You are on page 1of 80

Labour Market Vulnerability

and Labour Market Outcomes


during the Economic Upswing

Dr Vassilis Monastiriotis and Ioannis Laliotis


June – 2019
EUROPEAN COMMISSION
Directorate-General for Employment, Social Affairs and Inclusion
Directorate A — Employment and Social Governance
Directorate A, Unit EMPL. A4 — Thematic Analysis
Contact: Katarina Jaksic
E-mail: katarina.jaksic@ec.europa.eu

European Commission
B-1049 Brussels

2
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Labour Market Vulnerability


and Labour Market Outcomes
during the Economic Upswing

“The opinions expressed and arguments used in this publication are those of the author(s) and do not necessarily
reflect the official position of the European Commission. The European Commission does not guarantee the
accuracy of the data included in this publication. Neither the European Commission nor any person acting on the
European Commission’s behalf may be held responsible for the use which may be made of the information
contained therein.”

Directorate-General for Employment, Social Affairs and Inclusion


Social Situation Monitor
2019
LEGAL NOTICE
Manuscript completed in June 2019
Neither the European Commission nor any person acting on behalf of the European Commission is responsible
for the use that might be made of the following information. More information on the European Union is
available on the Internet (http://www.europa.eu).

Luxembourg: Publications Office of the European Union, 2020

PDF ISBN 978-92-76-14623-0 doi:10.2767/830095 KE-04-19-795-EN-N


© European Union, 2020

Reuse is authorised provided the source is acknowledged. The reuse policy of European Commission
documents is regulated by Decision 2011/833/EU (OJ L 330, 14.12.2011, p. 39). For any use or reproduction
of photos or other material that is not under the EU copyright, permission must be sought directly from the
copyright holders.
Contents
1. INTRODUCTION .................................................................................................. 6
2. VULNERABLE GROUPS AND LABOUR MARKET OUTCOMES ................................... 8
3. SELECTION OF CASES AND EMPIRICAL APPROACH ...........................................10
4. CHANGING LABOUR MARKET OUTCOMES FOR VULNERABLE GROUPS ................13
4.1 THE PREVALENCE OF VULNERABLE GROUPS ACROSS EMPLOYMENT STATUSES
..........................................................................................................................13
4.2 THE DISTRIBUTION OF LABOUR MARKET OUTCOMES FOR DIFFERENT GROUPS
..........................................................................................................................19
4.3 CHANGES IN THE INCIDENCE OF LABOUR MARKET STATUSES – MULTIVARIATE
ANALYSIS ..........................................................................................................26
5. A CLOSER LOOK – LABOUR MARKET OUTCOMES FOR HYPOTHETICAL PROFILES
..........................................................................................................................34
5.1 ESTONIA ............................................................................................................34
5.2 SPAIN ................................................................................................................38
5.3 PORTUGAL .........................................................................................................40
6. CONCLUSIONS ...................................................................................................43
7. BIBLIOGRAPHY .................................................................................................48

5
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

1. INTRODUCTION

The global financial crisis that erupted following the destabilization of the sub-prime
mortgage market in the USA in 2007 and the collapse of Lehman Brothers in September
2008, transmitted shock-waves to the global economy and had a particularly lasting
impact on the European economy. Owing in part to economic governance issues of the
Economic and Monetary Union and to a large extent to the fiscal derailment of Greece in
2009, the European economy, with the euro-area at its epicentre, entered a long period
of political, economic and institutional stress. Unemployment in the euro-area, which had
reached a historical low of 7.3% in 2007, increased to 10.3% in the second quarter of
2010 and to 12.1% in the second quarter of 2013, before starting subsiding more recently
(standing at 7.9% in the fourth quarter of 2018). 1

Although this cycle was followed by virtually all EU28 countries, at the national level there
have been some significant variations both in the timing and in the extent of
unemployment changes (Monastiriotis, 2018). For example, at its peak, unemployment
reached 27.7% in Greece and 26.3% in Spain, at a time when countries such as Austria,
Denmark, Germany and the Netherlands were close to full employment (unemployment
rates in the area of 5-7%). Similarly, the unemployment trajectories that have been
followed more recently also vary across the EU28. Some countries which had experienced
limited unemployment increases during the crisis, such as Austria, Finland and Luxemburg,
have seen their unemployment rates following an upward trend more recently. Other
countries, such as Belgium, Cyprus, France and Italy have been experiencing much higher
unemployment persistence compared to the overall EU28/EA19 trend. In a significant
number of countries, however, the recent period has been one of substantial downward
adjustments in unemployment – with 11 countries registering rates of decline in
unemployment of over 40% and 8 countries registering unemployment rates below 5% in
2017.

It is well known that during times of recession and heightened unemployment, the labour
market attachment of individuals belonging to vulnerable groups is disproportionately
affected. This is for at least two sets of reasons. First, fast rises in unemployment create
conditions of intensified job competition. A result of this is the so-called ‘bumping-down’
effect, whereby job losses – which may in the first instance be occurring anywhere in the
occupational/skills distribution – trickle down to the less skilled or to people with shorter
work histories or interrupted career paths (i.e., lower labour market attachment). In this
perspective, skilled individuals affected by unemployment (either via dismissals or as

1
Figures from Eurostat’s quarterly unemployment series (ure_rt_q) referring to the EA19.

6
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

existing job-seekers) displace individuals from jobs further down the occupation/skills
ladder – up to the point where the least qualified/competitive individuals are ‘sorted away’
from the labour market. Second, heightened unemployment also makes it more difficult
to find (or retain) a job even in the absence of job competition. Individuals with less
competitive skills and less ‘marketable’ characteristics are more likely to become
discouraged in their job-searches (the so-called ‘discouraged worker’ effect) and thus to
voluntarily drop out from the labour market. Empirical evidence on both of these effects
(bumping-down and discouraged-worker) is ample in the academic literature (Battu and
Sloane, 2000; McGuinness, 2006; O'Brien, 2011) and especially in relation to the crisis
(Congregado, 2011; Mossfeldt and Österholm, 2011; Monastiriotis and Martelli, 2013;
Gallie, 2013; Hogarth and Wilson, 2015; Vertova, 2016). 2

Do these conditions of labour market detachment reverse during economic upswings? In


the particular case of the recent European recovery, is there evidence to show that
individuals typically associated with labour market vulnerability have experienced
enhanced labour market attachment and improved labour market outcomes, especially in
those EU28 countries which have shown the strongest signs of recovery and/or labour
market conditions close to full employment? To put it differently, how have vulnerable
groups fared during the modest but noticeable economic upswing in parts of the EU28 in
the recent period? Where do they find jobs – in which sectors and occupations – and in
what kind of jobs?

This Research Note seeks to address these questions by investigating the labour market
outcomes, and degree of labour market attachment, that individuals belonging to different
vulnerable groups and/or possessing characteristics typically associated with labour
market vulnerability have experienced during the post-crisis economic upswing recorded
in many countries across Europe. To do so, it (a) identifies a number of characteristics and
groups of individuals commonly associated with heightened employment barriers and
unemployment risk and lower levels of labour market attachment; (b) selects a sub-group
of EU28 countries which have exhibited superior economic performance in their labour
markets in the post-crisis period (as measured by their national unemployment rate and
by the rate of decline of unemployment); and (c) examines how labour market outcomes
(employment status, occupational and sectoral distributions) have shifted for these
characteristics/groups of individuals in the selected countries experiencing an economic
upswing.

2
The literature also identifies an inverse – so-called ‘added-worker’ – effect, whereby heightened unemployment
and diminished household incomes push less qualified and previously detached individuals into the job
market (Congregado, 2011; Karamessini and Rubery, 2014; Bredtmann et al, 2018).

7
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

The structure of the Research Note is as follows. The next section discusses the issue of
labour market detachment and vulnerability, drawing on some selected literature on the
topic and relating in particular to the codification of labour market vulnerability offered
recently in the Faces of Joblessness project conducted jointly by the OECD, the World Bank
and the European Commission. Section 3 describes the data and approach for the empirical
analysis, including an explanation for the selection of country cases. Section 4 presents in
detail the analysis of shifts in labour market outcomes of specific groups of individuals in
the selected countries, through a series of descriptive and multivariate analyses. Section
5 directly connects to the definitions of vulnerability emanating from the Faces of
Joblessness project and examines, for specific indicative profiles, shifts in labour market
outcomes experienced in a subset of the selected countries for which data are available.
The last section concludes with some policy reflections in relation to questions both of
labour market inclusion and of potential labour market bottlenecks.

2. Vulnerable Groups and Labour Market Outcomes

The ILO defines “vulnerable populations” in relation to access to, and need of, social
protection and social service provision. This intersects with the wider concept of
“vulnerability” that focuses on the extent of poverty (or poverty risk) and informality.
Thus, vulnerable groups include women and children (related to issues of healthcare,
childcare and employment), the elderly (mainly on issues related to health and wellbeing),
and persons with disabilities (concerning both issues of healthcare and access to jobs) who
are in high risk of poverty. The OECD has a similar definition for vulnerable groups, which
emphasises the extent of need for multiple service supports: families in poor households,
children, youth with mental health issues, the homeless, and the frail elderly (OECD,
2015).

More recently, through their joint “Faces of Joblessness” project, the OECD, the European
Commission and the World Bank, shifted attention to the intersectionality of characteristics
linked to vulnerability and the central role that people’s position in (or out of) the labour
market may play in this, by developing a methodology to identify groups of individuals,
with different sets of characteristics, who face various barriers to employment and have
overall weak or nonexistent labour market attachment (Fernandez et al, 2016; Sundaram
et al, 2014). Through this, vulnerability is not only, or exclusively, defined on the basis of
fixed and exogenously acquired characteristics (e.g., female, elderly, migrant) but mainly
in relation to a number of pre-defined employment barriers. The latter include both
enabling and disenabling characteristics, i.e., characteristics that allow individuals to
detach themselves from the labour market (e.g., availability of ‘high’ non-labour incomes
and ‘high’ earnings replacement) and ones that do not allow them to attach themselves to

8
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

the labour market (e.g., low education or skills, health limitations, care responsibilities);
as well as characteristics that are specific to the person (e.g., limited/no work experience)
and ones that concern the demand conditions in the labour market in which the individual
lives (e.g., scarce job opportunities). Through these, a number of ‘profiles’, or groups,
have been derived, identifying individuals who possess different characteristics and face
different forms of vulnerability. Examples of these include: low-skilled economically
inactive mothers receiving benefits; well-educated mothers with working partners;
discouraged young adults with limited work experience; skilled/experienced older-age
males with limiting health issues; income-poor parents with no/limited past work
experience; and so forth.

In the wider literature concerning labour market exclusion and vulnerability, such
intersectionalities are usually examined through the analysis of labour market outcomes
(e.g., unemployment, inactivity, non-standard employment) for specific groups which are
known to exhibit high prevalence of, or significant barriers to, labour market attachment.
This concerns mainly: (a) females, (b) young adults, (c) individuals with low education,
(d) old-age individuals (pre- or post-retirement age), (e) migrants, and (f) individuals in
households with enhanced caring responsibilities (e.g., large number of dependents or
family members with poor health).

For women, labour market vulnerability is known to have multiple dimensions. Females
are disproportionately tied-up to household, parental and care responsibilities that
significantly limit their degree and intensity of labour market attachment (Ferrant et al,
2014); while they also restrict their effective job search intensity and job accessibility
(Matas et al, 2010). Career interruptions related to maternity also put females at a more
precarious position, as they dampen their employment opportunities and wage profiles
(Cebrián and Moreno, 2015) and restrict their career paths (Arntz et al, 2017). Females
are also known to respond more to business cycle fluctuations, constituting essentially a
‘reserve army’ of employment in times of recessions and negative household income
shocks (Karamessini, and Rubery, 2013).

The labour market vulnerability of migrants is also related to the economic cycle, with
unemployment rates among migrants rising typically faster in times of recession (Bonifazi
and Marini, 2014; Cebolla-Boado et al, 2015) and in-work vulnerabilities (low-pay, atypical
contracts, over-education) also rising (Krings et al, 2011; Khattab and Fox, 2016).
Migrants, however, face problems of access to, and discrimination in, the labour market
more generally and thus exhibit greater degrees of labour market vulnerability even in
good times. A large literature exists that shows how labour market discrimination affects
migrant’s wages and employment opportunities (see, inter alia, Bertrand and

9
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Mullainathan, 2004; Wood et al, 2009; Reyneri and Fullin, 2011; and Kaas and Manger,
2012).

Evidence of discrimination has also been found for older-age workers (Cheung et al, 2011;
Ahmed et al, 2012). For them, however, problems of labour market exclusion and
vulnerability often intersect with issues of ill-health and skills. Indeed, there is evidence
to suggest that technological change accelerates the deskilling of older-age cohorts of
employees (Peng et al, 2017) and particularly so in terms of ‘hard skills’, which are known
to matter more for older workers (Van Dalen et al, 2010). Health hazards and potential
occupational health risks are also a major force in the detachment of older-age individuals
from the labour market (Jones et al, 2013), besides of course the effect that individual ill-
health has on this (Schuring et al, 2013).

