1

Time for Men to Catch up on Women?

A Study of the Swedish Gender Wage Gap 1973-2012.

sa Lfstrm*
2014-06-11

*Department of Economics, Ume University, Ume, Sweden
Mail: asa.lofstrom@econ.umu.se

Abstract
The Swedish gender wage gap decreased substantially from the 1960s until the beginning of
the 1980s. At the same time women had been narrowing men in employment experience and
education. While women continued to catch up on men the average wage gap remained
almost the same as in the 1980s. The catch-up hypothesis was obviously not the sole answer
to the wage-gap. The purpose here was to discuss other factors of relevance for the evolution
of the average pay gap. Data for the period 1972-2012 is used in the analysis: The results are
mixed and firm conclusions are scarce. Some indications though, the older the women are at
first birth the smaller the pay gap and the same for female union membership while
unemployment, economic growth, fertility and time made the gap larger. It seems as time,
often reliable on issues such as changes in attitudes and prejudices, cannot settle this. One
finding, common in other studies as well, is the influence children may have on the wage
gap. If postponement of motherhood and/or fewer children is necessary to reduce the gender
wage gap the question whether this is desirable or not must be addressed more seriously. If
the answer is no it may be high time for men to catch up on women - through sharing the
full responsibility for children and household duties.
Key words: Gender wage-gap, education, employment, fertility, parental leave
JEL code: J30,J31,J38,J68

Financial support has been received from FORTE (previous FAS)
2


1 Introduction
A widespread view has long been that the gender wage gap is a consequence of women
lagging behind men. A reasonable opinion as long as the differences between men and women
were huge in labor market related issues such as experience and education. The catch-up
hypothesis stated that as soon as women had caught up with men the gap would more or less
disappear. The substantial gender-convergence on this matter has not been met in a way that
was expected. Today you may even say that the situation has become the reversed - women
are in some cases out-performing men but still the wage-gap remains. This makes it natural to
question catching-up as a relevant explanation, or even an issue, for the gender wage gap
today. It may be necessary to choose another approach in order to understand the gender wage
gap in the Swedish labor market.
One approach may be to look for other factors, other than those referring to women explicitly.
Factors that preserve the gap directly or indirectly through counteracting the full catch-up
effect of women. Such an approach should neither exclude the cost of children nor the
reward-system in the workplace. In such a context the idea of catching up may also play a
role, but now out of a male perspective. If the gender pay gap will ever be closed it is
necessary to address the fact that men are lagging behind women in almost all kind of family
related issues. Men catching up on women, in child care and other work related to the
household, may increase the probability for a smaller gender gap as they must be looked upon
as key factors today.
There is a large literature on the gender wage gap highlighting various reasons to the observed
wage differences and their power to explain the gap. Most studies focus differences in male
and female labor and their personal characteristics. In this paper I will study the gender gap
out of a long term perspective. The assumption is though that for the evolution of the gender
wage gap a blend of structural and political indicators may be as important as differences in
individuals capacity and qualifications. The aim is to investigate the gender wage gap in
Sweden for the period 1973-2012, identify changes of special relevance for the issue and
uncover their conceivable impact on the wage gap. The questions are whether the catching-up
hypothesis has been of any relevance in this context or not, what influence other factors may
have had on the gap and lastly if there are anything supporting the idea that it is high time for
men to catch up on women? The paper proceeds with a short historical background in section
3

2. Section 3 contains a descriptive analysis out of a catch-up perspective. In section 4 the
empirical model is presented, in section 5 the results and in 6 the conclusions.
2 History and previous studies a short background
In 1971 the Swedish parliament decided to replace the system of joint taxation with individual
taxation of spouses.
1
This decision was of great significance for women in general and
married women in particular. Before 1971 a womans income from work on the labor market
had been added to her husbands, which made her contribution heavily taxed. Womens work
was, in that respect, considered less profitable for the household. On these grounds the
economic incitements for most wives to participate on the labor market were weak.
2
An
immediate effect of an individualized assessment was therefore an increase in female
participation rate.
3

The reform itself was not only profitable for married women who wanted to work outside the
home, it was also an important step to recognize women, married or not, as part of the labor
force. This was crucial in many respects, not least since it challenged prejudices and
traditional views on women and their roles on different arenas. At that time, the early 1970s,
the conditions on the Swedish labor market were far from gender equal. Although work for, or
at least discussions about gender equality had become more common there were few concrete
results.
Much re-thinking among politicians, employers as well as among the public in general was
necessary before any fundamental change was possible. One important issue was to convince
a political majority that a social infrastructure would be necessary in order to enable women
to combine paid work and family in a reasonable way.
4
Such a majority was also formed and
a social infrastructure accommodated for a modern family was soon appearing. Together with
extensive investments in the educational system and an increasing interest in ameliorating

1
Swedish women could ask for separate taxation from 1965. From 1971 it became mandatory.
2
The decision to abandon joint taxation was, however, not taken without discussions. There were many
objections, not least from married women taking care of the family on a full time basis. As some tax-benefits
for families with one bread winner were going to disappear, the reform was met with reluctance and even
hostility. There were also general (ideological) worries concerning the family as such and its chances to survive
with the new tax system.
3
Selin (2009) shows that female employment would have been 10 percentage points lower if the 1969
statutory income tax system, which to a large part rested on joint taxation, would have been in place 1975.
These estimates point at that the reform effect was substantial and that most of it operated through the
increase in net wages. (p29)
4
A social infrastructure that facilitates for women (and men) to combine work and family must e.g. comprise
child care and a parental insurance that regulates length of leave and pay.
4

womens relative earnings this contributed in replacing the one-bread winner model with
the two-breadwinner model. Although many women entered the labor market on a part-time
basis the two-breadwinner model was surprisingly fast established. The supply of part-time
jobs during the 1970s was substantial and was, primarily due to lack of childcare, an accepted
alternative for many women, at least among those with pre-school children.
5

The wage gap between men and women was also paid attention to for reason of both national
and international concern. The gender gap was not only a Swedish phenomenon but a
worldwide one and the ILO-convention No. 100 from 1951 on Equal remuneration for work
of equal value made a lot to initiate a general discussion around this issue. The research in
this area has been, and still is, extensive in most countries where women are present on the
labor market and the results are strikingly similar: Women were previously and are still on
average paid less than men internationally, irrespective of whether they are working within
the same occupation, branch, sector or not.
6
Although the scope of the gender wage gap was
evident in Sweden, like in many other countries, the politicians were not ready for an
immediate ratification of the ILO-convention. However, the discussions were intense during
the whole of the 1950s and the disputes unmasked a society where it was apparent that men
were the dominant part, not only in the labor force but also in politics, trade unions,
employers confederations and businesses.
The debate came to an end in 1962, when the ILO-convention was approved by the Swedish
parliament.
7
In order to avoid this legislation the social partners had already in 1960 reached
an agreement that abolished separate wage tariffs for men and women and replaced them with
one joint tariff.
8,9
These important occurrences, however, did not reduce the attention on the
gender pay gap, rather the opposite. Thanks to the equal-pay debate numerous of previously
less well known facts had been brought to the surface, which made equal pay and the gender

