Research Report 109 The Gender Pay Gap PDF

Equality and Human Rights Commission
Research report 109 | Pay gaps research
The gender
pay gap
Malcolm Brynin
Institute for Social and Economic Research, University of
Essex
© 2017 Equality and Human Rights Commission
Published August 2017
ISBN 978-1-84206-686-7
Equality and Human Rights Commission Research Report Series

The Equality and Human Rights Commission Research Report Series publishes
research carried out for the Commission by commissioned researchers. The views
expressed in this report are those of the authors and do not necessarily represent
the views of the Commission. The Commission is publishing the report as a
contribution to discussion and debate.
Please contact the Research Team for further information about other Commission
research reports, or visit our website.
Post: Research Team

Equality and Human Rights Commission
Arndale House
The Arndale Centre
Manchester M4 3AQ
Email: research@equalityhumanrights.com
Telephone: 0161 829 8100
Website: www.equalityhumanrights.com
Access
You can download a copy of this report in Microsoft Word and PDF formats from
www.equalityhumanrights.com. If you require this publication in an alternative format,
please contact the Communications Team to discuss your needs at:
correspondence@equalityhumanrights.com.
Equality and Human Rights Commission – www.equalityhumanrights.com

Published: August 2017 1
The gender pay gap List of tables and figures
Contents
List of tables and figures 3

List of abbreviations and acronyms 5
Acknowledgements 6
Executive summary 7
Introduction 7
Key findings 7
Background 9
Methodology 10
1 |Introduction 11
2 |Methodology 13
2.1 Data 13
2.2 Definitions 13
2.3 Analysis 14
3 |The gender pay gap: what we already know 16

3.1 Theories of the causes of the gender pay gap 16
3.2 Measuring and explaining the gender pay gap 20
4 |Research findings: features of the gender pay gap 24

4.1 Patterns and trends in the gender pay gap 24
4.2 How the pay gap varies by characteristics 27
5 |Research findings: the drivers of the gender pay gap 44

5.1 Graduate status and occupational segregation 44
5.2 Explaining the gender pay gap: which factors carry more weight? 45
5.3 The problem of low pay 51
6 |Conclusions 54
Appendix 61
Bibliography 56

List of tables and figures
Tables
Table 4.1: Female mean and median hourly pay as a percentage of male hourly pay
Table 4.2: The gender pay gap for four age groups in 1994, 2004 and 2014
Table 4.3: Pay gaps (£ per hour) by part-time and full-time work over time
Table 4.4: The gender pay gap comparing married to unmarried people (£ per hour)
Table 4.5: The gender pay gap by marital status and age (£ per hour)
Table 4.6: The gender pay gap by presence of dependent children in household
Table 4.7: Housework and paid work hours (UKHLS; 45,533 observations)
Table 4.8: The pay gap by educational level
Table 4.9: Pay gap by public/private sector and graduate status (£ per hour)
Table 4.10: Pay gaps by region
Table 4.11: Pay gaps by region and over time
Table 5.1: The gender pay gap by graduate status and level of feminisation in
occupations
Table 5.2: Regression (OLS) results for men and women
Table 5.3: Gender pay gaps over time: unadjusted, adjusted and explained part
Table 5.4: Decomposition analysis of the explained part of the gender pay gap
Table A1: OLS regression results for men and women (full table)
Table A2: Decomposition analysis of the explained part of the gender pay gap
Figures
Figure 4.1: The distribution of male and female pay (£ per hour)
Figure 4.2: The general and occupational gender median pay gaps over time (£ per
hour)
Figure 4.3: The general gender median pay gap over time and by age (£ per hour)

Figure 4.4: The occupational gender pay gap over time and by age (£ per hour)
Figure 5.1: Decomposition analysis of the gender pay gap
Figure 5.2: Log odds of being in the bottom 10% of the wage distribution (logistic
regression)

The gender pay gap List of abbreviations and acronyms
List of abbreviations and acronyms
ASHE Annual Survey of Household Earnings

BHPS British Household Panel Survey
HMRC Her Majesty’s Revenue and Customs
LFS Labour Force Survey
OLS Ordinary least squares
ONS Office for National Statistics
PAYE Pay As You Earn
UKHLS UK Household Longitudinal Survey

The gender pay gap Acknowledgements
Acknowledgements
The authors would like to thank the following staff at the Equality and Human Rights
Commission: David Coulter, Anne Madden, Karen Hurrell and Gregory Crouch.
We are also grateful to the Office for National Statistics for providing access to the
Labour Force Survey via the UK Data Service. However, none of the original data
creators, depositors or copyright holders, the funders of the data collections, nor the
UK Data Service bear any responsibility for the analysis or the interpretation of the
results. Crown copyright material is reproduced with the permission of the Controller
of HMSO and the Queen's Printer for Scotland.

The gender pay gap Executive summary
Executive summary
Introduction
This research report explores the gender pay gap, which is defined as the difference
between the average hourly pay of men and women. As well as looking at basic
differences in pay, the research identifies the characteristics that explain those
differences such as age, occupation and level of education. The report is intended to
further debate and highlight areas where intervention may be needed.
Key findings
 According to the Annual Survey of Hours and Earnings, the median gender pay
gap for full and part-time workers in 2016 was 18.1%. The gap for full-time
employees only was smaller at 9.4%. While part-time women tended to earn
slightly more than part-time men (6%), part-time women earned 36.5% less than
full-time men. Women are much more likely to work part time than men.
 Based on the analysis of Labour Force Survey (LFS) data – the source for all of
the analysis in this report - the mean gender pay gap has reduced considerably.
As a percentage of male earnings, for full and part-time work, it fell from 27% in
1993 to 10% in 2014.
 The gender pay gap is a longstanding phenomenon and its causes are complex.
Social pressures and norms influence gender roles and often shape the types of
occupations and career paths which men and women follow, and therefore their
level of pay. Women are also more likely than men to work part-time and to take
time out from their careers for family reasons.
 The effect of ‘occupational segregation’ – the division of men and women into
different occupations – on pay has lessened. However, within occupations, on
average women are still paid less than men suggesting they are either being paid
less for doing broadly the same work or they have lower level jobs in the same
occupations. In 2014, the gender pay gap within occupations (the ‘occupational
gender pay gap’) was 15.3% based on the median. This has also declined, from
Equality and Human Rights Commission · www.equalityhumanrights.com

20.7% in 1993, however less so than the general gap. This disparity of pay within
occupations now explains a very large part of the gender pay gap overall.
 Women not only earn less than men overall, they are more likely to be low paid.
In 2014, 20.4% of men earned less than £8 per hour while 30.3% of women did
so. However, the proportion of women experiencing low pay has declined over
time.
 Nearly two-fifths of women in employment are part-time and four times as many
women as men work part-time. Male and female part-time workers generally earn
less per hour than full-time workers, but women who work part-time generally
earn more than men who do so.
 The pay gap widens with age: older women experience a larger pay gap
compared with their male peers than younger women with their male peers. This
is primarily because women are more likely than men to take time out of the
labour market to care for children. This may slow career development. The
statistical analysis found that women's shorter job tenure, a likely consequence of
starting a family, is a factor driving the pay gap.
 While younger married women earn more than unmarried women, this advantage
reverses with age. From their 40s onwards, married women experience a pay
disadvantage compared to unmarried women. This is likely to be linked with child-
rearing: the analysis found that having a child increases the pay gap considerably
for women. Married men, by contrast, earn substantially more than unmarried
men in all age groups. The ‘wage penalty’ for child-rearing, as a proportion of
women’s pay, has increased slightly over time. However, as with the gender pay
gap generally, the pay gap between men and women with children has also
declined over time.
 There appears to be a relationship between housework and the pay gap. Across
the whole sample, women do more housework than men, and the demands of
housework do not affect women and men in the same way. Where women work
fewer hours they do more housework but men do not vary their housework hours
relative to hours worked – their contribution tends to remain low regardless.
Women that do the largest amounts of housework experience a pay gap even
when compared with the small number of men who also do a lot of housework.
 Care responsibilities affect men as well as women in terms of pay. Men and
women with the most time-consuming care responsibilities tend to have similar
salaries. However, the majority of men and women do not have care
responsibilities.
 The gender pay gap varies according to where people live and the sector they
work in. The difference between central London and the rest of the country is

vast. London has a smaller gender pay gap compared to the UK as a whole. The
North of England, particularly North and East Yorkshire and Humberside,
followed by the South West, have the largest pay gaps. Scotland and Wales have
gender pay gaps that are similar to the UK average, although the gap is slightly
smaller in Wales than in Scotland.
 The gender pay gap is larger in the private sector, at £3.11 per hour over the
period 1993-2014, than in the public sector where the gap is £2.38 per hour
(adjusted for inflation). The pay gap between male and female graduates in the
public sector is comparatively small (£1.63 per hour) but higher in the private
sector (£2.77 per hour), suggesting that female graduates in the private sector
gain less from their education in terms of parity with men.
 In the period 1993–2014 the gender pay gap among graduates declined from
21% to 6%, whereas for women without A-levels the gap declined from 34% to
17%. Twenty-eight per cent of male employees are graduates compared with
33% of female employees. Regression analysis found that female graduates
have a higher pay return on their degree than male graduates – that is, controlling
for other factors such as age and job tenure – and also benefit more from working
in a graduate occupation.
 If a number of variables in the data, such as level of education for example, are
controlled for to remove their influence over the pay gap, the remaining ‘residual’
and unexplained pay gap is much smaller. However, the proportion of the gender
pay gap which remains unexplained has risen from just over half in the period
1993–1997 to over two-thirds between 2010-2014.
 Analysis of LFS data indicates that part-time work, occupational segregation and
the ‘glass ceiling’ are the main drivers of the gender pay gap. Other significant
factors include the over-representation of women in low-paid jobs, having children
and shorter job tenure (which is linked to child-rearing). As a result women face
barriers in their forties and in older age groups when it comes to progressing their
careers.
Background
In 2015 the Equality and Human Rights Commission (‘the Commission’)

commissioned research into the causes of, and potential solutions to, the gender,
ethnicity and disability pay gaps. A suite of reports has been produced including a
summary of findings from the entire project. These are available on the
Commission’s website: www.equalityhumanrights.com.

The Commission holds the view that pay gaps reflect broader inequalities in society
and tackling them is an important way to make Britain fairer. The analysis in this
report builds on our previous research on pay gaps and complements our extensive
online guidance on equal pay.
Methodology
Firstly, a brief literature review was conducted to establish what is already known
about the gender pay gap. The statistical analysis, which is based primarily on
historical data from the UK LFS covering the period 1993–2014, followed on from
this. The LFS is a quarterly household survey of the employment circumstances of
the UK population and is administered by the Office for National Statistics (ONS). A
small amount of additional analysis is based on the UK Household Longitudinal
Study (UKHLS, also known as Understanding Society).
The LFS data were used not only to establish the relative pay of different groups but
also to identify the personal characteristics that are associated with differences in
pay, such as level of education. These variables are referred to in the report as
'drivers' of pay gaps, although it is only possible to say that these are associated
with, rather than the causes of, pay gaps. This analysis reveals, based on the data
available, how much of the pay gap can be explained and how much is left
unexplained. For a full account of the methodology see Chapter two.

The gender pay gap Introduction
1 |Introduction
In 2015 the Equality and Human Rights Commission (‘the Commission’)

commissioned research into pay gaps. Its purpose was to explore the extent of
disparities of pay between certain groups, to elucidate their causes and to identify
ways to mitigate them. The research focused on pay gaps by gender, ethnicity and
disability. As well as statistical analysis of pay data, the project involved a literature
review of the causes of pay gaps, as well as governmental and employer
interventions that have attempted to address them. Workshop discussions about
practical solutions were also held with key stakeholders such as employers and
government bodies.
The main aims of the project were to:

 establish the size of pay gaps for women, ethnic minorities and disabled people
 review the evidence base and identify the causes of pay gaps
 analyse the relative impact of different variables on gender, ethnicity and
disability pay gaps, and
 review and discuss with stakeholders the effectiveness of certain interventions by
government, employers and other organisations to reduce pay gaps.
A suite of reports has been produced: one each for disability, ethnicity and gender,
covering literature reviews of the evidence base and statistical analysis, a report on
interventions covering a literature review and stakeholder consultation, plus a
summary of findings from the entire project. These are available on the Commission
website: www.equalityhumanrights.com.
This report focuses specifically on the gender pay gap, which is the difference
between the average hourly pay of men and women. It begins with a brief summary
of the evidence from the existing body of literature. This is followed by a statistical
analysis of data from the Labour Force Survey (LFS) which looks at the difference in
pay and how it has changed over time. It also explores the association with pay
levels of certain variables such as education, hours worked, occupation, age and so
on.

