Female entrepreneurship, Firm Productivity and Gender Pay Gap: Evidence from the UK
Yannis Galanakis
King’s College London
TPI Post-doc researcher workshop, Manchester
8 December 2022
This postdoc
- “Big picture” project: Data development
- Linking exercise
- FAME/Companies House
- IDBR
- HMRC
- Linking exercise
- Today’s presentation: smaller project linking FAME/Gender Pay Gap Service data
Motivation
Why Gender Pay Gap?
Government Equalities Office (GEO) vs. ONS
- Common definition of GPG: \[\frac{w_m-w_f}{w_m}\times 100\]
- Is it the same story?
Government Equalities Office (GEO)
- Since April 2017, companies with more than 250 employees report their GPG
- <250 employees on voluntary basis
- Gender pay gap calculations are based on employer payroll data drawn from a specific date each year (aka “snapshot date”)
- Most of public authority employers: 31 March
- private, voluntary, and all other public authority employers: 4 April
- How does pay transparency affect GPG and firm outcomes?
- Evidence from Denmark:
(Bennedsen et al. 2022)- GPG declines by 2 percentage points and
- no effect on firm profitability
- Evidence from Denmark:
Why female entrepreneurship?
- Female participation on boards is 33% in top public firms (Goodley, 2020) and 17.5% in SMEs (Shehata et al., 2017)
- US: 26% in top listed firms and 7% in private firms (Rivera et al., 2019)
- Female representation on committees: a more effective measure of board gender diversity, and a more direct effect on firm performance
- Diverse boards: better matching of skills to functions,
- Appointment of female directors to decision-making committees: competitive advantage
- How do women directors affect pay menus? Are they “fairer” employers?
Literature
GPG due to individual characteristics
(Dias, Joyce and Parodi, 2018; Cukrowska-Torzewska and Lovasz, 2016; Christofides et al., 2013; Manning and Swaffield, 2008)GPG due to firm-specific features:
- recruitment policies, training practices and employee allocation
(Heinze and Wolf, 2010; Simónand Russell, 2005) - Impact of workplace
(Mumford and Smith, 2016; 2007)
- recruitment policies, training practices and employee allocation
Firms can’t afford to discriminate due to market pressure
(Magda and Cukrowska-Torzewska, 2019; Li and Dong, 2011)Productivity dispersion decreases with competition
(Olley and Pakes, 1996; Syverson, 2004; Bloom and Van Reenen, 2007; Fox and Smeets, 2011)Board gender diversity and firm performance
(Sattar et al. 2022; Green and Homroy , 2018; Gregory-Smith et al., 2014)Directors and productivity
(Bamdiera et al. 2020)
This paper in a nutshell
Question
How do pay-menus treat employees?
- Do more productive firms offer an equal pay menu to both their male and female employees?
- Are female directors “fairer” employers?
Preliminary findings
- more productive firms have fewer female directors and employees
- firm productivity is positively correlated to GPG
- Mechanism:- Female under-representation: fewer women take up senior roles
- women are paid less than men in senior roles
- Female under-representation: fewer women take up senior roles
- companies with more female directors have a lower GPG by 4.2 percentage points
Data
- Gender Pay Gap
- Source: Government Equalities Office; gender-pay-gap.service.gov.uk
- Time: 2017/18 – today; Last accessed on: 30 Sept 2022
- Information: Company Number (harmonised with CH), address, SIC codes, measures of GPG:
- Mean/median % difference of hourly pay between men and women, \(\frac{w_m-w_f}{w_m} \cdot 100\)
- % of (fe)male employees paid a bonus
- % of (fe)male employees in each pay quarter
- Employer size
- Report due and submitted dates
- Directors and company financial details
- Source: FAME (Bureau van Dijk; BvD)
- Information: Companies registered with CH - financial details, corporate structures, shareholders and subsidiaries, total number of employees
Data II
Descriptive stats
- GEO: 11,193 unique companies report GPG
- FAME:
- 11,086 companies (99%) - 329 become inactive over the years
- 258,505 unique directors: individuals and companies
| gender | min | q1 | median | mean | sd | q3 | max | No of obs. |
|---|---|---|---|---|---|---|---|---|
| female | 0 | 1 | 3 | 5.63 | 7.62 | 7 | 195 | 55,006 |
| male | 0 | 8 | 16 | 19.95 | 17.07 | 27 | 297 | 195,040 |
| NA | 0 | 0 | 0 | 0.12 | 0.50 | 0 | 21 | 1,307 |
| Non-individual director | 0 | 0 | 0 | 0.67 | 1.25 | 1 | 31 | 7,152 |
|
Note: If director's gender is missing, `genderizeR` package imputes it based on their first name. Source: Own elaboration based on FAME and genderize.io |
Methodology
Calculate the share of current female directors for each company \(i\) \[\frac{n^f_i}{n^m_i + n^f_i}\] where \(n\) is the number of current directors, \(m\) stands for male and \(f\) for female
Reduced form: company \(i\), in year \(t\), Local Authority \(k\) and 2-digit SIC \(j\) \[\text{GPG}_{itkj} =\beta_0+\beta_1 \cdot \text{share of female directors}_{itkj} + \beta_2 \cdot log(\text{turnover per employee})_{itkj}+\beta_3 \boldsymbol x_{itkj}+\vartheta_t+\eta_k\cdot\delta_j+v_{itkj} \]
where \(\boldsymbol x\) is a vector of controls: firm age (and square), share of female employees, size of employer, profit per employee, liquidity ratio
Result I
Female directors vs. productivity
- Negative relationship: More productive firms have fewer female directors
Dep. var.: log(Turnover per employee) |
||||||
|---|---|---|---|---|---|---|
| Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | Model 6 | |
| Share of current female directors | −1.196*** | −0.255*** | −0.275*** | −0.139*** | −0.143*** | −0.165*** |
| (0.088) | (0.054) | (0.051) | (0.053) | (0.052) | (0.052) | |
| Num.Obs. | 54975 | 54975 | 54343 | 54102 | 54102 | 53835 |
| R2 Adj. | 0.100 | 0.426 | 0.433 | 0.458 | 0.458 | 0.488 |
| FE: local_authority_code | X | X | X | X | X | X |
| FE: year | X | X | X | X | X | X |
| FE: SIC.2 | X | X | X | X | X | |
| * p < 0.1, ** p < 0.05, *** p < 0.01 | ||||||
| Robust s.e. clustered at Local Authority level. Additional controls by specification: Firm age (and its square; models 3-6), share of female employees (models 4-6), employer size (models 5-6), Profit per employee in GBP (model 6) | ||||||
Result II
Firm productivity vs. GPG
- An 1% increase in turnover increases GPG by 2.3 p.p.
