Figure 12.16 Income inequality before and after social transfers in the EU15 and Spain, 2005 and 2012

4.  CASE STUDIES 4.1  Case Study 1: The Welfare State as ‘Middle-­class’ Job Provider Traditionally, the debate on the relationship between the welfare state and the middle class is addressed from the perspective of who benefits from welfare state programmes and who finances them. However, the relationship between the middle class and the welfare state goes beyond these perspectives. In most European countries the welfare state, whether directly or indirectly, is responsible for the creation of a large number of traditionally middle-­class occupations, and as such can be an important factor in the development of the middle class. In what follows we analyse the extent to which that is true for Spain. As mentioned in previous sections, Spain entered the last quarter of the twentieth century with a nascent welfare state. The two major employment generating activities of the welfare state are education and health care, both sectors with a large number of professional jobs traditionally identified with middle-­class status and income. Health workers are the largest single group of public employees in the Spanish Public Administration (over half a million) and together with public education employees add up

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 523

28/09/2016 08:25

524

Europe’s disappearing middle class?

to nearly half of total public employees (45 per cent in 2014). In order to identify the location of these employees and the rest of welfare state jobs in terms of job quality and wages, and the contribution of the consolidation of the Spanish welfare state to the development of middle-­class jobs, we have applied the methodology used in section 3.1.1 to allocate job c­ reation/ destruction along the wage distribution, differentiating now between welfare state jobs – measured by education, health care, social assistance and public administration – and the rest. The results, presented in Figure 12.17, are clear in terms of the important role played by the development of the welfare state in the upgrading of the Spanish employment structure and the development of a middle-­class job segment. Welfare state services – and public administration in general – employ a larger than average highly educated workforce with higher than average wages (Antón and Muñoz de Bustillo 2013). That shows in the relative important presence of welfare state employment in the upper quintiles. Together with the important role of the public sector as employer in the development of middle-­to high-­quality jobs in Spain in recent decades, the figure also shows its stability over time. During 1977–85, a period of employment destruction across the board, the development of the welfare state shows its imprint in quintiles 3–5, especially in the final quintile. This dynamic is also present in the following growth period, explaining more than two-­thirds of total employment growth in the upper quintile, and even during the 1993 recession. Although the high employment growth of the years of the economic boom dwarfs the contribution of the welfare state to total employment growth, its impact cannot be considered negligible, especially in the upper quintiles. Finally, during the crisis the contribution of welfare state services to employment follows a very distinctive pattern when we look at the first or the second stage of the crisis, coinciding with a major change in the Spanish anti-­crisis economic policy produced in 2010. During the first part of the crisis, Spain followed a countercyclical policy based on increasing public expenditure. In 2010, caught in a debt crisis and largely forced by the pressure exerted by the ‘markets’, the European Commission and the European Central Bank (ECB), Spain made a U-­turn in terms of economic policy and applied an intense (if not always successful) policy of fiscal consolidation. As a result, for the first time in the past three decades, the Spanish public sector, from being an employment generating actor (the only one in the first years of the crisis), became a major factor in the destruction of middle-­and high-­ quality employment.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 524

28/09/2016 08:25



Changes in the world of work and the middle class in Spain ­525 1977–85

500.0

Thousands of workers

0.0 –500.0 –1000.0 –1500.0

2000 1500 1000 500 0 –500 –1000

–2000.0

I

II

II

IV

1985–91

2500

V

Total

I

1977–85 total Employment in welfare state services

II

IV

V

Total

1985–91 total Employment in welfare state services

1992–93

200

II

1994–2008

8000

100

7000 Thousands of workers

0 –100 –200 –300 –400 –500 –600 –700

I

II

II

IV

V

6000 5000 4000 3000 2000 1000 0

Total

–1000

I

1992–93 total Employment in welfare state services

II

IV

V

Total

1994–2008 total Employment in welfare state services

2008–10

500

2011–13 0 Thousands of workers

0 –500 –1,000 –1,500 –2,000

II

I

II

II

IV

V

Total

–200 –400 –600 –800 –1000 –1200 –1400 –1600

2008–10 total Employment in welfare state services

I

II

II

IV

V

Total

2011–13 total Employment in welfare state services

Source:  Muñoz de Bustillo and Antón (2014: 33).

Figure 12.17 The role of the development of the welfare state in employment changes, Spain, 1977–2013

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 525

28/09/2016 08:25

526

Europe’s disappearing middle class?

2 000 000

Enrolment in tertiary education, all programmes Enrolment in tertiary education per 100 000 inhabitants

5000 4500 4000 3500 3000

1 500 000

2500 2000

1 000 000

1500 1000

500 000

500 0 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012

0

Enrolment per 100 000 inhabitants.

2 500 000

Source:  Authors’ analysis from UNESCO Statistical Institute and Anuario Estadístico de España (INE), various years.

Figure 12.18 Evolution of university students and tertiary enrolment rate, Spain, 1960–2012 4.2 Case Study 2: Education and Growth of the Middle Class: The Crisis as a Turning Point? As we mentioned in section 3.2.3, from the 1970s onward, Spain experienced a radical change in the nature of its tertiary education system. In two decades, from 1971 to 1991, tertiary education gross enrolment rate jumped from less than 10 per cent to more than 35 per cent, reaching 85 per cent in 2012.14 Such a transformation, as shown in Figure 12.18, implied a massive increase in the total number of university students, from less than a quarter of a million in 1970 to nearly 2 million in 2012. Within a few decades, for those students that finished secondary education, going to university became the norm, old universities expanded their capacity and new public universities were created across the country (reaching a total of 50 public universities with more than 800 establishments) to facilitate access to tertiary education. As argued by the US Middle Class Task Force (US Department of Commerce Economics and Statistics Administration 2010), if we define middle-­class families in terms of aspiration (instead of income), college education for their children is one of the half a dozen items that characterize middle-­class status. In this respect, the importance of tertiary education in the process of developing, consolidating and maintaining the middle class cannot be downplayed.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 526

28/09/2016 08:25



Changes in the world of work and the middle class in Spain ­527

In this section we study the role of tertiary education in the development of the Spanish middle class and the improvement of social mobility, as well as the potential impact of the changes in the Spanish university system during the crisis that could jeopardize the advances in this field of recent decades. Education plays an important role in the development of the middle class through two different mechanisms: facilitating social mobility and helping to reduce inequalities. Regarding the first item, in the last half of the twentieth century there was a substantial degree of educational mobility in Spain. The intensity of educational mobility can be summarized by looking at the correlation between the educational level of parents and children through different cohorts in different years. In this respect, for those who were born in 1945 and before, the correlation of parents’ and children’s education was 0.49; by contrast, the correlation for the cohort born from 1971 to 1975 was 0.34 (Raymond et al. 2009). Turning now to the distributional incidence of educational e­ xpenditure,15 according to recent estimates (Calero and Gil 2014) overall education expenditure is highly progressive, although with important differences by category, as the redistribution is highly progressive in pre-­primary, primary and secondary public schools, while being also progressive but with lower intensity in pre-­primary, primary and secondary private public-­financed schools (enseñanza concertada)16 and public tertiary education. In this respect, as we can see in Figure 12.19, families in the upper half of the income distribution, but also in the bottom decile, are clear beneficiaries of public tertiary education expenditure. This result highlights one of the dilemmas regarding tertiary education policy: on one side, from a redistributive perspective primary and secondary public education – with a much higher redistributive impact if only because the percentage of the population of the first quintile reaching university is much lower than among the upper two quintiles – should be given priority. On the other side, state underfinancing of tertiary education would certainly make it more difficult to increase the university attendance of children from low-­ income families, as well as alienate middle classes from the welfare state. Summing up, in the past three decades Spain has witnessed an unprecedented revolution in its university system, from an ‘elitist system’ addressed to a minority of the population to a ‘university of the masses’, with enrolment rates among the highest in the EU. This process has benefited the middle and upper classes and has also contributed to their increase in size by facilitating social mobility. In this context, the economic crisis of 2009–13 has affected public universities in different ways, with the corresponding negative impact on middle classes:

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 527

28/09/2016 08:25

528

Europe’s disappearing middle class?

20.0 18.0 16.0

Tertiary education expenditure

17.3

Total education expenditure 15.2

15.0

14.0 11.6

12.0 10.0

9.4 8.3

8.0

6.0

6.0

11.7 10.5 11.3 10.0 9.6 8.8 8.6

10.5 7.2

5.4

8.7

8.1

6.7

4.0 2.0 0.0

1

2

3

4

5

6

7

8

9

10

Source:  Authors’ analysis of Calero and Gil (2014: 20), corresponding to 2010.

Figure 12.19 Distribution of public education expenditure by type and income decile, Spain, 2010 ●●

●●

●●

There has been a significant reduction in the financing of public universities, with a nominal reduction of the overall budgets of Spanish public universities from 2010 to 2014 of 15 per cent in nominal terms (21 per cent in real terms). During the same period, the number of students increased, leading to a slightly larger reduction in terms of euros per student (22 per cent). Since the start of the crisis, university tuition has risen over the Consumer Prce Index (CPI), increasing average cost from the school year 2009–10 to the school year 2013–14 by an average of 47 per cent (37 per cent in real terms).17 The increase in tuition costs, in a time of recession and unemployment growth, has not been met with an increase in scholarships to ensure equality of opportunity. Quite the opposite. The requirements for scholarships have been tightened both in terms of required credits passed (100 per cent) and in terms of average grade (6.5/10). Some scholarships, such as the so-­called Beca Salario (wage-­ scholarship) to compensate the opportunity cost for low-­ income families of attending university instead of working, have been eliminated altogether (UGT 2014). These measures are very likely to damage equality of opportunity beyond their specific effects on the middle classes.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 528

28/09/2016 08:25



Changes in the world of work and the middle class in Spain ­529 ●●

●●

Changes in scholarships aimed at facilitating mobility between EU universities (Erasmus): perhaps the major change in relation to the middle class is that now only students with a general scholarship (income based) will be eligible for the state Erasmus scholarship. The rest will only receive the EU contribution and whatever contribution the regional government may make (only in some cases). The budget for this type of scholarship has been drastically reduced (by 75 per cent since 2011). In 2014 the Ministry of Education, Culture and Sports unilaterally changed the structure of university studies, making it possible now (although not compulsory) to offer university degrees of three years’ duration (instead of the standard four year degrees approved in the last reform). The reduction of one year in undergraduate studies is to be compensated by the possibility of increasing the duration of postgraduate (masters) studies, making a total of 3 + 2 years. It has been argued that the system will further increase the cost of university education, as the tuition per credit of postgraduate studies in 2013–14 was on average 45 per cent higher than the tuition fees of undergraduate studies. The overall cost of university would be lower with the new system, if only for those students not following postgraduate studies, but so would the quality of the education received. As the system has not been implemented yet, it remains to be see whether the labour market will welcome the new devalued (in terms of years of study) degrees or whether the change will lead to the creation of a new category of underpaid workers with university studies.

Altogether, these changes are affecting the purchasing power of Spanish families with increases in total expenditure allocated to university education of 18.3 per cent in 2012 and 33 per cent in 2013 (compared with −6.9  per cent and −10.3 per cent in total expenditure).18 It seems that, as with other items (such as public pensions or quality health-­care services), things that middle-­class families considered secure in the past cannot be taken for granted in the future.19

5.  CONCLUSION AND POLICY ISSUES According to Spanish public opinion (CIS 2011) the ideal type of society, in terms of income distribution, is one with a ‘diamond’ profile, with a low proportion of the population at the bottom and the top and a large proportion in the middle. In 2011, 49 per cent of Spaniards considered

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 529

28/09/2016 08:25

530

Europe’s disappearing middle class?

this to be their preferred model of society. However, when asked what the real pattern of distribution was, only 8 per cent thought that Spanish society roughly has this diamond shape, while 31 per cent believed it to be closer to a pyramid. This impression is related to the common belief that the middle-­income class – the so-­called ‘disappearing middle’ – has been the group more affected by the crisis. In the terms of a recent newspaper article: ‘the breaking of the middle class is a consequence of the crisis with particularly high economic, social and political implications’ (Velázquez-­Gaztelu 2014). This interpretation has come to be regarded as an unquestioned truth, regardless of what the data show. The data presented in this chapter (as well as in other research, such as FOESSA 2014) call this interpretation into question. The middle class has clearly not been immune to the crisis, but the lower-­income group has suffered more. From the perspective taken in this chapter, the middle-­income class’s sense of victimization can be explained by the rupture produced by the crisis and its management and their expectations about the future. Suddenly, nothing could be taken for granted anymore, from the stability of their jobs, to the middle-­class future of their children; from the quality and availability of services, such as health care and education, to the security of their future pensions. Everything was now being called into question. Out of the dynamics reviewed in these pages we call attention to two elements that might shape the future of the middle-­income class. The first is the evolution of the structure of employment. As we have seen, the structure of employment in Spain has experienced upgrading in the long run, with an important boost to middle-­income class jobs located in the central and upper wage quintiles. This process, which has been buttressed by the development of the welfare state and the increase of public employment in areas related to middle-­class jobs – such as health care, education or public administration – was reversed during the economic crisis of 2008–13, resulting in intense job destruction in the middle-­wage quintiles and the corresponding hollowing out of middle-­class jobs. It is impossible to know whether these changes mark the start of a new pattern of polarized employment creation in the future, or whether such change will be confined to the crisis. In any case, whatever happens will have major implications for the future of the middle class, as the conquest of a polarizing pattern will affect its employment base. The second issue is related to the impact of fiscal adjustment and structural reforms on the future prospects of the population in general, and of the middle-­income class in particular. The increase in temporary and involuntary part-­time employment, although more concentrated in unskilled and low-­pay jobs, is also affecting middle-­class jobs, especially in the case of the younger generation. The increase in unemployment across all

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 530

28/09/2016 08:25



Changes in the world of work and the middle class in Spain ­531

cohorts and occupations and the development of new forms of precarious employment not only imply higher uncertainty for middle-­income workers, but also higher risk of a sudden and sizeable decrease in income should they lose their jobs, as the conditions of re-­employment would be below those prevailing in previous jobs. Furthermore, austerity has translated into a deterioration of the core services of the welfare state, such as education and health care, while the debate on the future sustainability of pensions is generating new anxieties about the ability of future governments to honour the compromises made in terms of pension rights. A good example of this type of warning is the recent declaration by the governor of the Spanish Central Bank, who stated that ‘the current system does not guarantee the pension level expected by citizens’ (De Barrón 2015). This kind of statement clearly contributes to the uncertainty about the future in this crucial area of the welfare state. Until now, in Spain, as in most European countries, the welfare state has been a major element in the development and consolidation of the middle class and in the successful fulfilment of its aspirations. Many of the aspirations defining the ‘middle class’ (access to health services, tertiary education, pensions and even holidays) were granted by the welfare state, either directly through social spending or indirectly through the enactment of labour regulations favouring workers’ rights. The triumph of austerity and structural labour reforms all across Europe might jeopardize the role played by the state in satisfying such aspirations.

NOTES   1. We thank Sudipa Sarkar for excellent research assistance. We gratefully acknowledge funding from the Spanish Ministry of Economics and Competitiveness (research project CSO2013-­41828-­R).   2. This approach has the advantage of not being contingent on the specific (in terms of income brackets) definition of the middle class. On the down side, although the index is straightforward and easy to interpret – 0 means no polarization at all and 1 means maximum level of polarization – it is nevertheless more obscure as it cannot be interpreted in terms of the size of the middle class.   3. Owing to the way mobility is defined, there will always be a tendency for lower-­and upper-­income groups to show lower mobility. Low-­income groups have their downward mobility censored and the class itself is composed of only two deciles; the same is valid for the high-­income group, this time in terms of upward mobility.   4. The prevalence of downward mobility for the middle-­and high-­income groups might explain the growing pessimism of the Spanish middle class about the future faced by the younger generation. According to a 2012 survey conducted by the Pew Research Center (2012), although 71 per cent of Spaniards considered that their standard of living was better (32 per cent) or much better (39 per cent) than that of their parents, when asked about the future, about whether it was ‘easy or difficult for a young person to get a better job and to become wealthier than his or her parents’, only a meagre 5 per cent

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 531

28/09/2016 08:25

532

  5.

  6.

  7.

  8.

  9. 10. 11. 12.

13.

14.

15. 16. 17.

Europe’s disappearing middle class? considered that it was easy or very easy, while 69 per cent considered that it was very difficult. The homogamy rate for Spain in 2008 calculated from European Union Labour Force Survey was 63 per cent. According to this source Spain was in the middle of the distribution of the 18 OECD countries considered in the sample, with the Netherlands presenting the lowest rate, 53 per cent and Portugal the highest, 78 per cent. Other sources, such as the European Social Survey, show higher levels of homogamy, at around 74 per cent (Katrnák et al. 2012). Although immigration tends to be male dominated, in the Spanish case, and in relation to immigration from Latin America and the Caribbean, in 2007 women made up more than 54 per cent of immigrants (Muñoz de Bustillo and Antón 2010). This gender asymmetry was even stronger at the beginning of the immigration wave, as ten years earlier two out of three female migrants from this region were women (Martínez 2003). Nevertheless, as the arrival of foreigners has an effect on the native-­born population through several channels (from their labour market outcomes to mixed marriages, housing and household arrangements), it is not easy to figure out a proper counterfactual without the massive migrant inflows. Early leavers from education and training refer to persons aged 18 to 24 fulfilling the following two conditions: first, the highest level of education or training attained is ISCED 0, 1, 2 or 3c short (before competition of upper secondary), second, respondents declared not having received any education or training in the four weeks preceding the survey (numerator). The denominator consists of the total population of the same age group. For details on the socioeconomic factors behind early school abandonment in Spain see Fernández-­Macías et al. (2013). The role of education in the development of the middle class in Spain is discussed in more detail in section 4. The loss of public trust on the part of the population, as shown by various CIS surveys, is shared by the political parties and the employers’ organizations. According to our analysis of the ECVT 2010 (the Spanish working conditions survey) the group with the highest affiliation rate by income bracket − 27 per cent – has an income that roughly corresponds to our upper MiC. As the same dynamic affected the rest of the EU15, in terms of convergence the result is less impressive. In the EU15 social protection increased from 25.6 per cent of GDP in 2007 to 29.2 of GDP in 2012. In Ireland, the increase was much higher, from 17 to 31 per cent. We should bear in mind that half of Spanish social spending consists in in-­kind education and health-­care benefits, while the cash component of the national welfare state is dominated by unemployment insurance and contributory pensions. This leaves very little room for purely redistributive transfers. The gross tertiary enrolment rate is defined by the United Nations Educational, Scientific and Cultural Organization (UNESCO) Statistical Institute as the number of students enrolled in a given level of education, regardless of age, expressed as a ­percentage of the official school-­age population corresponding to the same level of education. For the tertiary level, the population used is the five-­year age group starting from the official secondary school graduation age. In 2011 public spending on education in Spain was lower than the EU28 average, at 4.82 per cent of GDP compared with 5.25 per cent. The difference in distributional impact in this case is due to the selection bias of children attending this type of schools, mostly from higher-­income level families (Urquizu 2008). In 2012 the Ministry of Education approved a change in the system of calculating university tuition; instead of using an annual revision formula that links the increase in tuition to the CPI (+ a maximum of 4 percentage points), the new system takes as reference a range from 15 to 25 per cent of the cost of university attendance in each region. The increase in tuition fees is based on the average cost of fees per credit in first

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 532

28/09/2016 08:25



Changes in the world of work and the middle class in Spain ­533

enrolment in undergraduate studies. Ministerio de Educación (2011) and Ministerio de Educación, Cultura y Deporte (2014). 18. Encuesta de Presupuestos Familiares (Household Budget Survey), base 2006. 19. In this respect, according to the comparative analysis of the cost of university for students in Europe by Sacristan (2014), and contrary to the widespread idea defended by the Spanish Ministry of Education that Spanish university tuition fees are relatively low, Spain is one of the most expensive European countries with regard to university tuition. In 11 of the countries reviewed by Sacristan (2014), undergraduate studies are completely free, while only in eight of the 38 covered in the study is tuition higher than in Spain. In parity power purchasing (PPP) terms, Spain ranks sixth out of 33 countries in terms of cost of attending university (after the United Kingdom – with the exception of Scotland – Slovenia, Ireland, Hungary and Lithuania). As for postgraduate studies, in most countries the tuition is the same as or similar to undergraduate studies. Spain is one of the seven countries in the sample in which postgraduate tuition is substantially higher.

