NÁRODOHOSPODÁŘSKÝ OBZOR – REVIEW OF ECONOMIC PERSPECTIVES,
VOL. 11, ISSUE 1, 2011, pp. 37–56, DOI: 10.2478/v10135-011-0006-y
WAGE DETERMINATION WITH SPECIAL REFERENCE TO ROLE IN
A FAMILY
Lenka Filipová, Zuzana Machová
VŠB – Technical University of Ostrava, Faculty of Economics, Sokolská třída 33, 701
21 Ostrava 1, lenka.filipova@vsb.cz, zuzana.machova@vsb.cz
Abstract: The paper has arisen as a part of the project dealing with the questions of
measuring of wage determinants and wage discrimination on the basis of different
subjective requirements of women and men on their wages. Using the ordered-response
model (Ordered Probit), it analysed wage determination on the basis of Mincer’s Wage
Regression including dummies for role in a family. The analysis was carried out with
the total sample of respondents and subsequently separately for men and women. The
data were gained from a questionnaire survey carried out in Ostrava city, Czech
Republic. In general, the analysis did not prove previous conclusions of theories or
empirical studies about prevailing influence of family characteristics on the wage rate.
Nevertheless, it proved statistically significant differences in the wage determination
between men and women, and showed that in the case of men, some family
characteristics may have been important for determining the wages.
Key words: Human Capital, Role in a Family, Wage Determinants, Wage
Discrimination.
JEL Classification: J24, J31, J71
DOI: 10.2478/v10135-011-0006-y
Volume XI, Issue I, 2011
37
WAGE DETERMINATION WITH SPECIAL
REFERENCE TO ROLE IN A FAMILY1
Lenka Filipová, Zuzana Machová2
Introduction
Wages are an important indicator in expressing human capital value. The standard Neoclassical
theory of the labour market states that wages (the marginal costs of labour)
should correspond with income from the marginal product of labour under
the conditions of ideal competition. This principle is the basis for standard human
capital theory which assumes that the amount of wages is influenced by the personality
characteristics affecting labour productivity. Mincer (1974) expressed the relation of
human capital to the amount of wages through a wage equation which has become
the methodology basis for simulating wage determinants up to the present.
More sophisticated theories have been developed (such as those by Dickens and Katz
(1987) or Krueger and Summers (1988), see further), modifying Mincer’s wage
regression by adding a labour status description (the position in one’s employment,
working hours, type of employment contract, and qualification requirements), corporate
factors (how large a company is, commercial and non-commercial sectors, and
industries), institutional factors (legislation protection of employment and the minimum
wage) and regional factors, which are now collectively referred to as wage determinants;
any differences in wages are due to these determinants.
If the wage differentiation is not explained by objective wage determinant
characteristics, it is referred to as wage discrimination.3
Nevertheless, no economic
studies have worked with the sociological aspect of different requirements of men and
women for the wage rate. According to sociological studies (Dudová, 2007; Hašková,
2003; Křížková, 2006; Moe, 2003 etc.), the unexplained differences in wages do not
have to be caused by discrimination but different preferences of men and women,
namely in relation to family existence. ‘General’ (gender) stereotype places an emphasis
on the ‘caring person type’ and the ‘family breadwinner type’, with these types
prevailingly matching the ‘simple’ division by gender. A shift in this perception is seen
in single-parent families. Men tend to put an emphasis on the caring type (part-time jobs
and lower wages) while women on the income for the family (full-time jobs and higher
wages). Where there are no children, the differences become blurred. The role of an
individual in the family can then be seen as one of the wage determinants. This can be
set in the preference theory which, as a part of the neoclassical economy, presents
1
The paper is supported by the SGS research project SP/201016 „Measuring Wage Determinants
and Wage Discrimination“, which is aimed at the analysis of wage determinants and
discrimination according to gender on the basis of different subjective requirements of women
and men on their wages.
2
VŠB – Technical University of Ostrava, Faculty of Economics, Sokolská třída 33, 701 21
Ostrava. Email: lenka.filipova@vsb.cz, zuzana.machova@vsb.cz.
3
Measuring the extent of wage discrimination on the basis of Oaxaca-Blinder decomposition of
wage differences is the objective of authors’ further research.
REVIEW OF ECONOMIC PERSPECTIVES
38
standard economic explanation of the decision-making process of individuals. Since
Gary Becker (Theory of preferences, 1997) this theory has gained its relevance for all
aspects of life decisions.
The aim of this paper is to analyze the influence of the family characteristics on the
wage determination of men and women.
For our purpose, it is necessary to connect a contemporary economic quantitative
research of wage determinants with a sociological qualitative research of the impact of
family roles and preferences on wage. The analysis is methodologically based on the
modified Mincer wage regression, where the vector of explanatory variables includes
three types of factors: personal, firm and family characteristics.
The first part of the article includes the summary of present knowledge of the issue in
question on both theoretical and empirical level. The second part focuses on
the description of data acquired by questionnaire survey carried out in the territory of
Ostrava city and methods of their processing. The third part includes the results of
the econometric analysis.
Theoretical and empirical background
The approach usually used for wage determinants measuring is Mincer’s approach
based on the theory of human capital which reflects personal characteristics of
individuals. According to this theory, an employee’s wage is determined by the quality
given by education, general working experience, specific working experience acquired
in the company and other qualities independent of education and experience. Empiric
literature1
proved Mincer’s wage function to be valid in basic and modified forms and
also in countries with different institutional structures. However, the proportion of
education and experience to explain differences in wages between employees varies by
country and it ranges approximately from 30 % to 50 %.
There are many studies using Mincer’s wage regression for wage differences
determination in the Czech and Slovak labour market. The detailed analysis was carried
out by, among others, Filer, Jurajda and Plánovský (1999). The rate of investment return
indicator into education in 1997 ranged from 6.5 % to 9.0 % per year and reached much
higher values than stated by Flanagan (1995) for 1991 and by Chase (1998) for 1993,
which means an increasing significance of education in the labour market. Later era
(around 2002) was elaborated in the study of Gottvald et al. (2002).