5
Some would probably say that women had become double-workers. One paid job in the market (full- or
part time) and one unpaid in the household (full-time).
6
Although the gender gap may be of equal size the level of salaries and wages may differ enormously between
countries, which makes global comparisons unreliable.
7
This is one of ILOs eight main conventions. Sweden did ratify it June 6
th
, 1962. In 2012, 168 out of ILOs 183
member states had approved this convention. The debate on equal pay was not new in Sweden. Already in
1947 the teachers reached an agreement on equal pay to male and female teachers.
8
It has been argued that threat on legislation made the social partners agree on this matter. A general
agreement between the social partners on the Swedish labor market in 1938 had made it a rule to negotiate
and reach agreements on a voluntary basis as a way to avoid legislation on labor market matters.
9
One reason to pay women less than men was that they were said to have a handicap, due to their main
tasks as mothers and wives.
5

wage gap an issue to be paid more, not less, attention to.
10
Furthermore, this was the right
time for action and change, since the Swedish economy was running extremely well during
the 1960s. This laid the foundations of many reforms, including the ones supporting gender
equality. The result was also soon visible. Figure 1 below illustrates the evolution of the
gender pay gap during the whole period, 1973-2012, and how fast the gap decreased at the
beginning of this period.
Figure 1: Gender wage gap 1973-2012. (Source: Statistics Sweden)

The reduction was mutual in the four dominant groups in the labor market from the 1960s
until the beginning of the 1980s, although from different levels. The development after that
has been more mixed. In the private sector the average gap between blue collar men and
women has been almost constant since the beginning of the 1980s, around ten percentage
points. Among white collar workers the gap was reduced until the end of the 1990s but after
that this trend stopped and the gap has since then been around 20 percentage points.

10
In 1965 the government decided e.g. to start an investigation on the situation for low income earners. The
full report was published in 1970. One result that paid much attention to was that womens wages and incomes
were extremely low at that time. (Lginkomstutredningen SOU 1970:24/Low Income Survey)
0
0,05
0,1
0,15
0,2
0,25
0,3
0,35
0,4
1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009 2012
(
1
-
F
e
m
a
l
e
/
M
a
l
e
)

White collar
(private)
Civil servants (state)
Blue collar (private)
Municipality
employees
6

In the public sector the reduction of the gap was also impressive until the 1980s. In the
municipalities it continued to shrink during the whole period with exceptions for a few years
in the 1980s. Today its average is approximately five percentage points. In the state sector the
prelude was as the previous groups but from the 1980s the reverse happened, the gap became
larger. This lasted until the beginning of the 1990s when a ten years period of constant gap
followed, around 15 percentage points. After that it decreased again and today the average
gap is around ten percentage points, i.e. the same as 30 years ago.
From this simple description it can be concluded that the gender pay gap was (i) substantial in
all sectors at the beginning of the 1970s, (ii) it was substantially reduced in all groups until the
beginning of the 1980s, (iii) the diminishing trend was permanently interrupted in one sector
and temporarily in two and lastly (iv) for the time being the gender wage gap remains in all
groups although the level varies.
Fifty years after the ILO-ratification and forty years after the introduction of the
individualized tax-reform it is clear that the reduction of the gender pay gap has been
substantial. The significance of these two reforms may, however, be hard to assess since their
impact is more indirect than direct which has, most likely, made other factors as important.
Previous studies
A large volume of literature on the gender wage gap has been published during the last four
decades. The majority of the studies are cross-sectional, while studies on the evolution of the
gap, time-series studies, are less common. The interest in the former is natural, due to the
dominating focus on explaining the wage gap out of a supply perspective, i.e. differentials in
mens and womens individual and personal characteristics and qualifications. Another reason
for this interest may have been the growing access to suitable data for this type of analysis.
The approach where political, economical and structural changes are in focus may be less
common but since they are intertwined with individual changes they might be hard to
separate. Decisions made by men and women on how much they will work on the labor
market, their level of education and if and when they will have children are not only a result
of their own preferences and priorities, but also of factors out of their control. In this paper I
will highlight some of these since I believe it is important in order to better understand the
complexity of the gender pay gap.
7

In a very early study where I investigated the effect of new laws and agreements from the
1960s and onwards on female relative wages in Sweden one finding was that an increase in
female relative wages was not, as expected, followed by a reduced demand for female labor.
11

Another was that technological development, productivity improvements and higher level of
education were important for rising wages, but it was also clear that labor related reforms,
such as laws prohibiting employers to discriminate women and the equal pay act from 1962,
contributed positively to reducing the gender gap. The solidarity wage policy, applied in the
1960s and 1970s, and its purpose on general wage-compression highlighted the low income
earners and as women constituted, and still constitute, a large portion of this group they were
now and then given top priority at the wage negotiations.
12

Many studies on Swedish data have been occupied with the effect of the solidarity wage
policy and its central components in particular: centralized wage setting and general wage-
compression.
13
The high degree of union density among men and women in Sweden across all
educational and occupational groups meant that the system covered most employees on the
labor market. This system was, however, questioned and in 1983, when two of the key figures
on the labor market actually broke up from the central negotiations, it may be claimed that a
new era began.
14
The arguments in favor of such a change varied from a wide spread
dissatisfaction among employers towards the unions strong position on the labor market to
direct recruit problems following too small wage dispersion within specific occupational
groups.
15

In a study where the prime interest was the evolution of the gender earning gap and the role of
wage compression panel data for four different years, 1968, 1974, 1981 and 1991, were
used.
16
This made it possible to say something about the evolution. A declining pay gap
during this period was found but the extent varied from period to period.
17
The tentative
conclusions were that solidarity wage policy may, through reducing residual wage
inequality, have played an important role in the decreasing gender wage gap in the late 1960s
and early 1970s. In the late 1970s, however, wage compression does not seem to be a major

11
Lfstrm (1989)
12
These wage-agreements could, occasionally, include extra to those with low pay (lglnepotter) in general,
and/or extra to women on grounds of low pay (kvinnolnepotter).
13
See e.g. Hibbs & Locking (1996), Calmfors & Drifill (1988), Edin & Richardsson (1999)
14
The union for metal workers and their employer counterpart.
15
This is hardly accidental that this change occurred at this point of time since it was preceded by a period
(1976-1982) of a non-Social Democrats governing in Sweden for the first time in 44 years.
16
Edin & Richardsson (1999)
17
See e.g Edin et al for a description.
8

factor.
18
The small wage-differentials also benefited low-paid women in particular, at least
until the mid-1980s.
Studies on international differences by Blau and Kahn do also show the role different wage-
setting institutions may have. They differ between the countries and so do their implications
for wage structures and hence for wage inequality and the gender wage gap. The US system
with a highly decentralized system of wage setting is also a country with larger gender wage
gap compared to other industrialized countries with a more centralized system and wirh
collective bargaining. Contrary to Sweden the long-lasting wage-gap in the US, the UK and
Australia was mainly interpreted as a result of discrimination. One reason for this was their
focus on minority groups in general, i.e. not women in particular. The studies have therefore
often focused on the effects of anti-discrimination legislation. This was for example the case
in the US in the 1960s and in the UK in the 1970s. The fundamental research on this matter
was performed by Gary Becker already in the 1950s which contributed substantially to
understanding all kind of discrimination on the labor market.
19
His thesis, which preceded the
American anti-discriminations laws, mainly dealt with the differences between white and
black labor but was as suitable for female labor. This made this legislation also beneficial for
them as well.
20