A much more extensive literature exists with regard to problems faced in the labour market
by young adults. Problems related to insufficient work experience, lack of soft-skills related
to typical entry-level jobs, as well as the nature of jobs available to new labour market
entrants (in relation in particular to labour market flexibility and dualisation) and family
characteristics such as parental education and past unemployment experience have long
been identified as key issues raising the labour market vulnerability of young adults
(O’Reilly et al, 2015; Zwysen, 2015). With the crisis, the issue of NEET youths (not in
employment, education or training) has received extensive policy attention and has
spurred a growing literature seeking to explain how youth employment opportunities
respond to business cycle fluctuations, finding evidence of increased precariousness
(involuntary part-time and temporary jobs) and labour market exclusion (inactivity, under-
employment) in times of recession and intensified job competition (see, inter-alia, Bell and
Blanchflower, 2015; Dolado et al, 2013; Kelly and McGuinness, 2015; and Quintano et al,
2018).

3. Selection of cases and empirical approach

To analyse the labour market trajectories of individuals during the recent economic
recovery, we draw on both of the classifications reviewed above: those that refer to fixed
individual characteristics as identified in the wider literature of labour market risks
(females, migrants, youth, old-age, etc); and those that derive from the intersectional
analysis of disadvantage (employment barriers) of the Faces of Joblessness project. We

use data from the EU-SILC 3 and concentrate on two years: 2012, which represents for

3
EU-SILC is the European Statistics on Income and Living Conditions. The survey is conducted annually by the national statistical authorities of
the EU member states and associated countries under the auspices of Eurostat. The data used here have been obtained from Eurostat under
license.

10
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

most countries the height of the crisis; and 2017, which is the latest year for which EU-
SILC micro-data are available.

Our analysis proceeds as follows. First, we construct eight major groups of individuals
potentially experiencing labour market risk. These are: Adult females (aged 15+); Youth
(15-24 years old); Mature-age individuals (55-64 years old); Old-age individuals (65-74

years old); Foreign-born 4 individuals (aged 15+); Low-educated individuals (with up to

primary education); Individuals living in large households (5+ members); and Individuals
living in households with at least one member reporting a limiting health condition (‘bad’
or ‘very bad health’). For each of the two years under study, we use the cross-sectional
information available in the data to examine: (a) the prevalence of each group across five
key labour market statuses (inactive, unemployed, salaried full-time employee; salaried
part-time employee; self-employed); (b) the incidence of each of these labour market
statuses in each group (e.g., the share of females who are inactive, unemployed, self-
employed, etc); (c) and the sectoral and occupational distribution of employment for each
of these groups (using 1-digit classifications). This analysis, presented in sections 4.1 and
4.2, allows us to document not only the position of these groups in (or out of) the labour
market, but also how this has shifted during the economic upswing.

Second, we take this investigation deeper by examining how the above distributions
change when we take into account simultaneously all other personal and household
characteristics of the individuals concerned. For example, for the employment outcomes
of females, we examine how these are distributed (percent inactive, unemployed, etc)
when we control for females’ education, age, ethnicity, marital status, household size, etc.
To do so, we estimate econometrically a multinomial logit model for the five labour market
status outcomes and derive the conditional predicted probabilities for each of these

outcomes for each of the groups under study. 5 We report these conditional probabilities

in section 4.3 and compare them with those obtained from the descriptive analysis.

Table 1. Unemployment rates (2017) and rates of change (2012-2017) across


the EU28
Unemployment rates Selection criteria
% change, Criterion
Country 2012 2017 Selected
2012-2017 fulfilled
Austria 4.9% 5.5% 12.2% Neither No
Belgium 7.6% 7.1% -6.6% Neither No
Bulgaria 12.3% 6.2% -49.6% Rate Yes
Croatia 15.8% 11.0% -30.4% Neither No
Cyprus 11.9% 11.1% -6.7% Neither No
Czech Republic 7.0% 2.9% -58.6% Both No
4
We use the place of birth information in our data to proxy for the migrant population in each country, as the available data do not directly
report the ethnicity or migration status of individuals. Naturally, the foreign-born group includes, besides the first-generation migrants, also the
nationals who were born abroad. Inversely, it does not include second- or third-generation migrants.
5
The method is explained in Appendix B.

11
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Denmark 7.5% 5.7% -24.0% Neither No


Estonia 10.0% 5.8% -42.0% Rate Yes
Finland 7.7% 8.6% 11.7% Neither No
France 9.8% 9.4% -4.1% Neither No
Germany 5.4% 3.8% -29.6% Level Yes
Greece 24.5% 21.5% -12.2% Neither No
Hungary 11.0% 4.2% -61.8% Both Yes
Ireland 15.5% 6.7% -56.8% Rate No
Italy 10.7% 11.2% 4.7% Neither No
Latvia 15.0% 8.7% -42.0% Rate No
Lithuania 13.4% 7.1% -47.0% Rate No
Luxembourg 5.1% 5.6% 9.8% Neither No
Malta 6.2% 4.0% -35.5% Level No
Netherlands 5.8% 4.9% -15.5% Level No
Poland 10.1% 4.9% -51.5% Both Yes
Portugal 15.8% 9.0% -43.0% Rate Yes
Romania 6.8% 4.9% -27.9% Level No
Slovakia 14.0% 8.1% -42.1% Rate No
Slovenia 8.9% 6.6% -25.8% Neither No
Spain 24.8% 17.2% -30.6% Neither Yes
Sweden 8.0% 6.7% -16.3% Neither No
United Kingdom 7.9% 4.4% -44.3% Both Yes
EA (19 countries) 11.4% 9.1 -20.2%
European Union - 28 10.5% 7.6 -27.6%
Source: Eurostat; authors’ calculations.

Last, utilising the information derived from the multivariate analysis, we implement an
innovative analysis which maps directly into the groups of labour market vulnerability
identified in the Faces of Joblessness project. We select three countries from the Faces of
Joblessness project, which belong to our list of eight countries studied (see below) for
which detailed groups of labour market vulnerability have been derived in the Faces of
Joblessness project. For each of these countries we use our estimated multinomial logit
models to derive predicted probabilities for the five labour market statuses mentioned
above for each of the distinctive (and country-specific) groups of labour market
vulnerability that have been reported in the relevant Faces of Joblessness studies (Browne
et al, 2018; Düll et al, 2018; Fernandez et al, 2018). We implement this analysis both for
2012 and for 2017 and examine the changes that have occurred for each of these groups
between the two years. This provides a direct measurement of how the economic upswing
has affected the selected groups (profiles) in the selected countries.

To identify the countries that have experienced an economic upswing in the recent period,
we rely on aggregate data on unemployment (from Eurostat) and pose two selection
criteria: the first relates to the extent to which countries are near full-employment, defined
here as an unemployment rate of 5% or less; the second concerns the extent to which
national economies have experienced a sizeable relative improvement in their labour
market performance, with the selection criterion set at cases where the unemployment
rate declined by more than 40% between 2012 and 2017. The performance of the EU28

12
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

countries on these two measures, as well as our selection decisions, are reported in Table
1.

As can be seen in Table 1, in 2017 there were eight countries in the EU28 with
unemployment rates at or below 5%; while eleven countries had experienced a drop in
their unemployment rates of more than 40%. Of these, we selected eight countries for our
subsequent analysis, trying to ensure that our selected group covers a range of
geographies (e.g., north, south, east), political economies (e.g., Nordic, Southern,
Continental) and labour market experiences across the EU28. Specifically, our selection
includes four of the eight EU28 countries which had in 2017 unemployment rates below
5% and six of the eleven countries which experienced a drop of more than 40% in their
unemployment rates between 2012 and 2017. These include: Hungary, Poland, Germany
and the United Kingdom, which exhibit unemployment rates below 5% in 2017; and
Bulgaria, Estonia and Portugal (as well as Hungary, Poland and the UK), which have
registered rates of decline of their unemployment rates by over 40% between 2012 and
2017. Spain is also included as the country with the largest absolute decline in the
unemployment rate in the same period (2012-2017), of over 7.5 percentage points. 6

4. Changing Labour Market Outcomes for Vulnerable Groups

4.1 The prevalence of vulnerable groups across employment


statuses

Measuring the prevalence of different vulnerable groups in each of the various labour
market statuses offers a picture concerning the exposure of individuals with different
characteristics to specific labour market outcomes. To demonstrate this, in Table 2 we
summarise the results concerning the prevalence across different labour market statuses
in 2017 for females and the youth (15-24 years old), two key groups of individuals that
are often associated to heightened labour market vulnerability. To clarify, prevalence
measures the concentration of individuals in any specific category, i.e., the frequency at
which any particular group of individuals is observed within any particular labour market
status (for example, the prevalence of females in unemployment is measured by the
percentage of unemployed people who are females).

Starting with the results for females (top panel of Table 2), we can see the common pattern
of females being disproportionately over-represented in the categories of inactivity and
part-time employment and under-represented mainly in the self-employment category.

6
Greece has exhibited a steeper decline in unemployment more recently (fourth quarter of 2018), both in absolute and in proportional terms (9.4
percentage points and 33.9%, respectively), compared to its peak in the third quarter of 2013). For the period analysed here, however (2012-
2017), the decline in unemployment in Greece was smaller than that seen in Spain.

13
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Country variations are quite notable, with the female prevalence in inactivity ranging from
a ‘low’ 54.9% in Germany to over 60% in Bulgaria, Estonia, Hungary and Poland, i.e., in
the countries belonging to the New Member States of the EU28 (and always above the
corresponding share of females to the total population). Differences in the prevalence of
females into part-time employment are even more marked, with values over 80% in
Germany (meaning that over 80% of those in part-time employment are females) and
above 75% in Spain and the United Kingdom; and values below 60% in Bulgaria.
Concerning self-employment, country values range between 32% and 42% (i.e., below
the share of females in the total population), while for full-time employment they range
between 41% and 49% (with the exception of Germany, which has much lower prevalence
of females in full-time employment, consistent with its very high female prevalence in
part-time employment).

Table 2. Prevalence of females and the youth in different labour market statuses
(2017)
Part- Self-
Full-time Unemployed Inactive
Country time employed
Females
Bulgaria 47.8% 59.3% 33.4% 41.3% 60.9%
Estonia 49.3% 69.5% 31.8% 36.3% 63.5%
Germany 34.2% 81.5% 41.8% 48.3% 54.9%
Hungary 46.5% 63.9% 37.6% 40.8% 62.7%
Poland 47.5% 70.4% 40.1% 55.2% 63.3%
Portugal 49.6% 69.1% 40.4% 50.3% 58.3%
Spain 43.4% 75.9% 34.0% 54.7% 57.6%
United
Kingdom 41.0% 75.9% 31.9% 36.3% 58.8%
Youth
Bulgaria 3.0% 5.3% 0.5% 9.7% 14.8%
Estonia 4.2% 8.3% 1.4% 13.4% 17.3%
Germany 7.1% 4.3% 1.2% 4.6% 16.2%
Hungary 4.4% 7.9% 0.7% 13.3% 19.2%
Poland 4.3% 10.6% 2.2% 8.6% 15.2%
Portugal 3.4% 8.6% 1.2% 9.7% 18.2%
Spain 2.3% 7.3% 1.1% 8.7% 17.8%
United
Kingdom 9.5% 13.2% 2.2% 24.7% 18.8%
Source: European Union Statistics on Income and Living Conditions (EU-SILC); authors’ calculations.

Significant cross-country variations also exist for the case of youth labour market
outcomes. In all countries, inactivity is the labour market status with the highest
prevalence of youth – as should be expected perhaps given the fact that many individuals
in this group are still in education. However, the prevalence of youth in the status of
unemployment varies substantially across countries – ranging from below 5% in Germany

14
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

to near 25% in the United Kingdom; with Hungary and Estonia also exhibiting
comparatively high values. Instead, the prevalence of youth in self-employment is much
lower in all countries and well-below their population shares. But, inversely, the prevalence
of youth is significantly higher in the part-time employment category (compare vis-à-vis
the shares reported for full-time employment) in practically every country under study
(with the exception of Germany). From these results it can be concluded that in the
majority of countries – essentially, with the exception of Germany – the youth face
heightened risks of labour market vulnerability, both with regard to labour market
detachment (inactivity, unemployment) and with regard to labour market precariousness
(as proxied by part-time employment).

It is impossible to present here in full length a complete picture concerning the prevalence
measures for all groups of vulnerability characteristics, across the eight countries, five
labour market statuses and two years of analysis. We thus report these results in Table
A.1 in the Appendix and, instead, we present here (Table 3) in summary form the changes
in the degree of prevalence of each group in each labour market status, relative to the
changes in its overall prevalence in the population, between 2012 and 2017. In Table 3, a
positive (negative) sign represents an increase (decline) in the relative prevalence of the
relevant group to the corresponding labour market status; with signs in parentheses
indicating cases where the change over time was at a rate of less than 5%.

