Relating firms’ gender-based board diversity and progress towards net-zero

Introduction

The World Economic Forum Global Risks Report 2021-2022 classifies “climate action failure”, “extreme weather” and “biodiversity loss” as potentially severe risks for the next decade. All sectors of civil society are attempting to minimise impact on the environment and forestall climate catastrophe. Additionally, more corporates are disclosing their environmental, social, and governance (ESG) performance for public scrutiny.

The topic of gender diversity has garnered media attention as firms champion the positive impact women in leadership positions can have on a company’s performance. McKinsey’s 2007 report titled Women Matter finds that companies where women are strongly represented at board and top-management levels are generally high performing. Specifically, the report details that companies in the top quartile in terms of women on the board outperformed those with the least in terms of measures such as Return on Equity and Return on Invested Capital. Furthermore, a 2017 McKinsey report discusses how companies with three or more women in their executive committees scored better on dimensions of organisational performance such as accountability, innovation, and motivation. Both reports urge stronger female representation and discuss the barriers that prevent women from pursuing leadership positions. Gender diversity mandates are now becoming the norm in the U.S., and laws in European countries such as Norway and Spain require that women comprise at least 40% of boards at publicly listed companies.

Mckinsey also advocates sustainability as a value-creator for firms. In fact, investors are increasingly looking at firms’ ESG metrics and demanding strong sustainability propositions, which “safeguard” a firm’s long-term success. A study at NYU Stern explores the positive correlation between a firm’s commitment to sustainability and financial performance. As evidenced, better ESG performance corresponds with higher equity returns and a reduction in downside risk, according to a 2019 McKinsey report.

With both gender diversity and sustainability metrics seeming to positively influence firm performance, the question arises: does sustainability relate to gender diversity? More specifically, does climate change performance increase with more women on the Board? The team is interested in revealing if there is any connection between the presence of women on the Board of Directors and the company’s commitment to climate targets (such as short-, medium- and long-term net-zero targets) and disclosure.

Research Topic

To what extent does gender diversity in corporate leadership impact a firm’s commitment and progress on climate targets? The project aims to study the relationship between the involvement of women in corporate leadership in European listed companies and the firms’ commitments to climate targets.

Data

A range of data sources were consulted to support the research topic, as outlined below.

Climate Action 100+

Climate Action 100+ is an investor coalition engaging companies with a combined USD 60 trillion in AUM on their decarbonisation strategies and climate disclosures. One initiative is an assessment of 167 companies, which account for over 80% of corporate industrial greenhouse gas emissions, across ten disclosure indicators, measuring the companies’ commitments to adopting net zero operations. Of these 167 companies, 52 companies headquartered in European countries were chosen to assess the relationship between female corporate leadership involvement and commitment to net-zero progress.

European Women on Boards (EWOB) - Gender Diversity Index 2020

For standardised Board Diversity data, the EWOB Gender Diversity Index Report (2020) was used. EWOB is a non-profit that reports on women’s participation in corporate governance in listed European firms. The Index is an aggregate indicator that reflects the share of women in leadership positions, on boards, and in executive functions. Accordingly, the scores for each of the aforementioned 52 companies were extracted. While there are several high scorers, overall, the data suggests that the chosen firms’ leadership is skewed towards men. To complement these scores, country-level data was sourced from the European Institute for Gender Equality.

European Institute for Gender Equality (EIGE) - The Gender Equality Index tool

EIGE describes itself as “an autonomous body of the European Union, established to contribute to and strengthen the promotion of gender equality”. The index tool measures the progress of gender equality by scoring countries across seven criteria.

The Climate Change Performance Index (CCPI)

CCPI is an annually published tool which assesses countries’ climate mitigation efforts across four categories - GHG emissions, Renewable Energy, Energy Use, and Climate Policy. This allows for an understanding of progress on climate targets at a country level.

