You are on page 1of 196

ISSN 1016-5444

Employment in Europe 2009

European Commission
Employment in Europe 2009

European Commission
Directorate-General for Employment, Social Affairs and Equal Opportunities
Unit D.1
Manuscript completed in October 2009
The “Employment in Europe 2009” report was prepared by the Directorate-General for Employment, Social Affairs
and Equal Opportunities; Employment Analysis Unit. Principal authors on this report include Matteo Governatori,
Magdalena Grzegorzewska, Christoph Maier, João Medeiros, Eric Meyermans, Paul Minty, Monika Šlebinger and Johan
Van der Valk (Eurostat). Important contributions were also provided by John Hurley (Eurofound) and Misa Labarile.

DG EMPL would also like to thank Eurostat and Eurofound’s European Monitoring Centre on Change for their close
collaboration and support in preparing the report. Comments from other services of the Commission are gratefully

Comments on the report would be gratefully received and should be sent to:
Unit D1
Directorate-General for Employment, Social Affairs and Equal Opportunities
Office J-27 05/80
B-1049 Brussels

Cover photos : © 123RF

For any use or reproduction of photos which are not under European Communities copyright,
permission must be sought directly from the copyright holder(s).

Europe Direct is a service to help you

find answers to your questions about
the European Union

Freephone number (*) :

00 800 6 7 8 9 10 11

(*) Certain mobile telephone operators do not allow access

to 00 800 numbers or these calls may be billed.

A great deal of additional information on the European Union is available on the Internet.
It can be accessed through the Europa server (

© European Communities, 2009

Reproduction is authorised provided the source is acknowledged.

Cataloguing data can be found at the end of this publication.

Luxembourg: Office for Official Publications of the European Communities, 2009

ISBN 978-92-79-13372-5
doi 10.2767/4837

Printed in Belgium

Printed on white chlorine-free paper

Chapter 5 The labour income share in the European Union

Responding to the crisis and preparing our labour markets for the future
The Commission’s annual Employment in Europe report, the 21st in the series, comes against a backdrop this year of
quite exceptional economic circumstances. Hot on the heels of last autumn’s global financial crisis, the worst economic
downturn Europe has seen since World War II has brought several years of relatively high economic growth and job
creation to an abrupt halt and thrown far too many businesses, households and workers into serious difficulties.

The European Union reacted swiftly to the financial and economic crisis, taking the steps necessary to prevent a meltdown
of the financial markets and adopting a European Economic Recovery Plan. With Europe’s labour markets already deeply
affected by the challenges of globalisation, technological change, ageing societies and climate change, the ensuing employ-
ment crisis has heightened the need for policies to help people keep their jobs or get them back into employment quickly.
The Union accordingly acted to stabilise labour markets, taking measures focused on maintaining existing employment and
creating new jobs, improving workers’ skills and matching labour-market demand and supply more closely.

In these turbulent times, the Commission has placed special emphasis on the monitoring and analysis of short-term develop-
ments and policy action. At the beginning of this year it launched the monthly Labour Market Monitor, a new short-term
monitoring tool that provides a useful guide for EU and Member State policymakers. The onset of the crisis should not mean,
however, that the broader structural issues affecting the EU labour market can be neglected. On the contrary, the Commission
must make sure that short-term policy measures are not at odds with long-term structural reforms, which in turn are a prereq-
uisite for the EU economy and labour markets to emerge well prepared for future challenges from the current downturn.

Bearing that challenge in mind, the 2009 Employment in Europe report takes a deeper look at two issues that are
important for EU employment policy in the future: the dynamics of European labour markets and the implications of
climate change for labour-market outcomes. A clearer understanding of labour-market dynamics is critical in a time of
crisis, when prompt policy responses are crucial. Measures to get laid-off workers back into employment and to curb
long-term unemployment are hugely important.

Meanwhile, climate change and the inescapable need to shift to a competitive low-carbon economy have become pri-
orities for urgent action. There is significant scope in Europe for creating new ‘green’ jobs and for ‘greening’ existing
jobs in many sectors and professions. But if these opportunities are to be grasped, the right policies, based on a sound
understanding of the key trends underlying efforts to respond to and mitigate the impact of climate change, must be
put in place. Careful analysis being a key ingredient of good policymaking, I am confident that this year’s Employment
in Europe report will provide useful insights for the employment policy debate.

Vladimír Špidla
Commissioner for Employment, Social Affairs and Equal Opportunities
Chapter 5 The labour income share in the European Union

Table of contents
Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

List of acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

Chapter 1: EU labour markets in times of economic crisis . . . . . . . . . . . . . . . . . 11

1. Introduction – the new economic context for EU labour markets . . . . . . . . . . . . . . . . . . . . . 11

2. Panorama of EU labour markets in 2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

2.1. EU labour market from a global perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2. Labour market situation in the EU . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2.1. Employment growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2.2. Employment rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.3. Contractual arrangements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.4. Unemployment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3. Labour market situation in the Member States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3.1. Employment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3.2. Unemployment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3.3. Contractual arrangements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3. The recent labour market downturn and its intensification since
the deepening of the financial crisis last autumn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.1. Economic activity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.1.1. Developments in GDP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2. How has the labour market adjusted to the economic downturn? . . . . . . . . . . . . . . . . . 25
3.2.1. Labour demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.2.2. Employment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.2.3. Unemployment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.2.4. Other labour market responses to the economic downturn . . . . . . . . . . . . . . . . . 33
3.3. Which population subgroups have been most affected? . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.3.1. Impact across various population subgroups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.4. What measures have been taken to mitigate
the impact of the crisis on labour markets? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.4.1. EU-level initiatives to promote employment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.4.2. Recent employment measures undertaken
by Member States to combat the employment effects of the crisis . . . . . . . . . . . 41
3.5. Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.5.1. Forecasts for the outlook to 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4. Summary and conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

Employment in Europe 2009

Chapter 2: Labour flows, transitions and unemployment duration . . . . . . . . . . 47

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

2. Labour flows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.1. Labour flows and transitions in the labour market . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.2. Job and labour flows - definitional and measurement issues . . . . . . . . . . . . . . . . . . . . . 50
2.3. The impact of cycle, firms’ and workers’ characteristics . . . . . . . . . . . . . . . . . . . . . . . . . 52
2.3.1. Other features of labour turnover: concentration,
persistence, role of dismissals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
2.3.2. The link between turnover and productivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
2.4. The size and characteristics of labour market flows in the EU . . . . . . . . . . . . . . . . . . . . 55
2.4.1. Time trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.4.2. Breakdowns by gender, age and education level . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.4.3. Institutions and labour turnover . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3. Labour market transitions in the EU . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.1. From flows to transitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.2. An empirical analysis of labour market transitions in the EU . . . . . . . . . . . . . . . . . . . . . 63
3.2.1. Trend versus cyclical components of transition rates . . . . . . . . . . . . . . . . . . . . . . 64
3.2.2. Country-by-country trend developments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
3.2.3. Breakdowns by gender, age and education level . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.2.4. A brief analysis of all labour market transitions, based on net flows . . . . . . . . . 70
4. Unemployment duration and long-term unemployment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4.1. Measurement issues with regard to LTU . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.1.1. Administrative versus survey data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.1.2. Duration of incomplete versus completed unemployment spells . . . . . . . . . . . . . 74
4.1.3. Longitudinal data: the importance of using detailed calendar information . . . . 74
4.1.4. Consideration of a broader concept of ‘joblessness’ . . . . . . . . . . . . . . . . . . . . . . . 75
4.1.5. Main characteristics of LTU . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.2. Relation over the economic cycle between unemployment and the incidence of LTU . 77
4.3. Statistics on the average duration of both completed
and incomplete unemployment spells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
4.4. Unemployment duration by reason of leaving/losing employment . . . . . . . . . . . . . . . . 83
4.5. Differences between the average duration
of CDU and IDU spells over the economic cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
4.6. Survival rates in unemployment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
4.7. Do CDU statistics and LTU rates convey the same type of information? . . . . . . . . . . . . . 88
4.8. Methods used during the previous four weeks to find work . . . . . . . . . . . . . . . . . . . . . 88
4.9. Use of longitudinal data to evaluate the incidence
of LTU and the recurrence of unemployment spells . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
4.10. Brief description of methods to estimate unemployment duration . . . . . . . . . . . . . . . . 93
4.11. Elasticities of unemployment duration to benefits
and duration dependence of hazard rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
4.12. Labour market institutions and unemployment duration . . . . . . . . . . . . . . . . . . . . . . . . 95
4.13. The consequences of LTU . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

Chapter 3: Climate change and labour market outcomes . . . . . . . . . . . . . . . . 105
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

2. Climate change and green jobs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

2.1. The challenge of climate change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
2.2. Total environment-related employment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
2.2.1. Definition and measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
2.2.2. Employment in the renewable energy sector . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
2.2.3. Analytical tools for the economic analysis of climate change . . . . . . . . . . . . . . . 113
3. Employment effects of adaptation to climate change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
3.1. The agricultural sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
3.2. The tourism sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
3.3. The power sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
3.4. The insurance sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
4. Employment effects of the transition towards a low-carbon economy . . . . . . . . . . . . . . . . 115
4.1. Improved energy efficiency and jobs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
4.1.1. Energy efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
4.1.2. Overall labour market outcomes of improved energy efficiency . . . . . . . . . . . . 117
4.1.3. Sectoral labour market outcomes of improved energy efficiency . . . . . . . . . . . 118
4.2. More renewable energy and jobs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
4.2.1. Employment impacts of RES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
4.2.2. Future labour market outcomes from renewable deployment . . . . . . . . . . . . . . 122
4.3. Carbon capture and storage and jobs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
4.4. Impacts on skills . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
5. Implications for labour market policies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
5.1. Employment policies in the climate change context . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
5.2. Assessment of climate policies against their employment impact . . . . . . . . . . . . . . . . . 129
5.3. The need for clear, predictable and coordinated climate policies . . . . . . . . . . . . . . . . . 129
6. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130

Statistical annex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

1. Macro economic indicators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

2. Labour market indicators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154

Introduction to labour market indicators tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
Notes for particular Member States/ tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
3. Data sources and definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186

List of acronyms
ADP Accelerated deployment policies European Globalisation Adjustment
ALMP Active labour market policy
EGSS Environmental goods and services sector
BAU Business as usual
E_I Employment to inactivity
CA Cluster analysis
EJT Excess job turnover
CCS Carbon capture and storage
ELT Excess labour turnover
CDU Completed duration of unemployment
EMAS Eco-Management and Audit Scheme
CF Churning flow
EMCO Employment Committee
CHP Combined heat and power
Energy Performance of Buildings Direc-
CL Closing tive

CO Contracting EPL Employment protection legislation

CSR Corporate social responsibility ERM European Restructuring Monitor

Directorate-General for Economic and ESF European Social Fund

Financial Affairs
EU European Union
Directorate-General for Employment,
Social Affairs and Equal Opportunities E_U Employment to unemployment

Directorate-General for Transport and All Member States that joined the EU
Energy on 1 May 2004

E Expanding All Member States that joined the EU

EU-12 on 1 May 2004 and 1 January 2007
ECHP European Community Households Panel (EU-10 + EU-2)

EEA European Economic Area All Member States forming part of the
EU before 1 May 2004
EERP European Economic Recovery Plan
All Member States forming part of the
EES European Employment Strategy EU before 1 January 2007

Employment in Europe 2009

EU-27 All EU Member States Non-accelerating wage rate of unem-

Belgium, Germany, Denmark, Greece,
Spain, Italy, Portugal and the UK NEG Net employment growth

EU ETS EU Emissions Trading Scheme NGV Natural gas vehicles

EURES European Employment Services O Opening

European Union Statistics on Income Organisation for Economic Co-opera-

and Living Conditions tion and Development

FTE Full-time equivalent OLS Ordinary least squares

GDP Gross domestic product PCA Principal component analysis

GHG Greenhouse gases PES Public employment services

H Hiring PPS Purchasing power standards

HP Hodrick-Prescott R&D Research and development

I Inactive RES Renewable energy sources

IDU Incomplete duration of unemployment S Separation

I_E Inactivity to employment SME Small and medium-sized enterprise

ILO International Labour Organization SSGI Social services of general interests

Intergovernmental Panel on Climate STWA Short-time working arrangements

TE Temporary employment
JC Job creation
TFP Total factor productivity
JD Job destruction
U Unemployed
JT Job turnover
U_E Unemployment to employment
LFS Labour Force Survey
UK United Kingdom
LPG Liquefied petroleum gas
United Nations Environmental Pro-
LT Labour turnover gramme

LTC Long-term care United Nations Framework Convention

on Climate Change
LTU Long-term unemployment
USA United States
METR Marginal effective tax rate
USB Users’ database
Non-accelerating inflation rate of
unemployment VAR Vector autoregressive systems

Chapter 1

EU labour markets in
times of economic crisis
1. Introduction – the begun to rise across all Member States. Acting in concert, the EU has already
new economic context In a number of European countries, taken important steps to address the
job losses have been rather restrained
for EU labour markets
fallout from the crisis, having taken
to date, largely due to recourse to action to prevent a meltdown in the
increased internal flexibility in the financial markets last autumn. In
The unprecedented crisis in global form of shorter hours or temporary December it agreed to put in place a
financial markets which gathered pace partial unemployment. However, even European Economic Recovery Plan(2)
in autumn last year has led to the most if labour markets have proven to be to lessen the effects of the downturn
severe recession since the Second World more resilient, the European Union and create the conditions for recovery.
War, affecting the wider economy and (EU) is still expected to lose some The top employment challenge for the
increasingly impacting on labour mar- 8.5 million jobs over 2009–10, with EU must be to minimise job losses, pre-
kets in the EU. After many years of unemployment potentially reaching vent unemployment from becoming
relatively high growth and job creation around 11% by 2010.(1) Indeed, histori- entrenched (i.e. becoming long-term
(9.7 million new jobs alone between cal experience shows that employment unemployment), favour transitions
2005 and 2008) – taking Europe back reacts to economic conditions with a back into employment and boost job
to employment levels not seen for dec- certain lag; hence labour market con- creation, and pave the way for econom-
ades – the global financial crisis and its ditions can be expected to worsen for ic renewal and for sustainable recovery
repercussions on the real economy are some time even after the trough in the and growth. This requires stronger
hitting businesses, jobs and households. economic situation has been reached. cooperation between all stakeholders,
They are thus increasingly affecting the better policy coordination and mutual
prospects and livelihoods of European At the same time, the crisis appears to learning – i.e. with a shared commit-
citizens. The sudden reversal of the be affecting some groups of workers ment to develop and implement the
previous period of employment growth more deeply than others. Although right policies and actions: to preserve
has set new challenges to policy-making men still have higher employment sustainable jobs in sound economic
and research. As unemployment con- rates than women, to date the former activities and help people into produc-
tinues to rise, the spotlight has fallen have been more affected by the tive employment; to support the most
more and more on limiting the effect downturn than the latter, reflecting vulnerable; and to prepare for the jobs
of the crisis on jobs and addressing the that many of the sectors hit hardest and skills of the future.
social impact. by the crisis are predominantly male-
oriented in terms of employment. In line with the European Economic
Although the picture varies across There has also been a continued Recovery Plan, the agreed fiscal stimu-
Member States, the crisis is expected strong rise in unemployment among lus and an acceleration of structur-
to have significant consequences for young people, with young men being al reforms will help boost demand,
all of their labour markets; for many particularly affected, highlighting a restore confidence and ensure that
this will manifest itself as a substan- rising need for support to tackle Europe emerges more strongly from
tial increase in unemployment. Initially youth unemployment. the crisis. Accelerating recovery must
the bulk of the negative impact on
labour markets was concentrated in (1) European Commission, ‘Economic fore- (2) European Commission, A European Eco-
Spain and the United Kingdom (UK), cast’, spring 2009, European economy, nomic Recovery Plan, COM(2008) 800, avail-
3/2009, Directorate-General for Economic able at:
but more recently unemployment has
and Financial Affairs. barroso/president/pdf/Comm_20081126.pdf

Employment in Europe 2009

be assisted by structural reforms to cre- shown the need for further research ­ roduct (GDP), leading to rather lim-
ate more flexible, secure and inclusive on ways to limit the negative impact of ited annual GDP growth of only 0.9%
labour markets. The Commission Com- the crisis on labour markets and better – much lower than the 2.9% growth
munication ‘Driving European recov- position them so they are well placed rate achieved in 2007 (Chart 1 and
ery’(3) outlined a number of elements to respond to the recovery when it Table 1). The economic crisis especially
to help Member States design and comes and to prepare them against affected output in industry, construc-
implement appropriate and effective any future crises. tion and retail trade. Indeed, by mid-
employment policies. On this basis, the 2008, these sectors had undergone a
Spring Council and the three employ- Even in these turbulent times, it is still decrease in production that became
ment workshops held in Madrid (ES), worthwhile to present the longer-term substantial by the end of the year.
Stockholm (SE) and Prague (CZ) in April picture to highlight the progress that
2009 helped define three key priori- has been made in European labour By end-2008, over half of the EU
ties: maintaining employment, creating markets over the last decade up until Member States were either in reces-
jobs and promoting mobility; upgrad- the global crisis hit last year (see sec- sion or in the process of entering
ing skills and matching labour market tion 2 below); however, in view of one: 18 of the EU-27 had entered
needs; and increasing access to employ- the rapidly changing situation, this recession – namely Austria, Den-
ment. The 7 May Employment Summit year’s report also presents a more mark, Estonia, ­Finland, France, Ger-
allowed for an exchange of views on up-to-date picture of the short-term many, Hungary, Ireland, Italy, Latvia,
these priorities and found common developments in labour markets since ­Lithuania, Luxembourg, the Neth-
ground on concrete actions.(4) last spring (section 3). erlands, Portugal, Romania, Spain,
Sweden and the UK – with the
Building on this mutual effort, the economy continuing to expand only
Commission recently published its Com-
munication entitled ‘A shared commit- 2. Panorama of EU in Cyprus, Greece and Slovakia.

ment for employment‘.(5) This aims to labour markets in 2008 The economic growth for the EU’s
strengthen cooperation between the main trading partners also showed a
EU and its Member States, as well decrease in 2008 compared with the
as among EU social partners, on the 2.1. EU labour market preceding year. For the United States
three key priorities mentioned above.
from a global perspective (USA), economic growth declined to
It focuses on concrete actions and is 1.1%, down from 2.0% in 2007 and
supported by all available Community from 2.8% in 2006. Over each of the
instruments, particularly the Europe- In 2008, the EU economy clearly suf- last two years, the USA has shown
an Social Fund and the Globalisation fered from the global economic down- pronounced declines in annual growth,
Adjustment Fund. turn which emanated from the crisis in while in the EU growth underwent a
financial markets and which resulted very fast decline in 2008. The latter pat-
Last but not least, the new economic in the EU having entered a reces- tern of a sharp deterioration in 2008
context arising from the global crisis sion by the end of the year. Over the was even more pronounced for Japan,
has highlighted a need for more up- course of 2008, expansion in the first where growth of –0.7% was recorded
to-date monitoring and analysis of the quarter was followed by three quar- for 2008, in comparison with the strong
labour market situation, which has ters of contraction in gross domestic 2.3% growth of the year before.
been addressed in part by the Com-
Chart 1: GDP growth in the EU, USA and Japan, 1998–2008
mission’s publication of a new monthly
monitoring report.(6) In addition, it has 5

(3) European Commission, Driving European EU-27

recovery, COM(2009) 114 final, available at: 4 US­/ JP
president/pdf/press_20090304_en.pdf 3
(4) For the Key Messages of the May 7 Sum-
mit, and a report on the three Work-
Annual % change

shops, see:
173&furtherEvents=yes 1

(5) European Commission, A shared commit-

ment for employment, COM(2009) 257 0
final, available at:
social/BlobServlet?docId=2798&langId=en -1

(6) Available on the website of the

Directorate-General for Employment, -2
Social Affairs and Equal Opportuni- 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
ties at
jsp?catId=120&langId=en Source: Eurostat, national accounts.

Chapter 1 EU labour markets in times of economic crisis

Table 1: International comparison of key indicators, 2006–08 The deteriorating economic climate
already affected the labour market in
2006 2007 2008 2008, with employment growth in the
Population (millions)
EU declining to 1.0% compared with
EU-27 493 495 497
1.8% in 2007 (Chart 2). However, the
EU-15 390 392 394
slowdown in employment growth
USA 299 302 304
Japan 128 128 128
was less pronounced than that in
GDP (in 1000 million PPS, current prices) GDP. This partly reflects the fact that
EU-27 11 684 12 360 12 512 labour markets usually respond with
EU-15 10 371 10 927 10 993 some delay to economic trends but
USA 10 927 11 416 11 560 also the fact that labour demand
Japan 3 400 3 568 3 558 started to adjust through flexible
Real GDP Growth (annual % change) working arrangements (e.g. shorter
EU-27 3.2 2.9 0.9
working hours) rather than through
EU-15 3.0 2.6 0.6
a reduction in employment.(7) As a
USA 2.8 2.0 1.1
result, EU employment growth in
Japan 2.0 2.3 -0.7
Employment Rate (as % of working age population)
2008 remained considerable given
EU-27 64.5 65.4 65.9 the extent of the economic down-
EU-15 66.2 67.0 67.3 turn, being substantially higher than
USA 71.9 71.7 70.9 that in the USA and Japan.
Japan 69.9 70.6 70.7
Employment Growth (annual % change) In the USA, the continued adverse
EU-27 1.6 1.8 1.0 economic situation, with GDP growth
EU-15 1.5 1.6 0.7 decreasing for the second year in a
USA 1.9 1.1 -0.5
row, had extensive repercussions for
Japan 0.4 0.4 -0.4
the labour market. Negative employ-
Unemployment Rate (as % of civilian labour force)
EU-27 8.2 7.1 7.0
ment growth of –0.5% was record-
EU-15 7.7 7.0 7.1 ed in 2008 (compared with positive
USA 4.6 4.6 5.8 growth of 1.1% in the preceding
Japan 4.1 3.9 4.0 year) similar to the level of employ-
Source: GDP and employment growth from national accounts, Eurostat (GDP for the ment contraction in Japan where
USA according to the System of National Accounts and employment growth for Japan employment growth also dropped
from AMECO database, Commission Services). GDP in purchasing power standards sharply in 2008, with a negative rate
(PPS) from national accounts, Eurostat. Employment rate from Eurostat (EU LFS) and of –0.4% being recorded. Over the
OECD data for USA and Japan. Unemployment rate from Eurostat series on unemploy-
ment. Population from demographic statistics, Eurostat, and for USA and Japan from last 10 years, employment growth
AMECO database, Commission Services. in Japan has been persistently lower
Note: Employment rates for the EU and Japan refer to persons aged 15–64; than that in the EU and USA.
USA employment rate refers to persons aged 16–64.
Growth in labour productivity (per
person employed) was slightly posi-
tive for the EU in 2008 (0.4%), over
Chart 2: Employment growth in the EU, USA and Japan, 1998–2008
one percentage point lower than
2.5 the previous year (Chart 3). This
was the result of a substantial slow-
down in economic growth while
1.5 employment levels increased slight-
ly. For Japan the downward trend in
labour productivity was even more
0.5 significant. Having been higher in
Annual % change

recent years compared with the EU,

productivity growth dropped to
-0.5 –0.3% in 2008. Productivity growth
in the USA rose to 1.6% in 2008,
-1.0 EU-27
US up from 0.9% the year before.
-1.5 JP

-2.0 (7) Employment growth is based on the

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 number of employed persons regardless
of their part- or full-time status, and not
Source: Eurostat, national accounts. on full-time equivalent employment.

Employment in Europe 2009

Chart 3: Growth in productivity per person employed This development was driven by
in the EU, USA and Japan, 1998–2008 the relatively strong contraction in
employment in conjunction with
positive GDP growth in the USA.
3.0 EU-27
2.5 JP
As indicated, compared with the year
before, total employment continued
to expand in the EU in 2008, while the
Annual % change

1.5 deteriorating economic situation and

consequent weakening of the labour
market was not yet fully reflected in
0.5 the annual EU labour market indica-
tors for that year. The EU employment
rate – i.e. the share of the population
-0.5 aged 15–64 years (the working-age
population) in employment – amount-
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 ed to 65.9% in 2008, up 0.5 percent-
age points compared with 2007. This
Source: Eurostat, national accounts.
trend contrasts with that in the USA,
where the share of the working-age
population in employment dropped
Chart 4: Employment rates in the EU, USA and Japan, 1975–2008 from 71.8% to 70.9%. This resulted
in virtually the same rate as in Japan,
where it remained broadly stable.
Even though the employment rate of
70 those countries remains considerably
% of working-age population

higher than that of the EU, it can be

noted that, over the last 10 years, the
EU rate shows a clear upward trend in
contrast to that of the USA (Chart 4).

During 2008 some 16.7 million per-

60 EU-27
sons – or 7.0% of the labour force
US – were unemployed in the EU-27.
This rate is very similar to that for
55 the preceding year but marks a halt
1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
in the decline in unemployment that
Source: DG EMPL calculations based on long-term trends in employment
had been observed since 2004. In
and population, Commission Services.
Japan a similar situation could be
seen in 2008, as the unemployment
rate increased only marginally to
Chart 5: Unemployment rates in the EU, USA and Japan, 1998–2008
4%. In the USA, however, unemploy-
10 ment grew strongly in 2008, rising
1.2 percentage points on the year
before to 5.8%, but still more than
8 one percentage point lower than the
EU-27 (Chart 5).
% of labour force

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

Source: Eurostat, series on unemployment.

Chapter 1 EU labour markets in times of economic crisis

2.2. Labour market Table 2: Change in employment in the EU by age, gender,

situation in the EU and type of employment, 2007–08 (% growth)

Relative (as % of 2007 level)

Total 1.0
2.2.1. Employment growth
Men 0.6
Employment in the EU increased by Women 1.5
2.3 million between 2007 and 2008. Age
However, the rise was not uniform 15-24 0.4
with respect to gender, age and type 25-54 0.6
of employment. The growth rate for 55-64 3.8
female employment was almost three 65+ 4.5
times that for male employment. With Type of employment
Employee versus self-employed Employee 1.2
regard to age, growth was strongest
Self-employed -0.1
for older workers, where employment
Full-time versus part-time Full-time job 1.0
increased by around 4%, much greater
Part-time job 1.2
than the 0.6% growth rate for prime- Permanent versus fixed-term employees Permanent 1.6
age workers (25–54 years) and 0.4% Fixed-term -2.7
for youth (15–24). In terms of type of
Source: Eurostat, national accounts, EU LFS and DG EMPL calculations.
employment, the relative growth in
part-time employment was similar to
that for full-time jobs, while the number
of employees with temporary contacts workers over this period was espe- atively small share ­associated with
decreased by around 3% in contrast to cially noteworthy (Table 3). Older ­self-employment. Part-time employ-
the 1.6% rise in employment for those workers aged over 54 accounted for ment made a significant contribution
with permanent contracts (Table 2). 40% of employment expansion, with to employment expansion, account-
broadly equal contributions from ing for 37% of the net rise in employ-
Taking a longer-term perspective, the men and women in this age group. In ment, although full-time jobs still
number of people in employment in contrast, the contribution of young accounted for the majority of the
the EU expanded by around 17 mil- people aged 15–24 to employment expansion (63%). Permanent jobs
lion between 2000 and 2008. Prime- expansion was negligible, in part accounted for almost three quarters
age workers aged 25–54 accounted reflecting the trend for youth to of employment growth, with fixed-
for almost 60% of the net increase stay longer in education. In terms term jobs accounting for a much
in employment, with women making of type of employment, almost 90% smaller but still important share,
a higher contribution than men. In of employment growth was made although there is a large underlying
addition, the contribution from older up of employees, with only a rel- heterogeneity across Member States.

Table 3: Contribution to net employment creation in the EU by age, gender and type of employment, 2000–2008

% contribution to employment creation 2000–08

Total Men Women
Age and gender
Total 41.0 59.0
15–24 1.3 0.9 0.3
25–54 58.2 19.1 39.1
55–64 37.9 18.9 19.0
65+ 2.6 2.0 0.6
Type of employment and gender
Employee versus self-employed Employee 89.8 34.8 55.0
Self-employed 10.2 5.7 4.5
Full-time versus part-time Full-time job 63.0 32.8 30.2
Part-time job 37.0 10.1 27.0
Permanent versus fixed-term employees Permanent 72.4 26.8 45.6
Fixed-term 27.6 12.3 15.3

Source: Eurostat, EU LFS.

Note: Data for RO 2002 instead of 2000. Data for full-time/part-time and permanent/ temporary indicators
for BG 2001 instead of 2000.

Employment in Europe 2009

2.2.2. Employment rate 25–54 also rose considerably, up 6 ket compared with younger gen-
percentage points compared with erations, as well as the difficulties in
The strong rise in the employment 2000 (Chart 6). re-entering the labour market after a
rate between 2000 and 2008 was long period of absence.
driven by specific subgroups of the In the EU-27, labour market partici-
population. For young persons and pation varies strongly by gender and While the gender and age gap in
men aged 25–54, employment rates age, with women and older persons employment rates is shrinking, employ-
remained broadly stable over this having a substantially lower employ- ment rates of women and older per-
period, and similarly showed little ment rate than average (Chart 7). In sons remain substantially lower than
change in 2008. In contrast, those 2008, the employment rate of women those of prime working-age men
among older men increased substan- aged 15–64 amounted to 59%, while (i.e. men aged 25–54). Indeed, the
tially (in 2008, 55% of men aged for men it was almost 73%. This gen- employment rate for men in the prime
55–64 were in employment com- der gap in employment rates exists working-age group is almost 15 per-
pared with 47% in 2000), while for for all age categories but is larg- centage points higher than that of
older women even larger rises were est for older persons. This reflects similarly aged women, while, in the
visible (up more than 9 percent- the fact that older generations of older age group, the gender gap was
age points, from 27.4% to 36.9%). women, in particular mothers, used even higher at 18 percentage points.
Employment rates for women aged to participate less in the labour mar- The difference in employment rates
between the prime working-age and
Chart 6: Employment rates in the EU by age group and gender, 2000–2008 older age group is even bigger. The
employment rate for people aged
55–64 is more than 30 percentage
points lower than that for those aged
25–54 – less than 46% of persons aged
55–64 were working compared with
% of respective population

almost 80% for persons aged 25–54.

Employment rates for the young are
60 also relatively low – less than 38%
of those aged 15–24 were in work in
2008, reflecting the fact that many are
still in full-time education.

15–24 Total
25–54 Men 55–64 Men
25–54 Women 55–64 Women 2.2.3. Contractual
2000 2001 2002 2003 2004 2005 2006 2007 2008 arrangements
Source: Eurostat, EU LFS.
In 2008 about 18% of those in
employment were working on a
part-time basis in the EU-27. After
Chart 7: Employment rates in the EU by age group and gender, 2008
rising by around 2 percentage points
100 between 2002 and 2006, this share
has remained more or less stable in
recent years. Part-time work remains
more common in the older Member
States than in the newly acceded
% of respective population

60 countries (and has been increasing

at a higher rate in the EU-15), with
the share for the EU-15 considerably
40 higher than that for the EU-27.

Some 14% of employees had a fixed-

term contract in 2008, represent-
ing a decrease of 0.4 percentage
0 points compared with 2007. While
15–24 25–54 55–64 Total 15–64
the change in the percentage of
Total 37.6 79.6 45.6 65.9 part-time workers seems to indicate
a long-term upward trend, the per-
Source: Eurostat, EU LFS.
centage of workers with a fixed-term

Chapter 1 EU labour markets in times of economic crisis

Chart 8: Part-time and fixed-term contracts, and self-employment in the EU, 2000–2008 This highlights one early distinguish-
ing feature of the current economic
downturn – namely that it has had
a noticeably larger impact on the
18 labour market situation of men than
% of employment / employees (fixed-term)

on that of women. This reflects the

fact that, so far, many of the sectors
hit hardest by the downturn – such
as construction, the car industry and
transport and storage – are pre-
dominantly male-oriented in terms
of employment.
12 Fixed-term employees
Part-time employed
2.3. Labour market
2000 2001 2002 2003 2004 2005 2006 2007 2008 situation in the Member
Source: Eurostat, EU LFS. States

Chart 9: Unemployment rates in the EU by gender, 2000–2008

2.3.1. Employment
10 Employment growth
Although employment continued to
9 expand in the EU in 2008, the rate of
growth decreased in the vast major-
ity of Member States compared with
% of labour force

2007. Furthermore, large differences

were visible between the Member
States. Employment growth was nega-
tive for Hungary, Ireland, Lithuania
7 and Spain in 2008, with the downturn
in the labour market brought about by
the deterioration in the economic situ-
6 ation being particularly severe in the
2000 2001 2002 2003 2004 2005 2006 2007 2008
last three countries, which had experi-
Source: Eurostat, EU LFS. enced employment growth of around
3% in 2007. Employment growth in
2008 was rather limited in Estonia,
contract is more sensitive to the eco- 2.2.4. Unemployment Romania, Italy and Portugal, at less
nomic situation. The latter increased than 0.5%, while in Latvia – a country
from 2003 until 2007, but declined in Unemployment in the EU-27 with one of the highest growth rates
2008, clearly following the business remained more or less stable in in the EU in previous years – employ-
cycle (Chart 8). 2008, as the deterioration in the ment growth dropped by almost three
economic situation brought an percentage points in 2008.
The extent of self-employment abrupt end to the strong down-
remained stable in 2008 compared ward trend in unemployment of Nevertheless, several Member States
with the year before, with 16% the previous three years (Chart 9). still recorded relatively high employ-
of workers being self-employed, However, the situation varies some- ment growth in 2008. Luxembourg
although the share has fallen since what according to gender. For men and Poland posted growth of 4%
2000 by the order of a percentage the unemployment rate in 2008, at or more, similar to the year before,
point. This suggests that the increase 6.6%, was exactly the same as the while employment growth in Bul-
in employment over that period, year before, while that for women garia increased to 3.3%. Among the
which was generally accounted for continued to decrease (falling to larger Member States, France record-
by women and older persons, were 7.5%, compared with 7.8% the year ed a substantial slowdown in its rate
mainly jobs as employees rather than before), although at a substantially of employment expansion, with the
as self-employed. slower pace. growth rate falling by 0.8 ­percentage

Employment in Europe 2009

points to around 0.5%, while the The Lisbon target of a 60% employ- This situation reflects the fact that
rate in Italy showed a similar devel- ment rate for women, however, within most Member States the gen-
opment. For Germany and the UK was much closer to being reached der gap in employment rates is still
the situation more or less stabilised in 2008, with 59.1% of working- substantial (Chart 11). This is par-
at around 1.5% and 0.7% respec- age women in work – a shortfall ticularly the case in Malta, Greece
tively (Chart 10). of less than 1 percentage point. and Italy, where the employment
Since 2000, considerable progress rate for men remains more than 20
Employment rates has been made in expanding female percentage points higher than that
employment, with the employment for women. In a further 15 Member
In the EU, the employment rate of the rate for women increasing by more States, the gap varies between 10%
working-age population (i.e. those than 5 percentage points over that and 20%. In contrast, in Sweden
aged 15–64) reached 65.9% in 2008, period. In 2008 the employment rate and Finland the employment rate for
an increase of 0.5 percentage points of women grew by 0.8 percentage men is less than 5 percentage points
compared with the preceding year. points compared with 2007, while higher than that for women.
The rate has risen by 3.7 percentage that for men rose only 0.3 percent-
points since 2000, when the Lisbon age points, thereby further narrow- The employment rate for older per-
target (Box 1) of an overall employ- ing the gender gap in employment. sons aged 55–64 in the EU increased
ment rate of 70% was set, leaving a Nonetheless, it remains considerably further in 2008, rising 1 percentage
gap of around 4 percentage points lower than the male employment point compared with 2007, but at
yet to be filled (Table 4). rate of 72.8%. 45.6% the rate remains relatively low.
Although the rate has risen substan-
tially since 2000, increasing by almost
9 percentage points, the current per-
Chart 10: Employment growth for Member States, 2007 and 2008 formance falls short of the target set
5 by the 2001 Stockholm Council of an
2008 employment rate of 50% by 2010 by
4 2007 4 percentage points.

Annual % change

Box 1: The Lisbon and Stockholm


1 The 2000 Lisbon European Council set a

strategic goal, over the decade 2000–10,
0 for the EU to become the most competi-
tive and dynamic knowledge-based econ-
omy in the world, capable of sustainable
LU PL BG SK SI CY MT AT BE FI NL DE EL EU-27 SE CZ DK LV UK FR PT IT RO EE LT HU ES IE economic growth with more and better
Source: Eurostat, national accounts. jobs and greater social cohesion.

It specifically stated that the overall aim

of employment and economic policies
Chart 11: Employment rates for Member States by gender, 2008 should be to raise the employment rate
to as close to 70% as possible by 2010,
and to increase the employment rate for
90 women to more than 60% by the same
year, not least in order to reinforce the
sustainability of social protection systems.
70 In addition to the 2010 Lisbon targets,
% of working-age population

the 2001 Stockholm European Council
set a new target of raising the average
50 EU employment rate for older men and
women (aged 55–64) to 50% by 2010.

In 2008, only eight Member States
recorded an employment rate of
10 more than 70%, the overall Lisbon
0 target (Chart 12) – namely Denmark
(78.1%), the Netherlands (77.2%),
Source: Eurostat, EU LFS. Sweden (74.3%), Austria (72.1%), the

Chapter 1 EU labour markets in times of economic crisis

Table 4: Employment rates in EU Member States in 2008 and progress towards Lisbon and Stockholm targets for 2010

Total employment rate Female employment rate Older people's employment rate
Gap Gap Gap
Change Change below Change Change below Change Change below
2008 2008 2008
2008–07 2008–00 2010 2008–07 2008–00 2010 2008–07 2008–00 2010
target target target
BE 62.4 0.4 1.9 7.6 56.2 0.8 4.7 3.8 34.5 0.1 8.2 15.5
BG 64.0 2.2 13.5 6.0 59.5 1.9 13.2 0.5 46.0 3.5 25.2 4.0
CZ 66.6 0.5 1.6 3.4 57.6 0.3 0.7 2.4 47.6 1.6 11.3 2.4
DK 78.1 0.9 1.8 > 74.3 1.0 2.6 > 57.0 -1.6 1.3 >
DE 70.7 1.4 5.2 > 65.4 1.5 7.3 > 53.8 2.3 16.2 >
EE 69.8 0.4 9.4 0.2 66.3 0.4 9.4 > 62.4 2.4 16.1 >
IE 67.6 -1.5 2.4 2.4 60.2 -0.4 6.3 > 53.6 -0.2 8.3 >
EL 61.9 0.5 5.4 8.1 48.7 0.9 7.0 11.3 42.8 0.3 3.8 7.2
ES 64.3 -1.3 8.1 5.7 54.9 0.2 13.6 5.1 45.6 1.0 8.6 4.4
FR 65.2 0.6 3.1 4.8 60.7 0.7 5.5 > 38.3 0.0 8.4 11.7
IT 58.7 0.1 5.0 11.3 47.2 0.6 7.7 12.8 34.4 0.7 6.8 15.6
CY 70.9 -0.1 5.2 > 62.9 0.4 9.3 > 54.8 -1.1 5.4 >
LV 68.6 0.3 11.2 1.4 65.4 1.0 11.7 > 59.4 1.8 23.4 >
LT 64.3 -0.6 5.3 5.7 61.8 -0.4 4.0 > 53.1 -0.3 12.7 >
LU 63.4 -0.8 0.7 6.6 55.1 -1.0 5.0 4.9 34.1 2.1 7.4 15.9
HU 56.7 -0.7 0.4 13.3 50.6 -0.3 0.8 9.4 31.4 -1.6 9.2 18.6
MT 55.2 0.6 1.0 14.8 37.4 1.7 4.2 22.6 29.1 0.6 0.6 20.9
NL 77.2 1.2 4.3 > 71.1 1.5 7.6 > 53.0 2.1 14.8 >
AT 72.1 0.7 3.6 > 65.8 1.4 6.1 > 41.0 2.4 12.2 9.0
PL 59.2 2.2 4.2 10.8 52.4 1.8 3.5 7.6 31.6 1.9 3.2 18.4
PT 68.2 0.4 -0.2 1.8 62.5 0.6 2.0 > 50.8 -0.1 0.1 >
RO 59.0 0.3 1.4 11.0 52.5 -0.3 0.7 7.5 43.1 1.7 5.9 6.9
SI 68.6 0.8 5.7 1.4 64.2 1.6 5.8 > 32.8 -0.7 10.0 17.2
SK 62.3 1.6 5.5 7.7 54.6 1.6 3.2 5.4 39.2 3.6 17.9 10.8
FI 71.1 0.8 3.9 > 69.0 0.5 4.8 > 56.5 1.4 14.8 >
SE 74.3 0.1 1.3 > 71.8 0.1 1.0 > 70.1 0.1 5.1 >
UK 71.5 0.0 0.3 > 65.8 0.3 1.1 > 58.0 0.6 7.3 >
EU-27 65.9 0.5 3.7 4.1 59.1 0.8 5.4 0.9 45.6 1.0 8.7 4.4
EU-15 67.3 0.3 3.9 2.7 60.4 0.7 6.3 > 47.4 0.9 9.6 2.6
2010 target 70% More than 60% 50%

Source: Eurostat, EU LFS.