Starting with the case of females, as can be seen, in countries such as Bulgaria, Estonia,
Hungary and the United Kingdom, the economic upswing has been accompanied by an
increase, albeit not a particularly high one, in the prevalence of this group in the inactivity
category – presumably as males experienced faster exits from inactivity relative to
females. In the remaining countries the opposite is true, with Spain recording a significant
decline in the prevalence of females in inactivity. Concerning unemployment, quite
inversely, the prevalence of females in this category declined significantly in Estonia,
Hungary and the United Kingdom, signifying lower labour market exclusion, but it
increased in Spain. Female prevalence in part-time employment increased sizably in
Bulgaria and Poland (and less significantly in Estonia and Hungary), but it declined in
Portugal (and, less so, in Germany, Spain and the United Kingdom). Instead, female
prevalence in self-employment increased in Germany, Spain, Hungary and Portugal but
declined in Estonia.

Concerning the youth group, we observe increases in their relative prevalence in the
inactivity category in all countries except for the United Kingdom – indicating perhaps a
move for this group towards more schooling (and away from labour force participation)
during the economic upswing. This is consistent with the fact that the group’s prevalence

15
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

in the unemployment category has instead declined in most countries (except for Estonia
and Hungary, where it rose sizably). Employment outcomes for this group have been more
varied across countries. The group’s prevalence in full-time employment declined in
Germany, Spain and Portugal (and, less so, in Bulgaria) but it increased in Estonia,
Hungary, Poland and the United Kingdom. Instead, the group’s relative prevalence in part-
time employment increased everywhere with the exception of Portugal; while its relative
prevalence in self-employment declined everywhere with the exception of Poland. Thus,
on the whole, the economic upswing was associated with mixed developments for youth
in the eight countries examined here. In many cases there is increased labour market
attachment and inclusion, but in other cases the opposite is true, while exposure to part-
time employment, which is often associated with more precarious jobs (Campbell and
Price, 2016, Lewchuk, 2017) seems to have been more commonly increased across
countries.

Unlike youth, for the mature-age workforce (55-64 years old) there is a consistent decline
in their prevalence within the group of inactive across all countries, but also a highly
consistent increase in their prevalence into unemployment. Still, across all countries this
group shows increased prevalence into full-time employment (as well as into part-time
employment in most countries) and thus, overall, the economic upswing seems to have
been associated with improved labour market outcomes (inclusion, attachment) for this
group. That said, the uniform increase in the prevalence of this group in the unemployment
category, under conditions of fast-declining unemployment or full employment, shows that
mature-aged individuals experience heightened vulnerabilities with regard to job-finding,
which are independent of the business cycle – if not amplified by the economic upswing.

A similar result with regard to unemployment is also found for the group of 65-74 year-
olds – with all countries registering increased relative prevalence in unemployment for this
group. The patterns with regard to inactivity have been more varied, with the relative
prevalence in inactivity for this group being increased in Spain, Hungary and the United
Kingdom but falling elsewhere. Regarding employment outcomes, however, in all countries
but the UK we find that the economic upswing coincided with increases in the prevalence
of this group both in full-time and in part-time employment (and in many cases also in
self-employment). Thus, albeit quite paradoxically, the economic upswing seems to have
contributed to a significant improvement in the labour market position of individuals above
the standard retirement age in Europe.

16
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Table 3. Change in the prevalence of workforce characteristics across different labour market statuses
Country Labour market Female Youth 55-64 65-74 Foreign Low Large Bad
status educated household health
household
Bulgaria Inactive (+) + - (-) + - (-) +
Unemployed (-) - + + - + + +
Full-time employee (+) (-) + + (-) - - -
Part-time employee + + + + - + + -
Self-employed (-) - + + + - - -
Germany Inactive (-) + - (-) - + - (-)
Unemployed (+) - + + + - - +
Full-time employee (+) - + + + - (+) (+)
Part-time employee (-) (+) (-) + + - + (+)
Self-employed + - + + - - (-) (+)
Estonia Inactive (+) (+) - - + + (-) +
Unemployed - + + + - - + -
Full-time employee (+) + + + (-) (+) (+) (+)
Part-time employee (+) + (+) + + - (+) +
Self-employed - - + - - + + (+)
Spain Inactive - + - (+) - + (+) (+)
Unemployed + (-) + + - (-) - +
Full-time employee (+) - + (+) (+) - (-) -
Part-time employee (-) + + + (+) + + -
Self-employed + - - + + - + (-)
Hungary Inactive (+) + - + + (-) (+) +
Unemployed - + + + + + - +
Full-time employee (+) + + (-) - + (+) +
Part-time employee (+) + + + - + - +
Self-employed + - + + + + + -
Poland Inactive (-) (+) - (-) - (+) (-) (+)
Unemployed (-) - + + + (+) (+) +
Full-time employee (+) + + + + - (-) (-)
Part-time employee + + + (+) - + - (+)
Self-employed (+) + + + + - + +
Portugal Inactive (-) + - (-) + + (-) (+)
Unemployed (-) - + + - - + +
Full-time employee (+) - + + (-) - (+) (-)
Part-time employee - (-) - + - - - -
Self-employed + - - - + - (-) -
United Inactive
Kingdom (+) (-) - (+) - + - (+)
Unemployed - - + + - + (-) +
Full-time employee (-) + + - + - + -
Part-time employee (-) (+) (+) - + - (-) +
Self-employed (+) - (-) - + - + (+)
Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-2017.

17
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

With regard to foreign-born individuals, we see a more mixed picture across countries,
both with regard to the shifts in their prevalence across labour market statuses and with
regard to their share in the overall population of each country (as shown in Table A.1:
declining in Estonia, Spain, Poland and Portugal, but increasing in the remaining four
countries). From Table 3 it appears that the economic upswing coincided with a higher
relative prevalence of this group into inactivity in half of the countries (Bulgaria, Estonia,
Hungary and Portugal) but with a lower one in the remaining four (Germany, Spain, Poland
and the United Kingdom). The group’s relative prevalence in unemployment increased in
Germany, Hungary and Poland but declined elsewhere. In Germany and Poland
employment outcomes also improved (at least with regard to full-time employment), but
in Hungary the relative prevalence of foreign-born individuals declined both for full-time
and part-time employment. Notable employment gains were also registered for this group
in the United Kingdom.

Yet a different picture is obtained concerning the shares of low-educated individuals.


Generally, this group’s relative prevalence into employment (especially full-time) declined
(in Bulgaria, Germany, Spain, Poland, Portugal and the United Kingdom); while a robust
increase in the group’s prevalence in full-time employment was only recorded in Hungary.
Similarly, the group’s prevalence in inactivity increased in the majority of countries
(Germany, Estonia, Spain, Poland, Portugal and the United Kingdom); and declined
robustly only in Bulgaria. In relation to unemployment the picture is somewhat more
balanced, with the group’s prevalence in this having increased in Bulgaria, Hungary and
the United Kingdom (and, less so, in Poland) but having declined in Germany, Estonia and
Portugal (and, less so, in Spain). On the whole, then, the picture is more in the direction
of increased labour market detachment for this group (especially with regard to rising
inactivity and declining full-time employment), signifying the increased vulnerabilities
faced by this group also with regard to newly created jobs. When looking at specific
countries, however, it is only in the United Kingdom and Germany that we observe an
across-the-board deterioration of labour market outcomes, as captured by the prevalence
indicators, for the low-educated during the economic upswing. Inversely, across-the-board
improvements in employment outcomes are only observed in Hungary.

Concerning individuals with potentially limiting household characteristics (large


households and households with members reporting limiting illness), the results in Table
3 reveal more variations and generally less sizeable changes with the economic upswing.
Prevalence in full-time employment for these groups has increased in Germany, Estonia
and Hungary but it declined Bulgaria, Spain and Poland (as well as in Portugal and the
United Kingdom with regard to the group associated to households reporting poor health).
For this latter group, evidence of heightened labour market detachment is, however,

18
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

observed in the majority of cases: the group’s relative prevalence in inactivity increased
in all countries except Germany while its relative prevalence in unemployment increased
in all countries except Estonia. Although this may be taken to suggest that the overall
labour market vulnerability of this group has increased during the economic upswing, it is
also likely that the group’s increased labour market detachment has to do with individual
choice (e.g., as employment and income conditions improve thus allowing secondary-
income earners to withdraw their labour from the job market).

4.2 The distribution of labour market outcomes for different


groups

In this section we turn our attention to the inverse of the picture analysed in section 4.1.
Specifically, section 4.1 examined the prevalence of each group/characteristic within each
labour market status; this section examines instead the incidence of each labour market
status within each group/characteristic. For example, whereas the analysis of the previous
section was showing how intensively females are observed within the group of inactive
people, the analysis of this section aims to show, comparatively, how intensively inactivity
is observed within the group of females. We extent this question beyond the analysis of
labour market statuses, as was done previously, looking this time also, for those in
employment, at the incidence of outcomes within sectors and occupations – i.e., looking
additionally at the occupational and sectoral distribution of employment for the groups
that are of interest here (women, youth, etc).

To begin, Figure 1 presents an illustrative example using the youth category (individuals
15-24 years old) for Portugal and the UK. We present the full results from this type of
analysis in the Appendix (Figures A.1, A.2 and A.3.a–A.3.h), covering an analysis of the
distribution of labour market outcomes across three dimensions (labour market statuses,
occupations and sectors, respectively), two periods (2012 and 2017), eight countries and
eight group profiles. Given the breadth of information provided in these results, we discuss
these only selectively in the text of this section, focusing on the main messages that can
be drawn from the observed patterns. Instead, our main focus remains with the
examination of the shifts that may have occurred during the economic upswing in the
incidence of various aggregate labour market statuses within each of our groups of
individuals. For this, we employ the same presentational tool as in section 4.1 (see Table
4).

Returning to our illustrative example, as noted, Figure 1 presents the distribution of


outcomes for youth in Portugal and the United Kingdom across labour market statuses,
sectors and occupations. Each spider-graph contains four pieces of information: the

19
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

distribution for the group of interest (youths) in 2012 and 2017 (red line and blue lines,
respectively) and, comparatively, the corresponding distributions for the total reference
population (e.g., working-age population for the analysis of labour market statuses and
people in employment for the sectoral and occupational analyses – continuous and dotted
black lines, respectively, for the two years).

Starting with the case of labour market statuses (top row of spider-graphs in Figure 1), it
can be seen that their distributions for youths are particularly skewed in both countries.
Compared to the working-age population, youths have a particularly high incidence of
inactivity and lower incidence of full-time employment (especially in Portugal) and self-
employment. Compared to 2012, the economic upswing shifted youths towards more
inactivity but less unemployment (especially in the UK). In the UK youths experience also
higher rates of full-time employment, whereas in Portugal the incidence of full-time
employment declined with the economic upswing. In both countries, the incidence of part-
time employment and self-employment does not seem to have been altered significantly.
It is also worth mentioning that, in both countries, the incidence of part-time employment
(which is notably greater in the UK than in Portugal) is not much different for youths
compared to the working-age population, once the incidence of inactivity for each group
is taken into account. Taking this into account, the incidence of unemployment also does
not appear particularly high in this group, at least during the economic upswing. These
observations reflect rather well a key point concerning the discussion about youth
unemployment and youth employment barriers more generally: while the rates of
unemployment and part-time employment for youths are generally higher than the
average ones in most countries, almost invariably the incidence of unemployment and
part-time employment is not higher for youths compared to the general working-age
population.

20
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Figure 1. Distribution of labour market outcomes for youth in Portugal and the
UK

Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-2017, authors’ elaboration.

Concerning the occupational distribution of employment (second row of spider-graphs), in


Portugal this does not seem to be particularly different for youths than for the general
working-age population. Youths in Portugal have a higher incidence of employment in
Services & Sales (both in 2012 and in 2017) and a lower incidence of employment in
occupations with senior job positions such as Managers, Professionals and Technicians &
associated professionals. The economic upswing has altered this pattern very little,
although the data do show a reduction of employment for youths in Craft & related trades
and in Clerical support; and a relative expansion in Services & sales and Elementary
occupations. For the UK, the youth distribution of employment across occupations is rather
more markedly different to that of the general population, resembling more the distribution
found in Portugal. Youths are under-represented in senior, non-manual, and manual
occupations such as Managers, Professionals, Technicians & associated professionals and
Plant & machine operators; and they have higher incidence of employment in Clerical
support, Services & sales and Elementary occupations. More so than in Portugal, the

21
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

economic upswing in the UK had no visible effects on the occupational distribution of


employment for youths.