In the UK the Equal Pay Act was introduced in 1970 and this law, together with the Sex
Discrimination Act, was passed in 1975. Results from different studies are mostly the same:
Womens wages did improve after the laws were passed and there were no negative effects on
their employment, rather the opposite.
21

Research performed during the last twenty years on the gender gap is still preoccupied with
supply related factors. Although the sex-differentials in labor market experience and level and
choice of education are very small, or non-existent, today the supply related factors are still
important. If the previous factors were easily accepted as reasonable explanation to the wage
gap, as mirroring actual productivity differences between men and women, the latter may be
harder to identify. However, one line of immense interest has been the effect on labor market
outcome of having a family, primarily children. Several studies have also confirmed that the
consequences for both men and women of having a family on their own are huge but

18
ibid p 26
19
Becker (1971)
20
The Equal Pay Act of 1963 and the Civil Rights Act of 1964.
21
See e.g Chiplin (1979), Chiplin et al (1980) for UK, Cain (1986), Blau & Kahn (1984) for US and Bonell (1987),
Gregory & Duncan (1981) for the Australian case.
9

different. There is, so far as I know, not a single study showing that women ever have been
rewarded in the labor market for giving birth, rather the opposite, while studies confirming the
positive effect of becoming a father are not uncommon.
The penalty women have experienced are due to a vast variety of conceptions and stereotypes
of women transforming into mothers. Women with one or more children are likely to work
fewer hours than men, they are assumed to make less of effort on the job, their human capital
will be out-dated since they spend time at home and they will, per definition, always place
family business before job business.
22
This is called statistical discrimination since the only
ground for this perception is the statistics showing that a female parent, on average, is
actually spending more time with the child than a male parent.
One way to reduce the female penalty is to make it possible for women not only to maintain
their jobs in the labor market but also to continue the career. The supply of childcare facilities
is in that case crucial since lack of places will put a limit for women to participate. The
governmental interventions will be decisive through e.g building child care centers,
facilitating entries and by subsidizing fees for child care.
23
This is fundamental but the actions
taken by the employers are also very important. How do they organize and remunerate jobs?
Is it done in a way that systematically penalizes those who temporary leave the labor market -
than women with children are at risk. In a recent paper by Claudia Goldin she expresses this
very clearly: The gender gap in pay would be considerably reduced and might vanish
altogether if firms did not have an incentive to disproportionately reward individuals who
labored long hours and worked particular hours.
24

Equal pay has been on the agenda in most western countries for a long time now, but
nevertheless no country has so far delivered a zero-result, i.e. no gender related wage gap. It
seems as if equal pay is an elusive, or you may say a constantly moving target. Equal pay
is however not just a question of money; it might as well be a question of social and political
factors. Are the anti-discrimination laws efficient enough, does the social infrastructure fulfill
its commitments towards men and women, and to what extent has the dominant culture at
different working sites and offices been changed to suit the new gender mix in the labor
market? These questions are still looking for unequivocal answers and for obvious reasons are
still lacking.

22
See e.g. Becker (1985)
23
Chevalier & Viitanen (2002)
24
Goldin (2014)
10

In the next section we will take a closer look at what was considered the major cause of the
gender wage gap for many decades: Women lagging behind men in education and labor
market experience.
3 Women catching up on men a descriptive analysis 1973-2012.
As already mentioned a common explanation to the gender wage gap has long been that
women were lagging behind men in essentially all aspects related to the labor market. The
differences between men and women, primarily in length and quality of work experience and
level of education, seemed therefore as a plausible reason to womens lower pay relative men.
The catch-up hypothesis, in its simplest form, stated however that the wage gap would, more
or less automatically, start to disappear as soon as a narrowing between women and men in
these respect were observed. We have already noticed from figure 1 a reduction in pay gap
and it seems reasonable to suspect that this is a consequence of a gender convergence in the
labor market. However, since the gender wage gap is far from closed the question remains
whether the catching up effect is just a temporary phenomenon, has ceased to work or is
counteracted by other factors? There is no simple answer to this but a closer look at the
narrowing process is necessary. We will therefore start the descriptive expos by illustrating
the narrowing between men and women in employment and education during the period under
study.
Figure 2: Female and male labor force participation rate 1973-2012. Sweden. Percent.

It is obvious that women catching-up on men in labor force attachment have been impressive.
The female participation rate rose from around 50 percent at the beginning of the 1960s to 60
50
60
70
80
90
100
Male
Female
11

percent at the beginning of the 1970s and by the end of the 1980s it exceeded 80 percent.
25

The figures for men, on the other hand was constant or slightly dropping during the same
period, mainly as an effect of lowering pension age and longer education. The drop in 1991-
1994 was an immediate result of the deep economic recession those years. The employment-
gap has been around 4-5 percent the last decades and compared to 25 percentage points in the
beginning of the 1970s the narrowing has been substantial.
The education-gap is measured in two ways: The student quota (StudQ), refers to number of
women in relation to men registered in post-secondary education. It is close to one, equal
number, at the beginning of the 1970s but from then on the number of women continually
exceeded the number of men and hence made the gap constantly growing during the whole
period.
Figure 3: Female quotas (women/men) in education. Percent.
StudQ=Registered students in post-secondary education 1973-2012; EduQ=Population with a
post-secondary education 1985-2012

Attending higher education will also be manifest by its positive impact on the level of
education in the population. The second measure is education quota, (EduQ) referring to

25
Among women with children the participation rate was even higher.
0,6
0,8
1
1,2
1,4
1,6
1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009 2012
W
o
m
e
n
/
m
e
n

StudQ EduQ
12

number of women to men with a post-secondary education. Until the mid-1990s there were
more men than women with a longer education in the population but in 1996 women had
caught up on men and since then have outnumbered them. This trend will continue at least as
long as the female student quota remains as high as it is today.
Figures 2 and 3 do, so far, support the catch-up hypothesis: (i) Women were lagging behind
men and (ii) women have made tremendous progress in catching up on them. These are
certainly important explanations to the shrinking wage gap, but it is as obvious that the
catching up effect has not been sufficient to eliminate the gap completely. The next step in
this analysis will therefore be to take a closer look at the wage gap related to female
employment rate and student quota respectively. For simplicity this is done only for white
collar employees in private sector and civil servant in state sector (see figures 4 and 5 below).
Figure 4: Gender wage gap in two sectors related to female employment quota (women/men) 1973-2012.