REFERENCES Aidt, T.S. and Z. Tzannatos, Z. (2008), ‘Trade unions, collective bargaining and macroeconomic performance: a review’, Industrial Relations Journal, 39 (4), 258–95. Antón, J.I. and R. Muñoz de Bustillo (2013), ‘Public-­private sector wage differentials in Spain. An updated picture in the midst of the Great Recession’, Investigación Económica, 74 (292), 115–57. Antón J.I., R. Muñoz de Bustillo and M. Carrera (2010), ‘From guests to hosts: immigrant-­native wage differentials in Spain’, International Journal of Manpower, 31 (6), 645–59. Ayala, L. (ed.) (2014), ‘Distribución de la renta, condiciones de vida y políticas redistributivas’, in VII Informe sobre exclusión y desarrollo social en España, Madrid: Fundación FOESSA. Bover, O. and M. Arellano (1995), ‘Female labour force participation rate in the 1980s: the case of Spain’, Investigaciones Económicas, 19 (2), 171–94. Calero, J. and M. Gil (2013), ‘La incidencia distributiva del gasto público en sanidad y educación en España’, in J. Ruiz-­Huerta (ed.), 1er Informe sobre la Desigualdad en España, Madrid: Fundación Alternativas, pp. 243–79. Calero, J. and M. Gil (2014), ‘Un análisis de la incidencia distributiva del gasto público en sanidad y educación en España’, Documento de Trabajo 2.8, VII Informe sobre exclusión y desarrollo social en España 2014, Madrid: FOESSA, accessed 4 March 2015 at http://www.foessa2014.es/informe/uploaded/­ documentos_trabajo/15102014141702_937.pdf. Carrasco, R., J.F. Jimeno and A.C. Ortega (2008), ‘The effects of immigration on the labour market performance of native-­born workers: some evidence for Spain’, Journal of Population Economics, 21 (3), 627–48. Centro de Investigaciones Sociológicas (CIS) (2005), Representación y Participación Política en España, Estudio 2588, Madrid: CIS. Centro de Investigaciones Sociológicas (CIS) (2014a), Barómetro de Abril 2014, Estudio 3021, Madrid: CIS. Centro de Investigaciones Sociológicas (CIS) (2014b), Barómetro de Diciembre 2014, Estudio 3047, Madrid: CIS. De Barrón, I. (2015), ‘Linde augura una segura bajada de las pensiones a largo

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 533

28/09/2016 08:25

534

Europe’s disappearing middle class?

plazo’, El País, 24 June, accessed 13 July 2015 at http://economia.elpais.com/ economia/2015/06/24/actualidad/1435161458_131212.html. Esteve, A. and C. Cortina (2006), ‘Changes in educational assortative mating in contemporary Spain’, Demographic Research, 14 (May), 405–28. Farré, L., L. González and F. Ortega (2011), ‘Immigration, family responsibilities and the labour supply of skilled native women’, The B.E. Journal of Economic Analysis & Policy, 11 (1), art. 34. Farré, L., F. Ortega and R. Tanaka (2015), ‘Immigration and school choices in the midst of the great recession’, IZA DP No. 9234, July, Institute for the Study of Labor, Bonn. Fernández-­Macias, E. (2012), ‘Job polarisation in Europe? Changes in employment structure and job quality, 1995–2007’, Work and Occupations, 39 (2), 157–82. Fernández-­Macías, E. and J. Hurley (2008), ‘More and better jobs: patterns of employment expansion in Europe’, European Foundation for the Improvement of Living and Working Conditions, ERM Report 2008, Dublin. Fernández-­Macias, E., J.I. Antón, F.J. Braña and R. Muñoz de Bustillo (2013), ‘Early school leaving in Spain: evolution, intensity and determinants’, European Journal of Education, 48 (1), 150–64. Fernández-­Macías, E., D. Storrie and J. Hurley (eds), (2012), Transformations of the employment structure in the EU and US, 1995–2007, Basingstoke: Palgrave. Fomento de Estudios Sociales y de Sociología Aplicada (FOESSA) (2014), VII  Informe sobre exclusión social y desarrollo social en España, Madrid: FOESSA. Foster, J.E. and M.C. Wolfson (2010), ‘Polarization and the decline of the ­middleclass: Canada and the U.S.’, Journal of Economic Inequality, 8 (2), 247–73. Garrido, L. and H. Cebolla (2010), ‘Rendimiento educativo y concentración de inmigrantes en las escuelas españolas: PISA 2006’, Presupuesto y Gasto Público, 61 (4), 159–76. Goldthorpe, J.H. (2012), ‘De vuelta a la clase y el estatus: por qué debe ­reivindicarse una perspectiva sociológica de la desigualdad social’, Reis, 137 (January–March), 43–58. Gonzalez, L. and F. Ortega, F. (2011), ‘How do very open economies adjust to large immigration flows? Evidence from Spanish regions’, Labour Economics, 18 (1), 57–70. Instituto de Mayores y Servicios Sociales (IMSERSO) (2005), El cuidado a las personas mayores en los hogares españoles. El entorno familiar, Madrid: Ministerio de Trabajo y Asuntos Sociales. Jaumotte, F. and C. Osorio (2015), ‘Inequality and labour market institutions’, Staff Discussion Notes, SDN/1/15/14, July, Washington DC: International Monetary Fund. Jiménez, S. and N. Jorgensen (2009), ‘Inmigración y demanda de seguros ­sanitarios’, Gaceta Sanitaria, 23 (1), 19–24. Katrnák, T., P. Fucík and R. Luijkx (2012), ‘The relationship between educational homogamy and educational mobility in 29 European countries’, International Sociology, 27 (4), 551–73. López, M. (2011), ‘Government reform collective bargaining rules’, EurWORK, European Observatory of Working Life, Eurofound, Dublin, accessed 12  April  2015 at http://eurofound.europa.eu/observatories/eurwork/articles/ other/government-­to-­reform-­collective-­bargaining-­rules. Madland, D., K. Walter and N. Bunker (2011), Unions Make the Middle Class.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 534

28/09/2016 08:25



Changes in the world of work and the middle class in Spain ­535

Without Unions, the Middle Class Withers, Washington, DC: Center for American Progress Action Fund. Manrique Arribas, J.C. (2007), ‘La familia como medio de inclusión de la mujer en la sociedad franquista’, Hispania Nova. Revista de Historia Contemporánea, no. 7, 193–221, accessed 11 February 2015 at http://hispanianova.rediris.es. Martínez García, J.S. (2014), ‘Cómo afecta la crisis a las clases sociales?’, ZOOM Político, 2014/20, Madrid: Fundación Alternativas. Martínez, R. (2003), La reciente emigración latinoamericana a España, Serie Población y Desarrollo 40, Santiago de Chile: CEPAL. Mercader, J.R. (2002), ‘Algunas reflexiones sobre el modelo social de la Constitución española, 25 años después’, Cuadrenos Constitucionales de la Cátedra Fadrique Furió, no. 40, Valencia. Ministerio de Educación (2011), Facts and Figure of the Spanish University System. Academic Year 2010–11, Madrid: Ministerio de Educación. Ministerio de Educación, Cultura y Deporte (2014), Datos Básicos del Sistema Universitario Español Curso 2013/2014, Madrid: Ministerio de Educación, Cultura y Deporte. Molina, O. and F. Miguelez (2013), ‘From negotiation to imposition: social dialogue in austerity times in Spain, International Labour Office (ILO), Working Paper, No. 51, Geneva. Morcillo, A. (2012), ‘Españolas con, contra, bajo, (d)el franquismo’, Por Aurora Morcillo, Desacuerdos. Sobre arte, políticas y esfera pública en el Estado español, no. 7, Madrid: Museo Nacional Centro de Arte Reina Sofia, pp. 42–63. Muñoz de Bustillo R. and J.I. Antón (2010), ‘De la España que emigra a la España que acoge: contexto, dimensión y características de la inmigración latinoamericana en España’, América Latina Hoy, 55 (August), 15–39. Muñoz de Bustillo, R. and J.I. Antón (2011), ‘From rags to riches? Immigration and poverty in Spain’, Population Research and Policy Review, 30 (October), 661–76. Muñoz de Bustillo, R. and J.I. Antón (2014), ‘Long-­term trends in the jobs structure in Spain: 1977–2013’, in R. Muñoz de Bustillo (ed.), Study and Analysis of Long-­Term Trends in the Jobs Structure in Four (4), European Countries, report prepared for the European Foundation for the Improvement of Living and Working Conditions, Dublin. Muñoz de Bustillo, R. and J.I. Antón (2015), ‘Turning back before ­arriving: the weakening of the emerging Spanish welfare state’, in D. Vaughan-­Whitehead (ed.), The European Social Model in Crisis: Is Europe Losing its Soul?, Cheltenham, UK and Northampton, MA, USA: Edward Elgar and Geneva: ILO, pp. 451–506. Muñoz de Bustillo, R., E. Fernández-­Macías and J.I. Antón (2008), El trabajo a tiempo parcial en España en el contexto de la UE, Madrid: Ministerio de Trabajo e Inmigración. Pew Research Center (2012), Global Attitude Project Spring 2012, Pew Research Center, Washington, DC. Novales, A. (1989), La incorporación de la mujer al mercado de trabajo en España: participación y ocupación’, Moneda y Crédito, 188 (1st semester), 457–78. Organisation for Economic Co-­ operation and Development (OECD) (2011), Doing Better for Families, Paris: OECD. Ortíz, M. (2012), ‘Efectos escolares de la inmigración: discursos sobre concentración’, Revista Iberoamericana de Educación/Revista Ibero-­americana de Educação, 59 (1), 1–10.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 535

28/09/2016 08:25

536

Europe’s disappearing middle class?

Parella, S. (2003), Mujer, inmigrante y trabajadora: la triple discriminación, Madrid: Editorial Anthropos. Pérez, J.A. (2014), ‘Este es el tiempo del cambio. España 1982–1996. Una sociedad en transformación’, in C. Zubeldia and D. Itirriaga (eds), España en democracia. Actas del IV Congreso Internacional de Historia de Nuestro Tiempo, Logroño: Universidad de La Rioja, pp. 63–82. Raymond, J.L., J.L. Roíg and Gómez, L. (2009), ‘Rendimientos de la educación en España y movilidad intergeneracional’, Papeles de Economía Española, 119 (April), 188–205. Sacristán, V. (2014), El coste de estudiar en Europa. Precios, becas y otras ayudas a las universidades europeas (2013–2014), Observatori Sistema Universitari, accessed 11 March 2015 at http://www.observatoriuniversitari.org/es/2014/06/19/ el-­coste-­de-­estudiar-­en-­europa/. Schumpeter, J.A. (1942), Capitalism, Socialism, and Democracy, 3rd edn, New York: Harper and Brothers. Unión General de Trabajadores (UGT) (2014), Becas y ayudas al estudio. Fin de la equidad, Madrid: Unión General de Trabajadores. Urquizu, I. (2008), ‘La selección de Escuela en España’, Revista de la asociación de Sociología de la Educación, 1 (2), 70–89. US Department of Commerce Economics and Statistics Administration (2010), Middle Class in America, prepared for the Office of the President of the United States Middle Class Task Force, Washington, DC: US Department of Commerce Economics and Statistics Administration. Velázquez-­Gaztelu, J.P. (2014), ‘Adiós a la clase media, bienvenidos al precariado’, El Diario, 18 December, accessed 8 February 2015 at http://www.eldiario.es/alternativaseconomicas/Adios-­clase-­media-­bienvenidos-­precariado_6_336376374. html. Wright, E.O. and R.E. Dwyer (2003), ‘Patterns of job expansion in the USA: a comparison of the 1960s and 1990s’, Socio-­Economic Review, 1 (3), 289–325.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 536

28/09/2016 08:25

537

VAUGHAN 9781786430595 PRINT (M4051).indd 537

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:25

51.1 29.8 20.0 18.9 21.2 10.0 29.5 4.1 9.6 27.0 29.8 7.3 8.0 38.8 43.0 10.2

Personal characteristics Share of women

Age 0–15 16–29 30–44 45–64 65 and over

Activity status Employed Unemployed Retired Another inactivity Aged less than 16 years old Share living in female-­headed house

Age of household head Less than 30 years old 30–44 years old 45–64 years old 65 and over

1980–81

5.3 33.5 46.0 15.3

31.9 5.5 14.2 27.3 21.1 9.9

21.1 23.6 19.2 23.1 13.0

50.8

1990–91

Middle class

Table 12A.1  Characteristics of middle class and total households, Spain

APPENDIX 12A.1

2.4 31.6 46.7 19.3

42.4 4.1 12.9 24.8 15.8 14.2

15.8 20.0 22.7 26.0 15.5

51.0

2004

7.7 37.1 43.0 12.2

29.1 4.3 10.3 27.0 29.2 8.4

29.2 19.4 18.3 21.7 11.4

51.4

1980–81

5.4 33.3 44.9 16.4

31.0 5.9 14.6 27.1 21.5 11.5

21.5 22.9 19.0 22.9 13.8

51.1

1990–91

Total

2.3 30.4 45.6 21.8

39.9 4.6 13.9 26.1 15.5 15.7

15.5 19.2 21.7 26.0 17.7

51.6

2004

538

VAUGHAN 9781786430595 PRINT (M4051).indd 538

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:25

88.7 6.5 4.8 80.0 4.3 14.2 1.5 1.6 0.4 9.9 30.3 57.9 4.5 1.4 0.3 1.3

Activity status of household head Employed Unemployed Retired Other inactivity

Type of household One–person household One adult with children Two adults without children Two adults with children Three or more adults Average household size Average number of children below 14 years old Average number of members aged 65 and more Average number of workers

1980–81

Education of household head Lower secondary education Upper secondary education and professional training Higher education

Table 12A.1  (continued)

2.0 0.2 11.5 21.3 65.0 4.2 0.8 0.4 1.3

70.8 4.1 23.7 1.4

84.5 8.3 7.2

1990–91

Middle class

3.6 0.5 16.7 20.7 58.6 3.6 0.5 0.4 1.5

70.7 2.2 23.1 4.0

65.3 16.9 17.8

2004

2.2 0.4 11.7 29.1 56.6 4.5 1.3 0.4 1.2

76.0 5.6 16.6 1.8

86.3 6.5 7.2

1980–81

2.9 0.4 12.9 21.6 62.1 4.1 0.9 0.4 1.3

67.6 5.5 25.1 1.8

82.9 7.6 9.5

1990–91

Total

4.6 0.5 18.4 19.9 56.6 3.5 0.5 0.4 1.4

66.6 3.3 25.1 5.0

65.7 15.5 18.8

2004

539

VAUGHAN 9781786430595 PRINT (M4051).indd 539

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:25

70.0 22.8 6.2 1.1

Housing tenure status Owned Rented Provided rent-­free Others 79.9 13.4 6.7 0.0

0.3 43.1 49.4 34.1 68.1 22.2 11.6 2.2 62.8 81.0 97.9 94.4 85.8 27.9 86.3 10.6 3.0 0.1

0.4 53.5 62.6 31.3 57.6 40.1 14.3 0.5 85.3 86.9 99.8 99.4 99.8 50.2 69.9 22.3 6.7 1.2

0.3 43.3 49.0 40.6 70.9 18.3 7.0 7.5 53.3 52.3 92.1 69.7 27.7 6.0 78.7 14.2 7.2 0.0

0.3 42.0 48.3 35.3 69.0 22.8 11.9 2.6 58.7 77.9 97.7 93.0 85.8 27.8

85.8 10.9 3.1 0.1

0.4 50.2 60.0 33.1 60.2 38.1 14.8 0.6 81.9 85.6 99.8 99.2 99.8 48.6

Source:  Authors’ analysis from Household Budget Surveys 1973–74, 1980–81, 1990–91, Continuous Household Budget Survey 1998–2004 and EU SILC.

0.3 44.2 49.6 39.9 70.5 17.1 6.4 7.0 58.1 55.4 94.9 74.7 28.6 6.1

Average share of members working Average share of members (+16) working Average share of members (16–64) working Average share of demographically dependent members Average share of economically dependent members Average share of working age women employed Household owns a secondary accommodation House overcrowding Household owns a car Household owns a phone Household owns a fridge Household owns a washing machine Household owns a TV House has heating

540

VAUGHAN 9781786430595 PRINT (M4051).indd 540

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:25

50.2 13.6 21.1 25.3 24.4 15.6 52.3 4.1 10.9 18.0 14.7 27.5 3.5 30.4 44.1 21.9

Personal characteristics Share of women

Age 0–15 16–29 30–44 45–64 65 and over

Activity status Employed Unemployed Retired Another inactivity Aged less than 16 years old

Characteristics of households Share living in female-­headed households Household head below 30 years of age Household head 30–44 years of age Household head 45–64 years of age Household head 65 years of age or over

2003

35.2 4.2 33.8 42.2 19.9

49.7 4.3 12.2 18.8 15.0

13.9 18.4 27.1 24.9 15.7

49.6

2007

Middle class

Table 12A.2 Characteristics of middle class and total households, Spain (final)

39.0 3.4 31.3 40.0 25.3

40.8 9.6 14.6 20.1 15.0

14.1 14.4 25.0 26.2 20.4

50.8

2011

28.6 3.3 31.1 41.8 23.8

47.5 5.1 11.7 20.2 15.6

14.5 19.5 24.7 23.5 17.8

50.9

2003

35.6 4.3 33.2 41.7 20.9

45.9 5.1 12.4 20.9 15.7

14.7 17.4 25.9 24.9 17.2

50.5

2007

Total

37.9 3.3 32.6 41.4 22.8

38.9 11.9 12.6 20.4 16.2

15.3 14.9 25.2 26.3 18.2

50.7

2011

541

VAUGHAN 9781786430595 PRINT (M4051).indd 541

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:25

61.2 17.7 21.1 66.2 3.5 18.8 11.5 9.0 0.7 27.5 27.3 35.6 3.6 0.5 0.4 1.8

Education of household head Lower secondary education Upper secondary education and prof. training Higher education

Activity status of household head Employed Unemployed Retired Another inactivity

Type of household One-­person household One adult with children Two adults without children Two adults with children Three or more adults Average household size Average number of children below 14 years of age Average number of members aged 65 or over Average number of workers 9.5 0.7 28.2 27.3 34.3 3.4 0.5 0.4 1.6

67.1 3.8 18.3 10.8

54.7 21.0 24.3

11.7 1.2 30.4 26.8 30.0 3.1 0.5 0.5 1.3

55.4 9.9 20.5 14.1

53.4 21.1 25.5

11.6 0.8 27.7 26.8 33.1 3.5 0.5 0.4 1.7

62.7 4.8 19.6 12.9

62.1 16.5 21.4

11.5 0.8 28.8 27.3 31.7 3.4 0.5 0.4 1.5

63.5 5.3 18.3 13.0

55.6 19.0 25.4

11.2 1.4 29.1 27.4 31.0 3.2 0.5 0.4 1.2

54.1 13.9 18.1 13.9

53.1 20.7 26.2

542

VAUGHAN 9781786430595 PRINT (M4051).indd 542

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:25

82.9 11.9 5.2

Housing tenure status Owned Rented Provided rent-­free

82.0 13.1 5.0

49.7 58.6 70.9 50.3 60.3 86.4 99.4 99.6 99.8 70.7 66.4 95.2

2007

82.1 12.7 5.3

40.8 49.2 62.4 59.2 55.9 85.0 98.4 99.8 99.7 77.9 56.4 92.7

2011

80.9 13.1 6.1

47.5 54.1 67.9 52.5 54.9 79.9 96.9 98.8 99.4 52.9 55.1 90.4

2003

80.3 14.2 5.5

45.9 54.4 67.1 54.1 56.1 82.8 99.0 99.3 99.7 67.2 63.9 94.1

2007

Total

79.3 14.9 5.8

38.9 46.8 58.1 61.1 51.7 83.7 97.8 99.5 99.6 76.9 53.8 91.1

2011

Source:  Authors’ analysis from Household Budget Surveys 1973–74, 1980–81, 1990–91, Continuous Household Budget Survey 1998–2004 and EU SILC.

