Gender board diversity and firm performance
grape_uw
99 views
42 slides
May 07, 2024
Slide 1 of 42
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
About This Presentation
Shift share IV identification of causal impact of gender board diversity on firm performance
Slide Content
Gender board diversity and firm performance:
a shift-share IV identification
Katarzyna Bech-Wysocka(FAME|GRAPE & SGH Warsaw School of Economics)
Joanna Tyrowicz(FAME|GRAPE, University of Regensburg, and IZA)
Sebastian Zalas(FAME|GRAPE, University of Warsaw)
Joint PhD Workshop
April 15, 2024
1
a shift-share IV identification
Katarzyna Bech-Wysocka(FAME|GRAPE & SGH Warsaw School of Economics)
Joanna Tyrowicz(FAME|GRAPE, University of Regensburg, and IZA)
Sebastian Zalas(FAME|GRAPE, University of Warsaw)
Joint PhD Workshop
April 15, 2024
1
Motivation
•We study the impact ofgender board diversityon firm performance
•Women directors tend to be under-represented on corporate boards:
•only 19% of board seats globally are occupied by women
•only 13% of companies have gender balanced boards
•less than 5% of the CEOs of the world’s largest corporations are women (International Labour
Office 2015a)
•Elgart (1983) ) predicts that the equal representation of women and men in top corporate
boardrooms will take around 200 years to achieve
2
•We study the impact ofgender board diversityon firm performance
•Women directors tend to be under-represented on corporate boards:
•only 19% of board seats globally are occupied by women
•only 13% of companies have gender balanced boards
•less than 5% of the CEOs of the world’s largest corporations are women (International Labour
Office 2015a)
•Elgart (1983) ) predicts that the equal representation of women and men in top corporate
boardrooms will take around 200 years to achieve
2
Motivation
•We study the impact ofgender board diversityon firm performance
•Women directors tend to be under-represented on corporate boards:
•only 19% of board seats globally are occupied by women
•only 13% of companies have gender balanced boards
•less than 5% of the CEOs of the world’s largest corporations are women (International Labour
Office 2015a)
•Elgart (1983) ) predicts that the equal representation of women and men in top corporate
boardrooms will take around 200 years to achieve
2
•We study the impact ofgender board diversityon firm performance
•Women directors tend to be under-represented on corporate boards:
•only 19% of board seats globally are occupied by women
•only 13% of companies have gender balanced boards
•less than 5% of the CEOs of the world’s largest corporations are women (International Labour
Office 2015a)
•Elgart (1983) ) predicts that the equal representation of women and men in top corporate
boardrooms will take around 200 years to achieve
2
Motivation
•We study the impact ofgender board diversityon firm performance
•Women directors tend to be under-represented on corporate boards:
•only 19% of board seats globally are occupied by women
•only 13% of companies have gender balanced boards
•less than 5% of the CEOs of the world’s largest corporations are women (International Labour
Office 2015a)
•Elgart (1983) ) predicts that the equal representation of women and men in top corporate
boardrooms will take around 200 years to achieve
2
•We study the impact ofgender board diversityon firm performance
•Women directors tend to be under-represented on corporate boards:
•only 19% of board seats globally are occupied by women
•only 13% of companies have gender balanced boards
•less than 5% of the CEOs of the world’s largest corporations are women (International Labour
Office 2015a)
•Elgart (1983) ) predicts that the equal representation of women and men in top corporate
boardrooms will take around 200 years to achieve
2
Motivation
•Gender quota legislation
•in 2003,Norwaywas the first country globally to adopt a gender quota requiring a 40% female
board representation in public limited and state-owned companies (implemented in 2006)
•legislated board quotas in many European countries, includingBelgium, France, Germany,
Iceland, Italy, the Netherlands, Spain and Sweden, but also inIsraelandIndia
•in 2020, theEuropean Unionstarted to review plans for mandatory gender quotas to speed up
the progress of women in gaining leadership positions
•Voluntary gender quotas
•in 2011, the Lord Davies report recommended adopting voluntary targets in theUK→the
representation of women on boards of FTSE 100 companies reached 32.4% in 2019 (International
Labour Office 2020)
•Deutsche Telekom, a leadingGermantelecommunications company, set voluntary quotas in 2010
→40% women on its group supervisory board in 2015 (International Labour Office 2015b)
•As a result of the voluntary measure,Australiahas reached 30% women on the boards of
Australian Securities Exchange companies in 2019 (International Labour Office 2020)
3
•Gender quota legislation
•in 2003,Norwaywas the first country globally to adopt a gender quota requiring a 40% female
board representation in public limited and state-owned companies (implemented in 2006)
•legislated board quotas in many European countries, includingBelgium, France, Germany,
Iceland, Italy, the Netherlands, Spain and Sweden, but also inIsraelandIndia
•in 2020, theEuropean Unionstarted to review plans for mandatory gender quotas to speed up
the progress of women in gaining leadership positions
•Voluntary gender quotas
•in 2011, the Lord Davies report recommended adopting voluntary targets in theUK→the
representation of women on boards of FTSE 100 companies reached 32.4% in 2019 (International
Labour Office 2020)
•Deutsche Telekom, a leadingGermantelecommunications company, set voluntary quotas in 2010
→40% women on its group supervisory board in 2015 (International Labour Office 2015b)