Note: Data for RO 2002 instead of 2000.

UK (71.5%), Finland (71.1%), Cyprus

(70.9%) and Germany (70.7%) – Chart 12: Employment rates for Member States, 2000 and 2008
with four Member States less than 80
2 percentage points short – Estonia
(69.8%), Latvia and Slovenia (both 70
68.6%) and Portugal (68.2%).
% of working-age population

However, five Member States remained 50

a considerable distance from the target
with rates over 10 percentage points 40

lower – namely Malta (55.2%), Hungary

(56.7%), Italy (58.7%), Romania (59%)
and Poland (59.2%). The low rates in 20
Italy and Poland have a substantial
impact on the EU average, although 10

Poland showed a strong increase in

the employment rate in 2008 (of more DK NL SE AT UK FI CY DE EE LV SI PT IE CZ EU-27 FR LT ES BG LU BE SK GR PL RO IT HU MT

than 2 percentage points compared Source: Eurostat, EU LFS.

with the previous year). Note: Data for RO 2002 instead of 2000.

Employment in Europe 2009

Chart 13: Female employment rates for Member States, 2000 and 2008 The particularly strong downturn in the
economic situation in Ireland prevented
it from reaching the target in 2008, as
70 the employment rate slid to 67.6%, a
decrease of 1.5 percentage points com-
% of female working-age population

pared with 2007. Other countries which

experienced a noticeable decline in
employment rates in 2008 were Spain,
40 Lithuania, Luxembourg and Hungary.

In 2008, as in 2007, a total of 15 Mem-
ber States had a female employment
rate at or above the Lisbon target of
10 60% (Chart 13). However, apart from
Bulgaria, the remaining Member
DK SE NL FI EE UK AT LV DE SI CY PT LT FR IE BG EU-27 CZ BE LU ES SK RO PL HU GR IT MT States were still far from the target,
with three more than 10 percentage
Source: Eurostat, EU LFS.
points short – namely Malta (37.4%),
Note: Data for RO 2002 instead of 2000. Italy (47.2%) and Greece (48.7%).
Nevertheless, Bulgaria, Poland, Malta
Chart 14: Employment rates for persons aged 55–64 for Member States, 2000 and 2008 and Slovakia showed considerable
progress in 2008, with an increase
in their female employment rates of
2008 more than 1.5 percentage points com-
pared with 2007.

As in 2007, only 12 Member States

% of population aged 55-64

had an employment rate for older
persons (aged 55–64) of more than
50% – the Stockholm target for 2010
30 – although, with a strong increase of
3.5 percentage points in 2008, Bul-
garia is fast approaching the target.
However, nine Member States are
more than 10 percentage points short
of the Stockholm target – Malta, Hun-
gary, Poland, Slovenia, Luxembourg,
Source: Eurostat, EU LFS. Italy, Belgium, France and Slovakia
Note: Data for RO 2002 instead of 2000. – although the latter made signifi-
cant progress in 2008, with the rate
increasing by 3.6 percentage points.
Chart 15: Activity rates for Member States by gender, 2008
Austria and Luxembourg also showed
100 substantial progress, with rates rising
90 more than 2 percentage points in the
Total last year. With a value of less than
30%, Malta had the lowest employ-
% of working-age population

70 ment rate for older persons among

60 all the Member States in 2008, having
not made any significant improve-
ment since 2000 (Chart 14).

30 Activity rates
In 2008, 71% of the working-age
10 population in the EU-27 was active
0 in the labour market (i.e. employed
or unemployed). Participation rates
Source: Eurostat, EU LFS. ranged from as high as almost 81% in

Chapter 1 EU labour markets in times of economic crisis

Denmark to as low as 59% in Malta. Chart 16: Unemployment rates for Member States, 2008 and 2007
More than half of the Member States
displayed rates in excess of 70%, while
Hungary, Romania, Italy and Poland 2007
also recorded relatively low rates of 10

less than 65%.

Activity rates vary significantly

% of labour force
according to gender. For women the
activity rate was less than 64% in
2008, compared with a rate of 78%
for men (Chart 15). This unequal
situation between men and women
varies considerably from country to 2

country. Large differences in male

and female activity rates can be 0
observed in Malta, Greece and Italy,
while the Nordic and Baltic States dis- Source: Eurostat, EU LFS.
play relatively small differences. The
Member States with the largest gen- Chart 17: Unemployment rates for Member States by gender, 2008
der differences in activity rates are 15
also those countries that are furthest Men
away from reaching the Stockholm Women

target on female employment. 12 Total

% of labour force

2.3.2. Unemployment

The unemployment rate for the EU 6

averaged 7.0% in 2008. Spain had the
highest unemployment rate (11.3%)
followed by Slovakia (9.5%). Other 3

countries with higher-than-average

unemployment rates in 2008 were
Hungary, France, Portugal, Greece, NL DK AT CY CZ SI LU EE BG UK LT RO MT SE IE FI IT BE EU-27 PL DE LV GR PT FR HU SK ES

Latvia, Germany and Poland. In con-

Source: Eurostat, EU LFS.
trast, very low unemployment rates,
of less than 4%, were recorded in the continued to decrease in 2008, with was also observed in the Baltic States,
Netherlands, Denmark, Austria and examples being Poland, Slovakia, Germany and the UK.
Cyprus (Chart 16). Bulgaria and Germany, which all saw
rates fall by more than 1 percentage In 2008, 2.6% of the labour force
Although the unemployment rate point compared with 2007. was in long-term unemployment (i.e.
at EU level remained stable for 2008 unemployed for a period of 12 months
compared with 2007, different devel- As in the past, unemployment was or more). Most Member States had
opments could be observed across generally higher among women than rates around the average or lower,
individual Member States. For some men in 2008 (Chart 17). This was espe- but some rates were considerably
unemployment increased in 2008, cially the case in Greece, where the higher, for example in Slovakia which
with this being especially the case labour market situation for women has by far the highest rate of long-
for Spain (where the rate rose by looks particularly challenging, with a term unemployment, at 6.6%.
3 percentage points) together with gender gap of 6 percentage points.
Ireland, Latvia and Lithuania. For However, the opposite situation (of Long-term unemployment is gener-
the first two countries, this reflects higher unemployment rates for men) ally more frequent among women
in particular a sharp drop in employ- is found in a few Member States. than men. In Slovakia almost 8% of
ment due to a strong contraction in This is notably the case in Ireland the female labour force was long-
the construction industry, caused by and Romania, where the unemploy- term unemployed in 2008, with
the marked downturn in the hous- ment rate for men was 2 percentage Greece (6%), Portugal (4%) and
ing markets in those Member States. points more than that for women in Italy (4%) also having relatively high
In other countries, unemployment 2008, while this ‘reverse’ gender gap rates (Chart 18).

Employment in Europe 2009

Chart 18: Long-term unemployment rates for Member States by gender, 2008 Youth unemployment (i.e. unem-
ployment among those aged 15–24)
remains a serious concern, leading
7 Women to renewed efforts to facilitate the
Total entry of young people into the labour
market and to support them as they
take their initial career steps. The
% of labour force

youth unemployment rate in the EU

4 amounted to 15.4% in 2008 – virtually
the same as in 2007 – and still more
than twice the rate for adults aged
25–54. In several Member States, the
problem is particularly severe, with
1 youth unemployment rates of 20%
or higher in Spain, Greece, Italy and

Source: Eurostat, EU LFS.

The relatively high unemployment
rate for young persons is to a certain
Chart 19: Youth unemployment rates for Member States by gender, 2008 extent a result of the fact that unem-
ployment is related to the labour force
(those who are employed or unem-
ployed). Since most young people are
in education and therefore in many
cases do not belong to the labour
force, this rate can become artificially
% of labour force aged 15-24

high. Furthermore, the labour market
behaviour of persons in education
differs considerably from country to
country, which makes it difficult to
compare youth unemployment rates
across Member States.(8)

To gain a fuller understanding of the

labour market situation for young
people, the youth unemployment
Source: Eurostat, EU LFS. ratio – i.e. the unemployment of per-
sons aged 15–24 relative to the total
population of the same age – is often
Chart 20: Youth unemployment ratio in the Member States, 2008 considered in parallel with the unem-
15 ployment rate. In 2008 on average
Men 6.9% of all persons aged 15–24 were
Women unemployed in the EU-27, with only a
12 Total
few Member States displaying ratios
higher than in the previous year. In
% of population aged 15-24

Spain and Sweden more than 10% of

young people were unemployed in
2008, while unemployment was also
6 relatively high among young persons
in the UK, Finland and France. In
most Member States young men were
more affected than women, with the
fraction of unemployed among young
men exceeding 11% in the UK and
Spain (Chart 20).
Source: Eurostat, EU LFS.

(8) For further discussion see the chapter on

unemployment in this report.

Chapter 1 EU labour markets in times of economic crisis

2.3.3. Contractual arrangements increased since 2000, but in some the about a third of Member States. The
share has decreased – namely in the share of employees on such contracts
Within the EU there are significant dif- Baltic States, Poland, Romania, Cyprus, was highest in Spain in 2008 (29%),
ferences across Member States regard- Czech Republic and Bulgaria, i.e. in but was also relatively elevated in
ing the incidence of part-time work. most of the new Member States. Poland, Portugal, the Netherlands,
Its share in total employment varied Slovenia and Sweden. In contrast,
from around 47% at the one extreme The incidence of part-time work is the incidence of temporary work was
in the Netherlands, where the share much higher for women than for men relatively low in Romania, the Bal-
of part-time employment continues in virtually all countries. In the most tic States, Malta and Slovakia. For
to be much higher than for any other extreme case, the Netherlands, more these Member States less than 5%
Member State, to only 2% in Bulgaria than 75% of female workers worked of employees had a fixed-term con-
in 2008 (Chart 21). Shares were also part time in 2008. Men working part tract in 2008. Fixed-term contracts are
relatively high (above 20%) in Austria, time are less common than women in more frequent among female than
Belgium, Denmark, Germany, Sweden all Member States, but compared with male employees. This is especially the
and the UK. In contrast, in most of the most countries the share of part-time case in Cyprus where the share of
new Member States, the overall share working men is relatively high in Den- fixed-term contracts for women was
of part-time employment remains rela- mark, the Netherlands, Sweden, the 20% in 2008, while for men it was
tively low, particularly in Bulgaria, the UK, Germany, Romania and Finland. only 8%. Sweden and Finland also
Czech Republic, Hungary and Slova- displayed relatively high shares of
kia. In most Member States the share The use of fixed-term contracts among fixed-term contracts among women
of part-time work has moderately employees is relatively common in compared with men (Chart 22).

Chart 21: Part-time employment for Member States by gender, 2008

3. The recent labour
70 Men market downturn and its
60 Total intensification since the
50 deepening of the financial
% of employment

crisis last autumn
Currently the EU is in the midst of
the deepest and most widespread
recession in the post-war era. After
10 several years of favourable growth,
and in particular good performance
NL SE DE UK DK AT BE IE EU-27 LU FR IT FI ES PT MT RO SI PL CY EE LT LV GR CZ HU SK BG in terms of employment creation, eco-
nomic and labour market conditions
Source: Eurostat, EU LFS.
deteriorated sharply in the second
part of 2008. This mainly occurred as
Chart 22: Fixed-term employment for Member States by gender, 2008
a result of the impact of the financial
35 crisis which deepened last autumn
and came on top of a correction in the
30 Men housing markets in many economies.
The ensuing weakening in global and
25 Total
domestic demand and a marked drop
in investor confidence together with
% of total employees

tighter financing conditions and a
reduction in availability of credit has
had a dramatic effect on the economy
and subsequently the labour market.

Employment growth in the EU had
already petered out in the second
0 quarter of last year. This is when the
level of employment in the EU peak-
Source: Eurostat, EU LFS. ed, but also when quarter-on-quarter

Employment in Europe 2009

GDP growth first turned negative. At Chart 23: GDP growth for the EU, USA, Japan and the larger EU Member States
the same time, the unemployment 6

rate troughed in spring 2008 and On previous quarter

then headed upwards. As such, the On previous year
second quarter of 2008 marks the 2
turning point in the previous positive
growth period of recent years. The 0

already negative trend was subse-

% change
quently bolstered in the third quar-
ter of 2008, as the financial crisis -4
deepened markedly in September
and October, leading to more sub- -6

stantial impacts in subsequent quar-

ters before initial signs of some eas-
ing started to appear in the second -10
Q2 Q3 Q4 Q1 Q2 Q2 Q3 Q4 Q1 Q2 Q2 Q3 Q4 Q1 Q2 Q2 Q3 Q4 Q1 Q2 Q2 Q3 Q4 Q1 Q2 Q2 Q3 Q4 Q1 Q2 Q2 Q3 Q4 Q1 Q2 Q2 Q3 Q4 Q1 Q2 Q2 Q3 Q4 Q1 Q2
quarter of 2009. This section focuses 2008 2009 2008 2009 2008 2009 2008 2009 2008 2009 2008 2009 2008 2009 2008 2009 2008 2009
on the recent period of employ- EU-27 US JP DE ES FR IT PL UK

ment contraction and rising unem- Source: Eurostat, national accounts. Data seasonally adjusted.
ployment from the second quarter
of 2008 on, with special emphasis on sales in the EU have plummeted(9), output in the EU had contracted by a
the intensification in the labour mar- leading to temporary closures of car substantial 4.9% by the second quar-
ket downturn since the deepening of manufacturing plants and partial ter of 2009 (Chart 23).(10)
the financial crisis last autumn. temporary unemployment and/or
widespread use of short-time work- The decline in EU GDP compares with
ing arrangements. In some Member a somewhat more limited decrease in
3.1. Economic activity States, such as Ireland and Spain, economic output in the USA (which
the ongoing correction in the hous- entered a recession in the fourth
The EU economy has clearly suf- ing market severely aggravated the quarter of 2008) but is much lower
fered from the ongoing global eco- already difficult situation, as the than the economic contraction in
nomic crisis, which deepened and negative wealth effect of falling Japan, which also entered recession
broadened in autumn last year due house prices dampened consumer in the third quarter of 2008. Year-on-
to the crisis in the financial markets, spending, as well as through employ- year, by the second quarter of 2009
and from the consequent further ment losses due to the sharp drop in economic output in the USA had
deterioration in the global eco- residential construction. contracted by 3.8%, while in Japan
nomic situation. As a result of the a sharp drop in exports combined
financial crisis, risk evasion became with weak domestic demand led to
pervasive with much tighter credit 3.1.1. Developments in GDP economic output contracting by a
conditions, and lending volumes to substantial 7.2%.
companies and individuals dropped. After entering a technical recession
In addition, exposure to the sub- (two quarters of negative quarter- The marked decline in economic
stantial ongoing housing-market on-quarter growth) in the third quar- output at EU level over the last
corrections or other country-specif- ter of 2008, the economic downturn year reflects strong contractions in
ic factors in several Member States in the EU worsened further in the GDP in Germany, Italy and the UK
brought a halt to growth in domes- following two quarters, although and slightly more moderate falls in
tic demand at the same time as there were signs that it was easing France and Spain. Within the EU,
external demand weakened. somewhat by mid-2009 – following most Member States had technically
on from the sharp 1.9% and 2.4% entered a recession by the first quar-
Faced with falling demand globally contractions recorded in the previ- ter of 2009, although some have
– as both developed and devel- ous two quarters, GDP declined by recently returned to positive growth.
oping countries have been hit by a more modest 0.3% in the second (In the second quarter of 2009, the
the downturn – and therefore poor quarter of 2009. Consequently, com- economies of the Czech Republic,
prospects for profits, firms sharply pared with a year earlier, economic France, Germany, Portugal, Slovenia
reduced investment. At the same and Sweden started to expand again
time, confronted by risks to employ- (9) Although in some Member States sup- to join Greece, Poland and Slovakia
port measures such as ‘scrapping pre-
ment and the need to rebuild sav- in posting positive growth.)
miums’ have mitigated the fall in sales
ings, households curtailed consump- (for example, car sales in Germany, the
tion, especially of durable items. As biggest EU car producer, have not plum- (10) Quarter-on-quarter and year-on-year
meted owing to the implementation of GDP growth is based on seasonally
a clear example of the latter, car
scrapping premiums). adjusted data.

Chapter 1 EU labour markets in times of economic crisis

Chart 24: Year-on-year GDP growth in the second quarter of 2009 the latter reflecting a sharp housing
and timeline of entering recession correction and its strong economic
Quarter of entering technical recession reliance on the financial sector.
Not in No q-on-q
2008 Q2 2008 Q3 2008 Q4 2009 Q1 2009 Q2 recession data
PL In comparison, declines in ­economic
CY EL output compared to the second
EU-27 MT
quarter of 2008 have been slightly
-4 less marked in Spain (down 4.2%)
Year-on-year GDP growth in 2009 Q2

-6 IT and France (down 2.8%), while in
-8 HU
RO contrast to the other larger Member
-10 States, the Polish economy expanded
year on year, though at a declining
pace. (In the second quarter, year-on-

year GDP growth fell to 1.4%, with
positive growth still driven by expan-
sion in service sectors and in con-
struction.) All the remaining Member
-22 States had experienced a contraction
in economic activity compared to the
Source: Eurostat, national accounts. Data seasonally adjusted.
second quarter of 2008.
Note: Data for BG non-seasonally adjusted.

3.2. How has the

There is quite a spread in the time turned positive again. The recession labour market adjusted
at which individual Member States has particularly hit manufacturing,
entered recession (Chart 24). Ireland while output also decreased notably to the economic
and the Baltic States of Estonia and in the trade, transport and commu- downturn?
Latvia were the first to enter reces- nication sector. Overall, despite the
sion, in the second quarter of 2008. recent improvement, by the second
This early entry may at least in part quarter of 2009 economic output 3.2.1. Labour demand
explain why they are among those was down 5.9% on the same quarter
that have suffered the greatest year- of the previous year, reflecting Ger- Demand for new workers has declined
on-year contraction in GDP up to the many’s strong economic dependence in line with the economic downturn.
second quarter of 2009. Most Member on foreign exports, which have plum- The EU job vacancy rate (i.e. the
States entered recession in the third meted due to the global downturn. number of vacancies relative to the
(including France, Germany, Italy and Similarly, strong declines in output sum of vacancies and occupied posts)
the UK) or fourth quarter (includ- have also occurred in Italy (down started to drop in the third quarter
ing Spain) of 2008, and only Greece, 6.0%) and the UK (down 6.0%), for of 2008 and subsequently fell to
Poland and Slovakia had still avoided
a technical recession by the second
Chart 25: Job vacancy statistics for EU Member States
quarter of 2009 (although Slovakia
experienced a very sharp contraction
in GDP in the first quarter).
2009 Q2
(vacancies / sum of vacancies and occupied posts)

2008 Q2
Although most of the larger Member
States had already experienced nega-
tive GDP growth by the third quar- 3
Job vacancy rate

ter of 2008, the main declines were

recorded in the fourth quarter of last
year and in the first quarter of 2009.
Among the larger Member States,
Germany is among those to have 1
suffered the strongest contraction
in economic output, with the reces-
sion deepening sharply since the last 0
quarter of 2008, although output Source: Eurostat, job vacancy statistics. Data non-seasonally adjusted.
recovered somewhat in the second Note: Firm size: total except for FR and IT (10 or more employees). Data for EU, DE, HU
quarter of this year as GDP growth and SI 2008-2009, RO 2008 and BG, FR, LV, LU and NL 2009 provisional.

Employment in Europe 2009

1.4% in the first and second quarters addition to France, labour demand 2008 on. After posting negligible
of 2009, a drop of 0.7 percentage was weakest in Latvia, Luxembourg growth in the second quarter, which
points compared to a year earlier. and Portugal. marks the high point in the previous
This drop in the job vacancy rate was period of employment expansion,
equivalent to a fall in demand for Temporary agency work has been employment in the EU contracted by
new workers of around one third hit particularly hard by the down- 0.2% and 0.3% over the third and
over the year for the EU as a whole, turn. Data from Eurociett over recent fourth quarters of 2008, and by a
but underlying this development is months has shown a sharp year-on- more substantial 0.8% (1.8  million)
significant variation in the size of the year contraction in the number of and 0.6% (1.4 million) over the first
decline in demand across individual hours invoiced by private employ- and second quarters of this year
Member States (Chart 25). ment agencies, ranging from the (Chart 26). As a result, employment
order of 30% in the Netherlands, in the EU had declined to 223 mil-
Among the larger Member States, Italy and Belgium, to 40% in France, lion by mid-2009, down by 4.3 mil-
the decline in the vacancy rate has and as much as 50% in Spain. lion (1.9%) compared with the level
been most pronounced in Poland a year earlier, and mainly reflecting
(down by 1 percentage points year Despite the clear downward adjust- falls in employment for men.(12)
on year, or around a half), reflecting ment in the demand for new work-
the recent cooling-off in employ- ers, it appears that for much of the Underlying this development at EU
ment expansion. The rate declined period to date, many firms have been level were deteriorating labour mar-
more moderately over the year in the reluctant to reduce the number of ket performances in the larger Mem-
UK (by 0.8 percentage points), Ger- existing employees even when the ber States, most notably in Spain and
many (by 0.6 percentage points) and demand for their output fell. Man- in the UK, but also more recently in
in France (by 0.3 percentage points), power Employment Outlook Sur- Germany and Poland where employ-
and remained unchanged in Spain veys(11) over recent quarters have con- ment levels had remained relatively
(but at an already low level). sistently indicated that the majority resistant to the effects of the crisis
of employers report that they intend over 2008 (in the former due to exten-
By the second quarter of 2009, the to make no changes in their staff- sive recourse to short-time working
rate stood at 0.5–0.7% in Italy, Spain ing levels, which has been a reflec- arrangements). Over the year to the
and Poland, and at only 0.3% in tion of employers’ concern of losing second quarter of 2009, labour mar-
France, the lowest in the EU. How- skilled workers who will be hard to ket performances deteriorated across
ever, it remained relatively high in replace. However, faced with the all Member States, most notably in
Germany (2.6%, the second highest more sustained period of weakness in the Baltic States, Ireland and Spain,
rate in the EU) and the UK (1.6%), demand and the ongoing tight credit which have all been affected by severe
reflecting continued labour shortages conditions that increased the need housing market downturns leading to
despite growing unemployment. Offi- to cut costs, including labour costs, substantial employment contraction
cial sources in Germany and the UK maintaining the resolve to remain in the construction sector.
confirm that, although in mid-2009 at existing staffing levels became
job vacancies were down around a increasingly difficult for many. Among the larger Member States,
fifth and a third respectively on a Spain has clearly experienced the
year earlier, overall vacancy levels most significant decline in employ-
remained reasonably high at between 3.2.2. Employment ment levels. In line with the slow-
400 000 and 500 000 in each country. down in economic activity, employ-
Employment growth ment growth in Spain progressively
All the other Member States for decelerated over the course of 2007
which vacancy data is available have In reaction to the economic down- and turned negative in the second
seen the rate fall relative to the same turn, the labour market in the EU quarter of 2008. The contraction in
quarter of the previous year, with the had already started to weaken con- employment accelerated over the
sole exception of Greece. The sharp- siderably in the spring/summer of following quarters, with quarter-
est falls have occurred in Cyprus, last year, as employment growth on-quarter growth posting –2.5%
the Czech Republic and the Baltic moderated from the high rates of in the first quarter of 2009 before
States. Apart from Germany and the 2006 and 2007. In the latter half of moderating to –1.3% in the second
UK, demand for new workers in the 2008, in response to the intensifica- quarter, and has been much stronger
second quarter of 2009 remained tion of the financial crisis, employ-
relatively strong in Cyprus, Finland ment growth subsequently deterio- (12) Quarter-on-quarter employment chang-
es and growth are based on season-
and the Netherlands (all with above rated even more strongly, turning
ally adjusted data, year-on-year employ-
average rates in excess of 1.5%), negative from the third quarter of ment changes and growth are based on
despite the strong declines relative non-seasonally adjusted data. Levels of
(11) For more information see the website: employment in 2009 are non-seasonally
to the year before. At under 0.5%, in adjusted.

Chapter 1 EU labour markets in times of economic crisis

Chart 26: Employment growth for the EU and larger Member States the second quarter of 2009, employ-
6 ment had contracted by 1.2% year
on year in France, by 0.9% in Italy,
4 and by 2.0% in the UK.

By contrast, in Germany the effects
of the severe economic recession on
% change

the labour market have so far been

mitigated by reductions in labour
intensity, as companies have used
internal adjustment measures such
as temporary suspension of pro-
On previous quarter duction and short-time working
On previous year arrangements rather than reducing
the workforce. Quarter-on-quarter
Q2 Q3 Q4 Q1 Q2 Q2 Q3 Q4 Q1 Q2 Q2 Q3 Q4 Q1 Q2 Q2 Q3 Q4 Q1 Q2 Q2 Q3 Q4 Q1 Q2 Q2 Q3 Q4 Q1 Q2 Q2 Q3 Q4 Q1 Q2
2008 2009 2008 2009 2008 2009 2008 2009 2008 2009 2008 2009 2008 2009 employment growth turned negative
(–0.1%) only in the first quarter of
Source: Eurostat, national accounts. Data seasonally adjusted for change on previous 2009 and declined only slightly fur-
quarter; data non-seasonally adjusted for change on previous year. ther (to –0.3%) in the second quarter,
Note: Seasonally adjusted data not available for PL. mainly reflecting falls in the financial
services/business activities and indus-
than in other larger Member States. Despite the recession being deeper try sectors. As a result, employment
Compared with the second quarter in Italy and the UK, and similar in levels in the second quarter of 2009
of the previous year, employment France, the deterioration in labour had hardly changed compared with
had contracted by 7.1%. The dete- markets in those Member States since those a year earlier.
rioration has been driven by labour the third quarter of last year has
reductions across all sectors apart been less pronounced than in Spain In Poland, the strong employment
from public-sector-based services, but (with employment declines at a sig- expansion observed in 2006 and 2007
has been particularly marked in the nificantly slower pace than declines started to moderate at the beginning
­construction and industry sectors. in economic activity – see Box 1). By of 2008. Despite still positive GDP

Table 5: Employment growth for EU Member States

% change on previous quarter % change on previous year
2008 Q1 2008 Q2 2008 Q3 2008 Q4 2009 Q1 2009 Q2 2008 Q1 2008 Q2 2008 Q3 2008 Q4 2009 Q1 2009 Q2
BE 0.5 0.3 0.3 0.0 -0.5 -0.5 1.9 1.7 1.7 1.2 0.1 -0.7
BG : : : : : : 4.8 3.4 3.0 2.1 -0.3 -1.8
CZ -0.5 0.9 0.6 -0.1 -1.0 -0.8 1.1 1.4 1.3 0.9 0.3 -1.4
DK 0.3 0.0 0.2 -0.6 -1.5 -0.5 1.6 1.1 0.9 -0.1 -1.9 -2.6
DE 0.6 0.2 0.2 0.1 -0.1 -0.3 1.7 1.5 1.4 1.1 0.4 -0.1
EE 0.4 -0.1 -0.2 -0.6 -7.2 -1.8 2.0 -0.5 -0.3 -0.2 -7.2 -10.2
IE -0.1 -0.8 -1.6 -1.5 -3.8 -1.5 1.6 -0.1 -2.1 -3.9 -7.5 -8.3
EL 0.4 0.1 0.1 0.6 -1.8 0.3 1.3 1.5 1.1 1.0 -0.6 -1.0
ES 0.7 -0.6 -1.5 -2.0 -2.5 -1.3 1.5 0.1 -0.9 -3.1 -6.5 -7.1
FR 0.2 0.0 -0.1 -0.2 -0.4 -0.5 1.2 0.8 0.3 -0.1 -0.7 -1.2
IT 0.1 0.1 -0.4 -0.2 -0.2 0.0 0.9 0.8 -0.2 -0.2 -0.6 -0.9
CY : : : : : : 2.4 2.7 3.5 1.9 1.4 -0.5
LV 0.1 -0.2 -2.3 -3.1 -3.3 -4.9 5.6 3.4 0.2 -5.4 -8.2 -13.1
LT -0.2 -0.6 0.2 -0.7 -4.5 -1.8 0.9 -0.6 -1.0 -1.2 -5.1 -6.7
LU 1.2 1.4 0.7 0.7 -0.5 0.4 5.2 4.9 4.7 4.0 2.3 1.3
HU : : : : : : -1.5 -1.8 -0.7 -0.9 -3.0 -4.5
MT : : : : : : 2.7 3.1 2.3 1.8 0.6 -0.8
NL 0.4 0.4 0.0 0.3 -0.4 -0.6 1.9 1.6 1.1 1.1 0.3 -0.8
AT 1.2 0.3 -0.1 -0.2 -0.4 -0.4 2.2 2.0 1.5 1.4 -0.4 -1.1
PL : : : : : : 7.0 5.4 3.7 3.0 -1.0 -0.7
PT 0.3 0.2 -0.9 0.4 -1.3 -0.9 0.9 1.2 -0.2 -0.1 -1.6 -2.7
RO : : : : : : : : : : : :
SI 0.7 0.7 0.6 0.4 -1.2 -1.4 3.2 3.1 2.9 2.4 0.5 -1.6
SK 0.4 0.2 1.7 -0.3 -1.9 -0.6 2.8 2.9 3.2 2.1 -0.4 -1.3
FI 1.1 0.5 -0.7 -0.2 -0.7 -1.2 2.5 2.1 1.0 0.8 -1.1 -3.0
SE : : : : : : 1.7 1.3 0.7 0.0 -1.2 -2.2
UK 0.3 0.0 -0.3 -0.2 -0.5 -0.9 1.5 1.2 0.4 -0.2 -1.1 -2.0
EU-27 0.5 0.1 -0.2 -0.3 -0.8 -0.6 1.8 1.4 0.7 0.2 -1.2 -1.9

Source: Eurostat, national accounts. Data seasonally adjusted for change on previous quarter;
data non-seasonally adjusted for change on previous year.

Employment in Europe 2009

growth, year-on-year employment severe in Ireland (–8.3%) and the Bal- been strongest in construction (which
growth in Poland turned negative tic States (Estonia, –10.2%; Latvia, had already been following a strong
in the first quarter of 2009 (post- –13.1%; and Lithuania, –6.7%), which downward trend in employment
ing a decline of –1%, a more severe – alongside Spain – have registered growth since the first quarter of 2007)
downward adjustment of year-on- the strongest deteriorations in their and industry, but has also been notice-
year growth compared to most other labour markets (Table 5). able in financial services and business
larger EU Member States except activities, and in the trade, transport
Spain) before moderating to –0.7% Sectoral employment and the communication sector.
in the second quarter. and ­restructuring
Over the year to the second quarter
In the remaining Member States, The moderation in year-on-year of 2009, total employment decreased
year-on-year employment growth employment growth over 2008 fol- by 4.3 million, mostly reflect-
had turned negative in all except lowed by the contraction in employ- ing significant drops of 1.3  million
Luxembourg by the second quarter ment in the first half of 2009 has in construction and 1.9 million in
of 2009. The decline in employment resulted from a broad deterioration non-construction-related indus-
over the last year has been ­particularly across almost all sectors. Decline has try (equivalent to a contraction in

Box 2: Employment declines versus falls in economic activity

The recent fall in employment in the EU has been much weaker than the fall in economic activity…

The recent fall in employment in the EU and most Member States has been significantly weaker than the fall in economic activity. This is in
part due to extensive recourse to short-time working arrangements or other measures to tackle the employment impact of the crisis in some
Member States, but also reflects the normal lags before the downturn in economic activity feeds through to the labour market. For the EU as
a whole, compared to a year earlier, economic output had contracted by a substantial 4.9% by the second quarter of 2009, while employment
had contracted by a much more limited 1.9%.

… however, in some Member States the response of employment to the decline in economic activity has been
much more pronounced…

By the second quarter of 2009, GDP had declined compared to a year earlier in all Member States except Poland, while employment had
declined in all except Luxembourg. However, in some Member States the response of employment to the decline in economic activity has been
much more pronounced than in others. The employment decline has been strongest in the Baltic States and Ireland, in line with the sharp
declines in economic activity, but also in Spain, even though the contraction in GDP has been more limited than for most other Member States.
Additionally, in Greece employment contracted much more noticeably than GDP.

The elasticities of employment declines to GDP declines (i.e. the decrease in employment over the year to the second quarter of 2009 divided by
the decrease in GDP over the same period) suggest a particularly strong reaction of employment to economic contraction in Greece, Ireland and
Spain, but also, although to a lesser extent, in Estonia, Hungary, Latvia and Portugal (Chart 27). In contrast, the elasticity of employment to the fall
in economic activity in countries such as Belgium, Italy, the Netherlands and Slovenia, and above all Germany, has been much more subdued.

Chart 27: Elasticity of employment declines to GDP contraction, 2008 Q2–2009 Q2

% change in employment / % change in GDP 2008 Q2-2009 Q2







Source: Eurostat, national accounts. Data seasonally adjusted.

Note: Data for BG, CY, HU, MT, PL and SE non-seasonally adjusted.

Chapter 1 EU labour markets in times of economic crisis

… particularly in Member States where the construction sector accounts for a relatively high share of employment.

There are several reasons for the comparatively stronger decline in employment in certain Member States. However, one key factor appears to
be the impact on and influence of the construction sector – one of the sectors hardest hit by the recent economic and financial crisis and which
accounts for an especially high share of national employment in the Baltic States, Ireland and Spain compared with other Member States (Chart
28). In this context, to a certain extent the variation across countries reflects productivity levels in the sectors which have been hit hardest. For
example, in Germany the manufacturing sector has been badly hit by plummeting exports but high productivity levels in this sector have led
to comparatively small falls in employment relative to GDP, while in Spain the large contraction in the relatively low-productivity construction
sector has led to large falls in employment relative to GDP.

Chart 28: Employment growth 2008 Q2–2009 Q2 versus share

of employment in construction in 2008 Q2

0 DE
% change in employment 2008 Q2–2009 Q2



-8 IE

-10 EE

0 5 10 15
Share of employment in construction 2008 Q2

Source: Eurostat, national accounts. Data non-seasonally adjusted.

Another reason is the widespread use of internal flexibility in certain Member States such as Belgium and Germany compared with relatively
limited use of such arrangements in the Baltic States, Ireland and Spain. Furthermore, in the case of Spain the high share of workers in temporary
contracts, who can be relatively easily dismissed, also in part explains its stronger employment reaction to the downturn. Indeed, countries with a
combination of very flexible short-term contracts and very inflexible permanent contracts may experience relatively large falls in employment.

sectoral ­employment of 7.6% and in other sectors has been more mod- trade, transport and communication
4.8% respectively). Indeed, these two erate, with employment declining by sector (down some 1.2 million, or
sectors combined account for more 0.8% (110 000) in agriculture and by 2.1%), and in financial services and
than 60% of all sectoral employment 0.6% in services (1 million). Within business activities (down 600 000, or
declines over this period. Contraction services, falls in employment in the 1.8%) were partially compensated by
employment expansion of 800 000
(or 1.3%) in other services (mainly in
Chart 29: Sectoral employment changes for the EU, 2008 Q2–2009 Q2 the public sector) (Chart 29).
2 2

Broadly similar sectoral trends are

0 0
reflected in the European Restruc-
turing Monitor data collected by
the European Monitoring Centre
-2 -2 on Change, which gives a picture
% change

of the labour market impact of the

crisis at enterprise level (Box 2). This
-4 -4
data source has in particular record-
ed substantial job losses announced
Change in levels (lhs)
-6 for the manufacturing sector, espe-
% change (rhs)
cially in auto-manufacturing and
related industries.
-8 -8
Agriculture Other industry, Construction Trade, transport Financial services and Other services Total
including energy and communication business activities

Source: Eurostat, national accounts. Data non-seasonally adjusted.

Employment in Europe 2009

Box 3: Restructuring developments in Europe

The European Restructuring Monitor (ERM) dataset is based on news and media reports of individual cases of restructuring, generally involv-
ing over 100 announced job losses or gains, identified by a network of national correspondents in the EU-27 and Norway. The following is a
summary analysis of the more than 3 000 ERM cases recorded during the 18-month period between 1 March 2008 and 30 August 2009.(1)

Announced job losses outnumbered job gains by a factor of 2.5 to 1

After recording greater announced job gains than losses in 2007 and the early months of 2008, ERM data from spring 2008 onwards clearly
registers the impact of the economic crisis. There were almost three cases of announced job loss for every one of job creation during the sub-
sequent period to August 2009. Excluding cases of transnational restructuring, total announced job losses from restructuring captured by the
ERM amounted to over 935 000 jobs. Just over 385 000 new jobs were announced. The impact of the crisis was most obvious in the months of
December 2008 and January 2009, in each of which over 100 000 job losses were announced. Since the turn of the year restructuring activity
has moderated significantly though job losses continue to outnumber job gains (Chart 30).

Chart 30: Announced job losses and gains for the EU


100 Net increase

Job losses
80 Job gains


Jobs (thousands)






Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug
2008 2009

Source: European Monitoring Centre on Change, European restructuring monitor.

Auto-manufacturing and related industries account for the highest number of job losses…

Auto-manufacturing recorded by some margin the highest number of cases of restructuring of any sector between March 2008 and August 2009 (268),
together with the greatest aggregate job losses (over 100 000). There has been a marked shift in the composition of this job loss from original equipment
manufacturers to motor parts and accessories producers in 2008–2009. It appears that employment in supply chain firms has proven more vulnerable to
the downturn than that in core producers. Over half of car sector job losses were in supply chain firms in 2008–09 (compared with a previous average of
25%). Related sectors – manufacture of basic metals and machinery/ equipment – also figure among the sectors most affected by job losses (Table 6).

Auto-manufacturing was also notable for the geographical spread of job losses and job gains. West European Member States recorded cases
of job loss almost exclusively while some Central and Eastern European Member States – notably Poland, but also Hungary and Slovakia –
recorded more job gains than job losses in the sector, even during the downturn.

Table 6: ERM announced job loss by NACE-2 sector (March 2008–August 2009)
2002-Feb 2008 Mar 2008-Aug 2009
Total job losses
Sector (NACE rev 1.1) % total job loss Cases % total job loss
Manufacturing: auto 8.2 268 11.2 105
Public administration 13.3 46 8.5 79
Retail 3.0 82 7.7 72
Post and telecommunications 12.0 69 7.5 70
Financial intermediation 7.5 82 7.3 68
Manufacture: machinery / equipment 2.0 163 5.4 51
Manufacture: basic metals 2.9 88 3.7 35
Manufacture: electrical machinery 2.3 97 3.0 28
Manufacture: food products 3.3 108 3.0 28
Manufacture: other non-metallic mineral products 1.0 80 2.8 26

Source: European Monitoring Centre on Change, European restructuring monitor.

(1) Summary based on extraction from ERM on 2 September 2009.

Chapter 1 EU labour markets in times of economic crisis

… while discount providers in the retail and hotels/restaurant sector announced job creation…

Over the period from March 2008 to August 2009, the largest restructuring announcements, both those involving job losses (excluding world
cases) and those involving job gains, were as summarised in Table 7.

Table 7: Top cases of announced job losses and job gains

Company Thousands Country Restructuring type Sector Announced date
Job loss
Army/National security 54 France Internal restructuring Public admin June 2008
Woolworths 27 the UK Bankruptcy/Closure Retail December 2008
Education Nationale 16 France Internal restructuring Education June 2009
T-System 12 Germany Internal restructuring Post/telecoms March 2008
TNT Post 11 the Netherlands Internal restructuring Post/telecoms July 2009
Commerzbank 9 the EU Merger/Acquisition Financial intermediation September 2008
PKP Cargo 9 Poland Internal restructuring Land transport January 2009
Royal Bank of Scotland 6.8 the UK Internal restructuring Financial intermediation February 2008
Commerzbank 6.5 Germany Merger/Acquisition Financial intermediation September 2008
Unicredit 5.9 Italy Merger/Acquisition Financial intermediation June 2008
Job gain
Edeka 25 Germany Retail October 2008
McDonald’s 12 the EU Horeca January 2009
Tesco 10 the UK Retail January 2009
Kentucky Fried Chicken 9 the UK Horeca February 2009
Subway 7 the UK Horeca January 2009
ASDA 7 the UK Retail January 2009
Source: European Monitoring Centre on Change, European restructuring monitor.

