Sunteți pe pagina 1din 18

Job satisfaction

of older workers
343
International Journal of Manpower,
Vol. 20 No. 6, 1999, pp. 343-360.
# MCB University Press, 0143-7720
Received January 1999
Revised April 1999
Accepted May 1999
Job satisfaction of older workers
Wim Groot
Maastricht University and ``Scholar'' Research Center for Education and
Labor Market, University of Amsterdam, The Netherlands
Henriette Maassen van den Brink
``Scholar'' Research Center for Education and Labor Market, University
of Amsterdam and Wageningen University, The Netherlands
Keywords Job satisfaction, Wages, Older employees
Abstract Using data for The Netherlands, this paper analyzes the relation between allocation,
wages and job satisfaction. Five conclusions emerge from the empirical analysis: satisfaction with
the job content is the main factor explaining overall job satisfaction; the effects of individual and
job characteristics on job satisfaction differ by the aspect of the job considered; the response to a
general question on job satisfaction differs from the response to questions on satisfaction with
different aspects of the job; it is relevant to consider the joint relation between wages and job
satisfaction; and skill mismatches do not seem to affect job satisfaction.
Introduction
One of the most remarkable trends in the labor force has been the decline in the
labor force participation rate of older workers. Until the 1960s the labor force
participation rate of men aged 60-64 was over 70 percent in most industrialized
countries. By 1996 this has fallen to less than 20 percent in some countries (for
details, see Gruber and Wise, 1998). In The Netherlands the overall labor force
participation rate of workers aged 55-64 declined from 39 percent in 1979 to 27
percent in 1996, while the male labor force participation rate decreased from 65
percent in 1979 to 41 percent in 1996. The fall in the labor force participation
rate can partly be explained by the introduction of new (early) retirement
schemes (see Kapteyn and De Vos, 1998). However, it may be wondered
whether the changing perception of work has also contributed to the decrease
in the labor force participation? Job demands and work related stress may have
increased over time, making workers less satisfied with the work they do and
more eager to retire at the earliest possible moment. Unfortunately, long-term
data on job demands, stress and job satisfaction of (older) workers are not
available to test this hypothesis.
In recent years a small but growing literature in the economics field has
emerged on the determinants of job satisfaction[1]. These studies include
Hartog and Mekkelholt (1989); Mekkelholt (1993); Clark (1993); Clark and
Oswald (1995). All of these studies take a cross-section of workers of all ages
The current issue and full text archive of this journal is available at
http://www.emerald-library.com
We would like to thank an anonymous referee for helpful comments on a previous version of
this paper. The data used in this paper were financed by a grant from NESTOR (the Dutch
Stimulation Program for Ageing Research) and from the Leiden University Speerpunten
Program. This paper is part of a research project on schooling and labour market flexibility of
older workers. Financial support by the ``Fonds for Maatschappijgericht Onderzoek'' of the
University of Amsterdam is gratefully acknowledged.
International
Journal of
Manpower
20,6
344
and analyze the factors that determine their level of job satisfaction. If one
looks through this literature three issues emerge: the differences in job
satisfaction between men and women, the effects of earnings on job satisfaction
and the effect of education on job satisfaction.
Clark (1993) analyzes differences in job satisfaction between men and
women. His main finding is that women report higher levels of overall job
satisfaction, even after controlling for a number of variables, than men. This
study rejects the explanation that women have more freedom of choice whether
or not to work and that women who are not satisfied with their job prefer to
leave the labor market. Rather it is concluded that women appear to be happier
in their job than men.
Clark and Oswald (1995) look at the effect of income on job satisfaction.
Their two main findings are: workers' reported satisfaction levels decline in
wages and holding income constant satisfaction levels are strongly
declining in the level of education. In Mekkelholt and Hartog (1989) it is found
that quasi rents or temporary wage differentials have a positive but small effect
on job satisfaction. Mekkelholt (1993) concludes that job satisfaction is mainly
determined by job characteristics and job rationing (e.g. local unemployment
rate and employment probabilities). The matching quality, individual
characteristics and job mobility costs are also important, but to a lesser extent.
In a number of papers it is found that, after controlling for wages, more
highly educated workers are less satisfied with their job than less educated
workers. The negative effect of education on job satisfaction found by Clark
and Oswald we already mentioned. Theeuwes and Woittiez (1995) find that
university graduates report significantly lower levels of well-being than the
less educated, and in Mekkelholt and Hartog (1989), Mekkelholt et al. (1991),
Clark (1993) and Clark and Oswald (1995) it is found that more highly educated
workers are less satisfied with their job than less educated workers.
From this brief survey of recent papers on job satisfaction using data of
workers of all age groups the following conclusions can be drawn:
.
Women appear to be happier with their job than men;
.
Higher earnings do not seem to increase job satisfaction;
.
More highly educated workers appear to be less satisfied with their job.
The question can be asked whether these conclusions also apply to older
workers?
In this paper we analyze to what extent individual characteristics, job
characteristics and the quality of the match between individual skills and skill
requirements of the job can explain differences in overall job satisfaction
between older workers. As indicators of the quality of the match between skills
supplied and skills demanded, the concepts of over- and under-education are
used.