•As a result of the voluntary measure,Australiahas reached 30% women on the boards of
Australian Securities Exchange companies in 2019 (International Labour Office 2020)
3
Motivation
•Gender quota legislation
•in 2003,Norwaywas the first country globally to adopt a gender quota requiring a 40% female
board representation in public limited and state-owned companies (implemented in 2006)
•legislated board quotas in many European countries, includingBelgium, France, Germany,
Iceland, Italy, the Netherlands, Spain and Sweden, but also inIsraelandIndia
•in 2020, theEuropean Unionstarted to review plans for mandatory gender quotas to speed up
the progress of women in gaining leadership positions
•Voluntary gender quotas
•in 2011, the Lord Davies report recommended adopting voluntary targets in theUK→the
representation of women on boards of FTSE 100 companies reached 32.4% in 2019 (International
Labour Office 2020)
•Deutsche Telekom, a leadingGermantelecommunications company, set voluntary quotas in 2010
→40% women on its group supervisory board in 2015 (International Labour Office 2015b)
•As a result of the voluntary measure,Australiahas reached 30% women on the boards of
Australian Securities Exchange companies in 2019 (International Labour Office 2020)
3
•Gender quota legislation
•in 2003,Norwaywas the first country globally to adopt a gender quota requiring a 40% female
board representation in public limited and state-owned companies (implemented in 2006)
•legislated board quotas in many European countries, includingBelgium, France, Germany,
Iceland, Italy, the Netherlands, Spain and Sweden, but also inIsraelandIndia
•in 2020, theEuropean Unionstarted to review plans for mandatory gender quotas to speed up
the progress of women in gaining leadership positions
•Voluntary gender quotas
•in 2011, the Lord Davies report recommended adopting voluntary targets in theUK→the
representation of women on boards of FTSE 100 companies reached 32.4% in 2019 (International
Labour Office 2020)
•Deutsche Telekom, a leadingGermantelecommunications company, set voluntary quotas in 2010
→40% women on its group supervisory board in 2015 (International Labour Office 2015b)
•As a result of the voluntary measure,Australiahas reached 30% women on the boards of
Australian Securities Exchange companies in 2019 (International Labour Office 2020)
3
What we know - theory
•human capital theory→positive impact
e.g. Becker (1964), Miller and Triana (2009), Faccio et al. (2016), Shaukat et al. (2016), Low et al. (2015), Smith et al. (2006)
•social identity theory→negative impact
e.g. Tajfel and Turner (1986)
•social network theory→negative impact
e.g. T¨onnies and Loomis (1959), Hambrick and Mason (1984), Ahern and Dittmar (2012)
•resource dependency theory→positive impact
e.g. Salancik and Pfeffer (1978), Hillman et al. (2000), D’Souza et al. (2010), Isidro and Sobral (2014), L¨uckerath-Rovers
(2013)
•agency theory→positive impact
e.g. Fama and Jensen (1983), Gul et al. (2008), Adams and Ferreira (2009), Simkins and Simpson (2003), Ararat et al. (2015),
Nguyen et al. (2015)
4
•human capital theory→positive impact
e.g. Becker (1964), Miller and Triana (2009), Faccio et al. (2016), Shaukat et al. (2016), Low et al. (2015), Smith et al. (2006)
•social identity theory→negative impact
e.g. Tajfel and Turner (1986)
•social network theory→negative impact
e.g. T¨onnies and Loomis (1959), Hambrick and Mason (1984), Ahern and Dittmar (2012)
•resource dependency theory→positive impact
e.g. Salancik and Pfeffer (1978), Hillman et al. (2000), D’Souza et al. (2010), Isidro and Sobral (2014), L¨uckerath-Rovers
(2013)
•agency theory→positive impact
e.g. Fama and Jensen (1983), Gul et al. (2008), Adams and Ferreira (2009), Simkins and Simpson (2003), Ararat et al. (2015),
Nguyen et al. (2015)
4
What we know - theory
•human capital theory→positive impact
e.g. Becker (1964), Miller and Triana (2009), Faccio et al. (2016), Shaukat et al. (2016), Low et al. (2015), Smith et al. (2006)
•social identity theory→negative impact
e.g. Tajfel and Turner (1986)
•social network theory→negative impact
e.g. T¨onnies and Loomis (1959), Hambrick and Mason (1984), Ahern and Dittmar (2012)
•resource dependency theory→positive impact
e.g. Salancik and Pfeffer (1978), Hillman et al. (2000), D’Souza et al. (2010), Isidro and Sobral (2014), L¨uckerath-Rovers
(2013)
•agency theory→positive impact
e.g. Fama and Jensen (1983), Gul et al. (2008), Adams and Ferreira (2009), Simkins and Simpson (2003), Ararat et al. (2015),
Nguyen et al. (2015)
4
•human capital theory→positive impact
e.g. Becker (1964), Miller and Triana (2009), Faccio et al. (2016), Shaukat et al. (2016), Low et al. (2015), Smith et al. (2006)
•social identity theory→negative impact
e.g. Tajfel and Turner (1986)
•social network theory→negative impact
e.g. T¨onnies and Loomis (1959), Hambrick and Mason (1984), Ahern and Dittmar (2012)
•resource dependency theory→positive impact
e.g. Salancik and Pfeffer (1978), Hillman et al. (2000), D’Souza et al. (2010), Isidro and Sobral (2014), L¨uckerath-Rovers
(2013)
•agency theory→positive impact
e.g. Fama and Jensen (1983), Gul et al. (2008), Adams and Ferreira (2009), Simkins and Simpson (2003), Ararat et al. (2015),
Nguyen et al. (2015)
4
What we know - theory
•human capital theory→positive impact
e.g. Becker (1964), Miller and Triana (2009), Faccio et al. (2016), Shaukat et al. (2016), Low et al. (2015), Smith et al. (2006)
•social identity theory→negative impact
e.g. Tajfel and Turner (1986)
•social network theory→negative impact
e.g. T¨onnies and Loomis (1959), Hambrick and Mason (1984), Ahern and Dittmar (2012)
•resource dependency theory→positive impact
e.g. Salancik and Pfeffer (1978), Hillman et al. (2000), D’Souza et al. (2010), Isidro and Sobral (2014), L¨uckerath-Rovers