The higher the female employment quota (EmpQ), the smaller the gender wage gap at least
among the white collar workers within the private sector. In the state sector this was only the
case until beginning of the 1980s, after that the gender gap has been constant or slightly
increasing. Already by the end of the 1980s the number of men and women on the labor
market was close to equal, 0.95, but at that stage the wage gap was no longer shrinking; it was
1973
1974
1975
1976
1977
1978
1979
1980
1981

1982
1983
1984

1985
1986

1987
1992
2003
2005
1983
1984
1985

1986
1988

1991

1997
2012
0
0,05
0,1
0,15
0,2
0,25
0,3
0,35
0,4
0,7 0,75 0,8 0,85 0,9 0,95 1
W
a
g
e

g
a
p

(
1
-
W
f
/
W
m
)

EmploymentQ (women/men)
Civil servants (state)
White collar (private sector)

13

substantial in the two groups but varying. Among white collar workers in private sector the
gap varied between 18 and 26 percentage points and in the state sector between 10 and 15
percentage points. The uneasy up-ward trend in the state sector is though inauspicious,
contrary to the largely downward trend in the private sector. A similar correlation is also
found for female student quota and wage gap (see figure 5).
Figure 5: Gender wage gap in two sectors related to female student quota (women/men) 1973-2012.

The number of women and men registered in higher education was close to equal in the
beginning of the 1970s but this gender balance was only ephemeral since it was gradually
replaced by an ever stronger female dominance. As this dominance became stronger the
gender pay gap decreased at least among white collar workers. Among the state sector
workers this was also the case until middle of the 1980s but after that the up-ward trend made
the development unclear.
A closer look at the actual level of education in the population during a somewhat shorter
period than the previous one, 1985-2012, reveals another pattern for men and women in the
two groups which figure 6 illustrates.

1973
1982

1996
2001
2005

2007
1973
1981

1984
1985
1988
1990
1995
1997
1998
1999
2007
2008

2009
2010
0
0,05
0,1
0,15
0,2
0,25
0,3
0,35
0,4
1 1,1 1,2 1,3 1,4 1,5 1,6 1,7
W
a
g
e

g
a
p

(
1
-
W
f
/
W
m
)

StudentQ (women/men)
White collar (private sector)

Civil servants (state)

14


Figure 6: Gender wage gap in two groups related to female education quota in the population 1985-2012.

The gradual increase of female students was also offset in a changed relation between women
and men with a post secondary education in the population. For this shorter period it was clear
that until 1996 there were fewer women than men with a higher education in the labor market,
but since then the situation is the reversed, women outnumber men with a higher education in
the population. Looking at the pay gap the trend is clear among white collar workers: The gap
is becoming smaller as the female education quota rise. The reduction though is hardly
impressive, on average five (5) percentage points, during the period under study. Among the
state employees the development is more mixed. During the first ten years, 1985-1996, the
gap increased, after that followed a period of fairly stable wage gap and it is not until the
beginning of 2000s the gap is actually decreasing.
The figures above illustrate differences and similarities in the two sectors but they do not say
anything about causality. What are the actual driving forces here? Does a smaller gender pay
gap make it more profitable for women to participate on the labor market and/or enroll in
higher education or is it the other way around? Do more women on the labor market and/or
women with higher education exercise pressure on employers and trade unions to negotiate
1985
1996

2000
2007

2012
1985
1999
2000
2007

2012
0
0,05
0,1
0,15
0,2
0,25
0,3
0,8 0,9 1 1,1 1,2 1,3 1,4 1,5
W
a
g
e

g
a
p

(
1
-
W
f
/
W
m
)

EducationQ (women/men)
White collar (private sector)

Civil servants (state)

15

about more equal pay? There is no simple answer to this but the fundamental truth, as pointed
out in the above descriptions, is that women catching up on men do not always coincide with
a decreasing gender pay gap. If anything, also according to the description above, one may
claim that this is true in the private sector but not in the state sector due to the latters
indecisive evolution of the pay gap.
Since women have improved their relative level of skills, as measured here, the narrowing of
the gap in work-experience and education also makes it reasonable to believe that they have
enhanced their level of unmeasured skills, as well. Increasing labor force attachment may for
example have encouraged women to acquire more on the job training. Womens commitment
to the labor force may also have improved their actual job skills, which may in turn have
undermined the rationale for statistical discrimination.
The crucial thing about the description is, however, that women have not only caught up on
men, they have also passed them. This raises a couple of questions, e.g. has womens
catching-up been a necessary condition to prevent the wage gap to become even bigger than it
is today? Another question concerns the difference between the private and the state sector.
Why is the negative correlation more conspicuous in the private than in the public sector?
The description above has shown that the catch-up hypothesis is of relevance in this context.
The gender wage gap has been decreasing as women, relative men, have become both more
experienced and more educated. But since there still is a gender wage gap and it seems
persistent, it may cast some doubts on its relevance today. Other factors may be more
influential and/or they might be counteracting a presumed positive catch-up effect. One
example may be the shift in Swedish wage policy, which occurred in the mid-1980s. The next
section will identify some additional variables, which may be useful in this context.
4 Model and data
In the previous section the grand gender convergence was confirmed but also that the
evolution has not resulted in a zero gender wage-gap. Is this a reason to dismiss the catch-up
hypothesis? A way to found out more about this is to include other aspects on this issue as
well. One approach is to widen the perspective, go beyond men and women, and look for
factors and structural events which might have had an impact on this issue. The variables I
will refer to here are presented below and for sake of simplicity they are sorted into four
separate boxes.
16


The first box consists of the two catch-up variables already presented above: female
employment quota (empQ) and female student quota (studQ). Their expected impact is
accordingly positive, i.e. as women are approaching men the smaller the gender wage gap will
be. The third variable in the box, time, goes without saying because if nothing else will
change, a common notion is that time will settle the matter. Gradual changes in attitudes
and prejudices towards women in general and working mothers in particular will, even if they
are slow, most likely contribute positively to a smaller gap. One example, where time may
serve as a proxy for a specific change of interest for the gender wage gap is the transformation
of the labor market from very sex-segregated to less. This has been going on for decades now,
it is a slow process indeed, and time may be an appropriate variable.
In the second box three different variables are used referring to the labor market situation
each year. Unemployment rate (unemp), measured in total and for men and women
separately. The expected effect of unemployment in general is that high make the gap larger
and vice versa. A large number of vacancies (vac) on the other hand may indicate a
prosperous economy and a growing labor market but it may also signal imbalances in supply
and demand i.e. matching problems in the labor market. The expectation though is that plenty
of vacancies ought to be beneficial for the supply-side and in that respect be positive for wage
levels. However, their immediate impact on the gender wage gap is less clear. The last
variable in the box is economic growth (growth). The expectation is that the higher the
growth, the smaller the gap and vise versa. There is, however, no such guarantee what so ever.
Economic growth may have a positive effect on both mens and womens earnings but
depending on the size of the wage-increase the actual outcome for the groups separately may
differ leaving the gap unaltered, smaller or larger.
The third box consists of two variables. The first is referring to the system for wage-setting
(cent). The Swedish wage-system has evolved from centralized to a more decentralized. This
transition started in the 1980s and by the end of the 1990s an intermediate system was in
place, where both local and central negotiations were accepted by the social partners. One
type of criticism of the old system was the idea of keeping the wage dispersion low. But, as
already mentioned, this was a way to keep overall wage inequality low and as such its
contribution to a smaller gender wage gap during the 1960s and 1970s has been documented.
The expectation, though, is that a general increase in wage differentials may contribute to a
17

larger gap between men and women at least as long as the labor market is as sex-segregated as
it is. One problem with this variable is, however, that this transition was a gradual process. No
fixed date exists for when this transition was covering the whole labor market. We have
therefore used a summary measure of centralization of wage bargaining, taking into account
both union authority and union concentration at multiple levels.
26