Hofstede Culture Index

To assess the role that culture plays in firms’ climate action and disclosure decisions, the latest Hofstede Culture Index was extracted from the Hofstede Insights website and applied to the firms based on their headquarter locations. Relevant dimensions of culture include individualism (does society prefer a loosely knit social framework or a tight-knit collective network?), masculinity (is there a preference for heroism and other traits considered as “masculine”?), uncertainty avoidance (how comfortable are members of society with uncertainty?) and long-term orientation (how does society prioritise traditions and view change?).

Methodology

To cross-reference the sources listed above, the dataset was narrowed down to 52 companies headquartered in Europe, providing a sufficient sample size for quantitative research, and comparable analysis. These companies were included in both the EWOB and Climate Action 100+ datasets.

The study utilises several quantitative analysis methods including linear regression, dummy variables, control variables, and joint hypothesis in assessing the data available and deriving insights about the relationship between the variables.

The Climate Action 100+ assessment evaluated companies in a three-level grading scale, “Yes”, “No” or “Partial”, against several dimensions including net-zero by 2050 ambition, long-term (2036-2050) GHG reduction target, medium-term (2026-2035) reduction target, short-term (present date to 2050) target, decarbonisation strategy, capital allocation on alignment, climate policy engagement, climate governance, and TCFD Disclosure. We converted the data set from a categorical value into numerical value (0=No, 1=Partial, 2=Yes) to reflect the degree of a company’s climate action.

For EWOB’s Gender Diversity Index, the companies’ raw scores were used as presented in the original report. In the rare case that a company was qualitatively assessed but not assigned a numerical score, the average GDI score for the report’s entire dataset was used.

To test our hypothesis, for the case of climate score and gender diversity, we are fitting two types of models of the following form:

\[ Performance_i=\beta_0 + \beta_1 GenderDiversity_i + \epsilon_i\] \[ ClimateScore_i=\beta_0 + \beta_1 GenderDiversity_i + \beta_2 X_i + \epsilon_i\]

where \(Performance_i\) are different outcome variables such as net-zero ambitions and decarbonisation strategy, \(ClimateScore_i\) is the overall added score of each individual performance indicator, \(GenderDiversity_i\) is the proportion of women on a company’s board, and \(X_i\) are various sets of control variables (either cultural dimensions or countries’ indexes).

Results

To analyse the relationship between a firm’s climate action and its gender diversity, each of the Climate 100+ performance indicators were regressed against EWOB’s gender diversity scores.

\[ Performance_i=\beta_0 + \beta_1 GenderDiversity_i + \epsilon_i\]

As seen in the plots above, there is no visible relationship between the score assigned to each dependent variable and gender diversity. This is also reflected in the summary of the regression models run.

Capital allocation alignment appears to be the only variable with a significant estimate (at 5% significance level) of 0.54, i.e., for every point increase in a company’s gender diversity, the climate score’s chance of going up increases by 0.54 percentage points. However, most firms are clustered at the Level 0 categorisation, indicating that the minimum benchmarks for this assessment metric were not satisfied by the relevant firms.