52.3 58.9 72.0 47.7 59.1 84.4 98.2 99.5 99.7 56.1 58.0 91.9

2003

Middle class

Average share of members working Average share of members (+16) working Average share of members (16–64) working Average share of economically dependent members Average share of working age women in employment (16–64) Household owns a car Household owns a phone Household owns a washing machine Household owns a TV Household owns a computer Household can go on holiday once a year Household can afford adequate heating in winter

Table 12A.2  (continued)

13. The rise and fall of the Swedish middle class? Dominique Anxo 1. INTRODUCTION Closely linked to the labour movement, the Swedish social democratic political project aimed to develop – through social and institutional reforms – an inclusive and cohesive society. Fundamentally egalitarian, this reformist project aimed also to reduce class inequalities through the implementation of an ambitious redistribution system and a series of measures to promote upward social mobility. From the 1950s up to the mid-­1970s, the political electoral success of the Swedish social democratic movement was intrinsically related to the mobilization of part of the middle class in its political project. That is, the electoral support and inclusion of the middle class in the social democratic political strategy was the condition sine qua non for its political success. On the other hand, and in a dialectic sense, this strategy and the objective of the social democratic system per se relied fundamentally on the implementation of political measures aimed at fostering the growth of the middle class through – among other things – mechanisms encouraging social mobility and reducing income inequalities. In contrast to other welfare state regimes, the Swedish social model has from its inception emphasized the principles of universalism, egalitarianism and individual citizens’ rights. These fundamental components of the Swedish model have been attractive to the middle class because, as stressed by Esping-­Andersen (1990), they have led to equality based on high standards rather than on minimal needs. The incremental development of an encompassing and generous welfare state, an ambitious redistributive and progressive tax and benefit system, a system of industrial relations limiting wage inequalities, and promoting good working conditions, massive investment in education and measures favouring occupational mobility, as well as full employment up to the end of the 1980s, are crucial factors explaining the rise and importance of the Swedish middle class, that remains one of the largest among modern societies, by international standards. However, in the wake of the severe economic crisis of the early 1990s 543

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 543

28/09/2016 08:25

544

Europe’s disappearing middle class?

and the implementation of several institutional and labour market reforms, income inequality has been on the rise in Sweden. The successive reforms of the social protection system, the deregulation of financial markets and the implementation of labour market reforms during the 1990s can partly be attributed to the diffusion of neoliberal ideas within the social democratic movement (‘making work pay’, work incentives, efficiency versus equity, privatization and market deregulation) and to changes in societal norms and attitudes regarding fairness and equity, in particular among the upper middle class. During this period, the labour market reforms – in particular, the reforms of the employment protection system but also the changes in the Swedish industrial relations system (tendency towards a decentralization of collective bargaining) and wage formation (introduction of a performance-­related individualized wage system) – have contributed to the rise of inequality in Sweden. Despite this clear tendency towards increasing inequality and a rolling back of the decommodification of the Swedish social model, the Swedish middle class has remained supportive of the Swedish welfare state and the financing that this involves, entailing a relatively high tax burden (Anxo 2015; Svallfors 2004). This continuous support has several explanations: a relatively efficient public administration and the provision of high-­ quality public services; the importance of the public sector for female employment; and awareness among the still large Swedish middle class that it is no longer exempt from the risk of unemployment in a context of increased globalization, more exposure to market forces and a reduction of employment stability. The objective of this chapter is twofold: to analyse, taking a broad historical perspective, the development over time of the Swedish middle class and to attempt to link this development to the transformations of the Swedish model and the major transformations in the world of work that Sweden has undergone in the past three decades. The chapter is organized as follows. After a description of the evolution of the Swedish middle class during the past three decades, in section 3 we attempt to link the major transformations in the Swedish labour market to changes in income distribution in general and the size of the middle class in particular. Particular attention is paid to the impact of the changes in the employment system – in particular the rise of unemployment and precariousness – on the Swedish middle class. Section 4 analyses the extent to which the long-­ term transformation of the Swedish occupational structure can explain the trends in the size of the Swedish middle class (case study 1). Section 5 analyses the potential impact of the changes in the composition of the Swedish labour force – in particular the substantial inflow of political refugees due to the Yugoslav civil war in the early 1990s – on the development of income

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 544

28/09/2016 08:25



The rise and fall of the Swedish middle class? ­545

inequality and the Swedish class structure (case study 2). The final section provides some concluding remarks.

2. THE SWEDISH MIDDLE CLASS: TRENDS AND EVOLUTION Different approaches have been used to estimate the size of the middle class and its evolution over time either by (1) identifying the middle class as the proportion of households with an equivalized disposable income ranging between two income limits (as a percentage of the median income), in this way defining the middle class as the share of the adult population falling within that range, or (2) defining the middle class as households’ share of total income between the twentieth and the eightieth percentiles1 by the position of individuals in the employment structure, defining the various social classes in terms of occupational prestige. In this context, the middle class or the intermediate class is defined as individuals with routine non-­ manual jobs, or lower-­ grade technicians, supervisors of manual workers and small employers or self-­employed workers (see, for example, Goldthorpe 2010). Following the common methodology of this volume, we follow the income-­based definition of the middle class set out in (1), that is, defining the middle class as the share of the adult population with an equivalized disposable income ranging between 60 and 200 per cent of the median equivalized disposable income. Before presenting our analysis of the long-­term evolution of the Swedish middle class based on this definition, the following section gives a brief overview of the literature on Sweden’s situation relative to other countries. 2.1 Previous Studies: The Swedish Middle Class in an International Perspective Using household income data from the Luxembourg Income Studies (LIS) and defining the middle class as the share of the adult population with an equivalized disposable income ranging between 75 and 125 per cent of median income, Pressman (2007) analysed the development of middle class between 1967 and 2000 in 11 modern economies. According to Pressman, the size of the Swedish middle class increased from 32.5 per cent at the end of the 1960s to 46.9 per cent at the turn of the twenty-­first century, displaying the largest middle class among the selected countries. Using the same dataset and a similar definition of the middle class, Atkinson and Brandolini (2011) analyse the changes in the size of the middle class in 15 countries and extend the period of analysis to 2005. According to

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 545

28/09/2016 08:25

546

Europe’s disappearing middle class?

the authors there is also a strong correlation between the extent of income inequality and the size of the middle class, the countries with the largest middle class displaying the lowest income inequality (as measured by the Gini coefficient). The authors show that Sweden, after Norway, exhibited the largest middle class in 2005 (47.4 per cent of the adult population). Analysing also the development of the middle class between the mid-­1980s and the mid-­2000s, the authors found that the size of the middle class fell in Sweden, the decline ranging between 1.5 to 4.1 percentage points, depending of the level of income cut-­offs for delimiting the middle class. Also worth noting is that this decline has coincided with an increase in the proportion of low-­income households2 (by 0.9 percentage points) and of rich households (by between 0.7 and 3.1 percentage points, depending on the choice of maximum income cut-­offs). In other words, like many other modern economies, Sweden – according to Atkinson and Brandolini – appears to have experienced a decline of its middle class in recent decades and a weak tendency to increased polarization in its class structure. In a more recent study, Dallinger (2013) analyses the development of the middle class in 19 post-­industrial countries during the period 1985–2005, also using LIS data. Contrasting with the two above-­mentioned studies, Dallinger instead defines social classes in terms of quintiles, more precisely as the share of income that individual quintiles have as a proportion of total income.3 Dallinger’s analysis shows that in 2005 the middle class as measured by its income share was largest in Denmark (57.8 per cent) followed by Sweden (56.8 per cent). In all the selected countries, except Denmark and Switzerland, the size of the middle class as measured by its share in total income declined between 1985 and 2005. In summary, Sweden, in spite of a weak decline of its middle class up to 2005, still exhibited the largest middle class among modern, post-­industrial economies, along with its Nordic counterparts, Denmark and Norway. 2.2  Evolution of the Swedish Middle Class, 1987–2012 Using household income data from the LIS for the period 1975 to 20054 and European Union Statistics on Income and Living Conditions (EU-­ SILC, hereafter Eurostat) for the period 2008 to 2012, we define, as previously mentioned, the middle class as the share of the adult population (18 years and older) with an equivalized disposable income ranging between 60 and 200 per cent of equivalized median household disposable income (hdi). We furthermore decompose the middle class into three income groups: lower middle-­income class (60–120 per cent of hdi), core middle-­ income class (120–160 per cent of hdi) and core upper-­middle class (160 to 200 per cent of hdi). Finally, we define the top-­income class as the share

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 546

28/09/2016 08:25



The rise and fall of the Swedish middle class? ­547

of the adult population with an equivalized disposable income above 200 per cent of household median disposable income. Figure 13.1 depicts the evolution of the Swedish class structure between 1987 and 2012. Extending the period of analysis up to 2012 by using EU-­ SILC (Eurostat) for 2008 and 2012, our estimations show that the Swedish middle class has declined by almost 8 percentage points since the late 1980s, confirming the long-­term trend for a decline of the Swedish middle class found in previous studies. As also shown by Figure 13.1’s upper panel, two broad phases could be distinguished: a phase of relative increase of the middle class between 1987 and 1995, the share of the Swedish middle class reaching a maximum in 1995 with 87 per cent of the adult population, followed, despite some fluctuations, by a phase of decline of the middle class after the mid-­1990s and an acceleration of this decline from 2005 onwards. Looking at the period as a whole, the share of the adult population in the lower-­income class grew by more than 5 percentage points, while the top-­income group increased its share by around 2 percentage points, implying that around two-­thirds of the decline of the Swedish middle class might be ascribed to some form of downward mobility and one-­third to upward mobility. As also shown in Figure 13.1 (lower panel) and Table 13.1, most of the decline of the middle class can be ascribed to the decrease of the core middle-­income group, a fall of 9.1 percentage points. Also worth noting is that the fall of the middle-­income group began at the start of the new millennium. The fall of the lower middle class was much more limited, at only 1 percentage point, while the share of the upper-­middle class increased by 2.4 percentage points over the period as a whole. As shown by Table 13A.1 (see the appendix to this chapter), not only has the population share of the middle class decreased but also its income share. Over the period as a whole the decline of the income share of the middle middle-­income group was particularly severe (–11.9 percentage points), while both the lower-­and upper-­income groups have seen their income share increase (by, respectively, +2.7 and + 6.6 percentage points). Drawing on Foster and Wolfson’s (2010) approach, we estimated a polarization index to assess the extent to which the Swedish class structure has experienced increased polarization during the past three decades. As shown by Table 13A.2 in the appendix, the evolution of this index between 1987 and 2012 indicates a slight increase in the level of polarization, confirming the above-­mentioned decrease of the Swedish middle class. In order to evaluate the redistributive impact of the Swedish welfare state we use the same two income categories as Atkinson and Brandolini (2010): factor income (income from labour and capital before tax and transfers) and net disposable income.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 547

28/09/2016 08:25

548

Europe’s disappearing middle class? a) Development of the Swedish class structure, 1987–2012

2012

16.9

2008

78.9

14.4

4.2

81.8

3.8

2005

12.0

84.2

3.8

2000

12.3

82.7

5.0

1995

10.0

87.0

1992

12.1

84.6

1987

11.5

86.6

0

20

40

3.0 3.3 1.9 60

100

80

Lower income class (200% of the median b) Development of the Swedish class structure, decomposition of the middle class in three income groups 2012

15.6

16.9

2008

17.0

14.4

34.2

29.1

4.2

35.8

29.1

3.8

27.5

3.8

2005

12.0

18.0

38.8

2000

12.3

17.1

38.7

1995

10.0

25.3

47.0

14.7

1992

12.1

16.1

1987

11.5

16.6

0

26.9

20

27.8

40.7

26.7

43.3 40

60

80

5.0 3.0 3.3 1.9 100

Lower income class (200% of the median Source:  LIS 1987–2005 and Eurostat for 2008–12 and author’s calculations.

Figure 13.1 Evolution of the middle class, Sweden, 1987–2012, aggregated (upper panel) and decomposed into three income classes (lower panel) (percentage)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 548

28/09/2016 08:25



The rise and fall of the Swedish middle class? ­549

Table 13.1 Variation over time in the share of the five income groups, Sweden, 1987–2012 (percentage points) Incomes groups/periods

Diff. 1987–2005

Lower-­income class, below 60% of median hdi Lower middle-­income class, 60–80% of median hdi Core middle-­income class, 80–120% median hdi Upper middle-­income class 120–200% of median hdi Middle class as a whole, 60–200% of median hdi Upper-­income class, above 200% of median hdi

Diff. Diff. 2005–12 1987–2012

0.5 1.4 −4.5 0.8 −4.8 2.0

4.9 −2.4 −4.6 1.6 −2.9 0.4

5.4 −1.0 −9.1 2.4 −7.7 2.3

Source:  LIS 1987–2005 and Eurostat for 2008–2 and author’s calculations.

100 90 80 70 60 50 40 30 20 10 0

9.6

1.9 26.7

29.9 19.5

14.1

3.0

11.3

25.3 28.2

27.7 43.3

15.8

20.5

47.0

6.9

8.2 32.7

16.6

1987 factor income

1987 disposable income

18.3

14.7

1995 factor income

1995 disposable income

31.4 2005 factor income

29.1

38.8

12.4 5.2

35.8

12.5 4.8

18.0

39.2

17.0

39.3

2008 factor income

2008 disposable income

2005 disposable income

4.2

19.5

29.1

23.8 34.2 15.6 16.9

14.4

12.0

10.0

3.8

25.0

8.6

35.5

11.5

3.8 27.5

Lower income class

Lower middle-income class

Upper middle-income class

Upper income class

2012 factor income

2012 disposable income

Middle middle-income class

Source:  LIS 1987–2005 and Eurostat for 2008–2012 and author’s calculations.

Figure 13.2 Redistributive impact of the welfare state, Sweden, 1987–2012 (percentage) Figure 13.2 shows that in 2012 the size of the middle class would have been much smaller without Sweden’s ambitious redistributive system; by around 41 per cent if we use factor income and by 79 per cent if we use disposable income. Also worth noting is that the Swedish tax and benefit system seems to be, among the middle class, more beneficial for the core middle-­income group and, as expected, less redistributive for the upper-­ middle class, reflecting the progressivity of the Swedish tax system. As also expected, the size of the lower-­and top-­income groups would have

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 549

28/09/2016 08:25

550

Europe’s disappearing middle class?

Table 13.2 Variation in the share of the five income groups measured by factor and disposable income, Sweden, 1987–2012 (in percentage points) Incomes groups/periods

Lower-­income class, below 60% of median hdi Lower middle-­income class, 60–80% of median hdi Core middle-­income class, 80–120% median hdi Upper middle-­income class 120–200% of median hdi Middle class as a whole, 60–200% of median hdi Upper-­income class, above 200% of median hdi

Variation-­ based factor income

Variation based on disposable income

6.6 −3.4 −7.0 −6.1 −16.5 9.9

5.4 −1.0 −9.1 2.4 −7.7 2.3

Source:  LIS 1987–2005 and Eurostat for 2008–12 and author’s calculations.

been much larger without the encompassing redistributive system. As also shown by Figure  13.2, the increase of the bottom-­and top-­income groups over the whole period would have been, as expected, much more pronounced without the prevailing redistributive measures, implying a much more polarized class structure. Conversely, the decline of the middle class over time would have been much more pronounced than the actual fall without redistributive measures (see Table 13.2). That is, the redistributive impact of the Swedish welfare state has significantly improved the situation of the core middle class as well as the lowest-­ income group and cushioned the increase of the top-­ income groups, thereby reducing the increasing tendency to class polarization in Sweden. 2.3 Changes Over Time in the Demographic Composition of the Swedish Middle Class An analysis of the Swedish class structure by age, gender, ethnicity and educational attainment shows that young people (18–24), senior citizens (65+), women, foreign-­born and low-­educated individuals are over-­represented among lower-­income groups and under-­represented among top-­income groups (see Figure 13.3). While the proportion of young people belonging to the lower-­income groups has been decreasing over time, the share of women, migrants and senior citizens has been increasing. There are strong reasons to believe that young people’s over-­representation among poor households is related both to the (longer) transitional phase from the education system to the labour market and also a high and growing rate of youth unemployment. For migrants also, lower employment rates, higher

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 550

28/09/2016 08:25



The rise and fall of the Swedish middle class? ­551 a) Distribution of income class by age groups, 1992–2012 All 2012 Senior 2012 (24.7) Old 2012 (32.1) Prime age 2012 (30.7 ) Youth 2012 (12.5)

16.9 15.6 24.4 9.2 9.1 14.4 13.0 28.5

All 2005 Senior 2005 (22.7) Old 2005 (36.1) Prime age 2005 (34.1) Youth 2005 (7.1)

12.5 18.3 20.1 5.5 10.6 10.1 14.5 35.5

All 1992 Senior 1992 (22.6) Old 1992 (29.7) Prime age 1992 (33.9) Youth 1992 (13.8)

13.5 19.3 5.4 8.2 11.9 6.7

34.2

29.1

27.0 30.2 45.4 15.9

25.1 24.0

29.0 36.9

2.6

28.0

34.2

31.7 42.8

33.2 45.5 25.2

4.2 12.1

27.3 31.5 29.0

37.4 32.1

35.9 48.3

29.8 47.1 38.1

2.0 7.9 2.7 7.8 0.0

3.8 11.6 1.2 8.3 8.3

31.2 32.3

18.6

20

3.2 7.5 2.1

44.0

16.3

0

4.2 17.4

28.0

40

10.8

60

Lower income class Middle middle-income class Upper income class

80

3.1 0.2 100

Lower middle-income class Upper middle-income class

b) Distribution of income class by gender, 1992–2012 All 2012

14.0

Female 2012 (51.4)

12.5

All 2005 Male 2005 (48.6)

10.4

All 1992

13.5

Male 1992 (48.8)

11.9

Female 1992 (51.2)

36.9

18.3

20

27.1

36.0 40

Lower income class Middle middle-income class Upper income class

4.2

31.1

38.8 18.5

28.0

29.0

37.4

16.3 14.0

3.9

25.9

35.2

20.4

60

4.2 4.5

27.5

30.3

38.7

16.1

15.0 0

32.5

16.5

14.5

Female 2005 (51.4)

30.8

36.0

14.7

19.6

29.1

34.2

15.6

16.9

Male 2012 (48.6)

80

4.5 3.9

3.8 4.2 3.5 100

Lower middle-income class Upper middle-income class

Note:  Share of various groups in total population within parentheses. Young people (18–24 years old), prime age (25–54 years old) and seniors (65 years and older). Source:  LIS 1992 and 2005 and Eurostat for 2012 and author’s calculations.

Figure 13.3 Demographic composition of income classes, Sweden, 1992–2012 (percentage)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 551

28/09/2016 08:25

552

Europe’s disappearing middle class? c) Distribution by income class and years, natives and immigrants

Natives 2005 (85.9)

4.2

19.4

33.8

22.8

22.3

Migrants 2005 (14.1)

2.8

4.6

29.5

37.4

17.6

10.9

28.0

36.9

18.3

12.5

All 2005

19.7

29.4

18.8

29.3

Migrants 2012 (13.5)

4.2 4.4

30.6

35.0

15.1

15.0

29.1

34.2

15.6

16.9

All 2012 Natives 2012 (86.5)

1.7

All 1992

13.5

16.3

37.4

29.0

3.8

Natives 1992 (94.8)

13.3

15.9

37.6

29.4

3.9

0

10

20

30

40

Lower income class Middle middle-income class Upper income class

50

60

2.3

22.9

33.2

24.0

17.5

Migrants 1992 (5.2)

70

80

90

100

Lower middle-income class Upper middle-income class

Note:  Share of various groups in total population within parentheses. Young people (18– 24 years old), prime age (25–54 years old) and seniors (65 years and older). Source:  LIS 1992 and 2005 and Eurostat for 2012 and author’s calculations.