The second variable in the same box is the share of female union membership (UM
fem
). It has
gradually increased, following the evolution of the female labor force participation rate. The
union density may have falling somewhat during the last 10-15 years in Sweden but it is still
high among men and women irrespective of branch, sector and occupation. The expected
effect of high density is positive i.e. a smaller gender wage gap.
The last box consists of two variables closely related to the private life: fertility rate (fert)
and womens age at first child birth (age). The total fertility rate (TFR) in Sweden has varied
quite dramatically during the period under study. From the mid-1960s until the beginning of
the 1980s the fertility rate dropped every year. This negative trend was, however, reversed
during the 1980s when the rate rose rapidly. In the early 1990s the Swedish fertility rate was
among the highest in Europe,
27
but after this peak the rate dropped again very quickly, mainly
as a consequence of the economic recession at the beginning of the 1990s.
28
By the end of the
1990s the fertility rate was as low as it was during the 1930s economic depression.
29
This
negative trend was again reversed in the beginning of the 2000s and in a few years time the
fertility rate was again among the highest in the Europe.
30

The reasons behind these dramatic shifts vary but some stand out more than others e.g.
considerable changes within the social infrastructure. The increase in the supply of publicly
subsidized childcare speaks for itself. In the 1960s there were almost no supply, during the
1970s there was a gradual increase and by the end of the 1980s the supply of places in
childcare centers were substantial, almost covering the demand. Naturally this was a
tremendous change for women with pre-school children in particular, wanting to be part of the
paid labor force.

26
The description for this measure is presented in Visser (2013).
27
TFR in Ireland and in Iceland was somewhat higher at that time.
28
Lfstrm (2003)
29
Total fertility rate in Sweden was at the beginning of the 1990s among the highest in Europe, 2.14, and at the
end of the same decade it was 1.5. This figure was of the same size as the lowest so far, in the middle of the
1930s. (Two (2) is the level for reproduction.)
30
In 2013 total fertility rate (TFR) in Sweden was 1.9.
18

Figure 7 summarize the historical impact this last change exerted in the case of Sweden. The
normal thing, still found in many countries, was that fertility and female employment rate was
negatively correlated: Rising female employment rate followed by falling fertility rate and
vise versa was also the case in Sweden until the end of the 1970s but after that point of time
the correlation turned out to be positive. Female labor force participation rate and fertility rate
was no longer interchangeable. This marks a very important shift and mirror a new era where
a place, a job in the labor market is most likely a necessary, if not always a sufficient
condition, for a young woman before she will decide to become a mother.
Figure 7: Female labor force participation rate (FLFP) and total fertility rate (TFR) in Sweden 1965-2012.

Several studies on the gender wage gap have confirmed the negative impact presence of
children has on womens wages. The question here is if this is visible in a long term macro
perspective as well? If this is the case, the effect of rising fertility rate is expected to be
negative, i.e. making the gap wider while the reverse will happen in case of falling fertility.
However, according to the illustration above the opposite may be as plausible.
The second variable, age of women at first birth, is interesting since women on average are
older today when starting a family than previous generations of women were. This is mainly a
consequence of women spending more time on education and/or on labor market activities
than before. The latter may also have been reinforced by the way the Swedish parental
0,5
0,7
0,9
1,1
1,3
1,5
1,7
1,9
2,1
2,3
2,5
50
55
60
65
70
75
80
85
1965 1970 1975 1980 1985 1990 1995 2000 2005 2010
T
F
R

F
L
F
P

(
%
)

r
FLFP TFR
19

insurance is constructed.
31
It encourages women to participate in the labor market before
having the first child, partly because they will be assured to have a job to return to after the
parental leave, partly because the paid parental leave will be more generous compared to not
have participated. It is hard to tell what impact a postponing of child birth may have on the
gender wage gap, but there are reasons to expect a positive effect i.e.
it will make the gap smaller.
Empirical specification
The gender wage gap (gwg) is the dependent variable, in logarithmic form, and the
independent variables are all continuous. All models are estimated separately for: (i) Blue
collar workers in private sector (ii) white collar workers in private sector, (iii) civil servants in
the state sector and (iv) municipality workers.
32

The four separate models can be specified in the following way.
(1) ln(gwg) =
0
+
1
empQ +
2
studQ +
3
time +
(2) ln(gwg) =
0
+
1
empQ +
2
studQ +
3
time +
4
unemp
fem
+
5
unemp
male
+
6
vac +
7
growth +
(3) ln(gwg) =
0
+
1
empQ +
2
studQ +
3
time +
4
unemp
fem
+
5
unemp
male
+
6
vac +
7
growth +
8
cent +

9
um
fem
+
(4) ln(gwg) =
0
+
1
empQ +
2
studQ +
3
time +
4
unemp
tot
+
5
vac +
6
growth +
7
cent +
8
um
fem
+

9
TFR +
10
ageW +
The models are estimated for the period 1973-2012.
A slightly modified version of model (4) is estimated for a shorter period: 1985-2012. In this
model the two catch-up variables in box I are excluded. Women lagging behind men in
employment are not considered an issue after 1985
33
and student quota will be replaced by
female education quota (eduQ) i.e. female ratio of high educated in the population. A new
variable, productivity (prod), is also included measured here as production per hour. The
expected effect of more women in higher education is positive i.e. it will contribute to a
smaller gender pay gap. The effect of rising productivity ought to be positive but this may be
in general terms, its effect on the gender wage gap may be more problematic. Rising

31
Parental insurance was introduced in 1974 replacing the (small) economic support directed to mothers only
when having a baby. The fathers are now also included in the insurance.
32
In this last group (iv) only those in primary municipalities are included. Employees in the secondary
municipalities (landstingen) are not included i.e. medical doctors, nurses etc.
33
This might be an over-simplification because more women than men are still working on a part-time basis
and can therefore be regarded as lagging behind men.
20

productivity is expected to be transformed into higher wages, but whether this will make the
gender gap smaller or not is an open question.
The empirical specification for the period 1985-2012 is:
(5) ln(gwg)=
0
+
1
eduQ +
2
time +
3
unemp
tot
+
4
vac +
5
prod +
6
growth +
7
cent
+

8
fert +
9
age +
Data are essentially from Statistics Sweden and the SPSS regression analysis package has
been used for the analysis. The results are presented in the section below.
5 Results
The gender wage gap measured as an average gap for the whole labor market is often
criticized. The main argument is that no consideration is taken to mens and womens
different characteristics and qualifications. Without denying these important aspects the crude
measure used here may still be useful, provided that the evolution of the gender gap is of
interest. However, as a way to mitigate the critics I do not use the average for the whole labor
market but the average for the four dominate labor groups. As the wage gap varies a lot
between them, already shown in the figures above, such a separation seems meaningful. The
purpose of this study was threefold: (i) study the impact of women catching-up on men, (ii)
find out the impact other factors may have exerted on the gender wage gap and lastly (iii) get
an indication on what the next step must be in order to further reduce the gender gap.
The parameter estimates in the tables below are only presented by their signs since the main
interest is concentrated on whether the variables have contributed to decrease or increase the
gap. This makes the interpretation simple: A plus (+) indicates an effect which makes the gap
larger, all else equal, while a minus () indicates the opposite. The results are presented in
table 1a-1d below.
34







