Gender Pay Gap - Faculty and Staff | Faculty of Artsfaculty.arts.ubc.ca/nfortin/Fortin... · 2016-06-07 · Earnings Inequality and the Gender Pay Gap Nicole Fortin Vancouver School

Earnings Inequality and theGender Pay Gap

Nicole Fortin Vancouver School of Economics andCanadian Institute for Advanced Research

1

State of the Art Lecture, CEA Meetings, June 4th 2016Click on Adobe Comment Tab to see Speaker’s Notes

Earnings Inequality and theGender Pay Gap

with the collaboration of Marie Drolet and Aneta Bonikowska

2

State of the Art Lecture, CEA Meetings, June 4th 2016

Presenter

Presentation Notes

Let me also thank my collaborators on this project, Marie Drolet and Aneta Boniskowska, from Statistics Canada, without whom this project would not have been feasible.

Gender Gap in Top Jobs

3

Presenter

Presentation Notes

At last year's CEA's, when Jean-Yes Duclos asks to deliver a state of the art address on the gender gap gap, little did he know at the time that he would become part of the first gender neutral cabinet, as Minister of Families, Children and Social Development

Gender Gap in Top Jobs

4

Presenter

Presentation Notes

Not only is this cabinet gender neutral, but likely has a favorable gender gap that is, if we include the person at the top

Earnings Inequality in Top Incomes

5

Presenter

Presentation Notes

Now the other important theme of this talk "earnings inequality in top incomes" has also become a hot political issue. Thanks to Bernie Sanders for picking up and graphically enhancing numbers for the work of Emmanual Saez and co-authors.

Earnings Inequality in Top Incomes

6Source: Veall (2012)

Presenter

Presentation Notes

(Slide 6 ) Indeed his Canadian co-author, Mike Veall made the topic of increasing inequality in top incomes the topic of is very successful presidential address in 2012. Although increases in the income share of the top 1% in Canada are lower than in the United States, they are nevertheless quite dramatic: the income share going to the top 1% was greater than 12% in 2010.

Increasing Earnings Inequality in Top Incomes and the Gender Pay Gap

• Apply the approach used in the analysis of earnings inequality in top incomes (developed by Thomas Piketty, Emmanuel Saez, and co-authors), as well as reweighing techniques à la DiNardo, Fortin and Lemieux (1996) [DFL] to the analysis of the gender pay gap

• Use all earnings data from the Canadian Longitudinal Worker Files (LWF, 1983-2010) supplemented by hourly wage data from the Labour Force Survey (LFS, 1997-2015)– Because couples file their income taxes separately, all earnings for T4

returns are available separately by gender in Canada

7

Presenter

Presentation Notes

So in the lecture today I'm gonna apply the approach use for the analysis of earnings inequality in top incomes, as well as reweighting techniques developped in DFL, to the analysis of the gender gap. To consider the earnings in top incomes we will appeal to Canadian longitudinal worker files, the LWF, but we will supplement with an analysis of the hourly wage data from more commonly used labour force survey

Increasing Earnings Inequality in Top Incomes and the Gender Pay Gap

• Questions of interest:1) What are the consequences of the under-representation of women in top

jobs for the overall gender pay gap?2) How is it contributing to the slowdown in the convergence of

female/male pay?3) What public policies and firm practices are effective to improve this

under-representation?

8

Presenter

Presentation Notes

I will focus on a few questions of interest: first what are the consequences of the under-representation of women in top jobs to the overall gender pay gap, second is this a factor in the slowdown of the convergence of male and female pay and third what kind of policies and firm practices would be effective to improve this under representation

Canadian DataLongitudinal Worker File (LWF)• LWF is a 10% random sample of all

Canadian workers• Years: 1983-2010• integrates data from the T1 and T4

files of Canada (CRA) and the LEAP (Statistics Canada)

• Annual earnings from all jobs, include bonuses, honorariums, etc.

• Selected if > half of minimum wage earnings equivalent

• Select workers age 25 to 64

Labour Force Survey (LFS) Public Use

• Monthly survey on approximately 100,000 individuals rotating 6-months panel sample design

• Years: 1997-2015• Hourly wage of employees from

main job• Selected if > half the minimum

wage• Select workers age 25 to 64

9

Presenter

Presentation Notes

Let me begin by giving you some details about the data that we will be using The longitudinal worker file that we are using is 10% random sample of all Canadian workers. We have access to the years 1983 to 2010 and the earnings data come from T1 and T4 slips therefore we have annual earnings from all jobs including bonuses. This contrasts with more traditional data sources used to study the gender gap which focuses on hourly wages on the main job, so there will be a bit of thinking outside the box. We select workers age 25 to 64 to avoid issues with students.

Canadian DataLongitudinal Worker File (LWF)

• No self-employment income• No labour supply information• Top coded at P99.99 ≈ $2,000,000

in 1983 to ≈$10,000,000 in 2000• Available covariates: union

coverage, age, industry• CPI adjusted to 2010$CAN

• No self-employment income• Number of weeks worked

unavailable• Top-coding (P99.9) from ≈$95/hour

in 1997 ≈ $125/hour in 2015 • At 2080 (=52wk*40hrs) hrs/year,

from $200,000 to $260,000• Available covariates: age, union,

education, occupation, industry, firm size, etc.

10

Labour Force Survey (LFS) Public Use

Presenter

Presentation Notes

The important disadvantage of the LWF is the absence of labour supply information, that being said even in the labor force survey the number of weeks work is unavailable. But an important limitation of the LFS survey is the fact that earnings or wage data is top coded percentile 99.9 but it has a wider set of the covariates that we take advantage off. Whereas in the LWF, we are only using industry at this point. All data are inflation adjusted using the CPI and in 2010 Canadian dollars.

Trends

1) Evolution of female/male labour force participationa) Extensive margin (LFP)b) Intensive margin (hours of work)

2) Evolution of female/male average wage and earnings ratios3) Evolution of female shares across top percentiles of the overall

distribution of wage and earnings4) Counterfactuals with alternative simulations

11

Presenter

Presentation Notes

To be sure that we are on the same page, I will begin by showing recent trends in different dimensions of gender gap issues First, we will look at the evolution of labour force participation, next at the evolution of wages differentials, and then I will focus on differences in the top percentiles and for the analysis, I will consider different sets of counterfactuals.

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1975 1979 1983 1987 1991 1995 1999 2003 2007 2011 2015

Part

icip

atio

n Ra

teCanadian Labour Force Participation Rate - Ages 25 to 64

Men Women

Steep Growth in Women’s Labour Force Participation* Followed by a Leveling-Off

Source: Fortin, Drolet and Bonikowska (2016), LFS Public use files, 1976-2015*Labour force participants include employed (at work or on-leave) and unemployed individuals 12

8.7

Presenter

Presentation Notes

The steep growth of women's labour force participation in the mid 20th century has been replaced in the 2000s by a substantial leveling-off, but nevertheless in Canada there has substantial convergence in labour force participation with male and female differences standing in the single digit.

Decline in LFP after the Great Recession in the US

13Source: Blau and Kahn (2016)

Presenter

Presentation Notes

This contrasts with the experience of thea U.S. where there has been actually a declining in participation for both genders following the great recession. The longer perspective in the U.S. graph (it starts in 1947) reminds us that this process of growth in female labor force participation has been longstanding with a long list of causes, including the introduction of washing machines.

The Women’s Liberation Movement of the 1960s and The “Pill”

• Goldin and Katz (2002) and Bailey (2006) point out to important changes in women’s LFP occurring in the 1960’s

• Women born after the mid-1950s had access to reliable contraception• More likely to pursue higher education and enter life-long careers• Accompanied by a decline in traditional gender roles attitudes which

stabilized in the mid-1990s in the U.S. (Fortin, 2015)• Before married women were more likely `secondary workers’ who entered

the labour market when kids were in school • Mulligan and Rubinstein (2013) argue that the closing of the gender pay

gap is largely due to changing selection of women into the labour market

14

Presenter

Presentation Notes

Nevertheless many think that the women's liberation movement of the sixties presented a shift in woman's attachment to the labor market, with many women switching from being secondary workers to career women.

Generational Effects in the Growth of Women’s LFP

Source: Fortin, Drolet and Bonikowska (2016), LFS public use files, ages 25 to 64 year 15

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1975 1979 1983 1987 1991 1995 1999 2003 2007 2011 2015

Part

icip

atio

n Ra

teWomen's Labour Force Participation by Synthetic Birth Cohort

<1920 1921-39 1940-45 1946-53 1954-58

1959-65 1966-75 1976-85 1985-90 All

Presenter

Presentation Notes

This implies that there are clear generational effects in the growth of female labour participation. The figure shows pre-Pill cohorts in the warmer colors, but later baby boomers in reens, and the generation Y in blue. As the pre-Pill cohorts reach retirement age, much of the growth in FLP coming from generational effects has been exhausted.