Different methods are used to define overall job satisfaction. This enables us
to test the robustness of the findings. The first job satisfaction variable used is
the response to the survey question on how satisfied the worker is with his/her
Job satisfaction
of older workers
345
job in general. The questionnaire also contains questions on the satisfaction
with different aspects of the job. Our second definition of overall job
satisfaction is derived from the sum of the responses on the different aspects of
the job. Further, factor analysis techniques are used to distinguish between the
separate items, and to distinguish the different aspects that underlie overall job
satisfaction. By this we are able to analyze how the effects of individual and job
characteristics differ between the job aspects distinguished, and how they
determine overall job satisfaction. Finally, we analyze how satisfaction with
different aspects of the job contributes to the overall job satisfaction.
Another contribution of this paper lies in the joint estimation of job satisfaction
and wages, of job satisfaction is determined by the wage rate. This joint
estimation allows us to identify the covariance between wages and job
satisfaction. The identification of this correlation coefficient between the residuals
of the wage equation and the job satisfaction equation enables us to indicate
whether workers who are paid more than their expected market wage are more (or
less) satisfied than workers who are underpaid relative to their expected market
wage rate. The issue of overpayment and underpayment and their effects on job
satisfaction was also taken up by Hartog and Mekkelholt (1989) and Mekkelholt
(1993), but not within a joint equation framework provided in this paper. They
used the deviation of predicted wages from actual wages as a variable in the job
satisfaction equation. Their results show that workers who earn more than
expected are more satisfied with their job (Mekkelholt, 1993, p. 75).
The joint estimation further enables us to separate the effect of variables on
job satisfaction from their effect on wages. By this approach some of the
somewhat puzzling and difficult to explain earlier results such as the
negative effect of education on job satisfaction may be better understood.
The outline of this paper is as follows. In the second section a brief
description of the data set is given. The data set consists of a cross-section of
the employed population aged 43 and older in The Netherlands in 1993. In this
section we further present some descriptive statistics on job satisfaction. In the
third section we describe the empirical model for analyzing skill utilization,
wages and job satisfaction. The variables in the equations are then discussed.
The empirical results are presented in the fifth section. The final section
concludes.
The data and the definition of job satisfaction
The data are taken from the first wave of a Dutch panel survey called CERRA-I
(Center for Economic Research on Retirement and Aging, wave I). The purpose
of this survey is to study retirement behavior. In structure and content this
survey resembles the Health and Retirement Survey (HRS) of the Michigan
Survey Center. The data set consists of approximately 4,700 households in
which the head of the household was between 43 and 63 years of age at the time
of interview in October 1993. The survey contains information on labor market
status, income, labor market history, housing, health status and a variety of
socio-economic variables of both the head of the household and his/her partner.
International
Journal of
Manpower
20,6
346
We limit our analysis to employed heads of households. The reason for
excluding employed partner was to avoid biases in the results due to the
endogeneity of the labor supply decision. Partners of heads of households may
have more choice in their labor supply decision than heads of households. This
may lead to biased results if only those partners participate in the labor market
for whom the satisfaction with paid work is large enough.
The questionnaire contains ten questions on satisfaction or dissatisfaction of
workers with particular aspects of their job. These specific job items involve
satisfaction with: the content of the job, the workload, the working conditions,
the pay, the number of hours of work, the time of day one has to work, the
commuting time to work, the supervisor, the co-workers, and the early
retirement arrangements. The response categories are: dissatisfied, neither
satisfied nor dissatisfied, and satisfied. After the valuation of the various
aspects of the job, overall job satisfaction is measured by the response to the
following question: ``how satisfied are you with your job in general?''
We will refer to the response to this question as the overall job satisfaction.
Overall job satisfaction is highly correlated with the satisfaction with the
content of the job (, = 0.508). Somewhat lower correlation coefficients are found
between overall job satisfaction and satisfaction with the work load (, = 0.280),
between overall job satisfaction and satisfaction with the working conditions (,
= 0.256), and between overall job satisfaction and satisfaction with the
supervisor (, = 0.276). The correlation coefficients between overall job
satisfaction and satisfaction with the other job aspects were all less than 0.2.
Further, high correlation coefficients are found between satisfaction with the
work load and satisfaction with the working conditions (, = 0.338), satisfaction
with the workload and satisfaction with the number of hours of work (, =
0.367), and between satisfaction with working conditions and satisfaction with
the supervisor (, = 0.305).
Our second job satisfaction is derived from the average value of the response
on the ten separate job aspects, divided by ten. We will refer to this variable as
the composite job satisfaction. This composite job satisfaction variable is used
as an alternative measure for overall job satisfaction.
In order to ascertain the validity of simply averaging the scores on the
individual items to obtain a measure of total job satisfaction, a factor analysis
was performed. The aim of this factor analysis was to establish how many
underlying factors the ten separate items have in common. Based on the
criterion of eigenvalues being larger than 1, three factors could be extracted.
The factor loadings for the three factors are found in Table I. The first factor
explains 24.7 percent of the total variance, the second factor 11.5 percent and
the third 10.2 percent. The three factors combined explain 46.3 percent of the
variance.