Other theories of wage determinants emphasize differences in occupation characteristics,
for instance the theory of job competition. The connection of the theory of human
capital and the theory of job competition is carried out by allocation theory which shows
that people are characterized by skills, and that job vacancies are characterized by
complexity. There can be comparative advantages when employing particular
individuals in particular jobs which are used when there is an effective labour market.
1
A very good overview of this literature can be found in the Handbook of Labor Economics,
North-Holland Publishing Company, 1999. Basic references include Gery S. Becker and B. R.
Chiswick, “Education and the Distribution of Earnings”, American Economic Review 56, 1966,
pp 358 – 369 and Jacob Mincer, “Schooling, Experience and Earnings”, New York: Columbia
University, 1974.
Volume XI, Issue I, 2011
39
The hypothesis of effective wages (Dickens and Katz, 1987; Davis and Haltiwanger,
1996) justifies the inclusion of indicators such as the branch of industry, how large
a company is and for how long it has existed. Active trade unions (Freeman and Medoff,
1986; Pencavel, 1991) justify the inclusion of the types of collective bargaining. Internal
wage structures (Krueger, 1993; Lazear, 1998) justify the inclusion of tenure. Thus,
wages are determined by personal characteristics, job vacancy characteristics and
institutional and market characteristics.
Much more present foreign research in the field of explanation of wage differences
between men and women uses not only economic approaches but also approaches from
other socioscientific disciplines, e.g. psychology or sociology. Their results show
different behaviour of men and women in relation to their employability in the labour
market (Jacobsen, 2007). In this context other factors influencing this behaviour are
studied as well, e.g. influence of the government and its institutions, existence of the
family and position of both genders in it (Boje, 2001; Moe, 2003). Length of maternity
leave is one of the factors of differences in wages. Although this subject is very
important from a political as well as an academic point of view, currently, there are only
a few studies that deal with this subject and analyze the impact of the length of
maternity leave on wages and employment of mothers and their results are not
conclusive, see Ejnaes and Kunze (2004), Schoenberg and Ludsteck (2007), Würtz
(2007) or Dustmann and Schoenberg (2008). The results of similar studies show
different levels of preferences of men and women in relation to the employment, for
instance different requirements for the length of employment but also in the negotiations
about wage rate. The theory of preferences then explains a search for optimum
combination of work and other activities in life that brings a person the highest
satisfaction possible. The position of each person on the labour market, as well as their
rate of participation in the other activities, is a result of their individual free choice
realized in certain social conditions.
So far, studies referring to the Czech Republic and other transitive economies have
focused especially on changes in the labour market during the transformation period and
their influence on wage differences between men and women. Jurajda (2001 and 2003)
says that women’s hourly wages are up to 30 % lower than men’s hourly wages in the
Czech Republic, whereas in the Slovak Republic, the wage difference is lower,
especially in the public sector. In the private sector, approximately two thirds of wage
differences are explained by discrimination factors in both republics. Jurajda (2005)
says that despite antidiscrimination policies have been introduced in the CZ, segregation
of women into low-wage jobs is the significant source of wage differences between men
and women.
Academic studies focusing on blending and harmonizing professional and family life
carried out in the Czech Republic and Slovakia point to the fact that organization of
careers and professional lives in couples still usually disadvantages professional life of
women to men’s career. Gender scheme or regulation is applied with respect to the
types of a ‘caring person type’ and the ‘family breadwinner type,’ which in absolute
majority correspond to the gender-based division. The studies also represent notions
about care distribution and paid work in the partners’ life. These notions are compared
to care actually taken and work assignment (Hašková et al., 2003; Křížková et al.,
2006.).
REVIEW OF ECONOMIC PERSPECTIVES
40
Ordered probit model of wage determinants: data, methodology and variables
In the project called Measuring Wage Determinants and Wage Discrimination
a questionnaire survey was carried out, the aim of which was to acquire data relating to
wage determinants of individual respondents according to Mincer’s theory, their
subjective requirements for wage rate and status in the family. The basic set researched
was the ’prime-age’ working population of the city of Ostrava. As the features of the
basic set are known from Population and Housing Census carried out in 2001,1
the
method chosen was quota selection which ensures consistent structure of the basic and
the selective set. Quotas 2
were provided for sex, age and field 3
which individual
respondents work in.
The survey was carried out in the city of Ostrava from May 26, 2010 to July 1, 2010.
399 questionnaires were distributed in total, out of which 391 were completed properly.
Subsequently, the selection of 300 questionnaires was made in order to meet the
established quotas. For further validation of consistency between the basic and selective
set structures with respect to defined quotas, a hypothesis of relative frequencies
equality was tested for each quota. Results of statistic tests confirmed relative
frequencies equality on the significant level of 5 % and the selection can therefore be
considered as representative.
The wage determinant analysis is based on a standard model stemming from Mincer’s
wage regression which is besides personal characteristics extended by traditional
institutional and corporate characteristics. However, the aim of the research is following
particularly other influences on the wage rate which are based on sociological studies
emphasising the influence of roles in the family on the wage rates. If the analysis
confirms the influence of family status and roles in the family on the wage rates, it will
change the view of gender-based discrimination. Part of the questionnaire would be also
questions related to the care of children and household.
Therefore, the distribution of selection set frequencies is shown in the following two
charts with respect to the care of children and household (hereinafter “care”) according
to the sex, education and wage rate before the econometric analysis. The question about
prevailing care offered respondents three types of responses: I take care mostly; my
partner takes care mostly, and we both take care equally. Only respondents having
children and with the marital status married or single living in common household were
included in the charts.
