5. Gender Pay Gap

Question 1

Key questions

Answer 1

There has been a narrowing of gender pay gap

What explains the narrowing?
Why do the gaps still remain?
How to close the remaining gap?

Question 2

Relevant factors that influence gender pay gap

Answer 2

Main driver used to be education - women have now overtaken men in education

BUT today education is a minor factor

Relevant factors today:
Characteristics of the jobs women do (occupational and firm/ industry sorting
Discrimination and social norms, which affect the gender distribution of labour and means of detecting discrimination
Gender differences in attitudes and psychological attributes and non-cognitive skills

Question 3

Gender Pay Gap (Unadjusted Gender Pay Gap)

Answer 3

Comparing male to female workers earnings not accounting for differences in worker characteristics, such as age, experience, education, etc…

Question 4

Adjusted Gender Pay Gap

Answer 4

Gender pay gap calculated while accounting for underlying differences in worker characteristics (education, age, experience…)

Idea is to compare workers who are roughly similar concerning their jobs, tenure and education
Allows us to assess the extent to which these different factors contribute to the observed inequalities

Question 5

Convergence

Answer 5

(unadjusted) GPG decreased from the 1970s up until 2000 and then remained roughly stable
Unadjusted GPG captures differences across multiple dimensions between men and women —> if women catch up then GPG decreases

Education gap between men and women has been decreasing —–> since higher educated individuals earn higher wages —–> expectation is GPG would decrease if education of women increases

Question 6

GPG 1970-2016

Answer 6

Unadjusted GPG decreased in US over time
Adjusting for human capital —-> adjusted GPG also shrank in the 1980s but remained roughly the same afterwards

Human capital explained large part of GPG in 1970s and 1980s - BUT following decades —> human capital now plays less of a role

Adjusted GPG substantially shrunk in 1980s but remained roughly constant thereafter
Most of convergence happened in the 1980s - remaining GPG not explained by differences in human capital

Question 7

Adjusted GPG break down - 1980 - 2010

Answer 7

Experience and education less important in explaining GPG over time
Occupation and industry became more important over time
Unexplained GPG has been decreasing —-> good news because listed factors explain more of GPG in 2010 than in 1980

Does this mean that discrimination against women has gone down?
not necessarily - unexplained residual could still contain productivity factors which are hard to measure
-while explained factors could be capturing discrimination - unexplained residuals provide some information about what is going on but we need much more detailed data and analysis to pin down the role of discrimination in pay differences

Question 8

Factors explaining the persistence of GPG

Answer 8

Individuals’ jobs and the sectors they work in
Maternity
Individual characteristics - psychological traits and social norms
Discrimination and pay secrecy

These explanations are NON-EXCLUSIVE - can reinforce eachother

Question 9

Jobs - Occupation x industry

Answer 9

Jobs are a combination of:
1) Your occupation (what you do)
2) Your employer/ sector / industry (were you work)

SO: GPG can be explained by sorting:
a) occupational sorting: women working in different occupations than men
OR
b) sector / industry sorting: women working in the same occupation but for different employers

Can be a combination of the two
SO: if men work in higher paying occupations and for better paying employers - sorting will lead to a GPG

Question 10

Card, Cardoso and Kline - 2015

Answer 10

Employer sorting: firms differ in the pay premiums that they pay

If women are less likely to work at higher paying firms or negotiate worse wages than men —> GPG

Results:
20% variance of log wages are explained by firm effects and 11.4% by positive assortative matching (high ability workers match with higher profit firms)

Sorting effect: women are underrepresented at high-wage firms - explains 15% of overall wage gap in
Portugal

Bargaining: 5% explained by women gaining less from higher-wage firms

Importance of the two components vary by education and occupation groups
Combined sorting-bargaining effect can explain 20% of total GDP in Portugal

Question 11

The role of occupations - Goldin - 2014

Answer 11

Shows that women choose substantially different jobs than men that are compatible with childcare and other family responsibilities
-accounting for occupational sorting accounts for 30% of the earnings gap - so men work in different occupations than women
-majority of earning gap is explained within occupation (so women taking other jobs than men conditional on the same occupation)

The within-occupation gap is explained by job characteristics - mostly hours worked and career interruptions
-certain occupational features require fixed working times and reduce the degree of substitutability across workers
- conversely, jobs with higher flexibility offer lower pay even for the same amount of hours worked —> large GPG

Conclusion: reduce cost of flexibility by adapting several changes (tech, economies of scale, …)

Note: these changes can also be in favour of the firm to save cost

Question 12

Role of occupations - Goldin & Katz - 2016

Answer 12

Pharmacy became highly remunerated female-majority profession with small gender earnings gap and low earnings dispersion

This change coincides with major changes to the occupation:
- tech changes in the field made flexible jobs more productive
- growth of pharmacy employment in retail chains and hospitals
-decline of independent pharmacies

All of these changes contribute to narrowing of GPG - difference in hours explains majority of GPG - hourly wages almost the same

Question 13

Occupation and motherhood - Adda et al (2017)

Answer 13

Previous discussion shows that women value jobs with higher flexibility than men - this is due to care giving responsibilities (given their occupation)

Adda, Dustman and Stevens (2017) quantify the career cost of children focusing on how do intended family plans (children) affect the types of careers women choose (sorting into occupation)

They decompose the career cost of children into loss of skill during career interruptions and lost earning opportunities