For the first factor there are high factor loadings for the content of the job,
the workload, the working conditions, the supervisor, and the co-workers. For
the second factor there are high factor loadings for: the workload, the working
conditions, the pay, the number of hours worked, the time of day one has to
Job satisfaction
of older workers
347
work, and the early retirement arrangements. For the third factor there are high
factor loadings for: the number of hours of work, the time of day one has to
work, and the commuting time to work. If the items with high factor loadings
are considered, the first factor consists of items that have to do with the content
of the job, while the third factor combines items that have to do with the hours
of work. The second factor is more mixed, but items that have high factor
loadings include those that have to do with the remuneration for the job.
Based on this factor analysis, we define three new job satisfaction variables:
satisfaction with the job content, satisfaction with the remuneration of the job,
and satisfaction with hours of work. The variable ``satisfaction with the job
content'' consists of the mean value, for every individual worker, of the
responses on the following items: content of the job, the workload, the working
conditions, the supervisor, and the co-workers. The variable ``satisfaction with
the remuneration of the job'' consists of the mean of the responses to the items:
satisfaction with pay and with the early retirement arrangements. The variable
``satisfaction with work time'' consists of the mean value of the responses to the
variables: satisfaction with the number of hours of work, the time of day one
has to work, and the commuting time to work.
Table II contains the frequency distribution of the response to the different job
satisfaction questions. Several noteworthy facts emerge from this Table. A
comparison of the distribution of responses on the separate job items shows that
workers are on average more satisfied with some aspects of the job than with
others. For example, only 1.3 percent of the workers are dissatisfied with
colleagues or co-workers, while more than 13 percent of the workers are
dissatisfied with the workload and with the early retirement arrangements. In
general it appears that workers are highly satisfied with the content of the job,
the time of day one has to work, the commuting time to work and the co-workers.
Workers are rather dissatisfied with the workload, the number of hours they
have to work, the supervisor, and the early retirement arrangements.
For the composite job satisfaction variables, workers are classified as being
dissatisfied if the sum of the responses to the different items divided by the
Satisfaction with Factor 1 Factor 2 Factor3
Content of the job 0.680 0.079 0.172
The workload 0.354 0.514 0.155
The working conditions 0.615 0.338 0.038
The pay 0.240 0.468 0.018
The number of hours of work 0.013 0.661 0.388
The time of day one has to work 0.076 0.402 0.479
The commuting time to work 0.050 0.089 0.827
The supervisor 0.691 0.080 0.119
The co-workers 0.504 0.177 0.123
The early retirement arrangements 0.020 0.671 0.148
Notes: Percentage of variance explained: 46.3 percent
Table I.
Rotated factor loading
matrix of factor
analysis on separate
job satisfaction items
International
Journal of
Manpower
20,6
348
number of items in the composite variables is less or equal to one. If the average
value of the responses is between one and two, the workers are classified as
being neither satisfied nor dissatisfied, while if it is more than two the worker is
classified as satisfied. This classification implies that one has to report being
satisfied with at least one item of the job without being dissatisfied with any
other item, in order to fall in the category ``satisfied with the job in general''.
A comparison of the overall job satisfaction level (own response) with the
composite job satisfaction level (average of responses to the ten separate job
items) shows that more than 90 percent of the workers report being satisfied
with their job in general, while the composite job satisfaction variable classifies
only 25.5 percent of the workers are satisfied with their job. Despite these
differences in the distribution of the variables, the overall job satisfaction
variable and the composite job satisfaction variable are highly correlated. The
correlation coefficient between the two variables is 0.438. A T-test shows that
the difference between the average values of the two variables is significantly
different from zero at the 1 percent level (the t-value is 17.9), however.
A well known problem in analyzing own reports to the job satisfaction
question is that workers give the socially acceptable answer even if this
deviates from their own evaluation. Workers say they are satisfied even if they
are not. Using the responses to a number of different items of the job may
reduce this socially acceptability bias, as it may be socially more acceptable to
be dissatisfied with some aspects of the job but not to be dissatisfied with the
job in general. In this sense the composite job satisfaction variable may give a
more accurate picture of overall job satisfaction.
Dissatisfied
Neither
satisfied nor
dissatisfied Satisfied
(%) (%) (%)
Overall job satisfaction level 2.1 7.8 90.1
Composite job satisfaction level 1.7 72.7 25.5
Composite satisfaction level job content 3.3 54.3 42.4
Composite satisfaction level pay 5.6 62.5 31.9
Composite satisfaction level hours of work 3.2 33.4 63.4
Satisfaction with different job items
The content of the job 3.2 7.9 89.0
The workload 13.3 24.5 62.2
The working conditions 7.1 16.8 76.1
The pay 8.9 16.7 74.3
The number of hours you have to work 10.1 15.3 74.6
The time of day you have to work 4.6 6.6 88.8
The commuting time to work 4.8 7.4 87.8
The supervisor 8.3 14.1 77.6
The co-workers 1.3 11.8 86.9
The early retirement arrangement 13.8 10.4 75.8
Table II.