The retail sector has been dynamic both in terms of job creation (accounting for nearly a quarter of new jobs announced) and job destruction
(the largest proportionate increase of announced job losses for any sector) in 2008–09. On the negative side, decreasing sales and profits
resulted from weak consumer confidence, tighter credit conditions and rising unemployment. A number of retail businesses that were previ-
ously experiencing difficulties have been unable to withstand further weakening of trading conditions (e.g. Woolworths in the UK and Ger-
many). On the positive side, some of the larger retail conglomerates have signalled aggressive growth plans with a view to securing market
share from failing retailers. Also noteworthy has been the expansion of discount retailers (e.g. Asda) and restaurant chains (e.g. McDonalds)
seeking to benefit from new ‘downshifting’ customers.

The share of announced job loss due to bankruptcy/closure increased while those due to offshoring
and relocation decreased …

Internal restructuring – something of a catch-all restructuring category – accounted for 70% of announced job losses in ERM restructuring
cases over 2008–2009. Bankruptcy/closure accounted for a sharply increased proportion of job losses over that period (up from 14% to 22%
of the total) (Table 8). At country level, the increased share of bankruptcy/closure-related job losses was notable in the UK, Italy, Finland,
Greece, Slovenia, Portugal and Bulgaria (>15 percentage points in each country).

Table 8: Share of job losses by restructuring type in the EU (%)

Restructuring type 2002-Feb 2008 Mar 2008-Aug 2009
Bankruptcy/Closure 14.1 21.8
Internal restructuring 72.9 70.0
Merger/Acquisition 4.1 3.9
Offshoring/Delocalisation 5.8 2.8
Other 0.4 0.3
Outsourcing 1.2 0.5
Relocation 1.6 0.6
Source: European Monitoring Centre on Change, European restructuring monitor.

At aggregate EU level, the increase in the share of bankruptcy/closure-related job losses was matched by a decline in the share of offshor-
ing/relocation/outsourcing. ERM data has been extensively used both by Eurofound and by external researchers(2) as a source for analysing
developments in relation to offshoring of activities and consequences for employment in the EU. One principal conclusion is that offshoring
has accounted for between 5% and 8% of announced job losses arising from major restructuring events in the EU since 2002, a perhaps
unexpectedly modest share given the interest of researchers and the general media in the phenomenon.

What is notable about the recent downturn is that the share of cases and of announced job losses attributable to offshoring/delocalisation – as
well as to related restructuring categories of relocation and outsourcing – have declined even from these modest levels. The only countries in
which offshoring represented a greater share of job losses in the most recent period were Latvia and Slovakia – countries that would previously
have been considered offshoring destinations – and Austria.

(2) e.g. Auer, P., G. Besse, and P. Meda, Offshoring and the internationalisation of employment, ILO, 2005 and ERM Annual Report 2007.

Employment in Europe 2009

3.2.3. Unemployment Chart 31: Unemployment rate versus GDP growth for the EU and the USA
GDP growth US (lhs) Unemployment rate US (rhs)

At EU level, the average activity rate 4

GDP growth EU (lhs) Unemployment rate EU (rhs)

has changed relatively little over the

last year and remains close to 71%.
This indicates that so far the effects
of the crisis on total labour supply

% change on previous year

have been very limited, reflecting 0

% of labour force
that recent labour market reforms in
many countries have strengthened
the labour market attachment of the
working-age population. As a conse-
quence, the crisis appears not to be -4

resulting in a noticeable reduction in

overall activity, but is rather focused
-6 4
in its impact on unemployment. Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2
2005 2006 2007 2008 2009

Despite clear signs of deterioration, Source: Eurostat, national accounts and series on unemployment.
Data seasonally adjusted.
so far the European labour market
has held up relatively well overall
to the economic downturn. Unem- reacted by reducing ­working time of the rise in unemployment var-
ployment has risen, but by less than instead where possible. For exam- ies considerably from country to
might have been feared given the ple, by May 2009 there were around country. Hungary, Ireland, Italy and
strength of the recession and the 1.5 million workers in Germany in Spain were the first Member States
sharp declines in confidence. For short-time working schemes. where unemployment rates started
example, despite the sharper eco- to rise, as early as in the first half
nomic downturn and stronger falls Nevertheless, the unemployment of 2007, followed by Estonia, Latvia,
in business confidence in the EU rate in the EU has remained on Lithuania and Luxembourg later in
compared with the USA, increas- an upward trend since it reached a 2007. All other Member States saw
es in the EU unemployment rate trough in spring of last year, climb- rates bottom out at the same time
have been less dramatic than in the ing with particular strength since last or later than the EU average. In
USA (Chart  31). By August 2009 the October to April in reaction to the Austria, Belgium, Denmark, Greece,
unemployment rate in the EU had heightening of the financial crisis. Finland, France, Portugal, Romania,
increased by 2.1  percentage points This reflects that up until the third Sweden and the UK, rates bottomed
compared to one year earlier, while quarter of 2008 unemployment in out in the second quarter of 2008.
in the USA it had risen by a more the EU as a whole held up well to the The remaining countries, including
marked 3.5 percentage points.(13) economic downturn, due in large part Germany and the Netherlands, saw
to Germany and Poland; however, it unemployment rates start to rise
The relative resilience of the EU then started to rise more strongly, only in the second half of 2008,
labour market to date reflects in in part reflecting the deterioration some one and a half years later than
part the usual lag of 2–3 quarters in the German labour market as its Spain and Italy.
before the sharp acceleration in exports began to be hit hard by the
the economic downturn in Octo- sharp slowdown in global demand. Although unemployment rates have
ber feeds through to the labour By August 2009 the unemployment been rising over the last year or so
market, but also has resulted from rate in the EU had increased to 9.1%, in all Member States, the severity
the increased use of internal adjust- a rise of 2.4 percentage points com- of the rise varies considerably across
ment measures (short-time work- pared to March/April 2008. Total countries. The increase in unemploy-
ing, temporary suspension of pro- unemployment rose to a seasonally ment has been precipitous in cer-
duction etc.) which has allowed adjusted 21.9 million (21.6 million tain Member States (unemployment
firms to use various means of inter- non-adjusted), an increase of 5.7 mil- rates have roughly doubled over the
nal adjustment rather than reduce lion (or more than a third) compared last year in Ireland and Spain, and
their workforce, especially in coun- to March/April 2008. tripled in the Baltic States) while
tries such as Belgium and Germany. even those Member States which
Indeed, while it seems that reducing Underlying the EU average are con- have been least affected so far (for
staff levels has been the immediate trasting developments across indi- example the rise has been relatively
response of US firms, EU firms have vidual Member States, both in terms limited in Germany and Poland) have
of the onset of the rise in unem- recently been reporting worsening
(13) Unemployment changes, rates and levels
ployment and its severity. The onset conditions. By August all had higher
are seasonally adjusted.

Chapter 1 EU labour markets in times of economic crisis

unemployment rates compared with Among the other Member States, in Austria and the Netherlands (4.7%
a year earlier. The most substantial only France (13%) and the UK (15%), and 3.5%, respectively) (Chart 32).
rises compared with August 2008 accounted for contributions to the
were in the Baltic States (of the order rise in unemployment over the last The long-term unemployment rate in
of around 9–11 percentage points), year of more than 10%. the EU had been decreasing up until
Ireland (up 6.2 percentage points) the third quarter of 2008 (when it
and Spain (up 7.1 percentage points). Some two years after unemployment affected 2.5% of the labour force),
In contrast the rise in unemploy- first started to rise there, Spain now but following the deterioration in
ment has been only marginal (with accounts for almost one in four of all the labour market over the latter
rates rising by less than 1 percentage unemployed persons in the EU-27, part of last year, it started to rise
point) in Austria, Belgium, Germany, with its unemployment rate reach- again to reach 2.8% in the second
Italy, the Netherlands and Romania ing 18.9% in August (with underly- quarter of 2009. However, this rise
(Chart 32). ing unemployment at 4.3 million), does not yet fully reflect the recent
twice as high as the EU average and weakening of EU labour markets and
Among the larger Member States, the highest in the EU. Among the the subsequent increase in unem-
unemployment has risen dramati- remaining Member States, by mid- ployment, and it is likely that the
cally over the last year or so in Spain, 2009 the unemployment rate was long-term unemployment rate will
accounting for more than 35% of the highest in Latvia (18.3%) and Esto- move higher in the quarters ahead as
total rise in unemployment in the EU nia, Ireland, Lithuania and Slovakia elements of the large influx of recent
since April 2008 and almost a third of (all with rates around 11–14%), but entrants to unemployment eventu-
the rise over the last year (Chart 33). in contrast remained remarkably low ally feed through to the stocks of the
long-term unemployed. Indeed, com-
pared with last year, the long-term
Chart 32: Unemployment rates for EU Member States unemployment rate has increased in
several Member States, most mark-
August 2009 edly (by over 1 percentage point) in
August 2008 Ireland, Spain and the Baltic States.

12 3.2.4. Other labour market

% of labour force

responses to the economic


Labour markets can also adjust

through other mechanisms than
reducing employment levels – there
are many established means of
0 adjusting production and aggre-
gate working time to counter tem-
Source: Eurostat, series on unemployment. Data seasonally adjusted.
porary slumps in demand. Indeed,
Note: Data for UK June 2008–June 2009, EE, EL, IT and RO 2008 Q2–2009 Q2.
as reported in a recent paper by
the European Foundation for the
Chart 33: Contribution to unemployment increase, August 2008–August 2009 Improvement of Living and Working
Conditions(14), there are signs that
Other many companies have made workers
redundant only as a last resort and
ES that a range of alternative responses
have been implemented. A common
CZ feature is negotiated reduction of
LT working time (‘short-time working’)
3% balanced by increased provision of
3% training. Other responses include
addressing labour costs (through pay
DE UK freezes or pay cuts, or reduced social
4% 15%
FR (14) European Foundation for the Improve-
ment of Living and Working Conditions,
Source: Eurostat, series on unemployment. Data seasonally adjusted. Europe in recession: Employment ini-
tiatives at company and Member State
Note: Data for UK June 2008–June 2009, EE, EL, IT and RO 2008 Q2–2009 Q2. level, Background paper.

Employment in Europe 2009

c­ ontributions by employers), paid/ Adjustment by type of Part-time employment has also

unpaid career breaks and, at the employment (temporary, part- moderated in response to eco-
aggregate level, an adjustment in time and self-employment) nomic conditions. The previous
the level and composition of employ- strong year-on-year growth of part-
ment in terms of temporary, part- Employment has adjusted first and time employment in the EU over
time and self-employment. foremost to the economic ­downturn 2006 and 2007 (in response to the
through the deceleration and increased economic activity in that
One sector that can serve as a show- ­subsequent contraction in tempo- period) weakened from the second
case in this regard is the automotive rary employment, which is the most quarter of 2008 onwards. Growth
sector, where demand has been espe- cyclical component of employment. in full-time employment likewise
cially badly hit by the credit crunch, In line with the downturn in over- dropped off from the second quar-
declining consumer sentiment and all economic activity, the previous ter of 2008. Similar growth rates for
increasing inventory levels, despite strong year-on-year growth in tem- both were observed in the second
recent initiatives taken in several porary employment of 4.5–5.5% over and third quarters of 2008; how-
Member States to support demand 2006 subsequently weakened over ever, from the fourth quarter of
(e.g. through ‘scrapping premiums’). 2007 and turned increasingly nega- last year onwards, the economic
A high proportion of companies in the tive over 2008 and into 2009. By the downturn had a stronger impact
automotive sector have resorted to second quarter of 2009, the number on full-time employment. The rate
collective redundancies and lay-offs in of employees in the EU with tempo- of year-on-year growth in full-time
the face of the unprecedented fall-off rary contracts had fallen by 1.7  mil- employment had dropped to –2.1%
in sales volume, but many have also lion (or around 6%) compared with by the second quarter of 2009,
implemented other measures to avoid the second quarter of 2008, mainly while for part-time employment
making workers redundant. Many of driven by falls in all the larger Mem- year-on-year growth still remained
the large automotive companies, espe- ber States and most notably by a relatively strong (at 1%). This tends
cially in western European Member decrease of 1 million in Spain. Growth to suggest that the decline in full-
States, extended scheduled seasonal in permanent employment, which had time employment has been par-
closures over Christmas 2008/New remained at a relatively stable rate of tially offset by a continued increase
Year 2009. Even after the resumption around 2% over 2008, also came to a in part-time employment, demon-
of production in 2009, many firms halt in the first quarter of 2009 and strating the potential role of part-
announced temporary plant closures subsequently turned negative in the time work as a ‘shock absorber’ in
during the year. Reduction or elimi- second quarter (Chart 34). the economic downturn.
nation of overtime and nightshifts
has also been a common response, as The downturn in temporary employ- Another area of employment where
has compulsory leave-taking where ment has led to a reduction in the the risks from the downturn may
workers are obliged to take annual share of employees in the EU with be quite different is among the
leave entitlements in periods specified fixed-term contracts relative to total self-employed. Indeed cash-flow dif-
by their employer (often in conjunc- dependent employment. This share ficulties and the ‘credit crunch’ may
tion with temporary plant closures). has broadly been decreasing since be creating particular problems for
Furthermore, either in combination the second half of 2007, falling to small businesses. On the other hand,
with or in addition to many of the 13.5% in the second quarter of 2009 people made redundant by their
above measures, obligatory periods of (down by 0.7 percentage points on a employer may see it as an opportu-
unpaid leave and shortened working year earlier) and reflecting declining nity to set up their own business, a
weeks (three- and four-day weeks) shares in most of the ­Member States, step ­encouraged by the recent pro-
have become widespread. most notably in Spain and Slovenia. posal by the European ­Commission

Chart 34: Employment growth (employees) and GDP growth for the EU
GDP growth and growth of employees with permanent and temporary contracts GDP growth and growth of full-time and part-time employment
6 6

4 4
% change on previous year

2 2

0 0

-2 -2
Employees in temporary contracts Employed part-time
-4 -4
Employees in permanent contracts Employed full-time
-6 GDP growth -6 GDP growth
-8 -8
Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2
2006 2007 2008 2009 2006 2007 2008 2009

Source: Eurostat, national accounts and EU LFS, DG EMPL calculations.

Data on GDP seasonally adjusted. Data on employment ­non-seasonally adjusted.

Chapter 1 EU labour markets in times of economic crisis

to establish a new €100 million Chart 35: Employment and hours worked in the industry sector in the EU
­ icro-finance facility, to ­provide
m 2

credit to small businesses and to

people who have lost their jobs and 0
want to start their own small busi-
nesses. Data from the EU labour
force survey, however, indicates that

% change on previous year

so far self-employment has been on
the decline, with such employment
dropping by close to half a million
in the year to the first quarter of -6

2009. Underlying this was strong

Hours worked
contraction in the third and fourth -8
quarters of 2008, compared with
much weaker contraction in the first -10
quarter of 2009. Q1 Q2
Q3 Q4 Q1 Q2
Q3 Q4 Q1 Q2
Q3 Q4 Q1 Q2
Q3 Q4 Q1

Source: Eurostat, short term business statistics. Data non-seasonally adjusted, hours
Working hours worked working day adjusted.
Note: Sector covered is industry (except construction), sewerage, waste management
The practice of promoting reduc- and remediation activities.
tions in working time is something
that has protected European jobs
from the initial impact of the reces- extends beyond – or has exhaust- ible in data from the EU labour
sion and helped to avoid the sharp ed – what has been collectively force survey on the hours worked
rises in unemployment seen for agreed, it can have recourse to the by those workers remaining in
example in the USA. In several federal Kurzarbeitgeld (short-time employment. For example, in Ger-
Member States public authorities working fund). Under the scheme, many and Austria, over the year to
have been involved in facilitating employers are subsidised up to the second quarter of 2009 there
companies’ recourse to short-time 67% of an employee’s wages by have been noticeable reductions
working: the Netherlands, Aus- the federal authorities in the case across most sectors in the hours
tria, Germany and France have in of temporary lay-off or reduced worked by full-time employees in
place short-time compensation pro- working hours occasioned by sharp their main job (Chart 36). Indeed,
grammes whereby employers can declines in demand, force majeure for those still in employment, aver-
apply for temporary state assistance or structural changes within the age working hours have declined
to top up the wages of workers company. By May 2009, around by more than 3% in these two
working reduced hours. 1.5  million workers in Germany countries, reflecting strong declines
were covered by the ­s hort-time across almost all sectors (for most
For example, in France, chomage working scheme. well above 2%) and including most
technique or chomage partiel is a service sectors. In Germany, the
publicly funded scheme that allows As highlighted previously, much of important manufacturing sector
companies in cases of exception- the recent decline in overall employ- has undergone a decline of around
al economic difficulties to have ment can be attributed to the indus- 6% on average in full-time employ-
recourse to state-governed funds try sector. Focusing on this sector, it ees’ working hours. In contrast, in
covering 60% of minimum hourly is nevertheless clear that total work- Spain and the UK, average hours
wages during periods when staff ing hours in the sector have declined of work by full-time employees
are temporarily laid off. In Ger- at a faster rate than employment has generally decreased less sig-
many, where it is common for since the last quarter of 2008, imply- nificantly across sectors, other than
sectoral collective agreements as ing there has been a substantial for the real estate activities sector
well as plant/company-level agree- adjustment in the labour market which declined by more than 5%,
ments to include options to reduce also through reducing working reflecting the effect of the strong
working hours in order to main- hours as opposed to laying people downturn in the housing market in
tain employment, the federal off (Chart 35). these two countries. Overall, aver-
Kurz­arbeit system provides a state- age hours worked declined by a
supported back-up for companies On a broader sectoral scale, the more limited 1.0% in the UK and
resorting to short-time working approach of certain Member States 1.6% in Spain, with most sectors
outside the provisions of collective to favour reductions in working seeing working hours fall by less
agreements. When a company’s time rather than reductions in the than 2% and very little change in
need for working-time flexibility level of employment is clearly vis- many service sectors.

Employment in Europe 2009

Chart 36: Relative change in working hours by sector, 2008 Q2–2009 Q2 potentially more significant employ-
United Kingdom ment reductions. The tendency
towards wage moderation is sup-
0 ported by recent developments in
the index of negotiated wages cal-
-3 culated by the ECB, which started to
% change

decline in 2009, falling to 2.7% for

the euro area in the second quarter
after picking up last year. Indeed,
the economic downturn has started
to affect labour markets not only
Spain through falling employment but also
more recently through weakening
0 of growth and even a decline in
employee compensation, especially
-3 the wages and salaries component
% change

(Chart 37).

Year-on-year growth in nominal com-

pensation per employee in the EU
slowed down significantly over 2008
Germany and declined sharply in the first quar-
ter of this year (reflecting a sharp
0 drop in the UK), before edging back
up to still register a decline of –0.8%
-3 in the second quarter. Recent falls
% change

have been driven mainly by decreases

in wages and salaries (which account
for around 80% of compensation),
while other labour costs (e.g. social
-12 contributions) declined less sharply.
In contrast, hourly nominal labour
0 costs have continued to increase,
though at a more subdued pace in
-3 the beginning of this year. After
% change

accelerating steadily in the second

half of 2008, the year-on-year growth
rate of hourly labour costs in the EU
dropped to 1.3% in the first quarter
-12 of 2009, compared with a rate of


Mining and quarrying



Water supply; sewerage


Wholesale and retail trade; repair

of motor vehicules and motorcycles

Transportation and storage

Accomodation and
food service activities

Information and comunication

Financial and insurance activities

Real estate activities


Administrative and
support service activities
Public administration and defence;
compulsory social security


Human health and

social work activities


Other service activities

4.7% in the fourth quarter of 2008,

but rebounded back to 3.7% in the
second quarter. Since the end of last
year, hourly labour costs have risen
the slowest (or even declined in the
Source: Eurostat, EU LFS. Data non-seasonally adjusted. first quarter of 2009) in the service
Note: Average number of actual weekly hours of work in main job for full-time employees. sector – rising at a rate of 2.9% in the
second quarter and remaining weak-
Labour costs this type of approach can be found in est (at 2.3%) in financial and insur-
the airline industry. ance activities – while they grew most
Another adjustment mechanism to significantly (by 5.1%) in industry
the slump in demand can be through On a more general level, there may (remaining far above average growth
concession bargaining, where employ- be greater wage moderation in the over recent years) and by 4.1% in
ers seek to link employment security face of the economic downturn and construction (although less than in
(e.g. the withdrawal of compulsory the heightened risk of unemploy- the previous two years). The trend
redundancy plans) to pay freezes or ment, and moreover lower wage in total hourly labour costs mainly
pay cuts. Recent high-profile cases of increases might have helped ­prevent resulted from developments in their

Chapter 1 EU labour markets in times of economic crisis

Chart 37: Growth in nominal compensation per employee and its components for the EU Men have suffered the brunt of the
5 contraction in employment, with
Compensation per employee
their employment falling by 2.7%
Wages and salaries (versus only a 0.3% decline for
Social contribution
3 women) and accounting for more
than 90% of the total net reduc-
tion over the year to the second
% change on previous year

1 quarter of 2009. This reflects that

the economic downturn has, so far,
predominantly hit male-oriented
-1 sectors in terms of employment,
such as the construction sector
and the automotive industry. In
-3 terms of age, youth (i.e. those
aged 15–24) has been proportion-
Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2
2006 2007 2008 2009 ately most affected, with a decline
Source: Eurostat, national accounts and labour cost statistics, DG EMPL calculations. in employment of 7.3% over this
Data non-seasonally adjusted. period. In contrast, employment
of older workers aged 55 and over
has held up rather well, and had
wage and salary component, which 3.3. Which population even increased compared to the
increased at a year-on-year rate of second quarter of 2008. The low-
3.6% in the second quarter of 2009, subgroups have been skilled have undergone a much
while the non-wage component (e.g. most affected? stronger reduction in employ-
social ­contributions) grew by 3.8%. ment than other skill levels: their
employment dropped by around
The recent variation in the growth in 3.3.1. Impact across various 4.9% compared with a fall of
hourly labour costs over the year to population subgroups only 2.6% for the medium-skilled,
the first or second quarters resulted while for the high-skilled employ-
from rather different growth patterns Employment ment actually expanded by 3.1%.
across Member States. In the UK a ­F inally, although nationals saw
sharp drop in the first quarter (down Results from the European labour their employment decline by 1.6%,
by 5.5% year on year) was followed by force survey highlight that certain third-country (i.e. non-EU) nation-
a subsequent rise in the second quar- population subgroups have clearly als experienced a stronger decline
ter (up by 0.9% year on year). More been affected more than others by of 2.7%, but in contrast nationals
moderate average EU growth has also employment contraction during the of other EU countries saw their
resulted from negligible growth in current recession (Chart 38). employment rise by 2.3%.
hourly labour costs in France and Italy
and a slowdown in Poland, where the
growth halved compared with the Chart 38: Relative change in employment in the EU by sex, age,
peak in the first quarter of 2008. On skill level and nationality, 2008 Q2–2009 Q2
the contrary, in Germany and Spain 4
the year-on-year rise in labour costs
continued to grow in the second quar-
ter of 2009, at a rate far above the EU
average. For Germany this perhaps
indicates an end to the previous period
of relatively strong wage restraint.
% change


Apart from Germany, Poland and

Spain, hourly labour costs and their -4

wage and salary component rose

notably (by 6% or more year on -6
year) in Austria, Greece, Hungary
and Slovenia and most substantially
in Bulgaria and Romania, indicating Total Men Women 15–24 25–54 55–64 65+ Low Medium High Nationals Other EU Non-EU
nationals nationals
continued strong convergence to lev- Total Sex Age Skill level Nationality

els in the other Member States. Source: Eurostat, EU LFS. Data non-seasonally adjusted.

Employment in Europe 2009

Chart 39: Relative change in employment in the EU by occupation group, focus of the initial impact of the
2008 Q2–2009 Q2 labour market downturn on the
2 manufacturing and construction sec-
tors and in particular the low-skilled.
Skilled service-sector-based occupa-
tions experienced significantly lower
fallout from the crisis, with continued
-2 growth in most of the high-skilled
% change

non-manual occupations – employ-

ment levels have risen over the last
year for legislators, senior officials
and managers and professionals.

The above developments are also
Legislators, senior
officials and managers


Technicians and
associate professionals


Service workers and shop

and market sales workers

Skilled agricultural
and fishery workers

Craft and related

trades workers

Plant and machine

operators and assemblers

Elementary occupations
reflected in the recent evolution of
unemployment for the various popu-
lation subgroups. While the overall
Source: Eurostat, EU LFS. Data non-seasonally adjusted. EU unemployment rate has risen by
2.4 percentage points since the low
In line with these developments, the percentage points, but the rate for of March/April 2008, there is signifi-
overall employment rate declined by older workers aged 55–64 rose by cant underlying variation according
1.2 percentage points over the year to 0.6 percentage points. to gender, age group, skill level and
the second quarter of 2009, but the nationality (Chart 40).
decline for men was much stronger In terms of occupations, the hardest
(2.1 percentage points) than that hit have been workers in manual and Focusing on gender, the increase
for women (0.3 percentage points). elementary occupations (Chart 39). in the overall unemployment rate
Among different age groups, the Craft and related trades workers, has been driven predominantly by
employment rate for young people plant and machine operators and the rise in the rate for men (Chart
aged 15–24 declined by 2.4 percent- assemblers and those in elementary 40a). Since the average unemploy-
age points over the year to the sec- occupations have seen employment ment rate in the EU troughed in
ond quarter of 2009, and for those levels decline by around 3.4–6.7% spring 2008, the rate for men had
of prime working age (25–54) by 1.4 over the last year, reflecting the increased by 2.9 percentage points

Chart 40: Unemployment rates in the EU for various groups

(a) according to sex (b) according to age
15-24 25-54 55+
Women Men Total
10 20

7.5 15
% of labour force
% of labour force

5 10

2.5 5

0 0
Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2

2005 2006 2007 2008 2009 2005 2006 2007 2008 2009

(c) according to education attainment (d) according to nationality

Non-EU Other EU
Low Medium High nationals nationals Nationals
15 20
% of labour force

12.5 15
% of labour force

7.5 10

3.5 5

0 0
Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2
2005 2006 2007 2008 2009 2007 2008 2009

Source: Eurostat, EU LFS. Data non-seasonally adjusted.

Chapter 1 EU labour markets in times of economic crisis

to 9.1% by August 2009, and for en’s greater concentration in part- aged 15–24 (Chart 40b). By the second
women by 1.7 percentage points to time work, lower-paid jobs, jobs with quarter of 2009 the youth unemploy-
9%. Consequently, the gender gap in shorter tenure and smaller firms will ment rate was up 4.5 percentage
­unemployment rates, still at 1.2 per- all have an impact on not only the rel- points compared with a year earlier,
centage points in the beginning of ative effects of the downturn but also in comparison with rises of 1.8 per-
2008, not only disappeared, but even the extent to which policy responses centage points for people of prime
reversed: in June 2009, for the first benefit or disadvantage different working age and 1.1 percentage
time, the male unemployment rate groups. At the same time, most fore- points for older people aged 55–64.
exceeded the female one. casts still expect much of the labour However, this partly reflects the fact
market adjustment to the recent crisis that rates for young people have
In terms of the absolute rise in unem- to lie ahead (see section 3.5). As the been rising over a longer period (they
ployment in the EU, men account for effects of the economic crisis broaden started growing in the first quarter
two thirds of the rise and women beyond the male-dominated sectors of 2008) while for prime working-age
one third. Clearly prime-age men immediately affected, and as these and older people, they only clearly
(aged 25–54) have been the worse in turn provoke some job losses in started to increase from the last quar-
affected in the current downturn in female-dominated services sectors, it ter of 2008. Nevertheless, the upturn
absolute terms, accounting for almost seems likely that there may be a more for youth has been particularly sharp
half (47%) of the overall increase significant effect on female employ- over the first quarter of 2009.
in unemployment. Furthermore, the ment in coming quarters than that
rise in male unemployment has been observed so far. As unemployment rates for young
more substantial than that for women people were already substantial-
across all age groups, generally being Turning to unemployment develop- ly higher than those for other age
around double of the rise in female ments for different age groups, in groups, the strong deterioration for
employment for all age groups (Chart absolute terms two thirds (69%) of this age group is of particular con-
41). In terms of relative increase, the the rise in unemployment over the cern. The increase at EU level has been
level of male unemployment rose by last year is attributable to the increase driven by a sharp rise in the unem-
almost 40% over the year to the in unemployment for those of prime ployment rate for young men, which
second quarter of 2009, while for working age, a quarter (23%) to has been much more pronounced
women it rose by around 20%. youth and around 8.7% to older peo- than the rise for young women. It
ple aged 55 and over. However, in also mainly reflects a strong jump
However, measures of unemployment relative terms, the picture is more in youth unemployment in Spain,
may not necessarily capture the full even. Relative to the levels in the sec- together with significant increases in
impact of the changing economic ond quarter of 2008, unemployment France and Poland, although youth
conditions for women, especially as for all groups had risen by around unemployment rates had also risen
women are more likely than men to 30%: 27% for young people, 31% for in all other Member States compared
leave the labour market altogether. prime-age workers and by 29% for with a year earlier, and especially so
In addition, the difference in the types older people aged 55 and over. in the Baltic States and Ireland.
of job that women and men perform
will be reflected in the impact of the Unemployment rates have recently As a consequence of the decrease in
recession on their employment and been rising for all age groups, but youth employment, the share of young
unemployment; for example wom- particularly strongly for young ­people people aged 15–24 not in employ-
ment, education or training (NEETs)
Chart 41: Composition of the rise in unemployment by sex and age, 2008 Q2–2009 Q2 increased to 13% by the first quarter
Women 55+
of 2009, from below 12% a year ear-
3% lier, and risks becoming a significant
problem as the recession continues.
Men 15–24
16% Focusing on skill levels (Chart 40c),
Women 25–54
22% in line with the strong decline in
employment for the low-skilled. their
unemployment rates have increased
Women 15–24
7% by 3.4 percentage points over the
Men 25–54 year to the second quarter of 2009,
Men 55–64 47%
5% compared with rises of 1.9 percent-
age points for the medium-skilled and
1.1 percentage points for the high-
skilled. This reflects the fact that the
Source: Eurostat, EU LFS. Data non-seasonally adjusted. majority of the rise in ­unemployment

Employment in Europe 2009

consists of low- or medium-skilled elementary occupations (much more discuss three key priorities to address
people (accounting for 40% and so than non-migrants), and as craft the current situation: maintaining
44% of the rise in unemployment, and trades workers – i.e. in the low- employment, creating jobs and pro-
respectively) and much less so of the skilled occupations which have been moting mobility; upgrading skills and
­high-skilled (less than 16%). most at risk in the downturn. matching labour market needs; and
increasing access to employment.
In terms of nationality groupings, In summary, the population subgroups The comprehensive initiatives which
migrants have been disproportion- that have so far been most affected have been launched at EU level are
ally affected by rising unemployment, by the rise in unemployment have consistent with the overall strategy
especially those migrants originating been young people, the low-skilled, aimed at addressing these three key
from outside the EU – traditionally migrants (especially those originating priorities and include the following:
one of the most vulnerable groups on from outside the EU), and men rather
the labour market (Chart 40d). While than women (Chart 42). • The European Economic Recovery
unemployment rates for nationals Plan (EERP) – a €200 billion recovery
rose by 1.8 percentage points over the package. It represents a compre-
last year, those for nationals of other 3.4. What measures hensive, coherent and coordinated
EU countries rose 2.8 percentage have been taken to response to help mitigate the impact
points and for third-country nation- of the financial market crisis on the
als by an even stronger 5 percentage mitigate the impact real economy, and calls for fiscal
points. In the last quarter of 2008 of the crisis on labour stimulus and structural reforms at
and first quarter of 2009, the rise in both Member State and EU level
the unemployment rate was particu-
larly steep for non-EU nationals; their • The Employment Summit on 7 May
unemployment rate has been at least 2009 to boost efforts to promote
7–8 percentage points higher than 3.4.1. EU-level initiatives to employment and social inclusion in
that of nationals over recent years, promote employment the light of the financial crisis
but the gap widened significantly
to around 11 percentage points by Through several recent initiatives • Proposing changes to the European
the second quarter of 2009, while it and by reinforcing existing activi- Social Fund (ESF) and the European
remained only around 3 percentage ties, the EU has strengthened its Globalisation Adjustment Fund
points higher for other EU nationals. efforts to promote employment and (EGF) to ensure greater access and
The disproportionally strong reaction social inclusion, as part of its strat- support to the labour market:
of migrants’ unemployment in part egy to deal with the economic and
reflects that they are overrepresented financial crisis. The 2009 Spring Euro- - €19 billion in support is allocat-
in sectors such as construction, which pean Council and the three employ- ed through the European Social
has been particularly strongly affected ment workshops held in Madrid, Fund for 2009-2010. Assistance
by the economic downturn. Further- Stockholm, and Prague in 2009 in is provided to Member States
more, in terms of occupations, a high ­partnership with Member States and to put in place rapid reaction
share of migrants are employed in Social Partners helped define and packages focussing on the three
key priorities, with speeded up
Chart 42: Rises in unemployment rates by sex, age, skill level and nationality, procedures when adoption of
2008 Q2–2009 Q2 the ESF programmes is needed

- Access to the EGF is being made

easier and more effective to cover
redundancies caused by the crisis,
by co-financing training and job
placements. The EU funding rate
Percentage points

has been increased from 50% to

65% until the end of 2011 and the
eligibility threshold for EGF applica-
tions has been lowered from 1 000
1 to 500 redundant workers. In addi-
tion, the duration of EGF support
has been extended from 12 to 24
Total Men Women 15–24 25–54 55+ Low Medium High Nationals Other EU Non-EU months to provide sufficient time
nationals nationals
Total Sex Age Skill level Nationality for re-integrating particularly the
Source: Eurostat, EU LFS. Data non-seasonally adjusted. most vulnerable into new jobs

Chapter 1 EU labour markets in times of economic crisis

• Proposing the establishment of a 3.4.2. Recent employment nesses to use temporary short-time
new micro-finance facility to pro- measures undertaken by working arrangements (STWAs)
vide credit to people who have lost Member States to combat instead of making employees
their jobs and want to start their the employment effects redundant. These measures allow
own small businesses. The initial of the crisis companies to temporarily reduce
budget of €100 million could lever- work levels or wages below what
age more than €500 million in a At the current juncture, the main is stated in the contractual agree-
joint initiative with international labour market challenge is to avoid ment or to momentarily suspend
financial institutions, in particular redundancies in basically healthy all, or part of, its activity. In these
the European Investment Bank firms/industries temporarily affect- cases, any loss of employee salary
ed by the short-term disturbance to is, in almost all cases, partly or fully
• The New skills for new jobs initia- demand levels, while ensuring appro- compensated by the state.
tive, improving the analysis and pre- priate and necessary labour re-alloca-
diction of the future skills require- tion across sectors. Employment poli- It appears that in a number of Mem-
ments of the European economy, cies therefore need more than ever ber States, job losses have indeed
thereby matching people to jobs to focus on implementing integrated been contained so far, largely due
more efficiently and providing an flexicurity pathways and better skills to these measures. Where STWAs
insight into training needs matching and upgrading. In view of have been at their most effective,
this, various measures to facilitate they include making eligibility con-
• Stepping up monitoring of the labour market transitions and to sup- ditional on pre-specified criteria,
employment and social situation, port employment are required. These one of which may be the coupling
including through the publication by include strengthened activation poli- of STWAs with work-related train-
the European Commission of a new cies and better matching through ing to improve the employability
series of monthly monitoring reports effective employment services, flex- of workers and ease their possible
on the rapidly changing situation ibility in working-time arrangements transition to new jobs, as is the case
and where appropriate, lower social in Germany and Belgium for exam-
• The European Commission has charges for employers and employ- ple. A number of Member States
adopted a “shared commitment ees, especially for low-paid jobs, ensured that these measures are
for employment”, which puts for- together with an adequate safety well targeted by minimising the
ward key priorities and actions to net for workers made redundant. risk of protecting non-viable firms
preserve jobs and help those facing in order to ensure the economy can
difficulties while paving the way Following the European Economic reallocate resources to more pro-
for recovery Recovery Plan, most Member States ductive uses. For example, Austria
have established a National Recovery and Hungary operate STWAs which
• The Commission Communication Plan including provisions for employ- can only be used by firms that dem-
‘Driving European Recovery’ in ment and social policies and, based onstrate that their long-term finan-
March 2009 outlined a number of on information collected by the Com- cial position is sufficiently sound
elements to help Member States mission, almost 300 measures have according to minimum thresholds
design and implement appropriate been identified as being introduced set by government, while the Neth-
and effective employment policies across the EU. The measures set out erlands has solved this problem
as a response to the economic crisis below, organised according to the by providing lending schemes to
three priorities highlighted previous- businesses instead of subsidies.
• And last but not least, the Euro- ly, demonstrate examples of targeted Some Member States partly sub-
pean Employment Strategy, one and productive ways forward to try sidise short-time working only if
of the pillars of the EU’s Strategy to soften the impact of the crisis. the worker is fully employed again
for Growth and Jobs, continues to after the STWA period is over.
provide a framework for Member Maintaining employment, creat-
States to take coordinated action ing jobs and promoting mobility Upgrading skills and matching
to promote employment in the con- labour market needs
text of the crisis, including through One of the main priorities in the
joint work in the Employment EU is to avoid job losses, particu- Essential in the short term, skills match-
Committee and the ­identification larly in sectors and firms that were ing and upgrading is also the best
of best practices under the mutual fundamentally sound prior to the way to address structural changes and
learning programme.(15) crisis. In response to the deteriorat- exploit new opportunities for sustain-
ing employment situation, several able jobs, such as those relating to the
Member States have introduced, or shift to a low-carbon economy, green
are introducing, different forms of jobs and the development of new
(15) See
public support to encourage busi- technologies. Training and retraining

Employment in Europe 2009

are essential to assist occupational/ Many Member States have intro- work of the crisis, notably in Germa-
professional mobility in a mid- to duced measures to improve the ny, Belgium, France and Sweden. In
longer-term perspective, as many design and capacity of their ALMPs the Netherlands, wage moderation
people who lose their job during the and training offers to respond to over the medium term was traded
current downturn may not be able to the new, pressing needs resulting against cuts in social security con-
get back into their old job, occupa- from the widening of the crisis. The tributions for both employers and
tion or industry after the crisis. majority of interventions have been employees, whilst Hungary froze
characterised by clear targeting, to minimum wages in an attempt to
On skills upgrading, training oppor- adequately respond to changing preserve employment.
tunities and incentives have been needs and priorities, with measures
expanded in most countries in the face explicitly designed to support and Improving incentives to work which
of the crisis, with on-the-job training ease the re-integration into the are embedded in tax and ben-
increasingly recognised as a key tool labour market of recently laid-off efit systems has also been high on
for improving the employability of workers being adopted in at least policy agendas, especially for low-
those already in employment. Roma- 10 Member States. One example of wage earners. Income supplements
nia, for example, introduced 50% such action is that in the Nether- and targeted in-work tax credits
support for the costs of continuous lands, where social partners have were reinforced in some Member
vocational training for both employ- agreed to ensure an apprentice- States, and in others the design
ees and the unemployed, while Ger- ship place for every school-leaver of unemployment insurance was
many established a new programme who has been unemployed for at modified so as to increase work
to support further vocational training least three months. Other unem- attractiveness.
for temporary workers, through a ployed people aged up to 27 will,
system of training vouchers. Portugal according to proposed legislation,
has expanded the provision of job receive a work/learning offer from 3.5. Outlook
training to recipients of minimum their municipality.
income, and enhanced financial sup- Since developments in the labour
port for access to education was intro- Increasing access to employment market tend to follow those in eco-
duced in Sweden and Austria. nomic activity with a certain lag, the
Rebates on social security contribu- main reaction to the sharp down-
On better matching labour market tions to boost labour demand dur- turn in economic activity in autumn
needs, modernising and improving ing the crisis have been introduced last year only started to take effect
the administrative capacity of Public in Belgium, Spain, France, Hungary, in the course of 2009. Indeed, the
Employment Services (PES) has been a Portugal, Sweden and Slovakia and relatively limited impact the eco-
central focus for a number of Member are typically made conditional upon nomic downturn has had so far on
States including Germany, Denmark, job creation. However, rebates can the labour markets in the EU may
Greece, Spain, France, Hungary, Swe- have an eroding effect on the long- worsen once the sharp acceleration
den, Slovakia and the UK. This has been term sustainability of social security in the economic downturn in Octo-
in response to the growing numbers of systems, meaning that it is critical ber fully feeds through to the labour
new unemployment benefit applicants therefore to ensure that any such market following the usual lag of
and clients in need of labour market rebates are temporary. Other meas- about 2–3 quarters.
assistance, as well as a higher skills ures that have been introduced by
profile for clients than before. Member States are often targeted Furthermore, the difficulties in find-
at those most difficult to employ, ing people with the appropriate
Some countries also focused on rein- sometimes to SMEs (France and Por- skills have encouraged employers
forcing the preventative arm of their tugal) or to the self-employed (Slo- in the EU to try to hold on to
active labour market policies (ALMPs). vakia and Slovenia). In a few cases, the experienced workers they have;
The Netherlands, for example, devel- the ­fiscal boost has been directed however, the longer the downturn
oped ‘mobility centres’, temporary towards sustaining employment in continues, the harder this will be
public–private partnerships aimed specific sectors, such as household- to maintain. The relative resilience
at intensifying timely assistance to related employment services and of the EU labour market, result-
jobseekers and businesses, to pre- building maintenance, or strategic ing from a combination of factors
vent forced lay-offs as far as possi- activities, or sectors such as research including widespread application of
ble. Employees who are threatened and development and investment internal adjustment measures aimed
by unemployment will be assisted and ­renewable energies. at increasing flexibility to adjust
in finding a new job, or temporarily to the drop in demand (e.g. short-
be sent on secondment with other Lowering labour costs for both time working, temporary suspension
employers, aided if necessary through employers and employees gained of production), skill shortages and
additional education and training. additional relevance in the frame- longer-term concerns over a shrink-

