The Consequences on Job Satisfaction of
Educational and Skill Mismatches
in the Spanish Labour Market:
A Panel Analysis
Lourdes Badillo-Amador †
Ángel López Nicolás ‡
Luis E. Vila ±
† Technical University of Cartagena
‡ Technical University of Cartagena, CRES and FEDEA
± University of Valencia
XXXIII, Symposium of Economic Analysis, Zaragoza, 2008
Objective
Analysis of the job satisfaction consequences
Educational mismatch
Skill mismatch
Relevance of unobserved heterogeneity
Dynamic structure of workers’ job satisfaction
Motivation
Job satisfaction can help to explain the
worker’s whole package of both pecuniary and
non-pecuniary rewards from work
Job satisfaction can clarify the consequences of
job-worker mismatches on benefits from work
Motivation
Scarce research has studied the job
satisfaction consequences of both
educational and skill mismatches, and crosssectional analyses has been carried out
Motivation
Two strong assumptions under
Cross-Sectional analyses
Workers’ observable characteristics are not
correlated with unobserved factors that also affect job
satisfaction
Job satisfaction has no inter-temporal dependence,
which implies that current scores of job satisfaction are
not influenced by previous experiences
Motivation
The present study considers
Longitudinal analyses in order to allow
for unobserved heterogeneity, and state
dependence
Attrition bias is also taken into account
Structure
1. Data
2. Job-worker matches: incidence and relationship
3. Models and Method
4. Results
5. Conclusions
1. Data
Spanish data
European Community Household Panel
(1994 – 2001)
Sample
Includes
Wage-earners aged between 16 and 64 years who work at
least 15 hours per week in their main job
Excludes
Trainees and workers in unpaid jobs
Those who either did not participate in the first wave or only took
part in this one
15,685 valid records
2. Job-worker matches
Education matches: modal definition
Required education (RE)
Educational mode of workers in the same occupational category
(ISCO88/2digit level)
Adequately educated worker
Worker’s education level = RE
Overeducated worker
Worker’s education level > RE
Undereducated worker
Worker’s education level < RE
2. Job-worker matches
The extend of a worker’s educational mismatch was
determined by comparing the number of years of
required education by his/her job with the number of
years of education level attained
Yrs. Overeducation = Schooling Yrs. – RE Yrs.
If Schooling Yrs. > RE Yrs, 0 otherwise
Yrs. Undereducation = RE Yrs. - Schooling Yrs.
If Schooling Yrs. < RE Yrs, 0 otherwise
2. Job-worker matches
Skill matches: workers’ self-assessment
i) Have your studies or your training provided you with the skills needed
for your current type of work?
ii) Do you feel that your skills or personal capabilities would allow you to
do a more demanding job than the one you do now?
i)
Yes
No
Yes
Overskilled
Wrongly skilled
No
Adequately
skilled
Underskilled
ii)
2. Job-worker matches
Proportion of skill and education job-worker matches and
association degree
Overeducated
Adequately
educated
Undereducated Total
Overskilled
Adequately skilled
Underskilled
12
6
11
13
10
16
10
9
13
35
25
40
Total
30
38
32
100
Pearson's c2
P-value
Cramér's V
123.246
0.000
0.063
3. Models and Method
(1)
js   mmit   xit  it
(2)
js   mmit   xit   i   it
(3)
js   mmit   xit   i   3 jsit 1  it
*
it
*
it
*
it
'
1
'
2
'
1
'
1
(i = 1,…,N)
'
2
'
2
(t = 2,…,Ti)
3. Models and Method
Parametrization of i
(Mundlak, 1978; Wooldridge, 2005)
(4)
i  0  1mmi  2 x i  3sji 1  ui
Sustituting (4) in (2) and (3)
(2a)
js*it  1' mmit   2' xit  0  1mmi  2 x i  3sji 1  ui   it
(3a) js*it   1' mmit   '2 xit   3 jsit 1  0  1mmi  2 x i  3sji 1  ui  it
3. Models and Method
What estimation model can be utilized to estimate the
job satisfaction when i is considered?