Frequency distribution
of job satisfaction
Job satisfaction
of older workers
349
What aspects of the job determine overall job satisfaction? To answer this
question we estimated an OLS equation with overall job satisfaction as the
dependent variable and the responses to the questions on satisfaction with
different job items as explanatory variables. The results of this can be found in
Table III. The most important factor that determines overall job satisfaction is
satisfaction with the content of the job. A one point higher score on the question
about satisfaction with the content of the job increases the response to the
overall job satisfaction question by nearly 0.4 points. Other factors that
determine overall job satisfaction are, in order of importance: satisfaction with
co-workers, satisfaction with the supervisor, satisfaction with the workload
and satisfaction with the early retirement arrangement. These variables all
have a significant effect on overall job satisfaction. However, the size of the
effects of these variables is much smaller than that of satisfaction with the
content of the job. All other items of the job do not significantly contribute to
the explanation of overall job satisfaction. Similar conclusions are drawn if we
include the four loadings from the factor analysis in the equation.
The empirical model
The starting point of the empirical model is the wage equation. We specify the
following semi-logarithmic wage equation:
Logw
i
= u
0
u
1
X
1i
u
2
X
2i
B
3
X
3i
c
1
= X
i
u c
1
(1)
With responses to
questions on
satisfaction with
different job items
With variables
derviced from factor
analyses
Intercept 0.242** (5.943) 0.370** (9.419)
Satisfaction with:
The content of the job 0.382** (19.461)
The workload 0.056** (4.442)
The working conditions 0.025 (1.631)
The earnings 0.020 (1.496)
The number of hours you have to work 0.025 (1.821)
The time of day you have to work 0.026 (1.486)
The commuting time to work 0.022 (1.345)
The supervisor 0.058** (4.122)
The co-workers 0.072** (3.339)
The early retirement arrangement 0.036** (3.069)
The content of the work 0.484** (18.261)
The working hours 0.059* (2.387)
The pay-off 0.041* (2.336)
Adj-R
2
0.321 0.226
Notes: *Significant at 5 percent level; ** significant at 1 per cent level
Table III.
OLS estimates of
overall job satisfaction
(t-values in brackets)
International
Journal of
Manpower
20,6
350
where w
i
is the net hourly wage rate of individual i. X
1
is a vector of human
capital variables and other individual characteristics, X
2
is a vector of job
characteristics and X
3
are variables for the quality of the job match. Further,
X = [1. X
1
. X
2
. X
3
[. c is a random error term and u is a vector of coefficients.
The utility generated by the job (U
i
) is supposed to be determined by the wage
rate received in the job, the quality of the match between skills supplied and
skills required (X
3
), job characteristics (Y
2
) and individual characteristics (Y
1
):
U
i
= U(lnw
i
. X
3i
. Y
1i
. Y
2i
) (2)
We assume that utility is linear in the variables referred to, i.e.:
U
i
= c
0
c
1
logw
i
c
2
X
3i
c
3
Y
1i
c
4
Y
2i
c
2
(3)
If we substitute equation (1) into equation (3) we obtain:
U
i
= c
0
c
1
[X
i
u[ c
2
X
3i
c
3
Y
1i
c
4
Y
2i
c
1
c
1
c
2
= W
i
c j
(4)
where W = [1. X. X
3
. Y
1
. Y
2
[.
The job satisfaction variable (JSV) is measured as an ordinal variable
ranging from 1 to 3. We assume that the utility levels in equation (4) are related
to the value of the JSV of individual i in the following way:
JSV
i
= 1 if U _ 0
JSV
i
= 2 if 0 < U
i
_ c
1
JSV
i
= 3 if c
1
< U
i
(5)
where c is a threshold coefficient whose value will be determined in the
empirical analysis. Ignoring individual specific effects for the moment, the
loglikelihood function (LogL) of the joint observation of wages and job
satisfaction is given by:
LogL =

JSV=2
JSV=1
Log

Wa

c(j c
j1
. c
1
. ,) c(j c
j
. c
1
)dj
=

JSV=3
JSV=1
Log((j
j1
) (j
j
))
1
2
Log(1 ,
2
) Log(o) Log(c(
c
1
o
))
(6)
where c
0
= 0 and c
2
= and:
j
j
Wc c
j
,
c
1
o

1 ,
2
p (7)
Job satisfaction
of older workers
351
and , is the correlation coefficient between c
1
and j, o and is the standard
deviation of the wage equation, and and c are the univariate standard
normal distribution and density functions, respectively.
The definition of the explanatory variables
According to Theeuwes and Woittiez (1995), in the literature on the
determinants of well-being, labor market status, income, personal
characteristics such as education, age, health, and gender are frequently
encountered as explanatory variables. Regarding labor market status and
gender, our sub-sample is restricted to employees. Besides the variables
mentioned earlier, we include job characteristics and variables for the quality of
the job match in the equations as well.