## [1] "Net-zero GHG Emissions by 2050 (or sooner) ambition"
## 
## Call:
## lm(formula = df[[i]] ~ GenderDiv, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -1.2746 -0.2567 -0.2169  0.7639  0.8391 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)   
## (Intercept)   1.1294     0.4115   2.745  0.00839 **
## GenderDiv     0.1750     0.6862   0.255  0.79980   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.7641 on 50 degrees of freedom
## Multiple R-squared:  0.001298,   Adjusted R-squared:  -0.01868 
## F-statistic: 0.065 on 1 and 50 DF,  p-value: 0.7998
## 
## [1] "Long-term (2036-2050) GHG reduction target(s)"
## 
## Call:
## lm(formula = df[[i]] ~ GenderDiv, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -1.2564 -0.2367 -0.2242  0.7610  0.8101 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)   
## (Intercept)   1.1715     0.3508   3.339  0.00159 **
## GenderDiv     0.1024     0.5850   0.175  0.86180   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.6514 on 50 degrees of freedom
## Multiple R-squared:  0.000612,   Adjusted R-squared:  -0.01938 
## F-statistic: 0.03062 on 1 and 50 DF,  p-value: 0.8618
## 
## [1] "Medium-term (2026-2035) GHG reduction target(s)"
## 
## Call:
## lm(formula = df[[i]] ~ GenderDiv, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -1.1831 -0.1632 -0.1487 -0.1160  0.8659 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)   1.0863     0.2925   3.714 0.000515 ***
## GenderDiv     0.1166     0.4878   0.239 0.812006    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.5432 on 50 degrees of freedom
## Multiple R-squared:  0.001142,   Adjusted R-squared:  -0.01884 
## F-statistic: 0.05717 on 1 and 50 DF,  p-value: 0.812
## 
## [1] "Short-term (up to 2025) GHG reduction target(s)"
## 
## Call:
## lm(formula = df[[i]] ~ GenderDiv, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -0.8359 -0.5900  0.2541  0.3875  1.4066 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)
## (Intercept)   0.2332     0.3183   0.733    0.467
## GenderDiv     0.6928     0.5307   1.305    0.198
## 
## Residual standard error: 0.591 on 50 degrees of freedom
## Multiple R-squared:  0.03296,    Adjusted R-squared:  0.01362 
## F-statistic: 1.704 on 1 and 50 DF,  p-value: 0.1977
## 
## [1] "Decarbonisation strategy"
## 
## Call:
## lm(formula = df[[i]] ~ GenderDiv, data = df)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -1.02827 -0.03964  0.10499  0.19924  1.18624 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)  
## (Intercept)   0.4887     0.2802   1.744   0.0872 .
## GenderDiv     0.6500     0.4672   1.391   0.1703  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.5203 on 50 degrees of freedom
## Multiple R-squared:  0.03727,    Adjusted R-squared:  0.01801 
## F-statistic: 1.936 on 1 and 50 DF,  p-value: 0.1703
## 
## [1] "Capital allocation allignment"
## 
## Call:
## lm(formula = df[[i]] ~ GenderDiv, data = df)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.37867 -0.12496 -0.08290 -0.02863  0.91439 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)  
## (Intercept)  -0.2183     0.1552  -1.406   0.1658  
## GenderDiv     0.5427     0.2589   2.097   0.0411 *
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.2882 on 50 degrees of freedom
## Multiple R-squared:  0.0808, Adjusted R-squared:  0.06242 
## F-statistic: 4.395 on 1 and 50 DF,  p-value: 0.04112
## 
## [1] "Climate policy engagement"
## 
## Call:
## lm(formula = df[[i]] ~ GenderDiv, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -0.8901 -0.7019  0.1996  0.2425  1.2425 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)  
## (Intercept)   0.5624     0.3092   1.819   0.0749 .
## GenderDiv     0.3901     0.5157   0.756   0.4529  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.5742 on 50 degrees of freedom
## Multiple R-squared:  0.01132,    Adjusted R-squared:  -0.008458 
## F-statistic: 0.5722 on 1 and 50 DF,  p-value: 0.4529
## 
## [1] "Climate Governance"
## 
## Call:
## lm(formula = df[[i]] ~ GenderDiv, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -1.3692 -0.3180 -0.2591  0.6230  0.7619 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)   
## (Intercept)   1.0233     0.2960   3.457  0.00112 **
## GenderDiv     0.5241     0.4936   1.062  0.29340   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.5496 on 50 degrees of freedom
## Multiple R-squared:  0.02205,    Adjusted R-squared:  0.002495 
## F-statistic: 1.128 on 1 and 50 DF,  p-value: 0.2934
## 
## [1] "TCFD Disclosure"
## 
## Call:
## lm(formula = df[[i]] ~ GenderDiv, data = df)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -1.07730 -0.08764 -0.05297 -0.02681  0.97624 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)   
## (Intercept)   0.9167     0.2720   3.370  0.00145 **
## GenderDiv     0.2434     0.4536   0.537  0.59397   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.505 on 50 degrees of freedom
## Multiple R-squared:  0.005724,   Adjusted R-squared:  -0.01416 
## F-statistic: 0.2879 on 1 and 50 DF,  p-value: 0.594

Alongside separately analysing each climate action, a composite score (without any specific rating) was created by summing up all the performance indicators. This was then regressed versus the firms’ gender diversity scores.

\[ ClimateScore_i=\beta_0 + \beta_1 GenderDiversity_i + \beta_2 X_i + \epsilon_i\]

The plot suggests a positive relationship i.e. more gender diversity is associated with greater climate score.