The gender pay gap Introduction
The Commission holds the view that pay gaps reflect broader inequalities in society
and tackling them is an important way to make Britain fairer. The analysis in this
report builds on previous Commission research on pay gaps and complements our
extensive online guidance on equal pay. It is intended to inform debate and highlight
areas where action may be needed.

The gender pay gap Methodology
2 |Methodology
2.1 Data
The analysis in this report is mainly based on the UK Labour Force Survey (LFS)
from 1993 to 2014, a repeated cross-section survey of individuals which supplies
detailed information on employment and has the advantage of a very large sample:
444,865 women and 429,295 men. However, there are some missing data which
means effective sample sizes are generally slightly lower.
A small amount of additional analysis is based on the UK Household Longitudinal

Study (UKHLS, also known as Understanding Society). This uses information on
time spent on household tasks, information which is not available in the LFS. The
data are from the first five waves (though pooled and therefore only treated as cross-
sectional). Wave 1 took place in 2009/10 and Wave 5 in 2014.
Both data sources rely on respondents providing details of their household income
and so provide comparable data. Other data sources provide different figures for the
pay gap usually for methodological reasons, for example, the Annual Survey of
Household Earnings (ASHE) is based on pay data received by HM Revenue and
Customs (HMRC). Thus it is not always possible to compare pay gap findings across
different sources.
2.2 Definitions
All the pay analysis is based on hourly wages and excludes extreme values (less
than £1 and over £80). Hourly wages are defined as normal gross pay divided by
normal hours of work. Those with a marginal commitment to the labour force, for
example students and semi-retired people, are not included in the analysis. This is
defined as working 10 hours a week or fewer. Except where the analysis is explicitly
of full-time or part-time work, the pay gap analysis is based on the entire sample,
that is, all people working at least 10 hours a week.

Self-employed people are not included in the analysis as we are primarily concerned
with the gender pay gap in salaried employment. There is also the problem of the
general unavailability of reliable pay data for the self-employed.
The age range for the data is 17-65. This starting age is one year older than usual in
response to legislative changes which distinguish a ‘participation age’, set at 17,
from the school leaving age of 16. Over 70% of 16 year olds are in full-time
education.
2.3 Analysis
The analysis uses the time series in a number of different ways. For many analyses
the data are pooled so that the results are averages for the entire period. Other
analysis compares the first year in the period (1993) with the last (2014) or splits the
period into groups of years so that very broad trends can be assessed. A finer view
of trends is given in graphical form where the distribution of pay or pay gaps across
all years is shown. In the regression analysis, ‘year’ is included as a control variable
to account for change over time. In effect this is limited because wages are deflated
(with 2014 as the base), so the trend is in real rather than money terms, showing the
rise in wages over and above price inflation.
There is a range of ways that the overall gender pay gap is calculated and
expressed in this report:
(i) The average female wage as a percentage of the average male wage.
This is the standard means of indicating the extent of inequality and
change in inequality, and is how the gender pay gap is normally
expressed.
(ii) The gap between average female and male pay as a percentage of the
average female wage. Decomposition analysis then shows what
proportion of the gender pay gap can be ‘explained’ by the data.
(iii) The raw wage gap, that is, in terms of money itself, expressed as an
hourly rate. This ‘real gap’ is perhaps more meaningful to most people as
we spend pounds and pence, not percentages. It shows the pay gap in
practical, everyday terms.
The averages are calculated in two ways, either as the mean or the median. The
mean is simply the total divided by the number of observations, which is how most of

us understand ‘average’. The median is the wage at the midway point in the wage
distribution such that 50% of people earn more than this and 50% earn less.
The median takes account of the skewed distribution of pay better than the mean. A
small proportion of employees earn very high pay and this skews the mean upwards,
so giving this group undue weight. For this reason the median is often used.
However, the mean is sometimes used as it reflects the fact that men are more likely
to earn very high wages and this should not be discounted. The mean is more widely
understood and, while the results based on the median will also be given here, the
mean will be used in much of the analysis.
It is generally found that the pay gap based on the mean is higher than that based on
the median (Perfect, 2013), simply because the high wages of some men skew the
distribution. However, in the following analysis this is not generally the case because
extreme high and low values have been removed (because these are generally
deemed to be unreliable and also describe a very small percentage of the working
population). Further, the analysis excludes those working less than 10 hours a week,
and these people are likely also to have low hourly earnings.
The report also uses a second measure called the ‘occupational pay gap’. It is
calculated by collecting data on the pay gap within individual occupations. These pay
gaps are then averaged to arrive at a single figure. There are around 80 occupations
in this analysis and the sample size, given at the start of this chapter, is large enough
to produce reliable estimates when averaged across all occupations.
The difference between the occupational pay gap and the ‘general’ pay gap depends
on the distribution of men and women across the entire economy. For example, if
men tended to be concentrated in highly paid occupations and women into poorly
paid ones, but within these men and women received the same pay for doing broadly
the same work, then the occupational pay gap would be zero. Most of the general
pay gap would therefore be the result of occupational segregation rather than
inequality within these. On the other hand, if men and women each accounted for
half of the workforce in every single occupation, but women were predominantly in
lower paid, more junior positions, then both the general and occupational gap would
be large – and the occupational pay gap would account for the entire general pay
gap.

The gender pay gap The gender pay gap: what we already know
3 |The gender pay gap: what we already

know
3.1 Theories of the causes of the gender pay gap
The gender pay gap is a longstanding feature of labour markets across the globe.
While it has narrowed in many countries, overall progress has been slow and is far
from complete. Several competing theories have been put forward to explain the
persistence of the gender pay gap. There is a broad distinction between economic,
sociological and psychological theories. The main ones are briefly outlined below.
3.1.1 Gender role theory

Differential gender roles are adopted early on in life and influence much of what
happens in the home, school, personal relationships, family life and employment
(see, for example: Lips, 2012; Ochsenfeld, 2014; Rubery, 2008). Therefore men and
women often follow different paths in education and employment, which lead to
overall differences in pay. Segregation into traditional gender roles is often not a
conscious 'choice' for either women or men. Rather, these choices are constrained
by social pressures and expectations, and are passed on from one generation to the
next.
3.1.2 Human capital theory

Some claim that the gender pay gap was, historically at least, chiefly the result of
women having lower 'human capital' than men – that is, lower knowledge, skills or
job experience (Becker, 1991; Polachek, 1981; Manning and Swaffield, 2008). This
gave rise to lower productivity and therefore to lower wages for women. Men, on the
other hand, had a comparative advantage when it came to investing time and
resources in their education and careers, reflecting the wider social context.
However, this male advantage in human capital has weakened over time. The
growth in women's education is often cited as a major driver of the decline in the
gender pay gap (Goldin, 2008; Weichselbaumer and Winter-Ebmer, 2005). In Britain,

Grimshaw and Rubery (2007) find that gender differences in human capital explain a
shrinking proportion of the pay gap. Olsen et al. (2010) show that education explains
little of the gap, while its impact is also declining. Brynin and Perales (2016) suggest
that among the highly qualified at least this might even be moving in women’s favour.
Manning and Swaffield (2008) show a substantial human capital impact on gender
inequality but this is the result of differences in on-the-job training.
Despite this progress, inequality persists. Access to education, as measured by

prestige of institution and field of study, remains gendered to the advantage of men
(Booth and Kee, 2011). Just as important, career breaks, primarily as a result of
having children, continue to reduce women’s work experience, and this is in turn
likely to reduce women’s career prospects over the life course (Goldin, 2008).
Although the research is now a little dated, Olsen and Walby (2004) found that 19%
of the gender pay gap could be attributed to differential work histories. On the other
hand, analysis by Dex et al. (2008) suggests that men do not necessarily do better
than women in the early stages of their career. Manning and Swaffield (2008) show
that the gender pay gap rises over women’s careers although the gap itself declines
from earlier to later cohorts.
3.1.3 Occupational segregation

Occupations that have high proportions of women working in them are often referred
to as 'feminised'. It has been found that the higher the proportion of women who
work in an occupation the lower the average pay within it (Blau and Kahn, 2003;
Bettio and Verashchagina, 2009; Levanon et al., 2009). Both men and women within
feminised occupations experience lower pay, although because women obviously
outnumber men they are disproportionately affected.
In the UK, Olsen et al. (2010) attribute 17% of the pay gap to occupational
segregation by gender. Other research on the combined effects of occupational
segregation and segregation within the workplace (that is, men and women
performing different roles in the same workplace) finds a larger gap (Mumford and
Smith, 2009), though in this case segregation is more important for the part-time
than for the full-time gender pay gap.
Segregation, however, is difficult to analyse because of different datasets and

definitions, while its effects are difficult to interpret because segregation is the result
of both supply (men and women tending to gravitate to different types of jobs) and
demand (for example, employers’ prejudices may act as a barrier). Explanations for
occupational segregation often include the gender role argument described above,

but other factors might include family constraints that, for instance, encourage entry
into part-time work, which mostly occurs in more routine occupations.
It is also unclear why average pay is lower in more feminised occupations. It might
be that women are paid less (for whatever reason), and that when they segregate
into specific occupations, average wages in these must by definition be lower.
However, both men and women who work in feminised occupations are paid less on
average than those in non-feminised ones. It is possible of course that family and
social pressures cause women to enter into low-paid occupations as a second-best
option. This would mean more women than men ending up in low-paid occupations
and could help explain why feminised occupations tend not to pay well.
Insofar as segregation is a problem, desegregation would seem to be the answer.

However, research in the United States found that the effect of desegregation on
equality is complex and inconclusive (Stainback and Tomaskovic-Devey, 2012).
Further evidence from the US shows that occupational gender segregation is in
decline (Hegewisch et al., 2010; Olsen et al., 2010, b; American Association of
University Women, 2012), as is the gender pay gap, but the relationship between the
two may also be weakening (Charles and Grusky, 2004). The effects of segregation
are not consistently negative for women. Highly skilled women, for example, are
sometimes very well paid in highly feminised occupations. This would mainly seem
to be the result of women entering highly professionalised public sector work, often
at management level, for instance in social work (Esping-Andersen, 1993; Brynin
and Perales, 2016).
Finally, inequality arises partly through ‘vertical segregation’, that is, men occupying
higher positions within an occupation, and partly through ‘horizontal segregation’ (in
other words occupational segregation) men and women working in different
occupations). Horizontal segregation suggests that women's work is often
undervalued (see 3.1.4 below). Vertical segregation highlights the problem of the
‘glass ceiling’ as a barrier to women reaching senior positions.
Whatever the exact cause of occupational segregation, it is clear that women tend to
be over-represented in the following low-paid jobs (the ‘five Cs’): cleaner, caterer,
carer, cashier, and clerical worker (Joint Negotiating Committee for Higher Education
Staff (JNCHES), 2011; Grimshaw and Rubery, 2007). This concentration of women
in jobs which do not require significant, if any, qualifications, and which are often
part-time, lowers women’s average pay relative to men’s.