In particular, we are interested in the effects of skill mismatches on wages
and job satisfaction. Two forms of skill mismatches are distinguished in this
paper. Both of them refer to the match between education required for the job
and the education level of the worker occupying the job. In the first form of skill
mismatches workers possess more education than is required for the job: i.e.
workers are over-educated for their job. In the second form workers have fewer
educational qualifications than required: they are under-educated. Both over-
and under-education may indicate that the allocation of skills over jobs is less
than optimal. Over-education may indicate that there is under-utlization of
skills (for a survey of earlier research on the incidence and wage effects of skill
mismatches, see Groot and Maassen van den Brink, 1997, 1999). The under-
utilization of skills may make workers less satisfied with the job they hold. We
know of only one paper that draws attention to the effects of skill mismatches
on job satisfaction. In Tsang and Levin (1985) it is argued that over-education
may lead to lower individual productivity by increasing job dissatisfaction and
adverse workplace behavior.
The data contain information on the highest education level attained on a
seven point scale ranging from ``only primary school'' to ``university''. The
education level required for the job is derived from the response on the
following question: ``Can you tell me what education level one at least has to
have completed to be considered for your current job?''. Respondents could
indicate the education level required on the same seven point scale as before. A
worker is classified as over-educated if the education level attained exceeds the
education level required, and as being under-educated if the education level
required exceeds the education level attained. On average 11 percent of the
workers in our data are classified as being over-educated, while 46 percent are
classified as under-educated. The large incidence of under-educated workers
may be due to the fact that this is a sample of older workers with enough labor
market experience to compensate for their lack of formal qualifications.
Besides the two skill mismatch variables, we include six dummy variables
for the highest education level attained by the workers. Other human capital
variables in the equations are: ten dummy variables for the type of education
acquired, and a dummy variable if the worker participated in on-the-job
International
Journal of
Manpower
20,6
352
training. Some human capital related variables are included as well: years of
tenure at the firm, years of labor market experience, a dummy if the worker has
been unemployed in the past, and a dummy if the worker has been out of the
labor force.
Other individual characteristics include: a dummy for gender (``male''), a
dummy if the worker is married or cohabiting (``married''), and for the country
of birth (``ethnicity''). A dummy variable for the health status was included as
well (1 = health is good or very good, 0 = otherwise).
Job characteristics include: five dummy variables for firm size, two dummy
variables for the number of workers the respondent is supervising, a dummy
variable if the firm has an early retirement arrangement, a dummy variable if
the worker is employed in a public sector firm, and a variable for hours of work.
Finally seven dummy variables are included for the industry in which the
worker is employed.
A first round of estimation revealed that some of these variables had a
highly insignificant effect on job satisfaction. These variables were therefore
omitted from the equations. The variables with a highly insignificant effect on
job satisfaction are: tenure, experience, on-the-job training, firm size, and type
of education. The predicted wage rate in the job satisfaction equation is thus
identified on variables that have an insignificant effect on job satisfaction.
It was further found that wages were linear in tenure and experience, and
that the variables ``experience squared'' and ``tenure squared'' had a highly
insignificant effect on wages. These variables were therefore not included in
the wage equations.
Estimation results
The parameter estimates of the job satisfaction cum wage equations are in
Table IV. There are a few differences between the results of the two job
satisfaction variables. The most remarkable difference is that the predicted
wage rate has a significantly positive effect on job satisfaction if the composed
job satisfaction variable is used (a higher wage rate increases job satisfaction),
and a positive but insignificant effect if the response to the direct question on
general job satisfaction is used.
It is further found that the education level variables have a significant effect
only in the composed job satisfaction equation where there is a significantly
positive effect of the predicted wage rate on job satisfaction. In the general job
satisfaction equation the education dummies are negative, but most of them are
not significantly different from zero. The significantly negative effect of
education level on job satisfaction confirms earlier findings mentioned in the
introduction, that higher educated workers are less satisfied with their job than
lower educated workers. However, if both the direct effects of education on job
satisfaction and the indirect effect through wages are considered, the negative
effect of education on job satisfaction disappears: the negative direct effects of
education on job satisfaction are canceled by the indirect or wage effects of
education. On balance, there are no significant differences in job satisfaction by
Job satisfaction
of older workers
353
Response to question of
general job satisfaction
Sum of responses to all
separate items of job
satisfaction
Job
satisfaction Wage rate
Job
satisfaction Wage rate
Intercept 0.431 2.490** 2.000 2.479**
(0.201) (55.292) (1.144) (45.737)
Human capital variables: education level (reference category: primary school)
Lower vocational education 0.235 0.035 0.185 0.057*
(1.382) (1.491) (1.330) (2.057)
Lower general education 0.485* 0.140** 0.551** 0.162**
(2.094) (5.652) (2.788) (5.626)
Intermediate general education 0.561 0.309** 0.873** 0.308**
(1.401) (9.811) (2.901) (8.231)
Intermediate vocational education 0.207 0.210** 0.536* 0.240**
(0.680) (8.029) (2.208) (7.876)
Higher vocational education 0.544 0.464** 1.040* 0.500**
(1.081) (15.945) (2.524 (14.777)
University 0.924 0.684** 1.436* 0.709**
(1.277) (19.509) (2.420) (16.820)
Job match quality
Over-educated 0.111 0.090** 0.024 0.061*
(0.617) (4.364) (0.160) (2.404)
Under-educated 0.037 0.116** 0.132 0.121**
(0.213) (6.624) (0.942) (5.871)
Specific training and employment record
Unemployed 0.194 0.075** 0.015 0.069**
(1.335) (4.533) (0.124) (3.435)
Out of labor force 0.026 0.053** 0.028 0.047**
(0.182) (2.978) (0.202) (2.286)
Other individual characteristics
Male 0.035 0.109** 0.125 0.098**
(0.155) (3.954) (0.556) (3.000)
Married 0.154 0.052* 0.135 0.049
(0.792) (0.241) (0.867) (1.894)
Ethnicity 0.467 0.000 0.095 0.022
(1.926) (0.002) (0.489) (0.708)
Health 0.423** 0.617**
(3.944) (5.739)
Job characteristics
Early retirement arrangement 0.139 0.661**
(1.140) (4.103)
Public sector 0.117 0.047** 0.204 0.037**
(0.934) (2.863) (1.953) (2.073)
(continued)
Table IV.