34
The complete results are found in the appendices.
21

Table 1a: Blue collar workers 1973-2012.
(1) (2) (3) (4)
BOX I empQ -* -* + -*
studQ -* - -* +
time +* + +* +*
BOX II unemp
fem
- -
unemp
male
+ +
unemp
tot
+
vac +* +* -
growth - -* -
BOX III cent -* -*
um (% fem) -* -
BOX IV fert +*
age at 1st birth (wom) -*
R
2
adj .71 .72 .89 .96
DW 0.68 0.88 2.11 1.76
*t-value >1.90

Table 1b: White collar workers in private sector. 1973-2012.
(1) (2) (3) (4)
BOX I empQ -* - +* +
studQ - -* -* +
time -* -* -* +*
BOX II unemp
fem
+ +*
unemp
male
- -
unemp
tot
+*
vac +* +* +*
growth - - +
BOX III cent - +
um(% fem) -* -*
BOX IV fert -
age at 1st birth (wom) -*
R
2
adj .92 .95 .97 .99
DW 0.74 1.25 2.12 1,90
*t-value >1.90






22

Table 1c: State employees 1973-2012.
(1) (2) (3) (4)
BOX I empQ -* -* -* -*
studQ -* -* -* -*
time +* +* +* +*
BOX II unemp
fem
-* -*
unemp
male
+* +*
unemp
tot
+*
vac -* -* -*
growth +* -* +*
BOX III cent + +
um (% fem) + +__
BOX IV fert +
age at 1st birth (women) -__
R
2
adj .24 .75 .91 .86
DW 0.20 0.94 1.79 1.65
*t-value >1.90

Table 1d: Municipality workers
a
1973-2012.
(1) (2) (3) (4)
BOX I empQ + - +* +*
studQ - -* -* -*
time -* - + +
BOX II unemp
fem
-* -
unemp
male
+ +
unemp
tot
+
vac + +* +
growth +* + +
BOX III cent - -
um (% fem) -* -*
BOX IV fert +
age at 1st birth (women) -
R
2
adj .90 .93 .95 .96
DW 0.46 0.81 1.55 1.63
a
Secondary municipalities (landstingen) not included. *t-value >1.90

Table 1a-1d shows mixed results. The results for model (1) and (2) exhibit some similarity
with a visible catch-up effect in all but one group. Rising employment rates and higher female
student quota have, as expected, negative signs indicating a smaller gap. Adding more
variables to the model, (3) and (4), the catching up effect became less pronounced since it is
23

only in the state sector the two estimates are consequently negative and significant. Although
the catch-up effect is dominant in all models an interesting piece is that the effect of student
quota is more often as expected as it is for employment quota. In the former case 14 out of
16 estimates were negative, whereas ten were significant, while in the latter case ten was
negative of which eight were significant. This might give a hint on the role level of education
may have in combating the wage gap.
The economic variables, box II, in model (2) were all significant in the state sector while just
a couple were significant in the other groups. The interpretation is not always straightforward.
More vacancies will apparently contribute to a larger gap in three out of four occupational
groups while economic growth will make it smaller in the private sector and larger in the
public sector. The latter may perhaps be a result of the outcome of wage negotiations where
men have been more successful than women. Some will probable say that women are not as
good as men to negotiate over their own salaries. The effect of unemployment on the gender
gap differs. In all groups, with exception for white collar workers, rising female
unemployment may reduce the gap while male unemployment rate make the opposite. This
seems contradictory but, if the ones losing their jobs are primarily low paid and those keeping
their jobs are better paid this might be a reasonable explanation.
In model (3) a measure on central wage negotiations and female trade union density was
included. The estimates are negative in all groups but one, indicating that a higher degree of
centralization and more women in the unions makes the gap smaller. The exception here is the
state sector since a positive estimate, although insignificant, indicates the opposite. In the last
model (4) total unemployment rate replaced the gender specific rates and it is obvious that
high unemployment in general makes the gap larger, the estimates are positive in all groups
although not always significant.
The two family related variables, fertility rate and age of women at first birth, performed
mixed results. The main impression is though: The higher the fertility rate the larger the gap
and the older the woman is at first birth the smaller the gap all else equal. However, this was
more evident for the latter variable than the former. The negative age-estimates in all groups
is of course interesting and indicates something previously less discussed: The consequences
of for example deliberately postponement of motherhood, of womens rising interest for
higher education and the development of the gender equality in general. I will come back to
this.
24

At last, it is often said that time settle most problems but in this particularly case time
seems not to be a trustworthy partner. Among white collar workers it may, to a certain degree,
be true but not in the other groups. Among employees in the state sector all estimates were
positive and significant and among blue collar workers three out of four estimates were
positive and significant indicating that the gender pay gap did increase not, as expected,
decrease as time went on, all else equal. Among municipality workers the time-effect was
more undecided.
What would this and other variables tell if the period under study was shorter? The reason to
shorten the period is that the reduction in the gender wage gap stopped in all groups in the
beginning of the 1980s (see figure 1). Another reason is that we can include a new variable,
productivity, and replace student quota with education quota in the model. The result of this is
presented below.
Table 2: Gender wage gap 1985-2012.
Blue collar workers (BCW); White collar workers (WCW); State employees (SW); Municipality workers (MW)
BCW WCW SW MW
a

BOX I eduQ - - -* -*
time + + +* +
BOX II unemp
tot
+* + - -
vac + +* -* +*
productivity +* + - +
growth -* - - -
BOX III cent - + +* +
BOX IV fert + - + +
age 1st birth - -* +* +*
R2 adj 0.75 .91 .94 .95
DW 2.61 2.72 1.86 2.50
a
Secondary municipalities (landstingen) not included. *t-value >1.90
The signs of the estimates are in many ways similar to the previous ones and the significant
results are few. The estimates have equal signs in three cases: More women in the population
with higher education and economic growth have contributed to a smaller gap during the
period under study while the estimate for the time-variable displays the opposite effect.
Accepting three estimates with similar signs as reasonable we may extend this and conclude
25

that number of vacancies, productivity and rising fertility may also have contributed to make
the gap larger, at least not smaller, during the period. Overall the regression analysis have
given more + signs than - signs which makes the gender wage gap more problematic and
the solutions not as simple as one could expect.
6 Discussions
The focus in this paper has been the evolution of the gender wage gap in Sweden during the
last forty years and what impact different factors may have exerted on the gap. The reason to
choose this angle was the long-lasting explanation to the gender wage-gap: Women lagging
behind men in labor market experience and level of education. The last forty years has shown
a grand gender convergence in these two respects but the implication for the gender pay gap
has not been as grand. Men and women are still rewarded unequally, and although the
findings in this study are mixed and in many cases inconsistent there are some indications on
what matters for the average gender gap. For sake of simplicity the estimates with equal signs
in at least three groups, significant or not, are summarized in the table below.
Table 3: Summary table 1973-2012 (Tab 1 mod (4)).
Blue collar workers (BCW); White collar workers (WCW); State employees (SW); Municipality workers (MW)
BCW WCW SEW MW
Decreasing gap: AgeW* AgeW* AgeW AgeW
Union Union* Union*

Increasing gap: Time* Time* Time* Time
Unemp Unemp* Unemp* Unemp
Growth Growth* Growth
Fert* Fert Fert

Obviously there is higher degree of consistency among factors increasing than decreasing the
gap. This is of course troublesome and especially that the time variable belongs to this group.
More research about this is needed before any firm conclusion is possible to draw.
35

The results for the two variables with a clear gender dimension, womens age at first birth and
fertility rate, deserve however special attention. I will therefore end by some reflections about
gender roles and gender equality and their impact on the pay.