- Mechanism:
- Women under-represented in senior positions – fewer women take up directorships
- Even if women are appointed, their earnings are lower than their male counterparts
- GPG comes from between-firms differences – i.e. women are occupied in lower-paying companies
Dep. var.: Median difference of hourly pay between men and women |
||||||
|---|---|---|---|---|---|---|
| Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | Model 6 | |
| log(Turnover per employee) | 0.019*** | 0.020*** | 0.019*** | 0.022*** | 0.023*** | 0.023*** |
| (0.001) | (0.001) | (0.001) | (0.001) | (0.001) | (0.001) | |
| Num.Obs. | 54975 | 54343 | 54343 | 54102 | 53835 | 53704 |
| R2 Adj. | 0.252 | 0.257 | 0.257 | 0.264 | 0.265 | 0.265 |
| FE: local_authority_code | X | X | X | X | X | X |
| FE: year | X | X | X | X | X | X |
| FE: SIC.2 | X | X | X | X | X | X |
| * p < 0.1, ** p < 0.05, *** p < 0.01 | ||||||
| Robust s.e. clustered at Local Authority level. Additional controls by specification: Firm age (and its square; models 2-6), share of female employees (models 3-6), employer size (models 4-6), Profit per employee in GBP (models 5-6), Liquidity ratio (model 6) | ||||||
Is representation an issue in the UK?
| Quartile | male | female |
|---|---|---|
| Lower | 47.0 | 53.0 |
| LowerMiddle | 52.0 | 48.0 |
| UpperMiddle | 58.0 | 42.0 |
| Top | 66.7 | 33.3 |
|
Note: The figures show the median percentage share of men and women across firms by quartile of payment. Source: Own elaboration based on gender-pay-gap.service.gov.uk |
Results III
Female directors vs. GPG
- An increase in the share of current female directors by 1pp decreases GPG by 4.2 pp
Dep. var.: Median difference of hourly pay between men and women |
||||||
|---|---|---|---|---|---|---|
| Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | Model 6 | |
| Share of current female directors | −0.023** | −0.003 | −0.031*** | −0.042*** | −0.042*** | −0.041*** |
| (0.010) | (0.009) | (0.008) | (0.008) | (0.008) | (0.008) | |
| log(Turnover per employee) | 0.016*** | 0.019*** | 0.022*** | 0.023*** | 0.023*** | |
| (0.002) | (0.001) | (0.001) | (0.001) | (0.001) | ||
| Num.Obs. | 54975 | 54975 | 54343 | 54102 | 53835 | 53704 |
| R2 Adj. | 0.044 | 0.054 | 0.257 | 0.265 | 0.266 | 0.266 |
| FE: local_authority_code | X | X | X | X | X | X |
| FE: year | X | X | X | X | X | X |
| FE: SIC.2 | X | X | X | X | ||
| * p < 0.1, ** p < 0.05, *** p < 0.01 | ||||||
| Robust s.e. clustered at Local Authority level. Additional controls by specification: Firm age (and its square; models 3-6), Employer size (models 4-6), Share of female employees (models 4-6), Profit per employee (models 5-6), Liquidity ratio (model 6) | ||||||
Conclusions
More productive firms have fewer female directors and employees
More productive firms report a higher GPG, because
- fewer women take up senior roles and they are paid less than men in senior roles, ceteris paribus
- Female under-representation: if women were equally represented in senior roles, their impact on pay menus could be greater
More diverse companies (i.e. with more female directors) have a lower GPG by 4.2 percentage points
Future work
- Sectoral mobility of directors
- People with Significant Control, shareholders gender and GPG
- Does nationality play any role?
- GPG: where does it come from? within- vs. between-firms differences
- How does the policy of reporting the GPG in 2017 changed the GPG incidence in the UK?
- Employee – employer data
- ASHE – feasible matching but small sample
- FAME-IDBR linking and HMRC
- Employee – employer data