Figure 13.3  (continued) unemployment rates and growing difficulties for newly arrived migrants and political refugees to get a foothold on the labour market also explain their larger share within the lower-­income group and the increasing proportion of migrants among poor households in Sweden. The larger share of senior citizens among the lower-­income class is partly related to the reduction of income during the retirement phase for some groups of senior citizens. Finally, the fact that the share of women is higher among lower-­ income groups is principally related to the higher share of part-­timers among female employees and by the gender wage gap. Considering the middle class as a whole, the share of young people among the middle class has been on the rise over time, while the proportion of migrants among the middle class has declined significantly, from around 80 per cent in 1992 to around 68 per cent per cent in 2012 (a fall of 12 percentage points). The corresponding figure for natives for the same period is a decline of around 3 percentage points, from 83 per cent in 1992 to 81 per cent in 2012. Sweden has one of the most educated labour forces among modern economies and has experienced a clear educational upgrade during the past half century. The educational attainment of the Swedish

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 552

28/09/2016 08:25



The rise and fall of the Swedish middle class? ­553

70 60

Compulsory or less

57.6

50 40

33.3

39.6 37.1

43.4 32.8

30

22.7

0

26.4

48.2

20.6

1985

1990

1995

2000

45.7 41.3

34.5

16.9

9.1

1975

Tertiary

47.3

30.2

26.0

17.6

20 10

46.2

Secondary

13.0

2005

2013

Source:  Statistics Sweden (2015b).

Figure 13.4 Trends in educational attainment as a percentage of the adult population, Sweden, 1975–2013 population is significantly higher than in most European member states and Organisation for Economic Co-­operation and Development (OECD) countries. Figure 13.4 depicts the trends in educational attainment of the adult population (25–64 years old) between 1975 and 2013. In 2013, more than 45 per cent of the adult population had upper secondary education and 41 per cent a tertiary education. During the school year 2010 to 2011 almost 45 per cent of young persons aged 19–26 years were enrolled in tertiary education in Sweden or abroad (Swedish Agency for Higher Education 2012). While younger generations are better educated than older generations, it should be noted that Swedish elderly workers (55–64 years old) have, by international standards, a high level of educational attainment, explaining their good record in the labour market with the highest senior employment rate and highest average age of exit in the EU28 (73 per cent and 64 years old). Over the past ten years there has been little change in school enrolment rates in compulsory schooling, while those for upper secondary school have increased significantly. In 2012, 98 per cent of each cohort was enrolled in upper secondary schools. During the late 1980s the university enrolment rate started to increase after having fallen continuously during the previous decade. When the youth labour market deteriorated in the early 1990s the university enrolment rate also increased as a consequence of a significant expansion in the volume of places at public universities. As previously stressed, one of the main features of the development of the youth labour market during the past two decades has been the gradual

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 553

28/09/2016 08:25

554

Europe’s disappearing middle class? All 2012

Upper-income class 1992

68.5 59.1

30.6 33.6

53.3

33.0

13.7

47.2

50.0

2.8

Upper middle-income class 2005 Upper middle-income class 1992

56.3

5.7

14.9

49.6

35.6

28.4

56.2

15.4

Lower middle-income 2005

22.3

58.0

19.7

Middle middle-income class 1992 Lower middle-income 2012

29.5 28.4

44.1

26.4

Middle middle-income class 2005

35.8

49.1

15.1

Middle middle-income class 2012

19.8

46.0

34.3

Upper-income class 2012 1.0 Upper-income class 2005 7.3

26.9

52.8

20.3

All 1992

Upper middle-income class 2012

39.3

52.3

8.4

All 2005

17.2

53.3

29.6

Lower middle-income 1992 Lower-income class 2012

20 Compulsory or less

9.8

47.9

42.3 0

16.8

52.5

30.7

Lower-income class 1992

31.0

50.0

18.9

Lower-income class 2005

8.7

42.4

48.9

40

60 Secondary education

80

100

Tertiary education

Source:  LIS 1992–2005 and Eurostat for 2012 author’s calculations for educational attainment by income classes.

Figure 13.5 Educational attainment trends by income groups, Sweden, 1992, 2005 and 2012 (percentage) postponement of entry into the labour market and a longer transitional phase from the education system to the labour market.5 Another characteristic of the Swedish education system is the considerable opportunity it provides to complete educational attainment through adult education. Students aged 20 years and above who fail their upper secondary school examinations have the opportunity to go back to school within the extensive adult educational lifelong learning system. As shown by Figure 13.5, the share of individuals with low educational attainment has been decreasing over the past two decades in each of our income classes, in particular among the lower-­income groups. Conversely, the share of highly educated individuals has been on the rise within each income class but, notably, relatively less in our core middle-­income class which, as stressed previously, has seen its size decreasing the most in recent decades. In order to assess the extent to which the transformation in the Swedish class structure was due to changes in the composition of the Swedish adult population we performed a shift share analysis for 1992 and 2012 (see Table 13.3). As shown by Table 13.3, the modifications over time in the age or gender

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 554

28/09/2016 08:25



The rise and fall of the Swedish middle class? ­555

Table 13.3 Impact of the changes in the composition of the population on the class structure, Sweden, 1992 and 2012 Lower-­ income class

Actual With constant age distribution With constant gender distribution With constant ethnic distribution With constant educational distribution

3.4 3.5 3.4 2.3 11.9

Middle-­income Top-­income class class

−3.8 −3.7 −3.7 −2.7 −10.4

0.4 0.5 0.4 0.5 −1.5

Source:  LIS 1987–2005 and Eurostat for 2008–12 and author’s calculations.

distribution can hardly explain the change in the Swedish class structure. The fact that modifications in the age composition of the population had limited effect does not mean that age or labour market behaviour and changes in the timing of transition over the life course between cohorts did not affect the class structure. The postponement of entry into the labour market, linked to the expansion of education in Sweden as well as the postponement of exit at the end of the job career, should affect both factor and disposable income. The later entry should imply over time a reduction of labour earnings for the current generation compared with previous generations who entered the labour market much earlier and conversely an increase of income among senior citizens working longer in comparison with previous cohorts. As also shown by Table 13.3, the ethnic and educational composition of the population has had a strong effect on the Swedish class structure. With a constant distribution of the population between natives and migrants, the proportion of individuals belonging to the lower-­income groups and the middle class should have been 1 per cent higher in 2012, implying that around one-­third of the decline in the middle class could be ascribed to change in the ethnic composition of the Swedish population. The impact of the changes in the distribution of educational attainment is significantly larger: the size of lower-­income groups should have been almost 8 ­percentage points higher and the middle class almost 7 ­percentage points lower if the distribution of education had been the same as in 1992. Owing to the importance of these two dimensions for the evolution of the Swedish class structure our two case studies (sections 4.1 and 4.2) will concentrate on changes in the skill structure and ethnic composition of the Swedish labour force. Before that, the next section describes the major transformation that the Swedish labour market has undergone during the past three decades.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 555

28/09/2016 08:25

556

Europe’s disappearing middle class?

3. MAJOR TRANSFORMATIONS IN THE WORLD OF WORK AND THEIR IMPACT ON THE MIDDLE CLASS 3.1 Changes in the Employment System: The Decline of Employment Rates and the Rise of Unemployment By any international standard, up to the 1992 crisis Sweden was remarkably successful in combining low unemployment – ranging between 1 and 3 per cent between the mid-­1960s and the end of the 1980s – not only with high and growing employment rates but limited gender disparities (see Figure  13.6, upper panel). In the wake of the deep recession of the early 1990s, the employment crisis became dramatic. In just three years – from 1990 to 1993 – the employment rate decreased by more than 10.5 ­percentage points to about 73 per cent, and the rate of open unemployment quintupled from less than 2 per cent to almost 10 per cent of the labour force. The employment rate started to grow again from the mid-­ 1990s onwards but never reached its 1980s level. The decline of employment rates and the inability of the Swedish economy since the early 1990s to preserve full employment is related to the 1990s’ fiscal consolidation measures and a re-­orientation of fiscal and monetary policy, giving priority to fighting inflation and budget balance (see Anxo 2013, 2015). Other factors might explain the decline of employment rates, such as the major investment in education during the 1990s,6 but it is clear that the ‘political acceptance’ of a higher ‘equilibrium rate of unemployment’ reflects the re-­ orientation of macroeconomic policy and not only a structural imbalance in the labour market and/or turbulence in the world economy. The fall in the youth employment rate has been particularly marked, from around 65 per cent at the end of the 1980s to less than 43 per cent in 2014. This is due to the above-­mentioned large investment in education that has entailed a lengthening of the education phase for young people and a postponement of entry into the labour market. By contrast, the employment rate of older workers (55–64 years old) has had a clear tendency to increase during the past decade, from around 64 per cent at the turn of the century to 74 per cent in 2014. The associated increase in the actual age of exit (64.5 years old in 2010), reflecting the postponement of exit at the end of the working career, has implied a lengthening of working life. Part of this evolution and change of labour market behaviour on the part of older and senior citizens is related to the comprehensive pension reforms initiated in the mid-­1990s and implemented at the turn of the century, which embodied strong de jure and de facto economic incentives to work longer.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 556

28/09/2016 08:25



The rise and fall of the Swedish middle class? ­557 Employment rates

90.0

Men

Women

85.0 77.7

80.0 75.0

74.2

70.0 65.0

1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

60.0

Unemployment rates 30 25

All (16–64 years old)

Youth 16–24 years old

22.9

20 15 8

10

0

1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

5

Source:  Statistics Sweden (2015b).

Figure 13.6 Trends in employment and unemployment rates by gender, Sweden, 1987–2014 (percentage of labour force) The employment prospects of young people have worsened significantly during the past two decades, particularly in connection with the two economic downturns of the early 1990s and the financial crisis of 2008. Youth unemployment rose from around 5 per cent at the end of 1991 to around 23 per cent in 2014 (see Figure 13.6, lower panel). The youth population does not constitute a homogeneous group and a successful transition from the education system to the labour market is, in Sweden as in other industrial countries, clearly dependent on educational attainment. The tendency to an increase in youth unemployment might be ascribed to several factors. Sweden is characterized by a combination of a relatively strict labour market regulation regime and relatively weak linkages between the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 557

28/09/2016 08:25

558

Europe’s disappearing middle class?

vocation-­oriented education system and the labour market, implying that the ‘barriers’ to labour market entry of young people, especially less educated young people, are relatively high compared with other industrialized countries, such as Germany.7 The reformulation of economic policy – that is, the priority given to fighting inflation and fiscal consolidation measures initiated in the mid-­ 1990s – reflects a weakening of the political commitment to the goal of full employment, as illustrated by the above-­mentioned decline in employment rates and also by unemployment rates well above those experienced during the Golden Age of the Swedish model (see Anxo 2013, 2015). As shown by Figure 13.6 (lower panel), the unemployment rate never returned to the level experienced before the 1990s economic downturn and in 2014 is still at a rate of 8 per cent, clearly above its long-­term equilibrium level. As shown by Figure 13.7, the proportion of people who are out of the labour force and unemployed is, as expected, higher among the lower-­ income class and lowest among the top-­income groups. Between 2005 and 2012, despite the increase in overall unemployment, the unemployment rate by income class has remained constant, at around 3 or 4 per  cent within the middle class and around 8 per cent among the low-­income group. However, the most prominent change during the period is the significant increase in the proportion of retirees at the two ends of the income distribution (11.3 and 7.7 percentage points, respectively). Thus, 100 90 80

31.3

37.7

70 60 50

33.5

40 30 20 10 0

8.3 27.0 Lowerincome class 2005

31.5

22.3

25.0

62.1

65.4

65.3

3.9

2.9

8.7

13.0

16.4

82.2

78.2

77.6

42.6

29.1

8.0

8.5

22.8

19.8

Lowerincome class 2008

22.2

Lowerincome class 2012

11.9

Out of the labour force

Unemployed

1.2

1.7

0.8

7.9 Topincome class 2005

7.2 Topincome class 2008

5.2 Topincome class 2012

3.4

9.4 6.3 Middle- Middle- Middleincome income income class class class 2005 2008 2012

Employed

Retired

Source:  LIS for 2005 and EU SILC for 2018 and 2012 and author’s calculations.

Figure 13.7 Labour force status by income class, Sweden, 2005–12 (percentage)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 558

28/09/2016 08:25



The rise and fall of the Swedish middle class? ­559

Table 13.4 Variation of the population share by income classes and labour force status, Sweden, 2005–12 (percentage points) Lower-­income class Employed Unemployed Retired Out of the labour force Total

1.8 12.3 10.3 16.3 4.9

Middle-­income class −1.5 −11.9 −11.2 −16.3 −5.3

Upper-­income class −0.3 −0.3 0.9 0.1 0.4

Source:  LIS for 2005 and EU SILC for 2008 and 2012 and author’s calculations.

the increase in class polarization in Sweden seems to have resulted mainly from the reduction of income among some retirees, mainly women and immigrants with lower labour market attachment across the life course. Recall that the pension reform implemented at the turn of the century implied a reduction of pension benefits, in particular for those with more erratic employment histories. There are therefore reasons to think that part of the decline of the middle class in Sweden can be ascribed to some form of downward mobility among the population of retirees. As shown by Table 13.4, the proportion of individuals belonging to the middle-­income groups was reduced between 2005 and 2012 by 5.4  ­percentage points. The share of individuals belonging to the lower-­ income class increased by 4.9 percentage points, while the proportion of individuals belonging to the upper-­income class increased slightly. Worth noticing is the strong fall in the share of unemployed, retired and persons out of the labour force among the middle-­income class and the corresponding increase in the lower-­income groups. There are therefore strong reasons to believe that during the past decade a part of the decline of the middle class can be ascribed to some form of downward mobility linked to a worsening of employment prospects for some vulnerable groups, but also to reforms of the social protection system, in particular the pension system and unemployment benefits, where more restrictive eligibility criteria and lower-­income replacement rates were introduced (see Anxo 2015). In summary, several factors have contributed to the relative decline of the middle class and the increased polarization of the class structure in Sweden: a reduction of employment rates among young people due to the postponement of entry into the labour market, a worsening of the employment prospects of low-­educated young people and low-­skilled workers, and an increase in the share of retired people among low-­income groups. The postponement of entry into the labour market has, in any case in the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 559

28/09/2016 08:25

560

Europe’s disappearing middle class?

short run, resulted in reduced earnings capacity for the younger cohort and contributed to a transitory downgrading of their position in the class structure, which also explains the relative decline of the middle class. For a large proportion of these young people, this downgrading is temporary and their reduced earnings capacity limited to the longer transition from school to work. We may assume that their investment in education and training will normally entail upward mobility later on in life. On the other hand, the worsening employment prospects of low-­skilled young people and the scarring effect of early periods of unemployment on subsequent earnings and future employment trajectory might have more long-­lasting negative consequences and strengthen class polarization in Sweden, even in the long run. There are also reasons to believe that the decrease in income of some retired people in the wake of the pension reforms has contributed to the decline of the middle class. By linking the income replacement rate to lifetime earnings this reform has penalized individuals with less continuous, more erratic employment trajectories, shorter employment spells and shorter working time across the life course. 3.2  Increased Labour Market Duality: The Rise of Short-­term Contracts The successive reforms of the employment protection system in Sweden, in particular the introduction of short-­term contracts not requiring justification from the employer, combined with the deregulation of employment intermediation in the early 1990s, has contributed to an increase of employment instability and growing duality in the labour market between insiders and outsiders. This growth of fixed-­term contracts may also have caused some rise in unemployment through higher labour turnover. International comparisons of employment protection indicate that Sweden has fairly restrictive legislation, although it does not stand out as extreme by European standards. Sweden combines stringent regulation of open-­ended contracts with relatively weak regulation of fixed-­term contracts and temporary agency work. In fact, Sweden exhibits among the strongest duality and the highest disparity between the stringency of regulation for open-­ended contracts and temporary jobs among OECD countries. The use of temporary agency work (TAW) can also be described as liberal, as the current legislation does not restrict TAW to certain areas of the labour market. In the aftermath of the 1990s crisis, Swedish companies have to a larger extent than before made use of external numerical flexibility (Anxo 2011). Against this background, short-­term contracts and agency work have increased significantly. The share of fixed-­term contracts rose by almost 40 per cent between 1990 and 1995. Since then, and up to the current

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 560

28/09/2016 08:25



The rise and fall of the Swedish middle class? ­561

18 17 16 15 14 13 12 11 10 9 2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

1994

1992

1993

1991

1990

1989

1988

1987

8

Source:  Statistics Sweden (2015b) and author’s calculations.

Figure 13.8 Trends in the share of fixed-term contracts among dependent employees, Sweden, 1987–2014 (percentage of dependent employees) recession, this share has continued to increase, reaching 17 per cent of total waged employment at the onset of the current recession (see Figure 13.8). In the wake of the 2008 crisis the share of short-­term contracts declined, but started growing again in 2010, almost reaching its pre-­crisis level in 2014. The share of short-­term contracts is also significantly higher among young people and low-­skilled workers. In 2012, half of young employees aged 20–24 years of age were on short-­term contracts compared with less than 10 per cent among prime age adults. As shown by Figure 13.9, the share of fixed-­term contracts is significantly higher among the lower-­income class, in which almost half of dependent employees are employed on fixed-­term contracts, whereas it is lowest among top-­income employees. This is not surprising owing to the higher incidence of short-­term contracts among young people and low-­educated or low-­skilled workers. Also note that the proportion of fixed-­term contracts has been increasing most in the lower tail of the income distribution. Among the middle class, the share of fixed-­term contracts is higher among the lower middle-­income class and it has increased more in this income group than in other middle-­ class income groups, where it has been more constant.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 561

28/09/2016 08:25

562 50 45 40 35 30 25 20 15 10 5 0

Europe’s disappearing middle class? 43.6

2005

38.3 37.1

2008 2012 18.5

22.4 23.3 10.1 9.9 10.8

Lower-income class

5.3 7.1 6.4

Lower-middle Middle-middle Upper-middle income class income class income class

10.6

15.5 17.3

3.9 3.9 4.3 Upper-income class

All

Source:  LIS for 2005, Eurostat for 2008 and 2012 and author’s calculations.

Figure 13.9 Fixed-term contacts by income groups, Sweden, 2005, 2007 and 2012 (percentage) 3.3 Rising Income Inequalities and Changes in the Industrial Relations System and Wage Setting Against the backdrop of the severe economic crisis of the early 1990s and the implementation of several structural reforms, income inequality has been on the rise in Sweden. While it is true that these reforms had the aim of enhancing work incentives and securing the long-­term sustainability of the social protection system, it is also clear that they strengthened the role of the market, reduced the degree of decommodification of the Swedish welfare state and contributed to an increase in income inequality in Sweden (see Anxo 2015 and Figure 13.10). While the reduction of the generosity of the social protection system and reform of taxation – in particular, capital taxation and the growth of the financial markets8 – might explain part of the rise in income inequality, other factors might also have contributed to the widening of income distribution during the 1990s, such as the changes in industrial and wage setting systems. Despite a decline in union density during the past decade, the two sides of industry remain the main actors regarding both labour market regulations and wage formation in Sweden.9 Since the breakdown of centralized bargaining in 1983, Sweden has witnessed a clear tendency towards decentralization of wage setting. This tendency was accompanied in both the public and private sectors by a marked trend towards differentiation and individualization of wages and terms and conditions of employment. There are strong reasons to believe that this tendency is related to deep societal changes affecting norms and attitudes concerning fairness. The reinforcement of individualistic values, in particular among individuals

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 562

28/09/2016 08:25



The rise and fall of the Swedish middle class? ­563

0.315

0.30

0.295

0.28

0.275 0.255

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

Gini coefficient with capital gains 1997

1991

0.195

Gini coefficient without capital gains 0.21 1996

0.215

0.23

1995

0.235

Source:  Statistics Sweden (2015a).

Figure 13.10 Trends in income inequalities (equivalized disposable income), Gini coefficient (lower panel), Sweden, 1991–2013 belonging to the upper middle class, explains the weakening of the ‘solidaristic’ wage policy (same wage for the same work), where the setting of wages was based primarily on job content and tasks. These normative and societal changes can be one element explaining the evolution towards a wage-­setting system based on individual performance. The long-­term decline in low-­skilled jobs and the concomitant increase in demand for high-­skilled jobs, combined with the higher return to education witnessed during the past two decades (see section 4), help to explain this increase in wage inequalities, because wage formation among high-­skilled occupations is more individualized and differentiated. The increase in income inequality can partly be ascribed to the weakening of the bargaining position of trade unions and partly to the decline of union density, but also to the deterioration of the situation in the labour market, with high and growing unemployment. The shift towards decentralized bargaining and individualization has been more pronounced among white-­collar and high-­skilled workers than among manual workers and low-­skilled employees, in both the public and private sectors. There are strong reasons to believe that the tendency towards more decentralized and individualized wage setting has increased wage dispersion, particularly for high-­skilled white-­collar workers. If the reduction in the share of the middle class and the corresponding increase in the share of lower-­and upper-­income groups explains the increased income inequality and polarization of Swedish society, the consequences of this increase are much more difficult to assess. However, the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 563

28/09/2016 08:25

564

Europe’s disappearing middle class?

decline of the demand for medium-­and low-­skilled jobs (see next section), and the increase of relative wages for high-­skilled workers might have contributed to this development.