## 
## Call:
## lm(formula = Score ~ GenderDiv, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -6.5617 -1.5842  0.6633  2.1032  4.0258 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)    6.393      1.379   4.637 2.57e-05 ***
## GenderDiv      3.437      2.299   1.495    0.141    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 2.56 on 50 degrees of freedom
## Multiple R-squared:  0.04279,    Adjusted R-squared:  0.02365 
## F-statistic: 2.235 on 1 and 50 DF,  p-value: 0.1412

When regressing the overall climate score of a company against the gender diversity score, a positive relationship between the variables was observed. The results exhibit an expected 3.437 points increase in the climate score for every increase by a unit in gender diversity. However, this estimate is not significant (p-value = 0.14 > 0.1).

As the results are not significant, no clear conclusions can be drawn. Other factors which may bias this result are investigated by introducing control variables to the regression, as detailed in the following sections.

Country-level Climate Change Progress

Our team hypothesised that the location of a company’s headquarters might influence their own climate action performance, due to local regulations on climate action. Country-level data from CCPI was introduced to the regression model as a control to test for this possibility.

\[ ClimateScore_i=\beta_0 + \beta_1 GenderDiversity_i + \beta_2 CountryClimateChangePerformanceIndex_i + \epsilon_i\]

## 
## Call:
## lm(formula = Score ~ GenderDiv + CountryCCPI, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -6.5659 -1.5795  0.6668  2.1077  4.0255 
## 
## Coefficients:
##              Estimate Std. Error t value Pr(>|t|)  
## (Intercept)  6.453557   2.546928   2.534   0.0145 *
## GenderDiv    3.453794   2.396263   1.441   0.1559  
## CountryCCPI -0.001216   0.042911  -0.028   0.9775  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 2.586 on 49 degrees of freedom
## Multiple R-squared:  0.04281,    Adjusted R-squared:  0.003736 
## F-statistic: 1.096 on 2 and 49 DF,  p-value: 0.3424

The estimate has changed from 3.437 to 3.454, and the country’s climate change performance index remains negatively correlated with the companies’ climate performance. This goes against the prior hypothesis, but the results are not significant.

Hofstede Culture Index

Next, an investigation into how the climate score of a company might be affected by gender diversity in combination with cultural dimensions was undertaken. Four of the six national culture dimensions defined by Professor Hofstede and his team were extracted for this study.

\[ ClimateScore_i=\beta_0 + \beta_1 GenderDiversity_i + \beta_2 Individualism_i + \epsilon_i\]

Firstly, the potential impact of individualistic culture on climate performance and gender diversity was considered. Hofstede’s definition of individualism reflects a society’s preference for a social framework whereby individuals only care for themselves and immediate family. Conversely, in a collectivist culture, individuals consider the benefit of the broader community. A higher score indicates a more individualistic culture. We hypothesised that a collectivistic culture is likely to have a higher climate performance score, due to the associated social benefit.

model <- lm(Score ~ GenderDiv + Individualism, df)
summary(model)

## 
## Call:
## lm(formula = Score ~ GenderDiv + Individualism, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -6.5015 -1.4825  0.4981  1.8271  3.9446 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)   
## (Intercept)    8.91665    2.60840   3.418  0.00128 **
## GenderDiv      4.37179    2.43486   1.796  0.07874 . 
## Individualism -0.04287    0.03766  -1.138  0.26051   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 2.552 on 49 degrees of freedom
## Multiple R-squared:  0.06745,    Adjusted R-squared:  0.02939 
## F-statistic: 1.772 on 2 and 49 DF,  p-value: 0.1807

The result shows that after adding the individualism score, the point estimate for gender diversity has increased from 3.437 to 4.372 and the result is statistically significant at a 10% level. The negative individualism coefficient backs the hypothesis that a more individualistic society will have a lower climate score (i.e., care less for the collective good). However, the latter estimate is not significant.