Parameter estimates
job satisfaction and
wages (t-values in
backets)
International
Journal of
Manpower
20,6
354
education level. If we re-estimate the composite job satisfaction equation with
the coefficient for the predicted wage rate set to zero i.e. if we omit the indirect
effects of education on satisfaction all education variables in the job
satisfaction equation become insignificantly different from zero.
The coefficients of the education level variables in the wage equations
indicate that both at the lower level and at the intermediate level the wage
effects of a general education are higher than those of a vocational education.
Compared to workers with only primary school, workers with lower
Response to question of
general job satisfaction
Sum of responses to all
separate items of job
satisfaction
Job
satisfaction Wage rate
Job
satisfaction Wage rate
Industry (reference category: industrial sector)
Agriculture 0.070 0.040 0.118 0.069
(0.147) (0.604) (0.203) (0.313)
Construction 0.250 0.064** 0.234 0.071**
(1.435) (2.748) (1.698) (2.844)
Transport 0.052 0.061** 0.364** 0.089**
(0.326) (2.861) (2.588) (3.590)
Commerce 0.140 0.042* 0.080 0.038
(0.899) (2.016) (0.573) (1.451)
Commercial services 0.115 0.041 0.078 0.037
(0.586) (1.749) (0.481) (1.277)
Public education 0.474** 0.029 0.576** 0.024
(2.799) (1.223) (3.218) (0.896)
Health care 0.175 0.077* 0.037 0.103**
(0.873) (2.488) (0.187) (2.625)
Hours of work 0.006 0.022**
(0.834) (2.737)
Predicted wage rate 0.545 1.603*
(0.634) (2.364)
c 0.816** 3.007**
(11.546) (26.151)
, 0.073 0.031
(1.425) (0.731)
o 0.230 0.227**
(94.584) (79.734)
Loglikelihood 542.733 704.648
Notes: *Significant at 5 percent level; **significant at 1 percent level; other control variables
in the wage equation include: ten dummy variables for type of education, on-the-job training,
tenure at the current firm, years of work experience, five dummy variables for firm size and
two dummy variables for managerial work
Table IV.
Job satisfaction
of older workers
355
vocational qualifications earn about 6 percent more, while workers with lower
general qualifications earn 16 percent more. At the intermediate level,
workers with vocational qualifications earn about 24 percent more than
workers with only primary school, while workers with intermediate general
qualifications earn 31 percent more. These findings are consistent with earlier
research for The Netherlands where it was also found that vocational
qualifications offer lower earnings perspectives than general qualifications
(see Groot, 1994).
Skill mismatches do not have a significant direct effect on job satisfaction.
This is contrary to some earlier studies for the USA, where a significantly
negative relation between over-education and job satisfaction was found
(Tsang, 1987; Tsang et al., 1991). The difference between our findings and some
of these previous studies may be due to the specific age group in our sample.
Skill mismatches do have a significant effect on wages. Over-educated
workers earn 6 to 9 percent less than workers with the same educational
qualifications earn in a job which matched their education level. Under-
educated workers earn 12 percent more than workers with a job that matches
their skills.
The employment history of the worker does not have an effect on his/her job
satisfaction. Neither tenure, experience, unemployment or out of the labor force
has a significant effect on job satisfaction. A year of tenure increases wages by
0.2 percent. Spells out of work decrease wages. This effect is larger for a spell of
unemployment than for a spell out of the labor force. Unemployment decreases
the wage rate by about 7.5 percent, a spell out of the labor force by about 5
percent. Spells of unemployment or out of the labor force depreciate human
capital: productive skills are lost or become obsolete while one is not employed.
Further, spells out of work may stigmatize workers. These stigmatizing effects
may be larger for involuntary spells (unemployment) than for voluntary spells
(out of the labor force). These stigmatizing effects of spells out of work may
explain the difference between the wage effects of unemployment and out of the
labor force.
Of the other individual characteristics only health status has a significant
effect on job satisfaction. Workers who report that their health is (very) good
are more satisfied with their job than workers who are in bad health.
Contrary to some of the earlier studies we do not find any significant
differences in job satisfaction between men and women. This may be due to the
fact that we include only heads of households in our analysis and therefore
avoid a possible source of labor supply bias in our results. The wage gap
between men and women is about 10 percent.