The case of the sectoral distribution of employment is much different, with some more
notable changes in the sectors of employment for youths in both countries between the
crisis period (2012) and the economic upswing (2017). In Portugal, the economic upswing
saw a relative expansion of youth employment in Education, Agriculture, Mining &
manufacturing, Arts & recreation and Accommodation & food; but also saw significant
declines in sectors such as Transport & storage, Financial & insurance activities,
Construction and Public administration. In the UK sectoral employment shifts were more
modest. Youth employment declined sizably in Agriculture and Mining & manufacturing
and less dramatically in Real estate, Arts & recreation and Financial & insurance activities;
but it expanded rather sizably in Transport & storage, Public administration, Construction
and Accommodation & food services. In both countries, youths seem to be under-
represented in only a sub-set of sectors, typically those associated with public sector
employment, such as Education, Health and Public administration. Inversely, in both
countries youth employment appears to be more heavily concentrated in Wholesale & retail
trade and in Accommodation & food, as well as in Real estate (in the UK) and Mining &
manufacturing (in Portugal). With the exception of the latter, this shows concentration of
youth generally in sectors with more precarious conditions of employment (more
temporary contracts and often less job and income security).

22
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Table 4. Change in the incidence of different labour market statuses for various workforce characteristics
Country Labour market status Working- Female Youth 55-64 65-74 Foreign Low Large Bad
age educated household health
household
Bulgaria Inactive - (-) + - (-) + - (-) (+)
Unemployed - - - (-) + - + + -
Full-time employee + (+) (+) + + (+) (-) - -
Part-time employee + + + + + - + + +
Self-employed + + - + + + + - -
Germany Inactive (-) (-) (+) - (-) - + - (-)
Unemployed - - - - + + - - (-)
Full-time employee (+) (+) - + + + - (+) (+)
Part-time employee + + + + + + - + +
Self-employed (-) + - + (+) - - (-) (-)
Estonia Inactive - (-) (-) - - (+) (+) - (+)
Unemployed - - - - - - - - -
Full-time employee + + + + + + + + +
Part-time employee + + + + + + - + +
Self-employed + + (+) + + - + + +
Spain Inactive - - (+) - (-) - + (-) (-)
Unemployed - (+) - + + - - - +
Full-time employee + + (+) + + + - (+) (-)
Part-time employee (+) (+) + + + + + + -
Self-employed + + - (-) + + - + (+)
Hungary Inactive - - (-) - (-) + - - (-)
Unemployed - - - - + - + - -
Full-time employee + + + + + (-) + + +
Part-time employee (-) (+) + + + - + - +
Self-employed + + - + + + + + (-)
Poland Inactive - (-) (+) - (-) - (+) (-) (-)
Unemployed - - - - + + - - -
Full-time employee + + + + + + - (+) (-)
Part-time employee - - (-) - - - + - -
Self-employed + + + + + + - + +
Portugal Inactive (-) (-) + - (-) + + (-) (+)
Unemployed - - - (-) + - - (+) (-)
Full-time employee (+) (+) - + + (+) - (+) (-)
Part-time employee + - (+) - + - - - -
Self-employed (+) + - - - + - (-) -
United Inactive (-) + (+) - + - + (-) +
Kingdom Unemployed - - - + + - + - -
Full-time employee (+) (-) + (+) - (+) - + -
Part-time employee (+) (-) (-) (-) - + - (-) (+)
Self-employed + + - + - + - + +
Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-2017.

23
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

As noted, we present the full picture concerning the distribution of labour market outcomes
for all countries and all groups of interest (females, youths, etc) in the Appendix. Rather
than discussing in detail these patterns here, however, we focus instead on the main
developments with regard to the distribution of labour market outcomes across the five
key employment statuses, for each country and group of interest, which we visually
summarise in Table 4. As in Table 3, Table 4 presents how the incidence of different labour
market statuses for various workforce characteristics changed, hence trying to depict both
the direction (plus/minus signs) and the intensity (in parentheses if the change is
proportionately of 5% or less) of the observed changes.

Starting with the first column, which depicts the results for the working-age population,
we see that unemployment declined strongly with the economic upswing in all countries
(almost by construction), but inactivity declined somewhat less strongly in Germany,
Portugal and the United Kingdom. In some countries (Bulgaria, Estonia) employment
expansion was strong across job types (full-time, part-time and self-employment), but in
other countries the trends for different job types were less homogenous. For example,
part-time employment increased robustly in Germany and Portugal, it increased rather
marginally in Spain and the UK, and it declined in Hungary and – especially – in Poland.
Germany saw also a decline in self-employment.

Interestingly, however, these patterns are not generally applicable to the different sub-
groups examined here. For females, we find again a general decline in the incidence of
inactivity, albeit to different degrees across countries and with the UK being an outlier with
inactivity among females rising during the economic upswing. A similar outlier behavior,
with regard to unemployment this time, is observed in the case of Spain, where the
incidence of unemployment for females actually rose, albeit marginally, during the
economic upswing. Concerning employment outcomes, employment gains for females
seem to have been more concentrated in non-standard jobs (part-time and self-
employment) in Bulgaria and Germany; they were more horizontal in Estonia, Spain and
Hungary; and they were more skewed towards self-employment in Poland, Portugal and
the UK (with all three countries registering sizable declines in the incidence of part-time
employment for females). Thus, on the whole, the economic upswing shows a general
improvement in the labour market position of females across all countries (less inactivity
and more employment), although with variable degrees of labour market attachment and
precariousness (declining/rising part-time employment).

In contrast, there was increased inactivity for youths in a number of countries (especially
Bulgaria and Poland, but also Germany, Spain, Poland and the UK), even if youth
unemployment declined robustly everywhere. For youths, employment gains were
registered in the sphere of part-time employment in Bulgaria, Germany, Estonia, Spain
and Hungary (and, less so, Portugal), while the incidence of self-employment declined in

24
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

the majority of countries and the incidence of full-time employment actually declined for
youths in Germany and Portugal. This picture of relative labour market detachment and
partial-only employment gains could be reflecting a pattern of choice among youths (e.g.,
opting to stay in education), although it is also possible that it signifies a diversification of
labour market risks as the labour market situation improves more robustly for other groups
of workers (presumably those with stronger labour market experience). This is in part
supported by the picture obtained for the group of mature working-age individuals, whose
labour market status shifts are more closely matching those seen for the full working-age
population (notable exceptions here include the rising incidence of unemployment in Spain
and the declining incidence of part-time employment and self-employment in Portugal).

Concerning the age-group of individuals above retirement age, we also observe a general
pattern of declining rates of inactivity, but in this case the incidence of unemployment
seems to have increased everywhere except for Estonia and the UK. The United Kingdom
is an outlier more generally for this group, with the rates of inactivity also rising during the
economic upswing and – uniquely – the rates of employment declining for this group across
all job types (against an almost universal increase of employment rates for this group in
all other countries). This pattern is somewhat counter-intuitive, as it suggests a heightened
degree of labour market attachment for ‘retirees’ post-crisis compared to the period during
the crisis – implying that this group of individuals behaved more in line with the
‘discouraged worker effect’ than the added worker effect’ during the crisis.

Aspects of both of these effects (discouraged and added worker) are also observed for the
two groups associated with the household situation of the individual (large households and
households with members reporting a limited illness – last two columns of Table 4). Here
patterns are much more difficult to decipher, reflecting perhaps the more idiosynchratic
nature of employment choices – and labour market position – of these groups across
countries. Also diverse are the patterns found for the remaining two groups, those of
foreign-born individuals and individuals with low education. For foreign-born individuals we
find that inactivity rose during the economic upswing in Bulgaria, Hungary and Portugal
(and, less so, in Estonia); unemployment declined almost everywhere (except for and
Poland, where it increased significantly); full-time employment increased everywhere more
robustly; while part-time employment declined in Bulgaria, Hungary, Poland and Portugal.
On the other hand, for the low-educated group, inactivity rose in Germany, Estonia, Spain,
Poland, Portugal and the UK (and declined in Bulgaria and Hungary); unemployment
increased in Bulgaria, Hungary and the UK (but declined quite robustly elsewhere); while

25
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

employment – across job types – declined sizably in Germany, Portugal and the United
Kingdom (but increased robustly in Hungary).

On the whole, it appears difficult to derive pattern that could be generalised across
countries for each of the groups considered here. The picture is more homogenous for the
mature working-age individuals, females and the youth (with increased labour market
attachment for the first two groups and more ambivalent patterns of labour market
attachment for youths); while for the remaining groups the patterns appear rather
idiosyncratic and country/case-specific.

4.3 Changes in the incidence of labour market statuses –


multivariate analysis

Notwithstanding the interesting patterns discussed in the previous sub-section, the


question emerges as to whether the observed changes are (a) significant in a statistical
sense and (b) conditional on changes that may have occurred intersectionally in other
characteristics. To examine this, this sub-section reports the results from a multivariate
multinomial logit analysis – for the case of labour market statuses only – which provides
direct estimates of how the probability of each labour market status changed for each
group (or characteristic) between 2012 and 2017 controlling for each individual’s range of
characteristics. We report these results in Figures 2a-2h. Each figure reports, for a given
individual characteristic (e.g., females), the estimated change in the conditional probability
of observing a particular labour market outcome, for each country and labour market
outcome, between the years 2012 and 2017. 7

7
Formally, the figures report the estimated marginal effects (i.e., the change in the predicted probability of an outcome if we move from year
2012 to year 2017, keeping everything else constant at mean sample values) together with their associated 95% confidence intervals (which in
turn give us a measure of the reliability of the precision of the indicator, i.e., the confidence one can have on the accuracy of the reported
estimate). For the central estimates (marginal effects), values closer to the top of the graph (e.g., near 0.05) show that in 2017 the probability of
the particular outcome for the particular group in the particular country was up to 5% higher than the same probability observed in 2012; while
values below the zero line show that this probability declined between 2012 and 2017. When confidence intervals cross the zero line, this shows
that the reported estimate is not statistically different from zero, i.e., that although the value of the estimate may be non-zero, our statistical
confidence in this value is very low (at non-acceptable levels statistically) and thus our prediction is that no (statistically significant) change has
occurred between 2012 and 2017. In this sense, statistical confidence in the reported estimates is only achieved for estimates whose confidence
intervals do not intersect the zero-line.
26
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-2017, authors’ elaboration.

Starting with the case of females (Figure 2a), we see that, when controlling simultaneously
for the full set of individual characteristics, observed differences between 2012 and 2017
– expressed now relative to the base category of males – are much more muted, while in
some cases they even reverse compared to the picture shown in Table 4. This is the case,
for example, in Hungary, where we observe – ceteris paribus – a rise in inactivity for
females with the economic upswing, which is statistically significant (in technical terms,
the 95% confidence interval lies above the zero-line). This means that when we control for
all other characteristics, women actually had a higher chance to become inactive
(compared to men) in 2017 than in 2012. For inactivity, statistically significant declines
are only observed for Germany and Spain (and marginally for Estonia); while increases in
full-time employment appear robust only in Germany, Spain and Poland (and, again,
marginally so in Estonia, with Hungary showing a statistically significant drop in female
full-time employment); and it is only in Spain that we observe a statistically significant
decline in female unemployment.

27
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Figure 2b. Estimated marginal effects for people aged 15-24 years old by country
and labour market status

Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-2017, authors’ elaboration.

Concerning the case of young individuals (Figure 2b), the results from the multivariate
analysis also present a more qualified picture compared to the one depicted in Table 4. As
can be seen, in this case the evidence of generalized increased inactivity is replicated, but
the increase is only statistically significant in Estonia, Hungary, Portugal and the UK.
Instead, the probability of full-time employment has declined everywhere, but only in
Spain, Poland and the UK the decline is statistically significant; while other changes are
neither statistically significant nor consistent across countries. Thus, the pattern prevailing
during the economic upswing is one of labour market detachment for youth with the crisis
(compared to the other age-groups in the workforce). In other words, although the
economic opportunities of this group may have been improving, in line with the overall
improvement in employment conditions in the countries under study, the performance of
young individuals is not improved disproportionately and if anything it has deteriorated
relative to other groups.

28
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Figure 2c. Estimated marginal effects for people aged 55-64 years old by country
and labour market status

Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-2017, authors’ elaboration.

Figure 2d. Estimated marginal effects for people aged 65-74 years old by country
and labour market status

Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-2017, authors’ elaboration.

29
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

For mature working-age individuals (again, relative to the other age groups) the changes
observed in the economic upswing are much stronger. Inactivity for this group has declined
everywhere quite significantly, both in statistical and in economic terms (by between 5%
and 15%). Unemployment has instead increased everywhere, albeit statistically
significantly only in Bulgaria, Spain, Hungary, Poland and Portugal. At the same time the
probability of full-time employment has also increased in a statistical sense in the majority
of countries (all except Spain); while the patterns for part-time employment and self-
employment are also in the same direction (albeit weaker statistically). A broadly similar
picture is also obtained for the group of post-retirement age individuals (Figure 2d).

Figure 2e. Estimated marginal effects for foreign-born individuals by country and
labour market status

Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-2017, authors’ elaboration.