Generational Effects in the Growth of Women’s LFP

Source: Fortin, Drolet and Bonikowska (2016), LFS public use files, ages 25 to 64 year 16

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1975 1979 1983 1987 1991 1995 1999 2003 2007 2011 2015

Part

icip

atio

n Ra

teWomen's Labour Force Participation by Synthetic Birth Cohort

1940-45 1959-65 All

4432

54

3434

47

Presenter

Presentation Notes

This is perhaps more clearly seen when we contrast the life-cycle pattern of LFP of women born in the early forties with those born in early sixties. For the early forties women, there is a clear hump shape of participation reaching its around 45 years old. Whereas for those born in the early sixties, their pattern of LFP is much more constant over the life-cycle with perhaps a small drop of the peak of childbearing years.

Continued Gender Convergence?

• According to the Mincer-Polachek hypothesis (1974), gender differences in experience and labour force attachment are the key determinants of the gender wage gap.

• Blau and Kahn (2016) found that declining gender differences in experience in the United States accounted for 18-31 % of wage convergence between men and women over the 1980-2000 period.

• Going forward, Goldin (2014) suggested that the impact of work force interruptions for family responsibilities should be understood in the context of temporal flexibility (or the lack thereof) in impacting the gender wage gap.

17

Presenter

Presentation Notes

This shift in labour force attachment was that the origin of the Mincer-Polachek hypothesis where it was believed that as gender differences in labour market experience would decrease with more women adopting male life-cycle patterns of LFP, but so would the gender pay gap. Indeed Blaud and Kahn in a recent paper and found that these declines in gender differences labor market experience do account for a sizable part of the convergence between men and women's wages. More recently, Claudia Goldin has suggested that gender differences in labour force attachment have to be understood in the context of temporal flexibility.

Less Convergence in Gender Gap in Hours

18Source: Fortin, Drolet and Bonikowska (2016), LFS data, ages 25 to 64 year, employed with positive hours of work, usual hours from all jobs

0.60

0.65

0.70

0.75

0.80

0.85

0.90

0.95

1.00

1975 1979 1983 1987 1991 1995 1999 2003 2007 2011 2015

Gender Ratio in Average Total Weekly Hours by Synthetic Birth Cohort

<1920 1921-39 1940-45 1946-53 1954-58

1959-65 1966-75 1976-85 1985-90 All

Presenter

Presentation Notes

This implies that more atttention has to be paid to gender differences in hours of work. As this graph shows focusing only on those with positive hours, even for younger cohorts there is substantially less convergence between men and women in LFP at the intensive marging. The gender ratio in hours of work stands around 85%.

Gender Gap in Hours andIncreasing Earnings Inequality in Top Incomes

• Kuhn and Lozano (2008) had shown increases in long hours of work (>48 hours a week) among highly educated highly-paid older men was greatest in detailed occupations and industries with larger increases in residual wage inequality.

• Pointing to some high penalty for flexibility in some high wage occupations, Goldin (2014) further conjectures that rewards to working long hours are an obstacle for the gender gap in pay to vanish

• Cortes and Pan (2015) find that highly competitive jobs (O*NET characteristics) also have long hours

• Cortes and Pan (2016) find that across countries long hours lowers the share of married women in corresponding occupations

19

Presenter

Presentation Notes

Now the link between long hours of work and increasing inequality was first noted by Kuhn and Lozano for men. More recently Claudia Goldin has argued that rewards to working long hours presented another obstacle for further progress in the gender pay gap. Further, Cortes and Pan has found a link between highly competitive jobs, long hours, and the share of married women in such occupations.

Trends

1) Evolution of female/male labour force participationa) Extensive margin (LFP) : substantial convergenceb) Intensive margin (hours of work): less convergence

2) Evolution of female/male “raw” wage and earnings ratios: Continuing progress?

3) Evolution of female shares across top percentiles of the overall distribution of wage and earnings

4) Counterfactuals with alternative simulations

20

Presenter

Presentation Notes

To recap what we have seen so far. We have seen substantial convergence in male and female labour participation at the extensive margin, less so at the intensive margin. The next question is to what extent this has lead to converging in male and female pay. Do we have continuing progress?

What is the ratio of women’s to men’s earnings on average in Canada?

Source: Drolet (2011) 25 to 54 year olds, various data sources.21

Presenter

Presentation Notes

Before talking about the gender pay gap, it is important to see how the basis of computation has a large impact on the actual numbers. When computed on hourly wages for all workers, the gender ratio hovers around 85%; when we include the annual earnings of full-time workers we are more around 70%, and then when we use the annual earnings of all workers we are down to 65%.


• “Hourly Wage” ratio ≈ 85% is the preferred measure to consider whether employers treat women fairly and should be used in statements such as

“women earn 85 cents out of every $1 men earn”• “Annual Earnings” ratio ≈ 65% mixes the number of hours worked with

how much is earned by hour provides a better measure of the welfare of women

• More women work part-time, many women working full-time full-year work less hours a week than men (clerical vs. industrial workers)

• But the “All Annual Earnings” measure is the only one available for the very top income groups

22

Presenter

Presentation Notes

When we are considering issues of fairness, do employers treat men and women the same, the hourly wage ratio is the preferred measure. This is because the annual earnings ratio mixes the number of hours worked with which how much is earned by the hours. On the other hand, it does provide a better measure of the welfare of women and importantly for our exercise today, this annual earnings measure is the only one available for the very top income groups.


• The focus of all annual earnings can also be justified by the less favorable gender earnings ratios found by Frenette (2014) over the life-cycle

• In terms of the present value of total cumulative earnings (1991-2000) from the LWF (combined with the 1991 Census), they are even lower

PV Cumulative Earnings ratio ≈ 57% for university graduates and college graduates, ≈ 53% for high school graduates

23

Presenter

Presentation Notes

More recently, Frenette showed that in terms of the cumulative earnings ratio is where getting dangerously close to 50% range.

Generational Effects in the Gender Pay Gap

Source: Fortin, Drolet and Bonikowska (2016), LFS data, ages 25 to 64 year, hourly wage on the main job 24

0.60

0.65

0.70

0.75

0.80

0.85

0.90

0.95

1.00

1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

Gender Gap in Hourly Wages by Synthetic Birth Cohort

1935-39 1940-45 1946-53 1954-58 1959-651966-75 1976-85 1986-88 All

3131

41

Presenter

Presentation Notes

Considering generational effects again, we see that for all cohorts (the black line) there as indeed been progress the gender ratio in hourly wages moving from about 80 in 1997 to about 85 in 2010. However when we look the path for generation Y, we see more of a flat line if not a declining line. Therefore part of the overall progress is due to their retirement of those pre-Baby Boom cohorts, whose larger gender pay gap can be partly explained by lower level of experience. We see much less progress for the younger cohorts.

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009

Gender Gap in Annual Earnings by Synthetic Birth Cohort

1935-39 1940-45 1946-53 1954-581959-65 1966-75 1976-85 All

Generational Effects in the Gender Pay Gap

25Source: Fortin, Drolet and Bonikowska (2016), LWF data, ages 25 to 64 year, 3-year moving average annual earnings from all jobs

3131

41

Presenter

Presentation Notes

Indeed turning to the gender gap in annual earnings, we see a decline in the gender ratio for younger cohorts that might be signaling a failure to progress along the corporate ladder during childbearing years.

Slowdown of the Progress in Gender Pay Ratio

Longitudinal Worker File (LWF) Labour Force Survey (LFS)

26

Annual Earnings

Start End %Δ %Δ/ year

1983-1996 0.58 0.64 9.8 0.81997-2010 0.63 0.67 6.2 0.5

Hourly Wage

Start End %Δ %Δ/ year

1997-2010 0.81 0.85 5.0 0.42011-2015 0.86 0.85 -0.3 -0.1

Presenter

Presentation Notes

Looking more precisely at the numbers, both in the LWF and the LFS, we see a slowdown in the progress of the gender pay ratio to about half percentage point from 1997 to 2010. However the labor force survey shows perhaps negative trend since 2011.

Standard Decomposition of the Gender Pay Gap

• The Oaxaca-Blinder decomposition starts with gender-specific OLS regressions of individual characteristics on (log) wages:

𝑌𝑌𝑔𝑔 = 𝑋𝑋𝑔𝑔′𝛽𝛽𝑔𝑔 + 𝜀𝜀𝑔𝑔, g = m, f• Constructs a counterfactual wage such as “what would be the average

wage of women if they had the same characteristics as men” 𝑌𝑌𝑓𝑓𝑚𝑚 = 𝑋𝑋𝑚𝑚′ 𝛽𝛽𝑓𝑓 = 𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑚𝑚 × 𝑝𝑝𝑝𝑝𝑞𝑞𝑝𝑝𝑞𝑞𝑓𝑓

• Divides the average gender pay gap into “explained” and “unexplained” part �𝑌𝑌𝑚𝑚 − �𝑌𝑌𝑓𝑓 = (𝑌𝑌𝑓𝑓𝑚𝑚 − �𝑌𝑌𝑓𝑓) + (�𝑌𝑌𝑚𝑚−𝑌𝑌𝑓𝑓𝑚𝑚) = ( 𝑋𝑋𝑚𝑚′ − 𝑋𝑋𝑓𝑓′)𝛽𝛽𝑓𝑓+ 𝑋𝑋𝑚𝑚′ (𝛽𝛽𝑚𝑚 − 𝛽𝛽𝑓𝑓)

explained unexplained

27

Presenter

Presentation Notes

So far we have considered in the "raw" gender gap, but what about the adjusted gap, that is the gap adjusted for differences gender differences in characteristics. The traditional Oaxaca-Blinder decomposition starts by constructing a counterfactual wage. The one that we emphasize here is "what would be the average wage of women be if they have the same characteristics as men" it corresponds to a situation where women would have been improving their characteristics such as educational attainment. It is computed as male quantities evaluated at the female prices. Using this counterfactual wage, the standard decomposition divides the difference in male and female wages into the differences in average male and female characteristics evaluated valued at female prices: this corresponds to the explained component of the gender gap while the differences in male and female returns evaluated at the male average characteristics correspond to the unexplained part.