3.1.4 Undervaluation theory

The persistence of the gender pay gap suggests the possibility of a stigma
associated with occupational feminisation – that work done by women is socially and
economically undervalued. This theory is most prevalent in the United States (see,
for example: England, 1992, 2005, 2010) though it is also supported in the UK (for
example Perales, 2013). The theory posits that society undervalues certain types of
work precisely because women do it.
Pay practices are ‘socially constructed’ and lead to undervaluation of women’s

labour in a range of ways – pay is heavily influenced by social pressures and norms,
as well as by the actions of employers, governments and trade unions. Pay is often
decided based on typically male behaviours such as performing long hours, working
continuously for a long time and an aggressive negotiating style. By not conforming
to these norms, some women lose out. On the other hand, women are still seen by
society as secondary earners, as well as likely to derive more intrinsic reward from
their work than men, thereby justifying lower salaries (Grimshaw and Rubery, 2007).
Given the overall trend towards a narrowing of the gender pay gap, however, it could
be that undervaluation theory is becoming less relevant (Jackson, 2008). The fact
that the gender pay gap varies across different countries suggests devaluation is not
universal or uniform (Bettio, 2002; Bettio and Verashchagina, 2009).
Further, many jobs in which women predominate are not stereotypically ‘feminine’ –
for instance, clerical work (Hakim, 1998) – and gender-integrated occupations are
generally better paid than both female-dominated and male-dominated occupations
(Hakim, 1998; Cotter et al., 2004; Magnusson, 2013). It also seems to be the case
that women have gained in wage terms from working in skilled occupations and that
there is an underlying trend in productivity in favour of women who are benefitting
from the shift to a skills-based, non-manual economy (Brynin and Perales, 2016).
This also undermines the theory in general but it remains possible, Brynin and
Perales argue, that the work of less qualified women is undervalued. In sum, proving
the empirical validity of undervaluation (or devaluation) theory has not been easy.
Insofar as the theory is correct, however, the solution is to assess the comparative
worth of feminised against less feminised occupations.
Finally, the pursuit of parity in gender pay has proven to be highly complex and has
been described as a ‘case of constantly moving goalposts’ (Rubery and Grimshaw,
2014). The authors argue that pay setting is linked with wider societal and economic
trends, such as the growing disparity in wealth and the waning influence of trade
unions. ‘Levelling down,’ whereby men’s pay is reduced by employers to the same
level as women’s, may reduce gender pay inequality but polarise incomes as a

whole. In this scenario, those on low and middle incomes experience wage
stagnation while those on high incomes have much faster wage growth.
3.2 Measuring and explaining the gender pay gap
3.2.1 The size of the pay gap and trends

The differences between the pay of men and women in employment have been
examined in detail in many countries for a number of years. A general consensus
exists as to the size of the UK gender pay gap, the direction of travel and the main
factors contributing to it.
The Annual Survey of Household Earnings (ASHE) is a key source of data on the
gender pay gap. Its findings are based on a 1% sample of employee jobs taken from
HM Revenue and Customs (HMRC) and Pay As You Earn (PAYE) records; it does
not cover the self-employed. The median, full-time gender pay gap decreased from
9.6% in 2014 to 9.4% in April 2015. This is the lowest since the survey began in
1997, although the gap has not changed very much in recent years. Including part-
time employees in the overall analysis, the pay gap in 2015 stood at 19.2%, the
same as in 2014. Comparing the pay of part-time men with part-time women, women
have a pay advantage of 6.5% in 2015, up from 5.5% the year before (ONS, 2015).
A previous Commission report (Perfect, 2013) found that, based on ASHE data, the
median, full-time gender pay gap had fallen from 17% in 1997 to 10% in 2010.1
An earlier Commission report (Metcalf, 2009) reviewed various studies based on

different datasets covering the late 1990s to the early 2000s. These gave an
unadjusted pay gap of around 20%-25%, but in one instance this was as low as 14%
and in another as high as 42%. This raw pay gap is the simple difference between
average male and female wages and does not therefore control for other factors.
When controlling for factors such as the average education of men and women or
the proportion in part-time work, the gap falls to around 10% in most cases. This
residual gap could be due to discrimination but also arises out of unexplained factors
(information not available in the data).
1
The ASHE is not used in the statistical analysis for this report because, unlike the Labour Force Survey (LFS), it
does not include other variables that may affect pay, such as level of education and family circumstances.

3.2.2 Factors linked with the gender pay gap

There is a consensus that two main factors associated with the pay gap are
segregation and work patterns, in particular, the distribution of part-time work
(Metcalf, 2009). There are two aspects here: the gap within the dimensions
examined and the distribution of women across the dimensions. When it comes to
graduates, for example, we must consider not only the differences in pay between
men and women but also the proportions of male and female graduates in
employment.
The gap varies considerably along various dimensions such as part-time work as
against full-time work (Metcalf, 2009). This can be interpreted causally, if only
loosely. If women who work part-time are paid less per hour than those who work
full-time, who are in turn paid less than men who work full-time, then female part-
time work contributes to the pay gap, even more so as part-time work forms a
substantial part of female employment. The women's part-time pay gap (comparing
the female part-time to the male full-time rate) is consistently very high, if falling.
Based on the British Household Panel Survey (BHPS) it declined from 36% in 1995-
97 to 31% in 2004-07 (Olsen et al., 2010). According to Perfect (2013), whose
analysis was based on ASHE data and therefore arrived at different figures, the gap
reduced from 43% in 1997 to 39% in 2012.
Age is another important factor, with the gap clearly narrower for younger age
groups (Harkness, 2005), reaching its widest point for women in their forties and
then declining slightly (JNCHES, 2011). This could be for a range of reasons. Firstly,
women’s educational achievement historically lagged behind that of men. Secondly,
motherhood in general greatly adds to the pay gap (Harkness, 2005; Rubery, 2008).
As a result it could be that women are less likely than men to be promoted late in
their careers.
Rising female educational attainment is eroding the pay gap. Compared to men,
women now have higher levels of education, and with the rise of female graduates in
employment the gender pay gap has declined. In the 2013-14 academic year, 56.1%
of university students in the UK were female (Universities UK, 2015).
Goldin (2008) has argued that, for the US, education is a key way to address
inequality because it effectively reverses the human capital deficit theory discussed
above. While some evidence suggests the possibility of a growing pay gap for
graduates in the UK (Joshi et al., 2007), this seems to vary considerably across
countries, with some exhibiting a greater gap at high levels of education, some the
reverse, and with the UK broadly speaking in the middle (Dolton et al., 2009). Other

evidence suggests strongly that rising levels of female education are reducing the
wage gap (Brynin and Perales, 2016).
There is also evidence in the literature of sex discrimination. A study of the financial
services sector (Metcalf and Rolfe, 2009) found evidence of gender bias in the
distribution of bonuses and performance-related pay such that in more than half the
cases the gap for discretionary performance-related pay was 45% or more rising to
80% in the 42 cases providing complete data .
Joint research in 2015 by the department for Business, Innovation and Skills (BIS)
and the Commission into pregnancy and maternity found evidence of poor treatment
of women: higher earners (mothers paid over £30,000 a year) were most likely to
report less favourable experiences in relation to career progression or financial
reward while on maternity leave, For example:
 6% said they failed to gain a promotion they felt they deserved
 6% said they received a pay rise or bonus that was less than their peers
 5% said they had a reduction in their salary or bonus.
A large research review across multiple organisations found that the sex differences
in rewards including salary, bonuses, and promotions were 14 times larger than the
sex differences in performance evaluations. The review also found that differences in
performance evaluations did not explain reward differences between men and
women (Joshi et al., 2015).
Other factors which are important include occupational segregation, reviewed above,
but also regional differences. There is little evidence on the latter; we contribute new
findings later on in the report.
Some factors, such as public sector employment, reduce the gap. However, it is not
clear whether this is a direct effect (resulting perhaps from a greater emphasis on
rigorous recruitment and promotion criteria and procedures in the public sector, but
also greater union representation), or an indirect effect stemming from the higher
proportion of graduates (at least half female) in the sector. Even here equality is far
from perfect, with women contributing 53% of all employment in the sector but
having only 33% of senior roles (JNCHES, 2011).
Further, it seems that while the pay gap in the private sector (for full-time workers)
was about 22% in 2009 compared to about 15% in the public sector, the public
sector gap has been rising while the private sector gap has been falling (JNCHES,
2011). This last report also shows that the gender pay gap for academic staff in a

sample of older universities in 2005-06 ranges from about 15% to 18%, but it is less
than half that in newer universities.
There are important occupational differences in the pay gap, and this makes clear
again that the distribution of women’s employment across occupations is an
important factor. For instance, for full-time work the gap in 2012 was only 4.9% in
sales and customer service occupations, which are low-paid. There is greater
equality in pay among those who are paid poorly. The highest gap, at 21.9%, was
among process and plant operatives, although in this occupation female employment
was only 18.2% for managers and senior officials and 9.3% for professionals. It was
somewhat higher (12.4%) for associate professional and technical employees
(Perfect, 2013).
There is consensus in the literature as to the persistence of the gender pay gap as
well as the fact that it seems to be declining over time. There is also a consensus in
respect of the main drivers of the gap, in particular the concentration of women in
relatively low-paid jobs and female reliance on part-time work (which lowers earnings
because part-time positions tend to pay less than full-time ones). These key facts
highlight the issue that many women are concentrated in poorly-paid jobs, even if the
pay gap among low-paid men and women is quite small. These themes emerge in
the analysis of the drivers of the pay gap, presented in the following chapters.

The gender pay gap Research findings: features of the gender pay gap
4 |Research findings: features of the

gender pay gap
We first look at the gender pay gap across time based on our analysis of the Labour
Force Survey (LFS) between 1993 and 2004. We then compare the gaps across
different geographical, social and employment categories described above.
4.1 Patterns and trends in the gender pay gap
The pay gap is generally expressed in the form of female pay as a percentage of
male pay, often, as here, for hourly pay. The data used in this research show that
female hourly pay as a percentage of male hourly pay has increased steadily over
time, and so, in common with the findings discussed above, the pay gap has been
shrinking. As explained in the methodology, the findings here are based on all
employees in the sample working at least 10 hours per week.
Table 4.1: Female mean and median hourly pay as a percentage of male hourly
pay
Difference in means Difference in medians
1993 73.0 70.6
2000 78.8 75.5
2006 85.0 80.6
2014 90.7 84.7
This narrowing of the gender pay gap has occurred concurrently with a decline in
male employment over the last 30 years. The decline began to level off after 1990,
while the employment rate for women has risen steadily (but less since 1990). In the
data used here, 47.9% of employees (therefore excluding the self-employed) were
female in 1993 and this had risen to 52.8% by 2014. In April to June 2013, around

67% of women aged 16 to 64 were in work, an increase from 53% in 1971. For men
the percentage fell to 76% in 2013 from 92% in 1971 (ONS, 2013).
Therefore the fall in the pay gap has accompanied a fall in the employment gap.
Based on the mean, the current pay gap between men and women is less than 10
percentage points whereas it was 27 percentage points in 1993. However, the use of
the median tells a different story. Not only is the improvement slightly less than in the
case of the mean but the pay gap is always larger. Women earn on this basis on
average not around 91% of the male wage but around 85%.
This is in part because, as stated above, the median better reflects the skewed
distribution of wages and women predominate at the lower end. Given this, and the
fact that the gap is consistently declining, a key issue is the problem of low pay. In
principle no-one should be paid less than the minimum wage and the published
‘Living Wage’2 was only in place for part of this period. In addition, there are age and
regional variants in these wage controls. Consequently, for simplicity, low pay is
defined arbitrarily here as below £8 an hour (please note this is not an official figure
of any kind). The amounts are deflated, so it is reasonable to use this yardstick for
the entire period. In 1993, 25.9% of men were earning below £8 per hour compared
with 51.0% of women. By 2014 both had fallen, to 20.4% and 30.3% respectively.
Yet the female rate was 50% higher than the male rate. We will return to this issue
later.
This outcome is demonstrated differently in Figure 4.1. This graph shows the
distribution of hourly pay (ranging from £2 to £80) for men and women separately,
and pools all the years of the survey covered here. The female wage is very much
skewed to the left, with a large proportion of women earning a few pounds per hour
to no more than £10, while far more men than women earn between £10 and £40.
This result is much the same as shown by Olsen et al. (2010). A major cause of the
pay gap is therefore the fact that far more women than men are low paid.
2
The Living Wage is set independently by the Living Wage Foundation and is based on the cost of living in the
UK. Paying the Living Wage is voluntary. It is higher than the National Minimum Wage, which is a legal
requirement for employers and is set by Government.

Figure 4.1: The distribution of male and female pay (£ per hour)
.15
Distribution of men/women
.1
.05
0
0 20 40 60 80
£ per hour
male pay female pay
If we look at the pay gap at different points in the pay distribution, this is higher at
higher levels of pay (because the range of pay is higher), but it is highest in relative
terms (that is, as a percentage of the female wage) at middle levels of pay (defined
arbitrarily as over £8 per hour and less than £15 per hour). Here it is 3.7% of female
pay (pooled over the period). Below that it is close to zero, and above that it is 2.6%.
These low figures compare with 21.5%, which is the pay gap as a percentage of the
average female wage for the whole female sample. By splitting wages into three
levels we are effectively controlling for the effects of the distribution and it is the latter
that is the central problem. Roughly equal proportions of men and women are in the
middle wage category over the period, but about 20% of men are in the lowest
category compared to 36% of women. To put it another way, women form 65% of the
lowest income bracket but only 38% of the highest.
Figure 4.2 describes the pay gap over the entire period. These are based on the
median. All of the graphs are smoothed, that is, they remove fluctuations in the data
to show trend lines. We construct the pay gaps here such that a negative amount
means that women earn less than men. Four outcomes are of note from Figure 4.2.
First, it can readily be seen that the gap is negative over the entire period on both
measures: women are on average always paid less than men both overall (the

general gender pay gap) and within occupations (the occupational pay gap). Second,
inequality on both measures has declined over time. Third, the occupational gap is
smaller than the general gap though it nevertheless remains quite large. Broadly
speaking, this means that inequality for women has two constituent elements.
Women are paid less than men for doing roughly the same type of work (though
possibly at a lower level of seniority), while they also tend to be concentrated in
lower paid occupations. The occupational gap therefore explains a part of the
general gap. Fourth, these two types of gap are converging, which implies that
inequality within occupations is of decreasing importance as an explanation of the
general pay gap.
Figure 4.2: The general and occupational gender median pay gaps over time (£
per hour)
0
-.5
-1
£ per hour
-2 -1.5
-2.5
-3
-3.5
1995 2000 2005 2010 2015

year
general occupational
4.2 How the pay gap varies by characteristics
4.2.1 The pay gap and age

A central feature of women’s employment relative to men’s is that the pay gap grows
with age. Wages rise for most people as their careers develop, although the extent of
wage growth varies by profession. It is likely that women benefit less than men from
older age, and therefore experience, as they are more likely to have interrupted

careers. The age profile is also influenced by the fact that women were, in the earlier
period of our analysis, less likely than men to have degrees. Women might also face
a ‘glass ceiling’ based in part on a lower probability of promotion than men. On
balance we would therefore expect the pay gap to rise over the life course.
Figure 4.3 compares four age groups (17-29, 30-39, 40-49 and 50-59) over the
period we examine (1993-2014). This analysis is not of the same women over time.
That is, we cannot tell from these data how women actually fare over their careers.
Instead, we interpret each successive cohort as experiencing the same conditions
the earlier cohort would probably experience if individuals within this were now 10
years older.
Figure 4.3: The general gender median pay gap over time and by age (£ per
hour)
0
-1 -2
£ per hour
-3 -4
-5
1995 2000 2005 2010 2015

year
Age 17-29 Age 30-39

Age 40-49 Age 50-59
The pay gap is higher with age (the lines of the older cohorts are below those of the
younger cohorts) but all age groups have seen a drop in the pay gap over time. At
the start of the period we can see, first, that the pay gap is high for all groups except
the youngest (who, as mentioned above, receive relatively low wages). The older the
group the greater the pay gap for three of the cohorts, but for those women in their
forties it would seem the pay gap is highest for most of the period – that is until
around 2010. As mentioned, in all cases there is a decline in the pay gap over time.