Attrition bias appears when participants either continues or stop
responding to the different survey waves for non-random reasons
Random effect
Pooled
ordered probit model ordered probit model
No
Attrition
bias
Yes
Consistent ,
but not efficient
Consistent and
efficient
Consistent with IPW
(Inverse probability
weight estimators)
Inconsistent
(IPW cannot be used)
3. Models and Method
Attrition bias tests
(Verbeek & Nijman, 1992)
Increasing the models by one of the following variables
Whether or not an individual responds in the next wave
Whether or not an individual answers in all waves
Number of waves that a worker is in the panel
There is Attrition bias if one is significante
(Estimation by Pooled ordered probit models with IPW)
3. Models and Method
Two kinds of
Inverse Probability Weight estimators (IPW)
Probability of response of each individual in each wave
Explanatory variables valued as in the first wave
Explanatory variables valued as in the previous wave
4. Results
Estimation results of job satisfaction. Equation (1)
Coefficients
Yrs. Overeducation
Yrs. Undereducation
-0.011 *
Robust
Std. Error
0.006
0.018 ***
0.006
Overskilled
-0.169 ***
0.025
Underskilled
-0.275 ***
0.030
Wrongly skilled
-0.389 ***
0.031
N
*p<0.10, **p<0.05, ***p<0.001
15685
4. Results
Estimated coefficients for Attrition tests
Ordered probit with unobserved heterogeneity
(Equation 2a)
Pooled model
Next wave
All waves
Number of waves
0.056 **
0.071 ***
0.008
Random effect model
0.061 **
0.076 ***
0.006
Dynamic ordered probit with unobserved
heterogeneity
(Equation 3a)
Pooled model
Next wave
All waves
Number of waves
0.024
0.059 ***
0.007
*p<0.10, **p<0.05, ***p<0.001
Random effect model
0.060 **
0.070 ***
0.007
4. Results
Consistent estimates are obtained by Pooled
ordered probit model with IPW estimators
IPW referred to previous wave were preferred
4. Results
Estimation results of job satisfaction. Equations 2a and 3a
Ordered probit with
unobserved heterogeneity
(Equation 2a)
Yrs. Overeducation
Yrs. Undereducation
Overskilled
Underskilled
Wrongly skilled
Satisfaction in wave=1: degree 1
Satisfaction in wave=1: degree 2
Satisfaction in wave=1: degree 3
Satisfaction in wave=1: degree 4
Satisfaction in wave=1: degree 6
Satisfaction in t-1: degree 1
Satisfaction in t-1: degree 2
Satisfaction in t-1: degree 3
Satisfaction in t-1: degree 4
Satisfaction in t-1: degree 6
-0.005
0.004
-0.126
-0.133
-0.186
-0.719
-0.416
-0.384
-0.193
0.376
***
***
***
***
***
***
***
***
Dynamic ordered probit with
unobserved heterogeneity
(Equation 3a)
-0.004
0.004
-0.118
-0.155
-0.175
-0.441
-0.234
-0.226
-0.116
0.265
-0.890
-0.704
-0.490
-0.257
0.308
13380
N
13380
Likelihood-ratio test (equations (2a) and (3a)): Chi-squared = 864.08 P-value = 0.000
*p<0.10, **p<0.05, ***p<0.01.
***
***
***
***
***
***
***
***
***
***
***
***
***
4. Results
Predicted Probability Distribution of Job Satisfaction for a
Reference Individual and Marginal Effects. Equation (3a)
Job satisfaction degree
1
2
3
4
5
6
0.006
0.023
0.101
0.232
0.445
0.193
Yrs. Overeducation
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
Yrs. Undereducation
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
Overskilled
0.002
0.006
0.018
0.018
-0.014
-0.031
Underskilled
0.003
0.009
0.024
0.024
-0.020
-0.040
Wrongly skilled
0.004
0.010
0.027
0.027
-0.023
-0.044
-0.001
Satisfaction in t-1: degree 1
0.013
0.031
0.072
0.057
-0.075
-0.098
Satisfaction in t-1: degree 2
0.005
0.014
0.037
0.035
-0.033
-0.057
Satisfaction in t-1: degree 3
0.005
0.014
0.035
0.034
-0.032
-0.056
Satisfaction in t-1: degree 4
0.002
0.006
0.017
0.018
-0.014
-0.030
Satisfaction in t-1: degree 6
-0.003
-0.011
-0.034
-0.046
0.013
0.081
Probability for a ref. individual
5. Conclusions
35% of workers have the same kind of fit under
education and skill mismatch criteria
Its association degree is lower than 0.1 in a
scale from 0 to 1
Weak relationship between education and skill
job-worker matches
5. Conclusions
Educational mismatches lost its influence in
workers’ job satisfaction after allowing for
unobserved heterogeneity
The influence of education mismatch in workers’ job
satisfaction is consequence of the unobserved
time-invariant characteristics of individuals
The cross-sectional analysis is misleading
5. Conclusions
After controlling for unobserved heterogeneity
the skill job-workers mismatches still affect
negatively to workers’ job satisfaction
Skill mismatches matter to workers
5. Conclusions
Workers’ current job perception depends on their
own previous job satisfaction
Persistence of job satisfaction