4. CASE STUDY 1: LONG-­TERM TRANSFORMATION IN THE SWEDISH EMPLOYMENT STRUCTURE: SOME IMPLICATIONS FOR THE MIDDLE CLASS 4.1  Changes in Employment Structure Like many modern societies, Sweden has undergone large structural changes during the past half-­century. The long-­term evolution of employment by broad sectors has entailed a large decline of employment in agriculture and manufacturing and a significant increase of employment in the services sector (see Figure 13.11). As shown by Figure 13.12, the distribution of employment by broad sectors among middle-­class people has exhibited the same tendency as in Swedish society as a whole. The share of middle-­class employed persons has increased significantly within the services sector and declined in agriculture and in manufacturing. 4.2  Changes in the Skill Structure The shift from a traditional industrial society towards a modern welfare state and knowledge-­intensive service economy in Sweden has also had important implications for the development of the skills and occupational structure. Following Goldthorpe and McKnight (2006) and Kuypers and Marx (Chapter 3 in this volume), we use International Standard Classification of Occupations (ISCO) occupational categories (see Table 13A.3 in the appendix) to assess the trends in the Swedish class structure. As shown by Table 13.5 and contrasting with the income cut-­ off approach, we found a lower decline of the middle class of –2.1 percentage points (compared with –7.7 percentage points when we use the income-­ based definition of the middle class). Also note that when we use an occupation-­based definition of social stratification, the decline of the middle class is overwhelmingly due to a decrease of the lower-­income class, that is, low-­skilled and low paid-­occupations, and a significant increase of the upper middle-­and top-­income class (high-­skilled occupations). This discrepancy between the two approaches might be ascribed to the fact that

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 564

28/09/2016 08:25



The rise and fall of the Swedish middle class? ­565 1975

2012 Agriculture 2%

Agriculture 6%

Construction 7%

Manufacturing industry 29%

Service 57%

Manufacturing industry 14%

Service 77%

Construction 8%

Agriculture Construction Total employment

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

1994

1993

1990

1985

Trends of employment by broad sectors, Sweden, 1975 = 100

1980

1975

180 160 140 120 100 80 60 40 20 0

Manufacturing Services

Source:  Statistics Sweden (2015b) and author’s calculations.

Figure 13.11 Share of employment by broad sectors, Sweden, 1975 (upper left panel) and 2012 (upper right panel), as well as trends of employment by broad sectors (lower panel) the occupational approach restricts the sample to employed persons, while the income-­cut approach includes the whole population (including inactive and unemployed persons). The observed decline of the middle class in Sweden and the increased polarization of its class structure when using the income cut-­off approach is therefore not primarily due to major changes in the world of work but to a deterioration of income of the unemployed (and out of the labour force); that is, a worsening of disposable income among both young people (entry, longer transitional phase from school to work) and the unemployed or those out of the labour force (retirees, students). The observed skill upgrading of the employed population has

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 565

28/09/2016 08:25

566

Europe’s disappearing middle class?

2012 3.6

21.6

74.9

2011 3.8

21.2

75.1

2008 3.1

23.5

2005 1.0

73.4

24.0

2000 1.3

75.1

25.9

72.8

1992 3.4

41.5

1987 3.4

55.1

43.5

1981

5.2

1975

5.1

53.2

44.5

50.3

53.0

0

20

41.8 40

Agriculture, forest, fisheries

60

100

80

Manufacturing industries

Service sector

Source:  LIS 1987–2005 and Eurostat for 2008–12 and author’s calculations.

Figure 13.12 Distribution of employment by sectors of the economy among the middle class, 20–64 year olds, Sweden, 1975–2012 (percentage) Table 13.5 Trends in the population share of social classes based on Goldthorpe’s occupational groups, Sweden, 1987–2012

Lower Lower middle Core middle Upper middle Upper Middle all

1987

1992

1995

2001

2005

2008

2012 Change in percentage points

25.0 21.7 19.8 18.6 14.9 60.1

23.1 19.1 20.0 22.3 15.6 61.4

20.7 18.8 20.5 22.8 17.2 62.1

16.6 20.0 20.5 19.5 23.3 60.0

16.2 19.9 20.1 19.9 23.4 59.9

17.7 19.1 19.5 19.7 24.0 58.3

17.1 17.3 19.5 21.2 25.0 58.0

−7.9 −4.4 −0.3 +2.6 +10.1 −2.1

Source:  Author’s calculations based on LIS (1987–2005, except 2001), Eurostat (2008–12) and Statistics Sweden (2015b) for 2001.

implied a shift of the employed population towards the upper tail of the income distribution (upper-­middle class and upper class). In other words, the decline of the Swedish middle class appears to be due mainly to an upward shift of the employed population from medium-­skilled towards high-­skilled occupations.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 566

28/09/2016 08:25



The rise and fall of the Swedish middle class? ­567

Following Fernández-­Macías et al. (2012) and in order to take into account the differences in the ‘quality of jobs’ we use a wage-­ based measure to rank existing jobs (job being defined as an occupation in an industry). More precisely, we estimate in each period a full-­time equivalent mean hourly wage for each job. We thereafter allocate the jobs into quintiles based on the wage ranking, where the lowest paid job is assigned to the first quintile and the best-­paid job to the fifth quintile. In order to assess the extent to which the change of employment is concentrated at the top, middle or bottom part of the occupational structure we sum up the net employment change between the starting and concluding year in each period. The following quintile charts give a graphical representation of the modification in the distribution of jobs and employment structure between 1985 and 2011, distinguishing between periods of job creation (Figure  13.13, upper panel) and job destruction (Figure 13.13, lower panel). In contrast with other advanced economies, such as the United States, the United Kingdom and Germany, which have faced a polarization of their job structure (an increase of both low-­skilled and high-­skilled jobs and a destruction of jobs in the middle layers of the occupational structure, see Eurofound 2015), during the past half century Sweden has experienced a consistent and pervasive pattern of skill upgrading over time. This pattern of occupational upgrading took the form of a shift of aggregate employment from the lower tail of the job distribution to the top, with an absolute decline of low-­paid low-­skilled jobs and an absolute increase of high-­skilled jobs. This has also been particularly marked during recessions, especially in the dramatic economic downturn of the early 1990s that resulted in a massive destruction of jobs in the middle and lower tail of job distribution and a weak increase of jobs in the upper tail of the occupational structure. This pattern of job upgrading has also characterized the current economic crisis, as shown by Figure 13.13 (lower panel), with a clear destruction of jobs in the lower tail but also by a significant increase of high-­skilled, high-­paid jobs. Looking now at the development of the occupational structure by broad sectors, the process of deindustrialization in Sweden, which has brought with it a continuous decline of employment in manufacturing after the mid-­1970s, has contributed positively to the overall job upgrading by a massive destruction of jobs in the middle layer and bottom of the job distribution and a weak increase of high-­skilled jobs. This pattern of job creation or destruction in manufacturing was particularly obvious during the last two deep recessions (early 1990s and the 2008 financial crisis, see Figure 13.14 lower panel). Like many other advanced economies, the Swedish post-­industrial society has been characterized by an increase of

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 567

28/09/2016 08:25

568

Europe’s disappearing middle class? Absolute changes of employment by quintiles, Sweden, 1997–2002

Absolute change of employment by quintiles, Sweden, 1985–1990 100 000

90 385

38 102

40 000 12 769

20 000

–20 000

45 602

Q1

Q2

Q3

Q4

Q5

Number of workers

Number of workers

120 000

60 000

0

–40 000 –60 000

100 000 80 000

0

8 125

4 427

Q1

Q2

Q3

Q4

70 000 16 612

20 000 Q2

Q3

–40 000 –60 000

–60 340

–80 000 –100 000

–106 963

–120 000 –133 155

–160 000

Q4

Q5

50 000 Number of workers

Q1

61 601

60 000

3 712

0

Q5

Absolute changes of employment by quintiles, Sweden, 2008–11

40 000

Number of workers

44 749

40 000

Absolute changes of employment by quintiles, Sweden, 1991–96

–140 000

67 834

60 000

20 000

–55 900

–80 000

–20 000

132 698

140 000

80 000

40 000 30 000 17 842

20 000 10 000 0 –10 000 –20 000 –30 000

Q1

Q2

Q3

Q4

Q5

–9 655 –3 935 –20 301

Source:  Anxo (2014).

Figure 13.13 Long-term changes in occupational structure, period of job creation (upper panel) and period of job destruction (lower panel). Average annual absolute change in job-wage quintiles, Sweden, 1985–2011 employment in the services sector. Two distinct periods should be distinguished, however. The first was characterized by employment growth mainly in the public sector (1975–91) (see Figure 13.15). From the early 1970s until the early 1990s the creation and development of a modern and encompassing welfare state and the related expansion of public sector employment in Sweden not only contributed to the strong feminization of the labour force and the increase of the overall employment rate, but also to the clear upgrading of the job structure in Sweden. The second period, from 1992 onwards, was characterized by a decline

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 568

28/09/2016 08:25

569

VAUGHAN 9781786430595 PRINT (M4051).indd 569

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:25

240

923

31 648 25 420

Q1

Q2

Q3

Q4

–1 984 –1 761 –7 788 –2 302 –5 902 –19 752 –12 388 –30 493

–4 730

–925

24 843

45 525

Q5

7 550 520

8 879

73 436

–138

Q2

Q3

–51 765

Q4

Q5

275

Q1

–18 971

624

26 196 14 518

16 362 7 874 8 340

Q2

Q3

Q4

Q5

–173 –3 630 –542 –2 056 –403 –10 451 –10 005 –22 175

902

40 449

82 538

128 161

–40 000

–20 000

0

20 000

40 000

60 000

–3 410 491 349

Q1

Q2

–23 231

–5 308 –13 366

1 784

12 735

Q3

–15 821

–339

8 466 3 760

Q4

–5 351

–97 –1 252

Q5

2 728 1 507 492

24 542

56 874

Absolute changes in employment by quintiles and industries, Sweden, 2008–11

140 000 120 000 100 000 80 000 60 000 40 000 20 000 0 –20 000 –40 000

Absolute changes in employment by quintiles and industries, Sweden, 1997–2002

Services

Construction

Manufacturing Ind

Agriculture

Figure 13.14 Long-term changes in the occupational structure by industries, period of job creation (upper panel) and period of job destruction (lower panel). Average annual absolute change in job-wage quintiles, Sweden, 1985–2011

Q1

21 858 7 411

–4 195 –371 –1 093 –1 657 –3 874 –12 281 –24 722 –11 565 –6 983 –16 714 –36 827 –27 549

–94 099

–33 531

–5 387

Source:  Anxo (2014).

–100 000

–80 000

–60 000

–40 000

–20 000

0

20 000

23 346

Absolute changes in employment by quintiles and industries, Sweden, 1991–96

40 000

–40 000

–20 000

0

20 000

40 000

60 000

80 000

Absolute changes in employment by quintiles and industries, Sweden, 1985–90

570

Europe’s disappearing middle class?

3000 2500

(000s)

2000 1500 1000 Public sector

500

Private sector

1963 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011

0

Note:  Dashed vertical lines, 1992 and 2008 recessions. Source:  Statistics Sweden (2015b) and author’s calculations.

Figure 13.15 Trends in employment in the private and public sector, Sweden, 1963–2011 of employment in the public sector and a significant increase of employment in the private services sector. The decline of employment in the public sector as a whole during the past decade – a reduction of 291 000 employees between 1990 and 2008 or 19.6 per cent – can be ascribed to early fiscal consolidation during the first half of the 1990s, the waves of liberalization or privatization, increased outsourcing via public procurement, as well as rationalization and downsizing processes initiated in the aftermath of the deep economic crisis of the early 1990s (see Anxo 2015). Paradoxically, the massive reduction of public sector employment during the 1990s had an upgrading impact since most of the employment destruction concerned the lower tail of the occupational structure. Taking a gender perspective, female employment in Sweden expanded at least as much as male employment in the top job-­quality quintiles with almost no expansion of female employment in the lowest quintiles. Globally, our analysis of the transformation of the Swedish skill structure shows that the expansion of female employment in Sweden has clearly contributed to the upgrading of the occupational structure, female employment being strongly associated with the expansion of high-­quality jobs (see Anxo 2014). The evolution of male employment, on the other hand, shows a decline in male employment in manufacturing and construction, and a pattern of job destruction concentrated principally at the bottom end of the job distribution (low-­skilled or low-­paid jobs). Job destruction in the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 570

28/09/2016 08:25



The rise and fall of the Swedish middle class? ­571

male-­dominated sectors of manufacturing and construction was particularly severe during the early 1990s deep recession. The process of deindustrialization and the move towards a post-­industrial society in Sweden resulted, therefore, in a destruction of low-­skilled, low paid jobs and a contraction of jobs in the middle layers of the industrial occupational structure. This destruction was compensated by a significant increase of high-­skilled, high paid jobs in the growing service sector, resulting in a clear pattern of overall skill upgrading in Sweden during the past 50 years. Focusing on the implications of these long-­term transformations of the occupational structure for Swedish class structure, Figure 13.16 displays the evolution of the skill distribution by income groups. Not surprisingly, the share of individuals with elementary occupations is much higher among low-­income groups compared with other income groups. However, the relative share of elementary occupations has been diminishing in all income groups, reflecting the above-­described trends in the Swedish occupational structure. Furthermore, the share of middle skilled occupations has been decreasing within the middle-­and upper-­ income classes, while the proportion of high-­skilled workers has been increasing significantly in these two income groups. That is, the overall upgrading of the skill structure has principally benefited the Swedish income and upper-­ middle middle class, particularly the core middle-­ class, implying a shift of the skill distribution to the right over time. Also worth noting is the significant increase in jobs in the middle layer of the skill distribution in the low-­income groups. That is, Sweden has experienced a shift from low to middle skill jobs in the lower tail of the income distribution. Focusing our attention on the transformation over time of the occupational structure within the middle-­income class groups (see Figure 13A.1 in the appendix), we see that the tendency of skill upgrading of the occupational structure of the middle class has been particularly marked among the upper-­middle class. The deep recession in the early 1990s and to a lesser extent the 2008 economic downturn also entailed a significant diminution, both in absolute and relative terms, of jobs in the lower tail and middle part of the job distribution and a clear increase of jobs in the upper tail of the distribution. If we assume that individuals in the middle class have jobs in the middle layer of the occupational distribution, the pattern of job destruction during the two large recessions partly explains the relative fall of the middle class. The development of a modern welfare state, the expansion of the public sector and the related feminization of the labour force during the 1970s and 1980s did not lead to increased job polarization but, on the contrary, contributed significantly to the process of skill upgrading in Sweden and went hand in hand with the development of the

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 571

28/09/2016 08:25

572

Europe’s disappearing middle class? a) Lower-income class, 20–64 years old, 1975–2012

2012

22.3

66.3

11.4

2011

18.5

2008

18.5

68.4

13.1

2005

19.0

67.0

14.0

1995

68.2

26.6

1992

45.3

36.4

1975

35.1

31.6

32.7 0

30.1

31.0

33.9

1981

28.1

39.8

30.1

1987

13.2

32.1 35.1

32.2 20

40

Elementary occupations

60 Medium-skilled workers

80

100

High-skilled workers

b) Middle-income class, 20–64 years old, 1975–2012 2012

12.5

28.4

59.2

2011

14.5

59.7

25.8

2008

14.8

60.3

24.9

2005

14.7

1995

15.7

1992

17.5

1987

18.2

1981

23.1

62.2

10

12.6

69.2

12.2

68.3 67.3

22.1 0

13.2

69.3

19.5

1975

15.4

68.9

20

30

Elementary occupations

40

50

10.6 60

Medium-skilled workers

70

80

90

100

90

100

High-skilled workers

c) Upper-income class, 20–64 years old, 1975–2012 2012

1.0 2.4

2011 2008

0.5

2005

0.8

1995

1.6

1992

1.4

1987

1.2

1981

1.9

1975

1.9 0

38.2

60.7

35.7

61.9 60.3

39.2

64.8

34.3

49.5

48.9

50.8

47.8

57.6

41.2

50.7

47.4

53.4

44.7 10

20

30

Elementary occupations

40

50

60

Medium-skilled workers

70

80

High-skilled workers

Source:  LIS 1975–2005 and Eurostat for 2008–12 and author’s calculations.

Figure 13.16 Skill distribution by income class, Sweden, 1975–2012 (percentage)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 572

28/09/2016 08:25



The rise and fall of the Swedish middle class? ­573

middle class. Our approach makes it possible also to analyse the modifications of the employment structure by better differentiating technological and institutional factors from short-­term cyclical factors. Sweden is a small open economy strongly exposed to international competition. Despite the above-­mentioned tendency towards rising wage inequalities, the relatively centralized and coordinated collective bargaining system and a still compressed wage structure have prevented the development of a low-­paid or low-­skilled sector in Sweden and instead have boosted policies favouring an upgrading of skills. In effect, the large investment in research and development, a well-­developed lifelong learning and training system, the above-­ described expansion of tertiary education during the past three decades, as well as more balanced bargaining power between the two sides of industry have limited the tendency towards job polarization that can be found, for example, in liberal market-­orientated welfare states and have promoted the development of a high-­skilled knowledge-­intensive service economy.

5. CASE STUDY 2: CHANGES IN THE ETHNIC COMPOSITION OF THE LABOUR FORCE – IMPACT ON INEQUALITY AND CLASS STRUCTURE Although the issue of the labour market integration of foreign nationals has been high on the list of government priorities, the situation of immigrants in the Swedish labour market has not improved significantly during the past two decades. Even though Swedish employment rates are, by international standards, among the highest in the EU,10 newly arrived migrants and people with low skills or low educational attainment exhibit a worse labour market situation in comparison with the working-­age population as a whole. In 2014, the employment rate among the native population was 83.4 per cent compared with 66.7 per cent among the foreign-­born population. At the same time, the unemployment rate stood at 5.1 per cent in the native population compared with 15.5 per cent among the foreign-­born population. Since the Second World War the proportion of foreign-­born citizens in Sweden has increased rapidly, reaching around 16.5 per cent of the population by 2014 (Statistics Sweden 2015a). However, the composition of the foreign-­born population has changed notably in recent decades. Until the mid-­1970s immigration to Sweden consisted essentially of labour migrants from the Nordic and European countries. In 1970, about 60 per cent of the foreign-­born individuals residing in Sweden were born in other Nordic countries (principally Finland) and more than 90 per cent in western

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 573

28/09/2016 08:25

574

Europe’s disappearing middle class?