Gender Pay Gap Largely “Unexplained” by Human Capital Variables

• For the United States, Blau and Kahn (2016) using human capital variables, including actual experience from the PSID, find a notable decline in the unexplained gap—from 0.341 log points in 1980 to 0.197 log points in 2010.

• But as a share of the gender gap in both years, the unexplained portion is actually a larger share of gap in 2010 (85%) than in 1980 (71%).

• For Canada, Baker and Drolet (2010) also report some progress in the unexplained gap from 0.163 log points in 1981 to 0.141 log points in 2008.

• But this represents an increase, from 1981 (61%) to 2008 (85%), in the share of gap that is unexplained by education, occupation and industry.

28

Presenter

Presentation Notes

What is a source of concern in this decomposition is the unexplained part, that is, the part that arises from differences in the wage structure that men and women face. Blau and Kahn in a recent paper haveI found using human capital variables, that include actual experience, that there has been a declining in the unexplained gap. However as a share of the gender gap, they find that the unexplained portion is actually increasing over time, 85% in 2010. For Canada, Baker and Drolet report results along the same line; they also find that the share of the gap that remains unexplained by education, occupation, and industry is also around 85% in 2008.

29Source: Blau and Kahn (2016)

Presenter

Presentation Notes

They argue that this is because over time education in particular has become a countervailing explanatory variable, that is women should be paid more give in the higher level of education.

Gender Pay Gap Largely “Unexplained” by Human Capital Variables

• Baker and Drolet (2010) explain that in many dimensions, such as education, women increasingly have an advantage over men.

• But because women’s wages have not seen commensurate increases, these are countervailing factors to explain the gap.

• They argue that most significant exception to this is the industrialdistribution of employment in which men maintain a significant advantage.

30

Presenter

Presentation Notes

They also argue that the industrial distribution of employment, remains all one of the characteristic in which men maintain a significant advantage.

Industry Composition Largest Single Explanatory Factor* in the Private Sector

• Yet, Schirle(2015) finds that in most provinces more than half of the gap is unexplained

31Source: Schirle (2015), LFS 2014, hourly wages of private sector full-time employees, ages 25-59* with the exception of Manitoba

Presenter

Presentation Notes

Indeed, Schirle who considered the gender gap at the provincial level, found that in 2014 industrial composition was the largest single explanatory factor. However most provinces a very substantial share the gap remained unexplained.

Trends


2) Evolution of female/male average wage and earnings ratios: Slower progress in recent years, share of the gap unexplained has increased, industry may remain a potent explanatory variable

3) Evolution of female shares across top percentiles of the overall distribution of wage and earnings

4) Counterfactuals with alternative simulations

32

Presenter

Presentation Notes

To summarize what we've seen in terms of the gender gap: we found a slowdown in the decline of the gender gap, an increase in the share of the gender gap there remains unexplained, although industrial composition may still have the explanatory power.

Increasing Earnings Inequality and the Gender Pay Gap

• When residual inequality experienced stupendous increases in the 1980s, Blau and Kahn (1997) coined the term “swimming upstream” to characterize women’s pursuit of pay equality in the face of countervailing currents.

• Have recent increases in top incomes lead to similar effects, therefore accounting for the slower progress in the gender pay and growing unexplained (by traditional factors) share?

• To the extent that some of the increases in top incomes are associated with excesses in rent seeking, curtailing those excesses would slow the countervailing currents

33

Presenter

Presentation Notes

We now turn to increasing top income inequality as the potential explanatory factor for this slower progress and growing unexplained share. It is important to mention that when residual inequality were increasing stupendously in the 1980s, Blau and Kahn coined the term "swimming upstream" to characterize women's pursuit of pay equality in the face of this countervailing current. So what we're asking is whether the recent increases in top income inequality could be leading to similar countervailing effects. Naturally we could also question the sources of the stunning increases in top incomes and ask whether all these countervailing currents can be slowdown.

Soaring Top Incomes in the United States

•

25%

30%

35%

40%

45%

50%

1917

1922

1927

1932

1937

1942

1947

1952

1957

1962

1967

1972

1977

1982

1987

1992

1997

2002

2007

2012

Top

10%

Inco

me

Shar

eTop 10% Pre-tax Income Share in the US, 1917-2013

Source: Piketty and Saez, 2003 updated to 2013. Series based on pre-tax cash market income including realized capital gains and excluding government transfers. 34

Presenter

Presentation Notes

The approach used by Emmanuel Saez to characterize the rise in top income inequality has been to display trends in the top 10% income share for example, as this picture showing that the share of the top 10% income almost reached 50% in 2012.

Mostly among the top 1%

0%

5%

10%

15%

20%

25%

1913

1918

1923

1928

1933

1938

1943

1948

1953

1958

1963

1968

1973

1978

1983

1988

1993

1998

2003

2008

2013Sh

are

of to

tal i

ncom

e fo

r eac

h gr

oup

Decomposing Top 10% into 3 Groups, 1913-2013

Top 1% (incomes above $392,000 in 2013)Top 5-1% (incomes between $165,500 and $392,000)Top 10-5% (incomes between $116,500 and $165,500)

Source: Piketty and Saez, 2003 updated to 2013. Series based on pre-tax cash market income including realized capital gains and excluding government transfers.

35

Presenter

Presentation Notes

He has also are focused on finer top income groups to show that the rise is highly concentrated among the very top. In this particular picture, the top-ten percent is divided in three groups, the top 10 to 5, the 5 to 1 and the top one, and most of the increase is seen in the top 1%.

Gender Gap in Top Incomes

• Follow Guvenen, Kaplan, and Song (2014) in using the thresholds of the wage and earnings distribution for men and women combined

• Depart from the traditional literature on the glass ceiling which compares the pay gap at percentiles of the gender-specific distributions

• Depart from most of the literature which uses the logarithm of wages or earnings in order to emphasize the top end

• Allow for the construction of counterfactuals to study the under-representation of women in top income groups

• Study the role of industrial segregation within income groups

36

Presenter

Presentation Notes

We will follow a similar methodology to the one of Guvenen, Kaplan and Song who use thresholds of the wage and earnings distribution of men and women combined. So in that way, we're doing now departed significantly from the traditional literature looking at the glass ceiling effect which usually compares the pay gap at percentiles of the gender-specific distribution. We will depart from the traditional literature use the logarithm of wages and use levels instead. This is needed in order to emphasize the top-end rather then using the logarithm which compress the top end. We construct counterfactual that will simulate alternative representation of women in top income groups and we will also study the impact of industrial segregation.

Thresholds of Top Incomes - 2010Longitudinal Worker File (LWF)

Annual Earnings for all jobs1) Top 0.1% > $662,8602) Top 1% > $206,7853) Top 10% > $92,000

Labour Force Survey (LFS)

Hourly wages on the main job1) Top 0.1% > $66 ( $128,705) 2) Top 1% >$53 ($116,922)3) Top 10% >$35 ($80,352)

at 2080 hours

37

Presenter

Presentation Notes

We use the following 3 thresholds for top income earners: the top 10%, the top 1%, and the top 0.1%. Now because of a lower top coding in the LFS, because of bonuses, because of all longer hours in top jobs, the annualized LFS thresholds do not correspond the LWF thresholds. So you should think of the thresholds for the different data source at quite different.

Larger Increases for Top Incomes!

300,000

500,000

700,000

900,000

1,100,000

1,300,000

1,500,000

1,700,000

1,900,000

2,100,000

0

50000

100000

150000

200000

250000

300000

350000

400000

450000

500000

1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009

Canadian All Earnings Trends

Bottom 90% Next 9% Next 0.9% Top 0.1% (right axis)

38Source: Fortin, Drolet and Bonikowska (2016), LWF 1983-2010, 25-64 years old, Annual earnings from all jobs

Presenter

Presentation Notes

Using these thresholds allows us to divide the earners into four distinct groups that add up: the bottom 90%, the next 9%, the next 0.9% and the top 0.1%. What is clear in this picture and what is well known is that there has been much higher increases for the top-end than in the bottom.

Larger Increases for Top Earners!

300,000

500,000

700,000

900,000

1,100,000

1,300,000

1,500,000

1,700,000

1,900,000

2,100,000

0

50000

100000

150000

200000

250000

300000

350000

400000

450000

500000

1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009

Canadian All Earnings Trends

Bottom 90% Next 9% Next 0.9% Top 0.1% (right axis)

162%

34%

81%

15%

39Source: Fortin, Drolet and Bonikowska (2016), LWF 1983-2010, 25-64 years old, Annual earnings from all jobs

Presenter

Presentation Notes

From 1982 to 2010, the top 0.1% has seen it earnings increase by 162%, the next 0.9 saw just half of that increase at 81%, the next nine say increaes of only 34%, and the bottom 90% only 15%.