The first chart shows that responses of common care prevail for both sexes and
education, the second position is prevailing care of the particular respondent and least
respondents granted prevailing care to the partner. The responses of university educated
men are interesting, they grant prevailing care for themselves in 29 %, only 7 % of them
responded the partner and the rest, i.e. 64 % responded equal care. Whereas in the case
1
ČSÚ, 2005.
2
With respect to the fact that survey respondents were only working people, it was not possible to
set quotas for education as ČSÚ provides only Population and Housing Census results for the
whole population and it can be assumed that the educational structure does not correspond with
the educational structure of working population, in other words there are more people with low
education among unemployed people.
3
See Population and Housing Census 2001.
Volume XI, Issue I, 2011
41
of university educated women, 12 % of respondents stated the partner as the prevailing
caretaker, 24 % granted the care for themselves and 65 % stated equal care with
the partner. The distribution of responses of skilled men is interesting as well, 22 % of
them granted the prevailing care to their partners, 19 % to themselves and 59 % equal
care.
From this data distribution it is obvious that in case of subjective characteristics of role
perception in the household it would be proper to specify and choose questions more
precisely, so as not to be psychologically marked. Besides the question about the care of
children, there could have been written the question about the risk losing the
employment in the questionnaire. Here, an individual prefers job security with usually
lower wage or prefers the employment with the higher prestige and higher wage but
with a larger degree of responsibility and competition in the field or profession and thus
a higher risk. The question about the care of children and household could have been
specified to average number of hours spent with children daily, not the vague
formulated question about prevailing care. Moreover, there might have been written so
called check questions which could have helped to eliminate too subjective views of the
question. For instance the question about the number of hours spent with children could
be supported by the question about the average working hours a day, etc. This should be
taken into consideration for the future research.
Table 1: Children care according to sex and education
Children care
Sex Education Me % of total Partner % of total Equally % of total Total
Primary 2 33 4 67 6
Training 5 19 6 22 16 59 27
High 1 7 1 7 12 86 14
Male
University 4 29 1 7 9 64 14
Total of Male 12 20 8 13 41 67 61
Primary 0 2 100 0 2
Training 6 25 0 18 75 24
High 12 30 2 5 26 65 40
Clollege 0 0 2 100 2
Female
University 4 24 2 12 11 65 17
Total of Female 22 26 6 7 57 67 85
Total 34 23 14 10 98 67 146
Note: figures in the table state absolute numbers of respondents. Only the % of total columns
represents the percentage of a particular category to the total number of respondents in
the particular row.
The distribution of responses to the question about the care of children and household
shows slightly rising proportion of responses granting prevailing care to the partner with
the wage rise. Though, it is not possible to assume any more significant ties between
wage and care of children and household from such data distribution.
REVIEW OF ECONOMIC PERSPECTIVES
42
Table 2 Children care according to net wage1
Children care
Net wage Me % of total Partner % of total Equally % of total Total
to 10 000 CZK 4 27 2 13 9 60 15
10 001 - 16 000 CZK 11 20 5 9 38 70 54
16 001 - 22 000 CZK 9 20 2 4 34 76 45
22 001 - 28 000 CZK 6 30 3 15 11 55 20
28 001 - 34 000 CZK 1 17 2 33 3 50 6
34 001 - 40 000 CZK 1 33 0 2 67 3
40 001 - 46 000 CZK 1 100 0 0 1
52 001 CZK and more 1 50 0 1 50 2
Total 34 23 14 10 98 67 146
Note: figures in the table state absolute numbers of respondents. Only the % of total columns
represents the percentage of a particular category to the total number of respondents in
the particular row.
Traditional model of wage determinants is simply expressed by the following equation:
ititit XW εβ +=ln ,
where i represents an individual and t time period. Dependent variable (wage) is
expressed in a logarithmic form and X represents independent variables vector, in this
case wage determinants. Theory of human capital is the theoretical foundation of this
specification and coefficients are interpreted as earnings from investments into the
human capital. ε represents an error term.
In this analysis, however, it is not possible to express wage in the logarithmic form and
to use the method of least squares as the wage is created in categories. Ordinal logit or
probit models are usually used for categorically explained variables. Eviews Software
uses the ordered-response model (Ordered Probit) with analytic second derivative
methods to obtain parameter and variance matrix of the estimated coefficient estimates
(Quadratic hill climbing) for the partivular variable type.
Hence, the model has the following form:
iii XW εβ += ,
where t index is missing as data only for one observation period are used. Explanatory
variables have the character of dummy variables. In total, 10 variables are used for wage
determinant definition, out of which 3 represent personal characteristics stemming out
of Mincer’s wage regression (sex, age, education), 4 characterize institutional and
corporate factors (ISCO2
, NACE3
, ownership, company size). For our purposes of
observation of the influence of family status on wage rate, there are 3 other variables
1
Net wage means disposable for respondents.
2
International Standard Classification of Occupations.
3
Nomenclature générale des Activités économiques dans les Communautés Européennes
(classification of economic studies).
Volume XI, Issue I, 2011
43
used in the analysis, 2 of which are objective (number of children, marital status) and
one variable represents subjective view of the role in the family in the question who is
involved more in care-taking of children and household.
Vector X involves the following variables.
A: personal characteristics stemming from Mincer’s wage equation:
• dummies for gender,
• 4 dummies for age (in categories),
• 5 dummies for education.
B: institutional and corporate factors:
• dummy variables for 10 ISCO categories,
• dummy variables for 21 NACE categories,1
• dummy variables for 2 types of ownership,
• dummy variables for 6 company size categories.
C: characteristics of family status and roles in the household:
• number of children as dummies for 6 categories,
• family status – single or married/partnership,
• care of children – dummies for 3 categories: me prevailing, partner prevailing, both
equally.
As mentioned above, the data sample includes 300 individuals: 146 men and 154
women. First, the analysis was carried out with the total sample of respondents,
subsequently separately for men and women. First, all the variables mentioned above
are included, then the separately personal characteristics, institutional and corporate
characteristics and family characteristics to verify economic significance including
variables into the analysis.