Europe (mainly from Italy, Greece and former Yugoslavia). In the wake of the first oil crisis and the decline of industrial and economic growth, the Swedish immigration policy became, as in other western economies, more restrictive. With the exception of Nordic workers and refugees from Latin America and eastern Europe, the immigration of workers came to a halt in the second half of the 1970s. The share of political and humanitarian refugees and tied-­movers (family reunification) increased, while the proportion of labour immigrants decreased significantly. From the early 1980s onwards, the bulk of immigrants came principally from the Middle East, Asia and Africa, and in the early 1990s refugees came from the former Yugoslavia. In 2014, around 15 per cent of resident immigrants were born in the Nordic countries, about 21 per cent in the rest of Europe and 64 per cent outside Europe. The past two decades, therefore, have seen a growing heterogeneity of the immigrant population with regard to ethnic origin and changes in the character of immigration, from labour immigration to humanitarian and political immigration. From the mid-­1970s onwards, the position of foreign-­born workers in the labour market deteriorated dramatically and the income differential between immigrants and natives, driven principally by the differences in employment level, increased notably (Hammarstedt and Shukur 2007). A large share of the political and humanitarian refugees that arrived during the 1980s never entered the labour market despite the then tight labour market and low overall unemployment rate. In the early 1990s, with the deep economic recession and unprecedented large upsurge of unemployment, the employment situation of immigrants deteriorated even further, while the inflow of refugees increased in connection with the conflicts in the Middle East (1989)11 and the civil war in the former Yugoslavia. According to Ekberg (2011), around 100 000 refugees came from the former Yugoslavia, mainly Bosnia-­Hercegovina (Ekberg 2011). Despite some improvement in the labour market situation from the second half of the 1990s, the extent of labour market integration of the foreign-­ born population, in particular among immigrants born outside Europe, remained very low up to the 2008 financial crisis and further deteriorated in connection with the economic downturn. As mentioned in section 2.3, lower employment rates, higher unemployment rates, lower average income12 and growing difficulties for newly arrived migrants or political refugees to get a foothold in the labour market also explain their larger share within the lower-­income group and the increasing proportion of migrants among poor households in Sweden. As shown in Figure 13.3 the proportion of foreign-­born people within the lower-­income class increased from 17.5 per cent in 1992 to 29.3 per cent in 2012.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 574

28/09/2016 08:25



The rise and fall of the Swedish middle class? ­575

30.0 25.1

25.0 19.3

20.0

10.0 5.0

17.5

15.4

15.0 8.0

11.4 9.1

8.4 4.7

2.8

12.3

23.4

23.0

12.9

7.5

14.1

10.7

9.7 5.8

5.0

16.3

15.6

8.1

12.5 9.0

13.5 11.6 9.2 9.1

1.9

0 1995

2000

2005

2008

Lower-income class

Lower middle-income class

Middle middle-income class

Upper middle-income class

Upper-income class

All

2012

Source:  LIS 1992 up to 2005, Eurostat for 2008 and 2012, and author’s calculations.

Figure 13.17 Share of foreign-born people in each income class and adult population, Sweden, 1992–2012 As also shown by Figure 13.3, the share of foreign-­ born people was unchanged in the top-­income bracket, while the share of migrants among the middle-­income class declined. As mentioned in section 2.3, with a given distribution of the population between natives and foreign born, the decline of the middle class between 1992 and 2012 should have been lower by 1 percentage point, meaning that almost one-­third of the actual decrease in the size of the middle class between 1992 and 2012 can be ascribed to the increase in the share of the migrant population. As shown by Figure 13.17, foreign-­born people are over-­represented in the lower tail of the income distribution, in particular in the lower-­income and lower middle-­income class. Conversely, they are under-­represented in the upper-­ middle and top-­income class. Using a multinomial logit, we have estimated the extent to which citizenship impacts on the likelihood of belonging to one of our three main income classes (lower-­, middle-­or top-­income class). Using pooled data for the years 1992 up to 2005 (LIS data) and for the years 2005–12 (Eurostat) and controlling for age, educational attainment and employment status, the results of our estimation (see Tables 13A.4 and 13A.5 in the appendix) show that to be foreign born increases – ceteris paribus – the likelihood of belonging to a lower-­income group (between 5 and 8 percentage points, depending on the period considered) and reduces the probability of being in the top-­income class (by almost 1 percentage point, independently of the period). Also worth noting is that the probability of a foreign-­born person

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 575

28/09/2016 08:25

576

Europe’s disappearing middle class?

belonging to the lower-­income class has increased since 1995, as shown by the sign of the interaction variable reported in Tables 13A.4 and 13A.5 in the appendix (Immigrant*year). In other words, part of the increase of income inequality and class polarization in Sweden can be ascribed to the changes in the ethnic composition of the population and by the worsening employment prospects of foreign born in the Swedish labour market. More globally the results of our estimations also show that, all things being equal, the probability of belonging to the low-­(versus top-­) income class has been increasing (decreasing) over time, confirming the tendency of a reduction of the size of the middle class and towards an increased polarization of the Swedish class structure. As also expected and corroborating our previous analysis, to be a student, retired or having low educational attainment ceteris paribus significantly increases the likelihood of belonging to the lower-­income category, confirming the crucial role of employment status and educational attainment or skills in understanding the transformation of Swedish class structure over time. In order to enhance our understanding of the transformations of the class structure and social stratification the use of panel data should be favoured in the future. Our approach based on cross-­sectional data suffers from the traditional drawbacks associated with this kind of data and limits the possibility of accurately assessing whether the transformation is actually the outcome of upward or downward mobility. The use of panel data would it make possible to better assess the dynamic of changes by analysing the transitions of individuals between various income classes over time.

6.  CONCLUSION AND POLICY ISSUES The analysis of the development of the Swedish class structure during the past three decades has clearly shown that Sweden experienced a relative decrease in its middle class and a stronger polarization of its class structure. We have shown that this decrease is due principally to a decline in the share of individuals belonging to the core middle class. Looking at the period as a whole, the share of the adult population belonging to the lower-­income group grew by more than 5 percentage points, while the top-­income groups increased by around 2 percentage points. This implies that around two-­thirds of the decline of the Swedish middle class might be ascribed to some form of downward mobility and one-­third to some form of upward mobility. During the same period, Sweden also experienced a concomitant increase in income inequality, mainly due to a weakening of the level of decommodification of the Swedish welfare state, changes in the balance of power between labour and capital, decentralization and

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 576

28/09/2016 08:25



The rise and fall of the Swedish middle class? ­577

individualization of wage setting, as well as increased returns to higher education and to capital income. The modification in the Swedish class structure and the relative diminution of the middle class are the outcome of complex causal configurations. We have identified three potential factors: changes in labour market behaviour, principally concerning labour market entry, with a lengthening of the transitional phase from school to work; a weakening of the goal of full employment and a concomitant deterioration of the labour market situation since the early 1990s, in particular the significant increase of youth unemployment and the low labour market attachment of newly arrived migrants; and last but not least, in recent decades, large structural changes in the Swedish employment and occupational structure. The patterns of job creation and destruction during the past three decades have implied a significant diminution, both in absolute and relative terms, of jobs in the lower tail of the job distribution and a clear increase of jobs in the upper tail of the distribution. The long-­term tendency towards an upgrading of occupational structure has particularly benefited the upper-­middle class and the top-­income group. Despite the above-­ mentioned rising wage inequalities, the relatively centralized and coordinated collective bargaining system and a still compressed wage structure have prevented the development of a low-­paid and low-­skilled sector in Sweden, and instead have boosted policies favouring skills upgrading. In effect, the large investment in research and development, the expansion of education during the past three decades as well as more balanced bargaining power between the two sides of industry have limited the tendency towards job polarization that can be found in liberal market-­orientated welfare states and have promoted the development of a high-­ skilled, knowledge-­intensive service economy. On the other hand, the reduction of demand for less qualified jobs has also impacted negatively on the possibility for low-­skilled or low educated workers to get a job and increased their propensity to become unemployed. In our view, the patterns of job creation and destruction in recent decades can scarcely explain the observed decline of the Swedish middle class. On the contrary, the increase in educational attainment and the associated skill upgrading of the employed population seem to have strengthened its position. Our results show that the decline of the middle class could be better explained by the reduction of employment rates among young people owing to the postponement of entry into the labour market and the worsening of the employment prospects of low-­educated young people and low-­skilled workers owing to the decline of the demand for low-­skilled jobs related to skill-­biased technological changes. Furthermore, the various ‘reforms’ of the social protection system – in particular in the pension system and unemployment benefits,

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 577

28/09/2016 08:25

578

Europe’s disappearing middle class?

where more restrictive eligibility criteria and lower-­income replacement rates were introduced – have without doubt reduced the disposable income of vulnerable groups in Swedish society and contributed to the increase of class polarization. However, in a cross-­country comparative perspective, it should be recalled that Sweden still counts among the countries with the lowest income inequalities and despite the relative decrease in its middle class still has one of the largest middle classes among modern post-­ industrial economies.

NOTES   1. Or the share of income of the three quintiles Q2–Q4 in total income (60 per cent of the adult population).   2. Defined by households with an equivalized income less than 75 per cent of median income.   3. The share of income of the three quintiles Q2, Q3 and Q4 (equalizing 60 per cent of the income distribution) defines the whole middle class, while the share of income for individuals in Q2 defines the lower-­middle class and the share of income of individuals belonging to Q4 defines the upper-­middle class.   4. Unfortunately, Statistics Sweden stopped providing data to LIS after 2005.   5. In 2008, the median age of entry into tertiary education was 22 and the age of graduation 28. The average entry age into the labour market was 20 years in 1990 and was almost 26 years in 2008 (see Anxo et al. 2011).  6. Besides the above-­mentioned large investment in secondary and tertiary education, during the period 1997–2002 the Swedish government implemented a new major adult education programme called the Adult Education Initiative or Knowledge Lift (KL) with the explicit objective of raising the educational level of low-­educated workers to three years’ upper-­secondary school. Participants were eligible for the same grants and loans as for ordinary adult education, and special education support was given to participants receiving unemployment insurance benefits. The size of the programme was unprecedented: in the period 1997–2002 more than 10 per cent of the labour force was enrolled in KL. To illustrate, in 1999 228 000 persons were enrolled in the KL ­programme – by comparison the number of pupils in the regular upper-­secondary school system was 300 000.   7. This also explains why the proportion of young unemployed engaged in youth-­targeted active labour market programmes has traditionally been high in Sweden. Youth programmes still constitute an important component of the transition between the education system and the labour market.   8. The stock market flourished during the second half of the 1990s and equities became a common means of saving for an increasing share of Swedish households. By the end of the 1990s around 60 per cent of all households were shareholders (Gustafsson and Palmer 2002).   9. In 2012 union density was around 70 per cent and the coverage rate of collective bargaining around 90 per cent (100 per cent in the public sector). 10. Alongside the Netherlands, Sweden has the highest employment rate in the EU. 11. In particular, political refugees from Iraq and Iran in 1989 and Bosnia-­Herzegovina between 1993 and 1994. 12. In 1992, the median household disposable income of foreign-­born people was 82 per cent of that of natives; it increased to 84 per cent in 2005 and fell again to 80 per cent in 2012.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 578

28/09/2016 08:25



The rise and fall of the Swedish middle class? ­579

REFERENCES Anxo, D. (2011), ‘Negotiated flexibility in Sweden: a more egalitarian response to the crisis?’, in D. Vaughan-­Whitehead (ed.), Work Inequalities in the Crisis: Evidence from Europe, Cheltenham, UK and Northampton, MA, USA: Edward Elgar and Geneva: ILO, pp. 445–76. Anxo, D. (2013), ‘Early fiscal consolidation and negotiated flexibility in Whitehead (ed.), Sweden:  a  fair way out of the crisis?’, in D. Vaughan-­ Public  Sector Shock: The Impact of Policy Retrenchment in Europe, Cheltenham, UK and Northampton, MA, USA: Edward Elgar and Geneva: ILO, pp. 543–75. Anxo, D. (2014), ‘Long-­ term trends in the Swedish jobs structure’, mimeo, Department of Economics and Statistics, Linnaeus University, Växjö and Kalmar, Sweden. Anxo, D. (2015), ‘The Swedish social model: resilience and success in turbulent times’, in D. Vaughan-­Whitehead (ed.), The European Social Model in Crisis, Cheltenham, UK and Northampton, MA, USA: Edward Elgar and Geneva: ILO, pp. 507–52. Anxo, D., G. Bosch and J. Rubery (eds) (2011), The Welfare State and Life Transition: A European Perspective, Cheltenham, UK and Northampton, MA, USA: Edward Elgar. Atkinson, A.B. and A. Brandolini (2011), ‘On the identification of the “middle class”’, ECINEQ working paper WP 2011 – 217, accessed January 2015 at at www.ecineq.org. Dallinger, U. (2013), ‘The endangered middle class? A comparative analysis of the role played by income redistribution’, Journal of European Social Policy, 23 (1), 83–101. Eurofound (2015), Upgrading or Polarization? Long-­Term and Global Shifts in the Employment Structure: European Jobs Monitor 2015, Luxembourg: Publications Office of the European Union. Eurostat (various years), EU Statistics on Income and Living Conditions EU-­ SILC, Sweden, 2008, 2011 and 2012, Eurostat, Luxembourg. Ekberg, J. (2011), ‘Will future immigration to Sweden make it easier to finance the welfare system?’, European Journal of Population, 27 (1), 103–24. Esping-­Andersen, G. (1990), The Three Worlds of Welfare Capitalism, Princeton, NJ: Princeton University Press. Fernández-­ Macías, E., T. Ward and R. Stehrer (2012), ‘Methodology’, in E.  Fernández-­Macías, J. Hurley and D. Storrie (eds), Transformation of the Employment Structure in the EU and USA, 1995–2007, Basingstoke: Palgrave, pp. 16–25. Foster, J.E. and M.C. Wolfson (2010), ‘Polarization and the decline of the middle class: Canada and the US’, Journal of Economic Inequality, 8 (2), 247–73. Goldthorpe, J.H. (2010), ‘Analysing social inequality: a critique of two recent contributions from economics and epidemiology’, European Sociological Review, 26 (6), 731–44. Goldthorpe, J.H. and A. Mcknight (2006), ‘The economic basis of social class’, in S. Morgan, D.B. Grusky and G.S. Fields (eds), Mobility and Inequality: Frontiers of Research from Sociology and Economics, Pala Alto, CA: Stanford University Press, pp. 109–136.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 579

28/09/2016 08:25

580

Europe’s disappearing middle class?

Gustafsson, B. and E. Palmer (2002), ‘Was the burden of the deep Swedish recession equally shared?’, Review of Income and Wealth, 48 (4), 537–60. Hammarsted, M., and G. Shukur (2007), ‘Immigrants’ relative earnings in Sweden – a quantile regression approach’, International Journal of Manpower, 28 (6), 456–73. Luxembourg Income Study (LIS), Database, Sweden Years 1975 to 2005, LIS, Luxembourg, accessed April–May 2015 at http://www.lisdatacenter.org. Pressman, S. (2007), ‘The decline of the middle class: an international perspective’, Journal of Economic Issues, 41 (1), 181–200. Statistics Sweden (2015a), ‘Household disposable income by deciles’, SCB, statistik/ Stockholm, accessed May 2015 at http://www.scb.se/sv_/Hitta-­ Statistik-­efter-­amne/Hushallens-­ekonomi/Inkomster-­och-­inkomstfordelning/Hus hallens-­ekonomi-­HEK/7289/7296/Disponibel-­inkomst-­19912011/163544/. Statistics Sweden (2015b), Labour Force Survey, Stockholm, various years, accessed April 2015 at www.scb.se. Svallfors, S. (2004), ‘Class, attitudes and the welfare state: Sweden in comparative perspective’, Social Policy and Administration, 38 (2), 119–38. Swedish National Agency for Higher Education (2012), Swedish Universities and University Colleges, Annual Report 2012, Högskoleverket, Stockholm.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 580

28/09/2016 08:25



The rise and fall of the Swedish middle class? ­581

APPENDIX 13A.1 Table 13A.1 Trends in population as a whole and income share by income classes, Sweden, 1987–2012 (18 years and above) 1987

1992

1995

2000

2005

2008

2012

Var pp

Population share Lower 11.5 Lower Mid 16.6 Mid Mid 43.3 Upper Mid 26.7 Middle all 86.6 Upper 1.9

12.1 16.1 40.7 27.8 84.6 3.3

10.0 14.7 47.0 25.3 87.0 3.0

12.3 17.1 38.7 26.9 82.7 5.0

12.0 18.0 38.8 27.5 84.3 3.8

14.4 17.0 35.8 29.1 81.9 3.8

16.9 15.6 34.2 29.1 78.9 4.2

+5.4 −1.0 −9.1 +2.4 −7.7 +2.3

Income share Lower Lower Mid Mid Mid Upper Mid Middle all Upper

4.3 10.9 40.8 36.8 88.5 7.2

3.0 10.4 45.0 34.6 90.0 7.0

4.6 10.8 37.6 35.8 84.2 11.2

4.8 11.4 38.3 36.4 86.1 9.0

6.1 11.2 33.5 40.0 84.7 9.3

7.1 10.3 32.0 40.3 82.6 10.3

+2.7 −1.2 −11.9 +3.8 −9.3 +6.6

4.4 11.5 43.9 36.5 91.9 3.7

Source:  Author’s calculations based on LIS and Eurostat.

Table 13A.2 Trends in income polarization, Foster and Wolfson Polarization Index, Sweden, 1987–2012

Factor inc Disp. inc

1987

1992

1995

2000

2005

2008

2011

2012

0.309 0.158

0.356 0.169

0.494 0.154

0.454 0.184

0.430 0.177

0.536 0.190

0.545 0.193

0.546 0.200

Source:  Author’s calculations based on LIS (1987–2005) and Eurostat (2008–12).

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 581

28/09/2016 08:25

582

Europe’s disappearing middle class?

Table 13A.3 Overview of Goldthorpe and ISCO classification of social class Goldthorpe schema

ISCO categories

Social class

Professional, administrative and  managerial employees, higher grade Professional, administrative and  managerial employees, lower grade; technicians, higher grade Routine non-­manual employees,  higher grade

1: Legislators, senior   officials and managers 2: Professionals (except 23:   teaching professionals)

Upper class

3: Technicians and associate  professionals 23: Teaching professionals 5: Service workers and shop  and market sales workers 4: Clerks 8: Plant and machine   operators and assemblers 6: Skilled agricultural and   fishery workers 7: Craft and related trade  workers 9: Elementary occupations

Upper-­middle  class

Small employers and self-­  employed workers Supervisors of manual workers;  technicians, lower grade Skilled manual workers, lower  grade Semi-­skilled and unskilled  manual workers

Core middle  class Lower  middle-­ income class Lower-­  income class

Source:  Goldthorpe and McKnight (2006) and Kuypers and Marx (Chapter 3 in this volume).

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 582

28/09/2016 08:25



The rise and fall of the Swedish middle class? ­583

Table 13A.4 Multinomial logit, marginal effects evaluated at sample mean, pooled sample, Sweden, 1992, 1995, 2000 and 2005 Variable

Marginal effect

Immigrant Age Age2 Female Not employed Retired In education Low education High education 1995 2000 2005 Immigrant*1995 Immigrant*2000 Immigrant*2005 Number of observations

Lower-­income class

Upper-­income class

0.032*** −0.013*** 0.000*** −0.009*** 0.145*** 0.036*** 0.199*** 0.032*** −0.014*** 0.023*** 0.038*** 0.081*** 0.002 0.050*** 0.013*** 76 941

−0.007*** 0.003*** −0.000*** −0.001*** −0.018*** −0.025*** −0.023*** −0.010*** 0.035*** 0.001 0.025*** 0.060*** −0.002 −0.013*** −0.040***

Note:  Standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. Source:  LIS database, author’s calculations.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 583

28/09/2016 08:25

584

Europe’s disappearing middle class?

Table 13A.5 Multinomial logit, marginal effects evaluated at sample mean, pooled sample, 2005, 2008. 2011 and 2012 Variable

Marginal effect

Immigrant Age Age2 Female Unemployed Retired Out of the labour force Low education High education 2008 2011 2012 Immigrant*2008 Immigrant*2011 Immigrant*2012 Number of observations

Lower-­income class

Upper-­income class

0.075*** −0.013*** 0.000*** 0.033*** 0.144*** 0.094*** 0.116*** 0.069*** −0.013*** 0.024*** 0.048*** 0.148*** 0.018* 0.007 0.014 51 535

−0.005*** 0.003*** −0.000*** −0.005*** −0.026*** −0.038*** −0.002*** −0.027*** 0.031*** −0.002 −0.003 −0.001 0.005 0.004 0.005

Note:  Standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. Source: 

LIS database, Eurostat for 2005, 2008 and 2012, and author’s calculations.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 584

28/09/2016 08:25



The rise and fall of the Swedish middle class? ­585 a) Low middle-income class 2012

17.5

2011

66.4

19.3

2008

19.5

2005

18.3

1995

16.1

62.8

17.9

66.5

14.0

70.1 25.6

1992

11.7

61.1

13.3

29.5

1987

24.9

1981

23.7

1975

61.0

10

12.6

63.9 32.8

0

9.5

62.4

12.4 53.4

20

30

40

Elementary occupations

50

60

13.8

70

Medium skilled workers

80

90

100

High skilled workers

b) Core middle-income class 2012

14.1

2011

24.9

60.9

16.7

62.7

2008

17.3

62.4

2005

17.1

63.0

1995

17.7

1992

20.3 19.9 71.3

19.6

1987

21.1

1981

23.2

1975 10

11.0

71.4

9.0

70.4

8.5

68.8

26.2

0

20.6

8.1

67.0

20

30

40

Elementary occupations

50

60

6.8

70

Medium skilled workers

80

90

100

High skilled workers

c) Upper middle-income class 2012

53.0

7.6

39.4

2011

9.1

54.1

36.8

2008

8.7

53.8

37.5

2005

9.5

1995

9.3

1992

9.8

1987

10.9

1981

56.9

20.8

69.9

19.3

69.4

12.6

0

22.6

69.4

11.5

1975

33.6 68.2

19.1

72.7

10

20

30

Elementary occupations

40

50

14.7

60

Medium skilled workers

70

80

90

100

High skilled workers

Source:  LIS 1975–2005 and Eurostat for 2008–2012 and author’s calculations.