Of the firm related variables we find significant effects of early retirement
arrangements, industry, and hours of work. If the firm has an early retirement
arrangement, workers are more satisfied with their job (only in the equation
using the composite variable).
There are few significant differences in job satisfaction between industries.
The exception is workers in public education. Workers in education are
International
Journal of
Manpower
20,6
356
significantly less satisfied with their work than workers in other industries. It
is well known that the teaching profession offers few opportunities for
advancement or promotion. Further the burden of work in education is heavy,
especially for teachers in secondary education, and as a result the outflow to
disability is relatively large. Finally, due to budgetary measures the earnings of
workers in education are lagging behind those of workers with the same level
of qualifications in other industries.
Job satisfaction decreases with hours of work: part-time workers are more
satisfied with their job than full-time workers.
Finally, in both equations the correlation coefficient between job satisfaction
and wages is not significantly different from zero. This suggests that relative
overpayment does not affect overall job satisfaction.
Table V contains the parameter estimates of the satisfaction with three job
aspects underlying total job satisfaction, i.e. the coefficients of satisfaction with
the job content, the remuneration and the work time. The coefficients in Table
V show that the effects of variables differ between the three job aspects
distinguished. Only the health variable has a significantly positive effect in all
three equations.
Education level has a significant effect on satisfaction with the
remuneration, but not on satisfaction with the job content or the work time. The
predicted wage rate has a significantly positive effect on satisfaction with the
remuneration only. The predicted wage rate has an insignificant effect on
satisfaction with the job content, and a significantly negative effect on
satisfaction with work time. Higher paid workers are less satisfied with the
hours they have to work.
In the satisfaction with remuneration equation the correlation coefficient
between the satisfaction equation and the wage equation is significantly
positive, indicating that workers who are relatively overpaid are more satisfied
with their remuneration than workers who are not overpaid. Relative
overpayment does not have a significant effect on satisfaction with the two
other job aspects distinguished.
Male workers are more satisfied with their work time than female workers.
Women with children may find it difficult to adjust their hours of work to
school hours of their children. Men and women do not differ in their satisfaction
with the job content and the remuneration of the job.
Workers who work part-time are more satisfied with their working hours
than workers who work full-time. In a sense this is remarkable, as commuting
time for part-time workers is relatively higher than for full-time workers. Part-
time workers are more satisfied with their job content as well. There is no
difference between part-time and full-time workers in their satisfaction with the
remuneration of the job.
Workers who have been unemployed in the past are less satisfied with their
hours of work, but not with the other two job aspects.
Job satisfaction
of older workers
357
Satisfaction job
content
Satisfaction
renumeration
Satisfaction
worktime
Intercept 1.856 4.457** 5.177**
(1.396) (2.881) (3.677)
Human capital variables: education level (reference category: primary school)
Lower vocational education 0.070 0.132 0.075
(0.648) (1.020) (0.664)
Lower general education 0.206 0.471** 0.122
(1.412) (2.719) (0.778)
Intermediate general education 0.228 0.579* 0.231
(0.952) (2.019) (0.874)
Intermediate vocational education 0.296 0.376 0.119
(1.582) (1.697) (0.604)
Higher vocational education 0.393 0.828* 0.309
(1.261) (2.290) (0.948)
University 0.529 1.308* 0.498
(1.196) (2.489) (1.057)
Job match quality
Overeducated 0.079 0.134 0.118
(0.692) (0.897) (0.950)
Undereducated 0.060 0.138 0.155
(0.554) (1.104) (1.353)
Specific training and employment record
Unemployed 0.175 0.100 0.276**
(1.918) (0.971) (2.948)
Out of labor force 0.183 0.163 0.140
(1.810) (1.326) (1.303)
Other individual characteristics
Male 0.196 0.040 0.519**
(1.256) (0.200) (3.042)
Married 0.063 0.248 0.102
(0.512) (1.662) (0.834)
Ethnicity 0.074 0.033 0.202
(0.522) (0.182) (1.503)
Health 0.573** 0.315** 0.412**
(7.235) (3.517) (5.272)
Job characteristics
Early retirement arrangement 0.034 1.040** 0.299**
(0.391) (7.658) (3.253)
Public sector 0.075 0.021 0.153
(0.912) (0.230) (1.771)
Industry (reference category: industrial sector)
Agriculture 0.345 0.254 0.142
(continued)
Table V.
Parameter estimates
satisfaction different
job aspects (t-values in
brackets)
International
Journal of
Manpower
20,6
358
Workers in firms with an early retirement arrangement are more satisfied with
the remuneration of the job and with the working hours, but not significantly
more satisfied with the job content.
Workers in public education are significantly less satisfied with their job
content and their remuneration, but not with their working hours. Workers in
public education usually have more holidays and shorter actual work weeks
than other workers.
Conclusion
Five conclusions emerge from our analysis of job satisfaction of older workers
in The Netherlands. The first conclusion is that satisfaction with the content of
the job is the main factor to explain overall job satisfaction. Satisfaction with
co-workers, supervisors, the workload and the early retirement arrangement
also plays a role, but to a minor degree.