35
See Goldscheider et al (2013) about attitudes and expectations on sharing of household tasks and child care
among men and women in Sweden.
26

The family, irrespective it is under construction, under enlargement or is going to break up, is
in most studies found to be more of a problem for women than for men when it comes to labor
market related issues such as pay and careers. Motherhood is still a hindrance to young
women's careers with long-lasting consequences. However, in our case it seems as the age of
the woman at first birth may mitigate this negative effect. The older she is at her first birth the
smaller the gender gap and vise versa. The reason for this, out of womens perspective, may
vary but it seems reasonable to assume that older first-time-mothers are more experienced
and more educated. They are probably also better informed and prepared to look after their
rights, towards the employers, than younger ones are when they get pregnant and step into
motherhood for the first time. A third reason may also be that the probability is higher that an
older first-time mother is living in a relationship where gender equality, equal sharing of
both paid and unpaid work, is more prevalent than among younger parents.
From the employers perspective, on the other hand, an older first time mother may also be
seen as less of problem within the firm and a period on leave, due to child care, will therefore
be less costly to the employer. Her competence is worth waiting for and the longer she has
been with the employer the more valuable she might be and the lower interest for the
employer to punish her in case of motherhood. Another positive effect, out of the
employers perspective, is that there are natural limits on number of children and starting late
may, most likely, indicate fewer children.
If the age of the first time mother may influence the wage gap by reducing it, the fertility rate
will do the opposite. Although the level of significance for the estimates is low the message is
clear. This is in line with previous findings, the cost of having children have always been
higher for women than for men. Women may have mitigated the negative effect through
constantly catching up on men, but now is the end of that road reached. Something new is
needed in order to further reduce the gender gap. The natural question is therefore whether it
is about time for men to catch up on women?
Men are still lagging behind women when it comes to time spent on children and other
household duties. This will, most likely, not be sustainable in a long-term perspective. The
convergence between men and women in this respect has started but the differences are still
huge. One reason, often referred to, behind the uneven division is the actual gender wage
differentials there are. As long as womens earnings are lower they will, on average, perform
more unpaid household work and as long as they keep doing it they will continue to receive
27

lower income. One way to break this vicious circle is to encourage and maintain the fathers
emotional commitment to the children direct from the start.
If the economic incentives were obvious for women to be part of the paid labor force, the
incentives for men to take part in the household work is almost non-existent. It is therefore
necessary to broaden the concept of rewards and include non-pecuniary ones as well. The
benefit of having a close relationship to own children from an early stage must be clarified. A
growing number of men have already experienced the backlash of not participating e.g. in
case of divorce. The fight about children has been devastating for many families and for
many fathers this has been a wake-up call. A way to avoid custody fights in the future may be
to take part right from the start - through sharing the emotional and economical responsibility
for children and household. The Swedish parental insurance, introduced in 1974, offered this
opportunity from start. However, equal sharing was hardly an issue 40 years ago. The figure
below gives a snapshot of the process of development so far, and a preliminary forecast for
the coming years.
Figure 8: The convergence in parental leave (percent of days) between parents.

The paid parental leave was in the beginning 180 days, 90 days to the mother and 90 to the
father with the option to transfer days to the other parent. Most fathers transferred their days
to the mothers. Today, 2014, the full paid days are 480 whereas 60 days are not transferable to
0
20
40
60
80
100
Women
Men
28

the other parent. The introduction of the two daddy-months, 1995 and 2002, have certainly
helped to increase the fathers share but still fathers on longer leave is uncommon. Mothers
are still using the lion part, 75 percent. A rising interest among the fathers themselves and
expectations on future fathers from the public in general may help to make fathers periods of
leave a normal thing. A very simple forecast where the future evolution proceeds at the same
speed as previous periods shows that in approximately twenty years the sharing of parental
leave may lie in between 40 to 60 percent, i.e. the definition of gender equality (solid line in
the figure). With a third daddy month this may be reached within 15 years (dotted line).
Whether this seems reasonable or not I leave to reader to judge.
The paper started with a reference to the introduction of the individualized tax-reform in the
beginning of the 1970s, a reform that was essential for womens liberation. From being
primarily unpaid and economically dependent they became an active part in the paid labor
force. It is interesting to remember though, as mentioned above, that this change was preceded
by a long time of discussions between and within groups, pro and against the tax-reform. The
present discussions about the future family policy seem to be in the same stage at the moment
as the arguments, pro and against equal sharing of parental leave, are in many cases the same.
However, the substantial difference is that men, not women, are in focus this time. It is my
belief that men, for many reasons, will catch up on women in the family related businesses
and when they do the economic gender-balance in the labor market will also be possible to
reach. Although I am convinced that this is a necessity I am not as convinced that it will be a
sufficient condition and referring back to Goldin a lot remains to be done within the labor
market and the firms as well.
36

References
Becker, G (1971) The Economics of Discrimination, 2
nd
ed. University of Chigago Press, Chigago
Blau, F.D & Kahn, L.M (1984) On Estimating Discrimination in the Economy: A comment.
Southern Economic Journal, 51.
Bonell, S.M (1987) The Effect of Equal Pay for Females on the Composition of Employment in
Australia. The Economic Record, vol 63, no 183, p 340-351.
Cain, G.G (1986) The Economic Analysis of Labor market Discrimination: A Survey. In Ashenfelter,
O & Layard, R (eds) Handbook of labor Economics,1, Elsevier Science Publishers BV.
Calmfors, L. & Drifill, J. (1988) Centralization of wage bargaining and macroeconomic
performance, Economic Policy 6

36
Goldin (2014)
29

Chevalier, A & Viitanen, T.K (2002) The Causality between Female Labour Force Participation and
the Availability of Childcare. Applied Economic Letters, 9, p 915-918.

