EPUNet Conference Barcelona, 8-9 May 2006

1
EPUNet Conference Barcelona, 8-9 May 2006
2
Unemployment risks in four European countries

• Antonio Schizzerotto Mario Lucchini
•  University of Milano Bicocca
• Italy

3
Aims of the study

• To analyse the effects of social classes on risks
  of unemployment and job stability.
• To check if the main features of the association
  between social classes, risks of unemployment and
  job security are stable across societies with
  different institutional arrangements.

4
Four EU countries

• Denmark, as representative of countries where the
  State plays an important role in the functioning
  of the whole society.
• Austria and Italy, as representatives of
  countries where family has a crucial position in
  the institutional arrangements of the society.
• United Kingdom, as representative of countries
  that attribute great importance to the market in
  the workings of the society.

5
Two hypotheses to be tested

• There is a strong relation between social class
  and both job security and risk of unemployment.
  In the case of dependent workers, this
  association depends on
• employment relation
• level of skill
• Despite differences in institutional
  arrangements, the above relation should hold
  across the four selected countries.

6
Data

• European Community Households Panel, eight waves
  (Jan 1994-Dec 2001) information regarding
  employment and unemployment episodes.

7
Methods used for the estimation of unemployment
risks.

• Model to estimate unemployment risks, we
  specified a random-coefficient Poisson regression
  model. The dependent variable in this model is
  the incidence rate ratio of being unemployed.
  Covariates as class and level of education were
  considered as causal variables.
• Class structure was represented using a recently
  developed nine-fold class scheme known as ESeC
  (European Socio-economic Classification)
• Education was used as coded in ECHP, according to
  ISCED scheme.
• We also controlled for civil status, gender,
  period effects, age, health condition,
  public/private sector of activity.

8
The ESeC class schema

• The ESeC (European Socio-economic Classification)
  is based on a widely-used social class schema
  devised by John Goldthorpe and Robert Erikson,
  known as the EGP schema.
• It enucleates nine socio-economic classes,
  resulting from the combination of the following
  factors
• Occupation, coded according to Isco88(com)
  classification
• Employment status, used to distinguish between
  employers, the self-employed, managers,
  supervisors and employees
• Size of organization, used to distinguish between
  large and small employers.

9
The ESeC classes

• Class 1 Large employers, higher grade
  professional, administrative and managerial
  occupations higher professionals and managers
• Class 2 Lower grade professional, administrative
  and managerial occupations higher grade
  supervisory and technician occupations lower
  professionals and managers
• Class 3 Intermediate occupations higher
  clerical, services and sales workers
• Classes 4 and 5 Small employers and
  self-employed in non-professional occupations
  small employers and self-employed (4) and
  farmers (5)
• Class 6 Lower supervisory and lower technician
  occupations lower supervisors and technicians
• Class 7 Lower clerical, services and sales
  occupations lower clerical, services and sales
  workers
• Class 8 Lower technical occupations skilled
  workers
• Class 9 Routine occupations semi- and
  unskilled workers.

10
The random-coefficient Poisson regression model

• We model the number of unemployment episodes
  experienced by an individual year after year.
• More specifically our dependent variable is the
  number of months of unemployment episodes during
  a year (or wave), standardized by the length in
  months of the overall participation in the labour
  market (i.e. number of months in unemployment
  number of months in employment).
• Since the total duration of the ECHP is 8 years
  (running from 1994-2001), each individual can be
  repeated until 8 times (from the wave 1 to wave
  8).
• We implemented a random intercept Poisson
  regression to model dependence and unobserved
  heterogeneity.
• In our model, the normally distributed random
  intercept for subject accommodates dependence
  among the repeated counts of unemployment
  episodes collected year after year.

11
Methods to estimate unemployment risks.

• Model to estimate unemployment risks, we
  specified a random-coefficient Poisson regression
  model. The dependent variable in this model is
  the incidence rate ratio of being unemployed.
• Class and education were considered as
  independent, causal variables.
• Class structure was represented using a recently
  developed nine-fold class scheme known as ESeC
  (European Socio-economic Classification)
• Education was used as coded in ECHP, according to
  ISCED scheme.
• civil status, gender, ECHP waves (assumed as
  expressing period effects), age, health
  condition, public/private sector of activity are
  treated as control variables

12
Methods used for the estimation of job stability.
To estimate job stability we carried out an event
history analysis in order to compute the survival
function of each class in each country.
13
Results of the estimation of unemployment risks
through a random-coefficient Poisson regression
model. Denmark, Italy, Austria and United
Kingdom-ECHP, 1994-2001.
14
(No Transcript)
15
(No Transcript)
16
(a) reference category p lt0.01 p lt0.05
p lt0.1
17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20
ESeC classes and the duration of employment
episodes
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
(No Transcript)
25
Conclusive remarks

• Education displays a protective effect against
  the risk of unemployment also in the UK. UK is a
  partial ecception but even there tertiary
  education shows this protective effect.
• Class exerts a quite strong effect on the risk of
  unemployment generally speaking, members of
  classes based on service employment relations are
  less likely to experience an unemployment spell
  than people belonging to classes based on labour
  contract.

26

• Yet between classes based on labour contract
  some differences can be observed.
• Namely, the risk of unemployment varies according
  to the level of technical skills the higher the
  technical expertise, the lower the risk of
  unemployment.
• Sometimes, the protective effect of technical
  skills overcomes the effect of the employment
  relations.

27

• Moving from incidence rate ratios of
  unemployment to job stability it can be seen
  that, by and large, occupational classes based on
  a service employment relationship display longer
  duration of employment episodes.