Secondly, the effects of individual and job characteristics on job satisfaction
differ by the aspect of the job considered. For example, financial variables
such as the expected wage rate and the indicator of relative overpayment
Satisfaction job
content
Satisfaction
renumeration
Satisfaction
worktime
(1.078) (0.673) (0.435)
Construction 0.021 0.095 0.079
(0.199) (0.759) (0.687)
Transport 0.011 0.235 0.080
(0.103) (1.828) (0.667)
Commerce 0.122 0.001 0.005
(1.090) (0.004) (0.045)
Commercial services 0.114 0.008 0.008
(0.886) (0.048) (0.063)
Public education 0.521** 0.548** 0.229
(4.092) (4.069) (1.883)
Health care 0.223 0.058 0.202
(1.683) (0.311) (1.380)
Hours of work 0.018** 0.002 0.018**
(3.095) (0.263) (3.327)
Predicted wage rate 0.157 1.894** 1.248**
(0.303) (3.127) (2.234)
c 2.137** 1.352** 1.624**
(31.454) (23.151) (24.851)
, 0.019 0.112** 0.010
(0.573) (2.846) (0.287)
Loglikelihood 1,172.30 994.156 1,203.16
Notes: *Significant at 5 percent level; **significant at 1 percent level; for reasons of space
the coefficients of the wage equations are not presented in this Table
Table V.
Job satisfaction
of older workers
359
have an effect on satisfaction with the remuneration of the job, but not on other
job aspects. Satisfaction with the job content and with working hours decrease
with hours actually worked, but hours worked has no effect on satisfaction
with the remuneration of the job.
The third conclusion that can be drawn is that the response to a general
question on job satisfaction may differ from the response to questions on
satisfaction with different aspects of the job. For example, if all ten aspects of
the job distinguished in this paper are considered jointly significant effects of
wages and education are found, while no such effects are found if the general
question on job satisfaction is considered.
The fourth conclusion that emerges is that it is relevant to consider the joint
relation between wages and job satisfaction. Evidence is found that the
negative relation between education level and job satisfaction found in
previous research can be ascribed to the opposite direct and indirect effects of
education on job satisfaction. It is found that on balance the negative direct
effects of education on job satisfaction are canceled by positive indirect effects
of education on job satisfaction through wages.
Finally, our analysis does not suggest that skill mismatches affect job
satisfaction. We do find, however, that skill mismatches affect wages: over-
educated workers receive a lower return to their education level attained while
under-educated workers receive a higher return than correctly allocated
workers.
References
Abramis, D. (1994),``Work role ambiguity, job satisfaction, and job performance: meta-analysis
and review'', Psychological Reports, Vol. 75, pp. 1411-33.
Clark, A. (1993), ``Job satisfaction and gender: why are women so happy at work'', Discussion
Paper 4, ESRC Research centre on micro-Social Change, University of Essex.
Clark, A. and Oswald, A. (1995), ``Satisfaction and comparison income'', paper presented at
CEPR/ESRC University of Essex Workshop, London.
Groot, W. (1994), ``Differences in rates of return by type of education'', Education Economics,
Vol. 2, pp. 209-14.
Groot, W. and Maassen van den Brink, H. (1997), ``Allocation and the returns to over-education in
the United Kingdom'', Education Economics, Vol. 5, pp. 169-83.
Groot, W. and Maassen van den Brink, H. (1999), ``Over-education in the labor market: a meta-
analysis'', Economics of Education Review, forthcoming.
Gruber, J. and Wise, D. (1998), ``Social security and retirement: an international comparison'',
American Economic Review, Vol. 88, pp. 158-63.
Hartog, J. and Mekkelholt, E. (1989), ``Everybody happy?'', mimeo, Department of Economics,
University of Amsterdam, Amsterdam.
Kapteyn, A. and de Vos, K. (1998), ``Social security and labor-force participation in The Netherlands'',
American Economic Review, Vol. 88, pp. 164-7.
Mekkelholt, E. (1993), ``Een sequentiele analyse van de baanmobiliteit in Nederland'', PhD-thesis,
Department of Economics, University of Amsterdam, Amsterdam.
Mekkelholt, E. and Hartog, J. (1989), ``Mobiliteit en beloning'', OSA-Werkdocument W59, Den
Haag.
International
Journal of
Manpower
20,6
360
Mekkelholt, E., Brouwer, E. and Praat, W. (1991), ``Arbeidsmobiliteit en beloning in Nederland'',
OSA Werkdocument W88, Den Haag.
Tsang, M. and Levin, H. (1985), ``The economics of over-education'', Economics of Education
Review, Vol. 4, pp. 93-104.
Theeuwes, J. and Woittiez, I. (1995), ``Well-being and labor market status'', Research
Memorandum 95.01, Department of Economics, Leiden University, Leiden.
Tsang, M. (1987), ``The impact of under-utilization of education on productivity: a case study of
the US Bell companies'', Economics of Education Review, Vol. 6, pp. 239-54.
Tsang, M., Rumberger, R. and Levin, H. (1991), ``The impact of surplus schooling on worker
productivity'', Industrial Relations, Vol. 30, pp. 209-28.

S-ar putea să vă placă și