(2013)
•agency theory→positive impact
e.g. Fama and Jensen (1983), Gul et al. (2008), Adams and Ferreira (2009), Simkins and Simpson (2003), Ararat et al. (2015),
Nguyen et al. (2015)
4
•human capital theory→positive impact
e.g. Becker (1964), Miller and Triana (2009), Faccio et al. (2016), Shaukat et al. (2016), Low et al. (2015), Smith et al. (2006)
•social identity theory→negative impact
e.g. Tajfel and Turner (1986)
•social network theory→negative impact
e.g. T¨onnies and Loomis (1959), Hambrick and Mason (1984), Ahern and Dittmar (2012)
•resource dependency theory→positive impact
e.g. Salancik and Pfeffer (1978), Hillman et al. (2000), D’Souza et al. (2010), Isidro and Sobral (2014), L¨uckerath-Rovers
(2013)
•agency theory→positive impact
e.g. Fama and Jensen (1983), Gul et al. (2008), Adams and Ferreira (2009), Simkins and Simpson (2003), Ararat et al. (2015),
Nguyen et al. (2015)
4
What we know - theory
•human capital theory→positive impact
e.g. Becker (1964), Miller and Triana (2009), Faccio et al. (2016), Shaukat et al. (2016), Low et al. (2015), Smith et al. (2006)
•social identity theory→negative impact
e.g. Tajfel and Turner (1986)
•social network theory→negative impact
e.g. T¨onnies and Loomis (1959), Hambrick and Mason (1984), Ahern and Dittmar (2012)
•resource dependency theory→positive impact
e.g. Salancik and Pfeffer (1978), Hillman et al. (2000), D’Souza et al. (2010), Isidro and Sobral (2014), L¨uckerath-Rovers
(2013)
•agency theory→positive impact
e.g. Fama and Jensen (1983), Gul et al. (2008), Adams and Ferreira (2009), Simkins and Simpson (2003), Ararat et al. (2015),
Nguyen et al. (2015)
4
•human capital theory→positive impact
e.g. Becker (1964), Miller and Triana (2009), Faccio et al. (2016), Shaukat et al. (2016), Low et al. (2015), Smith et al. (2006)
•social identity theory→negative impact
e.g. Tajfel and Turner (1986)
•social network theory→negative impact
e.g. T¨onnies and Loomis (1959), Hambrick and Mason (1984), Ahern and Dittmar (2012)
•resource dependency theory→positive impact
e.g. Salancik and Pfeffer (1978), Hillman et al. (2000), D’Souza et al. (2010), Isidro and Sobral (2014), L¨uckerath-Rovers
(2013)
•agency theory→positive impact
e.g. Fama and Jensen (1983), Gul et al. (2008), Adams and Ferreira (2009), Simkins and Simpson (2003), Ararat et al. (2015),
Nguyen et al. (2015)
4
What we know - theory
•human capital theory→positive impact
e.g. Becker (1964), Miller and Triana (2009), Faccio et al. (2016), Shaukat et al. (2016), Low et al. (2015), Smith et al. (2006)
•social identity theory→negative impact
e.g. Tajfel and Turner (1986)
•social network theory→negative impact
e.g. T¨onnies and Loomis (1959), Hambrick and Mason (1984), Ahern and Dittmar (2012)
•resource dependency theory→positive impact
e.g. Salancik and Pfeffer (1978), Hillman et al. (2000), D’Souza et al. (2010), Isidro and Sobral (2014), L¨uckerath-Rovers
(2013)
•agency theory→positive impact
e.g. Fama and Jensen (1983), Gul et al. (2008), Adams and Ferreira (2009), Simkins and Simpson (2003), Ararat et al. (2015),
Nguyen et al. (2015)
4
•human capital theory→positive impact
e.g. Becker (1964), Miller and Triana (2009), Faccio et al. (2016), Shaukat et al. (2016), Low et al. (2015), Smith et al. (2006)
•social identity theory→negative impact
e.g. Tajfel and Turner (1986)
•social network theory→negative impact
e.g. T¨onnies and Loomis (1959), Hambrick and Mason (1984), Ahern and Dittmar (2012)
•resource dependency theory→positive impact
e.g. Salancik and Pfeffer (1978), Hillman et al. (2000), D’Souza et al. (2010), Isidro and Sobral (2014), L¨uckerath-Rovers
(2013)
•agency theory→positive impact
e.g. Fama and Jensen (1983), Gul et al. (2008), Adams and Ferreira (2009), Simkins and Simpson (2003), Ararat et al. (2015),
Nguyen et al. (2015)
4
What we know - empirical analysis
•no impacte.g. Harasheh and Provasi (2021), Ming and Eam (2016), Marinova et al. (2010), Joecks et al. (2013), Wiley and
Tormos (2018)
•meta-analysis by Post and Byron (2015):mixed evidence•How?
•inconsistent OLS estimates, due to endogeneity (self-selection and/or reverse-causality)
•natural experimentse.g. Ahern and Dittmar (2012), Matsa and Miller (2013), Yang et al. (2019)•IV:
•Adams and Ferreira (2009) use social networks (instrument: fraction of male directors on the board
who sit on other boards on which there are female directors)
•Low et al. (2015) instrument: percentage of female (lower level) managers
•Smith et al. (2006) instrument: education level of spouses of other CEOs in same company
•Liu et al. (2014) instrument: percentage of women directors and the percentage of female
employment in its own industry
5
•no impacte.g. Harasheh and Provasi (2021), Ming and Eam (2016), Marinova et al. (2010), Joecks et al. (2013), Wiley and
Tormos (2018)
•meta-analysis by Post and Byron (2015):mixed evidence•How?
•inconsistent OLS estimates, due to endogeneity (self-selection and/or reverse-causality)
•natural experimentse.g. Ahern and Dittmar (2012), Matsa and Miller (2013), Yang et al. (2019)•IV:
•Adams and Ferreira (2009) use social networks (instrument: fraction of male directors on the board
who sit on other boards on which there are female directors)
•Low et al. (2015) instrument: percentage of female (lower level) managers
•Smith et al. (2006) instrument: education level of spouses of other CEOs in same company
•Liu et al. (2014) instrument: percentage of women directors and the percentage of female
employment in its own industry
5
What we know - empirical analysis
•no impacte.g. Harasheh and Provasi (2021), Ming and Eam (2016), Marinova et al. (2010), Joecks et al. (2013), Wiley and
Tormos (2018)
•meta-analysis by Post and Byron (2015):mixed evidence•How?