Chiplin, B (1979) An Evaluation of Sex Discrimination: Some problems and a Suggested
Reorientation. In Lloyd, C (ed) Women in the Labor market. New York Colombia
University Press.
Chiplin, B, Curran, M.M. & Parsley, CJ (1980) Relative Female Earnings in Great Britain and the
Impact of Legislation. In Sloane, PJ: Women and Low Pay. Macmillan, London.
Edin, PA & Richardson, K (1999) Swimming with the Tide: Solidarity Wage Policy and Gender
earnings Gap. IFAU 1999:3.
Goldin, C (2014) A Grand Gender Convergence: Its Last Chapter. American Economic Review,
104(4): 1-30.
Goldscheider, F, Bernhardt, E & Brandn (2013) Domestic gender equality and childbearing in
Sweden. Demographic Research, Vol 29, art 40, p1097-1126.
Gregory, R.G & Duncan, R.C (1981) Segmented Labor market Theories and the Australian
Experience of Equal pay for Women. Journ of post-Keynesian Economics,3 p 403-428.
Hibbs, D.A & Locking, H (1996) Wage Compression, Wage Drift and Wage Inflation in Sweden.
Labour Economics (3) p 109-141.
Lfstrm, (1989) Discrimination in the Swedish Labour Market. An Analysis of the Wage
Differentials between Men and Woman. Ume Economic Studies 1989:196 (In Swedish)
Lfstrm, (2003) Fertility and Business Cycle during the late 20th Century in Sweden. Ume
Economic Studies 2003:603 (In Swedish)
Selin, H (2009) The Rise in Female Employment and the Role of Tax Incentives. An Empirical
Analysis of the Swedish Individual tax Reform of 1971. CESifo Working paper No 269,
april 2009.
Visser, J. (2013) Data Base on Institutional Characteristics of Trade Unions, Wage Setting, State
Intervention and Social Pacts, 1960-2011 (ICTWSS). Amsterdam Institute for
Advanced Labor Studies, University of Amsterdam, April 2013 (version 4).









30

Appendices
Table A1a Blue collar workers 1973-2012 (standard errors within brackets)
(1) (2) (3) (4)
empQ -1.77 -1.85 1.59 -2.25
(.312) (.426) (1.68) (1.053)
studQ -.772 -.487 -.822 .022
(.256) (.357) (.283) (.216)
time .010 .006 .014 .036
(.004) (.006) (.005) (.010)
unempFem -.037 -.029
(.033) (.026)
unempMale .043 .045
(.032) (.025)
unempTot .013
(.008)
vac .000 .000 -.000
(.000) (.000) (.000)
growth -.011 -.019 -.006
(.007) (.007) .005)
cent -3.18 -1.31
(1.03) (.652)
unionFem -.076 -.010
(.028) (.018)
fert .196
(.074)
age at 1
st
birth (wom) -.252
(.076)
Constant -19.45 -12.24 -25.59 -63.74
(6.8) (11.49) (9.0) (18.42)
R
2
adj .71 .72 .89 .96
DW .68 .88 2.11 1.76


















31

Table A1b White collar workers 1973-2012 (standard errors within brackets)
(1) (2) (3) (4)
empQ -.787 -.290 1.88 .928
(.178) (.196) (.731) (.632)
studQ -.199 -.333 -.294 .206
(.146) (.164) (.123) (.129)
time -.008 -.008 -.006 .015
(.002) (.003) (.002) (.006)
unempFem .022 .025
(.015) (.011)
unempMale -.013 -.008
(.015) (.011)
unempTot .015
(.005)
vac .000 .000 -.000
(.000) (.000) (.000)
growth -.001 -.001 .000
(.003) (.003) (.003)
cent -.488 .180
(.448) (.391)
unionFem -.043 -.030
(.012) (.011)
fert -.059
(.044)
age at 1
st
birth (wom) -.199
(.046)
Constant 14.80 14.44 11.65 -26.54
(3.88) (5.27) (3.91) (11.05)
R
2
adj .92 .95 .97 .98
DW .74 1.25 2.12 1.90




















32

Table A1c State employees 1973-2012 (standard errors within brackets)
(1) (2) (3) (4)
empQ -1.949 -4.919 -4.634 -6.133
(.734) (.582) (2.23) (3.02)
studQ -1.278 -.973 -1.74 -1.637
(.603) (.487) (.375) (.620)
time .026 .038 .055 .059
(.008) (.008) (.006) (.029)
unempFem -.219 -.137
(.045) (.034)
unempMale .142 .074
(.044) (.033)
unempTot -.042
(.022)
vac -.000 -.000 -.000
(.000) (.000) (.000)
growth .030 .015 .029
(.010) (.009) (.015)
cent 1.237 2.792
(1.367) (1.872)
unionFem) .000 .038
(.037) (.051)
fert .231
(.211)
age at 1
st
birth (wom) -.060
(.219)
Constant -50.78 -71.06 -105.54 -113.505
(16.02) (15.69) (11.92) (52.88)
R
2
adj .24 .75 .91 .86
DW .20 .94 1.79 1.65




















33



Table A1d Municipality workers 1973-2012 (standard errors within brackets)
Secondary municipalities (landstingen) not included.
(1) (2) (3) (4)
empQ .415 -.314 4.904 3.713
(.428) (.521) (1.80) (1.784)
studQ -.634 -.880 -1.682 -1.101
(.352) (.436) (.302) (.366)
time -.023 .013 .008 .026
(.005) (.007) (.005) (.017)
unempFem -.102 -.028
(.041) (.027)
unempMale .070 .032
(.040) (.027)
unempTot .014
(.013)
vac .000 .000 -.000
(.000) (.000) (.000)
growth .024 .007 .009
(.009) (.007) (.009)
cent -1.283 -.047
(1.103) (1.105)
unionFem -.099 -.077
(.030) (.030)
fert .060
(.125)
age at 1
st
birth (wom) -.197
(.129)
Constant 43.90 25.38 -14.64 -47.70
(9.35) (14.05) (9.61) (31.22)
R
2
adj .90 .93 .95 .96
DW .46 .81 1.55 1.63

















34


Table A2 Gender pay gap 1985-2012.
BCW=blue collar workers, WCW=white collar workers, SW=state workers,
MW=municipality workers. (Standard errors within brackets)

BCW WCW SW MW
eduQ -4.53 -.096 -12.48 -14.36
(2.63) (1.75) (2.92) (3.58)
time .017 .007 .083 .043
(.018) (.012) (.020) (.024)
unempTot 0.017 .010 -.012 -.001
(.008) (.005) (.009) (.011)
vac .0001 .0002 -.0003 .0005
(.000) (.000) (.000) (.000)
productivity .007 .002 -.002 .001
(.002) (.002) (.003) (.003)
growth -.025 -.004 -.003 -.004
(.009) (.006) (.010) (.012)
cent -.170 2.57 10.13 1.63
(2.41) (1.61) (2.67) (3.27)
fert .051 -.086 .196 .154
(.139) (.093) (.155) (.189)
age at 1
st
birth (wom) -.137 -.217 .531 .489
(.103) (.069) (.115) (.141)
Constant -29.16 -10.92 -172.96 -88.19
(32.92) (21.97) (36.61) (44.86)
R
2
adj .75 .91 .94 .95
DW 2.61 2.72 1.87 2.50


Table A3: Summary table for the two periods and for private and public sectors separately.

1972-2012 1985-2012.
Private Public Private Public
time + + + +
unempTot + + + -
age at 1
st
birthWom - - - +
studQ + - .. ..
eduQ .. .. - -

growth ? + - -
fert ? + ? +
union - ? .. ..
vac ? ? + ?
cent ? ? ? +
productivity .. .. + ?
empQ ? ? .. ..
Gender wage gap: minus=smaller; plus=larger; ? = (+) and () within the same sector



35

Wage compression, wage drift and wage