Results of the total respondent sample (see Table 3 in the Appendix) show that the
model is statistically significant as a whole, as LR statistics came out statistically
significant. There had to be made slight adjustments in explanatory variables as for the
integration of some categories which were marked for great variability in observation
and the software could not process calculation of coefficients with such unbalanced
data. This category adjustment was used for all following analysis. Category integration
was made in the following cases:
• education (1 primary + 2 vocational education; 3 secondary with “maturita” exam
+ 4 college),
• ISCO (ISCO 1 + 2; ISCO 6 + 7; ISCO 8 + 9),
• number of children (category 4+5+6, which equals 3 and more children).
The variable explained is marked for great variability of the number of observations as
the following figure shows. Hence, categories were integrated here too, namely 5 – 9,
and analysis was carried out for original and new categories separately. The results did
1
NACE had to be excluded from the analysis. Ordered probit model did not allow to include this
variable into the calculation, as some categories include only very little observation sample and
category integration would not bring any economic significance in case of the field.
REVIEW OF ECONOMIC PERSPECTIVES
44
not differentiate much, which is the reason why the original wage categorization is
stated below and used for the result interpretation.
Figure 1: Frequency of net wage
0
20
40
60
80
100
120
1 2 3 4 5 6 7 8 9 10
Frequency
real wage
Note: real wage description equals net wage in our sense. Real wage was used in
the questionnaire for comprehensibility for respondents as the description of real wage on their
accounts.
Analysis and results
Mincerian presumptions about wage determinants were confirmed from the point of
view of variables, i.e. the influence of personal characteristics. Sex and age variables are
statistically significant. From corporate statistics only company size variable proved to
be slightly statistically significant, namely with categories 4 and 5 (250 to 5000
employees). From family statistics only the number of children variable is slightly
statistically significant, namely in the category 3 and more children only.
However, coefficient interpretation is more complex in cases of models with
qualitatively (categorically) explained variable given than in the case of the OLS
method. Here coefficient shows the probability to change the category of the explained
variable with the change of particular explanatory variable (result is calculated for
extreme, or end categories), whereas it is not possible to determine which category of
explained variable is referred to. The result only shows whether the category is going to
increase in case of coefficient plus sign (in our case it will be higher wage), or in case of
coefficient minus sign to decrease, i.e. the explained variable drops to lower wage. In
our case it is essential to take into account the fact that other (explanatory) variables are
expressed in the categories as dummy variables as well, i.e. in order to be clear, the
result needs to be interpreted in comparison to reference category which represents first
dummy variable (category). Thus, for instance the variable coefficient SEX_ 2 (women)
shows that women have higher probability that their wage will be in the low-wage
category than men; or for instance the variable coefficient EDU_5 (university
education) says that university educated people have higher probability that their wage
is in the high-wage category in comparison to people with primary education and
vocational education. This analysis does not include variables related to the subjective
Volume XI, Issue I, 2011
45
view of the care of children as it would mean to exclude childless households from the
analysis, thus the loss of the particular sample of individuals.
Furthermore, only analysis (see Table 4 in the Appendix) of personal characteristics on
the basis of Mincerian wage equation was carried out. Here the statistical significance
and the value of coefficients with particular variables do not differ from the previous
equation which points to the fact that personal characteristics play a significant role for
wage rate determination, as the theory of human capital suggests. The only exception in
comparison to the previous equation is an explicit, albeit very weak statistical
significance of age, namely with the category AGE_2 (30 – 39 years). The coefficient
shows that people in this age category have higher probability that their wage will range
within the higher categories than in case of the reference age category, which is 20 – 29
years.
When following institutional and corporate characteristics, all categories except for
ISCO_3 - Technicians and associate professionals proved (see Table 5 in the Appendix)
to be statistically significant. According to the presumptions, there is a coefficient minus
sign with all followed ISCO categories pointing at the probability of lower wage in
comparison to the reference category, which is ISCO 1 and 2 (Managers and
Professionals). The property variable is also statistically significant when employees in
the private sector have greater probability of higher wages compared to the public
sector. The company size indicator does not influence wages significantly, only
categories 4 and 5 (250 – 4999 employees) marked weak statistical significance with
plus coefficient values which represent the probability of higher wage in comparison to
reference category (up to 10 employees).
If we include only subjective and family characteristics into the equation, the model as
a whole is statistically insignificant (see Table 6 in the Appendix), for the reason that
except for family status, no category out of those statistically significant variables has
a significant impact on wage category. STATUS_2 category coefficient shows that
married couples or people living in partnership have higher probability that their wage
is in higher categories.
Analysis for men and women separately
The analysis carried out separately for men and women showed differences in statistical
significance for wage determinants. When including all variables, the model came out
statistically relevant for both men (see Table 7 in the Appendix) and women (see
Table 8 in the Appendix). With men, statistical relevance appeared in case of personal
characteristics with the age category 4 (40 – 49 years), with minus, then with both
educational categories with the presumed plus coefficient sign.
Furthermore, it was necessary to aggregate ISCO categories into larger units (ISCO
1,2,3; ISCO 4,5,6; ISCO 7,8,9) for the calculation than in the case of analysis of the
total sample. This aggregation is used for the rest of the component analysis. From
corporate characteristics, category ISCO 4,5,6 with plus sign is slightly statistically
significant; company ownership (employments in private sector have higher probability
of higher wage category than employments in public sector); company size categories 4
and 5 again with plus sign. From family characteristics the number of children variable
REVIEW OF ECONOMIC PERSPECTIVES
46
proved to be statistically significant, namely in categories 3 and more children, they
have also 1.5 times higher probability of higher wage category.