We can also see that the lines are far closer together at the end of the period than at
the start. Thus, it seems that while women's career development does not yet match
men's, the difference is declining over time.
Figure 4.4 depicts the occupational gender pay gap. The trend is broadly the same
but there are some differences. While the youngest group had a fairly flat profile over
time in terms of the general gap, within occupations this group's pay improves more
noticeably. At the end of the period, on average young women (aged 17-29) do not
face an occupational pay gap. This is possibly the result of general progress towards
greater equality of the sexes, driven in part by the fact that young women are likely to
have an education at least equal to that of young men. As shown in Figure 4.3,
women in their forties (historically the age group experiencing the biggest pay gap)
'overtake' women in their fifties, albeit earlier in the period.
Figure 4.4: The occupational gender pay gap over time and by age (£ per hour)
0
-.5
-3.5 -3 -2.5 -2 -1.5 -1
£ per hour
-4
1995 2000 2005 2010 2015

year
Age 17-29 Age 30-39

Age 40-49 Age 50-59
These trends are highlighted again in Table 4.2 below. In the above figures the gap
was shown as a female deficit, that is, with minus signs, as this is visually clearer.
They show the gap rising towards zero (equality). In the subsequent tables the gap is
expressed as the male minus the female wage, that is, as a male surplus. Therefore
the figures in the tables do not generally have minus signs, unless there is a female
advantage in pay.

The upper half of Table 4.2 shows the pay gap (the average female wage subtracted
from the average male wage) at three points in time, each 10 years apart. The lower
half of the table shows the pay gap as a percentage of the average female wage for
that year and that age group. In 1994, the gap for women in their forties was very
high, 48% of the average wage for women in that group. Further, while there has
been a big decline in the pay gap over time, for older women it is still 17% of their
average wage.
Table 4.2: The gender pay gap for four age groups in 1994, 2004 and 2014
Age
17-29 30-39 40-49 50-59
Pay gap (£) in:
1994 1.06 3.17 4.79 4.20
2004 0.59 2.72 3.73 3.46
2014 0.43 0.97 1.54 1.99
As % of female
wage
1994 12.8 30.9 48.5 46.1
2004 5.8 21.0 30.1 29.2
2014 4.5 7.7 12.1 16.8
If we use the median pay gap rather than the mean, the results are similar but the
gaps are larger – though with the consistent exception of the youngest age group.
The biggest gap is again for women in their forties, which was £5.09 in 1994, £4.34
in 2004, and £2.37 in 2014.
If we were to express this in terms of relative wages using the mean, in 1994 women
in their forties earned 67.3% of the wages of men in the same age group, rising to
76.9% in 2004 and 89.2% in 2014. The median, as before, worsens this picture to
61.6%, 71.0%, and 82.5% respectively. On this basis women in their forties were
earning only just over four-fifths of the equivalent male wage in the most recent year.

4.2.2 Part-time and full-time work

As mentioned above, the sample excludes those with only marginal hours (working
10 hours or less per week). Taking ‘usual working hours’ rather than actual hours
worked and defining part-time work as less than 30 hours a week, this produces an
average working week for full-time men of 39.4 hours and for full-time women of 37.1
hours over the period. In the case of part-time work, the figures are 19.9 in both
cases. However, 36.3% of women work part-time over the period compared to only
6.3% of men, though this figure has risen for both women and men. In 1993, the
figures were 35.7% for women and 4.3% for men, while in 2014 the figures were
37.2% and 8.8% respectively. Clearly, different cut-off points for defining part-time
work would produce different percentages.
Table 4.3 compares the pay of women and men working full-time and part-time for
two five-year periods. The gap between male full-time and female part-time workers
is given in the bottom row of the table. This has declined but remains large, more so
when calculated on the basis of the median. However, this is only part of the story.
Table 4.3: Pay gaps (£ per hour) by part-time and full-time work over time
Difference in means Difference in medians
1993-1997 2010-2014 1993-1997 2010-2014
Men/women full-time 2.77 1.04 2.57 1.29
Men/women part-time 2.13 -0.63 0.35 -0.97
Men full-time/men part-time 2.01 3.38 4.09 4.86
Women full-time/women part- 1.37 1.71 1.87 2.60

time
Men full-time/women part-time 4.14 2.75 4.44 3.89
Using the mean figures, men earn more than women in both full-time and part-time
work in 1993-1997, but there is little difference between these amounts (£2.77
against £2.13). However, male full-time workers have a bigger advantage over male
part-time workers (£2.01) than female full-time workers have over female part-time
workers (£1.37). Things are very different in 2010-2014. The gender differential for
full-time workers has reduced and it is actually in favour of women by 63p in the case
of part-time work. Part-time work has become more prevalent but it has also become
more poorly paid compared with full-time work, and this is affecting both men and
women. The gap between female part-time and full-time work has grown from £1.37

to £1.71, but the equivalent gap for men has seen a bigger increase from £2.01 to
£3.38.
When we use the median these outcomes are even starker. The current gap
between male full-time and part-time workers is very large at £4.86 per hour. The
situation for men in part-time work has become increasingly precarious. The problem
for male part-time workers is low pay. The problem for women is that a very high
proportion of women in employment (nearly two-fifths) are part-time.
Expressing the wage of female part-time workers as a percentage of the male full-
time wage produces a figure of 67.2% based on the mean for the earlier periods,
rising to 79.8% in the later period. The figures based on the median are 60.6% and
69.3% respectively. So, on the worst count, over 30 percentage points still separates
part-time women and full-time men in recent years.
4.2.3 The pay gap and family

In this section we see how the gender pay gap, based on the mean, varies by
whether people are married and by the number of children they have.
The highest pay gap is between married men and married women, at £3.30 per hour
(averaged over the entire period). In the case of cohabiting men and women it is
£1.94, while for those who are single it is only 40p. Single people tend to be younger
– their average age is 29 as opposed to 44 for married people. So pay is on average
lower for single men and women. For those who are separated, divorced or widowed
– only 9% of the sample – the pay gap is again high at £2.65.
As the gap is highest for married people and these form 58% of the sample, the
analysis of marital status now focuses only on married and unmarried people
(whether single, cohabiting, separated and so on). Table 4.4 shows pay gaps by
marital status between men and women but also between married and unmarried
men and similarly for women. It also shows the same distinctions comparing full-time
to part-time workers, as engagement in part-time work is in large measure the result
of family responsibilities.
Looking at men compared to women (the first two rows), the gap is higher for
married women whether they work full-time or part-time, but especially high for those
working full-time. That the gap is higher for married people clearly has a lot to do
with age – that is, married people tend to be older and men, who are generally less
likely to have family or caring responsibilities, are more advanced in their careers. It
is unclear, therefore, what effect marriage actually has – an issue that will be
examined further in the subsequent table.

Table 4.4: The gender pay gap comparing married to unmarried people (£ per
hour)
All Full-time Part-time
Unmarried men/women 1.10 0.72 0.51
Married men/women 3.30 2.51 0.79
Married/unmarried men 2.91 2.78 2.74
Married/unmarried women 0.71 0.99 1.44
Table 4.4 shows that the largest pay gap is between married and unmarried men.
The ‘marriage premium’, the fact that married men earn more than unmarried ones,
is a well-documented phenomenon, although there are competing theories which
attempt to explain it (this report does not elaborate on these as they are out of scope
of the research).
Table 4.5 breaks down the marital analysis by age, which, as pointed out above,
might be a confounding factor. Reflecting the earlier results, the first two rows show
that the pay gap is highest for women in their forties (remembering, though, that this
result is averaged over a 20-year period). This applies to both married and
unmarried women, however the former experience a larger pay gap (£4.02 per hour).
The pay gap between married and unmarried people, which is considerable, is not
solely a function of age. Indeed, when we look at the gap between the wages of
married and unmarried men, this is highest for those aged under 30. Over time,
therefore, increasing age reduces the pay advantage of married men over unmarried
men, but they maintain an advantage nevertheless. For women the effect of age
leads to unmarried women earning slightly more than married ones, as they advance
in years. The pay gap therefore becomes increasingly in favour of unmarried women
with age.

Table 4.5: The gender pay gap by marital status and age (£ per hour)
Age <30 Age 30-39 Age 40-49 Age 50+
Unmarried men/women 0.58 1.65 2.24 1.81
Married men/women 1.88 2.90 4.02 3.27
Married/unmarried men 2.57 1.23 1.46 1.01
Married/unmarried women 1.27 -0.02 -0.21 -0.44
Age plays a role in this, as does having children. Table 4.6 shows the latter outcome.
The figures do not show the effect on women’s pay of having children but rather
whether women with children (0, 1, 2 or 3+) experience greater pay gaps than men
with equivalent responsibilities. In fact, the number of children makes little difference
to women’s pay, which is £11.20 an hour where there are no children, £10.90 where
there is one child, £11.40 for two and £10.60 for three or more. The data suggests
that having children has a disproportionate impact on women’s pay, compared to
men’s. The results are shown for two widely separated five-year periods.
Table 4.6: The gender pay gap by presence of dependent children in

household
Number of children 1993-1997 2010-2014
0 2.43 1.08
1 3.40 1.63
2 4.66 2.03
3+ 4.63 1.79
In the earlier period it is clear that where women manage to remain at work, the pay
gap increases significantly when there are children in the family and further when
there is more than one child. While in the later period the gap is far lower it still
increases for women with one child. Compared with having no children, having a
child increases the pay gap considerably. This rises further with one more child but
then levels off or falls slightly. The difference between having no child and one child
is 97p in the earlier period and 55p in the later period. However, these increases are
40% and 51% respectively. The wage penalty for childbearing, as expressed in the
gender pay gap, therefore seems to have increased slightly. Although the penalty for
having three or more children is a little less than for two children, this is only a small

decline, indicating perhaps that men in this situation face as much a penalty as
women.
The LFS contains no information on caring or household responsibilities. To examine

this we use the UK Household Longitudinal Study (UKHLS), Waves 1-5, which
covers the period 2009-2014. We do not look at this within households so the
analysis is of all women against all men. First, upon divorce (or separation) women
lose or have reduced access to the generally higher income of their husbands. It is
unclear, however, whether this should affect hourly pay. As with the LFS we find that
the gender pay gap is highest for married people (£3.45 per hour) but only slightly
less for those who are separated (£3.38 per hour) and divorced (£3.20 per hour). It is
lower for cohabitees, who are of course younger on average, at £2.04, and quite
marginal for single and never-married people at 59p.
The survey asks respondents how long they spend on housework in a week. As
would be expected, women not only do more housework, but this amount is closely
(and negatively) related to work hours. The correlation between housework and work
hours for women is -0.29, indicating that where women work fewer hours they do
more housework. The correlation for men is much smaller, at –0.07, though still
statistically significant. Thus men do not vary their housework hours much according
to how many hours a week they do in paid work. Their contribution tends to remain
low regardless.
We have no means of knowing how reliable these housework amounts are.