Chapter 1 EU labour markets in times of economic crisis

ing labour force, may weaken if the 3.5.1. Forecasts for the expected to contract by more than
economy continues to contract and outlook to 2010 5%), while particularly strong falls
no signs of an improvement are (of around 7–9%) were foreseen for
forthcoming in the near future. Most forecasts available in early 2009 the Baltic States and Ireland. Even
painted a rather gloomy picture for countries with a traditionally strong
In this context, although the latest economic and employment pros- labour market performance such as
data available shows that the EU pects in the near future and pointed Denmark, the Netherlands and Swe-
labour market continues to dete- towards a sharp economic downturn den faced noticeable employment
riorate, there are increasing signs in 2009 and a long path to recovery contraction in 2009 and 2010.
that the pace of deterioration is (Table 9). All suggested that eco-
moderating. In particular, economic nomic activity in the EU would only The unemployment rate was expect-
sentiment, firm’s employment expec- recover slowly, and that much of ed to increase substantially in the EU,
tations and consumers’ unemploy- the adjustment in the labour market rising by about 4  percentage points
ment expectations, although remain- still lay ahead. Member States with on 2008 levels to around 11% in
ing pessimistic, show clear signs of stronger exposure to the effects of 2010. Furthermore, rates were pro-
improving. However, the unemploy- the housing and financial market jected to increase in all Member
ment rate continues to increase in corrections and those facing chal- States between 2008 and 2010, the
line with the prior worsening expec- lenges in terms of external competi- most marked rises being in the Baltic
tations; moreover, it can be expected tiveness were expected to perform States, Ireland and Spain (all up 8–10
to deteriorate further before the particularly weakly. percentage points), which all faced
lagged effect of the turnaround in substantial downturns in (construc-
economic sentiment and consumers’ In the Commission’s spring economic tion) activity. The strong increase in
expectations takes hold, alongside forecast(17), EU GDP was projected to unemployment in Spain was project-
any associated upturn in economic fall by 4% this year and to broadly sta- ed to lead to an unemployment rate
activity. Indeed, a recent Eurobarom- bilise in 2010 as the impact of fiscal and in excess of 20% by 2010, while rises
eter Survey(16) published in July shows monetary stimulus measures kick in. were also foreseen to be fairly sub-
that 61% of Europeans think further The economic downturn was expect- stantial (2–5 percentage points) in all
impacts of the economic crisis on ed to be broad-based across Member the other larger Member States.
jobs are to be expected, with close States, but particularly severe in Ire-
to a third (32%) of those in work land and in the Baltic States. Among In its June Economic Outlook the
concerned that they may lose their larger Member States, Germany and OECD also warned that the impact
jobs in the crisis. According to the Italy faced the strongest contraction in of the crisis on labour markets will
survey, around 9% of those polled 2009 (of 5.4% and 4.4% respectively), be felt for a long time. Unemploy-
had already personally experienced while for France, Spain and the UK it ment in the USA was expected to
job loss due to the crisis, although was expected to be slightly more mod- surpass 10% by the end of this year
some had since found a new job erate (around 3–4%). and to remain at that elevated level
resulting in an overall net rate of throughout 2010, while the unem-
job loss of 6%. Furthermore, 24% of The forecast projected that labour ployment rate in the euro area was
European citizens know a colleague markets would be severely affect- projected to soar to around 11% by
who has lost their job and 36% know ed by the downturn. Employment the end of this year and continue to
someone from among their friends growth in the EU (and in all Mem- rise to 12.3% by the end of 2010.
or family who has been affected, ber States except Luxembourg) was
confirming the widespread social expected to turn negative this year, However, more positive signs that
impact of the crisis. The ­expectations with overall employment contract- the global downturn might be bot-
of ­further deterioration in the labour ing by 2.6% in 2009 and by a fur- toming out started to be reported
market are confirmed by recent busi- ther 1.4% in 2010. This equates to in the second half of the year. In
ness surveys, which indicate that about 8.5 million job losses for the its September interim forecast, the
overall firms’ employment expecta- two years, in contrast with the net European Commission highlights
tions for the months ahead generally job creation of 9.7 million between that signs for an imminent economic
remain unfavourable. 2005 and 2008. All larger Member recovery are apparent, with fears of
States were projected to experience a prolonged and deep recession fad-
declining employment in 2009, espe- ing, although the sustainability of
cially Spain (where employment was the recovery remains to be tested.
GDP growth is set to turn posi-
(17) The Commission’s spring economic fore-
tive in the second half of the year.
cast was finalised in April 2009 (avail-
(16) Special Eurobarometer 316, European able at However, the forecast for 2009 as
Employment and Social Policy (see the finance/publications/publication_sum- a whole remains unchanged, with
website mary15046_en.htm). A new forecast will
GDP expected to fall by 4% in both
ion/archives/ebs/ebs_316_sum_en.pdf) be released in November.

Employment in Europe 2009

Table 9: Comparison of recent European Commission, OECD and IMF forecasts

European Commission Economic OECD Economic Outlook No 85, June IMF World Economic Outlook April 2009
Forecasts (Spring 2009) 2009
GDP Employment UR GDP Employment UR GDP Employment UR
2009 2010 2009 2010 2009 2010 2009 2010 2009 2010 2009 2010 2009 2010 2009 2010 2009 2010
Belgium -3.5 -0.2 -1.2 -1.5 8.5 10.3 -4.1 -0.5 -0.6 -1.8 8.3 10.6 –3.8 0.3 9.5 10.5
Germany -5.4 0.3 -1.5 -2.2 8.6 10.4 -6.1 0.2 -1.9 -3.2 8.7 11.6 –5.6 –1.0 –0.4 –2.0 9.0 10.8
Ireland -9.0 -2.6 -9.0 -4.0 13.3 16.0 -9.8 -1.5 -8.4 -3.7 12.2 14.8 –8.0 –3.0 12.0 13.0
Greece -0.9 0.1 -1.1 -0.1 9.1 9.7 -1.3 0.3 -1.6 -0.4 9.5 10.3 –0.2 –0.6 9.0 10.5
Spain -3.2 -1.0 -5.3 -2.7 17.3 20.5 -4.2 -0.9 -7.0 -3.2 18.1 19.6 –3.0 –0.7 –3.5 –1.0 17.7 19.3
France -3.0 -0.2 -2.2 -1.2 9.6 10.7 -3.0 0.2 -1.3 -1.2 9.7 11.2 –3.0 0.4 –1.8 –0.6 9.6 10.3
Italy -4.4 0.1 -3.3 -0.6 8.8 9.4 -5.5 0.4 -1.5 -2.0 8.4 10.2 –4.4 –0.4 –1.7 –1.5 8.9 10.5
Cyprus 0.3 0.7 -0.4 0.1 4.7 6.0 0.3 2.1 4.6 4.3
Luxembourg -3.0 0.1 0.5 -0.8 5.9 7.0 -4.0 -0.4 0.8 -1.0 6.0 7.2 –4.8 –0.2 6.8 6.0
Malta -0.9 0.2 -0.5 0.2 7.1 7.6 –1.5 1.1 6.9 7.6
Netherlands -3.5 -0.4 -1.0 -2.8 3.9 6.2 -4.9 -0.4 -1.0 -3.5 4.0 7.0 –4.8 –0.7 4.1 5.0
Austria -4.0 -0.1 -2.7 -0.9 6.0 7.1 -4.3 -0.1 -1.4 -1.8 6.1 7.9 –3.0 0.2 5.4 6.2
Portugal -3.7 -0.8 -1.4 -0.6 9.1 9.8 -4.5 -0.5 -2.8 -1.9 9.6 11.2 –4.1 –0.5 9.6 11.0
Slovenia -3.4 0.7 -4.7 -0.6 6.6 7.4 –2.7 1.4 6.2 6.1
Slovakia -2.6 0.7 -1.7 0.4 12.0 12.1 -5.0 3.1 -2.5 -2.2 11.8 13.6 –2.1 1.9 11.5 11.7
Finland -4.7 0.2 -2.9 -0.8 8.9 9.3 -4.7 0.8 -3.2 -3.3 8.7 10.8 –5.2 –1.2 8.5 9.3
Euro area -4.0 -0.1 -2.6 -1.5 9.9 11.5 -4.8 0.0 -2.5 -2.4 10.0 12.0 –4.2 –0.4 –1.6 –1.3 10.1 11.5
Bulgaria -1.6 -0.1 -2.2 -1.0 7.3 7.8
Czech Republic -2.7 0.3 -1.7 -1.3 6.1 7.4 -4.2 1.4 -2.1 -2.4 6.9 9.2 –3.5 0.1 5.5 5.7
Denmark -3.3 0.3 -2.2 -2.0 5.2 6.6 -4.0 0.1 -3.1 -3.5 6.0 7.9 –4.0 0.4 3.2 4.5
Estonia -10.3 -0.8 -7.3 -3.3 11.3 14.1
Latvia -13.1 -3.2 -8.9 -3.3 15.7 16.0
Lithuania -11.0 -4.7 -7.7 -2.4 13.8 15.9
Hungary -6.3 -0.3 -3.0 -2.0 9.5 11.2 -6.1 -2.2 -3.4 -1.5 10.7 11.7
Poland -1.4 0.8 -2.3 -1.4 9.9 12.1 -0.4 0.6 -1.5 -2.9 9.0 11.6
Romania -4.0 0.0 -2.2 0.6 8.0 7.7
Sweden -4.0 0.8 -2.4 -2.3 8.4 10.4 -5.5 0.2 -3.1 -3.9 8.7 11.4 –4.3 0.2 8.4 9.6
United Kingdom -3.8 0.1 -2.4 -0.9 8.2 9.4 -4.3 0.0 -2.3 -2.6 8.2 9.7 –4.1 –0.4 –1.7 –1.4 7.4 9.2
EU -4.0 -0.1 -2.6 -1.4 9.4 10.9
USA -2.9 0.9 -3.5 -0.9 8.9 10.2 -2.8 0.9 -3.2 0.1 9.3 10.1 –2.8 0.0 –2.6 0.1 8.9 10.1
Japan -5.3 0.1 -3.0 -1.2 5.8 6.3 -6.8 0.7 -1.5 -1.1 5.2 5.7 –6.2 0.5 –1.2 –1.6 4.6 5.6

Source: European Commission spring forecasts, OECD Economic Outlook No 85 (June 2009) and IMF World Economic Outlook (April 2009).

the EU and the euro area this year. the June forecast, the latest GDP 4. Summary and
Despite the encouraging recent eco- forecasts for 2009 provided a slightly
nomic signals, the labour market improved outlook for Japan and an conclusions
situation is expected to deteriorate unchanged one for the USA, and
further in the second half of 2009. revised up the projection for the The unprecedented crisis in global
The full impact of the economic euro area (up from –4.8% to –3.9%). financial markets which gathered
crisis on labour markets is, at least The OECD forecast for the EU reflects pace in autumn last year has led
partly, still to be faced. an improved outlook for France, Italy to the most severe recession since
and especially Germany, but points the Second World War, affecting
In its September Interim Assess- to a gloomier situation in the UK. As the wider economy and increasingly
ment of the Economic Outlook, the a consequence, the unprecedented impacting labour markets in the EU.
OECD confirms that recovery from rate of deterioration in labour mar- After many years of relatively high
the global recession is likely to be ket conditions witnessed in the EU growth and job creation, the global
earlier than had been expected a over the past year should ease, but financial crisis and its repercussions
few months ago, but that the pace the pace of labour market as well as on the real economy have brought
of recovery is likely to remain weak economic recovery will likely be lim- about a sudden reversal of the previ-
well into next year. Compared with ited for some time to come. ous period of positive employment

Chapter 1 EU labour markets in times of economic crisis

growth. This has set new challenges turn than the latter, reflecting that s­ uccessfully adjust to these chang-
for policy-making and research, and many of the sectors hit hardest by ing realities: to retain sound jobs,
as unemployment continues to rise, the crisis are predominantly male- enhance skills at all levels, get peo-
the spotlight has fallen more and oriented in terms of employment. ple back to work and set the condi-
more on limiting the effect of the There has also been a continued tions for new job creation. Flexicu-
crisis on jobs and addressing the strong rise in unemployment among rity remains the right approach to
social impact. young people, with young men both modernise labour markets and
being particularly affected, high- ensure a ­successful recovery.
Although the economic crisis has lighting a rising need for support to
had a major impact on economic tackle youth ­unemployment. In this context this year’s report
growth in the EU, the effect on the focuses on the themes of ‘Labour
labour market was rather limited in The top employment challenge for flows, transitions and unemployment
2008. This is in part due to the usual the EU must be to minimise job duration’ and of ‘Climate change
lags of six or more months before losses, prevent unemployment from and labour market outcomes’. With
output changes affect employment becoming entrenched (i.e. becom- regard to the former, the current
levels, together with the fact that ing long-term unemployment), recession and the sharp increase in
labour demand started to adjust favour transitions back into employ- unemployment have highlighted the
through flexible working arrange- ment and boost job creation, and importance of implementing meas-
ments (e.g. short-time working pave the way for economic renewal ures to facilitate transitions into and
schemes and shorter working hours, and for sustainable recovery and back to work and have increased
temporary closures etc.) together growth. In this context, acting in the potential pay-off of strength-
with nominal wage concessions in concert, the EU and Member States ening the effectiveness of existing
return for employment stability in have already taken important steps labour market policies. In particular
some sectors, rather than through to limit the impact of the crisis it underscores the importance of acti-
a reduction in employment. Signs on labour markets and create the vation and job search assistance serv-
point to substantial labour hoard- ­conditions for recovery. ices delivered through Public Employ-
ing, given that most of the adjust- ment Services, to reduce the labour
ment so far seems to have been in The ‘Employment in Europe 2009’ market impact of the recession.
terms of productivity declines rather report reflects two key factors influ-
than employment losses, and there encing the policy work of the Euro- In light of this, assessment of policies
is a risk that unless the economy pean Commission this year: first, targeted at reducing the incidence
picks up soon firms may start to shed the current context of the economic and duration of unemployment
jobs at a faster pace. The negative crisis and the need to respond to and fostering future job creation
impact on employment has become the expected rapid rise in unemploy- should receive added attention at
more manifest in 2009, and policies ment, and, second, the fact that the present juncture. This should be
to assist economic recovery and miti- the current Lisbon cycle is coming particularly the case for those that
gate the loss of employment have to an end, requiring assessment of increase moves from unemployment
gained in importance, remember- its achievements and shortfalls, and or inactivity back to employment and
ing, however, that any pick-up in an adequate reformulation of policy strengthen attachment to the labour
employment will also lag behind any priorities for post-2010. market of groups at the margin.
recovery in output. Moreover, the focus on unemploy-
Europe must not only tackle the reces- ment and its duration not only helps
Although already on the rise since sion but it must also turn it into an to better inform the public debate
March 2008, unemployment has opportunity to create a more produc- and underpin policy action, but also
been growing more strongly since tive, more innovative, better skilled represents an indispensable step in
last October in reaction to the and greener economy – one with the reassessment of the Lisbon Strat-
heightening of the economic crisis, open and inclusive labour markets egy in order to prepare the post-2010
before showing some signs of slow- offering a more cohesive society, bet- agenda. In line with flexicurity poli-
ing over mid-2009. Although felt in ter opportunities for all, and jobs that cies, the analysis focuses on condi-
all Member States, the onset of the are responsive to age, gender equality tions that favour ‘good’ transitions
increase and its severity vary widely and work/life balance concerns. This (e.g. from unemployment/inactivity
across countries. At the same time, cannot be a one-off effort but rather to employment) and limit the rise
the crisis appears to be affecting a continuous collective process. and effect of long-term unemploy-
some groups of workers more deep- ment. In the current recession, it
ly than others. Although men still European labour markets will be is crucial to assess the conditions
have higher employment rates than changed profoundly by the crisis (both structural and policies) that
women, to date the former have and workers and companies must can favour the speed and quality of
been more affected by the down- be given the necessary means to labour market transitions (e.g. job

Employment in Europe 2009

creation) in order to support a rapid competitiveness and improve pub- promoting labour market policies
economic recovery and minimise time lic health. The scope for the crea- along flexicurity principles so that
spent in unemployment, because of tion of new ‘green jobs’ and the workers can be smoothly reallocated
its lasting negative effects on human greening of existing jobs is signifi- towards less polluting activities and
capital and employability. cant and covers all types of worker. labour markets and workplaces can
Nevertheless, in order to exploit become more receptive to experi-
Regarding the second major theme – these opportunities effectively, ade- mental innovations. Special atten-
‘Climate change and labour market quate policies are required as mar- tion should also be paid to adequate
outcomes’ – by now there is a gen- kets may not necessarily tackle these training and education schemes in
eral consensus that a shift towards a problems because of market fail- order to avoid the emergence of
competitive low-carbon economy is ures. An adequate policy response skill gaps and shortages. In addition,
a pressing priority requiring imme- should be driven by an integrat- such policies should be complement-
diate action and that the current ed approach. Coordination should ed by social spending focused on
economic crisis should not hinder ensure that economic, employment, items that accommodate the tran-
this shift. This is particularly perti- social, energy, transportation and sition process in an active way. As
nent if one takes into consideration environmental policies are mutu- is the case with climate change,
that tackling climate change and ally reinforcing and in line with long-term care also constitutes a
other environmental challenges can the EU Lisbon Strategy for Growth structural challenge for European
be combined with major opportuni- and Jobs. It is important that short- labour markets, and is likely to have
ties to develop new technologies, term actions should reinforce long- a substantial impact in the years to
create new jobs, enhance energy term strategic goals. Overall, there come. For this reason, it is given a
security, increase ­i nternational is a strong case to be made for brief special focus in chapter 3.

Chapter 2

Labour flows, transitions

and unemployment
1. Introduction A successful flexicurity(4) strategy essen- number of workers being hired and
tially aims to balance the income insur- undergoing separations, respectively.
The European Union (EU) Common ance function of unemployment ben- Transition indicators have been devel-
Principles of Flexicurity(1), agreed efit systems with appropriate labour oped more recently in close connec-
among all Member States and sup- market ‘activation’ mechanisms. Such tion with the EU employment policy
ported by social partners, have been mechanisms are designed to facilitate debate, essentially to capture the likeli-
endorsed by the European Council.(2) the transition of displaced or mobile hood of an individual moving between
Recently, EU leaders have further workers into employment, and fos- different labour market statuses, con-
underlined the importance of those ter career development and upward tractual types and income levels. Thus,
principles in terms of helping manage mobility generally, rather than pro- they also provide vital information on
the employment effects and social tecting existing non-profitable jobs the quality of labour market dynam-
impact of the current recession, and through high dismissal costs. Meas- ics. Unemployment duration indicators
preparing for the economic upturn.(3) ures for activation include, inter alia, have also a long-established tradition
more and better spending on active as dynamic indicators, particularly to
The severity of the current recession labour market policies, access to qual- measure duration dependence(5) in
and the risks associated with protract- ity placement and counselling services unemployment and the effects of poli-
ed periods of high unemployment for the unemployed, a reduction in the cies and institutions on the ­duration of
have heightened the need to monitor disincentives of tax-benefit systems, unemployment.
movements in the labour market in and policies promoting labour market
ways that enable timely public policy attachment, particularly for those at It can be shown that there is an equi-
response. In particular, measures are the margin of the labour market or at librium relation between the unem-
required to foster re‑employment and risk of becoming prematurely inactive ployment rate, the job reallocation
avoid a rise in long-term unemploy- (e.g. early retirees). rate and unemployment duration,
ment, which could eventually lead to although it also depends on other
a permanent loss in welfare associ- The EU flexicurity agenda calls for an factors, such as institutional aspects
ated with the deterioration in human effective monitoring of the situation of the labour market. This provides a
capital and the reduced employability and of progress based on detailed strong argument for jointly analysing
of jobless people. analysis. Labour market dynamics can labour market flows and unemploy-
be characterised by various indicators, ment duration in this chapter. All other
(1) COM(2007)359 of 27 June 2007. such as flows, transitions and (unem- things being equal, job reallocation
(2) Council Conclusions (16201/07), as adopt- ployment) duration. All these indicators is inversely related to unemployment
ed at the Council (EPSCO) on 5/6 Decem- measure some aspect of adjustment in duration (and long-term unemploy-
ber 2007.
European Council Conclusions the labour market. The calculation of ment). In fact, low unemployment out-
(16616/1/07) of 14 December 2007. flow indicators, comprising both job flows resulting from factors, such as
(3) ‘Flexicurity in times of crisis’, Council Con- and labour indicators, has a long tradi- inadequate job matching or high dis-
clusions (10388/09), as adopted at the tion. They ­measure the number of jobs missal costs, are likely to be associated
Council (EPSCO) on 8/9 June 2009. It calls
being created and destroyed or the with high unemployment duration.
for measures supporting the adjustment
of European labour markets, emphasis-
ing investments in human capital, such as (4) DG EMPL (2007), ‘Towards Common Prin- (5) Duration dependence measures how the
retraining, skills-upgrading and improved ciples of Flexicurity: More and better jobs likelihood of leaving unemployment var-
matching of labour market needs. through flexibility and security’. ies with time already spent unemployed.

Employment in Europe 2009

In addition to the general concerns The third section calculates various indi- The causes of gross flows can be roughly
that motivate flexicurity principles, the cators of unemployment duration, using classified in terms of demand and sup-
severity of the current recession implies both EU LFS and EU SILC data. Based ply. The former reflects the need of firms
the risk of a prolonged rise in the on the former, it calculates indicators to adjust labour inputs to changes in
average duration of unemployment, on the average duration of incomplete final demand, competitiveness or tech-
and of a greater incidence of long- and completed spells of unemployment, nology requirements. This leads to the
term unemployment. As illustrated recognising that the former tends to destruction of jobs that are no longer
by the literature on unemployment overestimate the length of time spent in productive and to the creation of jobs in
hysteresis,(6) these developments may unemployment. It also goes beyond the expanding firms and sectors. In contrast,
in turn trigger an increase in the struc- use of cross-sectional data, using the EU labour supply factors reflect movements
tural unemployment rate (NAIRU(7)), SILC longitudinal component to evalu- of workers from one job to another, as
preventing a timely return to pre-cri- ate the incidence of long-term unem- well as between employment and non-
sis levels. This illustrates the need to ployment and the recurrence of unem- employment, in their search for better
closely monitor unemployment dura- ployment spells based on ­alternative pay, working conditions, and an overall
tion, alongside flows and transitions, in reference periods. improved work–life balance.
order to timely propose evidence-based
corrective action, such as better public In addition, the third section provides a Models that incorporate both demand
employment services (PES), or the right brief overview of the literature on the and supply factors (e.g. Mortensen and
­investments in human capital. relationship between levels and dura- Pissarides, 1994) highlight the central
tion of unemployment compensation role of such dynamics/transitions in
This chapter is divided into three sec- and the length of unemployment spells, the functioning of labour markets,
tions. The first section mainly uses data as well as the relationship between in which job separations – both quits
from the EU Labour Force Survey (LFS) to labour market institutional arrange- and dismissals - job vacancies, and job
calculate indicators of labour flows, with ments, notably in terms of employ- matching occur simultaneously. How-
results presented for EU aggregates and ment protection legislation and labour ever, due to the wide range of dif-
by country, as well as for workers with market policies, and the incidence of ferences in the capabilities and needs
different characteristics. An analysis of long-term unemployment. The chap- of both firms and workers, as well as
variance is carried out for differences in ter ends with estimates of long-term differences in their knowledge about
the rate of hiring, in an attempt to dis- unemployment up to 2010 based on conditions and possibilities open to
entangle the effects of different sectors recent macro-economic projections. them, there are significant mismatches
and country institutions. Detailed calen- and imbalances in the labour mar-
dar information from the longitudinal(8) ket. As a result, it is prevented from
component of the EU Survey on Income
and Living Conditions (SILC) is also used
2. Labour flows reaching an instantaneous ­equilibrium
­without adjustment costs.
to ­calculate indicators of labour flows.
2.1. Labour flows In practice, the matching process
The second section calculates time between persons who are unemployed
series for indicators of labour market and transitions in the and the job vacancies available in firms
transitions based on the EU LFS. Using labour market is lengthy and costly, with considerable
the Hodrick-Prescott (HP) filter, a sta- resources being devoted by firms, indi-
tistical technique to decompose a time The performance of labour markets is viduals and public agencies with the
series in trend and cycle, an attempt normally assessed by looking at static aim of forming productive matches. In
is also made to distinguish the effects variables, such as the share of the work- this context of job matching, it is clearly
of cyclical and structural changes on ing-age population or of the labour important to fully assess the character-
the transition indicators. In addition, force who are employed or unem- istics of both job and worker or labour
results are presented for EU aggre- ployed at a specific point in time (i.e. flows if the ‘allocative efficiency’ of our
gates and by country, as well as for the employment and unemployment labour markets is to be improved – i.e.
different categories of workers. rates), or changes in those shares over their ability to adequately match peo-
time. However, the literature shows that ple with available jobs as efficiently and
such figures of net employment growth as cheaply as possible.
(6) In the 1970s–1980s, the experience of or job ­creation can give a misleading
persistent high unemployment in Europe
impression of the underlying dynamics It is against this background that this
led to the development of theories of
unemployment centred on the notion of the labour market, given that the section of the report studies in detail
that the equilibrium unemployment rate gross flows into and out of employment what is known, or can be inferred,
depends on past unemployment rates
are much higher than the net results.(9) about the nature of labour market
(i.e. unemployment hysteresis).
flows and transitions in order to
(7) Non-accelerating inflation rate of unem- (9) These movements may take several forms,
such as the simultaneous creation and achieve a better understanding of the
ployment (NAIRU).
destruction of a large number of jobs or functioning of labour markets, which
(8) Longitudinal data track individuals significant flows of workers between jobs
can lead to more effective policies.
over time. or in and out of employment.
Chapter 2 Labour flows, transitions and unemployment duration

But beforehand, it might be ­useful between the unemployment rate, the rationale for jointly analysing in this
to present more formalised ­analytical job reallocation rate and unemploy- chapter labour market flows and
accounts of the relation in ­equilibrium ment duration, providing the basic ­unemployment duration (Box 1).

Box 1: Job flows, the unemployment rate and unemployment duration

In equilibrium, the relation between the unemployment rate, job flows and unemployment duration can be expressed using the following
identity (Layard et al., 1991, pp. 220-221).

(Equation 1)

Equation 1 assumes that the labour force is constant, and thereby in equilibrium labour inflows equal outflows (I=O).
According to Equation 1, the unemployment rate is directly related to the inflow rate and inversely related to the outflow rate, or equivalently
directly related to the product between the inflow rate and the duration of unemployment.
After straightforward manipulation of some basic identities, Garibaldi et al. (1997) derive the following equilibrium relation between job
­reallocation/turnover, the unemployment rate, and unemployment duration. Equation 2 is more general than Equation 1, because it does not
require in equilibrium that the labour force be constant, allowing instead for the labour force to grow at an exogenous rate n.

(Equation 2)

Equation 2 gives the equilibrium relation between job reallocation, the unemployment rate, and unemployment duration, providing a strong
argument for jointly addressing in this chapter issues of labour flows and unemployment duration.
However, Equation 2 should not be interpreted as a reduced form equation from which it would be possible to infer causal relations. In fact,
institutional settings and policies, such as tax-benefit policies, employment protection legislation, active labour market policies, retirement
policies are likely to simultaneously influence all the 5 variables in Equation 2: jr, u, d, q, n.
According to Equation 2, in a steady state equilibrium and all other things being equal, job reallocation is inversed related to unemployment
duration. For a given unemployment rate, Equation 2 identifies the existence of an apparent trade-off between job reallocation and unemploy-
ment duration (mediated by the unemployment rate). A(n) (exogenous) lowering (rise) of the equilibrium unemployment rate tends to alleviate
(worsen) that trade-off, shifting to the left (right) the locus of Equation 2 drawn in jr and d (Chart 1a).
Chart 1a: How changes in the equilibrium unemployment rate affect the ‘apparent’
trade-off between job reallocation and unemployment duration
jr: job reallocation rate

0.01 d: unemployment duration

Employment in Europe 2009

2.2. Job and labour job flows. Indicators of job flows market dynamics. Each job that is
are usually calculated using business created or destroyed corresponds to
flows - definitional and a hiring or a separation. However,
­surveys or administrative data.
measurement issues the reverse does not necessarily hold
On the other hand, Labour flows since hirings and separations may
A number of indicators have been are associated with developments in also occur because of independent
used in order to measure job and labour supply. The labour turnover decisions by workers without affect-
labour market flows. They are gener- (LT, or labour reallocation) indicator ing the overall number and distribu-
ally classified in terms of job flows, measures the number of workers tion of available jobs.
and flows of workers or labour. who either change employment sta-
tus (e.g. from employment to unem- JT corresponds to the share of LT
Job flows are essentially associ- ployment) or move between jobs. LT that is driven by labour demand fac-
ated with developments in labour is the sum of the two components tors or, in other words, the labour
demand. They include indicators of – the number of hirings (H) and flows that are needed in order to
job creation (JC), job destruction the number of separations (S) – and accommodate changes in the total
(JD) and job turnover (JT, or job real- is the conceptual equivalent of JT amount and distribution of jobs.
location). JC measures the number above. Indicators of labour flows are On the other hand, the difference
of jobs created in a given period by usually calculated using labour force between LT and JT is called the
opening and expanding firms, while surveys or administrative data. Churning Flow (Dale-Olsen, 2006),
JD indicates the number of jobs and represents the share of LT that
destroyed by contracting or closing Comparisons between job and labour is due to labour supply factors.
firms. JT, in this context, is the sum turnover indicators can shed light on Box 2 provides more formal and
of JC and JD and provides an over- the relative importance of supply ver- detailed information on how these
all quantitative measure of total sus demand factors in driving labour indicators are defined.

Box 2: Definition of job and labour flow indicators

This box explains some standard job and labour flow indicators common in the literature, together with their relationships. It largely follows
Davis and Haltiwanger (1995) and Bertola et al. (1999).

Job flows

JC is the sum of two components:

1. The number of jobs created in opening (O) establishments – i.e. establishments with zero employment in period t and positive
employment in t+1
2. The number of new jobs created in expanding (E) establishments – i.e. establishments existing both in period t and t+1 but with
larger employment in t+1

(1) Hence, JC = O + E

Symmetrically, JD is the sum of two components:

1. The number of jobs destroyed in closing (CL) establishments – i.e. establishments which are shut down between period t–1 and t.
2. The number of jobs destroyed in contracting (CO) establishments – i.e. continuing establishment registering a decline in
employment between period t–1 and t.

(2) Hence, JD = CL + CO

JT or gross job reallocation is the sum of JC and JD; therefore it adds all employment gains and losses which have occurred at establishment
level between period t–1 and t:

(3) JT = JC + JD

The difference between JC and JD equals net employment growth (NEG):

(4) NEG = JC – JD

Excess job turnover (EJT, or excess job reallocation) is the difference between JT and the absolute value of NEG. EJT represents the amount of
job flows over and above what is needed to accommodate net job growth.

(5) EJT = JT – ⎪NEG⎪

Chapter 2 Labour flows, transitions and unemployment duration

Labour flows

LT gives the number of persons who have changed labour status (e.g. from employment to unemployment) or moved between jobs between
period t–1 and t.

It is the sum of two components:

1. Hirings (H): the number of recruitments.
2. Separations (S): the number of dismissals/quits.

(6) LT = H + S

In this chapter, EU LFS data are used to calculate LT(10). H is calculated as the sum of three components:
1. The number of workers who moved from unemployment to employment (U_E)(11)
2. The number of workers who moved from inactivity to employment (I_E)
3. Job-to-job mobility – i.e. the number of workers who were employed both at time t and t–1, but with tenure lower than 1 year with
their employer in period t (E_E<1y)(12).

(7) H = U_E + I_E + E_E<1y

In this chapter, S are calculated as the sum of two components:

1. The number of workers who moved from employment to unemployment (E_U)
2. The number of workers who moved from employment to inactivity (E_I)

(8a) S = E_U + E_I

(8b) LT = U_E + I_E + E_E<1y + E_U + E_I

Similar to EJT, excess labour turnover (ELT) is the difference between total LT and the absolute value of NEG. It gives the number of job
matches created or destroyed in excess of what is needed to accommodate net employment growth.

(9) ELT = LT – ⎪NEG⎪

Indicators described above are normally presented as rates. The base is average employment in periods t–1 and t (Davis and Haltiwanger, 1996).

What is the relationship between labour and job turnover?

As stated in the main text, JT comprises all changes in the level and spatial distribution of employment deriving from firms’ labour input deci-
sions. In contrast, LT measures flows from the perspective of workers, thereby including both workers’ flows which are initiated by firms (i.e.
creation and destruction of jobs as well as dismissals followed by worker’s replacement in continuing jobs) and those resulting from workers’
decisions to move to different jobs, or in and out of employment. Hence by definition, JT is smaller than LT, because to each job created
(destroyed) there corresponds one hire (separation), whereas the opposite does not necessarily hold, as there are many separations and hires
which are not associated with changes in the existing stock of jobs. In conclusion, JT is equivalent to the number of workers changing job or
employment status as a result of firms’ decisions to change the level and distribution of employment opportunities within the economy.

The Churning Flow (CF, Dale-Olsen, 2006) is defined as the difference between LT and JT, representing the share of labour flows that do not
occur to accommodate firms’ job destruction/creation decisions or, in other words, the amount of workers’ reallocations which would take
place even in the absence of any change in the distribution of jobs across firms.

(10) CF = LT – JT

(10) LT and JT indicators can be calculated using different time units, such as a month, a quarter or the year. Indicators calculated using different
frequencies are not directly comparable as both jobs and workers’ flows can be reversed within a given period (i.e. between observations).
A worker may, for instance, go through a repeated sequence of short jobs, with unemployment spells in between, within a year. Such intra-
annual flows are not accounted for when using annual data. This implies that annual turnover rates systematically under-estimate the actual
number of job/labour flows/transitions. LT figures presented in this chapter are mainly based on EU LFS annual data, because quarterly EU
LFS data do not permit the calculation of LT indicators as the retrospective question on labour market status during the previous period is
not collected quarterly. This chapter includes also some preliminary calculations of LT indicators using EU SILC (and of the factors determining
some transitions) using the calendar information of labour market status.

(11) This is calculated by comparing the self-defined employment status in the current year (variable MAINSTAT in the EU LFS) with the self-defined
status in the previous year (WSTATY1). MAINSTAT is preferred to current employment status following the ILO definition (i.e. ILOSTAT) in order to
maximise comparability with WSTATY1. The same methodology is applied for EU LFS-based indicators of labour market transitions (see section 3).

(12) This excludes workers changing jobs while remaining with the same employer, as this information is not available from the EU LFS. However,
this is not a major omission, because the focus of the chapter (as in the turnover literature) is on external, rather than within-firm, employ-
ment flexibility.

Employment in Europe 2009

There is extensive empirical litera- of nearly 8% of employment in • The datasets used may differ in
ture, containing calculations regard- the USA private sector, with more the quality of their longitudinal
ing these indicators for developed than 8% of the US working-age links – i.e. the extent to which they
economies, together with analyses population changing job or their make appropriate adjustments for
of the main determinants of job and ­employment status every month. changes or breaks in the data series
labour flows. due to definitional changes. This
As regards European countries, leads to varying degrees of reliabil-
There is a considerable variation empirical analyses have also found ity in the corresponding measure-
between the detailed experiences large-scale gross job and labour flows ment of flows.
of different countries. However, a (Davis and Haltiwanger, 1999; Dale-
common characteristic is that gross Olsen, 2006). These results concern Recent work from the OECD (OECD,
flows, or movements, of employment not only countries such as the UK, 2009) has sought to overcome these
regularly and significantly exceed which is often considered to have a comparability issues by using new,
net flows. Moreover, high gross job particularly ‘flexible’ (external) labour internationally harmonised, data
and labour flows are present at all market, but also Germany, France or sources. The results of these analy-
phases of the economic cycle, which Italy. Abowd et al. (1999) estimated ses suggest that the size of both
is indicative of their structural – as that, in France, over one year, the job and labour turnover rates vary
opposed to cyclical – nature (Davis et creation of one job corresponds to quite substantially across countries.
al., 2006). In other words, this litera- the hiring of three persons and the Job turnover rates range from 25%
ture on labour market flows empha- separation of two, while Burda and or more in the USA and the UK to
sises the ‘fluid’ nature of labour Wiplosz (1994) report similar figures less than 15% in Germany, Slovenia
markets, going far beyond what for Germany, Spain and the UK. (14) and Sweden, whereas labour turno-
might otherwise be inferred by sim- ver rates vary from more than 40%
ply looking at the net results – i.e. However, comparisons of the results in the USA, the UK, Denmark and
changes in the level of ­employment of different cross-country studies of Spain to less than 30% in Hungary,
or ­unemployment. gross job and labour flows statis- Italy, Austria and Greece. Overall,
tics need to be made with care, labour and job turnover rates are
In a study of 11 Organisation for given possible differences in the data correlated across countries.
Economic Co-operation and Devel- sources used and the methodologies
opment (OECD) countries, the OECD applied to them. Davis et al. (2006)
Secretariat (OECD, 2009) found aver- identify three main problems affect- 2.3. The impact
age annual job turnover rates of ing international comparisons: of cycle, firms’ and
some 22% (of total employment)
over the period 1997–2004, and • Flows occurring between two peri- workers’ characteristics
annual average labour turnover ods may be measured using data
rates of 33% (of total employment) with a different frequency, e.g. Extensive empirical research has
between 2000 and 2005. Likewise, annual, quarterly, or monthly. also been undertaken regarding the
Haltiwanger et al. (2006) reported Annual data captures a smaller impact of a number of variables and
an average job turnover rate of 25% fraction of transitory employment factors on the magnitude and ‘cycli-
for a group of OECD countries dur- changes (i.e. missing those that cal profiles’ of job and labour flows
ing the 1990s(13), and Davis and Halti- occur but are reversed within the indicators.
wanger (1995) calculated that, in the time period considered) compared
USA, the average number of workers with quarterly or monthly data. As As regards the effects of the business
changing job or employment status a result, annual flow rates tend to cycle, Davis and Haltiwanger (1995),
represented more than a third of underestimate the actual amount Gomez-Salvador et al. (2004) and
total employment (36.8%), whereas of turnover. Dale-Olsen (2006) present results that
annual job turnover rates ranged suggest a complex and sometimes
from 20% to 30% of total employ- • The unit of analysis can vary, as counterintuitive picture, in which job
ment. The use of quarterly data flows may have been measured, turnover and labour turnover behave
has confirmed and strengthened the for other statistical purposes, at somewhat differently. Job turnover
results obtained using annual data. the level of an establishment, firm tends to exhibit a counter-cyclical
For instance using quarterly data or notional tax-paying entity. pattern in the USA and the UK,
Davis et al. (2006) found average and an acyclical pattern in Continen-
job creation and destruction rates tal Europe. As expected, however,
(14) Burda and Wiplosz (1994) argue that:
job destruction is counter-cyclical –
(13) OECD (2009) and Haltiwanger et al. (2006), “Even when compared with the US and
unlike previous studies, are based on inter- Japan, labour markets in Europe are far i.e. job losses increase in economic
nationally harmonised data sources, there- from stagnant. They are characterised downturns, while job creation is less
by providing more reliable cross-country by large flows between employment,
sensitive to the economic cycle.
estimates of job and labour flows. unemployment and nonparticipation.”