Source: Lemieux and Riddell (2015), LAD data 40

Presenter

Presentation Notes

These numbers display an even more unequal growth picture than the one reported by Lemieux and Riddell (2014) using weekly earnings data from the Census.

Gender Differences in Hourly Wage Distributions

41Source: Fortin, Drolet and Bonikowska (2016), LFS 1997-2015, 25-64 years old, Hourly wage from the main job

Top 10% Top 1% Top 0.1%

0.0

1.0

2.0

3.0

4.0

5D

ensi

ty

25 50 75 100 125Hourly Wage ($2010)

Women Men

B. 2011-2015Top 10% Top 1% Top 0.1%

0.0

1.0

2.0

3.0

4.0

5D

ensi

ty

25 50 75 100 125Hourly Wage ($2010)

Women Men

A. 1997-2002

Presenter

Presentation Notes

Another way to display increases in top incomes is to show how the thresholds for the top 1% and the top 0.1% have move further down in the right skew from the early 1997-2002 periods to the last five years. This picture clearly shows the top 1% and 0.1% pulled away from the bottom 90%.

Source: Fortin, Drolet and Bonikowska (2016) Computation, LFS 1997-2015, 25-64 years old, Hourly wage from the main job

Gender Differences in Hourly Wage Distributions

42

Top 10% Top 1% Top 0.1%

0.0

1.0

2.0

3.0

4.0

5D

ensi

ty

25 50 75 100 125Hourly Wage ($2010)

Women Men

B. 2011-2015Top 10% Top 1% Top 0.1%

0.0

1.0

2.0

3.0

4.0

5D

ensi

ty

25 50 75 100 125Hourly Wage ($2010)

Women Men

A. 1997-2002

3:2

4:18:1

Top 0.1%Top 10% Top 1%

Presenter

Presentation Notes

But importantly, the picture show the under-representation of women in top percentiles. At the 90%, the ratio of men to women 3:2. and the top 1% it is 4:1 and at the top 0.1% it is 8:1 of them.

Slower Convergence in Share of Women among Top Earners

43

0.00

0.10

0.20

0.30

0.40

0.50

0.60

1997 1999 2001 2003 2005 2007 2009 2011 2013 2015

Share of Women in Selected Percentiles of Hourly Wages

All Bottom 90% Next 9% Next 0.9% Top 0.1%

Source: Fortin, Drolet and Bonikowska (2016), LFS 1997-2015, 25-64 years old, Hourly wages from the main job

Presenter

Presentation Notes

This is best illustrated by showing the evolution of shares of women in our four income groups. In the LFS, among the bottom 90% women comprise more than 50% of workers. whereas let's say in the top are in the next 9%, they make up less than 40% of workers. Even though there's been substantial increases over the period, in 2015 women represent less than 20% of the top 0.1% (in the LFS, these are workers making more than $130,000 in 2010).

Slower Convergence in Share of Womenamong Top Earners

0.00

0.10

0.20

0.30

0.40

0.50

0.60

1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009

Share of Women in Selected Percentiles of Annual Earnings


Source: Fortin, Drolet and Bonikowska (2016), LWF 1983-2010, 25-64 years old, Annual earnings from all jobs

44

Presenter

Presentation Notes

In the LWF where we capture more high earners, the share of women in the top 0.1% (those making more than $600,000 in 2010) is about 10%, while the share of women in the next 0.9% is around 16% and about 26% in the next 9%.

Under-representation of women in top jobs makes for a less favorable overall gender pay ratio

Source: Fortin, Drolet and Bonikowska (2016), LFS 1997-2015, 25-64 years old, Hourly wages from the main job45

0.50

0.60

0.70

0.80

0.90

1.00

1.10

1997 1999 2001 2003 2005 2007 2009 2011 2013 2015

Female-Male Average Hourly Wages Ratios by Selected Percentiles

All Bottom 90% Next 9% Next 0.9% Top 0.1% Median

Presenter

Presentation Notes

Now this underrepresentation of women in top job makes for a less favorable overall gender ratio as shown in the figure the gender ratio in hourly wages in the bottom 90% is around 90% while for the entire distribution, it's more around 85%, that is a 8 percentage points difference.

Under-representation of women in top jobs slows progress in the overall gender pay ratio


0.40

0.50

0.60

0.70

0.80

0.90

1.00

1.10

1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009

Female-Male Earnings Ratios by Earnings Percentile


0.10

46

0.15

Presenter

Presentation Notes

Moreover when looking at the all annual earnings gender ratios, these slow entry of women in the top centiles makes for smaller progress in the gender pay gap in annual earnings. Whereas over the 1983 to 2010 period, there was an increase of 15 are percent in the gender ratio of all earnings in the bottom 90%, that increase for the overall distribution was only 10%.

Trends


2) Evolution of female/male average wage and earnings ratios:3) Evolution of female shares across top percentiles of the overall

distribution of wage and earnings4) Counterfactuals with alternative simulations

a) Using male shares in selected percentile earningsb) Using male industrial distribution

47

Presenter

Presentation Notes

Now let's get to the counterfactuals to better quantify the consequences of these low female shares in top percentiles. We will conduct two alternates simulations: one where we will give the male shares in our selected percentiles to women and another one where we will give to women of the male industrial distribution.

Counterfactual Gender Pay Gaps and Reweighting

• Kline (2011) shows that the counterfactual (letting 𝐷𝐷𝑖𝑖 = 1 denote male),𝜇𝜇01 = 𝐸𝐸[𝑋𝑋𝑖𝑖�𝐷𝐷𝑖𝑖 = 1]′𝛽𝛽0

can be computed from an OB regression𝜇𝜇01 = 𝐸𝐸[𝑋𝑋𝑖𝑖|𝐷𝐷𝑖𝑖 = 1]′ × 𝐸𝐸[𝑋𝑋𝑖𝑖𝑋𝑋𝑖𝑖′|𝐷𝐷𝑖𝑖 = 0] −1𝐸𝐸[𝑋𝑋𝑖𝑖𝑋𝑋𝑖𝑖′|𝐷𝐷𝑖𝑖 = 0]

• Or using reweighting à la DFL

𝜇𝜇01 = 𝐸𝐸[𝑤𝑤(𝑋𝑋𝑖𝑖)𝑌𝑌𝑖𝑖|𝐷𝐷𝑖𝑖 = 0] where 𝑤𝑤(𝑋𝑋𝑖𝑖) ≡P( |𝑋𝑋𝑖𝑖 𝐷𝐷𝑖𝑖=1)P( |𝑋𝑋𝑖𝑖 𝐷𝐷𝑖𝑖=0) = 1−𝜋𝜋

𝜋𝜋𝑒𝑒(𝑋𝑋𝑖𝑖)

(1−𝑒𝑒(𝑋𝑋𝑖𝑖))with 𝜋𝜋 ≡ P(𝐷𝐷𝑖𝑖 = 1) and 𝑞𝑞(𝑋𝑋𝑖𝑖)= P( |𝐷𝐷𝑖𝑖 = 1 𝑋𝑋𝑖𝑖), under the assumptions of common support 𝑞𝑞(𝑋𝑋𝑖𝑖)< 1 and conditional independence (𝑌𝑌𝑖𝑖1,𝑌𝑌𝑖𝑖0) ⊥ |𝐷𝐷𝑖𝑖 𝑋𝑋𝑖𝑖

48

Presenter

Presentation Notes

In these large samples, we will construct these counterfactuals using reweighting. It is important to note that Pat Kline has explicited shown the that our counterfactual of choice, the average earnings of women if they had the male characteristics, can be computed either from a regression framework as in Oaxaca-Blinder or using DFL reweighting, where the idea is to replace the female weights by the male weights in the computation of the average. This correspondence is a valid under the common support and the conditional independence assumption. I raise the commons support assumption, because it can indeed became an issue in some of our top income groups.

Counterfactual Gender Pay Gaps and Reweighting

• The sample analogues are: 𝜋𝜋 = 𝑁𝑁1𝑁𝑁

and 1−𝜋𝜋𝜋𝜋

= 𝑁𝑁0𝑁𝑁1

• If 𝑋𝑋𝑖𝑖 is a j-category variable, 𝑞𝑞(𝑋𝑋𝑖𝑖𝑗𝑗) = 𝑁𝑁1𝑗𝑗𝑁𝑁𝑗𝑗

and 𝑒𝑒(𝑋𝑋𝑖𝑖)(1−𝑒𝑒(𝑋𝑋𝑖𝑖))

= 𝑁𝑁1𝑗𝑗𝑁𝑁0𝑗𝑗

,

• So that reweighing observations requires only the ratio of shares in each j-category: 𝑤𝑤(𝑋𝑋𝑖𝑖𝑗𝑗)=𝑁𝑁0

𝑁𝑁1∗ 𝑁𝑁1𝑗𝑗𝑁𝑁0𝑗𝑗

= 𝑆𝑆1𝑗𝑗𝑆𝑆0𝑗𝑗

where 𝑆𝑆1𝑗𝑗 is the share of group 1 in category j • With conditional means, the overall mean is �𝑌𝑌0=∑𝑖𝑖 𝑆𝑆0𝑗𝑗𝑌𝑌0𝑗𝑗, so that

𝑌𝑌𝑜𝑜1 = ∑𝑗𝑗 𝑆𝑆1𝑗𝑗 ∑𝑖𝑖𝑆𝑆0𝑗𝑗𝑆𝑆0𝑗𝑗𝑌𝑌0𝑗𝑗 = ∑𝑗𝑗 𝑆𝑆1𝑗𝑗 ∑𝑖𝑖 𝑌𝑌0𝑗𝑗

49

Presenter

Presentation Notes

Translating of these population means formulas in sample analogs, we get the weights will correspond to the ratio of male share over female shares in the case of category-j of a categorical variable. In our case, the categorical of the variable of interest will be our 4 percentile groupings, as well as industry. Once we have computed the categorical means, for example the male categorical means, all we have to do to obtain our counterfactual is to multiply by the female shares .