However, we group hours into three categories: low (four or less hours), medium
(five to nine) and high (10 or more), so this will remove some of this unwanted
variation. As this question is only asked in three of the five Waves (which are pooled
in the analysis) the total sample is roughly halved. Table 4.7 shows the pay gaps for
these categories, the gap in paid typical work hours for men and women, and the
proportion of the entire sample in each category. Wages are not adjusted for inflation
as they only cover a five-year period. In fact, the pay gap barely changes over this
period (though it does not go down).

Table 4.7: Housework and paid work hours (UKHLS; 45,533 observations)
Housework Work hours

hours Hourly pay gap gap Percentage of sample
Men Women
Less than 5 2.26 3.77 27.6 12.8
5-9 2.10 5.51 13.0 17.6
10 or more 2.67 8.20 5.2 23.7

Note: The six cells in the last two columns total 100%.
Looking at the second column, it is clear that the greater the level of housework the
greater the gap in work hours, quite simply because women bear the brunt of this.
The last two columns show that 23.7% of the sample comprises women doing 10
hours a week or more of housework, compared to just 5.2% of the sample where
men do the same amount, while 27.6% of the sample comprises men who do less
than five hours. The proportion of men and women doing 5-9 hours of housework is
much more similar: men in this category comprise 13.0% of the sample and women
17.6%. Looking at the second column, it is clear that the greater the level of
housework the greater the gap in work hours, implying that men who do 10 or more
hours of housework may do it in addition to paid work, while women may do it
instead of paid work.
When we look at the pay gap we get an unexpected result. The gap varies little
across the first two categories but is higher at 10 hours or more housework a week.
The reason is that while men doing this amount of housework have a relatively low
hourly wage rate, the rate for women doing this much housework is proportionally
even lower. These women (comprising nearly a quarter of the sample in the table)
work less than other women and men and thereby have relatively low earnings, but
even compared to men doing similarly high average housework they are likely to be
poorly paid. Whatever the nature of the causal influences at work here, it suggests
that there is a relationship for working women between high housework demands
and low pay.
As for hours spent on care (whether in or out of the home, but excluding childcare),
82% of women and 87% of men do none at all. For men and women who do up to
and including four hours a week (7.3% of the sample) the pay gap is £2.57, falling to
£1.96 for 5-9 hours (but this is only 3.5% of the sample) and £1.77 for 10-19 hours
(2.0%). Thus, care responsibilities affect men as well as women, such that the higher
the number of hours spent on care the lower the gender inequality in terms of pay.

However, this lower level of inequality in experiencing such negative effects applies
to a very small proportion of the population.
4.2.4 The pay gap and education

In 1993, 17% of male employees had a degree (or equivalent higher education)
compared to 11% of women. In 2014, more women than men entered university and
this is reflected in the data with 28% of male employees being graduates compared
to 33% of female employees. Over the period, the gender pay gap for graduates is
£1.90 per hour, while it is £2.42 for those with A-levels or equivalent, and £2.18 for
those with lower qualifications. The pay gap is therefore greater in absolute and
relative terms for lower paid people.
The next table shows the pay gaps for these three educational levels for the same
periods as previously, both in absolute terms and as a proportion of the female wage
in each group. In the earlier period the biggest differential is for graduates, but
because relatively poorly qualified women earn less than others this differential is a
far greater proportion of women’s average wage – as much as 31.4%. In the later
period there is only a small differential remaining for graduates but it is far greater for
those with A-levels and not much lower for those with less than that. These figures
as a percentage of the female wage for each group are lower than before, especially
for women with no A-levels, but are still large. The main policy problem therefore is
that while highly educated women are to a large extent able to escape disadvantage
relative to men, less qualified women are not.
Table 4.8: The pay gap by educational level
Pay gap Pay gap

as % of as % of
female wage female wage
in each in each
1993-1997 group 2010-2014 group
Graduates 3.15 20.9 0.80 5.1
A-levels/equivalent 2.38 23.4 2.23 20.5
Less than A-level 2.44 31.4 1.50 15.7
Because analysing the gap within levels of education is in effect controlling for
education (in other words, comparing like with like) the skewedness of the wage

distribution is reduced and so the median results (not shown) are similar to the
above. Attending higher education in larger numbers is therefore enabling female
graduates to reduce gendered wage inequality.
Why do we get this discrepancy between education and income, with a high gender
pay gap for people with no A-levels and a low gap for those in low income brackets,
as shown earlier? It could be because a higher proportion of men with low
qualifications are in the higher wage brackets. For instance, 50% of women in the
highest wage bracket have a degree compared to only 37% of men. Many of these
will be in managerial jobs where a degree has in the past not been seen as a
prerequisite, or in highly skilled manual jobs based on vocational training. Another
factor is that the hourly wage of people with A-levels is only slightly higher than those
with lower or no qualifications.
If we express the female hourly wage as a percentage of the male wage, then in the
early period this is 82.7% for graduates, 81.0% for those with A-levels, and 76.1% for
those with lower qualifications than A-levels. However, in the later period these
figures are 95.2%, 83.0% and 86.5% respectively. The gaps shown in the second
and fourth columns of Table 4.8 show that these differences can be acute where
women’s pay is already low.
An additional human capital measure we can use is the percentage of graduates in

an occupation. The correlation between this and hourly pay is .50 for men and .61 for
women. It seems women, whether graduate or not, gain more than men from
working in a graduate occupation.
4.2.5 The public and private sectors

Over the entire period 20.8% of men worked in the public sector compared to 36.5%
of women. The public sector is therefore very important to women’s pay. Does public
sector pay help to reduce inequality for women? It is well known that wages are
more polarised in the private sector, with a greater disparity between the lowest and
the highest salaries. In the public sector, the pay ranges are narrower because of
collective wage agreements, the stronger presence of trade unions and more fixed
and transparent institutional arrangements (though some large private sector
organisations might be comparable in this regard).
However, it is not clear that these factors are of primary importance because it is
also the case that the public sector attracts more highly educated people, so the
comparison between the two sectors alone does not compare like with like. In Table

4.9 the pay gap by sector is shown in the first column, and then by graduate status
within sector, pooling all the years together.
Table 4.9: Pay gap by public/private sector and graduate status (£ per hour)
All Graduates Non-graduates
Public 2.38 1.63 2.27
Private 3.11 2.77 2.98
The gender pay gap is lower (averaged over the period) in the public sector than in
the private sector, though still substantial. The gap is also larger among non-
graduates than graduates, in both sectors. However, the pay gap between male and
female graduates in the public sector is comparatively small, at £1.63 per hour, while
the same gap in the private sector is £2.77.
These results imply that the difference in the pay gap between public and private
sectors is not entirely due to the fact that there are more graduates in the former. As
Table 4.9 shows, female graduates in the private sector gain little from their
education in terms of parity of pay with men. In the public sector, on the other hand,
women with degrees experience a much smaller pay gap than those without, when
compared with men within the same cohort.
In all, the public sector does not confer as much advantage to women as might be
thought, as only 10.5% of all female employees in the entire sample are graduates
working in the public sector, while 26.0% are non-graduates in the sector. The
equivalent figures for men are 7.1% and 13.8%. A large number of women are in
non-graduate public sector employment, which is associated with a substantial pay
gap.
Basing the gap on the median produces similar results except for graduates in the
private sector, which rises to £3.55.
Expressing the female hourly wage as a percentage of the male wage, female
graduates earn 84.8% of the male wage in the private sector and 91.5% in the public
sector. For non-graduates the figures are 75.6% and 83.2% respectively. Comparing
the whole private sector with the whole public sector produces figures of 76.3% and
84.5% respectively.

4.2.6 The pay gap across the UK’s regions

Table 4.10 shows the absolute pay gap within regions in the first column, based on
the mean. The second column shows this pay gap as a percentage of the average
female wage in the region. These results are ranked by this percentage, with the
lowest at the top of the table. It should be noted, however, that the regions are where
people live, not necessarily where they work. The regions reflect those used in the
LFS.
Table 4.10: Pay gaps by region
Pay gap as % of
Pay gap (£) female wage
Inner London 0.55 3.77
Northern Ireland 1.30 12.50
Outer London 1.95 14.42
West Yorkshire 2.02 18.88
Greater Manchester 2.03 18.92
Wales 2.00 19.07
Tyne and Wear 1.99 19.17
Strathclyde 2.20 20.07
Merseyside 2.13 20.96
West Midlands 2.22 21.20
Rest of Scotland 2.38 21.83
South Yorkshire 2.29 22.49
Rest of North West 2.45 22.96
East Anglia 2.50 23.43
East Midlands 2.47 23.48
Rest of West Midlands 2.60 24.69
Rest of South East 3.08 25.83
South West 2.72 25.59
Rest of Northern region 2.58 26.26

Rest of 2.66 26.26

Yorkshire/Humberside
UK 2.39 21.43
The pay gaps by region do not exhibit any clear overall pattern. However, London
has by far the lowest relative gaps. The gap is even lower in Inner London based on
the median (35p). This London difference runs counter to earlier work (Perfect, 2013)
using Annual Survey of Household Earnings (ASHE) data which shows that the full-
time gap based on the median is 12.0% in London as a whole in 2012 – higher than
the national average. Olsen et al. (2010) find the same using the British Household
Panel Survey (BHPS) which has a much smaller sample. Table 4.10 is based on the
mean rather than the median, however, and this might have made a difference.
(Regions are also areas of work rather than residential areas in the ASHE analysis.)
Table 4.10 is based on the gap as a percentage of the female wage (in each region)
rather than the female wage as a percentage of the male wage. Nevertheless, the
patterns should be the same.
This ‘London effect’ perhaps suggests that regions in which there are many highly
skilled jobs tend to have smaller gender pay gaps. The average age in London is
three years lower than the national average (in the data). The national percentage of
employees in the data with a degree is 21.2% compared to 42.8% in Inner London
and 27.6% in Outer London. Commuting areas such as the South East and East
Anglia, interestingly, have relatively large pay gaps. Comparing Inner London, where
the pay gap is 55p (3.8%) to the ‘Rest of South East’ (that is, excluding London),
where the gap is £3.08 (25.8%), the latter is five and a half times larger. The
absolute difference is nevertheless a matter of pennies between many other regions.
Just taking the regions with the lowest absolute gap (Inner London) and highest
(Rest of South East), the female wage is 96.4% of the male wage in the former and
79.5% in the latter.
How important is this in terms of the number of people affected? While the proportion
of men and women varies only slightly between regions, they clearly vary
enormously in number. London (Inner and Outer combined) accounts for 9.1% of the
sample, that is, nearly 1 in 10. The rest of the South East, where the gender gap is
highly unfavourable to women, accounts for 19.8%, slightly over double the
contribution of London. Northern Ireland contributes only 2.8%.
Table 4.10 pools the years across the entire period, but not only has the pay gap
declined, it is likely that differences between regions have changed. Table 4.11
shows results for two five-year periods, 1993-1997 compared to 2010-2014. In the

earlier period the ranking is again by the pay gap as a percentage of the female
wage. Looking at the second column shows how much individual regions have
shifted position over time. Several positions change, though overall in the later
period the picture is not very different from the earlier period. In particular, the three
regions where the pay gap is the lowest percentage of the female wage are the
same as in Table 4.10, suggesting some stability.
Of greatest note, however, is how large the percentages are in some regions in the
earlier period – as much as 40.4% in the South West. The highest in the later period
is 16.8% (in the Rest of Yorkshire and Humberside, that is, excluding West and
South Yorkshire). If we define ‘small’ as less than 10%, the rate is now small in
seven regions. Remarkably, in Inner London it is actually now in favour of women
while in Outer London it is only 4.0% in favour of men.
Is this a sign of the future, with London leading the way? In fact, taking 2014 alone
(not shown), the rate is very slightly in favour of women also in Outer London while in
Northern Ireland the male advantage is only 55p. In that year the female wage is
103% of the male wage in Inner London. However, it ought to be borne in mind that
the analysis excludes extremely high values, of which there are likely to be more in
Inner London, due to the higher number of very well-paid positions in industries such
as finance and banking.
Table 4.11: Pay gaps by region and over time
1993-1997 2010-2014
Pay gap (£) As % of Pay gap (£) As % of

female wage female
wage
Inner London 1.34 10.32 -0.34 -2.36
Northern Ireland 1.76 20.83 0.91 8.08
Outer London 2.94 25.13 0.57 4.04
Merseyside 2.53 27.56 1.46 13.18
West Midlands 2.75 31.14 1.09 9.67
West Yorkshire 2.76 31.15 1.37 12.02
Strathclyde 2.92 31.50 1.36 11.45
Tyne and Wear 2.79 32.37 0.99 8.79

Wales 2.81 32.52 1.08 9.40
South Yorkshire 2.79 32.75 1.54 14.23
Greater Manchester 2.95 32.96 1.13 9.83
Rest of 2.96 35.45 1.84 16.77

Yorkshire/Humberside
Rest of North West 3.20 35.59 1.28 11.04
East Anglia 3.24 36.82 1.78 15.59
Rest of Scotland 3.29 37.17 1.37 11.42
Rest of Northern region 3.19 37.31 1.87 17.12
East Midlands 3.28 37.96 1.61 14.07
Rest of West Midlands 3.43 39.15 1.64 14.25
Rest of South East 4.00 39.64 1.71 13.46
South West 3.50 40.41 1.61 13.68
UK 3.18 33.90 1.38 11.56
4.2.7 Features of the gender pay gap: main conclusions

The gender pay gap has declined considerably over the last two decades, but
women’s average hourly wage is still between roughly 85% and 90% of the average
male wage (depending on the measure used). The pay gap within occupations was
also high at the start of the period– in fact, this formed a substantial proportion of the
general wage gap. This suggests that women were then paid less for doing roughly
the same work as men, or had worse jobs than men within the same occupations.
While this within-occupation gap has declined, almost the entire general wage gap is
now taken up by this occupational gap.
Many women are in low-paid occupations, and not only that, tend to occupy the
lower-paid positions within these. The decline in the pay gap is reflected by the fact
that the gap is far higher for older people, implying an historical effect that might be
gradually fading away. It seems that the pay gap which affects older women is in part
a ‘career’ effect stemming from career breaks, and that this effect is itself declining.
Related to these trends, the pay gap for graduates is now close to negligible but
remains high for non-graduates and for those with A-levels or equivalent.