•inconsistent OLS estimates, due to endogeneity (self-selection and/or reverse-causality)
•natural experimentse.g. Ahern and Dittmar (2012), Matsa and Miller (2013), Yang et al. (2019)•IV:
•Adams and Ferreira (2009) use social networks (instrument: fraction of male directors on the board
who sit on other boards on which there are female directors)
•Low et al. (2015) instrument: percentage of female (lower level) managers
•Smith et al. (2006) instrument: education level of spouses of other CEOs in same company
•Liu et al. (2014) instrument: percentage of women directors and the percentage of female
employment in its own industry
5
•no impacte.g. Harasheh and Provasi (2021), Ming and Eam (2016), Marinova et al. (2010), Joecks et al. (2013), Wiley and
Tormos (2018)
•meta-analysis by Post and Byron (2015):mixed evidence•How?
•inconsistent OLS estimates, due to endogeneity (self-selection and/or reverse-causality)
•natural experimentse.g. Ahern and Dittmar (2012), Matsa and Miller (2013), Yang et al. (2019)•IV:
•Adams and Ferreira (2009) use social networks (instrument: fraction of male directors on the board
who sit on other boards on which there are female directors)
•Low et al. (2015) instrument: percentage of female (lower level) managers
•Smith et al. (2006) instrument: education level of spouses of other CEOs in same company
•Liu et al. (2014) instrument: percentage of women directors and the percentage of female
employment in its own industry
5
What we know - empirical analysis
•no impacte.g. Harasheh and Provasi (2021), Ming and Eam (2016), Marinova et al. (2010), Joecks et al. (2013), Wiley and
Tormos (2018)
•meta-analysis by Post and Byron (2015):mixed evidence•How?
•inconsistent OLS estimates, due to endogeneity (self-selection and/or reverse-causality)
•natural experimentse.g. Ahern and Dittmar (2012), Matsa and Miller (2013), Yang et al. (2019)•IV:
•Adams and Ferreira (2009) use social networks (instrument: fraction of male directors on the board
who sit on other boards on which there are female directors)
•Low et al. (2015) instrument: percentage of female (lower level) managers
•Smith et al. (2006) instrument: education level of spouses of other CEOs in same company
•Liu et al. (2014) instrument: percentage of women directors and the percentage of female
employment in its own industry
5
•no impacte.g. Harasheh and Provasi (2021), Ming and Eam (2016), Marinova et al. (2010), Joecks et al. (2013), Wiley and
Tormos (2018)
•meta-analysis by Post and Byron (2015):mixed evidence•How?
•inconsistent OLS estimates, due to endogeneity (self-selection and/or reverse-causality)
•natural experimentse.g. Ahern and Dittmar (2012), Matsa and Miller (2013), Yang et al. (2019)•IV:
•Adams and Ferreira (2009) use social networks (instrument: fraction of male directors on the board
who sit on other boards on which there are female directors)
•Low et al. (2015) instrument: percentage of female (lower level) managers
•Smith et al. (2006) instrument: education level of spouses of other CEOs in same company
•Liu et al. (2014) instrument: percentage of women directors and the percentage of female
employment in its own industry
5
What we know - empirical analysis
•no impacte.g. Harasheh and Provasi (2021), Ming and Eam (2016), Marinova et al. (2010), Joecks et al. (2013), Wiley and
Tormos (2018)
•meta-analysis by Post and Byron (2015):mixed evidence•How?
•inconsistent OLS estimates, due to endogeneity (self-selection and/or reverse-causality)
•natural experimentse.g. Ahern and Dittmar (2012), Matsa and Miller (2013), Yang et al. (2019)•IV:
•Adams and Ferreira (2009) use social networks (instrument: fraction of male directors on the board
who sit on other boards on which there are female directors)
•Low et al. (2015) instrument: percentage of female (lower level) managers
•Smith et al. (2006) instrument: education level of spouses of other CEOs in same company
•Liu et al. (2014) instrument: percentage of women directors and the percentage of female
employment in its own industry
5
•no impacte.g. Harasheh and Provasi (2021), Ming and Eam (2016), Marinova et al. (2010), Joecks et al. (2013), Wiley and
Tormos (2018)
•meta-analysis by Post and Byron (2015):mixed evidence•How?
•inconsistent OLS estimates, due to endogeneity (self-selection and/or reverse-causality)
•natural experimentse.g. Ahern and Dittmar (2012), Matsa and Miller (2013), Yang et al. (2019)•IV:
•Adams and Ferreira (2009) use social networks (instrument: fraction of male directors on the board
who sit on other boards on which there are female directors)
•Low et al. (2015) instrument: percentage of female (lower level) managers
•Smith et al. (2006) instrument: education level of spouses of other CEOs in same company
•Liu et al. (2014) instrument: percentage of women directors and the percentage of female
employment in its own industry
5
What we know - empirical analysis
•no impacte.g. Harasheh and Provasi (2021), Ming and Eam (2016), Marinova et al. (2010), Joecks et al. (2013), Wiley and
Tormos (2018)
•meta-analysis by Post and Byron (2015):mixed evidence•How?
•inconsistent OLS estimates, due to endogeneity (self-selection and/or reverse-causality)
•natural experimentse.g. Ahern and Dittmar (2012), Matsa and Miller (2013), Yang et al. (2019)•IV:
•Adams and Ferreira (2009) use social networks (instrument: fraction of male directors on the board
who sit on other boards on which there are female directors)
•Low et al. (2015) instrument: percentage of female (lower level) managers
•Smith et al. (2006) instrument: education level of spouses of other CEOs in same company
•Liu et al. (2014) instrument: percentage of women directors and the percentage of female
employment in its own industry
5
•no impacte.g. Harasheh and Provasi (2021), Ming and Eam (2016), Marinova et al. (2010), Joecks et al. (2013), Wiley and
Tormos (2018)
•meta-analysis by Post and Byron (2015):mixed evidence•How?