If the proportion of women across professorial ranks was identical to men, the overall counterfactual average female salary would be:

51.8/100×146048 + 30.7/100×114595 + 17.6/100×99709 =128259.3, and the overall ratio would be 128382/134955(*100)=95% The salary gap explained by rank is 128259.3-120623.1 =7636.2 More that 53% of the gap is accounted for by the gender differences in the proportion of

faculty members across rank.

Female/ GenderMale Ratio Gap

Men All 968 100 134955 0.89 14332Women All 419 100 30.2 120623 Men Full 501 51.8 152494 0.96 6446Women Full 130 31 20.6 146048Men Associate 297 30.7 121483 0.94 6888Women Associate 184 43.9 38.3 114595Men Assistant 170 17.6 106806 0.93 7097Women Assistant 105 25.1 38.2 99709

Gender Rank Numbers % of All % of women

Average Salary

Table 1. Average Professorial Salaries at UBC in 2010

Presenter

Presentation Notes

Let me give you an example that will be on familiar, that is one that uses of the salaries of professors, and even close to home the case of professors at UBC are in 2010. The overall female-male ratio was of a less than 90%, whereas the gender ratio in each professioriall rank, that is among Full, Associate, and Assistant Professors was closer to 95%. The proportion of male and female across the ranks was very different with 52% of men in the Full category vs. 31% of women, and correspondingly higher proportion of women at the Assistant level. So if we simply multiply the average female salary by the male shares, we obtain the average counterfactual salary that women would have earned if they weren't distributed across ranks as men are. Now this leads to an overall ratio of 95% and an explained gap of more than $7000 or 53 percent of the gap.

51

Variables: Model 1 % of gap Model 2 % of gap

Raw Gender Salary Differentials 14332.24 *** 14332.24 ***Accounted for by differences in characteristicsProfessorial Rank 7636.226 *** 53.28% 6647.376 *** 46.38%CRC, DUP 546.2663 * 3.81%Years in Rank 1180.126 ** 8.23%Departmental Dummies 3093.223 ** 21.58%Total Explained 7636.226 *** 53.28% 11466.99 *** 80.01%Total Unexplained 6696.018 *** 46.72% 2865.253 *** 19.99%

Table 2. Oaxaca-Blinder Decomposition of Average Professorial Salaries at UBC in 2010

Note: Using female coefficients. *** p<0.01, ** p<0.05, * p<0.1 See UBC (2011) for alternative specifications.

The more complete specification accounts for 80% of the gap, 46% of which from vertical segregation and 22% from horizontal segregation.

This leaves an unexplained gender gap of 2.2% of average professorial salary

Presenter

Presentation Notes

Now how does this measure of vertical segregation compares with other measures for example of horizontal segregation are captured by departmental dummies. We can turn to the familiar OB decomposition and we find that professorial rank still explain 46% of the gap while departmental dummies accoutn for 22% of gap. So it is a quite successful decomposition, with only 20% of the gap remaining unexplained.

If the shares of women in percentiles grouping* were the same as men’s, the gap in annual earnings would be 20 point lower

0.40

0.50

0.60

0.70

0.80

0.90

1.00

1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009

Female-Male Earnings Ratios by Earnings Percentile (All Earners)

Simulated Ratio Actual Ratio

0.42

0.19 = 45%

*percentiles grouping: bottom 90%, next 9%, next 0.9%, top 0.1% 52

0.33

0.19 = 58%


Presenter

Presentation Notes

(Slide 52) When we apply the same reweighting techniques to the average of all the annual earnings from the LWF, we find that the gender ratio would be almost 20 points lower if women had the same distribution across for all four percentiles groupings as men. Interestingly, this correspond to a growing portion of the gap. For example in 2010, 58% of the 33 percentage points gap would be explained by this measure of vertical segregation.

If the shares of women in percentiles grouping* were the same as men’s, the gap would be 6-9 points lower

Source: Fortin, Drolet and Bonikowska (2016), LFS 1997-2015, 25-64 years old, Hourly wages on the main job*percentiles grouping: bottom 90%, next 9%, next 0.9%, top 0.1% 53

0.70

0.75

0.80

0.85

0.90

0.95

1.00

1997 1999 2001 2003 2005 2007 2009 2011 2013 2015

Counterfactual Hourly Wage Ratio Substituting Male Shares in the Selected Wage Percentiles

Actual Ratio Simulated Ratio

0.07 = 45%

0.19

0.08 = 41%

0.15

Presenter

Presentation Notes

For the hourly wage ratio , the reweighting with male shares in percentile groupings brings us in the lower 90%. Although the number are less dramatic than for the all earnings ratio over this time period, consistent withs increasing top-end income inequality, the percentiles grouping counterfactuals explain a growing portion of the gap.

Explanatory Variables Model 1 % of gap

Model 1 % of gap

Model 2 % of gap

Model 2 % of gap

1997 2015 1997 2015Raw Gender Wage Gap 4.66 *** 3.93 *** 4.66 *** 3.93 ***Accounted for by differences in characteristicsSelected Centiles 0.83 *** 17.9% 1.67 *** 42.4% 0.77 *** 19.1% 1.46 *** 37.1%Demographics (age, marital status, kids) 0.04 *** 0.8% 0.00 *** 0.0% 0.01 *** 0.2% 0.00 *** 0.0%Education -0.17 *** -3.6% -0.54 ** -13.8% -0.05 *** -1.3% -0.10 *** -2.6%Part-time, Union, Tenure 0.40 *** 10.0% -0.01 *** -0.2%Industry 0.22 *** 5.6% 0.32 *** 8.2%Occupation 0.19 *** 4.8% 0.07 *** 1.8%Province 0.01 ** 0.2% 0.03 *** 0.01 0.1% 0.03 *** 0.7%Total Explained 0.71 *** 15.2% 1.16 *** 29.5% 1.55 *** 38.6% 1.77 *** 45.1%Total Unexplained 3.95 *** 84.8% 2.77 *** 70.5% 3.11 *** 77.1% 2.16 *** 54.9%Note: Entries are male/female differences in the explanatory variables multiplied by the corresponding female coefficients. All variables, except tenure are categorical. There are 4 marital status and 7 education classes, 11 industry, 47 occupation

O-B Decomposition in LFS 1997 and 2015

Source: Fortin, Drolet and Bonikowska (2016), LFS 1997-2015, 25-64 years old, Hourly wages on the main job

Presenter

Presentation Notes

Now for the LFS, we can conduct a full O-B decomposition and compare the impact of the under-representation of women in top centiles to standard explanatory variables. Here we find that even after controlling for education, union coverage, industry and occupation that the percentile groupings explaining a growing share of the gap, that is 37% in 2015 vs. 19% in 1997. Industry has become the second leading explanatory factor. Also notice the negative a coefficients on education and tenure showing that these human capital variables have become true countervailing factors.

Impact of Under-Representation in Top Jobs

• Over time, the under-representation of women in top jobs accounts for a growing share of the gender gap from 19% in 1997 to 37% in 2015, after accounting for the usual list of

factors (education, occupation, industry, etc.)• Even against industry and occupation, it is the most significantly

explanatory factor • It substantially reduces the unexplained portion of the gender gap which

had been growing over time in a puzzling way.

55

Presenter

Presentation Notes

To summarize the under-representation of women in top job accounts for a growing share of the gender gap. It is the most significant explanatory factors and reduces the unexplained portion of the gender gap, which have been growing over time home in a puzzling way.

Industry Composition by Gender (LFS)

56

0 5 10 15 20

Ext. Resources/Const.

Transp/WholeS/WhareH.

Manufacturing

Agri/Fish/Forest

Public Admin

Prof/Scien/Manag Serv.

Other Services

Retail Trade

Education

F.I.R.E.

Health Care/Soc. Ass.

A. 1997-2002

Men Women

0 5 10 15 20

Ext. Resources/Const.

Manufacturing

Transp/WholeS/WhareH.

Agri/Fish/Forest

Prof/Scien/Manag Serv.

Public Admin

Other Services

Retail Trade

F.I.R.E.

Education

Health Care/Soc. Ass.