Overall, this result shows that by adding the individualism control variable, a downward bias was present in the original regression. Research by Davis and Williamson have put forward the proposition that individualism promotes gender equality. As such, it implies a potential positive correlation between an individualistic culture and gender diversity on board level. Thus, the negative correlation between individualism and climate action score helps to explain the downward bias shown in the regression result.

\[ ClimateScore_i=\beta_0 + \beta_1 GenderDiversity_i + \beta_2 Masculinity_i + \epsilon_i\]

Secondly, we studied whether a masculine culture impacts climate performance and gender diversity. Hofstede defines a masculine culture as one which values achievement, heroism, assertiveness, and material rewards for success. Meanwhile, a lower score represents a culture that values femininity, with a preference for cooperation, modesty, quality of life, and caring for the weak. We hypothesised that a feminine culture is likely to score higher on climate performance, as this culture tends to prioritise quality of life.

model <- lm(Score ~ GenderDiv + Masculinity, df)
summary(model)

## 
## Call:
## lm(formula = Score ~ GenderDiv + Masculinity, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -6.0898 -1.6474  0.4954  1.8803  4.1968 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)   
## (Intercept)  5.47739    1.60155   3.420  0.00127 **
## GenderDiv    3.38384    2.29388   1.475  0.14657   
## Masculinity  0.01869    0.01676   1.116  0.27003   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 2.554 on 49 degrees of freedom
## Multiple R-squared:  0.0665, Adjusted R-squared:  0.0284 
## F-statistic: 1.745 on 2 and 49 DF,  p-value: 0.1853

The result shows that adding the masculinity score reduces the point estimate for gender diversity from 3.437to 3.384. Nevertheless, the result is not statistically significant, and does not suggest that masculinity scores of the country in which a firm is headquartered influences the climate performance of a firm.

\[ ClimateScore_i=\beta_0 + \beta_1 GenderDiversity_i + \beta_2 UncertaintyAvoidance_i + \epsilon_i\]

Thirdly, the uncertainty avoidance dimension expresses the degree to which members of a society feel uncomfortable with uncertainty and ambiguity. As per Hofstede, countries exhibiting strong uncertainty avoidance maintain a rigid code of belief and behaviour. We hypothesised that a culture with low uncertainty avoidance is likely to have a higher climate performance score, since this culture type is more open to new concepts and innovations to climate actions.

model <- lm(Score ~ GenderDiv + Uncertainty_Avoidance, df)
summary(model)

## 
## Call:
## lm(formula = Score ~ GenderDiv + Uncertainty_Avoidance, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -6.5351 -1.5501  0.6476  2.1096  4.0170 
## 
## Coefficients:
##                       Estimate Std. Error t value Pr(>|t|)   
## (Intercept)            6.21151    1.79133   3.468   0.0011 **
## GenderDiv              3.44553    2.32231   1.484   0.1443   
## Uncertainty_Avoidance  0.00287    0.01782   0.161   0.8727   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 2.585 on 49 degrees of freedom
## Multiple R-squared:  0.0433, Adjusted R-squared:  0.004247 
## F-statistic: 1.109 on 2 and 49 DF,  p-value: 0.3381

The result shows that after adding the uncertainty avoidance index, the point estimate for gender diversity (3.446) has remained at a similar level, and again the result is not statistically significant. It does not suggest that uncertainty avoidance scores influence the climate performance of a firm.

\[ ClimateScore_i=\beta_0 + \beta_1 GenderDiversity_i + \beta_2 LongTermOrientation_i + \epsilon_i\]

The last culture dimension assessed is long-term orientation. In the Hofstede study, a low scoring society that prefers to maintain traditions and norms while viewing societal change with suspicion. Those scoring high in this dimension, such as Germany and Switzerland, take a more long-term view of their decisions. Our team hypothesis was that cultures with high long-term orientation scores are likely to have higher climate performance scores, as they value the future more and might take greater climate action to prevent climate catastrophe.