The gender pay gap Research findings: the drivers of the gender pay gap
5 |Research findings: the drivers of the

gender pay gap
This section of the report explores variables which are associated with raising or
lowering pay. These are referred to as ‘drivers’ although it is only possible to say that
these are associated with, rather than the causes of, pay gaps. Various statistical
techniques are used to analyse these drivers. The following paragraphs outline the
main findings, and the methods used where appropriate.
5.1 Graduate status and occupational segregation
That pay is lower in occupations in which women predominate is well known and
confirmed in this report. Both men and women who work in female-dominated
occupations are paid less than those who work in male-dominated occupations.
Based on their larger numbers, women are obviously more likely to experience this
disadvantage than men.
We indicate this with simple correlations where a correlation of zero means no

relationship between pay and feminisation of an occupation (i.e. the number of
women), and a positive correlation of one shows a hundred percent relationship. A
negative correlation, in contrast, means that the two are inversely related: when one
is high the other is low. The correlation is –.01 for men and –.17 for women (both
correlations being highly statistically significant, that is at p=<.001). Therefore within
feminised occupations women tend to have much lower incomes than men.
Do female-dominated occupations pay less than male-dominated ones even when

the average educational level within both is high? We look at this through dividing
the sample into graduates and non-graduates, and dividing each group into three
levels of feminisation: 0 to 33% (low feminisation), 34% to 66% (medium), and
greater than 66% (high). These are clearly arbitrary divisions but have the advantage
of producing three groups roughly equal in size. In Table 5.1 we show the pay gap
for these six groups and also as a percentage of the female wage in each of these.

The lower half of the table shows that non-graduate women working in mixed or
male-dominated occupations experience the largest pay gaps.
The table also shows that the pay gap is smaller for graduates than non-graduates at
all levels of occupational feminisation. Compared to female graduates, women
without degrees therefore potentially face low pay in general and particularly low pay
in relation to men.
Table 5.1: The gender pay gap by graduate status and level of feminisation in
occupations
Gender pay gap (£ per As % of female wage in

% female in occupation hour) each group
Graduate
0-33 (low feminisation) 1.90 11.4
34-66 (medium) 1.69 10.4
> 66 (high) 0.77 5.0
Non-graduate
0-33 (low feminisation) 2.04 19.6
34-66 (medium) 2.36 21.9
> 66 (high) 0.76 8.1
Based on the median, these gaps are slightly higher and the median gap is in fact
lower for non-graduates in highly feminised occupations (not shown).
5.2 Explaining the gender pay gap: which factors carry more
weight?
In this section of the report, all the variables and many others are included in a
separate regression analysis for women and for men based on ordinary least
squares (OLS). The dependent variable is the log of hourly wages, which means that
the coefficients can be interpreted as roughly the percentage effect of the variable on
the outcome. (This is less true, however, the larger the effect.) Thus, Table 5.2
shows that each year of age increases the pay of men by roughly 6% and of women

by 4%. For simplicity the table excludes the effects for the large number of industries
and regions but the full table is available in the Appendix.
Many of the outcomes discussed above in simpler analyses are confirmed, now in a
framework including many controls. Higher education advantages women more than
men in that it provides a greater within-gender premium (that is, the effect is in
comparison with women with middle-level qualifications, not directly with men). Also
apparent is that having low (or no) qualifications disadvantages women less than
men, in comparison with women and men (respectively) with A-level qualifications or
equivalent. Also, as shown above, part-time work affects men and women
negatively, but men far more than women. The same applies to temporary work. The
public sector lifts women’s pay.
A key variable is the proportion of graduates in an occupation. The higher this is, the
higher the average wage in the occupation, as would be expected. This benefits
women more than men. However, the graduate squared term shows that this
positive effect tails off so that at high levels of graduate density the advantage to
women declines. As shown earlier, women gain more than men not only from having
a degree but from working in graduate occupations.
‘Proportion female’ works differently as by definition this distribution is very different

for men and women. The loss of pay is greater for women than for men but, as the
squared term indicates, the negative effect becomes positive. A separate calculation
(not shown) reveals that at high levels of feminisation (well over 60%) women benefit
from working in female occupations, controlling for education and other factors.
The full regression analysis – the statistical process for estimating the relationships
among variables – presented in the Appendix strongly confirms the regional
variations discussed earlier. Despite controlling for other variables, for instance
education, statistically speaking women’s pay is lifted by living in London.
In some other areas such as Tyne and Wear, and Wales, but above all in Northern
Ireland, a severe wage penalty for living in the region is compensated in large part
for women in that this affects men far more. The reference category is the South
East of England (excluding London), which is the dominant region in terms of
numbers. People in Northern Ireland earn far less than those living in the South East,
but in Northern Ireland in percentage terms it helps to be female. The last point is
important. Men still earn more than women in Northern Ireland but their deficit
relative to men in the South East is greater than the equivalent for women.

Table 5.2: Regression (OLS) results for men and women
Men Women
Age 0.06 0.04
Age squared -0.00 -0.00
Married 0.05 -0.01
Number dependent children under 19 0.00NS -0.01
Degree 0.13 0.17
Education GCSE or less -0.14 -0.11
Tenure 0.01 0.01
Temporary job -0.12 -0.08
Part-time -0.05 -0.02
Public sector 0.02 0.06
Proportion female in occupation -0.29 -0.56
Proportion female squared 0.12 0.49
Proportion graduate in occupation 1.48 1.89
Proportion graduate squared -1.16 -1.35
Constant -3.75 -5.45
Observations 365,285 379,090
R-squared 0.47 0.53
Note: All p<0.001 except NS=non-significant; reference category for education=A-level
qualifications or equivalent
It was shown in Table 4.1 that the pay gap has reduced over the period 1993-2014
from 27% to just over 9%. This is the unadjusted gap, which could be explained by a
large number of factors, many of which have been reviewed and tested above.
These ‘explanations’ are not necessarily causes but they give an indication of the
variables associated with the pay gap. It is higher for part-time workers, for instance,
and lower for public sector workers. According to the data, reducing women’s
reliance on part-time work and increasing public sector employment would probably
reduce the pay gap, but the causality is more complex. Women working part-time
tend to have lower education and public sector workers to have higher education, for
example, although of course this is not the case for every individual.
In Table 5.3 we show figures for the adjusted and unadjusted gaps, in both cases
taking periods of five years (1993-97 and 2010-14). The unadjusted gap is simply the
effect of being female without controlling for education and so on. The adjusted pay
gap takes account of various characteristics, using the same controls as in Table
5.2.

Table 5.3: Gender pay gaps over time: unadjusted, adjusted and explained part
Explained part
Gender Gender as %
pay gap pay gap Explained part of unadjusted
(unadjusted) (adjusted) of pay gap pay gap
1993-1997 27.9 14.8 13.1 47.0
2010-2014 12.6 8.6 4.0 31.7
These show the familiar reduction in the gender pay gap over time. The unadjusted
gender pay gap closely matches Table 4.1, while in comparison the adjusted gender
pay gap shows a reduction for both time periods. This is the ‘residual’ – the part of
the pay gap that cannot be explained by differences in characteristics between men
and women – and could be in part the effect of discrimination, although we cannot
know that for sure. It could be smaller still if there were more, appropriate
explanatory variables in the data.
While the adjusted pay gap itself is smaller, 8.6% in 2010-2014, perhaps of greater
interest is that in the recent period (2010-2014) the ‘Explained part’ (the part of the
pay gap reduced through the use of additional variables), has reduced dramatically –
from nearly a half to just under a third of the unadjusted gender pay gap is explained.
Therefore the proportion of the overall gender pay gap that is ‘residual’ and
unexplained is more than two-thirds. This is a much greater proportion of the gender
pay gap than in 1993-97 when it only accounted for just over half.
The data suggests that as the gender pay gap is becoming smaller over time, the
differences in characteristics for which we have data are explaining less of it. This
could mean that simply being female, disregarding differences in measured
characteristics between men and women, explains an increasing proportion of the
pay gap. As women become less different to men, in terms of variables that explain
pay, the amount these differences influence the pay gap reduces. The fact that there
is a smaller gender pay gap now, but an increasing part of that is not explainable by
gendered differences in employment patterns, type of work, age, children, and so on
may suggest that approaches to reduce the impact of these on the pay gap are
having an effect. This could also suggest that the remaining ‘residual’ unexplained
difference in pay between men and women could be the persistence of
discrimination, if not all of it then at least some.

A further factor has to be taken into account. Some of the explanatory variables
might obscure various levels of discrimination that work indirectly. For instance,
women might choose to undertake various forms of education or enter certain
occupations which they consider more accessible to them. This selection effect
might in turn be related to their perceptions of the problems of entry into the ‘male
world’ (Bettio and Verashchagina, 2009).
5.2.1 Decomposition analysis

We now examine the pay gap directly through an explicit form of decomposition
analysis. Table 5.2 presented analysis of the wages of men and women separately,
where some inferences can be made about gender effects simply through comparing
the male and female coefficients.
Table 5.2 used the female wage as an explanatory variable in an analysis of all
wages but controlling for a wide range of factors. However, these do not model the
pay gap directly. This can be undertaken through decomposition analysis. This
method has its limitations. A substantial proportion of the pay gap remains
unexplained – typically over 50%. This used to be interpreted as a residual effect
denoting discrimination but this is no longer accepted as the gap can always be
reduced with more information, as with the analysis presented in Table 5.3. Further,
when we then look at the factors that contribute to or reduce the explained part these
effects are (in this case) mostly very small and we cannot know for certain how they
relate to the effects of these same variables on the unexplained part (Olsen et al.,
2010a).
Table 5.4 is based on the standard Blinder-Oaxaca decomposition method (pooling

all the years). This first decomposes the gap into a part which is explained by the
characteristics of women relative to men and a part not explained by these but where
structural factors (which could include discrimination) mean that these characteristics
have different effects. It then tells us which characteristics contribute to the explained
part of the pay gap.
The gap between men’s and women’s pay is 20%. Information not given in the table
indicates that only about 27% of the pay gap is explained by the difference in
characteristics of men and women. About 42% is unexplained, which could be the
result of discrimination or alternatively unexplained because we do not have enough
information on other factors which might contribute to the gap. The remaining 31% is
the result of a combination between the two.