Figure 13A.1 Skill distribution by income groups, lower middle-income class, core middle class and upper middle-income class, Sweden, 1975–2012 (percentage)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 585

28/09/2016 08:25

14. How have middle-­income households fared in unequal Britain? A focus on work and employment trends Damian Grimshaw and Anthony Rafferty 1. INTRODUCTION The years of economic crisis and austerity in the United Kingdom have raised fundamental questions about the need for economic growth to be distributed more fairly, especially to combat rising poverty, to curtail runaway incomes among the wealthy, to re-­establish pathways for people aspiring to the middle class and to halt the erosion of income among incumbent middle-­class households (for example, Lansley 2009; Cribb et  al. 2013; Parker 2013; Plunkett et al. 2014; Hills 2015). Because the level of income inequality in the United Kingdom is high relative to most countries with a similar standard of living, the cost of downward instability among middle-­income households is arguably greater. Yet the empirical evidence suggests that a wide range of economic, political and labour market factors – some of which precede the 2008–09 economic crisis – have destabilized the place of the middle class in contemporary Britain and have contributed to a growing share of households at or just above the poverty income threshold. At stake is the character of UK society. Since 2010, welfare policy reforms associated first with the election of a Conservative-­led coalition and then with a majority Conservative government have contributed to a rise in absolute poverty (Belfield et al. 2014). Macroeconomic policy has, on the one hand, via the Bank of England’s policy of quantitative easing (QE), significantly boosted wealth among the richest (Bank of England 2012: 10) and on the other, via public spending cuts, downsized low-­and middle-­income jobs and pay levels in the public sector, damaging women’s employment situation especially (Grimshaw 2013; Rubery and Rafferty 2013). Meanwhile, longer-­term labour market trends have eaten away at 586

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 586

28/09/2016 08:25



How have middle-­income households fared in unequal Britain? ­587

the standard employment relationship and hollowed out prospects for relatively secure and stable middle-­income occupations. This chapter focuses on the effects of recent trends in work and employment on the position and character of middle-­ income households in Britain, focusing both on households at the middle of the income distribution and on individuals with mid-­ranking wages. As such, it complements other studies that focus specifically on low incomes (for example, Corlett and Whitakker 2014; Kumar et al. 2014) and top incomes (for example, Alvaredo et al. 2013) and contributes a stronger focus on the effects of work and employment trends. In common with other country chapters in this ILO programme of research, we focus on the effects of the 2008–09 economic crisis and subsequent period of austerity up to 2014 and set this against longer-­term trends in the world of work since 1995. The chapter is divided into three sections. Section 2 introduces key definitions and measures, and draws the basic pattern of distribution of households in Britain using ‘income cut-­off’ measures of lower, middle and upper classes. Section 3 interrogates the diverse ways in which trends in the world of work have impacted middle-­income households, focusing in particular on patterns of labour market participation, wage determination, polarized job growth and public sector downsizing. These trends are analysed against a wider backdrop of changes involving a radical shift in the economic settlement experienced by UK citizens (resulting in a lost decade of real income gains), erosion of the (still significant) distributive role of the tax and benefits system and the persistent high level of income inequality. Section 4 presents two case studies that provide further depth to our understanding of the contemporary experiences of the middle-­income classes in employment. The conclusion synthesizes the findings and presents recommendations for policy and practice.

2. MIDDLE-­INCOME HOUSEHOLDS – DIVERSE COMPOSITION AND CHANGING TRENDS 2.1  Defining Household Income and Middle-­class Incomes This chapter uses measures of household income to describe the changing fortunes of middle-­income groups. For household income we use disposable income adjusted for household size (equivalized using the modified Organisation for Economic Co-­ operation and Development, OECD, scale1) because this is probably the best approximation of material living standards. Following the UK Office for National Statistics (ONS), this measure includes original income (for example, from employment and

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 587

28/09/2016 08:25

588

Europe’s disappearing middle class?

investments), cash benefits (for example, state pension) and direct taxes (for example, social security, local tax and income tax). We exclude the effects of indirect tax (for example, value added tax, VAT), housing costs and benefits in kind (such as free health care and education services) (see ONS 2014: 3). Household income data are from the Family Resources Survey (FRS),2 a continuous cross-­sectional survey from 1992. Our analysis takes 1995 as the starting point and we focus on Britain only (England, Scotland and Wales). The dataset under-­reports the number and actual incomes of very high-­income households as compared with tax records (such as Her Majesty’s Revenue & Customs, HMRC, Survey of Personal Incomes3). Reported trend data adjust for 2012–13 prices using the recommended index.4 Our analysis reports findings for all households and, where more appropriate, for working-­age households where detailed analysis captures the characteristics of a ‘reference person’ for each household.5 Quantitative analyses of income classes tend to use one of two measures. The ‘people space’ measure segments the population into fixed units, such as decile or quintile groups, and typically charts the changing relative positions of each group. An alternative ‘income cut-­off’ measure segments the population according to fixed relative income bands and therefore facilitates the observation of changes in the size of each income-­class category. In common with other contributors to this International Labour Organization (ILO) research programme, we mostly rely on the income cut-­off measure. So, what are the relevant income thresholds for the United Kingdom? Here we follow the agreed ILO definitions for this research project and use the lower and upper income cut-­offs of 60 per cent and 200 per cent of median household income, distinguishing between the lower middle (60–80 per cent), mid middle (80–120 per cent) and upper middle (120–200 per cent). 2.2  Trends in the Size and Shape of Middle-­income Classes In 2012–13, nearly three in four households (73 per cent) were located along the middle-­income range of 60 per cent to 200 per cent of net equivalized median income (Figure 14.1). Using these income cut-­off definitions, there were more poor households than rich households, amounting to shares of 17 per cent (4.3 million) and 10 per cent (2.6 million), respectively. In terms of subdivisions of middle income households, around one-­quarter of all households were ‘upper middle’ (120–200 per cent), one-­third ‘mid middle’ and less than one-­fifth in the ‘lower middle’ income category. Since 1995–96, there has been a modest rise in the share of middle-­ income households, from 70 per cent to 73 per cent, with reductions among both the lower-­and upper-­income classes. The income classes that experienced the largest increases since 1995–96 were the mid-­middle

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 588

28/09/2016 08:25



How have middle-­income households fared in unequal Britain? ­589 a) All households

100 90 80 70 60 50 40 30

11.4

10.6

11.0

10.1

25.7

25.2

25.2

25.9

11.5

12.7

12.7

13.4

13.6

14.6

15.2

16.5

19.1

17.9

16.8

17.7

18.7

19.1

19.1

16.6

1995–96

2002–03

2007–08

2012–13

20 10 0

Below 60

60–80

80–100

100–120

120–200

200+

b) Working age households 100

14.3

13.3

13.3

11.7

30.8

30.0

29.3

28.1

11.9

13.3

12.9

12.8

12.6

13.4

13.5

15.1

20

14.6

13.4

13.7

15.9

10

15.9

16.6

17.3

16.4

1995–96

2002–03

2007–08

2012–13

90 80 70 60 50 40 30

0

Below 60

60–80

80–100

100–120

120–200

200+

Source:  Family Resources Survey; authors’ compilation.

Figure 14.1 Trend in the distribution of households by income class, United Kingdom, 1995–2013 (percentage) categories, at 80–100 per cent and 100–120 per cent of median income. By contrast, lower middle-­income households reduced their share (19.1 per cent to 17.7 per cent) and the upper middle remained remarkably constant in size (around 26 per cent). Together, the mid-­middle categories (80–120

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 589

28/09/2016 08:25

590

Europe’s disappearing middle class?

per cent of median income) expanded from 25 per cent to 30 per cent, suggesting a significant bulging of the middle of the income distribution over the 17-­year period. The five-­year period since 2007–08 captures the effects of the economic crisis and subsequent policy regime of ‘austerity’ measures. During this recent period, the bulging in size of middle-­income classes in fact occurred more among those households below the median income threshold – the 60–80 per cent and 80–100 per cent groups. The share of households below the poverty threshold shrank during the 2007–12 period as incomes among the poorest did not fall as quickly as incomes at the median point over this period; nevertheless, more households are poor in absolute terms (Belfield et al. 2014: 36–7, 78–9). Restricting the sample to working-­age households, we find similarities and differences in the trend (Figure 14.1b). The share of middle-­income households (retaining the same income cut-­ off thresholds) also rose slightly, from 70 per cent to 72 per cent. However, while the share of upper income households declined (more so than for all households), the share of poor households increased over the 17-­year period. Again, a bulging of mid middle income households is evident, with a rise from 24.5 per cent to 27.9 per cent of working-­age households. 2.3 High Inequality and the Shifting Position of Middle-­income Households Since 2007–08 the bulging is notable for its skewed emphasis among middle-­income households located between the poverty threshold and the median. Figure 14.1b shows a rise from 27 per cent to 31 per cent, while the share of middle-­income households above the median fell from 42 per cent to 41 per cent. The expansion of middle-­income households below median-­income levels is significant because it suggests that the United Kingdom is not moving from a pyramid-­shaped society, in which most people occupy the low-­income ranks, to a diamond shape, with a majority enjoying incomes clustered around the mean. Two trends have hindered this development – rising income inequality during the 1980s and a compression of middle-­class incomes around the lower half with an expanding share experiencing incomes between the poverty threshold and median income. The OECD ranks Britain seventh out of 35 countries using its Gini measure of net, equivalized household income inequality, placing it number one among European countries (Figure 14.2). However, unlike other countries which have recently experienced a period of rising income inequality, such as Spain, France, Sweden and

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 590

28/09/2016 08:25



How have middle-­income households fared in unequal Britain? ­591

0.40 United States

0.38 0.36

United Kingdom 0.34 Italy

Japan

0.32

Hungary

France

0.30 Netherlands 0.28 0.26

Germany

Sweden

0.24

Denmark

0.22

1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

0.20

Source:  Authors’ compilation from OECD equivalized net household income data.

Figure 14.2 International trends in the Gini measure of household income inequality, United Kingdom, 1985–2012 Denmark, inequality in the United Kingdom on this general measure has been relatively stable at its high level since the early 1990s. The most significant period of change in the United Kingdom was the 1980s when political, economic and welfare reforms under the Thatcher government forced through a rupture in the socio-­economic settlement and a dramatic increase in inequality (Gottschalk and Smeeding 1997), reversing the prior post-­war trend of a steady decline in inequality.6 Britain has in effect, therefore, already experienced high-­level inequality for a generation; high inequality appears to be locked into the social and economic fabric of contemporary British life. The alternative inter-­decile measure (P90/P10) reveals a similar trend. From a low level of 3.3 in 1978, the inter-­decile measure increased significantly to 5.0 in 1990, then fluctuated around the 4.5 level up to the crisis in 2008, since when it has in fact declined, measuring 4.1 in the most recent data for 2012–13. In general, this pattern reflects change in both halves of the income distribution. Median household income pulled away from the bottom decile level at a rapid rate during the 1980s (from a ratio of around 1.8 up to 2.2), as did the top decile from the median (from around 4.2 to 5.4). Indeed, a significant minority of households have experienced very high household incomes, which pulls the level of average income further

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 591

28/09/2016 08:25

592

Europe’s disappearing middle class?

apart from the median. Studies using tax records (for example, Atkinson et  al. 2011; Kenworthy 2013) show that ‘top-­ heavy inequality’ in the United Kingdom has soared since the 1980s. After a period of fluctuations the median fell against the bottom decile after the crisis, from 2.1 to 1.9 (2007–08 to 2012–13), signalling greater compression among the bottom half of household incomes, while the top decile sustained its differential (at 2.1 in 2012–13). Figure 14.3 shows that as the total median income fell – real net household income in Britain at the median level fell by 4 per cent, from £455 per week to £437 – the distribution of households moved to the left and 60–80% 80–100%

3.0

(2012/13)

(2012/13)

2007–08

100–120%

2012–13

(2012/13)

2.5

120–200% (2012/13)

Percentage

2.0 1.5 1.0 0.5 0 0

100

200

300

2007–08 2012–13 % change mean income % change share of households

400 500 600 Weekly net income 60–80% £273–364 £262–350 –4.4 5.2

700

80–100% £364–455 £350–437 –4.3 8.3

800

900

100–120% £455–546 £437–524 –4.4 5.6

1000

120–200% £546–910 £524–874 –4.7 2.4

Note:  Holding constant changes in population, FRS data before housing costs, 2012–13 prices. Source:  Authors’ compilation, after Carr et al. (2014: chart 2.4).

Figure 14.3 Change in distribution of real net household weekly income, United Kingdom, 2007–08 to 2012–13 (2012–13 prices)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 592

28/09/2016 08:25



How have middle-­income households fared in unequal Britain? ­593

increased in concentration around the lower part of the mid middle-­ income band, 80–100 per cent of median income. The shrinking share of households in the upper-­and lower-­income classes was distributed among the lower and mid-­middle classes, not so much the upper-­middle class households. Moreover, relative to the mean income of £536 in 2012–13, around two-­thirds of households had incomes below the average level of net equivalized income. The distribution is quite clearly skewed with a bulge (the mode) centred around 70–80 per cent of median household income; 34 per cent of households fall within the 60–100 per cent range of income. 2.4  Unequal Concentration of Income

11.5 13.4 10.5 12.2

10

80–100%

100–120%

7.0 5.7

15

13.6 16.5 10.1 12.3

19.1 17.7

20

11.1 10.3

25

18.7 16.6

30

11.4 10.1

25.7 25.9 32.7 32.6

35

28.6 27.0

As expected, there is an increasing concentration of income as we move from the poorest to richest households (Figure 14.4). In 2012–13, while the 10 per cent of households in the upper-­income class enjoyed more than

5 0

% of median

Below 60%

60–80%

120–200%

1995 % of households

2012 % of households

1995 % share of income

2012 % share of income

No. of households (millions)

% of households

Cumulative %

Total income (£ nearest million)

200+%

% of total income

Cumulative %

Below 60%

4.32

16.6

16.6

788

5.7

5.7

60–80%

4.61

17.7

34.2

1420

10.3

16.0

80–100% 100–120% 120–200%

4.30

16.5

50.7

1690

12.3

28.3

3.48

13.4

64.1

1680

12.2

40.4

6.74

25.9

89.9

4490

32.6

73.0

200+%

2.63

10.1

100.0

3 720

27.0

100.0

26.08

100.0

13 788

100.0

Total

Source:  FRS extended, authors’ compilation, deflated, equivalized, before housing costs, new Survey of Personal Incomes (SPI) income (OECD).

Figure 14.4 Income concentration by household income class, UK, 1995–96 and 2012–13 (percentage)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 593

28/09/2016 08:25

594

Europe’s disappearing middle class?

one-­quarter of total income (27 per cent), the 17 per cent of households in the lower class earned just 6 per cent of total net income. This means that while the richest enjoyed three times their share of income, the poorest only earned one-­third of their share; in numbers, 2.6 million rich households shared £3.7 billion, while 4.3 million poor households shared just £0.8 billion. Middle-­income households earned two-­thirds of total income (67 per cent), which in 2012–13 was lower than their share of households (73 per cent). However, this reflects compensating patterns among lower-­ middle and upper-­middle categories. The only category that earned close to its ‘fair share’ were those households with net incomes in the 100–120 per  cent range. Since 1995–96, changes in income concentration have shifted in line with the changing distribution of households by income category. Thus the bulging of mid middle income households (80–120 per cent) saw their share of total income rise from 21 per cent to 24 per cent. 2.5  Diverse Composition of Middle-­income Households Tracing the broad patterns and trends in household income categories using the methodology of equivalized incomes brushes over the fact that we are referring to many different types of household structure. Figure 14.5 plots the distribution by income class. The main differentiation across income classes is the distribution 100 90 Single female

80

Single male

70

Couple

60

Single with children 50 Couple with children 40

Single female pensioner

30

Single male pensioner

20

Pensioner couple

10 0

Source:  FRS extended, authors’ compilation, deflated, equivalized, before housing costs.

Figure 14.5 The distribution of household type by income class, United Kingdom, 2012–2013 (percentage)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 594

28/09/2016 08:25



How have middle-­income households fared in unequal Britain? ­595

of pensioners and non-­retired adults without children. The most even distribution is in the mid-­middle category (80–120 per cent of median income), where around one-­third of households are adults without children, another third are adults with children and a third are pensioner households. Among upper middle-­and upper-­income class households the share of non-­retired adults without children is far higher, accounting for a majority of both classes, owing to a lower representation of single-­parent households and pensioner households. Significantly, three-­quarters of all single parent households are in households below the median income level compared with half of dual-­parent households. Differentiating by education reveals a stark clustering of high school education at one end of the class spectrum and higher education at the other, as well as variation among our 60–200 per cent band of middle-­ income households (Figure 14.6). Among lower-­income households, two-­thirds of reference adults have only high school education, while in upper-­income households this is true 100 90

20.7

22.2

26.6 34.6

80

52.0

14.6

14.3

70

37.3

13.7

73.4

60

15.6

13.5

50

Higher education Further education High school

13.0

40 64.7

63.5

30

59.7 49.8

8.8

20

49.2

34.9

10

17.8

l ta To

+% 20 0

12

0– 2

00 %

% 10

0– 1

20

% 00

% 80 60 –

80 –1

Be

lo w

60 %

0

Source:  FRS extended, authors’ compilation, deflated, equivalized, before housing costs.

Figure 14.6 Level of education by household income class, United Kingdom, 2012–13 (percentage)

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 595

28/09/2016 08:25

596

Europe’s disappearing middle class?

100 90 80 70 60 50 40 30 20 10 0 Below 60%

60–80%

80–100%

100–120%

120–200%

200+%

Total

Managers Directors & Senior Officials

Professional Occupations

Associate Prof. & Technical Occupations

Admin & Secretarial Occupations

Skilled Trades Occupations

Caring, Leisure and Other Service Occupations

Sales & Customer Service

Process, Plant & Machine Operatives

Elementary Occupations

Undefined

Note:  Occupational status (according to SOC 1-digit) is registered for all persons who have ever worked; for each household, occupation is registered for the reference person (highestincome householder) without regard to gender. Source:  FRS 2012–13; authors’ compilation.

Figure 14.7 Occupational status by income class, United Kingdom, 2012–13 (percentage) of less than one in five adults. Reference adults with higher education account for a majority in the upper middle-­and upper-­class households and only around one in five in the lower-­and lower middle-­income classes. Restricting the sample to working-­age households reveals a similar pattern, albeit with lower shares in all income classes of adults with only high school education. Our use of household income cut-­offs in order to differentiate class partly aligns with differentiation by major occupational group, although the pattern is far more mixed. Figure 14.7 reports the general one-­ digit classification using the Standard Occupational Classification (SOC). Overall, professionals, managers and associate professionals account for around two in five household heads across Britain. Their share among the upper-­income class is far higher, at more than four in five households (84 per cent), and still high among the upper middle-­income class (60 per cent). For other income classes the share is below the average level,

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 596

28/09/2016 08:25



How have middle-­income households fared in unequal Britain? ­597

although surprisingly still constitutes around one-­fifth of households with incomes below and just above the poverty threshold. Conversely, elementary occupations tend to be clustered among the lower-­income households, accounting for around one-­fifth of household heads in lower-­income and lower middle-­ income households. Variation of occupational status by income category is likely to reflect differences in life-­stage (career and pay progression), as well as wage inequality within each of these very large one-­ digit occupational groups.