With women the situation is different. Only the age variable in the category 30 – 39
years with coefficient plus sign proved to be statistically significant from personal
characteristics. The education variable did not prove to be statistically significant with
women. Moreover, the university education category had to be excluded due to little
number of observations. Corporate characteristics play a more significant role with
women for wage rate determination. Basically, all observed variables and their
categories were marked for statistic significance. The most impact proved to be in ISCO
variable and in two other company size categories (4 and 5). Family characteristics do
not play a significant role with women for wage rate determination; none of the
variables is statistically significant.
When including personal characteristics stemming from Mincer’s equation, the model
with men came out statistically significant but only the education variable was
statistically significant, namely in the university education category with the suggested
plus sign showing the probability of higher wage categories. With women, the model
with personal characteristics came out only slightly statistically significant, whereas the
age category 30 – 39 years with plus sign was slightly statistically significant only (ask
authors for full results).
When analysing institutional and corporate characteristics separately (ask authors for
full results), the model did not come out statistically significant for men as category
ISCO 7, 8, 9 with the suggested minus sign was the only statistically significant variable
from the observed characteristics. The other variables and their categories did not play
any significant role with men for wage rate determination. With women the situation is
opposite. When analysing corporate characteristics, the model came out statistically
significant as all observed variables in all their categories are statistically significant.
High statistic relevance is present especially in variable ISCO with coefficient minus
sign, as if the ISCO variable compensated the importance of education. Company
ownership and company size variables except for categories 3 and 6 are also significant.
Nevertheless, the significance in other categories proves that categories according to
company size are relevant for wage rate determination.
The analysis of family characteristics did not prove statistical significance with both
men or women (ask authors for full results). With women none of the observed variable
proved to be statistically significant, with men the family status variable and number of
children variable in the category 3 and more children are slightly significant. Plus sign
in the category STATUS_2 says that married men or men living in partnership have
higher probability for higher wage category than single living men.
Conclusion
The aim of this paper was to carry out the analysis of wage determinants extended by
family status and family roles characteristics. The research itself stemmed out of
the hypothesis that the role in a family could have been the significant wage rate
indicator. According to sociological studies, individuals can be divided into those who
take care of children and household more, and those who work more and financially
provide for the family. On the basis of the role in a family, it is possible to expect
Volume XI, Issue I, 2011
47
different requirements for the wage rate when persons taking care more will prefer
greater job security usually connected with lower wage and vice versa persons working
more will prefer more prestigious jobs with promotion possibilities and with the demand
for a higher wage. This hypothesis corresponds to the theory of preferences which as
a part of the neoclassical economy presents standard economic explanation of the
decision-making process of individuals.
The analysis of wage determinants was based on the standard model stemming from
Mincer’s wage regression which was besides personal characteristics extended by
traditional institutional and corporate characteristics. The model was lately
complemented by other variables following the influence of family status and family
roles on wage rate. If the analysis confirmed their influence on wage height, it would
have also changed the view of gender-based discrimination which will be the subject of
the authors’ following research.
The results of the total respondent sample analysis showed that the model was
statistically significant as a whole. From the variables point of view, Mincerian
presumptions about wage determinants in particular proved to be right, i.e. the influence
of personal characteristics. When analysing personal, corporate and family statistics
separately, the models were statistically significant with statistically significant
variables in case of personal and corporate characteristics. The model for family
characteristics was not significant as the only statistically significant variable was the
family status in this case. The other two family characteristics did not prove their
significance.
The analysis carried out for men and women separately showed differences in statistical
significance of wage determinants. When including all variables, the model came out
statistically significant for both men and women. With men, personal characteristics
such as education and age played an important role. When the analysis was carried out
separately for personal, corporate and family characteristics, family status and the
number of children, namely in the category 3 and more children, variables came out
slightly statistically significant with men.
Statistical significance of education was not proven with women. Generally, the model
with personal characteristics came out slightly statistically significant. On the contrary,
corporate characteristics played an important role with women for wage rate
determination, especially in ISCO variable (job classification). The model for family
characteristics did not prove statistical significance; none of the variables was
statistically significant.
The separate analysis for men and women proved that the family status and number of
children could play a significant role in wage determination of men, i.e. men, unlike
women, had higher probability of having higher wage in case they were married and
had 3 and more children. If we consider men being the ‘family breadwinner type’ in
most, the findings can indicate that a part of the difference in wages of men and women
can be explained by the role in a family which reduces the extent of potential wage
discrimination according gender.
Nevertheless, the above mentioned seems to be very week in the context of the whole
analysis that in general did not prove previous conclusions of theories or empirical
REVIEW OF ECONOMIC PERSPECTIVES
48
studies about prevailing influence of personal characteristics on the wage rate. The
influence of family characteristics, especially family roles on the wage rate, failed to be
proved. On the other hand, the problem can be seen especially in asking questions about
the delicate topic of family roles which can bring rather psychologically marked and
little objective responses. It should be advisable to take into account this aspect in future
research. Furthermore, check questions which could help eliminate too subjective
viewpoints of the question. For instance, the question about number of hours spent with
children could be supported by the question about the average working hours a day, etc.
References
BOJE, T. P. (2001). Gender, Welfare State and the Market: Towards a New Division of
Labour (Routledge Research in Gender and Society, 4). Routledge, London. ISBN: 0-
203-99228-8.
ČSÚ. (2005). SLDB 2001. Retrieved from http://www.czso.cz/sldb/sldb2001.nsf/index.
DAVIS, S. J.; HALTIWANGER, J. (1996). Employre Size and the Wage Structure in
the United States. Annale d´Economie et de Statistique 41/42, 323-368.
DICKENS, W., KATZ, L. (1987). Inter-industry Wage Differences and Industry
Characteristics. In: LANG, K., et al. Unemployment and the Structure of Labor Market.
Basil-Blackwell, Oxford.
DUDOVÁ, R. et al. (2007). Development of Labour Market and Private, Family and
Partner Life. Sociological Studies 07(3). Sociological Institute AV ČR, Prague. 165 s.