Chapter 2 Labour flows, transitions and unemployment duration

Labour turnover is generally procyclical, organisational and demand factors. The turnover variance than the sector, with
in that hirings are particularly procycli- OECD (2009) also estimates that most job turnover tending to decline with
cal, although separations appear much of the total job and labour reallocation firm size, although more so in the USA
less sensitive to cyclical conditions. The that takes place occurs within sectors, than in EU Member States.
latter result reflects the outcome of rather than between sectors.(15)
two contrasting forces: dismissals tend Notwithstanding the importance of
to increase during downturns, while Using a number of US data sources, a firm’s characteristics, such as its size
voluntary quits tend to increase in Davis et al. (2006) found substantial and economic sector, both the OECD
more buoyant economic conditions. cross-sector variations in both job and (2009) and Haltiwanger et al.’s (2006)
labour flows in line with the results analyses leave a significant amount
Against this overall background, pat- of the OECD (2009) analysis – i.e. with of the total variability in turnover
terns of job and labour flows are seen construction and leisure/hospitality indicators unexplained, implying that
to vary considerably between firms registering much larger job and labour country-specific characteristics, such as
(in terms of their sector, size and flows than manufacturing. Moreover, national institutions and ­regulations,
age) and between workers (in terms they underlined the fact that the share could play a significant role.
of their gender, age and ­education). of dismissals relative to quits in total
separations also varies across sectors, In particular, OECD (2009) suggests that
Overall, job flows appear to be larger with construction and manufactur- countries with less stringent employ-
in younger and smaller firms, as well ing having a larger proportion of dis- ment protection legislation (EPL), such
as in the service sector, compared missals compared with the retail and as the USA, the UK and Denmark,
with manufacturing (Gomez-Salvador ­hospitality-leisure sectors. or countries with a particularly large
et al., 2004). Labour flows tend to share of temporary employment (e.g.
decrease with the age of the workers As regards the impact of firms’ char- Spain), tend to be characterised by
concerned (Haltiwanger and Vodop- acteristics (apart from sectoral com- higher job and labour turnover rates.(16)
icec, 2003), while the turnover of position effects), the OECD (2009) Haltiwanger et al. (2006) ran regres-
skilled workers tends to be higher results build on previous research by sions on the effects of EPL on turnover
than that of unskilled workers, main- also identifying significant effects rates which suggested that, after tak-
ly due to more frequent job-to-job resulting from the age and size of ing account of sector and firm size
moves by skilled workers (Dale-Olsen, the firms. The analysis concludes that effects, stringent regulations reduce
2006). Furthermore, levels of educa- a firm’s age plays a much larger role job turnover in general, particularly
tional attainment affect the ‘quality’ in explaining job flow variations than in sectors requiring more considerable
of workers’ moves by significantly its size, with younger firms tending labour adjustments. Within sectors,
boosting the chances for ‘upward’ to create more jobs. And, while older turnover in large firms is reduced to
mobility in the labour market – i.e. firms do tend to destroy more jobs a greater extent than in small firms,
mobility to more stable or better paid than younger ones, this relationship possibly because small businesses are
jobs (Muffels and Fouarge, 2006). between the age of the firm and job often exempt from EU or national leg-
destruction is generally weaker, and islation on dismissals. Garibaldi et al.
OECD (2009) carried out a compre- only valid for some countries, such as (1997) also find a negative correlation
hensive analysis of these issues using a the USA, the UK and Denmark. between an index of dismissal costs for
common methodology and a harmo- firms (encompassing rules on notice
nised dataset. Results show that job As with OECD (2009), Haltiwanger et and severance payments) and job real-
flows vary significantly across sectors, al. (2006) analysed job flows in 16 location for a sample of 10 OECD
and confirm previous findings that they developed and developing economies countries in the period 1982–89. The
are higher in construction and services based on a harmonised dataset and correlation is particularly strong when
(with a few exceptions, such as finan- found that the combined effect of a the sample is restricted to ­continuing
cial intermediation) than in manufac- firm size and its sector explain more firms, i.e. excluding entry and exit
turing. Cross-sector variations in rates than half of the overall variability in firms. As continuing firms tend to be
of labour turnover are even greater, job turnover between OECD countries. larger on average, this finding sug-
ranging from a high of 62% in hotels, In addition, this sectoral classification gests, again, that smaller firms can
restaurants, etc. (horeca) to 14% in is highly correlated across countries. more easily circumvent employment
electricity, gas and water supplies. However, their findings do differ protection regulations.
from those of OECD (1999) insofar as
Despite differences in national labour they argue that the firm’s size alone (16) This is consistent with recent findings
market institutional arrangements, var- accounts for a much larger share of job in Kugler and Pica (2003) and Gomez-
Salvador et al. (2004), which have used
iations in job and labour flows across
(15) On average within-industry job and harmonised cross-country datasets to
economic sectors tends to be similar labour turnover rates are equal to 18% reassess previous research by Bertola
across countries, ­suggesting the impor- and 30% of employment, respectively, and Rogerson (1997) and OECD (1999),
against overall job and labour turnover which was inconclusive, on the impact of
tance of sector-specific technological,
rates of 22% and 33%, respectively. EPL on job flows.

Employment in Europe 2009

Overall, this literature suggests that With respect to the relationship in terms of the total number of ‘sepa-
regulations that raise labour adjust- between job and labour flows, job rations’, but also in terms of the high-
ment costs reduce both job creation reallocation appears to account er share of dismissals over quits. Given
and destruction rates, thereby reduc- for a large proportion of labour the evidence that dismissed workers
ing turnover. This could adversely flows – e.g. between one third and are more likely to experience longer
affect the allocative efficiency of two thirds in the USA, according to unemployment spells than quitters
labour markets, at least in the short Davis and Haltiwanger (1995), and (see section 4.4), the high proportion
run, and potentially lower long-run two thirds on average across the 11 of dismissals in sharp downturns sig-
productivity growth rates (see 2.3.2). countries considered in OECD (2009). nals a medium-term increase in unem-
Nonetheless, these are ultimately This suggests that, from the perspec- ployment duration. Moreover, given
determined by a wide range of eco- tive of workers, labour dynamics are that labour flows often seem to be
nomic, technological and organisa- often involuntary, and result from associated with large-scale, spatially
tional factors, which are not taken a process of reallocation of employ- concentrated job reallocations, the
into account in these analyses. ment opportunities initiated by the potentially adverse consequences for
firms themselves, as a reaction to workers and local communities tend
As regards the effects of worker their market conditions, perceived to be more long-lasting.
characteristics, the OECD (2009) anal- opportunities or the initiatives of
ysis confirms that they significantly other competing firms.
affect potential labour mobility. In 2.3.2. The link between
practice, labour turnover appears to With respect to ‘spatial concentra- turnover and productivity
be higher for women than for men tion’ and ‘time persistence’ of job
in most countries, particularly with flows, a considerable share of overall As mentioned in section 2.1, labour
respect to hirings. A strong negative job creation and destruction appears market theory underlines the contri-
correlation is found between age to be the result of major contrac- bution of job and labour reallocation
and hiring rates, reflecting factors tions or expansions of employment in to the overall level of allocative effi-
such the entry of young people into a limited number of establishments ciency in an economy, by ensuring
the labour market and the accumula- (Davis and Haltiwanger, 1996; Davis that both jobs and workers move to
tion of firm-specific human capital. et al., 2006), rather than being spread those firms/sectors where they are
However, no significant relationship evenly across a large number of estab- most productive.
is observed between age and separa- lishments. In the USA, in the third
tion rates, except in a few countries quarter of 2001, Davis et al. (2006) The empirical literature on the impact
like Denmark, Finland, France and found that 68% of total job destruc- of job reallocation on total factor
Germany, where it is actually nega- tion occurred in establishments reduc- productivity (TFP) is rather small.
tive. Finally, separation rates tend ing their employment by at least 10%, However, a number of papers (Foster
to be higher for low-skilled workers, and 63% of job creation took place et al., 2001; Bartelsman et al., 2004;
while hiring rates vary to a lesser in establishments expanding their Brown and Earle, 2008) rely on firm-
extent by skill level. Labour turno- employment by 10% or more.(17) Fur- level datasets (e.g. business registers)
ver also appears to be larger at the thermore, the decision to create or available for several countries (both
opposite ends of the skill distribu- to destroy certain jobs at a particular developed and emerging economies)
tion, except in countries such as the point in time seemed to reflect non- to investigate the microeconomic
USA and Denmark, with high over- reversible – i.e. permanent – decisions determinants of aggregate produc-
all mobility, where the relationship at the level of the establishment. tivity performance at country or
between labour reallocation and skill industry level. Accordingly, changes
level is negative. Davis et al. (2006) found that the in aggregate TFP can be decomposed
relationship between the share of into a within-firm component, linked
separations due to dismissals, and to efficiency gains in the production
2.3.1. Other features changes in the level of employment at process, and a component deriving
of labour turnover: the establishment level, was not lin- from the reallocation of production
concentration, persistence, ear. Specifically, establishments tend factors (including labour) between
role of dismissals to accommodate moderate declines firms. The latter can involve the entry
in employment through quits, while of new firms and the exit of existing
Interesting findings have also been relying more on dismissals in the case ones or the employment expansion
highlighted in recent research with of large employment contractions. and contraction of continuing firms.
regards to the relationship between Overall, this implies that severe macr-
job and labour flows, on the one oeconomic recessions (like the current A consistent finding is that, while
hand, and the degree of ‘spatial con- one) differ from milder ones not only efficiency gains within firms often
centration’ and ‘time persistence’ of account for most changes in aggre-
(17) The degree of concentration appears to
job flows, on the other hand. gate productivity, a significant pro-
be lower for labour than for job flows.

Chapter 2 Labour flows, transitions and unemployment duration

portion of the latter comes from Table 1: Labour flow indicators by country (%)
job reallocation from contracting to
expanding firms, with a significant H S LT NEG ELT
DK 13.0 16.9 29.9 0.5 28.9
role being played by entering and
ES 16.2 13.2 29.4 4.0 25.4
exiting firms. This evidence signals
FI 13.5 15.0 28.6 0.5 27.8
that job reallocation tends to be
UK 14.4 11.1 25.6 0.8 24.8
associated with substantial ‘creative
EE 12.4 10.9 23.3 2.1 21.2
destruction’ forces, whereby new FR 11.8 11.2 23.0 0.8 22.2
and more productive firms replace EU-8 12.6 10.0 22.6 1.4 21.3
less productive ones and jobs move PL 12.1 10.3 22.4 1.3 19.9
from less efficient to more efficient DE 12.5 8.9 21.4 0.7 19.7
firms and sectors. HU 10.0 9.7 19.8 0.2 19.2
PT 9.2 10.0 19.3 0.2 18.8
The economic literature highlights CZ 8.7 10.4 19.1 0.8 17.9
the variability in the contribution RO 10.8 7.8 18.6 -1.4 16.3

of job reallocation to TFP growth BE 8.9 9.6 18.5 1.3 17.1

IT 10.3 7.5 17.9 1.4 16.5
(Bartelsman et al., 2004). However,
SE 7.8 7.9 15.7 -0.2 15.3
several papers have found that pro-
GR 6.8 7.1 13.9 1.6 12.3
ductivity-enhancing reallocation has
been particularly important in the Source: DG EMPL calculations using EU LFS.
transition economies of Central and Notes: H = Hirings, S = Separations, LT = Labour Turnover, NEG = Net Employment
Eastern Europe (Brown and Earle, Growth, ELT = Excess Labour Turnover. All figures are expressed as percentages of
2008; De Loecker and Konings, 2005; employment in the previous year (see Box 2 for the methodology). Averages for the
Masso et al., 2004). Initially, these period 2002–07, except for France (2002) and Sweden (2002–04).
economies were characterised by
both low job reallocation and TFP Chart 1b: Hirings, separations and labour turnover (% of employment)
growth rates. However, following 30
the opening up of markets during the
1990s, state-owned firms registered S
25 LT
large-scale job destruction, whereas
net entry of new private firms rose
significantly, accounting for a large 20

share of total job creation. Both

developments were associated with 15
substantial productivity gains, illus-
trating the critical role played by job
reallocation in successful transitions
towards a ­market economy.(18)

2.4. The size and 0

characteristics of labour DK ES FI UK EE FR EU-8 PL DE HU PT CZ RO BE IT SE GR

market flows in the EU Source: DG EMPL calculations using EU LFS.

Notes: Annual averages across 2002–07 period. 2002–04 for Sweden and 2002 only
for France.
Using the methodology presented
in Box 2, this section undertakes a labour turnover(19) in Europe based on Table 1 presents indicators for hiring,
review of the empirical evidence on the EU LFS and the EU SILC. Indicators separation and labour turnover rates,
are calculated for the EU as a whole as well as, for NEG and excess labour
(18) According to De Loecker and Konings and for a number of categories. Time turnover. These calculations confirm,
(2006), the net entry of firms accounted series for the indicators based on the for the EU, the three key findings of
for 17% of growth in aggregate pro-
EU LFS are presented, and an attempt the literature reviewed in section 2.3:
ductivity of the manufacturing sector in
Slovenia between 1994 and 2000. Using a is made to distinguish between cycli-
dataset covering six transition economies cal variations and long-term trends. • EU labour markets are very
in Central and Eastern Europe, Brown and
­dynamic. In the period 2002–07, the
Earle (2008) find that after market reforms (19) At EU level, job turnover indicators could
the contribution of job reallocation to not be calculated using harmonised labour turnover rate for the EU-8(20)
productivity growth reached much higher business surveys or matched employer-
levels than those reported for developed employee data sources, because data is (20) Belgium, Germany, Denmark, Greece,
economies, such as the USA or the UK. not available. Spain, Italy, Portugal and the UK.

Employment in Europe 2009

­ veraged 22.8%, ranging from a

a Table 2: Correlation between the hiring and separation rates
maximum close to 30% in Denmark
to a minimum of 14% in Greece. Pearson correlation coefficient 0.638 ***

*** significant at 1% level

• In all countries, labour turnover
Source: DG EMPL calculations using EU LFS.
rates exceed NEG by a significant
margin. In the EU-8, ‘excess’(21)
labour turnover averaged around Box 3: Labour turnover indicators based on the EU SILC – the impact
21% of initial employment. of the periodicity of observations on measurement

• Labour turnover indicators vary The users’ database of the EU SILC provides an opportunity to measure the effect of using
greatly across the EU (Chart 1b). data with different frequency (for example, annual versus monthly) for calculating labour
Denmark, Spain, Finland and the turnover indicators, in a framework that in principle controls for other sources of variation.
UK have the highest values, while The longitudinal component of the EU SILC includes both an annual variable on (the self-
Greece, Sweden and Italy have declared current) economic status (PL030) and a calendar of activity variable (PL0210A-L),
the lowest. reporting the monthly economic status.

As mentioned earlier in the text, there is a strong presumption that annual data captures
Across the panel of data, a significant
only part of all labour market transitions, ignoring those that occur between the observa-
positive correlation exists between tions dates, and which are reversed in the meantime. As a result, flow rates based on annual
hiring and separation rates (Table 2). data tend to systematically underestimate the actual values.
This is in line with findings in OECD
(2009), where hiring and separations Using longitudinal data for 2005–06, covering 24 European countries(1), hiring, separation
are highly correlated across indus- and turnover rates are calculated based on economic status variables having either an
annual or monthly frequency.
tries and countries(22), as they are
with respect to job creation and The results are as expected. Higher frequency observations ‘count’ more transitions (i.e.
destruction rates. both hirings and separations), although the results should be viewed as preliminary, taken
with caution, and may be subject to revision. On average, moving from annual observations
to a monthly calendar virtually ‘doubles’ the calculated values of the three labour turnover
indicators (Charts 2 to 4). (2)

(1) Austria, Belgium, the Czech Republic, Germany, Denmark, Estonia, Spain, Finland, France,
Greece, Hungary, Ireland, Italy, Lithuania, Luxembourg, Latvia, Norway, Poland, Portugal,
Sweden, Slovenia, Slovakia and the UK.

(2) The major exception to this general rule seems to be Germany, particularly for the hiring
rate. At present, the reason for this is unclear. In the cases of Sweden, Italy and Finland,
the ratio between the two labour turnover measures averages about 3.

(21) The number of workers who have

changed job or employment status in
excess of what is needed to accommo-
date net employment growth.

(22) After controlling for sectoral composition.

Chapter 2 Labour flows, transitions and unemployment duration

Chart 2: Hiring rates based on the longitudinal component of the EU SILC calculated using either annual
or monthly calendar variables for the economic status, 2005–06 (% of employment)








Month 0.310 0.292 0.245 0.201 0.197 0.182 0.176 0.176 0.174 0.173 0.168 0.159 0.151 0.151 0.150 0.146 0.141 0.140 0.136 0.129 0.115 0.112 0.103 0.046
Year 0.107 0.091 0.142 0.102 0.141 0.080 0.095 0.086 0.080 0.087 0.321 0.034 0.071 0.092 0.074 0.092 0.084 0.072 0.068 0.083 0.089 0.065 0.061 0.048

Source: DG EMPL calculations using the user’s database of EU SILC.

Note: Ranked in descending order of the indicator based on monthly observations.

Chart 3: Separation rates based on the longitudinal component of the EU SILC calculated using either annual or monthly
calendar variables for the economic status, 2005–06 (% of employment)







Month 0.298 0.264 0.200 0.195 0.186 0.175 0.172 0.167 0.155 0.153 0.150 0.147 0.142 0.141 0.125 0.12 0.119 0.117 0.115 0.114 0.106 0.099 0.098 0.054
Year 0.107 0.061 0.090 0.077 0.077 0.134 0.059 0.07 0.069 0.098 0.066 0.081 0.064 0.050 0.061 0.074 0.045 0.024 0.060 0.054 0.065 0.063 0.052 0.050

Source: DG EMPL calculations using the user’s database of EU SILC.

Note: Ranked in descending order of the indicator based on monthly observations.

Chart 4: Labour turnover rates based on the longitudinal component of the EU SILC calculated using either annual
or monthly calendar variables for the economic status, 2005–06 (% of employment)





Month 0.609 0.556 0.431 0.401 0.355 0.350 0.350 0.349 0.348 0.323 0.315 0.305 0.298 0.286 0.282 0.271 0.260 0.255 0.248 0.237 0.234 0.222 0.201 0.100
Year 0.214 0.152 0.219 0.192 0.111 0.229 0.240 0.149 0.145 0.153 0.130 0.143 0.152 0.345 0.135 0.166 0.146 0.144 0.127 0.126 0.132 0.155 0.113 0.099

Source: DG EMPL calculations using the user’s database of EU SILC.

Note: Ranked in descending order of the indicator based on monthly observations.

Employment in Europe 2009

Chart 5: Hirings, separations and labour turnover, change over time (%) • In Denmark, labour turnover
increased as a result of a rising
Hirings (%), change over time Average (1992–96) Average (1997–2001) Average (2002–07)
0.25 separation rate, being only partly
offset by a decreasing hiring rate.

0.15 In the period 1992–2007(23) and for the

EU-8(24), Chart 6 plots time series for
various labour turnover indicators. It
suggests the following comments:
• There is no evidence of significant
Separations (%), change over time
0.20 time trends in the period.

• Some indicators behave procycli-
cally, particularly the hiring and
labour turnover rates, as both reg-
0.05 ister increases in the late 1990s and
reductions in the early 2000s, fol-
DK FI ES UK EE CZ PL PT EU-8 HU BE DE SE IT GR FR lowed by a slight increase between
Labour turnover (%), change over time 2004 and 2007. The separation rate
shows a much more flat trajectory,
suggesting a lower sensitivity to
the economic cycle.
• Except during the 1992–93 eco-
0.10 nomic recession, the EU-8 hiring
rate has always been significantly
DK ES FI UK EE EU-8 PL DE HU PT CZ BE IT SE GR FR higher than the separation rate.
Source: DG EMPL calculations using EU LFS.
• Labour turnover regularly exceeds
NEG by a large margin.
Chart 6: Labour flows, change over time, EU-8 (%)

2.4.2. Breakdowns by
gender, age and education
0.15 The following three charts present
figures on hiring, separation and
0.10 labour turnover rates by gender, age,
and education level for those EU
Member States for which data is

Chart 7 shows that labour turnover
rates tend to be higher for women
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
than men, and that this is also the
case for hiring and separation rates if
taken separately. However, the extent
Source: DG EMPL calculations using EU LFS. of this gender gap varies considerably
across countries, being particularly
2.4.1. Time trends • Overall, labour turnover appears to large in Spain, the Czech Republic,
have peaked in the second period Greece, and Finland and very small in
Using EU LFS data, Chart 5 plots three (1997–2001).
period averages for labour turnover (23) Using the EU LFS, labour flows can only
indicators (1992–96, 1997–2001 and • In Spain, turnover fell with both hiring be calculated since 1992 for the EU-8
2002–07). The most notable develop- and separation rates having decreased
(24) Belgium, Germany, Denmark, Greece,
ments over these periods are as follows: substantially since the early 1990s.
Spain, Italy, Portugal and the UK.

Chapter 2 Labour flows, transitions and unemployment duration

Chart 7: Hirings, separations and labour turnover by gender (%)

Hirings Men Women










Labour turnover





Source: DG EMPL calculations using EU LFS.

Notes: Figures are six-year averages for the 2002–07 period, except for Sweden (2002–04) and France (2002).

Estonia, Poland, Romania and Swe- Chart 8: Hirings, separations and labour turnover by age (%)
den. Only in Estonia and Poland are Hirings (%) Old Prime Young
the hiring rates for men marginally 0.6

greater than they are for women.

Chart 8 provides evidence of the
large gap in labour turnover rates
that exists between younger workers
­(16–24 years old) on the one hand,
and prime-age workers (25–54 years FI ES PL FR EE UK EU-8 BE DE SE IT HU CZ DK PT GR RO

old) and older workers (55–64 years Separations (%)

old) on the other.(25) The turnover rate
for young workers is very high, above
60% in all but five Member States. 0.4

The high rates of mobility of younger

workers is due to many factors, such as 0.2
first entries into work, the widespread
use of temporary contracts, the way
firms screen newly recruited work-
ers, leading to frequent terminations Labour turnover (%)
of employment, as well as voluntary
quits by young people themselves in
search of better or more suitable jobs.
In the third part of this chapter, deal-
ing with unemployment duration, it 0.4

will be shown that the high turnover

rates of young workers are associated 0.0
with high unemployment rates, but
low unemployment duration. Source: DG EMPL calculations using EU LFS.
Notes: Figures are six-year averages for the 2002–07 period, except for Sweden
(2002–04) and France (2002).
(25) Excluding Romania.

Employment in Europe 2009

Chart 9: Hirings, separations and labour turnover by education level (%) Chart 9 suggests that labour turnover
tends to decrease with the level of
Low Medium High education, although there is a sig-
nificant and striking variation in this
relationship between countries. The
turnover rate of medium/high-skilled
workers is much lower than for low-
skilled workers in Germany, Estonia,
Denmark and Hungary. It is more
similar in Spain, France and Italy, and
the position is reversed in the UK,
0.30 Portugal and Greece, where labour
turnover of the more highly skilled is
0.20 greater than for the less skilled.

Charts 10 to 12 present data for the

EU-8 from 1992 to 2007 for hiring
and separation rates, broken down
DK FI CZ EE HU ES PL FR BE EU-8 DE UK PT RO SE IT GR by gender, age and skill levels.
Labour turnover
• The breakdown by gender shows
0.4 quite similar time profiles for both
hiring and separation rates – i.e.
the gender gap in both indicators
0.2 stays fairly constant.
• The breakdown by age illustrates
DK EE DE FI ES HU CZ PL UK RO EU-8 FR BE SE IT PT GR a significant increase in hiring
rates for younger workers, which
Source: DG EMPL calculations using EU LFS.
contrasts with the relative sta-
Notes: Figures are six-year averages for the 2002–07 period, except for Sweden
(2002–04) and France (2002).
bility for prime-age and older

Chart 10: Hirings and separations, change over time, by gender, EU-8 (%)

Hirings Separations Men Women

0.20 0.15

0.15 0.12

0.10 0.09

0.05 0.06
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Source: EU LFS, DG EMPL calculations.

Chart 11: Hirings and separations, change over time, by age, EU-8 (%)

Hirings Separations Young Prime Old

0.6 0.25




0.0 0.05
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Source: EU LFS, DG EMPL calculations

Chapter 2 Labour flows, transitions and unemployment duration

Chart 12: Hirings and separations, change over time, by skill level, EU-8 (%)

Hirings Separations Low Medium High

0.18 0.18

0.15 0.15

0.12 0.12

0.09 0.09

0.06 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 0.06 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Source: EU LFS, DG EMPL calculations.

workers. As regards separation • The breakdown by education level • For separation rates, the time series
rates, younger workers show suggests that the hiring rates of low- evidence does not suggest that
some increase overall, albeit skilled(26) workers fluctuate more over the business cycle has a dispropor-
much lower than for hiring rates, the economic cycle, while the varia- tionate effect on more vulnerable
while a substantial reduction has bility of separation rates appear to be groups, such as women, young
occurred for older workers. fairly similar across education levels. people or ­low-skilled workers.

Box 4: Analysis of variance of hiring rates using EU LFS data 2.4.3. Institutions and labour
The evidence presented so far focuses on the variation of labour flow indicators by major
labour force characteristics. As previously mentioned, the analytical literature also suggests Although sectoral composition plays
that labour flows vary significantly across economic sectors, and that such patterns are a major role in explaining the over-
quite stable across countries. In order to illustrate this point, an analysis of variance is car- all variability in labour flows, country
ried out for hiring rates only, because separation rates cannot be calculated by sector.(1) The effects still account for a significant part
total variance of hiring rates is broken down using three variables: time (2000–07), country
of total variability, which is seemingly
(25 Member States(2)), and sector (16 sectors, Table 3).(3)
consistent with research findings point-
These three explanatory variables account for 43% of the total variation in hiring rates. This ing to the important role of country-
result confirms for the EU, one of the key conclusions drawn in OECD (2009) – namely that specific institutional arrangements.
hiring rates vary substantially across sector, and that this factor tends to explain a larger
fraction of overall variability compared with country variability, respectively, almost one Chart 13 provides a graphical illus-
third against about 10%. tration of the partial correlations of
EPL and the incidence of temporary
Table 3: ANOVA (analysis of variance) of hiring rates in the EU employment (TE) on labour turnover
rates. The left panel suggests that
Dependent variable: hiring rates
Explanatory variables % overall variance Significance (F-test) R squared
countries with more stringent EPL(27)
Sector (NACE 2D) 0.31 *** tend to have lower turnover rates,
Country 0.10 *** while the right panel points to a posi-
Year 0.01 *** 0.428 tive relationship between the relative
*** coefficient significant at 1% level importance of temporary employment
contracts and labour turnover rates.
Source: DG EMPL calculations using EU LFS.
Table 4 presents a cross-country
Given that hiring and separations rates tend to be highly correlated (Table 2), it is reasonable
regression of labour turnover rates
to hypothesise that the general pattern of variation in labour turnover rates closely follows
that of hiring rates. on EPL and the share of temporary
employment, covering 15 EU Mem-
(1) The LFS includes a retrospective question on the sector of work in the previous year only
for those workers employed at the time of the survey, and no such question is asked of ber States for which data is avail-
people being currently unemployed or inactive, which makes it impossible to calculate able.(28) Albeit not controlling for
separation rates by sector.

(2) EU-27 excluding Bulgaria and Ireland. (27) The EPL indicator is calculated by the OECD
(OECD, 2004 and 2009). It ranges from 0 to
(3) NACE at two-digit level. The panel built for this analysis is unbalanced, presenting a 6, increasing with the level of rigidity of
significant number of empty cells, as the time coverage of available data is partial for a legislation on hirings and dismissals.
number of countries.
(28) Belgium, the Czech Republic, Germany,
Denmark, Estonia, Spain, Finland, France,
(26) The terms ‘education’ and ‘skills’ are used Greece, Hungary, Italy, Poland, Portugal,
as synonymous throughout the chapter. Sweden and the UK.

Employment in Europe 2009

Chart 13: Labour turnover, employment protection legislation and temporary employment

LT (%) LT (%)
0.30 0.30
0.26 0.26

0.22 0.22 PL
0.18 0.18 BE
0.14 0.14

0.10 0.10
1 1.5 2 2.5 3 3.5 4 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
EPL TE (%)

Source: DG EMPL calculations using EU LFS and OECD data.

Notes: Averages for 2000–05. EPL: average of 1998 and 2003 figures.

Table 4: OLS of labour turnover rates In this context, a perspective on the

frequency, type and time sequence
Explanatory variables: Coefficient Significance R squared
of any individual’s transitions in the
Constant 0.295
labour market sheds light, not only
EPL -0.724 ** on the extent – but also on the
TE 0.655 ** 0.413 overall quality – of mobility, includ-
** coefficient significant at 5% level ing whether ‘good’ moves, aimed
at improving labour market attach-
Source: DG EMPL calculations using EU LFS and OECD data.
ment/career progress, prevail over
Notes: Dependent variable: labour turnover rate. Explanatory variables: EPL = OECD ‘bad’ moves, namely those involving
index of strictness of EPL. TE in percentage of dependent employment. Figures for
turnover and temporary employment are five-year averages from 2000 to 2005. worsening working conditions. This
Figures for EPL are averages of 1998 and 2003. The cross-country regression includes is illustrated, for instance, by the
15 Member States. work of Muffels and Fouarge (2008)
who, based on European Community
cross-­country differences in sectoral permanent contracts or across jobs Households Panel (ECHP)(29) data, dis-
composition and workforce charac- with different pay levels) – as an tinguish between upward and down-
teristics, results confirm the initial alternative to measuring the overall ward transitions, calculating ‘dynam-
impression from the scatter plots. amount of mobility in the system. ic employment security’ indicators to
The regression coefficients are sig- measure the prevailing direction of
nificant and with the expected signs: The underlying assumption, accord- total labour market transitions.
negative for EPL, and positive for TE. ing to the Transitional Labour Mar-
kets theory (Schmidt and Gazier, The Commission services have reviewed
2002), is that the traditional pattern the evidence on labour market transi-

3. Labour market of holding the same job throughout

one’s working life is increasingly giv-
tions in the past. Based on the ECHP,
the Directorate General for Employ-
transitions in the EU ing way to more diversified work ment and Social Affairs (DG EMPL) has
histories, which are characterised monitored transitions by activity sta-
by more frequent moves between tus, type of employment contract and
3.1. From flows to employment, non-employment and pay level for a number of EU Member
transitions private and family responsibilities, as States (EiE 2004, chapter 4), and the
well as by intermediate states with Directorate-General for Economic and
shorter working hours (e.g. part- Financial Affairs (DG ECFIN) has also
Alternative approaches have been time) or phased transitions from a presented evidence on transition rates
developed to monitor labour market state to another (e.g. phased retire- by main employment statuses (DG
dynamics in which the transition of ment). These developments are con- ECFIN, 2008). Such indicators point to
individuals are followed – across dif- sidered to be natural consequences a structural improvement in EU labour
ferent types of employment status of, on the one hand, the need for market outcomes since the late 1990s,
(employed, unemployed or inactive); firms to respond quickly to changes mainly due to the increased propensity
between work, education/training, in their economic environment and,
private responsibilities and retire- on the other, the more varied needs (29) Use of longer longitudinal datasets ena-
ment; and across different types of and preferences of workers and their bles researchers to assess employment
dynamics and individual mobility choices
employment (e.g. from fixed-term to families and dependants.
within a lifecycle perspective.

Chapter 2 Labour flows, transitions and unemployment duration

of women to leave inactivity and seek Table 5: The transition matrix by activity status, EU-8, 2006–07 (%)
paid employment. They also confirm
that worker characteristics, such as age   State in 2007
State in 2006 E U I
and skills, have a significant impact on
E 94.7 2.5 2.8
transition probabilities – i.e. the way
U 32.5 54.2 13.3
different people progress both inside
I 10.5 4.4 85.2
and ­outside the labour force.
Source: DG EMPL calculations using EU LFS.
Notes: E = Employed, U = Unemployed, I = Inactive.
3.2. An empirical Each cell represents the percentage of workers that moved from an initial state (row)
analysis of labour to a final state (column) between two periods/years. Transition rates sum to 1 across
rows. They are calculated using the retrospective question included in the EU LFS on
market transitions the labour market status in the previous year.

in the EU Table 6: Transition rates in the EU, by country and over time (%)
This section presents some empirical   U_E  I_E 
evidence on labour market transi-   2002–07 1996–01 1990-95 change 2002–07 1996–01 1990-95 change
tions at both the aggregate level and UK 47.4 40.7 35.0 12.4 15.9 15.8 15.3 0.6
a number of breakdowns, such as by ES 44.3 37.7 31.6 12.7 8.6 5.6 5.0 3.6
gender, age and education levels, PT 40.7 40.3 38.8 1.9 6.7 6.9 6.1 0.6
based on the EU LFS. The analysis DK 39.3 36.3 37.6 1.7 15.4 20.7 23.2 -7.8
is basically descriptive, including the CZ 37.4 41.0 NA -3.6 6.2 9.9 NA -3.6

calculation of indicators, cross-tabula- EE 36.7 31.4 NA 5.2 8.7 9.3 NA -0.7

EU-8 31.5 31.2 28.8 2.7 9.8 9.4 7.7 2.1
tions and charts. Descriptive analyses
HU 30.7 31.5 NA -0.8 5.5 6.0 NA -0.5
can be misleading, however, because
SE 29.4 35.4 NA -6.0 16.3 18.5 NA -2.2
they do not identify the relative impor-
IT 26.7 28.3 25.2 1.5 4.5 5.2 4.8 -0.4
tance of the different factors affect- FI 26.6 26.7 28.2 -1.6 14.0 13.9 15.0 -1.0
ing labour market transitions. Such DE 25.2 26.1 25.1 0.1 13.1 13.2 8.9 4.2
identification would require the use RO 25.05 NA NA NA 8.04 NA NA NA
of multivariate regression techniques, GR 24.4 23.6 29.2 -4.8 3.0 3.2 3.5 -0.5
such as the estimation of probit mod- PL 23.0 28.5 NA -5.5 5.6 7.3 NA -1.7
els, which lies outside the scope of BE 16.9 17.4 26.1 -9.2 6.3 5.0 3.9 2.4
this chapter. In future, the Commis- FR NA 33.1 33.6 -1.7 NA 8.4 17.6 -9.3
sion services are planning to carry
Source: DG EMPL calculations using EU LFS.
out work, assisted by experts, using
microdata from the users’ EU LFS and Notes: The column ‘Change’ measures the difference between the averages of
2002–07 and 1990–95, or the difference between 2002–07 and 1996–2001 for those
EU SILC databases in order to better countries for which data for the latter period is not available. Only for France, the dif-
assess labour ­market ­transitions. ference between 1996–2001 and 1990–95 is displayed.

One-year(30) indicators are calcu- – i.e. ­transition rates in a given net flows, at the end of the section
lated for transitions between the year are calculated as shares rela- (see 3.2.4), will partly address these
statuses of employment, unemploy- tive to individuals’ labour market ­comparability issues.
ment and inactivity, with calcula- status in the previous year, which
tions presented in a 3-by-3 matrix. implies that they are not compara- In the remainder of this section, the
Table 5 provides an example for ble across different ­initial statuses(31) focus will be on two ‘good’ transi-
the EU-8 aggregate for transitions or with turnover rates calculated tion rates - from unemployment to
between 2006 and 2007. However, in section 2, because the latter use employment (U_E), and from inactiv-
the table also ­highlights a methodo- total employment as the denomi- ity to employment (I_E) – represent-
logical limitation of such ­indicators nator.(32) This caveat needs to be ing a favourable transition for those
kept in mind when looking at the concerned, and a successful mobilisa-
(30) Calculation of transition rates over longer empirical findings of this section and tion of under or unused labour/human
time horizons require longitudinal data
when comparing them with results resources for the economy as a whole.
sources following individuals for several
years. Such sources are currently available obtained for ­turnover. The use of
only to a very limited extent at the EU level Table 6 presents transition rates (U_E
(e.g. the EU SILC). However, the reader (31) As regards Table 5, comparisons can be and I_E) for EU Member States for which
should be aware that long-term transitions made across columns but not across rows.
are indispensable to reach a full understand- data is available and the EU-8 aggre-
ing of lifecycle effects of initial employment (32) The only exceptions are transition rates gate (annual averages for the periods:
conditions and changes in status for individ- from employment as they have the same
1990–95, 1996–01 and 2002–07).
ual earnings and labour market prospects. denominator as turnover.

Employment in Europe 2009

Chart 14: U_E and the economic cycle (%)

EU-8 EU-7 U_E cycle GDP cycle

1.4 1.4
Correlation 0,96 Correlation 0,98
1.0 1.0

0.6 0.6

0.2 0.2
0.0 0.0
-0.2 -0.2

-0.6 -0.6

-1.0 -1.0
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

Source: DG EMPL calculations using EU LFS.

Chart 15: I_E and the economic cycle (%)

EU-8 EU-7 I_E cycle GDP cycle

1.4 1.6
Correlation 0,06 Correlation 0,46
1.0 1.2

0.6 0.8
0.0 0.0
-0.6 -0.8
-1.0 -1.2
-1.4 -1.6
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

Source: DG EMPL calculations using EU LFS.

Table 7: Correlations between cyclical components of transitions and GDP

  U_E_cycle I_E_cycle E_U_cycle E_I_cycle

GDP_cycle 0.307 *** 0.160 *** -0.304 *** 0.026
Pearson correlation coefficients; *** coefficient significant at 1% level

Source: DG EMPL calculations using EU LFS.

These figures show considerable dif- time trends for both U_E and I_E indicator of GDP cyclical volatility for
ferences in country experiences. In transition rates for two EU aggre- two EU aggregates.(34)
the period 2002–07, the UK, Spain, gates. As discussed in the context
Portugal and Denmark registered of labour and job flows in section Chart 14 and Table 7 show a posi-
annual U_E transition rates around 2.3, such transition rates are also tive correlation between the cyclical
or above 40%, while at the lower end affected by the business cycle and, component of the U_E transition and
of the scale, the rates for Belgium, using the Hodrick-Prescott (HP) fil- GDP(35), while Chart 15 and Table 7
Poland, Greece and Germany did not ter, an attempt is made to identify suggest that the I_E transition rate
exceed 25%. The country ranking cyclical versus structural changes in is a leading and procyclical variable;
for I_E movements is quite different, ­transition indicators.(33) this possibly indicates a larger impact
however: in particular, Sweden, Fin-
land and Germany join the UK and Charts 14 and 15 plot the HP cyclical (34) For ease of graphical comparability, indi-
cators in both charts are standardised
Denmark at the upper end of the component of, respectively, the U_E and expressed as moving averages over
scale, with rates at around or above and I_E transition rates, against an three consecutive years. However, values
13%, while Spain and Portugal lose in Table 7 are based on comparisons of
non-standardised variables, which explain
several positions. (33) The HP filter is a mathematical tool, par- differences with correlation coefficients
ticularly used in macroeconomics to obtain displayed in Charts 14 and 15. 
a smoothed non-linear representation of EU-8: Belgium, Germany, Denmark,
a time series, one that is more sensitive to Greece, Spain, Italy, Portugal and the UK.
3.2.1. Trend versus cyclical long-term than to short-term fluctuations. EU-7: Belgium, Germany, Denmark,
The adjustment of the sensitivity of
components of transition rates the trend to short-term fluctuations is
Greece, Spain, Portugal and the UK.

achieved by modifying a parameter λ. For (35) This is in line with discussion in section
After briefly describing cross-country annual data, the parameter λ used in the 2.3 above underlining the overall pro-
HP method was 10. Maraval and del Rio cyclicality of hiring rates, U_E transitions
differences, this section focuses on
(2001) suggest 6<λ<14 for annual data. being an important component thereof.

Chapter 2 Labour flows, transitions and unemployment duration

Chart 16: Trends in transition rates, EU-8 (%)

0.35 0.12




0.25 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 0.06 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Source: DG EMPL calculations using EU LFS.

Chart 17: Trends in transition rates, EU-7 (%)

0.35 0.15




0.25 0.06
1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Source: DG EMPL calculations using EU LFS.

of other ‘structural’ factors, includ-

Chart 17b: Trends in employment to unemployment (E_U)
ing policy changes, on these transi-
transition rates, EU-8 and EU-7 (%)
tions relative to those affecting the
unemployed. Table 7 confirms this 0.040

finding for the two reverse transition

rates – i.e. those from employment- EU-8
to-unemployment (E_U) and from
employment-to-inactivity (I_E) – as
the former is negatively correlated
with GDP, while the latter appears to
be acyclical.