•inconsistent OLS estimates, due to endogeneity (self-selection and/or reverse-causality)
•natural experimentse.g. Ahern and Dittmar (2012), Matsa and Miller (2013), Yang et al. (2019)•IV:
•Adams and Ferreira (2009) use social networks (instrument: fraction of male directors on the board
who sit on other boards on which there are female directors)
•Low et al. (2015) instrument: percentage of female (lower level) managers
•Smith et al. (2006) instrument: education level of spouses of other CEOs in same company
•Liu et al. (2014) instrument: percentage of women directors and the percentage of female
employment in its own industry
5
Our contribution
The existing literature has some important limitations:
1. →not likely to reflect overall global trends
⇒data on all types of firms from many European countries
2. →most likely biased
⇒novel identification strategy based on shift-share IV (SSIV)
MAIN AIM:
To persuade decision makers that increasing the number of women in corporate boards is not only
the matter of gender equality, but it significantly boosts firm’s performance. Those reluctant to
social arguments might be convinced by economic ones.
6
The existing literature has some important limitations:
1. →not likely to reflect overall global trends
⇒data on all types of firms from many European countries
2. →most likely biased
⇒novel identification strategy based on shift-share IV (SSIV)
MAIN AIM:
To persuade decision makers that increasing the number of women in corporate boards is not only
the matter of gender equality, but it significantly boosts firm’s performance. Those reluctant to
social arguments might be convinced by economic ones.
6
Data
•9 Orbis-Amadeus data waves (2002 - 2020)
•each wave include data up to 10 years in the past
•all types of firms: listed, public, private, SMEs
•scarce information on gender and lack of detailed information of board membership
•Final sample:
•#countries: 30-40
•#of unique firms: 4.5 million
•#observations: over 20 million
•#years:≈25
7
•9 Orbis-Amadeus data waves (2002 - 2020)
•each wave include data up to 10 years in the past
•all types of firms: listed, public, private, SMEs
•scarce information on gender and lack of detailed information of board membership
•Final sample:
•#countries: 30-40
•#of unique firms: 4.5 million
•#observations: over 20 million
•#years:≈25
7
Data - problems solved
•Function assignment:
•distinguish executives and non-executives
•how? with legal information from each country
•Industry codes:
•NACE change in 2007⇒problematic matching before and after
•50% from original correspondence table
•rest hand-made classification
•Gender assingment:
•heuristics for some countries
•World Gender Names Database
•result: 99% for men and 97% for women
8
•Function assignment:
•distinguish executives and non-executives
•how? with legal information from each country
•Industry codes:
•NACE change in 2007⇒problematic matching before and after
•50% from original correspondence table
•rest hand-made classification
•Gender assingment:
•heuristics for some countries
•World Gender Names Database
•result: 99% for men and 97% for women
8
Data - remaining problems
•How to measure firm performance?
⇒salesvs. growth of sales, profitability, productivity, liquidity, market-share?
•How to measure gender board diversity?
•share of women among all board members in a given industry, country, and year
•average of the shares of women across firms in a given industry, country and year
•share of firms in a given industry, country and year which report no single woman on their board(s)
•Which boards to analyze?
⇒separate analysis for supervisory and management boards
9
•How to measure firm performance?
⇒salesvs. growth of sales, profitability, productivity, liquidity, market-share?
•How to measure gender board diversity?
•share of women among all board members in a given industry, country, and year
•average of the shares of women across firms in a given industry, country and year
•share of firms in a given industry, country and year which report no single woman on their board(s)
•Which boards to analyze?
⇒separate analysis for supervisory and management boards
9
Data - remaining problems
•How to measure firm performance?
⇒salesvs. growth of sales, profitability, productivity, liquidity, market-share?
•How to measure gender board diversity?
•share of women among all board members in a given industry, country, and year
•average of the shares of women across firms in a given industry, country and year
•share of firms in a given industry, country and year which report no single woman on their board(s)
•Which boards to analyze?
⇒separate analysis for supervisory and management boards
9
•How to measure firm performance?
⇒salesvs. growth of sales, profitability, productivity, liquidity, market-share?
•How to measure gender board diversity?
•share of women among all board members in a given industry, country, and year
•average of the shares of women across firms in a given industry, country and year
•share of firms in a given industry, country and year which report no single woman on their board(s)
•Which boards to analyze?
⇒separate analysis for supervisory and management boards
9
Data - remaining problems
•How to measure firm performance?
⇒salesvs. growth of sales, profitability, productivity, liquidity, market-share?
•How to measure gender board diversity?
•share of women among all board members in a given industry, country, and year
•average of the shares of women across firms in a given industry, country and year
•share of firms in a given industry, country and year which report no single woman on their board(s)
•Which boards to analyze?
⇒separate analysis for supervisory and management boards
9
•How to measure firm performance?
⇒salesvs. growth of sales, profitability, productivity, liquidity, market-share?
•How to measure gender board diversity?
•share of women among all board members in a given industry, country, and year
•average of the shares of women across firms in a given industry, country and year
•share of firms in a given industry, country and year which report no single woman on their board(s)
•Which boards to analyze?
⇒separate analysis for supervisory and management boards
9
Gender board diversity
10
10
Gender board diversity
Statistic Management Board Supervisory Board All board positions
Average Share
N 60,301.00 30,478.00 64,227.00
Mean 0.20 0.21 0.21
Std. Dev. 0.17 0.21 0.14
25th perc. 0.09 0.05 0.13
Median 0.17 0.18 0.19
75th perc. 0.27 0.30 0.28
Weighted Share
N 60,301.00 30,478.00 64,227.00
Mean 0.20 0.21 0.22
Std. Dev. 0.16 0.20 0.13
25th perc 0.10 0.06 0.13
Median 0.18 0.19 0.20
75th perc 0.27 0.30 0.28
Firms with no women
N 60,301.00 30,478.00 64,227.00
Mean 0.74 0.62 0.67
Std. Dev. 0.21 0.31 0.20
25th perc. 0.65 0.45 0.55
Median 0.78 0.67 0.69
75th perc. 0.88 0.90 0.80
11
Statistic Management Board Supervisory Board All board positions
Average Share
N 60,301.00 30,478.00 64,227.00
Mean 0.20 0.21 0.21
Std. Dev. 0.17 0.21 0.14
25th perc. 0.09 0.05 0.13
Median 0.17 0.18 0.19
75th perc. 0.27 0.30 0.28
Weighted Share
N 60,301.00 30,478.00 64,227.00
Mean 0.20 0.21 0.22
Std. Dev. 0.16 0.20 0.13
25th perc 0.10 0.06 0.13
Median 0.18 0.19 0.20
75th perc 0.27 0.30 0.28
Firms with no women
N 60,301.00 30,478.00 64,227.00
Mean 0.74 0.62 0.67
Std. Dev. 0.21 0.31 0.20
25th perc. 0.65 0.45 0.55
Median 0.78 0.67 0.69
75th perc. 0.88 0.90 0.80
11
Gender board diversity in time
12
12
Gender board diversity across countries
13
13
Methodology
•We want to estimate:
log(sales)i,t=β0+γGBDi,t+βklog(total assetsi,t) +βllog(employmenti,t) +ηi+βtt+εi,t,
but due to endogeneity ofGBDi,twe use IV set-up.