DUSTMANN, CH.; SCHOENBERG, U. (2008). The Effect of Expansions in Maternity
Leave. Coverage on Children’s Long-Term Outcomes. IZA Discussion Papers 3605.
Institute for the Study of Labor (IZA).
EJRNAES, M.; KUNZE, A. (2004). Wage Dips and Drops Around First Birth. IZA
Discussion Papers 1011. Institute for the Study of Labor (IZA).
FILER, R. K., JURAJDA, Š., PLÁNOVSKÝ, J. (1999). Returns to the Market: Valuing
Human Capital in the Post-Transition Czech and Slovak Republic. CERGE-EI, Prague.
FLANAGAN, R. J. (1995). Wage Structures in the Transition of the Czech Republic.
International Monetary Fund, IMF Staff Papers (4), 836-854.
FREEMAN, R. B.; MEDOFF, J. L. (1986). The Two Faces of Unionism. In:
REYNOLDS, L. G.; MASTERS, S. H.; MOSER, C. H. Reading in Labor Economics
and Labour Relations. Englewood Cliffs, Prentice Hall, N. J.
GOTTVALD, J. et al. (2002). Determinants of Individual Pay and Firms Pay Structures
in the Czech and Slovak Republic. Technical University of Ostrava, Ostrava. ISBN 80-
248-0150-7.
HAŠKOVÁ, H. et al. (2003). Equal Possibilities of Men and Women When
Harmonizing Work and Family?. Sociological Institute AV ČR, Prague. ISBN 80-7330-
048-6.
Volume XI, Issue I, 2011
49
CHASE, R. S. (1998). Markets for Communist Human Capital: Returns from Education
and Experience in the Czech Republic and Slovakia. Industrial and Labor Relation
Review 3, 401-423. Cornell University.
JACOBSEN, J. (2007). The Economics of Gender. Blackwell publishing. ISBN: 978-1-
4051-6182-4.
JURAJDA, Š. (2001). Gender Wage Gap and Segregation in Late Transition. Working
Paper 182, CERGE-EI, Prague.
JURAJDA, Š. (2003). Gender Wage Gap and Segregation in Enterprises and the Public
Sector in Late Transition Countries. Journal of Comparative Economics 31(2), 199-222.
Elsevier.
JURAJDA, Š., (2005). Gender Segregation and Wage Gap: An East-West Comparison.
Journal of the European Economic Association 3(2-3), 598-607. MIT Press.
KRUEGER, A. B. (1993). How Computers Have Changed the Wage Structure:
Evidence from Microdata, 1984-1989. Quarterly Journal of Economics 108 (1), 33-60.
KRUEGER, A. B., SUMMERS, L. H. (1988). Efficiency wages and the inter-industry
wage structure. Econometrica 56, pp. 259–293.
KŘÍŽKOVÁ, A. et al. (2006). Pracovní a rodinné role a jejich kombinace v životě
českých rodičů: plány versus realita. Sociologické studie/Sociological studies 06(14).
Sociologický ústav AV ČR, Praha.
LAZEAR, E. P. (1998). Personnel Economics for Managers. John Wiley and Sons,
New York.
MINCER, J. (1974). Schooling, Experience and Earnings. Columbia University, New
York.
MOE, K. (2003). Women, Family, and Work: Writings on the Economics of Gender.
Blackwell Publishing. ISBN: 0-631-22576-5.
PENCAVEL, J. (1991). Higher Education, Productivity and Earnings: A Review.
Journal of Economic Education 22 (4), 331-359.
SCHOENBERG, U.; LUDSTECK, J. (2007). Maternity Leave Legislation, Female
Labor Supply and the Family Wage Gap. IZA Discussion Papers 2699. Institute for the
Study of Labor (IZA).
WÜRTZ, A. (2007). The Long-Term Effecton Children of Increasing the Length of
Parents Birth-Related Leave. Department of Economics Working Paper 07-11. Aarhus
School of Business.
REVIEW OF ECONOMIC PERSPECTIVES
50
Appendix
Table 3
Sample: 1 300
Included observations: 300
Number of ordered indicator values: 8
Convergence achieved after 5 iterations
Coefficient Std. Error z-Statistic Prob.
SEX_2 -0.848307 0.162519 -5.219733 0.0000
AGE_2 0.187437 0.205926 0.910219 0.3627
AGE_3 -0.071725 0.234277 -0.306156 0.7595
AGE_4 -0.244319 0.240104 -1.017555 0.3089
EDU_3_4 0.746601 0.181814 4.106390 0.0000
EDU_5 1.490093 0.252873 5.892652 0.0000
STATUS_2 0.018132 0.151529 0.119662 0.9048
ISCO_3 -0.156457 0.236169 -0.662478 0.5077
ISCO_4 -0.276071 0.259500 -1.063858 0.2874
ISCO_5 -0.192397 0.298913 -0.643655 0.5198
ISCO_6_7 -0.254058 0.286624 -0.886380 0.3754
ISCO_8_9 -0.398214 0.286991 -1.387549 0.1653
OWNER_2 0.275483 0.160247 1.719112 0.0856
FIRM_SIZE_2 0.200496 0.237612 0.843797 0.3988
FIRM_SIZE_3 0.097574 0.222725 0.438091 0.6613
FIRM_SIZE_4 0.690253 0.244900 2.818511 0.0048
FIRM_SIZE_5 0.668688 0.260625 2.565705 0.0103
FIRM_SIZE_6 0.392722 0.260997 1.504699 0.1324
CHILDREN_2 0.100712 0.220422 0.456904 0.6477
CHILDREN_3 0.237133 0.218062 1.087458 0.2768
CHILDREN_4_5_6 0.741318 0.292991 2.530179 0.0114
Limit Points
LIMIT_2:C(22) -0.931091 0.396977 -2.345453 0.0190
LIMIT_3:C(23) 0.588157 0.396409 1.483713 0.1379
LIMIT_4:C(24) 1.716988 0.403456 4.255699 0.0000
LIMIT_5:C(25) 2.581186 0.415752 6.208467 0.0000
LIMIT_6:C(26) 3.083134 0.432680 7.125671 0.0000
LIMIT_7:C(27) 3.524463 0.464297 7.590967 0.0000
LIMIT_9:C(28) 3.658612 0.478055 7.653120 0.0000
Volume XI, Issue I, 2011
51
Pseudo R-squared 0.150241 Akaike info criterion 2.736209
Schwarz criterion 3.081895 Log likelihood -382.4313
Hannan-Quinn criter. 2.874553 Restr. log likelihood -450.0471
LR statistic 135.2315 Avg. log likelihood -1.274771
Prob (LR statistic) 0.000000
Table 4
Sample: 1 300
Included observations: 300
Number of ordered indicator values: 8
Convergence achieved after 4 iterations
Coefficient Std. Error z-Statistic Prob.