B. 2011-2015

Men Women

Source: Fortin, Drolet and Bonikowska (2016) computation, LFS 1997-2015, 25-64 years old

Presenter

Presentation Notes

Let me now turn to our second counterfactual as exercise where we ask how women would fare if they had the same industrial distribution as men. One way to understand how men and women are distributed differently across industry is to think of women are being well-represented in Service Industries, with Health Care being the one where they really dominate, followed by the Education Sector, while men really dominate in the primary and secondary sectors, as well as in the manufacturing and distribution sectors.

Industry Composition by Gender and Selected Centiles

57

A: 1997-2002Bottom

90%Next 9%

Next 0.9%

Top 0.1%

Bottom 90%

Next 9%

Next 0.9%

Top 0.1%

Agri/Fish/Forest 2.1 0.8 0.7 0.1 0.9 0.2 0.1 0.0Ext. Resources/Const. 11.4 10.7 8.7 9.3 2.1 2.4 2.7 0.0Manufacturing 26.4 18.7 19.9 23.0 11.1 5.4 7.7 5.3Transp/WholeS/WhareH 13.6 8.9 8.0 8.7 5.5 2.6 5.0 0.3Retail Trade 9.1 3.5 3.3 3.8 11.9 2.5 1.7 1.1 F.I.R.E. 3.7 8.0 11.5 12.1 8.9 8.0 14.2 14.0Prof/Scien/Manag Serv. 7.1 10.8 15.7 17.3 7.9 8.4 17.0 8.7Education 4.7 14.0 11.1 8.8 9.9 33.8 25.8 40.6Health Care/Soc. Ass. 3.8 2.8 3.1 3.2 20.4 19.1 7.6 12.2Other Services 11.4 8.2 8.8 7.7 14.5 7.4 10.9 12.9Public Admin 6.8 13.6 9.2 6.0 6.7 10.3 7.6 4.8

WomenMen


Presenter

Presentation Notes

Among the top earners, the distribution of men and women is quite similar as for the overall. With a substantial concentration of men in the top 0.1% in the Manufacturing sector and women in the top 0.1% in the Education and other Services sectors and a quasi- absence in topin the and primaries and resources extraction Sectors. However they have been substantial important changes from our early 1997 to 2002 period to the last five years.

Industry Composition by Gender and Selected Centiles

58

B: 2011-2015Bottom

90%Next 9%

Next 0.9%

Top 0.1%

Bottom 90%

Next 9%

Next 0.9%

Top 0.1%

Agri/Fish/Forest 1.7 0.3 0.3 0.0 0.7 0.1 0.0 0.0Ext. Resources/Const. 14.5 14.5 12.9 17.3 2.7 3.5 5.8 2.8Manufacturing 20.3 13.7 15.7 12.1 8.1 4.3 5.2 8.5Transp/WholeS/WhareH 13.8 9.2 10.3 9.0 5.8 2.8 3.5 6.8Retail Trade 9.3 3.8 2.5 3.4 12.1 3.5 4.3 2.5 F.I.R.E. 4.4 8.5 11.1 13.6 8.4 9.0 12.4 11.8Prof/Scien/Manag Serv. 9.1 14.1 19.9 18.1 8.8 7.4 14.3 13.5Education 5.0 11.0 8.7 9.0 10.5 26.1 21.1 25.0Health Care/Soc. Ass. 4.1 3.6 2.5 3.0 22.1 24.7 11.6 12.3Other Services 11.7 7.2 7.9 6.3 13.9 5.5 6.2 6.0Public Admin 6.3 14.0 8.2 8.1 6.8 13.2 15.6 10.7

Men Women


Presenter

Presentation Notes

In more recent years, they have been a clear shift of men in the top jobs from the manufacturing sectors to the extractive resource sector. There have also been important shifts of women in top jobs away from education and into a more diverse set of industrial sectors, in particular the distribution sector.

What if women worked in the same industrial sectors as men?

59

0.70

0.75

0.80

0.85

0.90

0.95

1.00

1.05

1997 1999 2001 2003 2005 2007 2009 2011 2013 2015

Gender Ratio in Average Hourly Wages by Selected Percentiles


Rw_All Rw_B90 Rw_N9 Rw_N0.0 Rw_T0.1

Source: Fortin, Drolet and Bonikowska (2016) Computation, LFS 1997-2015, 25-64 years old, Hourly wages on the main job

Presenter

Presentation Notes

So when we ask within each percentile grouping "what if women worked in the same industrial sectors such as men" we find well that for women in the bottom 90%, and thus the overall, this would have implied a less favorable gender ratio.

What if women worked in the same industrial sectors as men?

60

0.70

0.75

0.80

0.85

0.90

0.95

1.00

1.05

1997 1999 2001 2003 2005 2007 2009 2011 2013 2015

Gender Ratio in Average Hourly Wages by Selected Percentiles

Bottom 90% Next 9% Rw_B90 Rw_N9

Source: Fortin, Drolet and Bonikowska (2016) Computation, LFS 1997-2015, 25-64 years old, Hourly wages on the main job

Presenter

Presentation Notes

However removing on the last 2 percentile groupings for clarity, we find that this is not the case for the next 9%, if women in the next 9% moved to the same industrial sectors as men, this would imply a very favorable gender ratio, on very close to parity.

Similar Impact of Industrial Composition on the Annual Earnings Ratio

61

0.5

0.6

0.7

0.8

0.9

1

1.1

1991 1993 1995 1997 1999 2001 2003 2005 2007 2009

Gender Ratio in Annual Earnings by Selected Percentiles

All Bottom 90% Next 9% Next 0.9% Top 0.1%Rw_all Rw_B90 Rw_N9 Rw_N0.9% Rw_T0.1

Source: Fortin, Drolet and Bonikowska (2016) Computation, LWF 1991-2010, 25-64 years old, Annual earnings from all jobs

Presenter

Presentation Notes

In in the LWF all we find a similar result of the industrial distribution of men are 1000 are held all that much of women in the bottom 90%.

Similar Impact of Industrial Composition on the Annual Earnings Ratio

62

0.7

0.75

0.8

0.85

0.9

0.95

1

1991 1993 1995 1997 1999 2001 2003 2005 2007 2009

Gender Ratio in Annual Earnings by Selected Percentiles

Bottom 90% Next 9% Rw_B90 Rw_N9


Presenter

Presentation Notes

Again removing the top centiles all for which we have common support issues, we find are less favorable gender gap for the bottom and for the top which are is almost reaching parity.

Impact of Industrial Composition

• Although issues of common support limit the analysis for the top 1% and 0.1%, overall women’s own choice of industrial sectors seem appropriate

• Among the top 9%-1%, women would almost reach parity if they works in the same industrial sectors as men, but in the bottom 90% would do worse

• Largely due to the health care sector, which is a well-paying sector in the bottom 90%, but less so in the next 9% (among the salaried workers we observe)

• It could be arguably different if we included self-employment income

63

Presenter

Presentation Notes

We looked a little more closely at which industrial sectors were the source of the effect. Unlike education, the countervailing effects comes from differences in returns across percentiles groupings. Essentially when you're in the bottom 90% for women, the healthcare sectors is relatively well-paying sector. However for those in the next 9%, it is not relatively less well-paying, especially given that we observed only salaried workers. For example, we are missing the business earnings of doctors.

Bottom-Line

• Looking back at the transformation of women’s work in Canada over the 20th century, Fortin and Huberman (2002) had argued that the decline in vertical segregation had contributed more to the improvement of women’s labour market outcomes than changes in horizontal segregation.

• With increasing earnings inequality in top incomes, further improvements in vertical segregation, “more women in top jobs” will be likely be even more important for further decline in the gender pay gap in the 21st

century• But unlike in the 20th century, further educational attainment alone will

not yield those changes!

64

Presenter

Presentation Notes

So the bottom line: Does increasing inequality in top income contribute to the slowdown of progress in the gender gap and the answer is Yes, it does. Indeed, it contributed to a growing share of the gap falling into the unexplained category, that is unexplained by the usual factors. We conjecture that further decline in the gender gap in the 21st century will have to come from "more women in top jobs". So the next question is what are the best policies and firm practices that can facilitate an increased representation of women in top jobs.

Public Policy and Gender Pay Differentials

• Gender pay differentials “within” occupation “Equal Pay for Equal Work” • Gender pay differentials across “comparable” occupations, resulting from

horizontal segregation, are the focus of “Pay Equity” policies, implemented in the private sector of Canada’s two

most populous provinces: Ontario (1996) and Quebec (2001)• Gender pay differentials arising from the potential obstacles that women

face climbing (or not) the job ladder (vertical segregation) “Employment Equity”, enacted in the Federal jurisdiction in principle could

address disparities across the job ladder.

65

Presenter

Presentation Notes

The existing policies to foster gender equality in the workplace are unlikely to be very helpful. "Equal Pay for Equal Work" is within-occupation policy, "Pay Equity" work across comparable occupations in an horizontal fashion. Only "Employment Equity" is directed at vertical segregation, but its coverage is relatively narrow, and has focused in recent years on issues of ethnic and racial diversity, and disability.

More Women in Tops Jobs! What to Do?

• In recent years, many countries have pushed for more general gender equality in decision-making with bolder moves. Both in the political sphere and on corporate boards.

• Many European countries implemented female quotas on the board of directors of firms on public stock exchanges.

• Some emerging countries are doing the same.