•Shift-share IV, where instrumentzi,tis constructed as:
zi,t=
X
n
sharei,t,n×shockn
14
•We want to estimate:
log(sales)i,t=β0+γGBDi,t+βklog(total assetsi,t) +βllog(employmenti,t) +ηi+βtt+εi,t,
but due to endogeneity ofGBDi,twe use IV set-up.
•Shift-share IV, where instrumentzi,tis constructed as:
zi,t=
X
n
sharei,t,n×shockn
14
Methodology
•We want to estimate:
log(sales)i,t=β0+γGBDi,t+βklog(total assetsi,t) +βllog(employmenti,t) +ηi+βtt+εi,t,
but due to endogeneity ofGBDi,twe use IV set-up.
•Shift-share IV, where instrumentzi,tis constructed as:
zi,t=
X
n
sharei,t,n×shockn
14
•We want to estimate:
log(sales)i,t=β0+γGBDi,t+βklog(total assetsi,t) +βllog(employmenti,t) +ηi+βtt+εi,t,
but due to endogeneity ofGBDi,twe use IV set-up.
•Shift-share IV, where instrumentzi,tis constructed as:
zi,t=
X
n
sharei,t,n×shockn
14
Shift-share IV
•originated in Bartik (1991)
•recent empirical applications of SSIV across various fields in economics :
•Autor et al. (2013) - Chinese import exposure on manufacturing employment
•Boustan et al. (2013) - effect of income inequality on public taxation and expenditure
•Hummels et al. (2014) - impact of offshoring on wages
•Fitchett and Wesselbaum (2022) - impact of foreign aid payments on migration
•Qingyang (2023) - how industrial robots contribute to sustainable growth
•BUT notto analyze the impact of GBD on firm performance
15
•originated in Bartik (1991)
•recent empirical applications of SSIV across various fields in economics :
•Autor et al. (2013) - Chinese import exposure on manufacturing employment
•Boustan et al. (2013) - effect of income inequality on public taxation and expenditure
•Hummels et al. (2014) - impact of offshoring on wages
•Fitchett and Wesselbaum (2022) - impact of foreign aid payments on migration
•Qingyang (2023) - how industrial robots contribute to sustainable growth
•BUT notto analyze the impact of GBD on firm performance
15
Shift-share IV
•originated in Bartik (1991)
•recent empirical applications of SSIV across various fields in economics :
•Autor et al. (2013) - Chinese import exposure on manufacturing employment
•Boustan et al. (2013) - effect of income inequality on public taxation and expenditure
•Hummels et al. (2014) - impact of offshoring on wages
•Fitchett and Wesselbaum (2022) - impact of foreign aid payments on migration
•Qingyang (2023) - how industrial robots contribute to sustainable growth
•BUT notto analyze the impact of GBD on firm performance
15
•originated in Bartik (1991)
•recent empirical applications of SSIV across various fields in economics :
•Autor et al. (2013) - Chinese import exposure on manufacturing employment
•Boustan et al. (2013) - effect of income inequality on public taxation and expenditure
•Hummels et al. (2014) - impact of offshoring on wages
•Fitchett and Wesselbaum (2022) - impact of foreign aid payments on migration
•Qingyang (2023) - how industrial robots contribute to sustainable growth
•BUT notto analyze the impact of GBD on firm performance
15
Shift-share IV - details
zi,t=
X
n
sharei,t,n×shockn
•shock (shift) - global change in time, e.g. a universal trend of increasing number of women in
boards
•share - firm’siexposure to the shock
•shocks vary at a different level than the shares
•IMPORTANT: panel structure not enough, there must be an additional level of variation
16
zi,t=
X
n
sharei,t,n×shockn
•shock (shift) - global change in time, e.g. a universal trend of increasing number of women in
boards
•share - firm’siexposure to the shock
•shocks vary at a different level than the shares
•IMPORTANT: panel structure not enough, there must be an additional level of variation
16
Shift-share IV - examples
zi=
X
n
sharei,n×shockn
•Card (2001):x- relative employment (natives vs. immigrants) in regioni,shock- national
immigration growth from origin countryn,share- lagged shares of migrants from originnin
regioni
•Blanchard and Katz (1992):x- employment growth in regioni,shock- national growth of
industryn,share- lagged employment shares of industry in a region
17
zi=
X
n
sharei,n×shockn
•Card (2001):x- relative employment (natives vs. immigrants) in regioni,shock- national
immigration growth from origin countryn,share- lagged shares of migrants from originnin
regioni
•Blanchard and Katz (1992):x- employment growth in regioni,shock- national growth of
industryn,share- lagged employment shares of industry in a region
17
Shift-share IV - examples
zi=
X
n
sharei,n×shockn