Overall, the empirical evidence points 0.030

to significant business cycle effects for

transitions between unemployment
and employment, while transitions
between inactivity and employment
appear to be much less responsive to 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

the business cycle.

Source: DG EMPL calculations using EU LFS.

Charts 16 and 17 present the HP trend

components of both U_E and I_E rate(36), and appears to have been (E_U) and from employment to inac-
transitions, respectively, for the EU-8 concentrated in the ­second half of tivity (E_I) for both the EU-8 and
and EU-7 aggregates. These charts the 1990s. the EU-7. Scale differences prevent
suggest that EU labour markets have a direct comparison between ’good’
­substantially increased their capacity Although the main focus of this sec- and ’bad’ transitions.
to attract people into employment, tion is on ’good’ transitions, Charts
as reflected in the trend increase 17b and 17c plot the timeseries trends Chart 17b suggests a sizeable reduc-
in transition rates over the period. corresponding to the ’bad’ transitions tion in trend E_U transition rates
However, the improvement is more from employment to unemployment since the early 1990s, although results
pronounced in the I_E transition should be interpreted with some care
(36) For the EU-7.

Employment in Europe 2009

Chart 17c: Trends in employment to inactivity (E_I) findings for the trend component
transition rates, EU-8 and EU-7 (%) of E_I transition rates are somewhat
surprising as the latter is negatively
correlated with structural unemploy-
ment and positively correlated with
employment and participation rates.

3.2.2. Country-by-country
trend developments

Developments at EU level mask con-

siderable diversity across Member
States.(39) Chart 18 shows a large
trend increase in U_E transition rates
in Spain and the UK with, in the
latter case, the proportion of the
1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
unemployed who find a job within
Source: DG EMPL calculations using EU LFS. a year nearly doubling, between
1983 and 2007 – from about 28% to
almost 50%. Conversely, in Italy, U_E
Table 8: Correlations between the HP trend transitions, NAIRU, rates first improved then declined,
employment and participation rates, 1997–2007 while in Greece they have regis-
tered a continuous deterioration
  U_E_trend I_E_trend E_U_trend E_I_trend
over recent decades.
NAIRU -0.311 *** -0.304 *** 0.553 *** -0.162 ***
ER_trend 0.501 *** 0.863 *** -0.112 ** 0.491 ***
PR_trend 0.502 *** 0.801 *** 0.106 * 0.534 ***
Variety between Member States
Pearson correlation coefficients; ***, **, * coefficient significant at 1%, 5% and 10% level is also evident with regard to I_E
trends (Chart 19). Major improve-
Notes: NAIRU, non-accelerating inflation rate of unemployment. ments have been recorded in Ger-
Source: DG EMPL calculations using EU LFS. many and Spain, with the share of
inactive people moving to employ-
because the HP filter seems to have Improvements in trend developments ment within a year having almost
been only partly successful in sepa- for the ‘good’ transition rates (i.e. U_E tripled in Germany, from about 5%
rating cyclical from trend compo- and I_E) are likely to be associated to 15%, over a 20-year period, while
nents. In the EU-8, the E_U rate with a reduction in structural unem- in Italy and Greece trend declines
declined from 3.6% in 1993 to 2.7% ployment and a rise in trend partici- have been registered.
in 2007. As regards E_I trend rates, pation rates, contributing to the sig-
Chart 17c shows a more moderate nificant growth of employment reg-
decline, especially for the EU-7. istered in the EU since the mid-1990s.
As expected the trend components of
Overall, the evidence shows that the U_E and I_E transitions are nega-
increases in the trend components tively correlated with the structural
of ’good’ transitions (U_E and I_E) unemployment rate, as measured by
have been broadly accompanied DG ECFIN(37), and positively correlated
by reductions in the corresponding with both the employment and par-
’bad’ transitions. This signals that ticipation rates. Structural reductions
trend flows in both directions (from in unemployment rates seem to go
and to employment) have developed hand in hand with trend rises in the
in a way which is favourable to over- probability of ‘good’ transitions. Like-
all labour market performance in the wise, the trend component of E_U
EU. In particular, the recent evolution transition is positively correlated with
of E_I rates suggests that population structural unemployment and nega-
ageing has not resulted in a trend tively (albeit weakly) correlated with
rise in E_I transitions, partly reflect- the employment rate.(38) ­However,
ing the implementation of pension
reforms and a reduced reliance on
(38) However, correlation with participation (39) This was to some extent illustrated
early retirement schemes.
rate is, surprisingly, positive. already in Table 6 above.

Chapter 2 Labour flows, transitions and unemployment duration

Chart 18: Trends in U_E rates, selected Member States (%)

0.5 0.5

0.4 0.4

0.3 0.3

0.2 0.2
1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

0.30 0.40


0.25 0.30


0.20 0.20
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Source: DG EMPL calculations using EU LFS.

Chart 19: Trends in I_E rates, selected Member States (%)

0.15 0.12

0.12 0.10

0.09 0.08

0.06 0.06

0.03 0.04
1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

0.06 0.05

0.05 0.04

0.04 0.03

0.03 0.02
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Source: DG EMPL calculations using EU LFS.

3.2.3. Breakdowns by gender, • Overall U_E and I_E transition rates countries. Spain, Greece and Italy
age and education level tend to be higher for men than for register large gender gaps, espe-
women (Chart 20), unlike overall cially as regards I_E transition rates.
This section presents breakdowns of labour turnover and hiring rates In contrast, transition rates are
U_E and I_E transition rates by gen- (see Chart 7 above). higher for women for U_E rates in
der, age and education level (aver- the UK, Estonia and Finland, and
ages for the period 2002–07). The • However, the extent of the gen- for I_E rates in Sweden, Finland
main findings are: der gap varies substantially across and Hungary.

Employment in Europe 2009

Chart 20: Transitions by gender (%)

U_E I_E Men Women

60 20


30 10


0 0

Source: DG EMPL calculations using EU LFS.

Note: Six-year averages for 2002–07.

Chart 21:Transitions by age (%)

U_E I_E Young Prime Old

60 25

50 20

10 5

0 0

Source: DG EMPL calculations using EU LFS.

Note: Six-year averages for 2002–07.

Chart 22: Transitions by education level (%)

U_E I_E Medium Low High

80 35




0 0

Source: DG EMPL calculations using EU LFS.

Note: Six-year averages for 2002–07.

• Older workers (55–64) tend to have age workers, although this gap is While Charts 20–22 present average
considerably lower transition rates reversed for I_E transitions in most U_E and I_E transitions over 2002–
than young workers (15–24) and Member States, possibly due to 07, by country, Charts 23–25 apply
prime-age workers (25–54) (Chart 21), more of the younger worker age the HP filter to obtain time trends
especially with respect to I_E movers. group being in education. of U_E and I_E transitions broken
The proportion of unemployed older down by personal characteristics
workers finding a job within a year • As expected, higher education is since the late 1980s or early 1990s
is less than 20% in most Member associated with higher transition for, respectively, the EU-7 and EU-8.
States, with the notable exceptions rates, particularly I_E, where it The major findings are:
of the UK, Estonia and Italy. The larg- appears to be particularly criti-
est ‘age gaps’ in this respect are reg- cal in helping inactive persons • By gender (Chart 23), the trend rise
istered in Portugal, Spain, the Czech find a job. The widest ‘education in transition rates basically reflects
Republic and Denmark. gaps’ occur in the Czech Repub- gains made by women, leading
lic, the UK and Poland, and the to a substantial narrowing in the
• Young workers tend to have high- narrowest in Estonia, Greece and gender gap.
er U_E transition rates than prime- Germany.

Chapter 2 Labour flows, transitions and unemployment duration

Chart 23: U_E and I_E by gender, EU aggregates (%)

U_E, EU-8 I_E, EU-8 Men Women

0.40 0.15

0.35 0.12

0.30 0.09

0.25 0.06
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

U_E, EU-7 I_E, EU-7

0.40 0.15




0.20 0.06
1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Source: DG EMPL calculations using EU LFS.

Chart 24: U_E and I_E by age, EU aggregates (%)

U_E, EU-8 I_E, EU-8 Old Prime Young

0.4 0.15



0.1 0.00
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

U_E, EU-7 I_E, EU-7

0.5 0.20

0.4 0.15

0.3 0.10

0.2 0.05

0.1 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 0.00 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Source: DG EMPL calculations using EU LFS.

• By age groups (Chart 24), the trend tive gap for older workers which 2001. As regards the I_E transition
increase in the total U_E transi- existed at the start of the period rate, the rise reflects both develop-
tion rate largely reflects strong has widened further. ments in high and medium levels
improvements for young workers, of education. For both types of
while the rise in the total I_E transi- • By education level (Chart 25), the transition, the gap between high
tion rate basically reflects increases trend increase in the total U_E and low levels of education has
for prime-age workers, which in transition rate is concentrated in widened throughout the period.(40)
the case of the EU-7 aggregate the experiences of workers with
(40) It is interesting to note that the gap
almost tripled in the past 20 years. high education levels, while those between medium and low levels of edu-
Given the gains for young and with low education levels have cation is substantially lower than that
between high and medium levels of
prime-age workers, the large nega- experienced declining rates since

Employment in Europe 2009

Chart 25: Trends in transitions by education level, EU-8 (%)

U_E I_E Medium Low High

0.5 0.25




0.2 0.05
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Source: DG EMPL calculations using EU LFS.

Chart 26: Net flows between unemployment/inactivity and employment, EU aggregates (in thousands)

U_E - E_U I_E - E_I EU-8 EU-7

2000 3500
1500 3000

-1000 1000

-1500 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 500 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Source: DG EMPL calculations using EU LFS.

Chart 27: Net flows by activity status, by gender, EU-7 (in thousands)

U_E - E_U I_E - E_I Men Women

400 1500

200 1200

-600 300

-800 0
1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Source: DG EMPL calculations using EU LFS.

3.2.4. A brief analysis of all net flows.(41) However, there are cave- Chart 26 shows the evolution of net
labour market transitions, ats to such exercises since variables flows between unemployment and
based on net flows expressed in absolute numbers, unlike employment (U_E minus E_U) and
rates, do not take account of the between inactivity and employment
The analysis has largely focused on effect of changes in the overall size (I_E minus E_I) for two EU aggregates
‘good’ labour market transitions (i.e. of the relevant pool of workers. This (EU-7 and EU-8). Consideration of
U_E or I_E). However, it is also neces- implies that it is, a priori, difficult if net flows largely confirms the overall
sary to consider whether bad transi- not impossible to disentangle changes positive developments based only on
tions (i.e. from employment to unem- due to labour market efficiency from the ‘good’ transitions. In particular,
ployment or inactivity, E_U or I_U) those that are simply associated with improvements previously highlighted
may affect our overall conclusions. ­demographic, such as ageing. with respect to I_E transitions as well
as female workers are qualitatively
One basic way to assess both ‘good’ (41) Comparing transition rates (e.g. U_E and confirmed using net flows. The main
E_U) would be incorrect because they are
and ‘bad’ transitions is to calculate conclusions of this analysis are that:
calculated using different denominators.

Chapter 2 Labour flows, transitions and unemployment duration

Chart 28: Net flows by activity status, by age, EU-7 (in thousands)

U_E - E_U I_E - E_I Old Prime Young

600 3000
400 2500
0 1000
-200 500
-600 -1000
-800 -1500
1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Source: DG EMPL calculations using EU LFS.

Chart 29: Net flows by activity status, by skill level, EU-8 (in thousands)

U_E - E_U I_E - E_I Low Medium High

400 1500

200 1200

0 900

-200 600

-400 300

-600 0
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Source: DG EMPL calculations using EU LFS.

• For both types of transition (Chart • Although it is commonly thought 4. Unemployment

duration and long-term
26), net flows moved in a posi- that younger workers have the
tive direction, especially during the greatest difficulty in moving out of
mid-1990s, and showed consider- unemployment into employment unemployment
able resilience to the economic when there is a cyclical downturn
downturn of the early 2000s. in the economy, the data shows Although the European Employment
that prime-age workers are actu- Strategy (EES) has shifted the focus to
• By gender (Chart 27), changes in ally more adversely affected. employment, the study of unemploy-
net flows are more favourable for ment data remains important, espe-
women than men, particularly for • By skill level (Chart 29), the vari- cially during periods of recession. In
I_E net flows, corroborating ­previous ability in terms of U_E net flows is particular, it can help throw some
conclusions (e.g. Chart 23). largest for medium-skilled work- light on the factors determining its
ers, while I_E net flows show trend duration and the way employability
• By age (Chart 28), net flows have improvements for high and, more is affected, and better target policies
been positive for both young and markedly, medium skills. that facilitate re-employment.
prime-age workers, since 1995(42),
while registering negative values To summarise the information con-
for older workers throughout the tent of an economic variable, it is
period. The latter finding is clearly usually necessary to calculate at least
explained by demographic fac- two statistics: one of ‘location’ and
tors (i.e. retirements) as far as I_E another of ‘dispersion’. In the con-
flows are concerned, whereas it text of the EES, and with respect to
confirms the severe disadvantage unemployment, the unemployment
of unemployed older workers in rate can be seen as the ‘location’
terms of re-employment chances, statistic, and the long-term unem-
as ­reflected in U_E flows. ployment (LTU) rate (unemployment
duration of 12 months and more) can
be seen as playing the role of the
‘dispersion’ statistic.
(42) With the only exception of 2002 and 2003
for U_E net flows for prime-age workers.

Employment in Europe 2009

A simple example illustrates the need long-term unemployed, whose wage levels. This would be the case if
for unemployment statistics in rela- employability has fallen as a result variations in nominal wages depend-
tion to both ‘location’ and ‘disper- of their long absence from the ed solely on short-term unemploy-
sion’. A 10% unemployment rate labour market (Topel, 1990). ment, and not on the total stock of
can represent two entirely different the unemployed. Demand-side poli-
realities in terms of their implications As regards the latter hypothesis, a cies then have a potential impact on
for the (long-term) welfare of those number of studies find a negative the structural unemployment rate
affected: one where every individ- correlation between the duration of even in the long term, because of
ual in the labour force experiences unemployment and the rate of exit hysteresis effects. In this way, the
unemployment during 5 weeks per from unemployment even after con- long-term unemployed become a
year, and another where 10% of the trolling for ‘intrinsic’ characteristics major force creating high levels of
population are unemployed during of workers – i.e. personal factors unemployment.
the whole year. Provided adequate affecting their employability (Machin
income support is available, unem- and Manning, 1999). Ball (2009) finds empirical evidence sug-
ployment in the former situation is gesting that structural unemployment
a temporary state that is unlikely to It is also commonly argued that the is influenced by aggregate demand,
leave enduring ‘scars’ on the people structure of unemployment dura- reflecting the presence of hysteresis
concerned in terms of their employ- tion (or the incidence of LTU) is mechanisms. Since aggregate demand
ability. In the latter case, however, an important determinant of ‘search influences actual unemployment,
the consequences are likely to be effectiveness’ (Layard et al., 1991). hysteresis means that demand also
more significant and long-lasting. This is because, as the long-term influences structural unemployment.
unemployed become more and more Although some form of hysteresis
In order to distinguish between these detached from the labour market, seems to exist, it is not clear either
two situations it is necessary to cal- they play less and less of a role in what are the precise mechanisms at
culate a measure of unemployment competing for jobs and determining work or the policy ­implications. ­Further
‘dispersion’, such as unemployment
duration. To this end, this section Table 9: Incidence of LTU in OECD countries, 2007 (%)
uses both EU LFS and EU SILC data
Proportion unemployed Proportion unemployed Standardised
to calculate various statistics of more than 6 months more than 12 months unemployment rate
­unemployment duration and LTU. (43) AU 27.1 15.5 4.4
AT 44.2 26.8 4.4
The relationship between unemploy- BE 68.1 50.0 7.5
ment duration and employability is CA 14.8 7.5 6.0
central to our concerns. Some strands CZ 71.6 53.4 5.4
of economic theory suggest that DK 29.5 18.2 3.8
hysteresis effects play an important FI 37.9 23.0 6.8
role in labour market dynamics – i.e. FR 58.5 40.4 8.3
DE 71.3 56.6 8.4
that the long-run equilibrium level
EL 68.2 50.3 8.3
of unemployment depends on past
HU 64.0 47.6 7.3
levels of unemployment.(44) Two basic
IE 50.1 30.3 4.6
mechanisms have been put forward
IT 65.4 49.9 6.1
in support of this argument (Cahuc JP 48.2 32.0 3.9
and Zylberberg, 2004): KR 11.7 0.6 3.2
LU 54.7 33.5 4.2
• first, that the bargaining power of MX 5.4 2.7 3.7
insiders ensures that, even when NL 59.1 41.8 3.2
faced by a reduction in the demand NZ 16.7 5.7 3.7
for labour, they are nevertheless NO 25.8 8.8 2.5
able to maintain or increase wages PL 64.3 45.9 9.6
PT 67.6 47.3 8.1
ES 42.6 27.6 8.3
• second, the reduced bargaining
SK 82.3 70.8 11.2
power and expectations of the
SE 27.3 13.0 6.2
CH 56.6 40.8 3.6
(43) EU LFS group averages, and micro data TR 46.3 30.4 8.6
from the users’ database (USB) of EU SILC.
UK 41.5 24.7 5.3
(44) Related to this is the idea that a transi- US 17.6 10.0 4.6
tory shock has permanent effects and
that demand policies produce long-term Source: OECD.

Chapter 2 Labour flows, transitions and unemployment duration

Chart 30: Incidence of LTU and the unemployment rate of unemployment spells by duration
(Chart 31). In Europe, close to 45% of
all incomplete unemployment spells
70 SK (or spells in progress) last for longer
than 1 year compared with only
60 about 10% in the USA. Japan is in an
Proportion unemployed > 12 months

CZ intermediate position, but closer to
50 BE EL
the EU than the USA.
40 CH FR

On average over the economic cycle,
30 IE TR institutional labour market factors,
such as net replacement rates(46) and
EPL, are generally considered to play an
SE important role in explaining differences
NZ CA in the incidence of LTU (Box 8).(47)
0 2 4 6 8 10 12
Standardised unemployment rate
4.1. Measurement
Source: OECD.
issues with regard to LTU

Chart 31: Distribution of unemployment spells in progress by duration (%) 4.1.1. Administrative versus
survey data
1 year and over

There are two main sources of infor-

> 6 months & < 12 months
mation on the duration of unem-
ployment: survey and administrative
data. The latter is generally country-
> 3 months & < 6 months
2005-2007 US
specific – i.e. the data is obtained
2005-2007 JP from social security or public employ-
2005-2007 OECD-Europe
> 1 month & < 3 months ment agencies information systems,
and is sensitive to national idiosyn-
< 1 month
crasies, hampering international com-
parisons. It is also subject to frequent
0 10 20 30 40 50 breaks in series caused by changes in
administrative procedures, render-
Source: OECD.
ing timeseries analysis difficult even
within countries.
research on the topic is necessary, ­ articular cause of concern given
which has been neglected in recent extensive evidence (European Com- Given these problems, researchers
years, especially given the risks posed mission, 2007) that unemployment generally prefer to use survey-based
by the current recession. is closely associated with the risk of statistics, both cross-sectional and
poverty and higher levels of income longitudinal. In this respect, national
Table 9 presents some data on the inequality. labour force surveys, and the EU’s
proportion of the unemployed who LFS, have adopted the International
have been unemployed for more Chart 30 suggests a positive correla- Labour Organisation’s (ILO) definition
than 6 and 12 months in OECD coun- tion between total unemployment of unemployment, ensuring that data
tries, the latter being the official and the incidence of LTU, although on the duration of unemployment
measure of the incidence of LTU used some countries seem to be outliers.(45)
for policy analysis in the EU. In inter-
(46) These indicators are obtained by calculat-
national comparisons, a distinctive Overall, European countries score ing the ratio of net income when not
feature of EU labour markets is seen poorly in comparisons with the USA working (mainly unemployment benefits
to be the high incidence of LTU – a and Japan regarding the distribution if unemployed or means-tested benefits
if on social assistance) to net income in
factor that is cause for concern for work. A lower replacement rate is associ-
(45) For countries with unemployment rates
both equity and efficiency reasons. ated with a greater incentive to search for
ranging between 3 and 5%, the propor- and take up a job when unemployed.
tion of long-term unemployed varies
A high concentration of LTU among between less than 10% (New Zeeland, (47) In the USA, net replacement rates are
Mexico, Korea, the USA) and more than lower and EPL less stringent than on
households of working age is a
40% (the Netherlands and Switzerland). average across the EU.

Employment in Europe 2009

is comparable across countries.(48) unemployment’ if the exit rate from proposed which are more robust in
­ owever, in contrast to administrative
H unemployment declines with duration terms of taking account of short inter-
data, survey questions on the dura- (Salant, 1997).(50) ruptions to periods of unemployment,
tion of unemployment are seen to be such as the fraction of the year that
subject to ‘recall bias’ in that individu- EU official statistics on the incidence an individual spends unemployed. The
als appear to under-report short spells of LTU, measuring the proportion of OECD (2002) has also used detailed cal-
in unemployment. Moreover, longitu- the current unemployed who have endar information on labour market
dinal data (e.g. the ECHP) can suffer been unemployed for more than a status to calculate measures of inci-
from ‘seam effects’ – i.e. the tendency given time, are based on EU LFS infor- dence and duration of LTU for 11 EU
for changes in reported labour mar- mation on the duration of unemploy- Member States using ECHP data.
ket status to occur between the last ment spells in progress. The period
month covered by one interview and normally used for LTU is 12 months, Second, Akerlof and Main (1980) devel-
the first month covered by the next with 24 months used for the very LTU. op the following argument regarding
interview (OECD, 2002). A period of 6 months has also been the use of longitudinal data in assess-
used occasionally in the past. ing the incidence and duration of
LTU – namely that average statistics on
4.1.2. Duration of The decision to publish unemployment the duration of unemployment (either
incomplete versus completed duration statistics based on ‘in progress’ completed or in progress) based on
unemployment spells unemployment spells is basically a mat- LFS/cross-sectional data are meaning-
ter of convenience. Established meth- ful measures only if a large majority of
In terms of official data, the statistic ods are available (both parametric and individuals experience only one spell
usually quoted in relation to unem- non-parametric) to calculate statistics of unemployment during the period
ployment duration is actually the ‘aver- of average length of completed spells of observation. In practice, however,
age duration of unemployment spells of unemployment using the informa- they find that multiple unemployment
currently in progress (or interrupted/ tion on the spells in progress (Corak spells are a relatively common experi-
incomplete spells)’ based on cross-sec- and Heisz, 1995). Box 5 and section ence, at least in the USA. Consequent-
tional labour surveys. Unfortunately, 4.3 present and apply, respectively, a ly, statistics based on average duration
this measure tends to overestimate non-parametric method to calculate of (completed) unemployment spells
the average duration of completed statistics on the average duration of seriously misrepresent the unemploy-
unemployment spells – a statistic that completed unemployment spells using ment experience of all groups: people
would be a more appropriate indicator quarterly EU LFS data. Results will then with single spells because on average
for assessing the welfare implications be compared with average duration single spells are longer than multi-
of unemployment.(49) In fact, the mean of incomplete unemployment spells, ple spells; and people with multiple
average length of ‘unemployment in ­highlighting the main differences. spells because, on average, individual
progress’ spells exceeds the mean aver- spells are shorter than for persons with
age length of ‘completed spells of a single spell overall, although total
4.1.3. Longitudinal data: the time spent in unemployment might
(48) Eurostat uses the following statistical importance of using detailed be longer for persons with multiple
Unemployed persons are all persons calendar information spells, and because the unemployment
aged 15–74 who were not employed experience of these persons includes
during the reference week, had actively Statistics based on cross-sectional sur- the multiplicity of their spells which is
sought work during the past four weeks
and were ready to begin working imme- vey data (either for incomplete or not considered in the analysis.
diately or within two weeks.  completed unemployment spells) suf-
Employed persons are all persons who fer from two major drawbacks. First, In order to address issues related
worked at least one hour for pay or
profit during the reference week or were they are very sensitive to short breaks to the multiplicity of unemployment
temporarily absent from such work.  in unemployment – e.g. in principle spells, detailed calendar information
The unemployment rate is the number a single day’s employment during a on labour market status is used in
of people unemployed as a percentage
of the labour force.  period of unemployment will auto- section 4.9 to calculate the incidence
The labour force is the total number of matically reset the clock to zero (OECD, and duration of LTU for 14 European
people employed and unemployed. 2002). Other definitions/statistics of countries using EU SILC. The results of
(49) In this regard, an analogy with popula- unemployment duration have been this empirical analysis confirm Aker-
tion data might be helpful (Akerlof and
lof and Main’s (1980) main insight –
Main, 1980). The difference between
measures of average duration of com- (50) Empirical analysis find that exit rates from namely that the use of cross-sectional
plete and in progress unemployment unemployment decline with unemploy- (EU LFS) versus longitudinal (ECHP,
spells is analogous to the difference ment duration, regardless of the cause for
EU SILC) data to measure the inci-
between mean life span (equivalent to this decline being either unobserved het-
the average length of completed life) erogeneity or pure duration dependence. dence/duration of LTU implies a major
and the mean age of the population For more on the relations between statis- drawback, because the group average
(equivalent to the average length of tics on average complete and in progress
nature of the former does not capture
lives currently in progress). unemployment spells see Box 5.

Chapter 2 Labour flows, transitions and unemployment duration

important information on the distri- Table 10: Long-term unemployment and joblessness in the EU, 2005–07 (%)
bution of unemployment duration by
number of spells in unemployment. Joblessness/not employed Unemployed
of which of which
Have not Have not Have not Have not
Total worked for the worked for the Total worked for the worked for the
4.1.4. Consideration past year past two years past year past two years
of a broader concept of Males
‘joblessness’ 15–24 52.3 1.7 0.7 16.8 2.5 1.0
25–54 8.1 4.7 4.0 6.4 3.3 2.2
55–64 43.8 38.3 33.5 6.1 4.3 3.2
In seeking to better assess the over-
all attachment to the labour force
15–24 59.3 3.0 1.7 17.4 2.7 1.2
of the working-age population, an
25–54 23.6 14.2 12.4 8.0 4.4 3.3
exclusive focus on the unemployed, 55–64 63.0 49.9 46.0 5.9 4.2 3.2
or the long-term unemployed, can be
seen as unduly restrictive as it ignores Source: DG EMPL calculations based on EU LFS data. LFS variable used: LEAVTIME.
those who are not working for a Note: Joblessness is defined as the absence of employment during the period shown.
variety of reasons – notably because
they are ‘discouraged’ from looking
for work because they believe no
work is available. For that reason, the Chart 32: Incidence of LTU in the EU – more than 6, 12 or 24 months (%)
OECD (2002) has calculated a broader IR6M IR24M IR12M
concept of ‘long-term joblessness’, 0.8

which goes beyond the standard ILO

definition, and is defined as the work- 0.7
ing-age population who were not in
employment at the time of the labour 0.6
force interview and have not worked
within the last one or two years.

Comparison between data for the

long-term unemployed, as traditionally 0.4

defined, and this measure of long-term

joblessness/non-employment are poten- 0.3
tially more significant for prime-age
men (25–54), who are generally expect- 0.2
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
ed to be in employment unless there
are special circumstances, such as health Source: DG EMPL calculations based on EU LFS. Variable used: DURUNE.
problems or being in lifelong education.
In this respect, the proportion of the
prime-age population (25–54), especially
women, who were without a job in the Chart 33: Incidence of LTU by country (%)
previous one or two years is significantly
higher than the numbers recorded as
unemployed (Table 10). PL
0.7 IT
4.1.5. Main characteristics 0.6
of LTU
This section presents information on the
incidence of LTU(51) over time for an EU
aggregate.(52) Chart 32 suggests that, on
(51) Unemployment lasting for 12 and more 0.3
months over total unemployment.

(52) This aggregate is made of all the 27 EU

Member States for which data is avail- 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
able. It varies during the 1992–2007
period; therefore charts should be inter- Source: DG EMPL calculations based on EU LFS. Variable used: DURUNE.
preted with caution.

Employment in Europe 2009

average across the EU, the distribution Chart 34: Incidence of LTU by gender (%)
of unemployment spells by duration has
not changed markedly since 1992 – the
first year for which data is available. Males

Chart 33 presents data for the five-

largest EU economies. In Germany 0.5
and France, the incidence of LTU
increased between 2002 and 2007.
Since the early 1990s, among the
largest EU economies the incidence
of LTU has significantly fallen in Spain
and the UK, while in Poland, the inci-
dence of LTU rose between 1999 and
2005, having since declined.
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
Across the EU, there has been a con-
vergence in the incidence of LTU by Source: DG EMPL calculations based on EU LFS. Variable used: DURUNE.
gender (Chart 34).

Despite the high rate of unem- Chart 35: Incidence of LTU by age (%)
ployment of younger workers (see
Table  10), the incidence of LTU is
low relative to older age groups
(Chart 35), in line with their low 0.7
average duration of unemployment
(see Chart 47). It has also fallen since 0.6
the mid-1990s.
Chart 36 suggests that since the mid-
1990s there has been on average
across the EU a significant reduction
in the incidence of LTU for the unem-
ployed with high levels of education.
This compares with persistent higher
levels for those with low and medium 0.2
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
levels of education, although for the
latter two groups there has been some Source: DG EMPL calculations based on EU LFS. Variable used: DURUNE.
­modest improvement in recent years.

Chart 36: Incidence of LTU by education level (%)





1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Source: DG EMPL calculations based on EU LFS. Variable used: DURUNE.

Chapter 2 Labour flows, transitions and unemployment duration

Table 11: Composition of LTU (1 year and over) by major breakdowns (average values 2005–07) (%)
Males Females Youth Prime-age Older workers Low education Medium education High education
AT 27.3 25.6 14.1 28.9 58.2 29.7 24.4 25.8
BE 49.9 52.1 28.4 56.5 79.7 59.8 47.6 38.5
BG 56.9 59.5 43.9 60.6 67.0 66.8 52.7 51.0
CY 19.7 20.1 na 21.0 41.4 23.3 20.8 21.0
CZ 52.0 54.2 36.1 57.8 54.2 70.3 48.3 33.0
DE 54.7 54.8 31.4 56.2 72.7 56.5 54.4 51.8
DK 20.1 20.1 na 20.9 45.4 17.8 21.4 21.9
EE 50.4 50.0 32.2 56.0 70.8 48.6 52.9 49.5
ES 18.6 25.0 11.8 23.3 47.9 23.7 19.6 21.1
FI 27.6 21.2 5.6 28.1 49.3 25.0 21.7 30.4
FR 40.5 40.7 23.8 44.4 65.0 48.3 36.2 33.2
GR 43.4 57.2 44.9 53.9 56.7 52.0 54.3 47.1
HU 46.2 44.9 36.0 47.6 54.1 49.8 44.5 36.7
IE 38.5 21.3 21.5 35.8 44.4 43.7 25.1 18.3
IT 45.8 49.7 41.9 49.5 55.4 50.8 46.8 38.9
LT 42.9 43.0 23.6 46.1 55.9 47.3 43.1 38.5
LU 33.5 24.3 20.7 28.9 54.0 29.8 22.2 29.5
LV 40.0 32.7 20.7 40.2 55.0 34.9 39.2 34.2
MT 49.5 32.4 31.5 53.5 na 46.4 29.8 na
NL 42.7 37.4 15.7 45.9 69.5 36.9 42.2 43.5
PL 53.9 56.3 40.6 59.6 65.2 62.5 55.3 39.0
PT 48.6 48.1 29.4 51.1 69.1 52.3 42.3 32.2
RO 55.5 52.9 49.2 56.6 58.7 54.9 54.8 49.1
SE 16.2 12.4 4.7 18.4 30.8 11.0 16.5 17.6
SI 47.8 47.1 34.4 51.0 64.8 54.2 47.2 38.1
SK 74.7 73.6 59.5 77.8 83.3 87.6 70.5 44.9
UK 26.6 16.3 13.9 26.8 37.0 28.1 18.6 17.0
EU 44.5 44.5 28.8 48.0 63.0 46.1 46.1 35.0
Source: DG EMPL calculations based on EU LFS. Variable used: DURUNE.
Legend: Youth (16-24); Prime-age (25-54); Older workers (55-64);
Low education (isced 00-21); Medium education (isced 22-43); High education (51-60).

4.2. Relation over the the relationship between unemploy- flows into unemployment and the
economic cycle between ment and the incidence/proportion unemployment rate fall, the stock of
of LTU displays counter-clockwise unemployed becomes more heavily
unemployment and the loops (Charts 37–43). weighted with individuals who are in
incidence of LTU the midst of longer unemployment
At the onset of an economic down- spells which had begun during the
Machin and Manning (1999) mention turn, higher inflows into unem- downturn phase. As a result, for a
that, over the economic cycle, the ployment tend to reduce the aver- given level of unemployment, the
incidence of LTU lags actual unem- age incidence of LTU. Conversely, incidence of LTU is usually higher
ployment or, in more technical terms, during economic expansion, while during upturns than downturns.

Chart 37: The unemployment rate and the incidence of LTU in the EU –
more than 6 months, top panel; more than 12 months, bottom panel (%)
Proportion unemployed > 6 months

0.68 97
95 98 96
0.66 94
01 02
0.64 00
03 93
0.62 05 99
0.60 06
0.07 0.08 0.09 0.1 0.11 0.12 0.13
Unemployment rate
Proportion unemployed > 12 months

98 97
0.49 96

0.47 01 95
00 94
06 05 02 99
0.45 03
07 93
0.07 0.08 0.09 0.1 0.11 0.12 0.13
Unemployment rate

Source: EU LFS.

Employment in Europe 2009

Chart 38: The unemployment rate and the incidence of LTU in Germany –
more than 6 months, top panel; more than 12 months, bottom panel (%)
Proportion unemployed > 6 months

0.70 07
98 05
00 95 99 97
0.65 04
01 96
02 94

0.06 0.07 0.08 0.09 0.1 0.11 0.12
Unemployment rate
Proportion unemployed > 12 months

07 06
99 98 05
0.50 00 04
01 95 03 97
0.45 02 96
0.06 0.07 0.08 0.09 0.10 0.11 0.12
Unemployment rate

Source: EU LFS.

Chart 39: The unemployment rate and the incidence of LTU in France –
more than 6 months, top panel; more than 12 months, bottom panel (%)
Proportion unemployed > 6 months


0.60 06 98 97
05 95
04 94
0.58 96
07 00
0.56 03
01 92
0.075 0.085 0.095 0.105 0.115 0.125 0.135
Unemployment rate
Proportion unemployed > 12 months

0.42 98
05 00 95
0.40 04
07 97
99 96
03 94
0.36 01
02 93
0.075 0.085 0.095 0.105 0.115 0.125 0.135
Unemployment rate

Source: EU LFS.

Chart 40: The unemployment rate and the incidence of LTU in Italy –
more than 6 months, top panel; more than 12 months, bottom panel (%)
Proportion unemployed > 6 months

01 00 99 97
0.75 98
02 96
03 94 95
06 05 92
07 04

0.06 0.07 0.08 0.09 0.10 0.11 0.12
Unemployment rate
Proportion unemployed > 12 months

01 00 99
0.60 98
03 02
0.55 96
94 95
0.50 06 05 93
07 04 92

0.06 0.07 0.08 0.09 0.10 0.11 0.12
Unemployment rate

Source: EU LFS.

Chapter 2 Labour flows, transitions and unemployment duration

Chart 41: The unemployment rate and the incidence of LTU in the UK –
more than 6 months, top panel; more than 12 months, bottom panel (%)
United Kingdom
Proportion unemployed > 6 months

95 94 93
96 92

00 99
01 07
0.39 06
0.34 04 03
0.045 0.055 0.065 0.075 0.085 0.095 0.105
Unemployment rate
United Kingdom
Proportion unemployed > 12 months

0.35 92
00 99
04 05 03 06
0.20 02
0.045 0.055 0.065 0.075 0.085 0.095 0.105
Unemployment rate

Source: EU LFS.

Chart 42: The unemployment rate and the incidence of LTU in Spain –
more than 6 months, top panel; more than 12 months, bottom panel (%)
Proportion unemployed > 6 months

97 96 94
98 93
0.60 92
01 00
0.50 02

06 05
0.08 0.1 0.12 0.14 0.16 0.18 0.2 0.22 0.24
Unemployment rate
Proportion unemployed > 12 months

97 95 94
0.5 96
0.45 93
00 92
0.35 02
0.3 04

07 06
0.08 0.1 0.12 0.14 0.16 0.18 0.2 0.22 0.24
Unemployment rate

Source: EU LFS.

Chart 43: The unemployment rate and the incidence of LTU in Poland –
more than 6 months, top panel; more than 12 months, bottom panel (%)
Proportion unemployed > 6 months

0.76 05
04 03
06 02
07 97
98 00

0.095 0.115 0.135 0.155 0.175 0.195
Unemployment rate
Proportion unemployed > 12 months

06 03
04 02
0.52 07
0.48 98
97 00
0.09 0.11 0.13 0.15 0.17 0.19 0.21
Unemployment rate

Source: EU LFS.

Employment in Europe 2009

Box 5: Methodology to estimate an unbiased measure of the average duration of unemployment

The average incomplete duration of unemployment (or of spells currently in-progress) is a biased measure of the
completed duration of unemployment

The welfare of the unemployed depends on the adequacy of replacement income and the conditional probability (or hazard rate) of leaving unemployment
and obtaining a job. Success, in turn, depends on personal circumstances (notably age, education, social networks), the (dis)incentives of the tax-benefit
system (e.g. entitlement to and duration of unemployment benefits), and the availability and quality of public employment services. The duration of unem-
ployment is thus an important determinant of the welfare cost of unemployment, probably more important than the unemployment status itself.

The EU LFS does not capture the complete length of an unemployment spell, only the length of time spent unemployed up to the reference
week. The EU LFS records the duration of incomplete unemployment spells (or spells in-progress) as the shorter of the following two periods:
the length of time since last employment or the duration of search for work.

The average duration of incomplete unemployment spells can be calculated as the sum of all in-progress spells divided by the number of
unemployed persons. A preferable indicator to measure the duration of unemployment would be the average expected completed duration
of unemployment (of a synthetic cohort entering unemployment at the same time). The average duration of incomplete unemployment spells
is a biased measure of the average completed duration because of the presence of two biases: a length bias and a sampling bias (Chart 44).
1. The length bias arises because a large fraction of the unemployment spells is right-censored – i.e. duration is above a certain
value but by exactly how much it is unknown. This tends to under-estimate the complete spell length.
2. The sampling bias reflects the fact that there is a higher probability of sampling individuals with longer unemployment durations – i.e. the
‘stock sampling problem’. This tends to overestimate the complete spell length.

Salant (1977) shows that the average incomplete duration of unemployment will be greater than the average completed duration if the
probability of moving out of unemployment (i.e. the hazard rate) declines with the time spent unemployed. This condition is generally satisfied.
As a result, the effect of the sampling bias is seen to outweigh the effect of the length bias.

Given that the EU LFS is a continuous survey (data collection in all weeks of the year), while the reference period for unemployment is four
weeks, the sampling bias should affect only individuals in unemployment for very short periods (i.e. less than a month).
Chart 44: Duration data and biases
Sampled spells





Qt Qt+1 Qt+2 Calendar time

Qt, Qt+1, Qt+2: date of survey (in consecutive quarters)
t1, t2: completed spells (sampling bias)
t3, t4: right-censored spells (length bias)

A non-parametric methodology to calculate the average completed duration of unemployment using the EU LFS

The objective is to estimate the length of completed unemployment spells from data on spells in-progress. The methodology used in this
chapter follows Sider (1985) and Corak and Heisz (1995) in using a synthetic cohort approach and a non-parametric method based on lifetime
and actuarial methods, such as those of Kaplan and Meier (1958). The calculation is based on aggregate probabilities (or group averages) that
individuals continue in unemployment from one survey period (month/quarter) to the next.

Let f(x) be the number of individuals remaining in unemployment after x periods 0<x<n, where n is the maximum length in unemploy-
ment. The average duration of a completed spell is simply the sum of all spells weighted by the probability of leaving unemployment at each
­duration. The average duration of a completed spell is given by S:

(Equation 1)

Chapter 2 Labour flows, transitions and unemployment duration

where pj is the continuation rate between periods j-1 and j. The product of the pjs and (1-px) is the proportion of the original cohort of unem-
ployed leaving unemployment after x periods.

Evaluating Equation 1 using current continuation rates yields an estimate of the expected unemployment duration for a synthetic cohort of
individuals entering unemployment. It reflects the average completed spell duration that would be incurred if current continuation rates were
maintained into the future.