model <- lm(Score ~ GenderDiv + LTOrientation, df)
summary(model)

## 
## Call:
## lm(formula = Score ~ GenderDiv + LTOrientation, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -6.6031 -1.6654  0.5675  1.9821  4.3759 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)   
## (Intercept)    7.97315    2.35639   3.384  0.00141 **
## GenderDiv      2.81977    2.42370   1.163  0.25030   
## LTOrientation -0.01983    0.02395  -0.828  0.41160   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 2.568 on 49 degrees of freedom
## Multiple R-squared:  0.056,  Adjusted R-squared:  0.01747 
## F-statistic: 1.453 on 2 and 49 DF,  p-value: 0.2437

The result shows that after adding the long-term orientation score, the point estimate for gender diversity falls from 3.437 to 2.82, but, again, the result is not statistically significant. The result goes against our hypothesis that the higher score on long-term orientation should have a higher score for climate action score. Thus, it does not suggest that long-term orientation scores influence the climate performance of a firm.

Country-level Gender Equality

As a final control, the scores from EIGE’s Gender Equality Index were added to the model, to assess whether a company’s high gender diversity score was due to the firm’s internal commitment to gender diversity, or due to the firm being headquartered in a country that naturally promotes gender equality. In this case, there was the potential for endogeneity, whereby the country’s progress on gender equality could affect both the firms’ climate and gender equality scores.

\[ ClimateScore_i=\beta_0 + \beta_1 GenderDiversity_i + \beta_2 CountryGenderEquityIndex_i + \epsilon_i\]

model <- lm(Score ~ GenderDiv + CountryGEI, df)
summary(model)

## 
## Call:
## lm(formula = Score ~ GenderDiv + CountryGEI, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -6.0441 -1.4240  0.3394  1.9351  4.0340 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)  
## (Intercept) 10.68593    5.10902   2.092   0.0422 *
## GenderDiv    4.35043    2.42994   1.790   0.0801 .
## CountryGEI  -0.06994    0.07391  -0.946   0.3490  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 2.588 on 45 degrees of freedom
##   (4 observations deleted due to missingness)
## Multiple R-squared:  0.07135,    Adjusted R-squared:  0.03008 
## F-statistic: 1.729 on 2 and 45 DF,  p-value: 0.1891

A point estimate of 4.35 was found for the gender diversity at a significance level of 10%. Therefore, adding the country gender equity index as a control variable shows that there was a downward bias in our original model. Accounting for inherent country variability in attitudes towards gender equality highlights that firms’ commitments to gender diversity – and the ensuing action on climate progress – cannot be viewed as equal; some firms face higher barriers to progress due to the fabrics of the societies they exist in, while others will be bolstered by policies and processes which promote equality at a country level. Therefore, accounting for these differences presents a more substantial relationship between female corporate leadership and action on climate progress.

A joint hypothesis test was run to check whether the result could be considered significant:

linearHypothesis(model, c("GenderDiv=0", "CountryGEI=0"))

## Linear hypothesis test
## 
## Hypothesis:
## GenderDiv = 0
## CountryGEI = 0
## 
## Model 1: restricted model
## Model 2: Score ~ GenderDiv + CountryGEI
## 
##   Res.Df    RSS Df Sum of Sq      F Pr(>F)
## 1     47 324.48                           
## 2     45 301.33  2    23.152 1.7287 0.1891

The results show an F value smaller than 10, so it cannot be concluded that the result is significant.

Conclusion

The EU regulatory environment is rapidly changing, requiring more firms to disclose and report on sustainability. In April 2021, the European Union proposed the Corporate Sustainability Reporting Directive, which requires large corporates listed on regulated markets to report by 2024. With more regulation in place, higher quality data will allow for a better understanding of how local regulations, culture, and gender diversity at board level could all contribute to firms’ progress on climate action.

The simple linear regressions run for the firms’ climate progress scores yield results that are positively correlated with the firms’ gender diversity in corporate governance, however, the results were not significant. The exception was the regression of capital alignment allocation versus gender diversity, which was significant at the 5% significance level.