SEX_2 -0.925409 0.138460 -6.683601 0.0000
AGE_2 0.401255 0.173230 2.316323 0.0205
AGE_3 0.242185 0.180751 1.339876 0.1803
AGE_4 0.131601 0.181200 0.726276 0.4677
EDU_3_4 0.726254 0.148295 4.897353 0.0000
EDU_5 1.504578 0.176837 8.508264 0.0000
Limit Points
LIMIT_2:C(7) -1.186187 0.169214 -7.009968 0.0000
LIMIT_3:C(8) 0.280483 0.154185 1.819125 0.0689
LIMIT_4:C(9) 1.356898 0.166480 8.150536 0.0000
LIMIT_5:C(10) 2.181798 0.195026 11.18720 0.0000
LIMIT_6:C(11) 2.657638 0.228974 11.60675 0.0000
LIMIT_7:C(12) 3.066424 0.276517 11.08947 0.0000
LIMIT_9:C(13) 3.187081 0.294803 10.81089 0.0000
Pseudo R-squared 0.125076 Akaike info criterion 2.711714
Schwarz criterion 2.872212 Log likelihood -393.7572
Hannan-Quinn criter. 2.775946 Restr. log likelihood -450.0471
LR statistic 112.5798 Avg. log likelihood -1.312524
Prob (LR statistic) 0.000000
REVIEW OF ECONOMIC PERSPECTIVES
52
Table 5
Sample: 1 300
Included observations: 300
Number of ordered indicator values: 8
Convergence achieved after 4 iterations
Coefficient Std. Error z-Statistic Prob.
ISCO_3 -0.155146 0.226620 -0.684611 0.4936
ISCO_4 -0.808845 0.238228 -3.395251 0.0007
ISCO_5 -1.078983 0.258581 -4.172714 0.0000
ISCO_6_7 -0.746683 0.234634 -3.182328 0.0015
ISCO_8_9 -1.132370 0.232137 -4.878025 0.0000
OWNER_2 0.404857 0.149855 2.701660 0.0069
FIRM_SIZE_2 0.101602 0.230116 0.441527 0.6588
FIRM_SIZE_3 0.029972 0.215751 0.138918 0.8895
FIRM_SIZE_4 0.466645 0.233534 1.998190 0.0457
FIRM_SIZE_5 0.485584 0.247600 1.961167 0.0499
FIRM_SIZE_6 0.251072 0.248718 1.009463 0.3128
Limit Points
LIMIT_2:C(12) -1.521981 0.289194 -5.262848 0.0000
LIMIT_3:C(13) -0.221075 0.284558 -0.776907 0.4372
LIMIT_4:C(14) 0.738595 0.286500 2.577996 0.0099
LIMIT_5:C(15) 1.486671 0.292679 5.079525 0.0000
LIMIT_6:C(16) 1.916914 0.304605 6.293109 0.0000
LIMIT_7:C(17) 2.283128 0.328822 6.943352 0.0000
LIMIT_9:C(18) 2.401618 0.342391 7.014259 0.0000
Pseudo R-squared 0.058549 Akaike info criterion 2.944648
Schwarz criterion 3.166875 Log likelihood -423.6972
Hannan-Quinn criter. 3.033584 Restr. log likelihood -450.0471
LR statistic 52.69969 Avg. log likelihood -1.412324
Prob (LR statistic) 0.000000
Volume XI, Issue I, 2011
53
Table 6
Sample (adjusted): 2 300
Included observations: 202 after adjustments
Number of ordered indicator values: 8
Convergence achieved after 4 iterations
Coefficient Std. Error z-Statistic Prob.
STATUS_2 0.463549 0.204076 2.271454 0.0231
CHILDREN_3 0.128385 0.168174 0.763402 0.4452
CHILDREN_4_5_6 0.257828 0.235916 1.092883 0.2744
CH_CARE_2 -0.065076 0.311325 -0.209028 0.8344
CH_CARE_3 -0.137547 0.186816 -0.736268 0.4616
Limit Points
LIMIT_2:C(6) -0.763718 0.190913 -4.000344 0.0001
LIMIT_3:C(7) 0.417802 0.185373 2.253840 0.0242
LIMIT_4:C(8) 1.213909 0.195244 6.217397 0.0000
LIMIT_5:C(9) 1.920934 0.221731 8.663353 0.0000
LIMIT_6:C(10) 2.228114 0.244531 9.111774 0.0000
LIMIT_7:C(11) 2.477420 0.272864 9.079324 0.0000
LIMIT_9:C(12) 2.595962 0.290776 8.927709 0.0000
Pseudo R-squared 0.012444 Akaike info criterion 3.124460
Schwarz criterion 3.320991 Log likelihood -303.5704
Hannan-Quinn criter. 3.203977 Restr. log likelihood -307.3955
LR statistic 7.650198 Avg. log likelihood -1.502824
Prob (LR statistic) 0.176597
REVIEW OF ECONOMIC PERSPECTIVES
54
Table 7
Sample: 1 146
Included observations: 146 - male
Number of ordered indicator values: 8
Convergence achieved after 5 iterations
Coefficient Std. Error z-Statistic Prob.