Note that Equation 1 can be rewritten in a slightly more familiar form, involving the hazard and survival(53) functions:

(Equation 2)

Where h(x) is the hazard function: the conditional probability of leaving unemployment after x periods. s(x-1) gives the probability of surviving
x-1 periods in unemployment.

This chapter uses quarterly data from the EU LFS specifically the derived variable LEAVTIME, measuring the duration of unemployment for an
unemployed person with previous work experience, in order to calculate the following continuation rates, which are calculated as the ratios
of the number of individuals in each of the following categories:

0–2 months in quarter t–1 to 3–5 months in quarter t

3–5 months in quarter t–1 to 6–8 months in quarter t
6–11 months in quarter t–2 to 12–17 months in quarter t
12–17 months in quarter t–2 to 18–23 months in quarter t
18–29 months in quarter t–3 to 27–38 months in quarter t
30+ months in quarter t–6 to 48+ in quarter t

Continuation rates are then converted to monthly equivalents (by raising them to the ‘appropriate’ power). This assumes that monthly continu-
ation rates are constant within each interval.

Some notes of caution

It should be stressed that the methodology presented in this box was originally developed for labour force surveys with a monthly periodicity in
which the duration of unemployment is measured in weeks. In this respect, the EU LFS presents a number of limitations since it has a quarterly
periodicity and measures unemployment duration in months. Moreover, exploitable statistics are only available since around the turn of the
century, although the question on the duration of unemployment/inactivity was included in the 1992 survey.

Corak and Heisz (1995) point to the need to smooth raw data on durations prior to applying the methodology developed above, because
they find significant spikes in the frequency distribution of the reported incomplete duration spells associated with survey respondents’ ‘digit
preference’ – namely the fact that individuals taking part in the survey tend to report even, rather than odd, numbers.

Furthermore, results are subject to three major qualifications. First, they should be interpreted in terms of the average unemployment experi-
ence which may include multiple spells as opposed to the single spell stereotype. Second, the underlying data do not permit us to distinguish
between spells of unemployment at the end of which the person concerned finds employment, and those where they end up leaving the labour
force altogether. Third, no information is available concerning the reason for their becoming unemployed, which might have enabled us to
draw some conclusions in relation to their subsequent unemployment histories.(54)


This box adapted a non-parametric methodology, first developed for USA and Canada in the early 1970s (Kaitz, 1970), to calculate an unbiased
measured of the average duration of unemployment for EU Member States using the EU LFS. The resulting measure reflects the completed
duration of the unemployment spell for a synthetic cohort that enters unemployment today and which faces current economic conditions
throughout the entire duration of the unemployment spell.

Although the empirical calculations and conclusions are tentative and provisional, and therefore should be interpreted with caution, they do give
us a better understanding of the factors shaping unemployment duration and hazard rates, potentially leading to more effective policies.

(53) The survival function shows what proportion of a cohort of people who become unemployed remains unemployed as time passes.

(54) Section 4.4 gives information about the distribution of the duration of incomplete unemployment spells by main reason of leaving/losing the
last job.
Employment in Europe 2009

4.3. Statistics on Chart 45: Statistics on the average duration of CDU and IDU spells,
the average duration average in period 2005 Q1 to 2008 Q3 (months)

of both completed 35

and incomplete 30

unemployment spells

Using the methodology presented in 20

Box 5, quarterly EU LFS data is used

to calculate statistics for both the 15
average duration of completed and
incomplete unemployment spells. An 10
attempt is made to compute quar-
terly values from 1992 Q1 to 2008 5
Q3, with breakdowns by country,
gender, education and age. 0
CDU 13.9 12.2 12.1 11.9 10.4 10.2 10.1 9.9 9.7 9.6 9.5 9.1 8.9 8.6 8.4 8.2 8.1 7.9 7.5 6.9 6.9 6.8 6.8 6.6 6.6 6.4 3.5
IDU 21.8 31.9 26.1 21.8 22.1 24.5 26.2 27.8 27.1 24.2 18.4 24.7 18.8 22.4 19.2 25.1 21.6 20.8 15.6 15.9 19.5 16.8 12.5 14.3 15 13.3 26.4 20.5
This exercise highlights two prac-
tical obstacles to the calculation Source: DG EMPL calculations based on EU LFS data.
of average completed unemploy- Note: Countries are ranked in descending order of CDU.
ment spells using a non-parametric
Chart 46: Statistics on the average duration of completed (CDU) unemployment
1. The statistic of completed spells by gender, average in period 2005 Q1 to 2008 Q3 (months)
unemployment spells can only 15
effectively be calculated after
2000 Q1, and even then, on
average just for about one 12

quarter of the maximum

number of possible data points,
which compares with a much

longer time coverage for statis-

tics of incomplete duration of 6
­unemployment spells.

2. In practice, some dataseries on 3

unemployment duration are

degenerate – i.e. they contain
cases of unemployment that are HU SK DE EU GR CZ BG PL RO BE IE SI UK EE FR NL LT IT FI LU LV AT CY SE DK ES MT
Women 13.6 14.7 12.3 12 11.5 10.8 10.7 10.3 9.8 8.7 8.5 9.3 8.9 6 8.6 7.6 7.9 8.5 7 5.9 8.6 6.9 7.2 6.4 6.7 7.1 4.2
of infinite duration (Portugal, Men 13.8 12 12.1 11.7 9.1 9.5 9.2 8.9 10.5 9.6 10.5 7.6 9.2 6.1 8.2 7.9 7.3 7.4 7.7 6.3 5.5 6.6 6.1 6.8 6.3 6.2 6.5

2008), implying that a signifi-

Source: DG EMPL calculations based on EU LFS data.
cant fraction of the unemployed
Note: Countries are ranked in descending order of totals (not shown).
may never find a job and leave
unemployment. This presum-
ably reflects either very few job In order to better evaluate the poten- duration is significant higher than
offers being available in the tial biases involved in the calculation the completed duration by a factor
labour market and/or very little of indicators on unemployment dura- of around 2 for the EU. However,
pressure on those recorded as tion, Chart 45 compares average val- these results are preliminary requir-
unemployed to seek work, pos- ues for the period 2005 Q1 to 2008 ing further research. In particular,
sibly because the unemployment Q3 for statistics on both the average the large gap between IDU and
compensation scheme is being completed duration of unemploy- CDU statistics – besides reflecting
used as a more general form of ment (CDU) and average incomplete the effects of the two sampling and
income support. Obviously, sta- duration of unemployment (IDU). length bias – results from problems
tistics on the completed duration with the method used to calculate
of unemployment cannot be cal- As expected, the sampling bias CDU statistics, namely the need to
culated for such ­degenerate dis- outweighs the length bias; conse- smooth the data prior to applying
tributions. quently the average incomplete the methodology.

Chapter 2 Labour flows, transitions and unemployment duration

Chart 47: Statistics on the average duration of completed (CDU) unemployment education. In the EU during 2005
spells by age, average in period 2005 Q1 to 2008 Q3 (months) Q1–2008 Q3, the average duration
of unemployment spells for unem-
ployed persons with low level of edu-
cation was 12.3 months, compared
with only 8.1 months for those in the
15 high education group.

4.4. Unemployment


duration by reason
of leaving/losing
Based on the EU LFS, Charts 49 and 50
present information about the distri-
15-24 9.9 8.8 9.9
25-54 14.2 12.8 12.3
10.3 10.6 7.6 7.1 9 8.6 6.4 7.6
12.2 10.6 10.4 9.4 9.4 9.6 10.7 11.2
7.3 8.1 4.2
9 9.7 7.3
7.2 5.7
8.7 8.2
6.1 3 6.1 6 6.1
7.1 7.3 6.5 6.2 6.5
bution of incomplete unemployment
55-64 13.7 9.8 15 14.7 7 11.8 11.2 10.2 7.5 6.2 9.9 3.3 9.3 15.6 8.6 13.1 6.8 8.1 9.9 5.3 5.8 4.6 8.4 6.2 7.7 spells by the main reason of leaving
(or having lost) the last job.
Source: DG EMPL calculations based on EU LFS data.
Note: Countries are ranked in descending order of totals (not shown).
Chart 49 shows that the reasons dis-
missal/redundancy and own illness/
Chart 48: Statistics on the average duration of completed (CDU) unemployment disability are associated with more
spells by education level, average in period 2005 Q1 to 2008 Q3 (in months) flat distributions, which correspond
to relatively high average durations
of unemployment.(55)

On average, becoming unemployed

at the end of a fixed-term contract
seems to be associated with relatively
9 low unemployment duration, par-
ticularly when compared with hav-

ing been dismissed or made redun-

dant (Chart 50). In fact, in terms of
unemployment duration, it looks as
if dismissal/redundancy is as ‘bad’ as
having lost/left a job on account of
own illness/disability (Chart 49).
Low 14.2 13.7 13.3 12.3 9.5 9.6 8.9 8.3 6.6 9 10.7 8.8 10.4 3.4 9.9 7.2 7.3 7.7 7.4 6.2 6.2 6.7 7.5 6 5.1 5.8
Medium 13.6 13.2 12.2 11.8 10.6 9.9 9.5 9.8 10.9 9.6 9 8.3 8.7 7.8 7.9 7.8 7.9 8.1 7 6.4 6.3 6.2 4.1 6.7 6.6 6.4
High 11.8 10.1 9.7 8.1 12.4 9.2 8.4 7.7 9.5 5.7 6.3 6.8 7.4 5.9 8.1 7.4 5.2 7.1 6.7 6.4 5.9 5.6 5.6 6.1 6.3 6.6

Source: DG EMPL calculations based on EU LFS data.

Note: Countries are ranked in descending order of totals (not shown).

Despite country differences, in the EU (25–54) (e.g. Slovakia, Greece, Roma-

as a whole, gender gaps for the aver- nia and Belgium), this may reflect
age duration of completed unem- stronger ‘discouragement’ effects for
(55) Average values are not calculated,
ployment spells are small (Chart 46). the former. In fact, low U_E transi- because EU LFS’s DURUNE variable
tion rates reflecting limited prospects reports unemployment duration in inter-
vals. The EU LFS records 10 main reasons
Although there are country differ- of finding a job (see section 3 on
for leaving (or having lost) employment:
ences, average completed durations transitions) may lead older workers ‘a job of limited duration has ended’,
of unemployment increase with age to exit the labour force altogether. ‘compulsory military or community serv-
ice’, ‘dismissed or made redundant’,
in a majority of Member States (Chart
‘early retirement’, ‘education or train-
47). In countries where the average Chart 48 suggests that the average ing’, ‘looking for children or incapacitat-
duration for older workers (55–64) duration of completed unemploy- ed children’, ‘normal retirement’, ‘other
personal or family responsibilities’, ‘own
is lower than for prime-age workers ment decreases with the level of
illness or disability’, and ‘other reasons’.

Employment in Europe 2009

Chart 49: Distribution of the duration of incomplete unemployment 4.5. Differences

spells by main reasons for leaving (or having lost) employment, between the average
average in period 2006–08 for the EU
duration of CDU and
IDU spells over the

economic cycle

Given the importance of monitor-

ing cyclical developments for policy
purposes, this section follows Corak
and Heisz (1995) in characterising
the cyclical variation of the statistics
used to measure CDU and IDU.(56) As
an example, time-series are plotted
0.00 only for a few selected countries
0–2 months 3–5 months 6–11 months 12–17 months 18–23 months 24–47 months 4 years and longer
A job of limited duration has ended
(Spain, the Netherlands and the UK).
Dismissed or made redundant
Own illness or disability
0.09 As already mentioned in section 4.2,
the average IDU shows that changes
Source: EU LFS, DG EMPL calculations. Variable used: DURUNE. in the composition of unemployment
by duration over the economic cycle
follow a marked counter-clockwise
Chart 50: Distribution of the duration of incomplete unemployment spells
loop (Charts 51–53). This implies that
initiated either at the end of a temporary contract or after dismissal/redundancy,
besides being a biased indicator of
average in period 2006–08 for the EU
completed spells, the IDU statistic is
0.40 also a lagging cyclical indicator.

With respect to monitoring cycli-

0.30 cal developments, Corak and Heisz
(1995) suggest that turning points in
the CDU statistic closely follow the
evolution of cyclical conditions in
the labour market, although there
is a loop in the data, but this time

0–2 months 3–5 months 6–11 months 12–17 months 18–23 months 24–47 months 4 years and longer
A job of limited duration has ended 0.38 0.19 0.17 0.09 0.05 0.08 0.04
Dismissed or made redundant 0.23 0.14 0.17 0.12 0.07 0.17 0.1

(56) Although using significantly shorter time


Chapter 2 Labour flows, transitions and unemployment duration

Chart 51: Average IDU (top) and CDU (bottom) incomplete/completed duration of unemployment (in months)
in Spain against the unemployment rate, period 1992 Q1 to 2008 Q3
98 97 96 95
20 94
00 93

16 01
05 02
14 06
04 03
12 08
0.07 0.09 0.11 0.13 0.15 0.17 0.19 0.21 0.23 0.25 0.27
Unemployment rate

9 02

04 00
07 06
0.08 0.09 0.1 0.11 0.12 0.13 0.14 0.15
Unemployment rate

Source: DG EMPL calculations based on EU LFS data.

Chart 52: Average IDU (top) and CDU (bottom) incomplete/completed duration of unemployment (in months)
in the Netherlands against the unemployment rate, period 1992 Q1 to 2008 Q3
The Netherlands
34 98
32 99 97 96
30 93 95

26 00 07
24 08 05
22 01
02 04
20 03
0.02 0.03 0.04 0.05 0.06 0.07 0.08
Unemployment rate
The Netherlands
11 04

08 05
8 07 06
7 02
6 01 00
0.02 0.03 0.03 0.04 0.04 0.05 0.05
Unemployment rate

Source: DG EMPL calculations based on EU LFS data.

Chart 53: Average IDU (top) and CDU (bottom) incomplete/completed duration of unemployment (in months)
in the UK against the unemployment rate, period 1992 Q1 to 2008 Q3
24 95 94
97 96
22 98

21 99 93
01 07 08
19 04 06 92
18 02
17 05 03
0.04 0.05 0.06 0.07 0.08 0.09 0.10 0.11
Unemployment rate

9.5 UK
05 07

8 02
04 03
01 00
0.047 0.049 0.051 0.053 0.055 0.057 0.059
Unemployment rate

Source: DG EMPL calculations based on EU LFS data.

Employment in Europe 2009

4.6. Survival rates Chart 54: Survival function in unemployment in the largest EU Member States,
in unemployment average in period 2005 Q1 to 2008 Q3

The calculation of statistics for the UK

average CDU spells involves the cal- 0.8
culation of hazard and survival func- FR
tions (see Equation 2 in Box  5). The EU
survival function shows, in effect, 0.6 ES

what proportion of a cohort of DE

people who become unemployed

remains unemployed as time passes.
This section presents a few survival
functions for various breakdowns – 0.2
countries, gender, etc.

The country ranking resulting from 1M 2M 3M 4M 5M 6M 7M 8M 9M 10M 11M 12M 13M 14M 15M 16M 17M 18M 19M 20M 21M 22M 23M 24M
the survival function in unemploy-
ment (Chart 54) is identical to that Source: DG EMPL calculations based on LFS data.
obtained using the CDU statistic
(Chart 45). The survival curve of a Chart 55: Survival function in unemployment in a selected number
country with a higher (lower) value of EU Member States, average in period 2005 Q1 to 2008 Q3
for CDU lies above (below) that of 1.0
a country with lower (higher) CDU.
Amongst the largest EU Member SK

States, Spain and Italy have the low- 0.8 SE

est values for CDU as their survival
rates draw nearer to the x-axis. The
0.6 DK
situation in Spain and Italy is thought
to partly reflect the high incidence
of atypical labour contracts, result- 0.4

ing from the easing of employment

protection legislation for temporary
contracts in recent years.

For illustrative purposes, Chart 55 0.0

presents survival curves for a few 1M 2M 3M 4M 5M 6M 7M 8M 9M 10M 11M 12M 13M 14M 15M 16M 17M 18M 19M 20M 21M 22M 23M 24M

selected Member States. The rela-

Source: DG EMPL calculations based on LFS data.
tively high transition rates from
unemployment and inactivity into Chart 56: Survival function in unemployment in the EU by gender,
employment observed in Denmark average in period 2005 Q1 to 2008 Q3
(see section 3) go together with low
survival rates in unemployment – i.e. 1.0

low CDU. Conversely Hungary and Men

Slovakia show above EU average sur-
vival rates/CDU.

Chart 56 confirms the results shown 0.6

in Chart 46, namely that gender dif-

ferences in CDU/survival rates are
relatively minor.(57)


(57) In the same period (2005 Q1 to 2008

Q3), the long-term unemployment rate 1M 2M 3M 4M 5M 6M 7M 8M 9M 10M 11M 12M 13M 14M 15M 16M 17M 18M 19M 20M 21M 22M 23M 24M
(incomplete duration of at least 12
months) in the EU was 3.7% and 3.1% Source: DG EMPL calculations based on LFS data.
for women and men, respectively.

Chapter 2 Labour flows, transitions and unemployment duration

Chart 57: Survival function in unemployment in the EU by age, The CDU of young workers (15–24) is
average in period 2005 Q1 to 2008 Q3 the lowest of the three age groups
(Chart 57), although young people
have both the highest unemploy-
ment and LTU rates (Table 12). Given
0.8 Prime
the short duration of youth unem-
Old ployment, it is particularly important
to ensure that it is ‘accurately’ meas-
0.6 ured, recognising that unemploy-
ment rate indicators may overstate
the relative welfare impact of youth
0.4 unemployment.(58)

The apparent contradiction between,

on the one hand, the high youth
unemployment rate, and on the
other, the relative low value for the
1M 2M 3M 4M 5M 6M 7M 8M 9M 10M 11M 12M 13M 14M 15M 16M 17M 18M 19M 20M 21M 22M 23M 24M
average duration of completed spells
Source: DG EMPL calculations based on LFS data. can be explained as follows. The
high unemployment rate reflects the
high inflow rate into unemployment,
which is not reflected in the CDU
Table 12: CDU, unemployment and LTU rates, EU average statistic because of the low average
in the period 2005 Q1 to 2008 Q3 duration of unemployment.(59)
15–24 10.3 16.7 4.6
Chart 58 suggests that education is a
25–54 12.2 6.9 3.2 major factor affecting the duration
55–64 14.7 5.8 3.6 of unemployment.

Chart 58: Survival function in unemployment in the EU by education,

average in period 2005 Q1 to 2008 Q3

0.8 Low




(58) The coefficients of correlation (Pearson’s

and Spearman’s rank) for country differ-
0.0 ences between youth and the working
1M 2M 3M 4M 5M 6M 7M 8M 9M 10M 11M 12M 13M 14M 15M 16M 17M 18M 19M 20M 21M 22M 23M 24M
age population (15–64) of CDU and LTU
were not found to be significantly differ-
Source: DG EMPL calculations based on LFS data. ent from zero.

(59) In the steady state, the following identity

is satisfied: . Where µ is the aver-
age duration of unemployment and i the
inflow rate into unemployment. A high
inflow rate into unemployment does not
necessarily imply a high unemployment
rate, because it can be offset by a lower
average duration of unemployment (i.e.
a high outflow rate).

Employment in Europe 2009

4.7. Do CDU statistics Table 13: Average completed duration of unemployment, unemployment rate
and LTU rates convey and three LTU rates (more than 6, 12 and 24 months of incomplete duration),
country averages in the period 2005 Q1 to 2008 Q3
the same type of
information? AT
BE 9.6 7.8 5.1 3.9 2.6
The correlation between the com-
BG 10.1 7.9 5.6 4.5 3.2
pleted duration of unemployment
CY 6.8 4.5 1.7 0.9 0.3
(CDU) and the unemployment rate is
CZ 10.2 6.2 4.5 3.3 2.0
significantly different from zero. The DE 12.1 9.5 6.5 5.1 3.5
correlation between CDU and various DK 6.6 4.0 1.3 0.8 0.3
LTU rates is both significantly differ- EE 8.6 6.1 3.6 2.8 1.7
ent from zero and high (Table 13). ES 6.4 9.4 3.4 2.0 0.9
FI 7.5 7.4 2.8 1.7 0.9
The scatter plot shown in Chart 59 FR 8.4 8.3 4.8 3.3 1.7
also suggests that the LTU rate and GR 10.4 8.8 6.0 4.5 2.6
the CDU essentially convey the same HU 13.9 7.5 5.1 3.4 1.7

information, although there are a IE 9.5 4.9 2.3 1.5 0.8

IT 7.9 6.9 4.2 3.3 2.0
few exceptions or country outliers.(60)
LT 8.1 6.1 3.4 2.4 1.5
LU 6.9 4.6 2.3 1.3 0.4

4.8. Methods used LV

during the previous four NL 8.2 3.7 2.0 1.4 0.8

weeks to find work PL

RO 9.7 7.0 4.6 3.6 1.9
To be recorded as unemployed SE 6.6 6.8 1.9 0.9 0.3
according to the ILO definition used SI 9.1 5.5 3.5 2.6 1.5
in the EU LFS, an individual must have SK 12.2 12.6 10.7 9.2 7.0
searched actively for a job during the UK 8.9 5.3 2.1 1.2 0.6

last four weeks. The EU LFS also col- EU 11.9 7.9 4.6 3.4 2.0
Correlation coefficient
lects information on the methods
CDU Pearson a) 0.45 0.62 0.62 0.61
used to find work, and some nation-
CDU Spearman a) 0.48 0.76 0.76 0.75
al statistic institutes additionally ask
workers to identify the job-finding Source: DG EMPL calculations based on EU LFS data.
method that produced results. Notes: (a) Values in bold are significantly different from 0 with a significance level
alpha=0.05. EU-27, excluding Portugal.
The EU LFS lists a total of 13 methods,
with the unemployed typically using Chart 59: Standardised values of LTU rate and CDU, country averages
more than one method. Charts 60–64 in period 2005 Q1 to 2008 Q3
present information on the propor-
tion of unemployed persons using SK
a few selected methods – namely
‘Contacted public employment office 3

to find work’, ‘Contacted private

employment agency to find work’, 2
‘Applied to employers directly’, ‘Asked

friends, relatives, trade unions, etc.’, 1


and ‘Inserted, answered or studied BE GR

advertisements in newspapers’.(61) IT
0 RO
(60) Note that measures of unemployment MT LT
duration or LTU rates are essentially meas- AT FI NL IE
ures of ‘dispersion’ of unemployment, ver- -1 LU
sus the unemployment rate which can be
seen as a measure of ‘localisation’ or of the
average incidence of unemployment. -2
-3 -2 -1 0 1 2 3
(61) The latter combines EU LFS’s search CDU_standardised
methods E and F, respectively, ‘Inserted
or answered advertisements in newspa-
pers or journals’ and ‘Studied advertise- Source: DG EMPL calculations based on EU LFS data.
ments in newspapers and journals’.

Chapter 2 Labour flows, transitions and unemployment duration

Chart 60: Country breakdown of the proportion of unemployed persons using

different job search methods, average in period 2004–07

Inserted, answered or studied Asked friends, relatives, Applied to employers directly

1.0 advertisements in newspapers trade unions, etc.
Contacted private employment Contacted public employment
agency to find work office to find work






Source: EU LFS, DG EMPL calculations.

Chart 61: Gender breakdown of the proportion of unemployed persons using

different job search methods, average in period 2004–07

Men Women








Contacted public employment Contacted private employment Applied to employers directly Asked friends, relatives, Inserted, answered or studied
office to find work agency to find work trade unions, etc. advertisements in newspapers

Source: EU LFS, DG EMPL calculations.

Chart 62: Age breakdown of the proportion of unemployed persons using

different job search methods, average in period 2004–07

15–24 25–54 55–64









Contacted public employment Contacted private employment Applied to employers directly Asked friends, relatives, Inserted, answered or studied
office to find work agency to find work trade unions, etc. advertisements in newspapers

Source: EU LFS, DG EMPL calculations.

Employment in Europe 2009

Chart 63: Breakdown by level of education of the proportion of unemployed persons using
different job search methods, average in period 2004–07

Low Medium High









Contacted public employment Contacted private employment Applied to employers directly Asked friends, relatives, Inserted, answered or studied
office to find work agency to find work trade unions, etc. advertisements in newspapers

Source: EU LFS, DG EMPL calculations.

Chart 64: Breakdown by unemployment duration of the proportion of unemployed persons

using different job search methods, average in period 2004–07









Contacted public employment Contacted private employment Applied to employers directly Asked friends, relatives, Inserted, answered or studied
office to find work agency to find work trade unions, etc. advertisements in newspapers

Source: EU LFS, DG EMPL calculations.

On average in the EU, the two • Men and women use various meth- • Overall, some degree of ‘dura-
most frequently used methods of ods in much the same proportions tion dependence’ seems to exist in
job search are ‘Contacted public (Chart 61) the choice of search method with
employment office to find work’ those in LTU using some methods
and ‘Inserted, answered or stud- • Older workers tend to use pub- less frequently, especially ‘applied
ied advertisements in newspapers’. lic employment services more and to employers directly’ and ‘asked
Nonetheless, considerable varia- other methods less than other age friends, relatives, trade unions,
tion remains across countries, with groups (Chart 62) etc.’ (Chart 64), suggesting some
Italy and Spain relying more on discouragement.
direct contacts with employers and • Unemployed persons with high
­personal networks. levels of education tend to make
more use of private employment
The breakdowns provided in Charts agencies, apply directly to employ-
61–64 suggest the following about ers, and insert advertisements/
‘preferences’ for search methods: study the press (Chart 63).

Chapter 2 Labour flows, transitions and unemployment duration

4.9. Use of c­ ross-sectional data (i.e. EU LFS) is not experience of individuals, are more
able to monitor important aspects resilient to short term interruptions
longitudinal data
of unemployment experiences, such in unemployment spells. However,
to evaluate the as the multiplicity of unemploy- one should acknowledge that the
incidence of LTU and ment spells. In addition, they lack relative small sample size of EU SILC
the detailed information needed to raises issues regarding the accuracy/
the recurrence of evaluate the impact on the measure- representativeness of its results.
unemployment spells ment of unemployment (and of its
duration) of choosing between alter- To calculate measures of incidence
This section reports evidence on native reference periods. Measures and duration of LTU, this section uses
the incidence of LTU and the recur- of unemployment based on longi- detailed calendar information on the
rence (or multiplicity) of unemploy- tudinal/calendar data, because they labour market status for 14 European
ment spells. Results suggest that take into consideration the complete countries using micro-data from the
users’ EU SILC database.(62) Longitu-
Table 14: Percentage of all persons unemployed in July 2005 who experienced dinal data is also utilised to calculate
at least 12 months of unemployment as measured by: the distribution of the duration of
total unemployment by the number
Incomplete duration of Completed duration of Total unemployment in a
of spells in unemployment.
the current spell the current spell period of 36 months b)
AT 35.1 70.6 81.6
BE 69.3 90.4 94.5 Following the OECD (2002), Table 14
DK 20.1 60.3 86.3 shows the proportion of all individu-
EE 59.5 81.0 85.8 als unemployed at a given point in
ES 27.0 73.3 86.3 time – July(63) 2005 – who experienced
FI 39.1 61.2 73.6 12 months or more of unemploy-
FR 54.9 84.9 90.5 ment, measured using alternative
GR 44.8 78.0 89.1 reference periods.
IE 52.2 75.8 79.4
IT 45.5 84.3 91.5
As noted earlier, the use of calendar/
LU 29.0 100.0 100.0
monthly data from EU SILC is likely to
NO 7.4 54.7 57.3
be associated with a relatively small
PT 50.6 85.7 91.5
SE 22.4 37.0 55.3
sampling bias compared with the
Average a) 39.8 74.1 83.1 length bias (see Box 5). Therefore,
in this case, statistics of completed
Source: UDB EU SILC. DG EMPL calculations. unemployment duration are likely to
Notes: a) Non-weighted average. b) Except Denmark, Greece, Luxembourg and Nor- exceed those based on incomplete
way for which data covers 48 months. spells, contrary to calculations using
quarterly EU LFS data.(64)
Table 15: Average months of unemployment experienced
by persons unemployed in July 2005 as measured by:

Incomplete duration Completed duration

(62) Data used cover the period 2004–06 for
of the current spell of the current spell
the countries: Austria, Belgium, Esto-
AT 10.0 19.0 nia, Spain, Finland, France, Ireland, Italy,
BE 14.8 27.3 Portugal and Sweden; and the period
DK 7.7 16.2 2003–06 for the countries: Denmark,
EE 13.2 23.2 Greece, Luxembourg and Norway.

ES 9.1 16.7 (63) July 2005 was chosen as reference point

FI 10.0 17.9 because it is the middle of the period
covered for a majority of countries from
FR 12.3 23.7
January 2004 to December 2006.
GR 12.6 22.1
IE 12.0 21.8 (64) In principle, the higher the frequency
of observations the lower the sampling
IT 11.8 22.5
bias in the measurement of unemploy-
LU 11.7 21.8 ment duration – i.e. the tendency for
NO 6.2 11.4 unemployment of shorter/longer dura-
PT 11.8 22.2 tion to be under/over-represented in
the sample. Although depending on the
SE 7.3 11.3
exact details of the data collection meth-
Average a) 10.7 19.8 odology used (or sampling design), mov-
ing from quarterly (EU LFS) to monthly/
Source: UDB EU SILC. DG EMPL calculations. calendar data (EU SILC) is likely to signifi-
Note: a) Non-weighted average. cantly reduce the sampling bias involved
in calculating unemployment duration.

Employment in Europe 2009

Table 16: Average duration of total unemployment by number of spells (in months), period January 2003/04 to December 2006

Average dura-
Average dura- Number Average dura- Number Number Total Average Number
tion of 3 or
tion of 1 spell of cases tion of 2 spells of cases of cases duration of cases
more spells
AT 22.0 56 9.5 18 3.6 6 18.0 80
BE 30.0 159 11.7 27 5.8 10 26.4 196
DK 19.8 47 10.0 12 12.8 2 16.3 61
EE 25.1 284 11.1 56 7.1 13 22.3 353
ES 18.9 543 9.9 124 5.9 107 15.6 774
FI 23.3 122 8.7 69 5.5 36 16.3 227
FR 26.8 492 10.6 103 5.8 50 22.6 645
GR 26.6 163 12.1 47 5.8 32 21.0 242
IE 26.8 59 5.3 9 4.9 7 21.4 75
IT 25.6 772 9.6 185 4.8 74 21.5 1 031
LU 22.4 139 12.0 15 21.2 154
NO 11.3 43 7.8 9 6.7 1 10.7 53
PT 24.1 180 11.8 36 6.5 10 21.4 226
SE 11.7 57 9.2 33 5.5 28 9.6 118
Average/Sum a) 22.5 3 116 9.9 743 6.2 376 18.9 4 235

Source: UDB EU SILC. DG EMPL calculations.

Note: a) Non-weighted average.

data (Chart 45). Use of calendar/

Chart 65: Average duration of total unemployment by number monthly data potentially reduces the
of spells in unemployment (months) sampling bias (Box 5), although EU
SILC data is released only once a
year and involves smaller samples
Average duration of 3 or more spells
than EU LFS data, which is available
Average duration of 2 spells
Average duration of 1 spell
The monthly calendar of labour mar-
ket status in EU SILC enables the
average (completed) unemployment

15 duration to be calculated by number

of spells in unemployment (Table 16
10 and Chart 65).(66)

5 Multiple spells in unemployment are

a relatively common experience. In
the sample used, about one quarter
of all unemployed persons experi-
Source: UBD EU SILC. DG EMPL calculations. enced more than one spell of unem-
Notes: Countries ranked in descending order of totals (not shown). ployment. In addition, the distribu-
tion of average duration of total
unemployment by spell is strongly
Using the conventional measure of to December 2006(65), more than four skewed; therefore, statistics based on
LTU, on average some 40% of unem- out of each five persons in unemploy- the average duration misrepresent
ployed persons had already been ment (in July 2005) went on to spend the situation of all groups. This sug-
unemployed for 12 months and 12 or more months in unemployment gests that the calculation of unem-
more. However, about 35% of those over a 3–4-year period. ployment duration statistics (based
without a job for less than 12 months on the calendar variables of longi-
(in July 2005) ended up experiencing It should be noted that the CDU sta- tudinal data) be used to comple-
a completed spell of unemployment tistic calculated using longitudinal ment the information on unemploy-
lasting 12 months and more. On data (Table 15) is considerable higher ment duration calculated using LFS
this basis, nearly 75% were long- than the one based on cross-sectional ­cross-sectional data.
term unemployed. Counting the
total unemployment that occurred (65) From January 2003 to December 2006 for (66) Note that this point calculates the aver-
the countries: Denmark, Greece, Luxem- age duration of total unemployment,
in the period from January 2003/04
bourg and Norway. not the average duration of LTU.

Chapter 2 Labour flows, transitions and unemployment duration

4.10. Brief description Box 6: Brief reference to profiling techniques

of methods to estimate
unemployment duration Profiling mainly uses econometric techniques (sometimes in combination with the judge-
ment of case handlers) to identify those most at-risk of becoming long-term unemployed
from among the newly unemployed. Profiling is more common in Anglo-Saxon countries.(1)
The average duration of unemploy- Those profiled are referred to various active labour market policies, such as counselling
ment is a key element in the evaluation and job-search assistance, where a counsellor then assists the jobseeker in tailoring re-
of the labour market experience of the employment services to their specific needs.
unemployed and of the welfare impli-
cations of those experiences. There are The aim of profiling is to facilitate the allocation (or the rationing) of scarce staff and finan-
cial resources (e.g. the use of limited public employment services and training slots) to those
essentially three methods to estimate
most in need and might be justified even if many non-referred workers end up in LTU as
the average duration of completed well.(2) According to the OECD (1999):
unemployment spells: non-parametric,
semi-parametric and parametric. evidence suggests that operational profiling systems are subject to a varying degree of
inaccuracy and misprediction. However, possible deadweight losses arising from early
interventions in favour of wrongly profiled at-risk workers must be weighted against the
Non-parametric methods have
costs of delaying assistance until jobseekers are actually ‘scarred’ by the experience of LTU,
already been extensively used in risking losing human capital and employability.
this chapter. A well-known problem
with them, however, is that they
frequently introduce an undesirable
level of ‘noise’ in the empirical esti- (1) OECD (1998) reviewed experiences with different profiling approaches in Australia,
mates of hazard/survival functions. Canada, the UK and the USA. It should be noted that there are strong differences of view
Sider (1985) states that the estima- about the relevance and reliability of formal profiling methods, and how central a role
profiling can play in making active labour market programmes more effective.
tion of average completed unem-
ployment spells from raw data on (2) If jobseekers at risk of LTU are correctly identified and offered appropriate active labour
market policies, resources will ultimately be freed up to help those in LTU, assuming that
incomplete unemployment spells is doing the latter is more expensive than the former.
particularly hazardous given the mul-
tiple problems and irregularities in applied to the estimation of hazard The notion of duration dependence
the data.(67) While acknowledging functions – i.e. the conditional prob- is central. It measures how the prob-
the possible need to use smoothing ability of exiting a particular state. ability of leaving unemployment (the
techniques, Baker and Trivedi (1985) Current practice consists of postu- hazard rate) varies in line with the
in a comparative study favour the use lating an a priori form for the haz- time already spent in unemployment.
of non-parametric over parametric ard function, depending on a limited If the hazard rate increases (decreas-
methods, because of their simplicity set of parameters; hence this strand es) with the amount of time already
and the fact that there is no need to of analysis is called the parametric spent unemployed, there is positive
make assumptions on the statistical approach. The hazard function is gen- (negative) duration dependence.
distribution of duration models. erally estimated as a function of the
duration of unemployment and a set Parametric/regression methods have
Semi-parametric methods establish a of explanatory variables that have an the potential advantage to identify
compromise between a strictly non- impact on both labour market policies major polices and risk factors associat-
parametric approach and a paramet- (e.g. unemployment benefit system) ed with unemployment duration. This
ric one. Unemployment durations are and the characteristics of individuals is particularly relevant in assessing the
grouped together into a relatively (e.g. gender, education, family sta- impact of unemployment benefit sys-
small number of time intervals. A spe- tus, age). The Weibull distribution tems on both average duration and
cific (piecewise constant) hazard func- is the simplest functional form in duration dependence, and in identify-
tion is estimated for each interval. which it is possible to distinguish aver- ing the groups that are most vulnera-
age duration from duration depend- ble to long unemployment spells, and
Regression analyses using duration ence, although the latter is taken into which may require targeted policies
models try to explain the duration of account in a monotonic fashion.(68) and/or profiling (Box 6).
a given state (e.g. the duration of an
unemployment spell). This is normally (68) The Weibull function assumes that the However, implementation of regres-
hazard rate out of unemployment h(t)
sion methods is associated with con-
is governed by: . There are
(67) The full schedule of in-progress spells is two parameters µ and α. µ determines siderable technical difficulties. A
dominated by a pattern of spikes that the average duration of unemploy- major problem is that unobserved
reflect response bias among individuals ment spells. If α=1, there is no duration
differences between individuals in
in the sample. Local modes occur corre- dependence and the hazard rate is
sponding roughly to monthly, quarterly, equal to µ at all durations. If α>1 (α<1) the data set (i.e. differences that
half-year, and yearly points in the sched- there is positive (negative) duration are not easily documented like age,
ule. This pattern must be accounted for dependence (Cahuc and Zylberberg,
gender, education level) can cause a
and smoothed… 2004).

Employment in Europe 2009

Box 7: Some elements on job search theory 4.11. Elasticities of

unemployment duration
Job search theory predicts that, under certain conditions, both higher levels and longer to benefits and duration
periods of unemployment benefits lower the hazard rate of leaving unemployment, and
therefore result in higher average unemployment duration (Bover and al., 2002). dependence of hazard
In its simplest formulation of job search theory(1), an unemployed worker accepts all job rates
offers above the reservation wage ξ, which is the lowest wage at which a worker will accept
a job offer. The reservation wage solves the equation: Most available empirical studies find
positive (but modest) elasticities of
the average duration of unemploy-
(Equation 1)
ment to the level of unemployment
benefits, in line with theoretical pre-
dictions (Devine and Keifer, 1991).
where b is the value of the unemployment benefit net of the costs of looking for a job, p is
Conversely, empirical results show
the discount rate, F(w) is the distribution of available wages, λ the arrival rate of job offers,
that the duration of unemployment
and c the intensity of job search.
is more sensitive to the extension of
Equation 1 can be interpreted as follows. The reservation wage just covers unemployment the entitlement period to benefits,
benefits (net of search costs) plus the expected gain from waiting for a better offer. The than it is to an increase in the level
optimal strategy of an unemployed person is to accept (reject) any job offer above (below) of benefits (Katz and Meyer, 1988),
the reservation wage ξ. and that unemployment hazard rates
The hazard rate - h(t) – can be written as: tend to rise in the period immedi-
(Equation 2a) ately preceding the expiry of benefit
entitlement (Meyer, 1990; Dormont
While the average duration of unemployment - µ(t) – is given by: et al., 2001) (see Box 7 for a synthetic
(Equation 2b) presentation of search theory).

An increase in the unemployment benefit raises the reservation wage (and/or reduces the In a survey on unemployment dura-
intensity of the job search). This translates into a reduction in the probability of a job offer tion studies for the Portuguese econ-
being accepted, , leading to a fall in the hazard rate and therefore to an increase omy, Portugal (2008) lists a number
in unemployment duration.
of personal characteristics of the
Equations 2a and 2b can be used to identify the following possible sources of duration unemployed which are statistically
dependence (Machin and Manning, 1999): i) job offer rate; ii) search intensity; iii) wage offer significant in predicting hazard rates/
distribution; and iv) reservation wage. unemployment duration. For exam-
ple, older workers have lower (higher)
An important caveat is that the disincentive effects described above regarding the rise in
hazard rates (unemployment dura-
benefits concern only those unemployed persons who are entitled to unemployment ben-
tion) and hazard rates are higher(70)
efits. For unemployed persons not entitled to benefits, their increase might raise hazard
rates as employment becomes more valuable on account of enhanced future rights to for those who are married, educated,
unemployment benefits. and have work experience.

(1) Portugal (2008). For the classical framework of job search theory, see Mortensen (1977). A number of empirical studies find
that hazard rates decrease with the
systematic bias in estimation, favour- The methods employed to disen- duration of unemployment(71), but
ing findings of negative duration tangle ‘true’ duration dependence the scale of this decline is limited,
dependence – i.e. that exit rates out from unobserved heterogeneity, especially after controlling for indi-
of unemployment fall with duration. thereby correcting for the estima- vidual heterogeneity. ‘True’ negative
In fact, such unobserved heterogene- tion bias induced by the latter, are duration dependence has important
ity always leads to negative duration not entirely convincing in terms of policy implications, suggesting that
dependence. The reason for this bias their underlying economic-theoret- the employability of jobless per-
stems from the fact that, as time ical rationale being largely deter- sons deteriorates with the duration
passes, the pool of the unemployed mined by mathematical/computa- of unemployment itself – i.e. after
is progressively made up of the less tional convenience (Machin and controlling for both policies and
employable individuals.(69) Manning, 1999). ­observable personal characteristics.