To account for the effect of cultural differences on firms’ gender diversity, Hofstede’s cultural dimension scores were used as control variables. This did not lead to much change, except when controlling for individualism which produced a statistically significant result at a 10% level. This suggests that the original regression presented a downward bias.

Assessments were also conducted to contextualise a firm’s performance relative to its headquarter country and to assess any potential confounding in the original regression (Climate Score versus Gender Diversity). The regressions yielded that the gender diversity estimate was significant at 10% level, indicating that we originally had a downward bias.

Overall, although all the regressions presented a positive correlation between progress on climate targets and firms’ gender diversity in corporate leadership, most results did not register much significance. One potential limitation which may have contributed to the insignificant scores is the small dataset of 52 companies that was assessed. A larger dataset could address this limitation and is likely to become more accessible as regulations surrounding both gender diversity and climate disclosures increase. A larger dataset would also enable more granular assessments on variables such as the industry of the companies, which may affect the climate progress and gender diversity scores alike.

References

Burck et al. (2019) The Climate Change Performance Index https://ccpi.org/download/the-climate-change-performance-index-2020/ Accessed on 16 January 2022

Chattopadhyay, R., & Duflo, E. (2004). Women as Policy Makers: Evidence from a Randomized Policy Experiment in India. Econometrica, 72(5), 1409–1443. http://www.jstor.org/stable/3598894 Accessed on 16 January 2022

Climate Action 100+ (2021) Company Assessment https://www.climateaction100.org/ Accessed on 10 December 2021

Creary SJ et. al (2019) When and Why Diversity Improves Your Board’s Performance. Harvard Business Review https://hbr.org/2019/03/when-and-why-diversity-improves-your-boards-performance Accessed on 16 January 2022

Davis L.S. & Williamson C.R. (2019) Does individualism promote gender equality? World Development. Volume 123. https://doi.org/10.1016/j.worlddev.2019.104627 Accessed on 22 January 2022

European Institute for Gender Equality (2020)The Gender Equality Index Tool https://eige.europa.eu/gender-equality-index/2020/PL Accessed on 10 January 2022

European Women on Boards (2020) European Women on Boards Gender Diversity Index 2020 https://europeanwomenonboards.eu/wp-content/uploads/2021/01/Gender-Equality-Index-Final-report-2020-210120.pdf Accessed on 10 January 2022

Hofstede G. et al. (2022) Country Comparison Tools https://www.hofstede-insights.com/ Accessed 10 January 2022

McKinsey&Company (2019) Five ways that ESG creates value https://www.mckinsey.com/~/media/McKinsey/BusinessFunctions/StrategyandCorporateFinance/OurInsights/FivewaysthatESGcreatesvalue/Five-ways-that-ESG-creates-value.ashx Accessed on 18 January 2022

Mckinsey&Company (2007) Women Matter: Gender diversity, a corporate performance driver https://www.mckinsey.com/~/media/mckinsey/businessfunctions/peopleandorganizationalperformance/ourinsights/genderdiversityacorporateperformancedriver/genderdiversityacorporateperformancedriver.pdf Accessed on 17 January 2022

Mckinsey&Company (2017) Women Matter: Time to accelerate. Ten years of insights into gender diveristy. Retrieved from https://initiative-chefsache.de/content/uploads/2019/08/Women-Matter-2017-Time-to-accelerate-Ten-years-of-insights-into-gender-diversity.pdf Accessed on 17 January 2022

Whelan T. et. Al (2021) ESG and Financial Performance Uncovering the Relationship by Aggregating Evidence from 1,000 Plus Studies Published between 2015 – 2020 https://www.stern.nyu.edu/experience-stern/about/departments-centers-initiatives/centers-of-research/center-sustainable-business/research/research-initiatives/esg-and-financial-performance Accessed on 18 January 2022

World Economic Forum (2022) The Global Risks Report 2022 17th Edition Insight Report https://www3.weforum.org/docs/WEF_The_Global_Risks_Report_2022.pdf Accessed on 17 January 2022