•Card (2001):x- relative employment (natives vs. immigrants) in regioni,shock- national
immigration growth from origin countryn,share- lagged shares of migrants from originnin
regioni
•Blanchard and Katz (1992):x- employment growth in regioni,shock- national growth of
industryn,share- lagged employment shares of industry in a region
17
zi=
X
n
sharei,n×shockn
•Card (2001):x- relative employment (natives vs. immigrants) in regioni,shock- national
immigration growth from origin countryn,share- lagged shares of migrants from originnin
regioni
•Blanchard and Katz (1992):x- employment growth in regioni,shock- national growth of
industryn,share- lagged employment shares of industry in a region
17
Shift-share IV - well established properties
•similar to standard IV: relevance & exogeneity conditions
•shock exogeneity - Borusyak et al. (2022), Ad˜ao et al. (2019) - similar to RCT framework
•share exogeneity - Goldsmith-Pinkham et al. (2020) - similar to standard IV/2SLS
•ex ante decision
18
•similar to standard IV: relevance & exogeneity conditions
•shock exogeneity - Borusyak et al. (2022), Ad˜ao et al. (2019) - similar to RCT framework
•share exogeneity - Goldsmith-Pinkham et al. (2020) - similar to standard IV/2SLS
•ex ante decision
18
Shift-share IV - well established properties
•similar to standard IV: relevance & exogeneity conditions
•shock exogeneity - Borusyak et al. (2022), Ad˜ao et al. (2019) - similar to RCT framework
•share exogeneity - Goldsmith-Pinkham et al. (2020) - similar to standard IV/2SLS
•ex ante decision
18
•similar to standard IV: relevance & exogeneity conditions
•shock exogeneity - Borusyak et al. (2022), Ad˜ao et al. (2019) - similar to RCT framework
•share exogeneity - Goldsmith-Pinkham et al. (2020) - similar to standard IV/2SLS
•ex ante decision
18
Our SSIV
SSIV formulated as:
IVk,t=
X
c
#womenc,k,t−10
#managersk,t−10
| {z }
share
·∆women presence measure
c,t,t−10
| {z }
shift
OR
IVc,t=
X
k
#womenk,c,t−10
#managersc,t−10
| {z }
share
·∆women presence measure
k,t,t−10
| {z }
shift
c- country,k- sector,t- time.
•women presence measure: can beaverage women shareorweighted average women shareor
average number of women
•shift factor: change over 10 years (or compare to one initial point in the past)
•share: lagged (initial)
19
SSIV formulated as:
IVk,t=
X
c
#womenc,k,t−10
#managersk,t−10
| {z }
share
·∆women presence measure
c,t,t−10
| {z }
shift
OR
IVc,t=
X
k
#womenk,c,t−10
#managersc,t−10
| {z }
share
·∆women presence measure
k,t,t−10
| {z }
shift
c- country,k- sector,t- time.
•women presence measure: can beaverage women shareorweighted average women shareor
average number of women
•shift factor: change over 10 years (or compare to one initial point in the past)
•share: lagged (initial)
19
Preliminary results
(1) (2) (3) (4) (5) (6) (7) (8) (9)
OLS IV IV OLS IV IV OLS IV IV
Dependent variable: log(operating revenue)
Gender Board Diversity 0.005 0.283
∗∗∗
0.154
∗∗
0.005 0.120 -0.077 -0.008 0.253
∗∗
0.312
∗∗
(0.005) (0.051) (0.075) (0.005) (0.089) (0.135) (0.012) (0.111) (0.139)
log(total assets) 0.527
∗∗∗
0.527
∗∗∗
0.560
∗∗∗
0.523
∗∗∗
0.523
∗∗∗
0.556
∗∗∗
0.464
∗∗∗
0.465
∗∗∗
0.509
∗∗∗
(0.003) (0.003) (0.005) (0.003) (0.003) (0.005) (0.009) (0.008) (0.012)
log(employment) 0.447
∗∗∗
0.446
∗∗∗
0.406
∗∗∗
0.446
∗∗∗
0.446
∗∗∗
0.401
∗∗∗
0.476
∗∗∗
0.475
∗∗∗
0.413
∗∗∗
(0.003) (0.003) (0.005) (0.003) (0.003) (0.005) (0.009) (0.008) (0.012)
Dependent variable: GBD (first stage)
iv -0.608 -0.591 -0.183 -0.158 -0.328 -0.318
[0.074] [0.083] [0.017] [0.018] [0.043] [0.040]
iv
2
0.056 0.052 0.009 0.007 0.005 0.004
[0.018] [0.017] [0.002] [0.002] [0.001] [0.001]
Instruments F test 143.540 80.189 135.702 65.608 28.851 30.984
[0.000] [0.000] [0.000] [0.000] [0.000] [0.000]
Overid. test 1.509 1.550 0.021 0.309 2.142 1.766
[0.219] [0.213] [0.886] [0.578] [0.143] [0.184]
Endogeneity test 28.548 3.441 1.700 0.741 5.054 4.163
[0.000] [0.064] [0.192] [0.389] [0.025] [0.041]
First Stage F stat. 1752.22 1345.50 1496.42 1146.70 354.37 298.28
Firm FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
Year trend Yes Yes Yes Yes Yes Yes Yes Yes Yes
Country weights Yes Yes Yes
Board definition all all all all all all MB MB MB
NACE level 2 digits 2 digits 2 digits 4 digits 4 digits 4 digits 2 digits 2 digits 2 digits
Observations 2,198,743 2,198,743 2,198,743 2,125,301 2,125,301 2,125,301 369,921 369,921 369,921
Notes: s.e. in parentheses, p-values in brackets, * p¡0.1, ** p¡0.05, *** p¡0.01. Standard errors are clustered at the firm level.