In contrast, the shifts observed with the economic upswing for those foreign-born (Figure
2e) are much smaller and generally not statistically significant. Exceptions to this are the
sizable decline in the probability of unemployment in Bulgaria, the much more marginal
decline in unemployment in Estonia, Spain and the UK, the decline in inactivity in Germany
and, inversely, the increase in inactivity in Estonia. Other changes, for example the reduced
probabilities for full-time employment for this group in Hungary and Poland, are not
statistically significant. Overall, the economic upswing does not seem to have
systematically altered the degree of labour market attachment for foreign-born individuals
relative to the native population. As the economic situation of natives has improved, this

30
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

also suggests an absolute – albeit not disproportional – improvement of labour market


opportunities also for those foreign-born.

Figure 2f. Estimated marginal effects for low-education individuals by country


and labour market status

Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-2017, authors’ elaboration.

In turn, among those low-educated (Figure 2f), the key patterns are as follows: a sizeable
increase in the probability of inactivity, with a simultaneous decline in the probability of
unemployment in Germany; an increase in inactivity also in Spain, this time accompanied
by statistically significant declines in full-time employment and in self-employment; a
substantial decline in inactivity for this group in Hungary with a statistically significant shift
towards more full-time employment; and increases in the probability of inactivity for this
group in Poland and Portugal with concurrent deterioration in the probabilities for full-time
employment and self-employment. Thus, for this group, inactivity increased statistically
significantly in four of the eight countries and full-time employment declined significantly
in three countries, broadly signifying a relative deterioration of economic opportunities for
this group with the economic upswing. In Germany, however, the main development is the
decline in the unemployment probability – and thus heightened labour market attachment
– for this group.

31
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Figure 2g. Estimated marginal effects for individuals in large households by


country and labour market status

Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-2017, authors’ elaboration.

As was the case with the univariate analysis of Table 4, the estimated shifts in the
multivariate analysis for the groups of individuals defined by their household characteristics
(large households – Figure 2g; and households with members reporting limiting illness –
Figure 2h) appear less systematic and on the whole not statistically significant. For those
in large households we find a statistically significant decline in inactivity in Germany and
Hungary; a statistically significant decline in unemployment and increase in full-time
employment in Hungary; and a statistically significant increase of part-time employment
in Germany and of full-time employment in the UK. For individuals living in households
with limiting health conditions, the main pattern is the observed increase in the probability
of inactivity in Estonia, Hungary, Portugal and the UK; and a broad trend of declining full-
time employment, which is however statistically significant only in Spain, Poland and the
UK. It thus appears that, broadly speaking, the labour market situation of individuals
belonging to these two groups has in fact deteriorated with the economic upswing, relative
to other groups, a result that is most probably reflecting the fact that for these individuals
labour market participation is driven mainly by need rather than by opportunity (i.e., it
expands, in relative terms during recessions and contracts, relative to others, when
economic opportunities become more ample).

32
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Figure 2h. Estimated marginal effects for individuals in households reporting


limiting health by country and labour market status

Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-2017, authors’ elaboration.

Concluding this analysis, the following inferences can perhaps be derived. On the whole,
despite the rather sizable changes observed during the economic upswing in the eight EU28
labour markets studied here, the labour market position of individuals belonging to
generally considered vulnerable groups has not altered dramatically. In some cases
(females, mature working-age individuals), notable improvements have been recorded.
For other groups, however, improvements in labour market attachment have either been
broadly in line with those in the general population (mainly the youth and those foreign-
born) or have been worse than those of the general population – suggesting, in this latter
case, a dynamic whereby economic upswings represent an opportunity for individuals in
particular vulnerable groups to detach themselves from the labour market. In any case,
the general message is that the economic upswing is not sufficient, in and by itself, to
improve the relative labour market position of vulnerable groups in the economy. Examples
of improvement exist, but they are idiosyncratic to a large extent, concerning specific
labour market outcomes for specific groups in specific countries. As was seen in our
discussion in section 4.2, and is more fully demonstrated in the Appendix, the same broadly
applies also with regard to the occupational and sectoral distribution of employment for
these groups.

33
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

5. A Closer Look – Labour Market Outcomes for Hypothetical


Profiles

The evidence collected from the analysis of individual characteristics associated with labour
market vulnerability suggests that, on the whole, risks of labour market detachment
(inactivity, unemployment, part-time employment) are rather ‘sticky’ in the sense that
they largely persist even in conditions of economic upswings. In this section we turn our
attention to examining whether this also holds true for types of individuals linked to
particular labour market risks, as identified through the study of employment barriers and
labour market vulnerability under the Faces of Joblessness project. Drawing on the
estimates from the multinomial logit analysis presented in section 4.3, we reconstruct
some key profiles identified in this project and derive their associated predicted
probabilities of inactivity, unemployment, full-time employment, part-time employment
and self-employment, comparing how these probabilities shifted during the economic
upswing. We focus on a sub-set of our study countries – namely Estonia, Spain and

Portugal – as detailed profiles for the remaining countries are not currently available. 8

5.1 Estonia

For Estonia, the key three profiles identified in the Faces of Joblessness project where as
follows: (a) Older adults with health limitations, low skills and limited work experience; (b)
Working poor males; and (c) Unskilled mothers with care responsibilities and limited work
experience.

As should be expected, the first group (henceforth, ‘older adults’) had a very high incident
of labour market detachment in 2012, with predicted probabilities of inactivity and
unemployment near 60% and 30%, respectively. The probabilities of full-time
employment, part-time employment and self-employment were all very low and
cumulatively just above 10%. As the economy heated up, by 2017, the incidence of
employment for this group increased only marginally (at the backdrop of a 42% decline in
unemployment nationally – see Table 1), with full-time employment rising but remaining
below 10%, and part-time employment and self-employment also remaining below 5%

8
The Faces of Joblessness project has also produced specific profiles for Lithuania, Italy and Ireland. Of them,
Italy fails to meet either of our selection criteria, while Ireland and Lithuania meet the rate-of-change criterion
but not the unemployment-level criterion.

34
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

each (all figures are in terms of predicted probabilities for the average group
characteristics). A sizeable shift occurred, however, in relation to the statuses with low
labour market attachment. While the predicted probability of unemployment for the older
adults group declined by some 20 percentage points (from near 30% to near 10%), its
predicted probability of inactivity increased significantly (from 60% to well-over 70%). In
relative terms the decline in the predicted probability for unemployment appears the most
sizeable, although the increase in the incidence (predicted probability) of part-time
employment is also quite dramatic (over-doubling between 2012 and 2017). It thus seems
that, for this group, the economic upswing facilitated two types of shifts: on the one hand,
some employment gains, across all types of jobs, with a sizeable decline in unemployment;
on the other hand, a sizeable reduction in labour market attachment, presumably signifying
intensified exclusion (including self-selection out of the job market) as economic conditions
improved and individuals with more competitive skills started becoming more successful in
the job market.

35
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-2017.

Inversely, the second group (working poor males), already had a high prevalence of
employment in 2012 (predicted probability of full-time employment at near 55%). Given
the definition of this group, however, the prevalence of inactivity and unemployment was
also quite high, both standing at double-digit numbers in 2012 and jointly accounting for
some 40% of individuals in this group. The economic upswing had by 2017 changed this
picture quite significantly too, but this time positive labour market outcomes were much
more visible. Inactivity and unemployment (in terms of predicted probabilities) both fell –
to below 20% and below 5%, respectively. At the same time there were increases in all
types of employment, especially regarding the predicted probability for full-time
employment. Proportionately, however, the largest increases were registered for part-time
36
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

and self-employment, both of which almost doubled, with the predicted probability for self-
employment rising to near 10% in 2017. For this group of individuals, the economic
upswing was characterised by increased labour market attachment, which however was
disproportionately directed towards less standard forms of employment.

The situation for the group related to unskilled mothers with care responsibilities presents
instead a more positive picture. In terms of predicted probabilities for the average group
characteristics, the incidence of inactivity fell even more dramatically, declining from just
short of 40% in 2012 to just above 20% in 2017. The decline in the predicted probability
of unemployment was even more sizeable, from near 20% in 2012 to less than 5% in
2017. Even more dramatic, by over 25 percentage points, was the increase in the predicted
probability of full-time employment that reached a value of over 60% in 2017. Importantly,
employment gains for this group do not seem to have been associated with increased
precariousness, as the probabilities of part-time employment and self-employment have
not shifted – each remaining at below 10% and 5%, respectively. Thus, the economic
upswing saw increased labour market attachment and inclusion for this profile of
individuals, with a relative decline both of unemployment/inactivity and of non-standard
forms of employment.

Overall, the economic upswing seems to have facilitated increased labour market
attachment in Estonia, at least for some vulnerable groups as they were identified in the
Faces of Joblessness project. In some cases, this seems to have worked through a
proportionately more enhanced access to non-standard jobs. But even for those, the
incidence of unemployment declined very notably and significant gains were recorded in
terms of access to full-time employment (especially for unskilled mothers). Thus, it seems
that demand factors, as reflected by the overall decline, by 42%, in national unemployment
during the economic upswing, play an important role in re-engaging vulnerable individuals
into the labour market. The evidence of increased labour market detachment for older
workers with health limitations, low skills and limited work experience does not necessarily
refute this conclusion, as it possibly reflects the fact that such individuals were either able
to get into jobs (declining unemployment) or to detach themselves, perhaps voluntarily,
from the labour market.

37
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

5.2 Spain

Reflecting the specific conditions of the crisis in the country, the profile of groups facing
labour market vulnerability in Spain are different. In this case, the Faces of Joblessness
project identified two main vulnerability groups: (a) Young and prime-age adults with low
work experience; and (b) Low-skilled women in unstable jobs.

The first of these groups had, in 2012, exceptionally high predicted probabilities of
unemployment (over 60%) but – given the age composition of this group, which includes
also prime-age males – not particularly high probabilities for inactivity (below 10%).
Among the three employment outcomes, the incidence of full-time employment was the
highest, but predicted probabilities for part-time employment were also high, at
comparable levels to those for full-time employment. Self-employment was also relatively
high in proportional terms. By 2017 the situation for this group had shifted somehow,
although not remarkably.

38
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-2017.

The probabilities of inactivity and unemployment declined, with the latter remaining still at
very high levels (above 50%). Importantly, employment expansion seems to have come
predominantly through full-time jobs, with the incidence for this having increased by over
5 percentage points. Increases in part-time employment and in self-employment were
much more modest.

A somewhat different picture is obtained for the second group, of low-skilled women in
(and out of) unstable jobs. For them, predicted probabilities for unemployment were also
high in 2012, although significantly lower than in the first group, while predicted
probabilities for inactivity were significantly higher (at over 15%). With the improvement
in the economic situation by 2017, the predicted probability for inactivity declined
noticeably, falling below 15% but, crucially, the predicted probability of unemployment for
this group remained practically unchanged, despite the overall decline in unemployment
rates nationally by some 7.5 percentage points (see Table 1). Despite this, the economic
upswing is in fact associated with employment gains for this group. The predicted
probability of full-time employment reached 30% and the predicted probabilities of part-
time employment and self-employment increased more modestly (reaching levels near
17% and 7% respectively).

Thus, the picture for Spain suggests a partial labour market improvement. Some degree
of labour market re-attachment is evident, with declining inactivity and a sizeable part of

39
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

the employment gains coming from increased full-time employment. Given the fact that
the labour market conditions in Spain remain far from full employment (the national
unemployment rate in 2017 was at 17.2% – see Table 1), labour market vulnerability has
remained high for both of the groups considered here, with unemployment persisting at
levels well above the national average. It thus appears that, besides the problems of
effective demand and labour market disequilibrium, other structural factors may also be
playing a role in creating barriers for the identified groups in the labour market.

5.3 Portugal

The vulnerable groups identified in the Faces of Joblessness project for the case of Portugal
are in some respects narrower than in the other two countries considered here (Estonia
and Spain). They refer to: (a) Prime-age individuals with low education and scarce job
opportunities; (b) Young individuals without any work experience and with scarce job
opportunities; and (c) Young individuals with limited recent work experience and low skills.

For the first group (prime-age individuals), labour market inclusion has been reasonably
high already during the crisis. In 2012, the predicted probability of full-time employment
was above 50%, while the predicted incidence of non-standard employment was in relative
terms quite low (below 5% and below 10% for part-time employment and for self-
employment, respectively). At the same time, given the age profile of this group, inactivity
rates were also relatively high (at near 10%), while the predicted probability of
unemployment was much higher than the national average, at around 25%. The

40
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

substantial drop in national unemployment (by some 43% - see Table 1) seems to have
affected little, in proportional terms, this group. By 2017, the group had similar (and
marginally increased) predicted probabilities for inactivity, while the predicted probability
of unemployment remained high at near 20%. The predicted probability for full-time
employment increased only marginally, remaining in the area of 55% and, while the
predicted probability for part-time employment increased proportionately by a higher rate,
it remained overall at low levels, below 10% – similar to the predicted probability for self-
employment.