66

Presenter

Presentation Notes

In recent years, many countries have implemented much bolder measures are to foster more gender equality in this decision making, both in political sphere and on corporate boards.

Quotas for corporate boards advance?

67

Source: Dizik, 2015

Presenter

Presentation Notes

In particular many European countries have implemented female minimum or female quotas on the board of directors of firms on public stock exchanges and some emerging countries are doing the same.

More Women in Tops Jobs! What to Do?

• Short of calling for gender quotas, the Canadian Securities Administrators of seven provinces and territories (CSA, 2015) implemented “comply-or-explain” female representation rules on January 1, 2015 (Shecter, 2014; McFarland, 2015).

• These rules require companies listed on their stock exchanges to disclose how many women they have on their boards and in their executive ranks.

• But many companies have shown bare `technical compliance’ with the reporting rules introduced last year and it is "simply not good enough," says Howard Wetston, the Ontario Securities Commission chair.

68

Presenter

Presentation Notes

Despite the hopeful note on the earlier slide, the administrators of Canadian securities exchanges instead of calling for quotas have implemented some disclosure rules. One year after these disclosure rules were put in place their implementation of was thought to be quite lackluster.

CAUS CAUS CAUS

AU

NZAUNZAUNZ

HK

ID

JP

MY

PH

SG

THHK

JP

MY

PHTH

ID

JP

MY PH SGTH AT

BEFRDE

GR

IE

IT

NLES

CH

GB

ATBECZFR

DEGR

HU IE

IT

NL

PL

PT

RUESCH GB

ATBECZ

FR

DE

GR HUIE

IT

NLPL

PTRU

ES CHGBAT

BE

FR

DE

GRHU

IEIT

NLPL

PTRU

ESCHGB

DKFI

NO

SE

DK

FI

NO

SE

DK

FI

NO

SE

DK

FI

NO

SE0

1020

3040

Wom

en o

n Bo

ards

(%)

40 45 50Female Employment Share (%)

Can/US AU/NZ AsiaEurope Nordic Countries

Years: 2006, 2009, 2011, 2013, 2014

Women on Boards and Employment Share

Sources: GMI, European PWN, Deloitte (2015), World Bank Indicators

Presenter

Presentation Notes

When looking at a set of European and Asian countries, the median female employment share is above 45% while the share of women on board around 9-11%.

AU

AUAU

ATAT

ATAT

BE BEBE BE

CZCZ

DK

DKDKDK

FIFI

FIFI

FRFR

FRFR

DEDE

DEDE

GR

GR GR GR HKHKHU

HUHUID

IDIE IEIEIE

ITIT IT

IT

JPJPJP

MYMYMY NL

NLNL

NL

NZNZ

NZ

NO

NONONO

PH

PH

PH PL

PLPL

PT PTPT

RURURUSG SGES

ESES ES

SESESESE

CHCH CHCHTHTHTH

GBGB

GBGB

AUATATBE BE

DK

FRFR

ITIT

MY NL

NLNL

ES

AUAUATATBE BE

DKDKDKFRFR

GR

GR GR GRIT

ITMYMY

NL

NO

NONONO

ESES ES

FI

HKHK

JPJP

NZNZ

PL

PLGB

GBGB

FIFIFI

JP

NZPL GB

010

2030

40W

omen

on

Boar

ds (%

)

40 45 50Female Employment Share (%)

No Regulation or Quota Pre-Quota Post-QuotaPre-Regulation Post-Regulation Fitted values

Years: 2006, 2009, 2011, 2013, 2014

Women on Boards and Employment Share

Sources: GMI, European PWN, Deloitte (2015), World Bank Indicators

Presenter

Presentation Notes

However many of these countries have implemented quotas or regulation and both seem quite effective in a country fixed effect model at increasing the share of women on board by 3 to 5%.

More Women in Tops Jobs! Does it Help?

• Female CEOs/Directors have mixed results on firm performance (returns on asset, on equity, profits, Tobin Q, etc.) in firm-fixed effects models– Positive or no effect in Denmark [Smith et al. (2006), Parotta and

Smith (2013)] and in Italy [Amore et al. (2013)]– Negative or no effect in the US [Wolfers (2006), Adams and Ferreira

(2009)], in Norway [Ahern and Dittmar (2012)]• Bertrand, Black, Jensen, and Lleras-Muney (2014) show mixed results on

relative female wages: the Norwegian quotas increased representation of women among top 5 highest earners, but had no effect at other points in the distribution or on the gender pay gap.

71

Presenter

Presentation Notes

However the impact of these quotas either on firm of performance or on relative female wages had been shown to be quite mixed.

ATAT

ATATAU

AUAUAUBE

BE

BE BECHCHCZCZ

CZCZ DEDEDEDE

DKDK

DKDK

ESES

ESES

FI FI

FIFI

FRFRFR

FRGB

GB

GBGB

GR

GR

GR

GR

HUHUHUHU

IDID

IE IE

IEIE

IT IT IT IT

MY

MYMY

NLNL

NL

NL

NONO

NONO

PHPH

PLPLPLPL

PT

PT

PTPT

SESESESE

TH

TH

ATAT

AUAUBE

BE

DK

ES

FRFR

IT IT

MYNL

NLNL

ATAT

AUAUBE BE

DK

DKDK

ES

ESES

FR

FR

GR

GR

GR

GR

IT ITMYMYNL

NONO

NONOFIGB

GB

GB

PLPLPLFI

FIFI

GB

PL15

2535

45W

omen

in S

enio

r Man

agem

ent (

%)

35 40 45 50Female Employment Share (%)

No Regulation or Quota Pre-Quota Post-QuotaPre-Regulation Post-Regulation Fitted values

Years: 2006, 2009, 2011, 2013, 2014

Women in Senior Management and Employment Share

Sources: ILO, World Bank Indicators

Presenter

Presentation Notes

Even the simple relationship between female employment share and share of women in senior management home across all quotas and non-quotas countries, pre and post quota years does not show as clear a relationship as for women on corporate boards.

Women Fail to Move from Bottom 90% to Next 9%in Early Career (age 30)

73

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.04083

_84

84_8

5

85_8

6

86_8

7

87_8

8

88_8

9

89_9

0

90_9

1

91_9

2

92_9

3

93_9

4

94_9

5

95_9

6

96_9

7

97_9

8

98_9

9

99_0

0

00_0

1

01_0

2

02_0

3

03_0

4

04_0

5

05_0

6

06_0

7

07_0

8

08_0

9

09_1

0

Share of Men and Women Moving from Bottom 90% to Next 9%- Recent Synthetic Cohorts

Men 1960 Men 1970 Men 1980 Men All

Women 1960 Women 1970 Women 1980 Women All

30

3030

30

30

30


Presenter

Presentation Notes

Now when we look at yearly transitions from the bottom 90% to the next 9% are using our LWF Data. The dotted line show the transitions for everybody but what is more interesting the focus on a recent cohorts which show that by age 30 men are beginning to make transitions are into the next 9% at much higher rates than women.

Public Policy and Gender Pay Differentials

• Likely the better policies are those that “level the playing field”, but without lowering women’s attachment to the labour marketo Maternity-leave benefits, parental leave provisions But gender neutrality is an issue! (Antecol, Bedard, and Stearns, 2016)o Affordable high-quality child care

• Firm practices are likely also important: o On-site child care and o Flexible hours of work o Paying attention to gender biases at work

74

Presenter

Presentation Notes

So when asked which are the best policies to foster gender equality in the workplace, this leads me to believe that policies and firm practices that help women with the work-life balance will likely home be the more hepful. This is not to say that to women all may be facing on gender bias and other obstacles as the are trying to climb on the organizational ladder wrong but nevertheless on it's a beginning.

Why so Few Women in Top Jobs? Paths for Future Research

Hypotheses from Labour/Behavioral Economics

— Women shy away from competition (Gneezy, Niederle and Rustichini, 2003; Niederle and Vesterlund, 2007)

— Women cannot say “no” to non-promotable tasks (Vesterlund, 2015)— Negotiating divide (Babcock and Laschever, 2003, 2009) — The importance of money vs. people (Fortin, 2008), vs. work flexibility

(Blau and Kahn, 2016; Goldin, 2014)— Differential treatment by customers (i.e. discrimination) (80 cents eBay,

Kricheli-Katz and Regev, 2016)

75

Presenter

Presentation Notes

Now they are a number of interesting hypotheses to explore in future research: women's lack of competitiveness, the negotiating divide, the importance of money vs. people, the importance of long work flexibility.

Why so Few Women in Top Jobs? Paths for Future Research

Hypotheses from Identity Theory/Social Psychology— Glass cliff phenomenon (Ryan and Haslam, 2007), — Failure of romance of leadership to take hold (Kulich, Ryan,and Haslam,

2007) — Recognition deficit emanating from role incongruity (Eagly and Karau,

2002) — Weak work networks within the firm or the industry (Lalanne and

Seabright, 2011) — Preferences for deontological over utilitarian judgments, weaker team

spirit (Kennedy and Kray, 2013; Friesdorf, Conway, and Gawronski, 2015)

76

Why so Few Women in Top Jobs? Recognition Deficit

Source: Maller and Kossof, McKinsey & Company, 2013 *C-level: Chief Executive Officer, Chief Financial Officer, etc.77

Why so Few Women in Top Jobs? Preferences ….