From now on the analysis only focuses on what can be explained. Moreover, only
the most interesting outcomes are shown in Table 5.4 and in Figure 5.1, whether
large or small (no large effects are excluded from the table).
Table 5.4: Decomposition analysis of the explained part of the gender pay gap
Age -0.004
Graduate 0.000*
GCSE or less 0.001
Job tenure 0.013
Temporary job 0.001
Part-time 0.017
Public sector -0.011
Proportion female in occupation 0.020
Proportion graduate in occupation -0.008
Observations 744,335
Note: All p<0.001 except *=p <.05; other controls not shown.
Figure 5.1: Decomposition analysis of the gender pay gap
25%
20%
20%
15%
10%
5%
5%
2% 2% 1%
0% 0%
0%
0% -1% -1%
-5%

In the decomposition analysis it is clear that a large part of the pay gap is
unexplained by the data, and that even the most important drivers of pay have only a
small impact on the size of the pay gap.
Each year of age reduces the gap, which goes against the earlier discussion,
including Table 5.2, but age confounds many important factors such as career
effects and work experience. It is possible that when we control for some of these,
age does not by itself reduce women’s wages relative to those of men.
The effect appears small but is not. Every year of age reduces the pay gap by half of
one per cent. The effects of marriage and children are statistically significant but so
small that they are not given here, yet that in itself is an interesting finding.
Collectively this suggests that, were it not for other factors, women ought to be on a
more level footing with men or even possibly ahead of them in terms of pay, as they
age. The fact that they are not indicates other important issues are at play, such as
job tenure and work experience (see below). Obviously, as previously discussed, as
women age they are more likely to take time out of the labour market for child-
rearing purposes.
As shown in the previous table, being a graduate lifts pay for women more than it
does for men. Table 5.4 suggests, however, that it does not reduce the pay gap.
Women’s relatively low education explains a part of the gap as does their lower job
tenure (which can perhaps be treated as a partial proxy for work experience). This
conforms to earlier research, which finds some, if not a large effect (see, for
example: Manning and Swaffield, 2008). Career breaks reduce both job experience
and tenure in a specific job.
Temporary work has a very small effect on the pay gap; part-time work explains
much more of it. Working in the public sector reduces the pay gap. The fact that
women are concentrated in occupations with high levels of feminisation increases
the gap by a considerable amount, while working in graduate occupations reduces
this. The biggest effects are part-time work and occupational segregation. However,
as shown in Table 5.1, it is probably the case that high levels of feminisation reduce
the gap for graduates. This does not show up here because they are outnumbered
by non-graduates.
5.3 The problem of low pay
We finally return to the issue of low pay, which was discussed earlier in the report.
This is a variant of the pay gap argument insofar as it focuses attention on one part

of the wage distribution. Much of the analysis reviewed above shows a steadily
improving situation for women generally but with the caveat that this might not apply
to the lower part of the distribution. Put simply, women are more likely to be low paid
than men. If we define low pay as £8 per hour over the entire period (adjusted for
inflation), 51% of women were low paid in 1993 compared to 26% of men. By 2014
the figures had become 30% and 20% respectively. We now define low pay slightly
more precisely, and purely for purposes of this analysis, as being in the bottom 10%
of the pay distribution (but pooling years into two periods so that the 10% is the
group below this threshold before 2000 on the one hand and after 2000 on the
other). This shows that, taking both periods together, 8.4% of men and 15.8% of
women were low paid.
Figure 5.2 shows the probability of being low paid in the entire period on this
definition, based on the log of odds ratios. If a figure is positive it means this variable
raises the odds of being low paid, if negative it reduces it. Men and women are
analysed together with interaction terms between being female and (1) graduate
status, (2) having low or no qualifications, and (3) part-time work, to see if being
female intensifies their effects on the probability of being low-paid.

Figure 5.2: Log odds of being in the bottom 10% of the wage distribution
(logistic regression)
1.2
1
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
A woman is far more likely to be low paid than a man, controlling for other factors
such as education and part-time work in fact, this nearly doubles the odds.)
Occupational feminisation also raises the odds. Being a graduate reduces the odds,
far more for women than for men, but having low or no qualifications increases the
odds more for women. Finally, male part-time workers are extremely likely to be low
paid, even more so than female part-time workers.

The gender pay gap Conclusions
6 |Conclusions
This research report opened by looking at trends in the gender pay gap over time.
While the gap between average female and male pay remains significant, it has
been in continuous decline over the period. In 1993 women (part-time and full-time
combined) earned on average 73% of the male hourly wage; in 2014 the equivalent
figure was 90%. The gap is larger using the median rather than the mean but it has
still declined considerably, although it has narrowed in some regions more than in
others. The gender pay gap is smaller in London than elsewhere in the UK.
When we take the average pay gap within occupations we see that this too is in
favour of men. This means that women tend to receive less pay than men even
when working in the same occupation. This does not necessarily mean that pay
discrimination is happening. It could, for example, be the result of women being in
less senior positions.
However, the occupational gap is lower than the general pay gap. This means that
unequal pay within occupations explains only a small part of the general gap. A more
important factor is the disproportionately high number of women in poorly paid
occupations, many of whom work part-time. On average part-time positions do not
pay as well, per hour, as full-time ones. Women who work part-time earn less per
hour than women who work full-time; women who work full-time earn less per hour
than the equivalent men. However, part-time women generally earn more than part-
time men.
The pay gap is higher for older women, particularly for those in their forties. There
are a few main reasons for this. Firstly, older women have a lower level of education
on average than younger women. Secondly, the statistical analysis found that
women's shorter job tenure is a key factor in lowering their pay, relative to men's.
Women’s career progression does not therefore normally match men’s and women
are less likely to be promoted later in their careers.
Married men earn more than married women and the gender pay gap is far bigger
among married than unmarried people. The gap also rises with the number of
children but levels off at two children, so it is no wider for women with three or more
children than it is for those with two.

The gender pay gap Conclusions
Taking into account education, the gender pay gap is highest among non-graduates.
The gap is lower in the public sector than in the private sector and especially low – if
still substantial – for graduates in the public sector. In other words, although degree-
educated men maintain a pay advantage over their female peers, university
education has reduced the pay gap for women. The pay gap is therefore lower for
younger, graduate women.
Another factor, discussed in the introduction, is the role of occupational segregation,

whereby men and women are concentrated in separate occupations. This has
negative effects on wages for both men and women, because in general the better
paying occupations tend to be those in which there is a fairly even mix of men and
women. That said, women working in highly feminised sectors, especially if
graduates, experience the smallest sectoral pay gap. The same applies to non-
graduate women in highly feminised occupations, although this is likely to be
because wages are similarly low for both men and women.
Controlling for a wide range of characteristics and circumstances in a regression

analysis confirms the finding that higher education is an important way to address
the gender pay gap. Women have a higher return to having a degree – that is the
graduate premium relative to women with lower-level qualifications – than men
(again relative to men with lower qualifications). They also benefit more from working
in a graduate occupation.
Overall, the gender pay gap is exacerbated by the large numbers of women that
work part-time and the many women in highly feminised, low-paid occupations. It is
mitigated by the growing numbers of women who go to university. When it comes to
exploring policy solutions it is important to consider both ends of the pay spectrum.
This means tackling the ‘glass ceiling’ for high-achieving women but at the same
time ensuring that women, and men for that matter, are paid adequately and fairly for
less skilled work.

The gender pay gap Bibliography
Bibliography
American Association of University Women (2012), ‘The Simple Truth about the
Gender Pay Gap’
Becker, G. (1991), A Treatise on the Family. Cambridge, Mass.: Harvard University

Press
Bettio, F. (2002), ‘The pros and cons of occupational gender segregation in Europe’,
Canadian Public Policy, vol. 28, S1, pp. 65–84
Bettio, F. and Verashchagina, A. (2009), ‘Gender Segregation in the Labour Market:

Root Causes, Implications and Policy Responses in the EU’. European Commission.
Available at:
http://ec.europa.eu/social/main.jsp?catId=738&furtherPubs=yes&langId=en&pubId=3
64 [accessed: 1 September 2016]
Blau, F.D. and Kahn, L. M. (2003), ‘Understanding international differences in the

gender pay gap’, Journal of Labour Economics, vol. 21, no. 1, pp. 106–44
Booth, A. and Kee, H. J. (2011), ‘A long-run view of the university gender gap in
Australia’, Australian Economic History Review, vol. 51, no. 3, pp. 254-76
Brynin, M. and Perales, F. (2016), ‘Gender wage inequality: the de-gendering of the
occupational structure’, European Sociological Review, vol. 32, no. 1, pp. 162-74
Charles, M. and Grusky, D. (2004), Occupational Ghettos: the Worldwide

Segregation of Women and Men. Stanford: Stanford University Press
Cotter, D. A., Hermsen, J. M. and Vanneman, R. (2004), ‘Gender Inequality at Work’.

Russell Sage Foundation.
Department for Business, Innovation and Skills and the Equality and Human Rights
Commission (2015), ‘Pregnancy and Maternity-Related Discrimination and
Disadvantage. First findings: Surveys of Employers and Mothers’. Available at:
https://www.equalityhumanrights.com/en/managing-pregnancy-and-maternity-

workplace/pregnancy-and-maternity-discrimination-research-findings [accessed: 1
September 2016]
Dex, S., Ward, K. and Joshi, H. (2008), ‘Changes in women’s occupations and
occupational mobility over 25 years’, in Scott, J., Dex, S. and Joshi, H. (eds.),
Women and Employment: Changing Lives and New Challenges. Cheltenham:
Edward Elgar, pp. 54-80
Dolton, P., Marcenaro-Gutierrez, O. and Skalli, A. (2009), ‘Gender wage differentials

in Europe’, in Dolton, P., Asplund, R. and Barth, E. (eds.), Education and Inequality
across Europe. Cheltenham: Edward Elgar, pp. 192-219
England, P. (1992), Comparable Worth: Theories and Evidence. New York: Aldine
de Gruyter
England, P. (2005), ‘Emerging theories of care work’, Annual Review of Sociology,

vol. 31, pp. 381-99
England, P. (2010), ‘The gender revolution uneven and stalled’, Gender & Society,
vol. 24, no. 2, pp. 149-66
Esping-Andersen, G. (1993), Changing Classes: Stratification and Mobility in Post-

Industrial Societies. London: Sage
Goldin, C. (2008), ‘The rising (and then declining) significance of gender’, in Blau, F.,
Brinton, M. and Grusky, D. (eds.) The Declining Significance of Gender? New York:
Russell Sage Foundation, pp. 67–95
Grimshaw, D. and Rubery, J. (2007), ‘Undervaluing Women's Work’. Equal

Opportunities Commission Working Paper Series no. 53.Available at:
http://www.njl.nu/uploads/Jill_formiddag.pdf [accessed: 1 September 2016]
Hakim, C. (1998), Social Change and Innovation in the Labour Market: Evidence
from the Census SARs on Occupational Segregation and Labour Mobility, Part-time
Working and Student Jobs. Oxford: Oxford University Press
Harkness, S. E. (2005), ‘Pay Inequality in Gender’, in Delorenzi, S., Reed, J. and

Robinson, P. (eds.), Maintaining Momentum: Promoting Social Mobility and Life
Changes from Early Years to Adulthood. London: Institute for Public Policy
Research, pp. 93-103
Hegewisch, A., Liepmann, H., Hayes, J. and Hartmann, H. (2010), ‘Separate and not
equal? Gender segregation in the labor market and the gender wage gap’. Institute
for Women’s Policy Research, Briefing Paper IWPR C377. Available at:

http://www.iwpr.org/publications/pubs/separate-and-not-equal-gender-segregation-
in-the-labor-market-and-the-gender-wage-gap [accessed: 1 September 2016]
Jackson, R. (2008), ‘Opposing forces: how, why, and when will gender inequality
disappear?’, in Blau, F., Brinton, M. and Grusky, D. (eds.), The Declining
Significance of Gender? New York: Russell Sage Foundation, pp. 215–41
JNCHES (2011), ‘The Gender Pay Gap – a Literature Review’. New Joint
Negotiating Committee for Higher Education Staff. Available at:
http://www.ucea.ac.uk/en/publications/index.cfm/njgpygap [accessed: 1 September
2016]
Joshi, H., Makepeace, G. and Dolton, P. (2007), ‘More or less unequal? Evidence on
the pay of men and women from the British Birth Cohort Studies’, Gender, Work and
Organization, vol. 14, no. 1, pp. 37-55
Levanon, A., England, P. and Allison, P. (2009), ‘Occupational feminization and pay:
assessing causal dynamics using 1950-2000 Census data’, Social Forces, vol. 88,
no. 2, pp. 865–92
Joshi, A., Son, J. and Roh, H. (2015), ‘When Can Women Close the Gap? A Meta-
Analytic Test of Sex Differences in Performance and Rewards’, Academy of
Management, vol. 58, no. 5, pp. 1516-45
Lips, H. M. (2012), ‘The gender pay gap: challenging the rationalizations. Perceived
equity, discrimination, and the limits of human capital models’, Sex Roles, vol. 68,
no. 3-4, pp. 169-85
Magnusson, C. (2013), ‘More women, lower pay? Occupational sex composition,

wages and wage growth’, Acta Sociologica, vol. 56, no. 3, pp. 227-45
Manning, A. and Swaffield, J. (2008), ‘The gender gap in early-career wage growth’,
The Economic Journal, vol. 118, no. 530, pp. 983-1024
Metcalf, H. (2009), ‘Pay Gaps across the Equality Strands: a Review’. Equality and
Human Rights Commission Research Report no. 14. Available at:
https://www.equalityhumanrights.com/en/publication-download/research-report-14-
pay-gaps-across-equality-strands-review [accessed: 1 September 2016]
Metcalf, H. and Rolfe, H. (2009) ‘Employment and earnings in the finance sector: A
gender analysis’. Equality and Human Rights Commission Research Report no. 17.
Available at: https://www.equalityhumanrights.com/en/publication-