For the group of young individuals with scarce job opportunities things were (almost by
definition) much worse during the crisis. In 2012 the predicted unemployment probability
was over 50% and, while the predicted probability for inactivity was by any standard very
low (at below 3%), the predicted probability for full-time employment was also low in
relative terms (at around 35%). The predicted probabilities for part-time employment and
for self-employment were, in proportional terms, higher than those for other groups,
standing jointly at around 28% of total employment for this group. Crucially, the
improvement in the economic situation of the country more recently does not seem to have
been translated into significant shifts in labour market attachment for this group. Rates of
inactivity have remained stable, although quite low, while the predicted unemployment
probability declined only marginally remaining at levels near 50%. The predicted incidence
of full-time employment in fact declined and a shift towards part-time employment and
self-employment was observed instead – with the latter rising to over 10% of the total for
the group – signifying a heightened degree of labour market precariousness despite the
economic upswing.

41
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-2017.

The other group of young individuals had a somewhat different labour market situation in
2012. For them, predicted rates of inactivity were even lower and this time the predicted
incidence of unemployment was also relatively low (at around 18%). Instead, the predicted
incidence of full-time employment was very high (at above 70%) and the rates of part-
time employment and of self-employment were low (at below 5% each). As with the
previous group, however, the economic upswing is not associated with a significant
improvement in the labour market situation of this group. Predicted rates of unemployment
declined somewhat, but remained above the national average; while the main employment
gains seem to have come in the form of part-time employment which increased
proportionately by over 30% (while still remaining at below 10%, however).

Thus, overall, the picture portrayed for Portugal is somewhat subdued. Vulnerable groups
have benefitted from the nation-wide decline in unemployment, but less-than-
proportionately; all groups experienced employment gains, but for most of them this came
more in the form of non-standard employment (part-time jobs or self-employment); and
for the groups that already had a significant degree of labour market detachment, overall
labour market attachment did not increase dramatically despite the notable improvement
in the economic situation of the country in the 2012-2017 period. Thus, it seems that in
this case employment barriers and labour market vulnerability is more structural and not
solely attributable to the phase of the business cycle.

42
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

6. Conclusions

The aim of this Research Note was to offer an in-depth examination of how the labour
market position of individuals with specific characteristics associated to employment
barriers and labour market vulnerability changed in EU countries which have experienced
fast declining unemployment since the crisis and/or conditions close to full-employment.
To do so, we first identified eight countries for which unemployment conditions were
significantly improved (i.e., decline in unemployment rates by more than 40%, and/or
unemployment rates were at below 5%) between 2012 and 2017 (the latest year with
information on individuals from the EU-SILC survey). Then, we examined how individuals
falling within eight groups of vulnerability characteristics (females, youths, mature
working-age individuals, individuals above retirement age, foreign-born individuals, those
low educated, and those living in large households or in households with members
reporting limiting health) performed in those countries’ labour markets during the
economic upswing. Complementarily to this investigation, which essentially sought to
answer the question “how do individuals from vulnerable groups fare in the labour market
during economic upswings” (in other words, do they find jobs?), we also examined how
those individuals have fared with regard to their sectoral and occupational distribution of
employment (asking, in other words, where do they find jobs?). These questions were
examined through an extensive analysis of frequencies from observational data
(descriptive analysis), that included both an investigation into the prevalence of each
individual group/characteristic within each labour market status and an investigation into
the incidence of each labour market status (outcome) within each individual group.

Acknowledging the importance of the issue of the intersectionality of disadvantage, in the


labour market and beyond, we additionally proceeded with a more detailed examination of
labour market outcomes for vulnerable groups, in two ways. First, by estimating for each
country a multinomial logit model of the labour market status/outcome of each individual
in each of the two years under analysis, i.e. 2012 and 2017, corresponding to the periods
of crisis and economic upswing, respectively. In other words, the probability for each
individual to be inactive, unemployed, full-time employed, part-time employed or self-
employed was modelled as a function of a rich set of individual and household
characteristics that were allowed to bear different effects on the individual’s labour market
status in each country, in each of the two years. Based on this, we subsequently examined
how the conditional probability of each labour market status for each of the eight
characteristics of interest changed between 2012 and 2017 in each of the eight countries
studied here. To do so, we measured the marginal effects (valuated at mean sample
43
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

values) associated to the interaction term between a year 2017 binary indicator and each
characteristic of interest. This allowed us to examine not only whether the observed
changes from the descriptive analysis hold once we control for each individual’s full range
of characteristics (e.g., examining the shifts in a female’s labour market status conditional
on her age, level of education, marital status, household composition, etc) but also whether
they are significant in a statistical sense. Second, we used the multinomial logit model
estimates, derived as conditional fitted values (predicted probabilities, also evaluated at
mean sample values) for specific individual profiles, i.e., set of individual and household
characteristics, that were borrowed from the Faces of Joblessness project. As discussed in
Section 2, that project derived a number of individual profiles carrying an intersection of
characteristics that make them particularly vulnerable in the labour market, i.e., profiles
of individuals facing significant employment barriers, where each profile was country-
specific, directly derived from observational data from each of the six countries covered in
the project. Three of those countries (Estonia, Spain and Portugal) were also in our group
of eight countries with superior labour market performance. For them, it was thus possible
to examine how individuals with profiles similar to the ones identified in the Faces of
Joblessness studies have performed, comparatively, in 2012 and 2017 with regard to their
labour market status. This offered a unique insight into how the economic upswing may
have contributed to the economic position of individuals taking into account the
intersectionality of characteristics associated to labour market vulnerability.

The overall picture emanating from our analysis – along all the dimensions summarized
above – is one of varied patterns and trends across groups and countries. Starting with
the measures of prevalence, which show how much each group is proportionately
over/under-represented in each labour market outcome, we found for example instances
of increased prevalence of inactivity and unemployment in some countries but reduced
prevalence in others – with differences cutting across country groups (north-south, east-
west, etc). Concerning youths, for example, we observed a broadly homogenous trend
across countries with regard to patterns of labour market attachment (declining prevalence
in unemployment but increasing in inactivity) but at the same time rather great cross-
country variation with regard to the types of jobs in which young individuals exhibit higher
prevalence (full time, part time and self-employment). An inverse pattern of increased
prevalence in unemployment but declining prevalence in inactivity was found for the two
older age groups (mature working age and post retirement age) although no systematic
patterns across countries were found here either. More generally, the picture deriving from
the prevalence measures suggests that the economic upswing did not lead to a generalised

44
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

improvement in the labour market position of all vulnerable groups. As unemployment


declined nationally, some of these groups (e.g., youths) detached themselves more
intensely from the labour market, signifying perhaps a choice towards more education;
while others (e.g., the low-educated) experienced a higher prevalence of unemployment,
reflecting apparently their diminished competitive position with regard to job-finding.

The analysis concerning the incidence of labour market outcomes for each group of
individuals produced broadly similar results. Females and youths have a significantly higher
incidence of inactivity and lower incidence of full-time employment everywhere, as do most
of the groups considered in the analysis. With the economic upswing, inactivity seems to
have declined for females, those living in large households and those over 55 years old in
the majority of countries but it increased for youths and the low-educated in the majority
of cases. For those foreign-born or those living in households with limiting health the results
are more mixed. On the other hand, the incidence of unemployment declined almost
everywhere for females, the youth, the mature working-age individuals, those foreign-
born, those living in households with limiting health and, with a few exceptions, for the
low-educated; but it increased almost universally for the older age-group. On the whole,
then, the evidence concerning the question as to how the economic upswing affected the
labour market attachment of individuals belonging to vulnerable groups is somewhat
ambivalent: most groups experienced improvements with regard to unemployment and
inactivity, but in most cases this improvement was below what was seen for the general
population while some groups (especially those with low labour market experience – the
youth – and those with low skills – the low educated) experienced enhanced labour market
detachment (rising inactivity).

The same applies broadly with regard to the employment outcomes for such individuals.
While in the general population we observe a universal increase in the incidence of full-
time employment (and, less so, of part-time employment and self-employment),
employment gains (especially full-time) are weaker for females, the youth, and for those
foreign-born; while for some groups our results show a deteriorating employment
performance in 2017. In most countries, this is the case for low educated and those living
in households with limiting health; for the youth in Portugal and Germany; and for the
older workers (past retirement age) in the United Kingdom. A more thorough multivariate
analysis conditioning these effects on the distribution of all other individual characteristics,
returns a rather more muted picture of labour market attachment for the groups examined
here. Employment gains are restricted to the cases of Germany, Spain and Poland for
females and the past-retirement-age individuals; they are more universal for the group of
45
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

mature working-age individuals (who also experience a robust decline in inactivity); they
are always statistically zero or negative for the low-educated (with the exception of
Hungary), those living in large households (with the exception of Hungary and the UK),
those living in households with limiting health and those foreign-born; while for the youth
the majority of countries show a decline in full-time employment (and in employment
overall). In other words, when controlling for the characteristics of individuals which may
make them more successful in the labour market (both in terms of labour market
attachment – inactivity, unemployment – and in terms of types of employment), our
evidence shows that the economic upswing was combined with a general deterioration of
the position of individuals belonging to the main vulnerability groups examined here. In
absolute terms, many of them experienced an improvement in the labour market position;
but the improvement in the economy overall – and for groups with lower perceived
vulnerability – was much superior.

Reflecting on this point, our analysis went one step further and examined for a sub-set of
countries, for which it was possible to do so, the changing labour market position of
individuals belonging to specific labour market vulnerability profiles as these have been
identified through the Faces of Joblessness project. For these groups/profiles, the overall
picture was rather mixed and dependent on the whole on the particularities of each
group/profile. For ‘older adults with health limitations, low skills and limited work
experience’ in Estonia, the evidence suggests a pattern of polarisation, with some
experiencing improved employment probabilities but others experiencing increased labour
market detachment (inactivity). In the same country, the group of ‘working poor males’
experienced instead improved labour market attachment, but with employment gains
predominantly directed towards less standard jobs (part-time and self-employment). The
third vulnerable group, i.e. ‘unskilled mothers with care responsibilities and limited work
experience’, saw both increased labour market attachment and improvements in their job
position (towards full-time employment). In the case of Spain, the two vulnerability
groups/profiles identified there had more modest changes in their economic situation –
reflecting perhaps the fact that, despite the dramatic decline of unemployment in the
country between 2012 and 2017, unemployment remained high also in 2017. Both groups
(‘young and prime-age adults with low work experience’ and ‘low-skilled women in unstable
jobs’) had declining probabilities of inactivity and increased probabilities for full-time
employment, but these improvements were quite marginal and nowhere near those seen
across the labour market for the country as a whole. A similar conclusion is drawn for the
vulnerability profiles of Portugal. Thus, whereas some notable improvement was recorded

46
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

in Estonia, suggesting perhaps that the economic upswing – and thus demand conditions
– are sufficient to uplift the labour market attachment and employment performance of
vulnerable groups, individuals in Spain and Portugal benefitted much less from the
economic upswing suggesting in turn more structural barriers to employment and labour
market attachment for these groups.

Our analysis examined additionally the sectoral and employment distribution for specific
characteristics associated to labour market vulnerability. On the whole, the shifts observed
in these cases (sectoral and occupational distribution of employment) were much more
modest and rather idiosyncratic – varying greatly across cases (countries, characteristics)
both in a static (picture in any particular year) and in a dynamic sense (shifts between
years). As with the main thrust of our analysis, the examination of sectoral and
occupational shifts suggests on the whole the absence of any radical changes that would
have altered significantly the employment patterns and job positions of individuals carrying
the characteristics that were of focus here. The economic upswing helped some individuals
carrying these characteristics to improve their labour market position, but the patterns of
employment (with regard to job-types, as well as sectors and occupations of employment)
have been rather ‘sticky’, even as the countries examined here moved closer to conditions
of full employment.

The general conclusion emanating from the large set of empirical evidence put together in
this Research Note is thus that improvements in the economic cycle are broadly helpful but
nowhere near sufficient by themselves to address the problems of vulnerability, labour
market detachment/exclusion and job precariousness observed for specific groups of
individuals. For some groups, the economic upswing has even created conditions – either
in the form of ‘pull’ factors or in the form of ‘push’ factors – contributing to enhanced labour
market detachment (especially for the low educated and, in some cases, for the youth).

47
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

7. BIBLIOGRAPHY

Ahmed, A.M., Andersson, L. and Hammarstedt, M., 2012. Does age matter for
employability? A field experiment on ageism in the Swedish labour market. Applied
Economics Letters, 19(4), pp.403-406.

Arntz, M., Dlugosz, S. and Wilke, R.A., 2017. The sorting of female careers after first birth:
A competing risks analysis of maternity leave duration. Oxford bulletin of economics and
statistics, 79(5), pp.689-716.