(70) Unemployment duration is lower.

(69) Individuals with (unobserved) character-
istics that hinder their exit from unem- (71) i.e. there is negative duration depend-
ployment. ence.

Chapter 2 Labour flows, transitions and unemployment duration

Table 17: Average duration of unemployment and duration dependence It is a particularly complex issue, part-
ly because EPL interacts with other
Average duration of unemployment labour market policies, such as wage
Duration dependence (α)
(in months)
bargaining and unemployment ben-
1960s-1970s 1980s-1990s 1960s-1970s 1980s-1990s
efits. According to flexicurity princi-
6.2 15.1 0.39 0.58
Belgium ples, the design of an unemployment
(0.07) (0.06) (0.002) (0.002)
3.6 12.7 0.54 0.93 benefit system has to be considered
(0.01) (0.01) (0.001) (0.001) alongside other labour market insti-
4.2 5.3 0.86 0.58 tutional arrangements (e.g. EPL and
(0.01) (0.01) (0.001) (0.001)
ALMPs) as well as the taxation sys-
2.4 13.7 0.68 0.66
Netherlands tem. Full treatment of these issues is
(0.01) (0.04) (0.002) (0.002)
2.3 17.7 0.58 0.91 outside the scope of this chapter.
(0.37) (0.17) (0.06) (0.01)
0.8 6.5 0.35 0.57 In a nutshell, theoretical arguments
United Kingdom
(0.14) (0.36) (0.02) (0.02)
suggest that high EPL is likely to
1.2 6.5 0.72 0.79
Australia depress labour demand because fir-
(0.22) (0.56) (0.10) (0.10)
1.1 1.2 0.61 0.52 ing costs increase total production
United States
(0.04) (0.03) (0.01) (0.01) costs, reducing wages that firms can
offer. EPL tends also to boost work-
Standard errors are in parenthesis.
ers’ wage demands because high
Source: Machin and Manning (1999, table 4).
EPL, particularly when combined
with a generous unemployment ben-
In these circumstances, transitory Machin and Manning’s (1999) esti- efit system, boosts their bargaining
economic shocks that increase unem- mates do not control for country-spe- power. Therefore, high EPL on the
ployment can lead to a more perma- cific policies and the heterogeneity one hand tends to shift downwards
nent deterioration in the functioning of individuals. Hence the results are labour demand, while on the other it
of the labour market as the reduction likely to be biased towards finding tends to move upwards labour sup-
in hazard rates, and the consequence negative duration dependence. How- ply, overall resulting in higher equi-
rise in the average duration of unem- ever, a number of papers suggest librium unemployment and unem-
ployment, become entrenched. Ade- that negative duration dependence ployment duration.
quate policy responses may involve, persists even after controlling for the
inter alia, better targeting of active intrinsic characteristics of workers (van However, empirical results suggest
labour market policy (ALMP) spend- den Berg and van Ours; 1994, 1996). that strict EPL does not have signifi-
ing, possibly making use of profiling cant effects on overall labour market
techniques, in an effort to maximise variables, although it is seen to dete-
the efficiency of spending. 4.12. Labour market riorate labour market prospects of
institutions and particularly vulnerable groups, such
Using data spanning four decades as young and older workers.
(1960s to 1990s), Machin and Man- unemployment duration
ning (1999) estimate Weibull dura- As regards the impact of wage-
tion models for a group of OECD In an often cited paper, Blanchard and bargaining structures on unemploy-
countries (Table 17). Although aver- Portugal (2001) compare Portuguese ment, Calmfors and Driffill (1988)
age duration of unemployment and and US labour markets, highlight- argue that there is a non-linear
the incidence of LTU increased from ing major differences. Using micro- relationship between the degree of
the 1960s–70s to 1980s–90s, this data, the authors find significantly centralisation/coordination of collec-
outcome did not result from any lower labour flows and higher unem- tive wage bargaining in an economy
marked change in duration depend- ployment duration in Portugal com- and the level of unemployment(73),
ence which remained negative in pared to the USA, which the authors roughly postulating an ‘inverted U’
all the countries considered.(72) The attribute to stricter EPL in the former. relationship: with low and high lev-
authors argue that the rise in aver- The statistical evidence also suggests els of centralisation/coordination in
age unemployment duration reflects that there is a higher incidence of LTU wage bargaining being preferable
a downward shift in hazard rates at in Portugal, and that unemployment
all durations, without changing the duration tends to be higher in coun- (73) The rationale for this is related to Olson’s
slope of the curve. tries with strict EPL (see Box 8). idea (1982) – namely that organised inter-
ests can be most harmful when they do
not internalise a significant proportion
(72) Note that α<1 means negative duration The impact of EPL on labour market of the costs they impose on society, but
dependence. However, negative dura- outcomes has been the subject of become less harmful as their interests
tion dependence seems to have declined, become encompassing enough to inter-
considerable research (OECD, 2006).
especially in France and Spain. nalise the costs they impose on society.

Employment in Europe 2009

to an intermediate one. Market pres- consumption patterns. They also sub- However, a tax-benefit system can
sure tends to discipline outcomes sidise job search, giving sufficient time turn out to be unnecessarily inef-
when wages are negotiated at firm for an unemployed person to search ficient, resulting in lower tran-
level, being compatible with favour- for and accept an adequate job offer.(74) sitions into employment, higher
able overall unemployment develop- Using a general equilibrium model, unemployment duration and over-
ments. Unions that take into con- Acemoglu and Shimer (1999) show that all lack of incentives for vocational
sideration the wider effects of their economies with moderate unemploy- training (OECD, 2006, chapter 3). (75)
actions when bargaining at national ment benefits can have higher output For a given level of income sup-
level also secure favourable overall and welfare than those with lower port, higher benefit levels cou-
unemployment outcomes. However, levels of unemployment insurance, pled with shorter eligibility periods
unions are not seen to internalise because unemployment insurance are preferable in order to facili-
significant amounts of the costs that encourages workers to look for more tate the return to employment
their decisions impose on society productive, albeit more vulnerable jobs. and improve the fluidity of labour
when wages are negotiated at sec- Although too generous unemployment markets (Cahuc and Zylberberg,
toral level, resulting in unfavourable benefit systems can increase unemploy- 2004). (76) Alternatively, the adverse
overall unemployment outcomes. ment and lengthen its duration, mod- employment effects of long ben-
erate unemployment benefits can raise efit duration could be (at least
The aims of unemployment ben- the quality of job matches and labour partly) offset by the introduction
efit systems are basically to protect productivity, outweighing the effects of stricter job search obligations,
workers against cyclical fluctuations of the rise in unemployment and its entailing financial penalties in case
in employment, and thereby smooth­ duration on output and welfare. of ­non-compliance. (77)

(75) OECD finds moderately sized labour mar-

ket effects to changes in financial (dis)
incentives. The marginal effective tax
rate (METR), which is a comprehensive
indicator of the leakage between gross
earnings gains and the resulting rise in
disposable income, provides a useful
measure of these disincentives (Carone
et al., 2004). Some calculations carried
out by the OECD suggest that a 20%
reduction of METRs, roughly represent-
ing the size of some of the more ambi-
tious reforms implemented in recent
years, imply a rise in the one-year transi-
tion probability for moving from unem-
ployment to employment from 45% to
49%. However, large effects are found
for the unemployed with a working part-
ner, whose re-employment probability is
estimated to increase by 7 percentage
points, from 51% to nearly 58%. The
evidence for the effect of tax disincen-
tives on transitions from inactive to work
is more mixed, with large effects found
only for single women: for this group,
the probability to move from inactivity
to work would increase by almost 13%.

(76) A number of cross-country panel data

studies find that the impact of unem-
ployment benefits on aggregate labour
market outcomes is significant, with ben-
efit duration having a more detrimen-
tal impact than high replacement rates
(OECD, 2006, chapter 3).

(77) Several recent empirical studies conclude

that labour supply disincentives from
generous unemployment benefits can be
offset, at least to a significant degree, by
(74) Obviously, together with other elements benefit administration practices that use
of tax-benefit systems, unemployment financial sanctions (i.e. benefit cuts) to
benefits contribute to the overarching enforce an obligation to actively search for
objective of social cohesion and the work and to accept reasonable job offers
reduction of poverty risks. (Boone et al., 2004; Hasselpflug, 2005).

Chapter 2 Labour flows, transitions and unemployment duration

Box 8: Cross-country variation in LTU

This box presents a simple cross-country regression of the incidence of LTU(78) on a number of labour market ‘institutions’. Averages for
22 countries in the period 2000/01 to 2006/07 are used. The list of variables is the following:
Dependent variable:
IR12: lncidence of long-term unemployment of 12 months and more
Period: 2000-2007
Source: LFS and OECD
Explanatory variables:
UT: Unemployment trap: Marginal effective tax rate for an unemployed person (67% of average wage, one-earner couple with 2 children)
NRR6: Net replacement rate after 6 months - 1 earner 2 children, 67% of average wage
NRR12: Net replacement rate after 12 months - 1 earner 2 children, 67% of the average wage
NRR60: Net replacement rate after 60 months - 1 earner 2 children, 67% the average wage
NRR12_6 = NRR12 / NRR6
NRR60_6 = NRR60 / NRR6
Period: 2001-2007
Source: DG ECFIN (2009), ‘Recent reforms of tax and benefit systems in the framework of flexicurity’
EPL: Employment protection legislation indicator
Period: 2003
Source: OECD
ALMP: Active labour market policies in % of GDP
PLMP: Passive labour market policies in % of GDP
Period: 2000-2006
Source: LFS and OECD
SAR: Strictness in availability rules that unemployed must fulfill in order to be entitled to unemployment benefits
Period: 2004
Source: Søren Hasselpflug (2005), ‘Availability criteria in 25 countries’, wp 12/2005 (Finansministeriet)
Countries: BE, DK, DE, GR, ES, FR, IE, IT, NL, AT, PT, FI, SE, UK, CZ, EE, HU, LT, PL, SK, SI, US

The set of explanatory variables includes:

• an unemployment trap and a number of net replacement rate indicators calculated by Carone et al. (2009) for an analysis of recent
reforms of tax-benefit systems, as well as an assessment of their impact on financial incentives to work and on labour supply
• OECD’s (2004) EPL indicator
• an indicator on the strictness of unemployment benefit entitlement rules (Hasselpflung, 2005).

Given the high multicollinearity, a principal component analysis (PCA) was run on the 10 (potential) explanatory variables. The country scores
of the first three dimensions were used as explanatory variables in an ordinary least squares (OLS) regression where the incidence rate of LTU
(12 months and more) is the dependent variable.

Table 18: Percentage of variance after Varimax rotation The first factor summarises the ‘financial incentives’ on labour market
supply of tax-benefit systems. The second factor is highly correlated with
Percentage of variance after Varimax rotation: spending in labour market policies (LMP). The third factor is picking up
D1 D2 D3
the effects of EPL.
Variability (%) 53.2 24.4 13.2
Cumulative % 53.2 77.6 90.8
Regressing IR12 on the three principal components gives:
Factor loadings after Varimax rotation:
D1 D2 D3
UT 0.78 0.45 0.26 Table 19: OLS regression – standardised coefficients
NRR6 0.90 0.28 0.17
NRR12 0.97 0.21 0.05 OLS regression - standardised coefficients
NRR60 0.92 0.25 -0.18 Source Value Standard error t Pr > |t|
NRR12_6 0.91 -0.09 -0.14 D1: Financial Incentives -0.189 0.189 -0.997 0.332
NRR60_6 0.84 0.02 -0.39 D2: LMP -0.384 0.189 -2.031 0.057
EPL -0.06 0.12 0.94 D3: EPL 0.416 0.189 2.202 0.041
ALMP 0.13 0.96 0.06
PLMP 0.20 0.93 0.09 Source: DG EMPL calculations.
SAR a) 0.10 -0.03 0.01
Results suggest the following tentative interpretation. Strict EPL tends to
Source:DG EMPL calculations. raise the incidence of LTU, while spending on LMPs tends to reduce it. The
Note: (a) Supplementary variable. Figures in bold signify impact of financial incentives linked to tax-benefit systems is not found to
factor loadings larger than 0.5 in absolute value. be significantly different from zero.

Flexicurity principles make a strong case for protecting/supporting workers, i.e. their transitions, instead of particular jobs. The econometric
result above intuitively supports this approach. Most flexicurity packages include reductions in job protection (EPL) in exchange for greater
support of workers’ job-to-job transitions by increasing spending on both active and passive LMPs. According to the regression estimates
presented above, such a strategy is likely to contribute to a reduction in LTU.

(78) The LTU statistic used is based on the incomplete duration of unemployment, as a high number of missing quarterly observations prevents
use of a statistic based on the completed duration of unemployment.

Employment in Europe 2009

4.13. The consequences Box 9: In the current economic situation, what are the prospects for LTU?
of LTU
The European Commission publishes biannual economic forecasts (e.g. DG ECFIN, 2009),
including forecasts of unemployment rates, but they do not provide details of the structure
A distinction is commonly made in of unemployment by duration. Given the special problems associated with LTU, this box
the economic literature between the attempts to estimate its evolution in the period up to 2010, based on the latest European
effects of the experience of being Commission spring 2009 economic forecast.
unemployed on the employment
capacity of individuals, and the more Using quarterly data, three vector autoregressive systems (VAR) equations are estimated. The
general effects of unemployment on endogenous variables are: real GDP(1), the incidence rate of LTU(2), and the unemployment rate. The
VAR systems do not include exogenous variables (only a constant term).(3) In principle, data avail-
the economy.
ability allowed for the estimation of 16 VAR systems, covering the following EU Member States:
Austria, Belgium, the Czech Republic, Denmark, Estonia, Finland, France, Italy, Lithuania, Latvia,
As regards the economy as a whole, the Netherlands, Poland, Portugal, Slovenia, Slovakia and the UK.(4) The lag length of VAR systems
several labour market theories(79) varies country by country depending on statistical criteria/tests on the optimal lag order and stability
suggest that the long-term unem- requirements – namely that the inverse roots of the characteristic polynomial lie inside the unit root.
ployed put upward pressure on In the end, models were retained for 11 EU Member States: Belgium, the Czech Republic, Denmark,
wages because of their relatively low Estonia, Italy, Latvia, the Netherlands, Poland, Portugal, Slovenia, Slovakia and the UK.(5) In 2008,
search effectiveness. As regards the these 11 EU Member States represented about 40% of the EU labour force.
impact on the unemployed them- The estimation period of VAR systems covers the whole period for which data is available
selves, the consequences are seen as (up to 2008 Q4). The model is simulated outside of the estimation range between 2009 Q1
multiple. LTU is often associated with and 2012 Q4. In 2009 and 2010, GDP and unemployment rate variables are ‘forced’ to equal
a high concentration of unemploy- DG ECFIN’s spring 2009 economic forecast.(6) Therefore, the only variable free to adjust
ment across a relatively few number according to the VAR dynamics is the incidence rate of LTU (or equivalently the LTU rate).
of individuals, who move back and
As could be anticipated by the definition of LTU, the incidence of LTU (or the LTU rate) lags
forth between unemployment and
the unemployment rate (Chart 66).
temporary jobs. LTU is an impor-
tant risk factor of poverty among Chart 66: Unemployment and LTU rates for an aggregate of 11 EU Member
the working-age population. The lit- States, period 2003 Q1 to 2010 Q4 (%)
erature suggests (e.g. Akerlof and 10 5
Main, 1980; Machin and Manning,
1999; OECD, 2002) that a more accu-
rate indicator to assess the poverty/ 9

inequality risks associated with LTU 4

might be a concentration measures

of unemployment based on longitu-


dinal data.(80) It should be noted that

LTU is also associated with deteriorat- 3

ing physical and mental health, and 7

an increased propensity to engage in UR

shadow economy or illegal activities.
6 2
Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4
2003 2004 2005 2006 2007 2008 2009 2010
Given the potential relevance of LTU,
Source: DG EMPL calculations.
over and above that of unemploy-
ment, Box 9 presents some estimates
of the incidence of LTU and of the (1) The first difference of the log of real GDP. Quarterly real GDP values are seasonally and
LTUR in the period 2009 and 2010. working days adjusted.

(2) Instead, the LTU rate could have been used. 

The incidence rate of LTU is based on the incomplete duration of unemployment, as a
high number of missing quarterly observations prevents use of a statistic based on the
completed duration of unemployment.

(3) Inclusion of quarterly dummies was tested but they were not found to be significant.

(4) Historical data is from Eurostat.

(5) Computations were made using EViews software.

(79) e.g. the efficient wage model (Shapiro (6) In the jargon of EViews, this requires first to include add factors in the GDP and unem-
and Stiglitz, 1984), and the matching ployment rate equations, and second to solve the model in a way that the endogenous
model (Blanchard and Diamond, 1994). variable in these equations match pre-determined trajectories for them. DG ECFIN’s
spring 2009 economic forecast includes quarterly profiles for GDP and annual averages
(80) e.g. the fraction of total unemployment for the unemployment rate. This information was used in setting target trajectories for
accounted for by those experiencing it GDP and unemployment rate in the period 2009 Q1–2010 Q4.
for 12 months and more.

Chapter 2 Labour flows, transitions and unemployment duration

LTU lags total unemployment (Table 20). In a majority of countries, and despite the strong out of jobs, the extent to which they
rise in the unemployment rate, the incidence of LTU is still expected to fall during 2009, can or cannot quickly find alternative
although the LTU rate is already projected to increase but by a considerably smaller margin employment if they lose their current
than the unemployment rate. The LTU rate is expected to increase until the end of 2010. jobs, and the extent to which differ-
ent sections of the labour force are
Table 20: Incidence of LTU, LTU and unemployment rates more affected than others.
(non-weighted annual averages)
Such differences have major implica-
tions for policies. A labour market in
BE 2008 47.7 3.4 7.0
which relatively few people become
2009 41.5 3.5 8.5
unemployed, but where those who
2010 48.0 4.9 10.2
CZ 2008 49.3 2.2 4.4
do are likely to remain unemployed
2009 43.2 2.6 6.1 for a very long time, is likely to be
2010 42.1 3.1 7.4 more damaging to the long-term
DK 2008 13.7 0.5 3.4 employability of jobless individuals
2009 14.5 0.8 5.2 than one in which there are many
2010 21.3 1.4 6.6 more who become unemployed, but
EE 2008 31.3 1.8 5.6 remain in that position for only a
2009 27.9 3.2 11.3 short period of time.
2010 39.3 5.5 14.1
IT 2008 45.6 3.1 6.8
In this context, the first section of this
2009 44.1 3.9 8.8
chapter focused on labour turnover –
2010 49.4 4.6 9.4
i.e. the gross movements of people
LV 2008 26.0 2.0 7.8
in and out of specific jobs together
2009 41.9 6.6 15.7
2010 55.2 8.8 16.0
with moves in and out of employ-
PL 2008 33.5 2.4 7.2 ment altogether. Using data from
2009 32.8 3.2 9.9 the EU LFS, it finds that such gross
2010 37.0 4.5 12.1 labour flows exceed net flows or
PT 2008 47.4 3.8 8.1 employment growth by a significant
2009 44.8 4.1 9.1 margin. In the EU, average annual
2010 49.2 4.8 9.8 labour turnover between 2002 and
SI 2008 42.2 1.9 4.5 2007 amounted to 22% of employ-
2009 52.1 3.4 6.6 ment, compared with net employ-
2010 58.6 4.3 7.4
ment growth of just 1.4% a year.
SK 2008 69.3 6.6 9.6
2009 60.7 7.3 12.0
On average every year, between one
2010 53.8 6.5 12.0
UK 2008 24.2 1.4 5.7
fifth and one quarter of all Euro-
2009 28.0 2.3 8.2 pean workers separate from their
2010 37.1 3.5 9.4 current job and/or are hired to a new
EU11 2008 37.6 2.4 6.4 one. Such labour market dynamism
2009 36.8 3.2 8.7 is not just limited to countries usually
2010 42.2 4.2 9.9 considered as ‘flexible’, such as the
UK or Denmark – but rather it con-
Source: Eurostat and DG ECFIN’s spring 2009 economic forecast, DG EMPL calculations. cerns all Member States, although
annual labour turnover, relative to
total employment, ranges from 14%
in Greece and 16% in Sweden to
between 25% and 30% in the UK,
5. Conclusions In the public debate, it is common Finland, Spain and Denmark.
to discuss labour market perform-
This chapter has focused on analys- ance on the basis of essentially static EU LFS data is fundamental to our
ing the dynamics of EU labour mar- variables, such as levels and rates knowledge of EU labour markets,
kets in order to assess their degree of employment and unemployment. but annual or quarterly data can-
of ‘fluidity’. It has considered three However, labour markets of econ- not pick up flows which occur and
main groups of indicators, cover- omies with similar employment or are reversed between surveys. Using
ing aspects related to labour turno- unemployment rates may be work- both monthly and annual data from
ver, labour market transitions, and ing in very different ways in terms the longitudinal component of EU
­unemployment duration. of the movement of people into and SILC, it is found that, on average,

Employment in Europe 2009

indicators of labour turnover based and inactivity to employment (U_E transition rates have increased sub-
on monthly data are twice as large as and I_E, respectively). Compared with stantially in Germany (since the late
indicators based on annual data. data on turnover rates and gross 1980s), and in Spain (since the mid-
flows, such transition rates can pro- 1990s), while decreasing in Greece
The size of labour turnover also var- vide much more detailed informa- (since the early 1980s), in France and
ies substantially between different tion on the ‘quality’ of these labour Denmark (since the early 1990s), and
groups of workers. Flows tend to be market transitions, i.e. on the preva- in Italy (since 2001).
substantially higher for women than lence of ‘good’ versus ‘bad’ moves.
for men (5 percentage points differ- In relation to worker characteristics,
ence at aggregate EU level) with only In line with the evidence on labour the U_E transition rates for older
a few country exceptions (Estonia, turnover, the EU is seen to be char- workers (aged 55-64) are less than a
Poland and Sweden). Turnover rates acterised by relatively large annual third of those of prime-age workers
for young workers (15–24) stand at transition rates. On average during (aged 24-54). Moreover, more highly
about 70% of their employment the period 2002–07, nearly a third of educated unemployed or inactive
level, being much larger than those unemployed people, and about 10% people have a substantially high-
of prime-age and older workers. At of inactive people, found a job in the er probability of moving back into
the same time, turnover rates tend to following year. However, such U_E employment, especially if they had
decrease with the level of education, transition rates range from 40% or previously been inactive.
although to different degrees across more in the UK, Spain, Portugal and
Member States. the Netherlands to 25% or below in The third section of this chapter cov-
Germany, Greece, Poland and Bel- ers labour market aspects related to
In addition, labour turnover varies gium; while, I_E rates range from the incidence of LTU and its duration.
according to firms’ characteristics, 15% or more in Sweden, the UK and A distinctive feature of EU labour
and the chapter also notes that sec- Denmark to 3% and 4.5% in Greece markets is the high incidence of LTU.
toral differences explain a much and Italy. In Europe, close to 45% of all unem-
larger fraction of overall variability ployment spells last longer than 1
in EU hiring rates compared with dif- This chapter also breaks down transi- year, compared with only about 10%
ferences between countries or the tion rates into cyclical and trend com- in the USA, raising concerns on both
effects of the economic cycle. This ponents. The U_E transition rate and equity and efficiency grounds.
underlies the importance of sector- GDP are positively correlated, while
specific technological, organisational the I_E transition rate appears to be Based on the EU LFS, two indicators
and demand factors in driving labour a procyclical and leading variable. of the average duration of unemploy-
dynamics. As regards the trend components ment are calculated: firstly, the official
of both U_E and I_E transition rates, statistic that measures the duration of
Furthermore, the evidence available a sustained rise occurred since the incomplete spells (i.e. spells still in
suggests that Member States with second half of the 1990s in the EU, progress at the time of observation),
less stringent EPL, such as the UK suggesting a fundamental structural and secondly the duration of com-
and Denmark, or with a higher share improvement in our labour markets pleted unemployment spells.
of temporary employment, such as during this period.
Spain, tend to have higher labour The official statistic of unemploy-
turnover rates. It is nevertheless dif- Moreover, U_E and I_E trend transi- ment duration based on incomplete
ficult to draw definite conclusions tion rates are seen to be negatively spells tends to overlook the large
about the desired or ‘optimal’ levels correlated with structural unemploy- number of short spells of unemploy-
of labour turnover. While more rigid ment, and positively correlated with ment that occur between observa-
labour market institutions tend to participation and employment rates, tion periods, leading to an overesti-
create obstacles to the reallocation suggesting that positive develop- mation of the average duration of
of labour from declining to expand- ments in those transitions have con- completed spells. The chapter finds
ing activities, high labour turnover tributed significantly to the improved that, in the EU over the 2005–08
can also be associated with welfare labour market performance. period, the average duration of com-
costs, such as high frictional unem- pleted spells in unemployment was
ployment, matching costs, a loss of Trends in the development of U_E just about one half of the measure
specific human capital, as well as pos- transition rates vary considerably calculated using incomplete spells –
sible higher spending on unemploy- across Member States. In the UK, the i.e. the official statistic.
ment benefits. share of the unemployed finding a
job within a year nearly doubled from Although the measure based on com-
The second section of this chapter 1983 to 2007, while in Greece there has pleted spells would be more appro-
looks at indicators of transition rates been a significant decline over much priate for assessing the welfare impli-
from unemployment to employment, the same period. In ­contrast, trend I_E cations of unemployment, the official

Chapter 2 Labour flows, transitions and unemployment duration

statistic of unemployment duration in unemployment over a number of A number of studies point to the
has a number of practical advantages years. Third, the data can be utilised importance of labour market insti-
in terms of timeliness, transparency, to monitor multiple unemployment tutional arrangements, notably the
data availability and ease of calcula- spells, allowing the calculation of extent of EPL or the incidence of
tion, which justify its continued use, unemployment duration statistics per temporary work, in explaining major
despite the possible bias. number of spells. cross-country differences in the inci-
dence of LTU and in the duration
Breaking down figures based on com- Finally, the chapter also investigates of unemployment. This chapter car-
plete unemployment spells by work- the impact of certain policies on the ries out a simple econometric cross-
er characteristics, the chapter shows incidence of LTU. In general, econo- country analysis of the impact of a
that gender gaps in average unem- metric studies suggest that higher number of policies on the incidence
ployment duration are very small levels of unemployment benefits of LTU. The results suggest that strict
across the EU (12 months for women and, especially, longer periods of EPL tends to raise LTU, while spend-
against 11.7 for men). However, access to such benefits tend to be ing on labour market policies tends
unemployment duration tends to associated with longer periods of to reduce it.(83)
increase with age and decreases with unemployment, with the probability
the level of education (12.3  months of leaving unemployment tending Flexicurity principles argue for focus-
for the low skilled unemployed, com- to increase significantly just before ing on protecting and supporting
pared with 8.1 months for the skilled their expiry. workers in undertaking ‘good’ tran-
unemployed). sitions in the labour market, rather
The literature also suggests that the than preserving particular jobs. The
An evaluation of a number of aspects probability of leaving unemployment econometric results provide broad
related to unemployment duration, decreases with duration (i.e. ‘neg- support for this approach, with the
such as the incidence of LTU and ative duration dependence’).(82) In specific prospect of reductions in the
multiplicity of unemployment spells, part because employability tends to general level of LTU.
draws on longitudinal data from EU decline the longer people are away
SILC.(81) Three particular aspects are from the labour market. Thus a rise In this respect, and based on the Com-
associated with the use of longitudi- in unemployment duration, follow- mission’s spring 2009 economic fore-
nal data to measure unemployment ing an economic downturn/recession, cast, the chapter provides estimates
duration. may become entrenched, transform- of LTU for the period up to 2010.
ing a cyclical/transitory problem into For an aggregate of 11 EU Member
First, their use allows for a more com- a permanent one – i.e. unemploy- States and representing about 40%
prehensive coverage of shorter spells ment hysteresis. In this context, ade- of the EU labour force, the LTU
of unemployment. Second, they ena- quate policy responses may involve, rate is projected to increase from
ble alternative indicators of LTU to be inter alia, better targeting of ALMPs 2.4% in 2008 to 4.2% in 2010, lag-
calculated, which are robust to short spending towards those most at risk ging the expected evolution of total
interruptions of unemployment, such of staying unemployed for long peri- unemployment, which is expected to
as the fraction of unemployed spend- ods, or of becoming inactive, possibly increase from 6.4% in 2008 to 9.9%
ing a total of 12 and more months using profiling techniques. in 2010.(84)

(83) Financial incentives linked to tax-benefit

systems do not seem to have any sig-
nificant effect on long-term unemploy-

(81) The calendar/monthly information of (82) Although after controlling for individual (84) This unemployment rate is based on cal-
the labour market status variable in the characteristics, the magnitude of this culations carried out in Box 9, which do
longitudinal component of EU SILC. effect seems to be rather limited. not include all 27 EU Member States.

Employment in Europe 2009

References Boone, J. and J. van Ours (2004), Davis, S.J. and J. Haltiwanger (1995),
‘Effective active labor market poli- ‘Measuring gross worker and job flows’,
Acemoglu, D. and R. Shimer (1999), cies’, IZA Discussion Paper, No. 1335, NBER WP 5133, Cambridge, MA.
‘Efficient unemployment insurance’, Bonn, November.
Journal of Political Economy, Vol. Davis, S.J. and J. Haltiwanger (1999),
107, No. 5, pp. 893-928. Bover, O., M. Arellano and S. Bentolila ‘Gross job flows’, in Ashenfelter, O.
(2002), ‘Unemployment duration, ben- and D.Card (eds.), Handbook of Labour
Abowd, J.M., P. Corbel and F. Kra- efit duration and the business cycle’, Economics, Vol. 3, Chapter 41.
marz (1999), ‘The Entry and Exit of The Economic Journal, pp. 223-265.
Workers and the Growth of Employ- De Loecker, J. and J. Konings (2006),
ment: An Analysis of French Estab- Brown, J.D. and J.S. Earle (2008), ‘Under- ‘Job reallocation and productivity
lishments’, Review of Economics and standing the Contributions of Realloca- growth in a post-socialist economy:
Statistics, Vol. 81, No. 2, pp. 170-187. tion to Productivity Growth: Lessons evidence from Slovenian manufac-
from a Comparative Firm-Level Analy- turing’, European Journal of Political
Akerlof, G. and B. Main (1980), ‘Unem- sis’, IZA Working Paper No. 3683. Economy, Vol. 22, pp. 388–408.
ployment spells and the unemploy-
ment experience’, The American Eco- Burda, M. and C. Wyplosz (1994), European Commission (2007),
nomic Review, Vol. 70, pp. 885-893. ‘Gross worker and job flows in ‘Towards common principles of flexi-
Europe’, European Economic Review, curity: more and better jobs through
Baker, G. and P. Trivedi (1985), ‘Esti- Vol. 38, pp. 1287-1315. flexibility and security’, Directorate-
mation of unemployment duration General for Employment, Social
from grouped data: a comparative Cahuc, P. and A. Zylberberg (2004), Affairs and Equal Opportunities,
study’, Journal of Labor Economic, Labor economics, MIT Press, Cam- Brussels.
Vol. 3, No.2. bridge, MA.
European Commission (2007), ‘The
Baker, M. (1992), ‘Unemployment Calmfors, L. and J. Driffill (1988), ‘Bar- social situation in the European Union
duration: compositional effects and gaining structure, corporatism and 2007 – social cohesion through equal
cyclical variability’, The American Eco- macroeconomic performance’, CEPR. opportunities’, Directorate-General
nomic Review, Vol. 82, pp. 313-321. for Employment, Social Affairs and
Carone, G., H. Immervoll, D. Paturot Equal Opportunities, Brussels.
Ball, L. (2009), ‘Hysteresis in unem- and A. Salomäki (2004), ‘Indicators of
ployment: old and new evidence’, unemployment and low-wage traps’, European Commission (2008), ‘Labour
NBER Working Paper No. 14818. OECD Social, Employment and Migra- market and wage development in
tion Working Paper, No. 18, Paris. 2007’, European Economy, Vol. 5,
Bartelsman, E., J. Haltiwanger and Box 3, pp. 48-58.
S. Scarpetta (2004), ‘Microeconomic Carone, G., K. Stovicek, F. Pierini
Evidence of Creative Destruction in and E. Sail (2009), ‘Recent reforms Foster, L., J. Haltiwanger and C. J. Kri-
Industrial and Developing Countries’, of the tax and benefit systems in the zan (2001), ‘Aggregate productivity
IZA Working Paper No. 1374. framework of flexicurity’, European growth: lessons from microeconomic
Economy, Occasional Papers No. 43. evidence’ in Dean, E., M. Harper
Bertola, G., T. Boeri and S. Cazes and C. Hulten, eds., New Develop-
(1999), ‘Employment protection and Corak, M. and A. Heisz (1995), ‘Alter- ments in Productivity Analysis, Chi-
labour market adjustment in OECD native measures of the average dura- cago, University of Chicago Press, pp
countries: evolving institutions and tion of unemployment’, manuscript. 303–372.
variable enforcement’, Employment
and Training paper 48, ILO, Geneva. Dale-Olsen, H. (2006), ‘Labour turno- Garibaldi, P., C. Pissarides and J. Kon-
ver and mobility in beginning of ings (1997), ‘Gross job reallocation
Bertola, G. and R. Rogerson (1997), the 21st century – a survey of the and labour market policy’ in Snower,
‘Institutions and labor reallocation’, literature on job and worker flows’, D. and G. De la Dehesa, Unemploy-
European Economic Review, Elsevier, working paper. ment policy: government options for
Vol. 41, No.6, June, pp.1147-1171. the labour market, Cambridge Uni-
Davis, S., R. Faberman and J. Halti- versity Press, pp. 467-489.
Blanchard, O. and P. Portugal (2001), wanger (2006), ‘The flow approach
‘What hides behind an unemploy- to labor markets: new data sources Gomez-Salvador, R., J. Messina and
ment rate: comparing Portuguese and micro–macro links’, Journal of G. Vallanti, (2004), ‘Gross job flows
and U.S. labor markets’, The Ameri- Economic Perspectives, American and institutions in Europe’, Working
can Economic Review, Vol. 91 No.1, Economic Association, Vol. 20, No.3, Paper 318, European Central Bank,
pp. 187-207. summer, pp. 3-26. Frankfurt.

Chapter 2 Labour flows, transitions and unemployment duration

Haltiwanger, J. and M. Vodopivec Martin, J. (2000), ‘What works OECD (2006), ‘General policies to
(2003), ‘Worker flows, job flows and among active labour market poli- improve employment opportunities for
firm wage policies’, The Economics cies: evidence from OECD countries’ all’, Employment Outlook, OECD, Paris.
of Transition, The European Bank for experiences’, OECD Economic Studies
Reconstruction and Development, No.30, 2000/I. OECD (2009), ‘Looking inside the per-
Vol. 11, No.2, June, pp.253–290. petual motion machine: how indus-
Masso, J., R. Eamets and K. Philips try, firm and worker characteristics
Haltiwanger, J., S. Scarpetta and (2004), ‘Creative destruction and tran- shape job and worker flows’, Employ-
H. Schweiger (2006), ‘Assessing job sition: the effects of firm entry and ment Outlook, OECD, Paris.
flows across countries: the role of exit on productivity growth in Esto-
industry, firm size, and regulations’, nia’, IZA Working Paper No. 1243. Olsen M. (1982), The rise and decline
Policy Research Working Paper Series of nations – economic growth, stag-
4070, The World Bank. Mortensen, D. (1977), ‘Unemploy- flation, and social rigidities, Yale
ment insurance and job search deci- University Press.
Hasselpflug, S. (2005), ‘Availability sions’, Industrial and Labor Relations
criteria in 25 countries’, Finansmin- Review, Vol. 30, No.4. Pissarides C. (2000), Equilibrium
isteriet Working Paper, No. 12/2005, unemployment theory, MIT Press,
Copenhagen. Mortensen, D. (ed) (1994), ‘Job crea- Cambridge, MA.
tion and job destruction in the the-
Kaplan, L. and P. Meyer (1958), ‘Non- ory of unemployment’, Review of Portugal, P. (2008), ‘Unemployment
parametric estimation from incom- Economic Studies, Vol. 61, July, pp. duration in the Portuguese labour
plete observations’, Journal of the 397-415. market’, Economic Bulletin, Banco
American Statistical Association, Vol. de Portugal.
53, pp. 457-481. Muffels, R. and D. Fouarge (2008),
Labour market flexibility and employ- Salant, S. (1977), ‘Search theory and
Katz, L. and B. Meyer (1988), ‘The ment security in Europe: a Siamese duration data: a theory of sorts’,
impact of the potential duration of twin?, Edward Elgar. Quarterly Journal of Economics, Vol
unemployment benefits on the dura- 91, No.1-4.
tion of unemployment’, NBER Work- Muffels, R., T. Wilthagen and N. van
ing Paper No.2741. den Heuvel (2002), ‘Labour market Sider, H. (1985), ‘Unemployment
transitions and employment regimes’, duration and incidence: 1968-82’,
Kiefer, N. (1988), ‘Economic duration Discussion Paper FS I 02–204. The American Economic Review, Vol
data and hazard functions’, Journal 75, pp. 461-472.
of Economic Literature, Vol. 26, pp. OECD (1998), ‘Early identification of
646-679. jobseekers at risk of LTU – The role of Schmidt, G. and B. Gazier (2002), The
profiling’, OECD Proceedings. dynamic of full employment: social
Kugler, A. and G. Pica (2003), ‘Effects integration through transitional
of employment protection on worker OECD (1999), ‘Employment protec- labour market, Edward Elgar, 2002.
and job flows: evidence from the tion and labour market perform-
1990 Italian reform’, Labour Econom- ance’, Employment Outlook, OECD, van den Berg, G. and J. van Ours
ics, Vol. 15, pp. 78-95. Paris. (1994), ‘Unemployment Dynamics
and Duration Dependence in France,
Layard, R., S. Nickell and R. Jackman OECD (2002), ‘The ins and outs of the Netherlands and the UK’, Eco-
(1991), ‘Unemployment’, Oxford Uni- long-term unemployment’, Employ- nomic Journal, Vol. 104, pp. 432-443.
versity Press. ment Outlook, Chapter 4.
van den Berg, G. and J. van Ours
Machin, S. and A. Manning (1999), ‘The OECD (2004), ‘Employment protec- (1996), ‘Unemployment Dynamics
causes and consequences of longterm tion regulation and labour market and Duration Dependence’, Journal
unemployment in Europe’, Handbook performance’, Employment Outlook, of Labor Economics, Vol. 14, pp.
of Labor Economics, Vol. 47. OECD, Paris. 100-125.

Chapter 3

Climate change and

labour market outcomes
1. Introduction needed if we are to avoid dramatic, into ­consideration that tackling cli-
irreversible and self-reinforcing mate change(5) provides significant
The International Panel for Climate changes in the world’s climate.(4) opportunities to develop new tech-
Change (IPCC, 2007d) and Stern nologies, create new jobs, enhance
Review (2006)(1) provide convincing The current global economic crisis energy security, increase interna-
evidence that the world is already may be slowing the rate of increase tional competitiveness and improve
experiencing global warming, and in energy use and carbon emis- public health.
that – in view of the fact that since sions, but it does not change the
the onset of the industrial era the significance of this long-term chal- This chapter analyses how European
human impact on climate greatly lenge. Furthermore, it should not labour markets are affected by cli-
exceeds the impact from natu- hinder the direction and speed of mate change, adaptation to it, and
ral factors(2) – deep and significant the transition towards a competitive policies to mitigate further climate
cuts in anthropogenic greenhouse low-carbon and resource-efficient change, with a view to strengthening
gas (GHG)(3) emissions are urgently­ economy – certainly not if one takes the development of labour market
policies that can support the creation
(4) Although the consensus is quite wide-
of more and better jobs for all in a
spread, nonconformist views on the
urgency and size of climate change sustainable economy.
action do exist , see for instance Lawson
(2008) and Nordhaus (2007 and 2008). In
The second section begins with a
its assessment of the effects of climate
change, the Stern Review (2006) proposes brief summary of the main envi-
to use a low social discount rate implying ronmental challenges faced by the
that costs carried by future generations
European Union (EU) and the world,
have a high actual value. On compar-
ing the costs and benefits of action on and examines how new economic
climate change the Stern Review (2006) concepts related to these challenges,
finds that the benefits of strong, early