download/research-report-17-employment-and-earnings-finance-sector-gender-
analysis [accessed: 1 September 2016]
Mumford, K. and Smith, P. (2009), ‘What determines the part-time and gender
earnings gaps in Britain: evidence from the workplace’, Oxford Economic Papers,
vol. 61, Supplement 1, pp. 56-75
Ochsenfeld, F. (2014), ‘Why do women’s fields of study pay less? A test of

devaluation, human capital, and gender role theory’, European Sociological Review,
vol. 30, no. 4, pp. 536-48
ONS (2015), Annual Survey of Hours and Earnings: 2015 Provisional Results.
Available at:
http://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/earningsandworki
nghours/bulletins/annualsurveyofhoursandearnings/2015provisionalresults
[accessed: 1 September 2016]
ONS. 2013. Women in the labour market: 2013. [ONLINE]. Available at:
http://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/employmentande
mployeetypes/articles/womeninthelabourmarket/2013-09-25 [accessed: 1 September
2016]
Olsen, W. and Walby, S. (2004), ‘Modelling Gender Pay Gaps’. Equal Opportunities
Commission Working Paper Series no. 18.
Olsen, W., Gash, V., Vandecasteele, L., Walthery, P. and Heuvelman, H. (2010),
‘The Gender Pay Gap in the UK: 1995-2007: Part 1 – Research Report’.
Government Equalities Office
Olsen, W., Heuvelman, H., Gash, V. and Vandecasteele, L. (2010a), ‘The Gender
Pay Gap in the UK: 1995-2007: Part 2 – Policy Related Factors Offsetting Women’s
Low Pay in the UK, 2004-07’. Government Equalities Office. Available at:
http://webarchive.nationalarchives.gov.uk/20100505211508/http:/www.equalities.gov
.uk/pdf/301113_GEO_GenderPayGap_Part2_acc.pdf [accessed: 1 September 2016]
Perales, F. (2013), ‘Occupational feminization, specialized human capital and

wages: evidence from the British labour market’, Work, Employment and Society,
vol. 27, no. 4, pp. 600-20
Perfect, D. (2013), ‘Gender Pay Gaps, 2012’. Equality and Human Rights
Commission Briefing Paper no. 6 Available at:
https://www.equalityhumanrights.com/en/publication-download/briefing-paper-6-
gender-pay-gap-2012 [accessed: 1 September 2016]

Polachek, S. (1981), ‘Occupational self-selection: a human capital approach to sex

differences in occupational structure’, Reviews of Economics and Statistics, vol. 63,
no. 1, pp. 60–9
Rubery, J. (2008), ‘Women and work in the UK: the need for modernisation of labour
market institutions’, in Scott, J., Dex, S. and Joshi, H. (eds.), Women and
Employment: Changing Lives and New Challenges. Cheltenham: Edward Elgar, pp.
289-312
Rubery, J. and Grimshaw, D. (2014), ‘The 40-year pursuit of equal pay: a case of
constantly moving goalposts’, Cambridge Journal of Economics, vol. 39, no. 2, pp.
319-43
Stainback, K. and Tomaskovic-Devey, D. (2012), Documenting Desegregation:

Racial and Gender Segregation in Private-Sector Employment Since the Civil Rights
Act. New York: Russell Sage Foundation
Universities UK (2015), ‘Patterns and Trends in UK Higher Education’. Available at:

http://www.universitiesuk.ac.uk/facts-and-stats/data-and-analysis/Pages/patterns-
and-trends-uk-higher-education-2015.aspx [accessed: 1 September 2016]
Weichselbaumer, D. and Winter-Ebmer, R. (2005), ‘A meta-analysis of the

international gender wage gap’, Journal of Economic Surveys, vol. 19, no. 3, pp.
479-511

The gender pay gap Appendix
Appendix
Table A1: OLS regression results for men and women (full table)
Men Women
Age 0.06 0.04
Age squared -0.00 -0.00
Married 0.05 -0.01
Number dependent children under 19 0.00NS -0.01
Degree 0.13 0.17
Education GCSE or less -0.14 -0.11
Tenure 0.01 0.01
Temporary job -0.12 -0.08
Part-time -0.05 -0.02
Public sector 0.02 0.06
Industry=agriculture -0.09 -0.07
Industry=energy-water 0.11 0.15
Industry=mining 0.18 0.20
Industry=manufacturing 0.06 0.06
Industry=construction 0.08 0.07
Industry=distribution-hotels -0.05 -0.04
Industry=transport-communications 0.04 0.09
Industry=finance-property 0.09 0.13
Region=Tyne and Wear -0.14 -0.10
Region=Rest of Northern Region -0.12 -0.11
Region=South Yorkshire -0.13 -0.11
Region=West Yorkshire -0.11 -0.08
Region=Rest of Yorkshire and Humberside -0.11 -0.11
Region=East Midlands -0.10 -0.09
Region=East Anglia -0.09 -0.07
Region=Inner London 0.03 0.13
Region=Outer London 0.04 0.10

Region=South West -0.09 -0.08

Region=West Midlands (Met county) -0.11 -0.09
Region=Rest of West Midlands -0.10 -0.08
Region=Greater Manchester -0.12 -0.08
Region=Merseyside -0.12 -0.10
Region=Rest of North West -0.12 -0.09
Region=Wales -0.14 -0.10
Region=Strathclyde -0.10 -0.07
Region=Rest of Scotland -0.09 -0.07
Region=Northern Ireland -0.18 -0.11
Firm size 0.04 0.03
Year 0.00 0.00
Proportion female in occupation -0.29 -0.56
Proportion female squared 0.12 0.49
Proportion graduate in occupation 1.48 1.89
Proportion graduate squared -1.16 -1.35
Constant -3.75 -5.45
Observations 365,285 379,090
R-squared 0.47 0.53
Note: All p<0.001 except NS=non-significant; reference categories: education=school-leaving qualifications or

equivalent; industry=‘other services’; region=rest of South East.

Table A2 Decomposition analysis of the explained part of the gender pay gap
Explained Standard
part error
Age -0.004*** (0.001)
Age squared -0.001 (0.001)
Married 0.000 (0.000)
Number dependent children under 19 -0.000*** (0.000)
Degree 0.000* (0.000)
Education GCSE or less 0.001*** (0.000)
Tenure 0.013*** (0.000)
Temporary job 0.001*** (0.000)
Part-time 0.017*** (0.000)
Public sector -0.011*** (0.000)
Industry=agriculture -0.000*** (0.000)
Industry=energy-water 0.002*** (0.000)
Industry=mining 0.001*** (0.000)
Industry=manufacturing 0.006*** (0.000)
Industry=construction 0.007*** (0.000)
Industry=distribution-hotels 0.002*** (0.000)
Industry=transport-communications 0.006*** (0.000)
Industry=finance-property -0.002*** (0.000)
Region=Tyne and Wear -0.000 (0.000)
Region=Rest of Northern Region -0.000 (0.000)
Region=South Yorkshire -0.000 (0.000)
Region=West Yorkshire -0.000 (0.000)
Region=Rest of Yorkshire and Humberside -0.000 (0.000)
Region=East Midlands -0.000*** (0.000)
Region=East Anglia -0.000*** (0.000)
Region=Inner London -0.000*** (0.000)
Region=Outer London -0.000*** (0.000)
Region=South West -0.000 (0.000)
Region=West Midlands (Met county) -0.000*** (0.000)
Region=Rest of West Midlands -0.000*** (0.000)
Region=Greater Manchester 0.000 (0.000)
Region=Merseyside 0.000*** (0.000)
Region=Rest of North West 0.000 (0.000)

Region=Wales 0.000* (0.000)

Region=Strathclyde 0.000*** (0.000)
Region=Rest of Scotland 0.000*** (0.000)
Region=Northern Ireland 0.000 (0.000)
Firm size 0.005*** (0.000)
Year -0.002*** (0.000)
Proportion female in occupation 0.020*** (0.001)
Proportion graduate in occupation -0.008*** (0.000)
Total explained 0.053*** (0.001)
Pay gap 19.8%
Observations 744,335
Note: Figures rounded to three decimal places. *** p<0.001, ** p<0.01, * p<0.05
Standard errors in brackets.

The gender pay gap Contacts
Contacts
This publication and related equality and human rights resources are available from
the Commission’s website: www.equalityhumanrights.com.
For advice, information or guidance on equality, discrimination or human rights

issues, please contact the Equality Advisory and Support Service, a free and
independent service.
Website www.equalityadvisoryservice.com
Telephone 0808 800 0082
Textphone 0808 800 0084
Hours 09:00 to 19:00 (Monday to Friday)
10:00 to 14:00 (Saturday)
Post FREEPOST Equality Advisory Support Service FPN4431
Questions and comments regarding this publication may be addressed to:

correspondence@equalityhumanrights.com. The Commission welcomes your
feedback.
Alternative formats
This publication is also available as a Microsoft Word file from
www.equalityhumanrights.com. For information on accessing a Commission
publication in an alternative format, please contact:
correspondence@equalityhumanrights.com.

Published August 2017

You can download this publication from
www.equalityhumanrights.com

Published: August 2017

Research Report 109 The Gender Pay Gap PDF

Uploaded by

Copyright:

Available Formats

Research Report 109 The Gender Pay Gap PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Research Report 109 The Gender Pay Gap PDF

Uploaded by

Copyright:

Available Formats

What is the gender pay gap and how is it measured?

What is the gender pay gap and how is it measured?

What factors contribute to the gender pay gap?

What factors contribute to the gender pay gap?

Equality and Human Rights Commission

Research report 109 | Pay gaps research

Equality and Human Rights Commission Research Report Series

Post: Research Team

Equality and Human Rights Commission – www.equalityhumanrights.com

List of tables and figures 3

3 |The gender pay gap: what we already know 16

4 |Research findings: features of the gender pay gap 24

5 |Research findings: the drivers of the gender pay gap 44

Equality and Human Rights Commission – www.equalityhumanrights.com

List of tables and figures

Equality and Human Rights Commission – www.equalityhumanrights.com

Equality and Human Rights Commission – www.equalityhumanrights.com

List of abbreviations and acronyms

ASHE Annual Survey of Household Earnings

Equality and Human Rights Commission – www.equalityhumanrights.com

Equality and Human Rights Commission – www.equalityhumanrights.com

Equality and Human Rights Commission · www.equalityhumanrights.com

Equality and Human Rights Commission · www.equalityhumanrights.com

In 2015 the Equality and Human Rights Commission (‘the Commission’)

Equality and Human Rights Commission · www.equalityhumanrights.com

Equality and Human Rights Commission · www.equalityhumanrights.com

In 2015 the Equality and Human Rights Commission (‘the Commission’)

The main aims of the project were to:

Equality and Human Rights Commission · www.equalityhumanrights.com

Equality and Human Rights Commission · www.equalityhumanrights.com

A small amount of additional analysis is based on the UK Household Longitudinal

Equality and Human Rights Commission · www.equalityhumanrights.com

Equality and Human Rights Commission · www.equalityhumanrights.com

Equality and Human Rights Commission · www.equalityhumanrights.com

3 |The gender pay gap: what we already

3.1 Theories of the causes of the gender pay gap

3.1.1 Gender role theory

3.1.2 Human capital theory

Equality and Human Rights Commission · www.equalityhumanrights.com

Despite this progress, inequality persists. Access to education, as measured by

3.1.3 Occupational segregation

Segregation, however, is difficult to analyse because of different datasets and

Equality and Human Rights Commission · www.equalityhumanrights.com

Insofar as segregation is a problem, desegregation would seem to be the answer.

Equality and Human Rights Commission · www.equalityhumanrights.com

3.1.4 Undervaluation theory

Pay practices are ‘socially constructed’ and lead to undervaluation of women’s

Equality and Human Rights Commission · www.equalityhumanrights.com

3.2 Measuring and explaining the gender pay gap

3.2.1 The size of the pay gap and trends

An earlier Commission report (Metcalf, 2009) reviewed various studies based on

Equality and Human Rights Commission · www.equalityhumanrights.com

3.2.2 Factors linked with the gender pay gap

Equality and Human Rights Commission · www.equalityhumanrights.com

Equality and Human Rights Commission · www.equalityhumanrights.com

Equality and Human Rights Commission · www.equalityhumanrights.com

4 |Research findings: features of the

4.1 Patterns and trends in the gender pay gap

Difference in means Difference in medians