Battu, H. and Sloane, P., 2000. Overeducation and crowding out in Britain. The
overeducated worker, pp.157-174.

Bell, D. N. F., & Blanchflower, D. G. (2015). Youth unemployment in Greece: measuring


the challenge. IZA Journal of European Labor Studies, 4(1), 1-25.

Bertrand, M. and S.Mullainathan (2004), ‘Are Emily and Greg More Employable than
Lakisha and Jamal? A Field Experiment on Labor Market Discrimination’, American
Economic Review94, 991–1013.

Bonifazi, C. and Marini, C., 2014. The impact of the economic crisis on foreigners in the
Italian labour market. Journal of Ethnic and Migration Studies, 40(3), pp.493-511.

Bredtmann, J., Otten, S. and Rulff, C., 2018. Husband’s unemployment and wife’s labor
supply: the added worker effect across Europe. ILR Review, 71(5), pp.1201-1231.

Browne J., Immervoll H., Fernandez R., Neumann D., Pacifico D. and Thévenot C. (2018),
Faces of Joblessness in Estonia: A People-centred perspective on employment barriers and
policies, OECD Social, Employment and Migration Working Paper No206, OECD, Paris.

Campbell, I. and Price, R., 2016. Precarious work and precarious workers: Towards an
improved conceptualisation. The Economic and Labour Relations Review, 27(3), pp.314-
332.

Cebolla-Boado, H., Miyar-Busto, M. and Muñoz-Comet, J., 2015. Is the Spanish recession
increasing inequality? Male migrant-native differences in educational returns against
unemployment. Journal of Ethnic and Migration Studies, 41(5), pp.710-728.

Cebrián, I. and Moreno, G., 2015. The effects of gender differences in career interruptions
on the gender wage gap in Spain. Feminist Economics, 21(4), pp.1-27.
48
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Cheung, C.K., Kam, P.K. and Man-hung Ngan, R., 2011. Age discrimination in the labour
market from the perspectives of employers and older workers. International Social Work,
54(1), pp.118-136.

Congregado, E., Golpe, A.A. and van Stel, A., 2011. Exploring the big jump in the Spanish
unemployment rate: Evidence on an ‘added-worker’effect. Economic Modelling, 28(3),
pp.1099-1105.

Dolado, J. J., Jansen, M., Felgueroso, F., Fuentes, A., & Wölfl, A. (2013). Youth Labour
Market Performance in Spain and its Determinants: A Micro-Level Perspective. OECD
Economics Department Working Papers, No. 1039, OECD Publishing.

Düll N., Thévenot C., Immervoll H., Browne J., Fernandez R., Neumann D. and Pacifico D.
(2018), Faces of Joblessness in Portugal: A People-centred perspective on employment
barriers and policies, OECD Social, Employment and Migration Working Paper No210,
OECD, Paris.

Fernandez R., Immervoll H., Pacifico D., Browne J., Neumann D. and Thévenot C. (2018),
Faces of Joblessness in Spain: A People-centred perspective on employment barriers and
policies, OECD Social, Employment and Migration Working Paper No207, OECD, Paris.

Fernandez, R., Immervoll H., Pacifico D. and Thévenot C. (2016), Faces of Joblessness:
Characterising Employment Barriers to Inform Policy, OECD Social, Employment and
Migration Working Paper No192.

Ferrant, G., Pesando, L.M. and Nowacka, K., 2014. Unpaid Care Work: The missing link in
the analysis of gender gaps in labour outcomes. OECD Development Centre. http://www.
oecd. org/dev/development-gender/Unpaid_care_work. pdf. Accessed April, 20, p.2016.

Gallie, D., 2013. Economic crisis, country variations, and institutional structures (pp. 1-
29). Oxford: Oxford University Press.

Hogarth, T.E.R.E.N.C.E. and Wilson, R., 2015. The outlook for skills demand and supply in
Europe. Technology, globalisation and the future of work in Europe: Essays on employment
in a digitised economy, Institute for Public Policy Research, London, pp.17-24.

Jones, M.K., Latreille, P.L., Sloane, P.J. and Staneva, A.V., 2013. Work-related health risks
in Europe: Are older workers more vulnerable?. Social Science & Medicine, 88, pp.18-29.

49
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Kaas, L. and Manger, C., 2012. Ethnic discrimination in Germany's labour market: a field
experiment. German economic review, 13(1), pp.1-20.

Karamessini, M. and Rubery, J. eds., 2013. Women and austerity: The economic crisis and
the future for gender equality. Routledge.

Kelly, E., & McGuinness, S. (2015). Impact of the Great Recession on unemployed and
NEET individuals’ labour market transitions in Ireland. Economic Systems, 39(1), 59-71.

Khattab, N. and Fox, J., 2016. East-European immigrants responding to the recession in
Britain: is there a trade-off between unemployment and over-qualification?. Journal of
Ethnic and Migration Studies, 42(11), pp.1774-1789.

Krings, T., Bobek, A., Moriarty, E., Salamońska, J. and Wickham, J., 2011. From boom to
bust: Migrant labour and employers in the Irish construction sector. Economic and
Industrial Democracy, 32(3), pp.459-476.

Lewchuk, W., 2017. Precarious jobs: Where are they, and how do they affect well-being?.
The Economic and Labour Relations Review, 28(3), pp.402-419.

Matas, A., Raymond, J.L. and Roig, J.L., 2010. Job accessibility and female employment
probability: the cases of Barcelona and Madrid. Urban Studies, 47(4), pp.769-787.

McGuinness, S., 2006. Overeducation in the labour market. Journal of economic surveys,
20(3), pp.387-418.

Monastiriotis V. (2018), Labour market adjustments during the crisis and the role of
flexibility, Social Situation Monitor Research Note, European Commission (DG EMPL),
Luxemburg (DOI: 10.2767/947941)

Monastiriotis, V. and Martelli, A., 2013. Beyond rising unemployment: unemployment risk,
crisis and regional adjustments in Greece.

Mossfeldt, M. and Österholm, P., 2011. The persistent labour-market effects of the financial
crisis. Applied Economics Letters, 18(7), pp.637-642.

O’Reilly, J., Eichhorst, W., Gábos, A., Hadjivassiliou, K., Lain, D., Leschke, J., McGuinness,
S., Kureková, L.M., Nazio, T., Ortlieb, R. and Russell, H., 2015. Five characteristics of youth
unemployment in Europe: Flexibility, education, migration, family legacies, and EU policy.
Sage Open, 5(1), p.2158244015574962.
50
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

O'Brien, M., 2011. Discouraged older male workers and the discouraged worker effect.
Australian Journal of Labour Economics

OECD (2015), Integrating Social Services for Vulnerable Groups: bridging sectors for better
service delivery, OECD Publishing, Paris.

Peng, F., Anwar, S. and Kang, L., 2017. New technology and old institutions: An empirical
analysis of the skill-biased demand for older workers in Europe. Economic Modelling, 64,
pp.1-19.

Quintano, C., Mazzocchi, P., & Rocca, A. (2018). The determinants of Italian NEETs and
the effects of the economic crisis. Genus, 74(1), 5.

Reyneri, E. and Fullin, G., 2011. Labour market penalties of new immigrants in new and
old receiving West European countries. International Migration, 49(1), pp.31-57.

Schuring, M., Robroek, S.J., Otten, F.W., Arts, C.H. and Burdorf, A., 2013. The effect of ill
health and socioeconomic status on labor force exit and re-employment: a prospective
study with ten years follow-up in the Netherlands. Scandinavian journal of work,
environment & health, pp.134-143.

Sundaram R., Hoerning U., De Andrade Falcao N., Millan N., Tokman C., and Zini M. (2014),
Portraits of Labor Market Exclusion, The World Bank.

Van Dalen, H.P., Henkens, K. and Schippers, J., 2010. Productivity of older workers:
Perceptions of employers and employees. Population and development review, 36(2),
pp.309-330.

Vertova, G., 2016. Crisis, policy responses and gender: the Italian experience. In
Economics and Austerity in Europe (pp. 81-97). Routledge.

Wood, M., J.Hales, S.Purdon, T.Sejersen and O.Hayllar (2009), ‘A Test for Racial
Discrimination in Recruitment Practice in British Cities’, Department for Work and Pensions
Research Report No. 607.

Zwysen, W. (2015). The effects of father’s worklessness on young adults in the UK. IZA
Journal of European Labor Studies, 4(1), 2.

51
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

APPENDIX A

Table A1. Prevalence of characteristics across labour market statuses, by country and year
Labour 15-24 55-64 65-74
Low Large Bad
Country Year market Female years years years Foreign
educated household health
status old old old
Bulgaria 2012 Inactive 60.6% 17.1% 14.2% 28.7% 0.4% 9.5% 17.2% 35.8%
Unemployed 42.2% 14.8% 16.3% 1.1% 0.9% 10.6% 28.6% 23.8%
Full-time 47.6% 3.8% 18.1% 1.4% 0.4% 1.7% 19.4% 13.1%
Part-time 55.7% 6.1% 23.7% 9.7% 2.0% 9.0% 24.3% 25.7%
Self-
employed 33.4% 2.1% 19.6% 3.9% 0.5% 2.4% 23.8% 15.5%
TOTAL 51.8% 10.5% 16.4% 12.8% 0.5% 6.0% 19.8% 23.9%
2017 Inactive 60.9% 14.8% 11.2% 32.0% 0.7% 9.1% 12.7% 31.4%
Unemployed 41.3% 9.7% 19.3% 3.2% 0.3% 17.5% 31.3% 21.9%
Full-time 47.8% 3.0% 19.0% 2.5% 0.5% 1.6% 13.0% 9.7%
Part-time 59.3% 5.3% 26.0% 12.6% 1.6% 9.9% 25.0% 17.5%
Self-
employed 33.4% 0.5% 23.0% 4.9% 0.8% 2.3% 14.0% 9.0%
TOTAL 51.9% 8.3% 16.2% 14.8% 0.6% 6.3% 14.7% 19.7%
Germany 2012 Inactive 56.7% 16.4% 10.0% 30.0% 17.2% 3.9% 4.7% 20.5%
Unemployed 48.0% 5.8% 29.1% 1.8% 13.6% 8.5% 5.0% 28.7%
Full-time 33.9% 8.4% 17.1% 0.7% 11.5% 1.4% 5.3% 7.4%
Part-time 83.9% 4.4% 23.8% 1.2% 13.5% 2.7% 5.9% 9.6%
Self-
employed 37.0% 1.5% 21.1% 5.9% 16.6% 0.6% 6.0% 7.2%
TOTAL 51.2% 10.9% 15.7% 13.2% 14.4% 2.9% 5.1% 14.1%
2017 Inactive 54.9% 16.2% 8.9% 27.0% 17.4% 6.3% 4.0% 20.2%
Unemployed 48.3% 4.6% 33.3% 2.4% 21.2% 6.7% 3.2% 30.8%
52
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Full-time 34.2% 7.1% 20.1% 0.9% 13.9% 1.4% 5.1% 7.4%


Part-time 81.5% 4.3% 25.8% 3.0% 16.0% 2.3% 6.9% 9.8%
Self-
employed 41.8% 1.2% 26.2% 5.9% 16.2% 0.6% 5.7% 7.4%
TOTAL 50.7% 10.2% 17.1% 12.2% 16.1% 3.7% 4.9% 14.0%
Estonia 2012 Inactive 63.0% 21.9% 12.4% 23.1% 20.7% 7.0% 9.2% 34.8%
Unemployed 44.1% 12.0% 16.8% 0.1% 15.6% 1.3% 10.3% 21.8%
Full-time 49.9% 5.0% 17.9% 2.1% 12.0% 0.4% 8.3% 9.8%
Part-time 70.0% 8.7% 18.5% 13.0% 15.6% 1.2% 7.3% 12.8%
Self-
employed 34.9% 2.5% 15.5% 4.9% 12.9% 0.5% 8.7% 9.1%
TOTAL 55.1% 12.3% 15.5% 11.0% 15.9% 3.1% 8.7% 20.7%
2017 Inactive 63.5% 17.3% 9.4% 24.4% 20.7% 5.1% 9.7% 30.6%
Unemployed 36.3% 13.4% 20.8% 0.2% 9.1% 0.8% 13.7% 16.1%
Full-time 49.3% 4.2% 19.3% 4.3% 10.7% 0.3% 9.3% 8.1%
Part-time 69.5% 8.3% 19.7% 16.0% 14.9% 0.5% 8.2% 12.1%
Self-
employed 31.8% 1.4% 22.6% 4.5% 7.1% 0.4% 10.9% 7.6%
TOTAL 54.2% 9.4% 15.9% 12.3% 14.4% 2.1% 9.6% 16.9%
Spain 2012 Inactive 60.4% 17.5% 13.1% 24.2% 9.1% 39.5% 10.4% 25.7%
Unemployed 48.5% 9.6% 14.9% 1.2% 22.6% 21.9% 15.2% 14.1%
Full-time 41.7% 2.6% 13.4% 1.1% 15.3% 8.9% 9.6% 7.3%
Part-time 75.9% 6.4% 9.2% 1.3% 23.0% 11.6% 11.7% 10.7%
Self-
employed 30.0% 1.4% 21.6% 3.2% 9.3% 16.3% 9.2% 9.1%
TOTAL 51.2% 9.7% 13.8% 10.5% 13.9% 23.6% 10.8% 16.0%
2017 Inactive 57.6% 17.8% 11.7% 28.0% 7.2% 43.9% 9.2% 20.8%
Unemployed 54.7% 8.7% 18.7% 3.7% 15.3% 21.1% 12.6% 14.7%
Full-time 43.4% 2.3% 16.3% 1.3% 13.8% 7.0% 8.1% 5.3%