Source: Barsch and Yee, McKinsey & Company, 2012 *C-suite: Chief Executive Officer, Chief Financial Officer, etc.78

Presenter

Presentation Notes

The consulting firm McKinsey & Company which conducts quite interesting survey among its clients and finds in differences in attitudes towards female leaders.

Stay Tuned!

Thank you!

79

References

• Adams, R.B. and Ferreira, D., 2009. “Women in the Boardroom and their Impact on Governance and Performance." Journal of Financial Economics, 94, 291-309.

• Ahern, Kenneth R., and Amy K. Dittmar (2012) “The Changing of the Boards: The Impact on Firm Valuation of Mandated Female Board Representation,” Quarterly Journal of Economics, 127: 137-197.

• Amore M. D., O. Garofalo and A. Minichilli (2013) “Gender Interactions Within the Family Firm." Management Science, forthcoming.

• Antecol, Heather, Kelly Bedard, and Jenna Stearns (2016). “Equal but Inequitable: Who Benefits from Gender-Neutral Tenure Clock Stopping Policies?” Paper presented at the SOLE 2016.

• Bailey, Martha J. (2006) More power to the pill: the impact of contraceptive freedom on women's life cycle labor supply. Quarterly Journal of Economics, pp.289-320.

• Baker, Michael and Drolet, Marie (2010) “A new view of the male/female pay gap,” Canadian Public Policy, 36: 429-464.

• Bertrand, Marianne, Sandra E. Black, Sissel Jensen, and Adriana Lleras-Muney, “Breaking the Glass Ceiling? The Effect of Board Quotas on Female Labor Market Outcomes in Norway,” NBER working paper, June 2014.

80

References

• Blau, Francine D., and Lawrence M. Kahn (1997) "Swimming upstream: Trends in the gender wage differential in the 1980s." Journal of labor Economics: 1-42.

• Blau, Francine D. and Lawrence Kahn (2016) The Gender Wage Gap: Extent, Trends, and Explanations, IZA Discussion Paper No. 9656.

• Canadian Securities Administrators [CSA] (2015) Multilateral Staff Notice 58-307 Staff Review of Women on Boards and in Executive Officer Positions –Compliance with NI 58-101 Disclosure of Corporate Governance Practice.

• Coulombe, Serge and Marc Frenette (2007) “Has Higher Education Among Young Women Substantially Reduced the Gender Gap in Employment and Earnings?” (No. 2007301e). Statistics Canada, Analytical Studies Branch.

• Cortés, Patricia, and Jessica Pan (2015) “When Time Binds: Returns to Working Long Hours and the Gender Wage Gap among the Highly Skilled,” Paper presented at the SOLE 2015.

• Cortés, Patricia, and Jessica Pan (2016) “Prevalence of Long Hours and Women’ Job Choices: Evidence across Countries and within the U.S.” Paper presented at the SOLE 2016.

• Dinardo, John, Nicole M. Fortin and Thomas Lemieux (1996),“Labor Market Institutions and the Distribution of Wages, 1973–1992: a Semiparametric Approach”, Econometrica, 64: 1001-1046.

81

References

• Dizik, Alina, “Do quotas for corporate boards help women advance?” Capital Ideas Magazine, Spring 2015,

• Drolet, Marie (2001) “The Persistent Gap: New Evidence on the Canadian Gender Wage Gap,” Statistics Canada Analytical Studies Branch Working Paper, (157).

• Drolet, M. (2011) Why has the gender wage gap narrowed?. Perspectives on Labour and Income, 23(1), p.5.

• Eagly, Alice H., and Steven J. Karau (2002) “Role Congruity Theory of Prejudice toward Female Leaders,” Psychological Review, 109: 573-598.

• Fortin, Nicole M. and Michael Huberman (2002) “Occupational Gender Segregation and Women’s Wages in Canada: an Historical Perspective” Canadian Public Policy, (May 2002), S11–S39.

• Fortin, Nicole M. (2015) “Gender Role Attitudes and Women's Labor Market Participation: Opting-Out, AIDS, and the Persistent Appeal of Housewifery,” Annals of Economics and Statistics, 117-118, Special Issue on the Economics of Gender, 379-401.

82

References

• Friesdorf, Rebecca, Paul Conway, and Bertram Gawronski (2015) “Gender Differences in Responses to Moral Dilemmas A Process Dissociation Analysis,” Personality and Social Psychology Bulletin, 41: 696-713.

• Frenette, Marc (2014) “An Investment of a Lifetime? The Long-term Labour Market Premiums Associated with a Postsecondary Education,” Analytical Studies Branch Research Paper Series, no. 359. Statistics Canada Catalogue no. 11F0019M. Ottawa: Statistics Canada

• Gagliarducci , Stefano and M. Daniele Paserman (2015b) “The Effect of Female Leadership on Establishment and Employee Outcomes: Evidence from Linked Employer-Employee Data,” in Solomon W. Polachek , Konstantinos Tatsiramos , Klaus F. Zimmermann (ed.) Gender Convergence in the Labor Market, Research in Labor Economics, 41: 343 – 375.

• Goldin, Claudia, and Lawrence Katz (2002) "The Power of the Pill: Oral Contraceptives and Women's Career and Marriage Decisions," Journal of Political Economy, CX (2002), 730-770.

• Goldin, Claudia (2014) “A Grand Gender Convergence: Its Last Chapter” American Economic Review, 104(4): 1091-1119.

• Guvenen, Fatih, Greg Kaplan, and Jae Song. (2014) “The Glass Ceiling and the Paper Floor: Gender Differences among Top Earners, 1981-2012.” No. w20560. National Bureau of Economic Research

83

References

• Kennedy, Jessica A., and Laura J. Kray (2014) "Who is Willing to Sacrifice Ethical Values for Money and Social Status? Gender Differences in Reactions to Ethical Compromises." Social Psychological and Personality Science, 5: 152-159.

• Kuhn, Peter, and Fernando Lozano (2008) “The Expanding Workweek? Understanding Trends in Long Work Hours among US men, 1979-2004.” Journal of Labor Economics, 2008, 26(2): 311-343.

• Kline, Patrick (2011) "Oaxaca-Blinder as a reweighting estimator." American Economic Review, 101(3): 532-537.• Lalanne, Marie, and Paul Seabright (2011) “The Old Boy Network: Gender Differences in the Impact of Social

Networks on Remuneration in Top Executive Jobs,” Toulouse School of Economics.• Lemieux, Thomas, and W. Craig Riddell (2015) Top incomes in Canada: Evidence from the Census. No. w21347.

National Bureau of Economic Research.• Lesser, Kim M., and Janet Lever (2011) “Does Gender Bias against Female Leaders Persist? Quantitative and

Qualitative Data from a Large-Scale Survey,” Human Relations, 64: 1555-1578.• MacFarland, Janet (2015) “OSC Rebukes Firms for Lack of Action on Gender-Diversity Rules,” The Globe and Mail,

June 16 2015.

84

References

• Matsa, David A., and Amalia R. Miller (2013) “A Female Style in Corporate Leadership? Evidence from Quotas." American Economic Journal: Applied Economics, 5: 136-69.

• Mincer, Jacob, and Solomon Polachek (1974) “Family investments in human capital: Earnings of women." In Marriage, family, human capital, and fertility, Journal of Political Economy, 82: 76-110.

• Mulligan, Casey B., and Yona Rubinstein (2008) "Selection, Investment, and Women's Relative Wages over Time." Quarterly Journal of Economics : 1061-1110.

• Niederle, Muriel and Lise Vesterlund (2007) “Do Women Shy Away from Competition? Do Men Compete too Much?” Quarterly Journal of Economics, 122: 1067-1101.

• Parrotta P. and Smith, N. (2013) “Female-Led Firms: Performance and Risk Attitudes.“ IZA Discussion Paper No. 7613.

• Piketty, Thomas, and Emmanuel Saez (2003) "Income Inequality in the United States, 1913–1998*." The Quarterly journal of economics 118 (1): 1-41.

• Piketty, Thomas, and Emmanuel Saez. "Top incomes and the great recession: Recent evolutions and policy implications." IMF economic review 61, no. 3 (2013): 456-478.

85

References

• Shecter, Barbara (2014) “OSC proposes rules for women on boards, but stops short of quotas,” Financial Post, January 16, 2014.

• Schirle, Tammy (2015) “The Gender Wage Gap in the Canadian Provinces, 1997-2014 (No. 0088)” Laurier Centre for Economic Research and Policy Analysis.

• Smith, N., Smith, V. and Verner, M. (2006) “Do Women in Top Management Affect Firm Performance? A Panel Study of 2500 Danish Firms." International Journal of Productivity and Performance Management, 55(7), 569-593.

• Veall, Michael R. (2012) “Top income shares in Canada: recent trends and policy implications,” Canadian Journal of Economics, 45(4), pp.1247-1272.

86

Documents

Gender Pay Gap - Faculty and Staff | Faculty of Artsfaculty.arts.ubc.ca/nfortin/Fortin... · 2016-06-07 · Earnings Inequality and the Gender Pay Gap Nicole Fortin Vancouver School