3. HOW HAVE CHANGES IN THE WORLD OF WORK IMPACTED ON MIDDLE-­CLASS INCOMES? Income from employment is fundamental in shaping the changing fortunes of the British middle classes. It is therefore clear that a wide array of interrelated changes in the world of work are continuously at play in changing the fortunes of middle-­ income households. These include patterns of labour market participation, pay practices, pensions, the scale and shape of job growth across different occupations, employer use of non-­standard employment contracts, government investment in the public sector, income protection for those seeking work or in work but with household incomes below the poverty threshold, patterns of gender equality in employment and the roles of trade unions and employers in crafting the quality of employment and trends in career mobility. This section is organized around six sub-­headings, summarized in Table 14.1. For each factor, our analysis reflects on four questions of common concern. First, is there a distinction between longer-­term conditions that precede the crisis and conditions that emerged post-­crisis as a result of the recession, subsequent period of slow growth and government ‘austerity’ policies? Second, what role have the main social actors played, namely, the government, employers and trade unions? Third, to what extent do changes in the particular employment conditions help explain the patterns of middle-­class incomes identified above? Fourth, what are the linkages with broader socio-­economic factors? 3.1  Labour Market Participation and the Problem of Precarious Work Participation in the labour market provides a key explanatory factor for the income-­class status of working-­age households. Households with at least one person in full-­time employment constitute the vast majority of upper middle-­and upper-­income classes but barely figure among the lower-­income class (Figure 14.8). For example, more than four in five

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 597

28/09/2016 08:25

598

VAUGHAN 9781786430595 PRINT (M4051).indd 598

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:25

Welfare reforms encourage  employment participation (and self-­employment) Stronger employment rights  for people employed in non-­standard jobs (part-­ time, temporary and agency work)

Statutory minimum wage  introduced (1999); raised relative value 2003–07 Limited reforms of  shareholder/company finance rules

Wages: limited real  growth Wage stagnation since 2002 Persistent wide gender   pay gap Growing share paid  60–80% of median wage Stable, high rate of productivity  growth (approximately 3 per cent) but 15–20 per cent gap with

State policy reforms

Labour market  participation and precarious work Strong job growth  performance for men and women (employment rate rose from 71 to 73 per cent) Unemployment rate  steady decline from 7 to 5 per cent

Pre-­crisis (1997–2008)

Employment drivers

Extensive use of  flexible employment contracts across all economic sectors Some evidence  of exclusion of workers with part-­ time, temporary and agency contracts from decent employment prospects Compression of  pay differentials in low and middle job positions Emphasis on pay  rises for highest paid (bonuses, stock options, and so on)

Employer actions

Extended labour  market connections underpin social inclusion of all households Risk of growing  use of precarious employment forms Stagnant wage  growth hinders growth in household income Successful  negotiation of new pay structure for national health sector (2004) Failure to arrest  falling membership density (27 per cent to 25 per cent)

Impact on middle-­ income households

Significant shift  in strategy to mobilize workers with non-­standard employment contracts as part of new encompassing approach

Union actions

Table 14.1 The changing world of work, role of social actors and impacts on middle-­income households, United Kingdom, 1997–2009

599

VAUGHAN 9781786430595 PRINT (M4051).indd 599

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:25

Job structure:  polarization Evidence of polarization  during 2001–10 – loss of middle-­wage jobs Highest middle-­wage job  losses among male and female full-­timers and high-­ technology/low-­technology manufacturing sectors

Tax and benefits

 France, Germany, USA (per hour) Extension and uprating of  wage top-­ups (tax credits/ in-­work benefits) New 50 per cent income tax  on high earnings; income thresholds rise with inflation; reduced starting rate (20 per cent to 10 per cent) but then abolished in 2008 Failure to shore up  manufacturing sector, especially damaging for balanced regional growth Delayering of  mid-­level positions contributes to loss of middle-­wage jobs

Over-­use of  minimum wage as normal rate in many low-­wage jobs n.a.

Mostly positive  effect on household income albeit increasing reliance of low and mid earners on welfare (tax credits) Restricted mid-­wage  job opportunities and reduced career ladders for lower middle-­class households

n.a.

Decline in density in  manufacturing (33 to 21 per cent) and construction (30 to 17 per cent) weakens ability to shore up demand

600

VAUGHAN 9781786430595 PRINT (M4051).indd 600

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:25

Labour market  participation and precarious work Fall in male and female  employment rates, then recovery from 2010 (male), 2012 (female)

Abolished state subsidies for Extensive use of  flexible employment  job creation during contracts across all recovery (2010) and economic sectors, more restrictive welfare especially agency conditionality rules on job work and zero-­hours seekers contracts Limited policy response to  massive rise in use of zero-­ hours contracts

Strategic emphasis  on protecting workers with precarious employment contracts

Weaker labour  market connections lowers household income and raises risk of indebtedness

Very strong  underpinning of M-­income households

Several significant  successes in negotiating better pay deals (especially health), harmonized conditions (health and local government), skill development (health) and gender equality (all sectors)

Relatively positive  approach towards trade union engagement coupled with increasingly performance-­ oriented management techniques

Major government  programme of investment supports strong job growth, especially in health and education

Public services:  expansion Massive job expansion. Major source of employment  for highly educated, especially significant for women

Post-­crisis (after 2008–09)

Impact on middle-­ income households

Union actions

Employer actions

State policy reforms

Employment drivers

Table 14.1  (continued)

601

VAUGHAN 9781786430595 PRINT (M4051).indd 601

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:25

Job structure: balanced growth   but more precarious Post-­2010 relatively balanced  job growth across the wage distribution Positive job growth in  middle and high paid manufacturing jobs

Wages: falling real pay Collapse in real wages   since 2009 Persistent gender pay gap Growing share paid  60–80 per cent of median Stagnant productivity  (per hour), widened gap with France, Germany, USA (20–30 per cent) Tax and benefits n.a.

Explicit policy post-­2010 to  reduce earnings threshold for 40 per cent income tax (by 15 per cent nominal over four years); cut 50 per cent high earner rate to 45 per cent; raised personal allowance by 30 per cent New political consensus to  rebalance the economy, but mostly involves devolution of social spending (for example, ‘northern powerhouse’) rather than new investment Major job cuts in  FTSE companies during 2008–2009; partial shift from standard employment contracts to precarious contracts post-­2010

Pay lags behind  change in output per worker/hour Continued high pay  practices Continued misuse   of minimum wage

Fall in relative value of  statutory minimum wage until 2014–15 rise

Weakened presence  in private sector workplaces and limited role in restructuring practices

Stable level of  collective bargaining coverage: low in private sector 17 per cent, high in public sector 64 per cent Few effective pay campaigns n.a.

Positive mid-­wage  job opportunities but high share require self-­ employment and other precarious contractual forms

Negative effect on  household incomes

Falling wages stifles  household income and reduces upwards mobility

602

VAUGHAN 9781786430595 PRINT (M4051).indd 602

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting

28/09/2016 08:25

Accelerated rise in share of  self-­employed (13 to 15%), with falling average real incomes Rising share of part-­  time jobs and growth of ‘involuntary part-­timers’ Public services: downsizing   and privatization Falling job demand, real wage   cuts and pension cuts

Employment drivers

Table 14.1  (continued)

Major government  programme of ‘retrenchment’ – spending cuts, capital investment cuts Public sector pay freeze  (2011–2013) then capped at 1 per cent (below inflation)

State policy reforms

Negotiated worse   pension conditions Continued  outsourcing of public services to private sector (some living wage concessions)

Employer actions

Many public protests  and strikes but failed to arrest pay cuts, downsizing and pension cuts.

Union actions

Very strong  undermining of M-­income households, especially female members

Impact on middle-­ income households



How have middle-­income households fared in unequal Britain? ­603

100 90 80 Workless, other inactive

70

Workless, head or spouse unemployed

60

Workless, head or spouse aged 60+

50

1+ part-time, no full time 1 full-time & 1 not employed (couple)

40

1 full-time & 1 part-time (couple) 30

All full-time (single/couple)

20

One or more self employed

10 Total

200+

120–200

100–120

80–100

60–80

Below 60

0

Source:  FRS; working-age households 2012–13; authors’ compilation.

Figure 14.8 Patterns of employment participation among different income classes, United Kingdom, 2012–13 (percentage) (83  per cent) upper middle-­income households have at least one person in full-­time employment in the labour market, but this is true of only one in five (21 per cent) lower-­income class households. Similarly, patterns of worklessness are found in less than 10 per cent of households with incomes above the median level, but they are found in one-­third of lower middle-­ income households (34 per cent) and one-­half of lower-­income households (50 per cent). Overall, weaker and less effective connections with the labour market are characteristic of low-­income households. Traditional gendered patterns of employment participation, which divide a couple between full-­ time and part-­ time or full-­ time and no employment, are most likely to be found among mid middle-­ income households.7 While such household working patterns constitute only one in four of all households, their share is highest among the 80–100 per cent and 100–120 per cent mid middle-­income classes, at 31 per cent and 30 per cent, respectively. Since the crisis and subsequent jobs recovery, middle-­income households have experienced notable shifts in patterns of labour market participation. Compared with 2007–08, full-­time employment plays a smaller role – it has fallen from a share of 46 to 43 per cent of households (lower middle), 67 to 64 per cent (mid middle 80–100 per cent) and 76 to 73 per cent (mid middle 100–120 per cent). Conversely, the share of middle-­income households where connection to the labour market is only established via part-­time

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 603

28/09/2016 08:25

604

Europe’s disappearing middle class?

Table 14.2 Trends of full-­time and part-­time working pre-­and post-­crisis by household income class, United Kingdom, 1995–2013 Below 60%

60–80%

80–100%

100–120%

120–200%

200+%

Total

At least one person in full-­time employment 1995–96 11.3 33.5 58.1 2007–08 19.9 46.2 66.7 2012–13 20.9 43.1 63.5

72.5 76.0 73.0

80.8 83.3 82.5

77.3 77.7 77.7

58.5 53.3 51.6

1 + in part-­time employment, no full-­time 1995–96 9.8 12.2 8.2 2007–08 14.8 14.3 9.6 2012–13 15.8 14.4 12.6

4.7 6.1 9.4

3.4 3.3 4.9

2.9 2.7 2.0

6.4 7.9 9.6

Source:  FRS data; authors’ compilation.

work has grown – from 10 to 13 per cent (mid middle 80–100 per cent), 6 to 9 per cent (mid middle 100–120 per cent) and 3 to 5 per cent (upper middle).8 This post-­crisis trend reverses the trend prior to 2007–08 which was characterized by an increasing distribution of full-­time employment across all income classes (Table 14.2). The share of households with at least one person engaged in full-­time work grew from 59 per cent to 63 per cent from 1995–96 to 2007–08 and this expansion was especially beneficial among the lower-­ middle classes, among whom it has since diminished. Especially notable is the rising share of households below the poverty threshold (60  per  cent) with at least one person in full-­time employment over the 17-­year period, which is a sure reflection of Britain’s enduring problem of low pay. Spread of precarious employment post-­crisis The evidence fits with wider reports concerning the rising share of involuntary part-­time employment during the post-­2010 jobs recovery (Rubery and Rafferty 2013), as well as the high share of non-­standard forms of employment among net job growth since the crisis (OECD 2015). Until 2011, the share of temporary contracts had remained relatively stable at around 5–6 per cent over the previous two decades with a slight bias towards lower-­wage jobs than middle-­and higher-­wage jobs. But during the current period of rapid job expansion in an ongoing context of economic uncertainty, employers have expanded their use of temporary contracts, especially for jobs at the lower end of the wage spectrum. Related research shows that temporary jobs contributed as much as one-­fifth of net job expansion among the bottom two quintiles of the wage distribution during 2011–14 (Grimshaw and Rafferty 2014: fig. 3.17).

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 604

28/09/2016 08:25



How have middle-­income households fared in unequal Britain? ­605

100 90 80

9.8

14.1

25.1

70 60

35.9 13.0

50 40 30

16.7

14.6

14.3

15.9

21.2

12.3

Temporary

Permanent

20 10

7.2

0 Below 60

60–80

80–100

100–120

120–200

200+

Source:  FRS working-age households; main job of the reference person; authors’ compilation.

Figure 14.9 Distribution of temporary and permanent jobs by household income class, UK, 2012–13 (percentage) Translated to their effects on middle-­income households, it appears that the incidence of temporary employment is skewed towards the low-­and lower middle-­ income households (below median income). Figure 14.9 shows that compared with the distribution of permanent jobs, temporary jobs are far more over-­represented among lower-­income households but also to a lesser extent among lower middle-­income households. Perhaps the most striking feature of the post-­2010 jobs recovery has been the significant growth in self-­employment. The share of workers in self-­employment was stable prior to the crisis (2003–09), but then rapidly accelerated so that by 2014 it accounted for nearly one in six workers (15 per cent). During 2011–14 the increased incidence of self-­employment was associated with several factors, including tougher welfare entitlement rules, new technologies that enable employers and customers to access ‘crowdsourced’ employment and straightforward income tax advantages (Bergvall-­Kåreborn and Howcroft 2014; Etherington and Daguerre 2015). Overall, jobs for the self-­employed accounted for 55 per cent of net job expansion during 2011–14 (540 000 of 990 000 net job growth). At the same time, estimated real median income has massively declined since the crisis by almost one-­third (Figure 14.10), far more than the drop experienced by people in standard paid employment.

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 605

28/09/2016 08:25

15

£350

15

£330

14

£310

14

£290

13

£270

13

£250

12

£230

12

£210

11 11

£190 Self-employed

Real median weekly income

10

£170

Real median weekly income

Europe’s disappearing middle class?

£150

Jan–Mar 1993 Oct–Dec 1993 Jul–Sep 1994 Apr–Jun 1995 Jan–Mar 1996 Oct–Dec 1996 Jul–Sep 1997 Apr–Jun 1998 Jan–Mar 1999 Oct–Dec 1999 Jul–Sep 2000 Apr–Jun 2001 Jan–Mar 2002 Oct–Dec 2002 Jul–Sep 2003 Apr–Jun 2004 Jan–Mar 2005 Oct–Dec 2005 Jul–Sep 2006 Apr–Jun 2007 Jan–Mar 2008 Oct–Dec 2008 Jul–Sep 2009 Apr–Jun 2010 Jan–Mar 2011 Oct–Dec 2011 Jul–Sep 2012 Apr–Jun 2013 Jan–Mar 2014

Percentage of workers in self-employment

606

Source:  Number of self-employed from Labour Force Survey (ONS); income data from Family Resource Survey, authors’ compilation.

Figure 14.10 Trends in the share of workers in self-employment and real median weekly income, United Kingdom, 1993–2014 (percentage) 3.2  Wages: Falling Pay and Weak Worker Representation Our first factor – gross hourly pay – is a key variable in any explanation of what has happened to middle-­class incomes over the past two decades, given that wage income accounts for the largest portion of income among all British households (see also ILO 2015: fig. 29). However, for some years real wages have exerted a downward drag on income growth. Many commentators express concern about the problem of falling and stagnant pay levels, its alliance with low productivity growth and the absence of appropriate institutional levers to boost pay and productivity, such as through support for trade unions and worker voice in pay determination. Figure 14.11 presents trends in gross hourly pay for the main workforce groups (male/female, full-­time/part-­time) alongside gross domestic product (GDP) per capita and median household income for the period since 1997. It is notable that stagnation of real wage growth was established several years before the 2008–09 economic crisis. The year 2002 appears to be a key turning point; average real median wage growth during 2002–07 was just 0.2 per cent, compared with real GDP per capita average annual growth of

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 606

28/09/2016 08:25



How have middle-­income households fared in unequal Britain? ­607

135 130 125 120 115 110 105 100 1997 Q1 1997 Q3 1998 Q1 1998 Q3 1999 Q1 1999 Q3 2000 Q1 2000 Q3 2001 Q1 2001 Q3 2002 Q1 2002 Q3 2003 Q1 2003 Q3 2004 Q1 2004 Q3 2005 Q1 2005 Q3 2006 Q1 2006 Q3 2007 Q1 2007 Q3 2008 Q1 2008 Q3 2009 Q1 2009 Q3 2010 Q1 2010 Q3 2011 Q1 2011 Q3 2012 Q1 2012 Q3 2013 Q1 2013 Q3 2014 Q1

95

GDP per capita

Total pay

MFT pay

MPT pay

FFT pay

FPT pay

Median household income

Source:  GDP per head, chained volume measure (CVM, 2011 prices) from Office of National Statistics; real earnings from Annual Survey of Hours and Earnings (gross hourly median pay, all adult employees) and ONS Retail Price Index (2013 prices); real net median household income (2012–13 prices, Great Britain, before housing costs) from Institute of Fiscal Studies compilation of FES and Family Resources Survey data, accessed June 2015 at http://www.ifs.org.uk/tools_and_resources/incomes_in_uk; authors’ compilation.

Figure 14.11 Trends in real median hourly pay, GDP per capita and net household income, United Kingdom, 1997–2014 2.4 per cent. From this point UK households have witnessed a bifurcation between their living standards and the growth in GDP per capita, evident in the slowing down in growth of median household income and in stagnation of total pay. Since 2009,9 real pay at the median level has been in freefall, losing 10 per cent in real value over the next four years; the remarkable result is that, according to the Retail Price Index (RPI) measure of inflation, median gross hourly pay in 2013 is equivalent to the level earned in 1999. Overall, this is the longest period of falling real wages since wage records began in 1964 (Taylor et al. 2014). Male and female, full-­time and part-­time employees shared this general experience, although the greatest degree of stagnation and decline has been experienced by male full-­timers whose median real pay in 2013 in fact matched that in 1997. Moreover, relative to the top and bottom of the wage distribution, middle earners have fared the worst both before and since the crisis. From 1997 up to the crisis in 2008, median real pay had only increased by slightly

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 607

28/09/2016 08:25

608

Europe’s disappearing middle class?

more than 10 per cent compared with around 15 per cent at the bottom and top decile levels (while GDP per capita increased by around 32 per cent). Then, after 2009, real median pay sank to a level just 4 per cent higher than its 1997 value – £11.56 and £11.12, respectively – adjusting for inflation. Two reasons can be suggested for the relative gains (and smaller losses) experienced at the upper and lower ends of the wage distribution. High earners benefited most from employer practices to reward staff with bonuses, as well as other non-­pay rewards, such as stock options. Low earners, on the other hand, benefited from the upward pressure at the bottom exerted by the statutory minimum wage introduced in 1999 and especially the period 2003–07, when the Low Pay Commission acted purposefully to raise the relative value of the minimum wage. Workers’ absent voice in pay determination Britain is well known among its European neighbours for the limited capacity of workers to defend their pay collectively against private sector employers, so the weakly embedded institutions for collective bargaining are likely to have played a role. In the public sector, too, despite strong joint regulatory mechanisms to set pay, restrictive government rules since 2010 have restrained pay growth. The share of workers in unions has been diminishing for many years,10 especially among men: in the private sector, union density among men fell from 26 per cent in 1995 to 16 per cent in 2013 and among women the drop has been smaller but from a much lower starting point, from 15 per cent to 12 per cent. With such low levels of union representation, prospects for collective bargaining are of course restricted: the share of employees whose pay is agreed through negotiations between employer(s) and union(s) fell from 23 per cent to just 17 per cent over this period. These low levels of union membership are worse among the low and high paid, given that middle earners, especially men, are more likely to join a union (Figure 14.12). The withdrawal of state support for collective bargaining since the early 1980s and its fragmentation across workplaces and up and down the supply chain have led many advocates and academics in Britain to call for new policy reforms to reinvigorate its purpose and role in shaping wage settlements and other conditions of employment (for example, Ewing and Hendy 2013; Tailby and Moore 2014). However, the re-­election of a right-­ wing Conservative majority government in 2015 means that the prospects for joint negotiation of wages will certainly be worse over the next five years. As part of a new clamp-­down on union activities, the government has responded to lobbying from the private sector employer group, the Confederation of British Industry (CBI), to implement stricter rules on

File supplied to Vaughan-Whitehead for personal use - Not for sharing or posting VAUGHAN 9781786430595 PRINT (M4051).indd 608

28/09/2016 08:25



0

15.90

31.30

17.40

28.60

34.40

32.40 7.60

10

8.20

20

17.50

30

17.40

23.80

40

23.20

32.40

50

32.20

49.50

60

48.10

How have middle-­income households fared in unequal Britain? ­609

2010

2013

2010

Male

2013 Female