Use a structural model to estimate these effects:
-allows them to compute counterfactual by studying the same individual and only changing a single parameter e.g. shutting down fertility
-comes at the cost of imposing more structure on the problem (more assumptions) - compared to reduced-form estimations

Results:
-shutting down the fertility of women (no children):
-women are more likely to choose “abstract-thinking” jobs and less likely to pick routine or manual occupations
-At any given age - women in the no-fertility scenario are more likely to work
-women are more likely to work full-time at any age
SO: experience is greater and so are wages

• career cost of children are calculated as the net-present value of lifetime income at the start of the career
• skill depreciation due to time out of the labour force accounts for 20% of lower wages
  -occupational choice at beginning of career accounts for 19% of cost (via wages)

OVERALL: career cost of children explains 1/3 of the gender wage gap

Question 14

The childhood / motherhood penalty - Kleven, Landais, Sogard , 2018

Answer 14

Motherhood penalty: women earn lower hourly earnings because they take lower paying jobs closer to home (due to flexibility)

Study uses Danish admin data from 1980 - 2013 to track men and women

findings:
-women’s earnings drop sharply after birth of first child and never fully recover
- no drop for men or women without children

Note: Denmark is a country that ranks highly on gender equality measures
-BUT despite family-friendly policies - gender pay gap still persists
SO: they are part of the solution but as long as women take on more care-giving responsibilities - inequalities will persist

Question 15

Individual characteristics, psychological traits and social norms

Answer 15

So far: differences in choices
What explains these differences in choices?
Why do women want job flexibility and take a disproportional amount of unpaid care work?

First instinct: differences in preferences and abilities
LIMITED EVIDENCE for this argument

In contrast: social norms and culture are key factors for understanding the gender differences in labour force participation and wages

Question 16

Blau and Kahn - 2017

Answer 16

There is evidence that men and women differ in some attributes that affect labour market outcomes:
-women found to be more risk averse and less willing to negotiate
- women more likely to engage in low-promotion tasks
-men more likely to compete and tend to be overconfident
- men value money to a greater extent

HOWEVER - these differences not all biological - they are results of social norms which affect upbringing - shaping preferences and skills from an early age
Hence: gender differences may still be malleable
Also depends on whether a certain trait is an advantage or not

Overall: these factors seem to contribute only modestly to explain GPG

Question 17

Social norms - Blau and Kahn - 2017

Answer 17

Men might be rewarded for showing a certain trait - women might have a negative response for showing the same trait

Possible explanation: social norms
-we expect certain behaviours from men and others from women - deviating from these norms may be penalised
-social norms shape preferences, behaviour and incentives to foster certain skills

Issue: social norms have large effect on today’s outcomes but they change very slowly —–> limited room for policy interventions

Question 18

Gender norms and marriage - Bertrand, Kamenica and Pan (2015)

Answer 18

The more likely (randomly chosen) woman is to out-earn a randomly chosen man - the lower the marriage rates

If the wife out-earns the husband - marriage satisfaction is lower
If woman’s income is likely to exceed her husband’s income (predicted income) - she is less likely to work- and if she works, she earns less (than predicted)

Evidence that women state a too low salary in survey’s if their income exceeds their husbands income

If a woman out-earns her husband - she also spends more time in non-market work -
HOWEVER: the strength of this social norm is falling

Question 19

Discrimination and pay secrecy

Answer 19

Residual in wage regression often considered to be a measure of discrimination (unexplained wage gap)

This is problematic because:
- some variables not observed by researcher might explain the difference but end up being attributed to discrimination

• some included variables might capture discrimination but we would attribute that to occupation - not discrimination

Hence, when doing field / lab experiment:
- advantage: estimate of discrimination less contaminated by unmeasured factors
- disadvantage: the results are hard to generalise beyond the specific setting

Fact: unexplained GPG exists - which can be attributed to discrimination
Problem: this gap is often not observable due to pay secrecy
Pay transparency legislation become popular to address this issue:
- firms have to produce pay reports detailing pay statistics
- pay reports need to be shared with employees
-sometimes pay reports require an audit or mandate taking action upon discovering unjustified differences

Reasoning: if employees finds out about unfair differences in their remuneration - they can challenge their employer on that basis demanding equal pay

Question 20

Pay transparency

Answer 20

Literature shows mixed evidence on effect of pay transparency:
- some papers find reduction in GPG which can be driven by male wages decrease or female wages increase
- other papers find no effect

How do we explain these differences:
-awareness of pay transparency, non compliance by firms
-bargaining: do employees have power to negotiate wages, how likely are people to negotiate wages
- information conveyed by statistics:
- do they add any additional info?
- do they allow for discovering discrimination?
Sometimes these statistics are even at firm level and not at occupation level - e.g. comparisons of cleaners vs managers not insightful

Labour market institutions / policy design:
- are firms required to act upon pay differences?
- what are the sanctions for non-compliance or continued large gaps?
- are these reports (exogenously) provided vs do workers have to request them?

Question 21

Strategies for closing GPG

Answer 21

Policies at detecting and fighting discrimination
Family friendly labour market policies
Structural change of jobs

These strategies could even have a positive feedback loop: family-friendly policies increase labour force participation and salaries for women —> increases returns to education investments —-> further generations of women are more likely to invest in education

Yet: these strategies only part of the solution —–> social norms and culture matter as well:
- changing norms and stereotypes that limit the choices available to both men and women imperative to achieve equal opportunities
- difficult and slow process but can be done