20
(1) (2) (3) (4) (5) (6) (7) (8) (9)
OLS IV IV OLS IV IV OLS IV IV
Dependent variable: log(operating revenue)
Gender Board Diversity 0.005 0.283
∗∗∗
0.154
∗∗
0.005 0.120 -0.077 -0.008 0.253
∗∗
0.312
∗∗
(0.005) (0.051) (0.075) (0.005) (0.089) (0.135) (0.012) (0.111) (0.139)
log(total assets) 0.527
∗∗∗
0.527
∗∗∗
0.560
∗∗∗
0.523
∗∗∗
0.523
∗∗∗
0.556
∗∗∗
0.464
∗∗∗
0.465
∗∗∗
0.509
∗∗∗
(0.003) (0.003) (0.005) (0.003) (0.003) (0.005) (0.009) (0.008) (0.012)
log(employment) 0.447
∗∗∗
0.446
∗∗∗
0.406
∗∗∗
0.446
∗∗∗
0.446
∗∗∗
0.401
∗∗∗
0.476
∗∗∗
0.475
∗∗∗
0.413
∗∗∗
(0.003) (0.003) (0.005) (0.003) (0.003) (0.005) (0.009) (0.008) (0.012)
Dependent variable: GBD (first stage)
iv -0.608 -0.591 -0.183 -0.158 -0.328 -0.318
[0.074] [0.083] [0.017] [0.018] [0.043] [0.040]
iv
2
0.056 0.052 0.009 0.007 0.005 0.004
[0.018] [0.017] [0.002] [0.002] [0.001] [0.001]
Instruments F test 143.540 80.189 135.702 65.608 28.851 30.984
[0.000] [0.000] [0.000] [0.000] [0.000] [0.000]
Overid. test 1.509 1.550 0.021 0.309 2.142 1.766
[0.219] [0.213] [0.886] [0.578] [0.143] [0.184]
Endogeneity test 28.548 3.441 1.700 0.741 5.054 4.163
[0.000] [0.064] [0.192] [0.389] [0.025] [0.041]
First Stage F stat. 1752.22 1345.50 1496.42 1146.70 354.37 298.28
Firm FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
Year trend Yes Yes Yes Yes Yes Yes Yes Yes Yes
Country weights Yes Yes Yes
Board definition all all all all all all MB MB MB
NACE level 2 digits 2 digits 2 digits 4 digits 4 digits 4 digits 2 digits 2 digits 2 digits
Observations 2,198,743 2,198,743 2,198,743 2,125,301 2,125,301 2,125,301 369,921 369,921 369,921
Notes: s.e. in parentheses, p-values in brackets, * p¡0.1, ** p¡0.05, *** p¡0.01. Standard errors are clustered at the firm level.
20
Problems
•negative predictors of proportions from the first stage
•solution: zero-one inflated beta distribution
Yit∼0 ( or 1) with probability 1−pit
Yit∼Beta(µitφ,(1−µit)φ) with probabilitypit
•use MLE in:
logit(pit) =α0+X
T
itα+ait
logit(µit) =γ0+W
T
itγ+bit
•BUTforbidden regressions: nonlinear first stage, solution: e.g. Adams et al. (2009) use a
three-step procedure:
21
•negative predictors of proportions from the first stage
•solution: zero-one inflated beta distribution
Yit∼0 ( or 1) with probability 1−pit
Yit∼Beta(µitφ,(1−µit)φ) with probabilitypit
•use MLE in:
logit(pit) =α0+X
T
itα+ait
logit(µit) =γ0+W
T
itγ+bit
•BUTforbidden regressions: nonlinear first stage, solution: e.g. Adams et al. (2009) use a
three-step procedure:
21
Problems
•negative predictors of proportions from the first stage
•solution: zero-one inflated beta distribution
Yit∼0 ( or 1) with probability 1−pit
Yit∼Beta(µitφ,(1−µit)φ) with probabilitypit
•use MLE in:
logit(pit) =α0+X
T
itα+ait
logit(µit) =γ0+W
T
itγ+bit
•BUTforbidden regressions: nonlinear first stage, solution: e.g. Adams et al. (2009) use a
three-step procedure:
21
•negative predictors of proportions from the first stage
•solution: zero-one inflated beta distribution
Yit∼0 ( or 1) with probability 1−pit
Yit∼Beta(µitφ,(1−µit)φ) with probabilitypit
•use MLE in:
logit(pit) =α0+X
T
itα+ait
logit(µit) =γ0+W
T
itγ+bit
•BUTforbidden regressions: nonlinear first stage, solution: e.g. Adams et al. (2009) use a
three-step procedure:
21
Further challenges
•single instrumental variable→cannot verify the instrumental exogeneity condition (e.g. via
overidentifying restrictions test)
•for identification all shares (within the instrument) need to fulfill an exclusion restriction, which
in practice often fails
•our trick - useinstrumentandinstrument
2
•alternative solution: Apfel (2022) - ML procedure based on (i) adaptive Lasso or (ii) Confidence
Interval Method
22
•single instrumental variable→cannot verify the instrumental exogeneity condition (e.g. via
overidentifying restrictions test)
•for identification all shares (within the instrument) need to fulfill an exclusion restriction, which
in practice often fails
•our trick - useinstrumentandinstrument
2
•alternative solution: Apfel (2022) - ML procedure based on (i) adaptive Lasso or (ii) Confidence
Interval Method
22
Conclusions
What do we have so far?
1.
2.Positiveimpact of GBD on firm performance→human capital theory & reseource dependence
theory & agency theory effects dominate.
23
What do we have so far?
1.
2.Positiveimpact of GBD on firm performance→human capital theory & reseource dependence
theory & agency theory effects dominate.
23
Thank you!
Questions or suggestions?
w: grape.org.pl
t: grapeorg
f: grape.org
e: [email protected]
24
Questions or suggestions?
w: grape.org.pl
t: grapeorg
f: grape.org
e: [email protected]
24
Categories
GeneralDownload
Download Slideshow Get the original presentation fileQuick Actions
Statistics
- Views 99
- Slides 42
- Age 575 days
Related Slideshows
22
Pray For The Peace Of Jerusalem and You Will Prosper
RodolfoMoralesMarcuc
32 views
26
Don_t_Waste_Your_Life_God.....powerpoint
chalobrido8
34 views
31
VILLASUR_FACTORS_TO_CONSIDER_IN_PLATING_SALAD_10-13.pdf
JaiJai148317
31 views
14
Fertility awareness methods for women in the society
Isaiah47
30 views
35
Chapter 5 Arithmetic Functions Computer Organisation and Architecture
RitikSharma297999
28 views
5
syakira bhasa inggris (1) (1).pptx.......
ourcommunity56
30 views