53
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Part-time 75.9% 7.3% 15.7% 1.9% 20.8% 12.1% 11.2% 7.6%


Self-
employed 34.0% 1.1% 21.4% 4.2% 16.2% 9.7% 9.8% 6.9%
TOTAL 51.4% 9.2% 15.2% 11.9% 12.1% 23.4% 9.4% 12.7%
Hungary 2012 Inactive 61.4% 20.4% 20.5% 23.0% 0.6% 6.4% 13.2% 39.4%
Unemployed 46.8% 13.7% 20.0% 0.0% 0.9% 3.5% 19.8% 32.2%
Full-time 46.4% 4.1% 19.0% 0.4% 1.2% 0.5% 13.3% 14.8%
Part-time 63.1% 6.7% 23.5% 2.6% 2.9% 0.6% 14.0% 28.7%
Self-
employed 35.9% 1.6% 28.0% 2.9% 0.6% 0.1% 8.2% 13.6%
TOTAL 53.4% 12.5% 20.4% 11.3% 0.9% 3.5% 13.4% 27.9%
2017 Inactive 62.7% 19.2% 15.3% 29.1% 1.4% 4.5% 12.1% 32.7%
Unemployed 40.8% 13.3% 23.9% 0.8% 2.1% 6.4% 14.5% 27.7%
Full-time 46.5% 4.4% 19.4% 0.4% 1.6% 0.8% 11.9% 12.5%
Part-time 63.9% 7.9% 28.3% 5.3% 0.7% 2.1% 10.8% 24.4%
Self-
employed 37.6% 0.7% 29.9% 4.2% 1.4% 0.1% 7.8% 9.9%
TOTAL 53.1% 11.0% 18.6% 13.1% 1.5% 2.5% 11.8% 21.8%
Poland 2012 Inactive 62.7% 19.7% 22.5% 21.6% 1.9% 28.5% 23.3% 40.8%
Unemployed 55.7% 16.0% 17.0% 0.4% 0.3% 15.0% 30.1% 28.6%
Full-time 45.7% 5.1% 12.3% 0.5% 0.3% 4.9% 22.7% 16.0%
Part-time 65.6% 11.9% 19.5% 6.0% 0.4% 7.0% 20.9% 20.4%
Self-
employed 39.5% 2.2% 18.9% 1.9% 0.3% 11.5% 35.3% 22.0%
TOTAL 53.5% 11.9% 17.8% 9.8% 1.0% 16.4% 24.7% 28.1%
2017 Inactive 63.3% 15.2% 19.2% 28.5% 1.3% 25.0% 21.8% 37.2%
Unemployed 55.2% 8.6% 20.8% 2.6% 0.7% 12.6% 30.0% 29.1%
Full-time 47.5% 4.3% 15.9% 1.4% 0.6% 3.5% 22.0% 13.6%
Part-time 70.4% 10.6% 23.1% 8.2% 0.3% 9.2% 19.2% 18.5%

54
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Self-
employed 40.1% 2.2% 22.1% 3.0% 0.5% 8.5% 37.2% 21.3%
TOTAL 54.3% 9.0% 18.4% 13.2% 0.9% 13.7% 23.9% 25.3%
Portugal 2012 Inactive 59.0% 16.4% 15.3% 27.5% 3.5% 46.4% 8.2% 48.1%
Unemployed 51.0% 11.7% 17.8% 1.3% 10.2% 45.3% 12.3% 30.1%
Full-time 48.3% 4.9% 12.0% 1.0% 8.9% 33.3% 9.5% 17.4%
Part-time 78.1% 8.7% 21.2% 2.5% 11.9% 48.0% 11.6% 28.7%
Self-
employed 35.5% 1.6% 24.3% 6.3% 7.5% 53.0% 10.7% 21.4%
TOTAL 53.0% 10.2% 14.9% 12.4% 6.8% 41.5% 9.4% 31.9%
2017 Inactive 58.3% 18.2% 12.6% 28.4% 3.9% 60.7% 7.8% 41.6%
Unemployed 50.3% 9.7% 21.8% 2.7% 9.0% 37.9% 14.5% 29.9%
Full-time 49.6% 3.4% 16.4% 1.5% 8.7% 23.8% 9.3% 14.6%
Part-time 69.1% 8.6% 20.4% 3.4% 8.7% 40.4% 10.2% 20.4%
Self-
employed 40.4% 1.2% 23.5% 5.5% 8.4% 38.6% 10.4% 16.9%
TOTAL 53.3% 10.1% 15.8% 13.3% 6.7% 41.9% 9.2% 27.5%
United
Inactive
Kingdom 2012 58.8% 20.3% 13.1% 23.6% 12.4% 12.0% 10.0% 24.4%
Unemployed 39.7% 33.7% 10.1% 0.5% 17.6% 15.2% 15.8% 18.2%
Full-time 41.3% 8.5% 13.6% 2.8% 14.4% 4.2% 8.1% 8.1%
Part-time 77.0% 13.8% 18.7% 8.0% 12.2% 1.5% 10.4% 9.2%
Self-
employed 31.4% 3.8% 20.2% 9.1% 15.8% 5.7% 10.9% 6.7%
TOTAL 51.3% 14.0% 14.4% 11.5% 13.6% 7.2% 9.6% 14.5%
2017 Inactive 58.8% 18.8% 11.7% 25.7% 13.0% 0.1% 10.2% 20.6%
Unemployed 36.3% 24.7% 16.2% 34.4% 17.3% 0.2% 17.0% 17.0%
Full-time 41.0% 9.5% 14.3% 2.3% 18.3% 0.0% 10.1% 5.8%
Part-time 75.9% 13.2% 18.8% 6.3% 15.8% 0.0% 11.4% 8.1%

55
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Self-
employed 31.9% 2.2% 19.7% 0.0% 19.9% 0.0% 12.9% 5.8%
TOTAL 51.2% 13.3% 14.3% 12.4% 16.0% 0.0% 10.7% 12.1%
Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-2017.

56
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

57
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Figure A.1. Distribution of labour market statuses by group

58
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

59
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

60
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

61
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Figure A.2. Distribution of occupations by group

62
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

63
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

64
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

65
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Figure A.3.a. Distribution of women by sector of economic activity.


Bulgaria Germany

Estonia Spain

Hungary Poland

Portugal United Kingdom

66
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-
2017.

Figure A.3.b. Distribution of 15-24 years old by sector of economic activity.


Bulgaria Germany

Estonia Spain

Hungary Poland

67
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Portugal United Kingdom

Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-
2017.

Figure A.3.c. Distribution of 55-64 years old by sector of economic activity.


Bulgaria Germany

Estonia Spain

68
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Hungary Poland

Portugal United Kingdom

Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-
2017.

69
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Figure A.3.d. Distribution of 65-74 years old by sector of economic activity.


Bulgaria Germany

Estonia Spain

Hungary Poland

Portugal United Kingdom

70
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-
2017.

Figure A.3.e. Distribution of foreign by sector of economic activity.


Bulgaria Germany

Estonia Spain

Hungary Poland

71
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Portugal United Kingdom

Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-
2017.

Figure A.3.f. Distribution of low educated by sector of economic activity.


Bulgaria Germany

Estonia Spain

72
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Hungary Poland

Portugal United Kingdom

Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-
2017.

73
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Figure A.3.g. Distribution of people from large households by sector of


economic activity.
Bulgaria Germany

Estonia Spain

Hungary Poland

Portugal United Kingdom

74
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-
2017.

Figure A.3.h. Distribution of people from households with bad health by sector
of economic activity.
Bulgaria Germany

Estonia Spain

Hungary Poland

75
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Portugal United Kingdom

Source: European Union Statistics on Income and Living Conditions (EU-SILC), 2012-
2017.

76
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

Appendix B

Multinomial logit analysis and marginal effects (results presented


in section 4.3)

To examine how individual outcomes corresponding to different group characteristics


(female, youth, etc) changed between 2012 and 2017 we employ a regression analysis
using a multinomial logit (polytomous logistic regression) model, where we introduce a full
set of interaction terms between the categorical variables of interest (e.g., 1 if female, 0
otherwise) and a year dummy taking the value of 1 if the observation is taken from the
2017 sample (and zero if the observation belongs to the 2012 sample).

Specifically, the multinomial logit model uses a maximum likelihood estimator to estimate
the probability of the realisation of an event K in relation to the probability of all other
events (K − 1) , as follows:

1
Pr(Yi = K | X ) = (1)
1 + ∑k =1 e
K −1 β X
k i

with K=1,2,3,4,5 representing the five possible outcomes with regard to an individual’s
labour market status (inactive, unemployed, and working full-time, part-time or as self-
employed).

We estimate this model jointly for the two years in our sample (2012 and 2017), applying
population weights as provided in the EU-SILC. The vector X, of individual and household
characteristics, includes the following dichotomous indicators (categorical variables):
female, young (15-24 years old), over 55 years old, over 65 years old, foreign-born, low
educated (primary or less), never married, living in household where at least 1 member
reports bad health status, and living in large household (5 or more people). To capture the
change in the associated probabilities between 2012 and 2017 we interact the variables of
interest (female, youth, low-educated, poor-health households, large households, mature
individuals and old-age individuals) with a year dummy for 2017. Thus, the linearized form
of the underlying equation that is being estimated is:

𝑌𝑌𝑖𝑖𝑖𝑖 = 𝛼𝛼 + 𝛽𝛽 𝜃𝜃 𝑋𝑋𝑖𝑖𝑖𝑖𝜃𝜃 + 𝛾𝛾 𝜃𝜃 �𝑋𝑋𝑖𝑖𝑖𝑖𝜃𝜃 𝐷𝐷2017 � + 𝐷𝐷2017 + 𝜀𝜀𝑖𝑖𝑖𝑖 (2)

where Xθ stands for the θ variables listed above (female, youth, etc), D2017 is the year
dummy for 2017, i indexes individuals, t indexes time (the years 2012 and 2017) and ε is
77
Labour Market Vulnerability and Labour Market Outcomes during the Economic Upswing

a randomly distributed error term. Our analysis focuses on the coefficients γθ, which
capture the change in the relationship between each Xθ variable and the Y outcome
between 2012 and 2017.

Given the non-linearity of the estimating model (as depicted in equation (1)), we report
results in the form of marginal effects, which are evaluated at average pooled sample
values in order to allow cross-period comparisons that are independent from compositional
changes in the sampled population. Marginal effects are transformations of the estimated
coefficients which measure (for the case of categorical variables, such as the case here)
the change in the predicted probability of an event (here, labour market outcome/status)
for a discrete change in the corresponding variable (i.e., as the variable of interest changes
from 0 to 1). Given the non-linearity of the logit model, these changes in predicted
probabilities (marginal effects) are not constant across the range of values of the other
variables entering the estimated model (e.g., the marginal effect of the female indicator
when youth equals 1 is not the same as when youth equals zero), the marginal effects are
calculated at some fixed values of all other covariates. Our approach follows common
practice in the literature and calculates the marginal effects at the mean sample values of
all the covariates in the model.

78
Getting in touch with the EU
In person

All over the European Union there are hundreds of Europe Direct Information Centres. You can find the address of the
centre nearest you at: http://europa.eu/contact

On the phone or by e-mail

Europe Direct is a service that answers your questions about the European Union. You can contact this service

– by freephone: 00 800 6 7 8 9 10 11 (certain operators may charge for these calls),

– at the following standard number: +32 22999696 or

– by electronic mail via: http://europa.eu/contact

Finding information about the EU


Online

Information about the European Union in all the official languages of the EU is available on the Europa website at:
http://europa.eu

EU Publications

You can download or order free and priced EU publications from EU Bookshop at: http://bookshop.europa.eu.
Multiple copies of free publications may be obtained by contacting Europe Direct or your local information centre
(see http://europa.eu/contact)

EU law and related documents

For access to legal information from the EU, including all EU law since 1951 in all the official language versions, go
to EUR-Lex at: http://eur-lex.europa.eu

Open data from the EU

The EU Open Data Portal (http://data.europa.eu/euodp/en/data) provides access to datasets from the EU. Data can be
downloaded and reused for free, both for commercial and non-commercial purposes.

You might also like