AGE_2 -0.040494 0.273289 -0.148172 0.8822
AGE_3 -0.130870 0.350611 -0.373262 0.7090
AGE_4 -0.714121 0.352733 -2.024537 0.0429
EDU_3_4 0.608964 0.252131 2.415269 0.0157
EDU_5 1.854249 0.373598 4.963227 0.0000
STATUS_2 -0.034835 0.248388 -0.140243 0.8885
ISCO_4_5_6 0.931966 0.383374 2.430958 0.0151
ISCO_7_8_9 0.369204 0.312338 1.182068 0.2372
OWNER_2 0.589960 0.284267 2.075374 0.0380
FIRM_SIZE_2 0.538385 0.356204 1.511453 0.1307
FIRM_SIZE_3 0.182242 0.323809 0.562807 0.5736
FIRM_SIZE_4 0.886421 0.363459 2.438849 0.0147
FIRM_SIZE_5 1.022033 0.367040 2.784526 0.0054
FIRM_SIZE_6 0.555240 0.361222 1.537114 0.1243
CHILDREN_2 0.460653 0.346107 1.330956 0.1832
CHILDREN_3 0.548022 0.347806 1.575655 0.1151
CHILDREN_4_5_6 1.519670 0.452838 3.355883 0.0008
Limit Points
LIMIT_2:C(18) 0.127222 0.580897 0.219010 0.8266
LIMIT_3:C(19) 1.643437 0.582259 2.822519 0.0048
LIMIT_4:C(20) 2.792842 0.598179 4.668909 0.0000
LIMIT_5:C(21) 3.736509 0.621885 6.008357 0.0000
LIMIT_6:C(22) 4.217387 0.642792 6.561041 0.0000
LIMIT_7:C(23) 4.952239 0.721809 6.860876 0.0000
LIMIT_9:C(24) 5.329673 0.790798 6.739617 0.0000
Pseudo R-squared 0.128990 Akaike info criterion 3.013529
Schwarz criterion 3.503984 Log likelihood -195.9876
Hannan-Quinn criter. 3.212812 Restr. log likelihood -225.0118
LR statistic 58.04852 Avg. log likelihood -1.342381
Prob (LR statistic) 0.000002
Volume XI, Issue I, 2011
55
Table 8
Sample: 1 154
Included observations: 154 - female
Number of ordered indicator values: 6
Convergence achieved after 5 iterations
Coefficient Std. Error z-Statistic Prob.
AGE_2 0.930515 0.342273 2.718633 0.0066
AGE_3 0.416318 0.354528 1.174288 0.2403
AGE_4 0.318844 0.351262 0.907710 0.3640
EDU_3_4 0.114108 0.193134 0.590822 0.5546
STATUS_2 -0.015613 0.205039 -0.076149 0.9393
ISCO_4_5_6 -1.010093 0.226047 -4.468511 0.0000
ISCO_7_8_9 -2.266329 0.346404 -6.542446 0.0000
OWNER_2 0.441097 0.219085 2.013362 0.0441
FIRM_SIZE_2 0.718095 0.352526 2.036998 0.0417
FIRM_SIZE_3 0.236085 0.320149 0.737421 0.4609
FIRM_SIZE_4 1.325448 0.355698 3.726329 0.0002
FIRM_SIZE_5 1.086004 0.395652 2.744848 0.0061
FIRM_SIZE_6 0.416711 0.408374 1.020415 0.3075
CHILDREN_2 -0.456011 0.314831 -1.448432 0.1475
CHILDREN_3 -0.214396 0.301759 -0.710487 0.4774
CHILDREN_4_5_6 -0.252446 0.409450 -0.616550 0.5375
Limit Points
LIMIT_2:C(17) -1.088943 0.402408 -2.706069 0.0068
LIMIT_3:C(18) 0.547916 0.397043 1.379991 0.1676
LIMIT_4:C(19) 1.779681 0.421870 4.218552 0.0000
LIMIT_5:C(20) 2.620059 0.471338 5.558769 0.0000
LIMIT_9:C(21) 3.231512 0.548540 5.891118 0.0000
Pseudo R-squared 0.168019 Akaike info criterion 2.493625
Schwarz criterion 2.907755 Log likelihood -171.0091
Hannan-Quinn criter. 2.661844 Restr. log likelihood -205.5444
LR statistic 69.07059 Avg. log likelihood -1.110449
Prob (LR statistic) 0.000000
REVIEW OF ECONOMIC PERSPECTIVES
56
WAGE DETERMINATION WITH SPECIAL REFERENCE TO ROLE IN
A FAMILY
Lenka Filipová, Zuzana Machová
VŠB – Technical University of Ostrava, Faculty of Economics, Sokolská třída 33, 701
21 Ostrava 1, lenka.filipova@vsb.cz, zuzana.machova@vsb.cz
Abstract: The paper has arisen as a part of the project dealing with the questions of
measuring of wage determinants and wage discrimination on the basis of different
subjective requirements of women and men on their wages. Using the ordered-response
model (Ordered Probit), it analysed wage determination on the basis of Mincer’s Wage
Regression including dummies for role in a family. The analysis was carried out with
the total sample of respondents and subsequently separately for men and women. The
data were gained from a questionnaire survey carried out in Ostrava city, Czech
Republic. In general, the analysis did not prove previous conclusions of theories or
empirical studies about prevailing influence of family characteristics on the wage rate.
Nevertheless, it proved statistically significant differences in the wage determination
between men and women, and showed that in the case of men, some family
characteristics may have been important for determining the wages.
Key words: Human Capital, Role in a Family, Wage Determinants, Wage
Discrimination.
JEL Classification: J24, J31, J71
DOI: 10.2478/v10135-011-0006-y