Gender-Diverse Boards and Environmental Accountability in Business

Abstract

Businesses are increasingly expected to disclose their environmental impacts, set measurable reduction targets, and align their operations with these commitments. Transparent communication of such efforts to stakeholders has become essential in meeting growing societal and regulatory demands. This study investigates the contribution of women on corporate boards to environmental transparency, particularly given the challenges they often face in board participation and strategic decision-making. We analyze the relationship between the proportion of female directors and firms’ environmental scores across four activity groups identified by Kenessey, accounting for the varying environmental impacts of these industries. To the best of our knowledge, this is the first study to examine how board gender diversity relates to corporate environmental transparency across Kenessey’s four activity groups—an environmentally meaningful taxonomy grounded in differences in natural-resource use and environmental pressure rather than purely economic similarity (e.g., GICS). The results demonstrate a positive and significant association across all activity groups, with stronger effects in sectors characterized by higher environmental impacts. Moreover, the presence of strong governance, social welfare policies, effective oversight, climate commitments, and regulatory frameworks further reinforces this relationship.

JEL Classification: G34, M14, Q56.

Introduction

Women remain underrepresented in corporate leadership and boardrooms despite long-run gains in labor-force participation. Because boards set strategy and oversee disclosure, changes in women’s board representation (WOB) may shape how firms respond to rising environmental reporting expectations. At the same time, minority representation on male-dominated boards can limit voice and influence, making it important to account for boardroom dynamics and thresholds. These considerations motivate our focus on whether and when WOB is associated with greater disclosure-based environmental transparency.

Integrated Theoretical Framework and Conceptual Model

Early quota-style initiatives often resulted in one or two female directors, which can create tokenism and constrain influence (Kanter, 1977). When coalition-building is difficult and appointments are perceived as compliance rather than expertise, women’s impact on board decisions may be limited; critical mass arguments therefore emphasize thresholds (commonly ≥3 women) for substantive participation (Torchia et al., 2011; Yarram & Adapa, 2021). In parallel, EU initiatives have aimed to raise WOB (e.g., the 2012 proposal targeting 40%), though progress remains uneven across countries. These developments make WOB a policy-relevant governance attribute with potential financial and non-financial implications (Baghdadi et al., 2023; Lückerath-Rovers, 2013; Nielsen et al., 2010).

We contribute by testing whether the WOB—environmental transparency association varies across industries that differ in environmental intensity and natural-resource dependence. Prior studies link WOB to environmental disclosure and related ESG outcomes (e.g., García Martín & Herrero, 2020; Khan et al., 2021; Nadeem et al., 2020; Nuber et al., 2021; Özparlak et al., 2025), yet systematic cross-industry heterogeneity remains underexplored. Because stakeholder scrutiny and disclosure demands are stronger in environmentally sensitive industries, “one-size-fits-all” inferences may be misleading (Coles et al., 2008; Hawn & Ioannou, 2016). We therefore examine whether the association is stronger in higher-intensity activity groups.

Regulators and standard setters increasingly recognize sector heterogeneity in sustainability reporting (e.g., GRI sector standards and SASB materiality guidance), implying that disclosure incentives and benchmarks differ across industries. Related work also highlights that environmental metrics can be context-specific (e.g., water-quality index frameworks; Arzhangi & Partani, 2025; Arzhangi et al., 2025). From a corporate-finance perspective, disclosure-based environmental information can reduce information asymmetry and influence investor beliefs and capital allocation. Our research question therefore requires an industry taxonomy that captures environmental intensity and natural-resource dependence, not only economic similarity. While GICS provides detailed economic segmentation, Kenessey’s four activity groups operationalize differences in resource use and environmental pressure (Kuznets, 1973; Kenessey, 1987). This aggregation also preserves statistical power relative to 11-sector splits, enabling fixed-effects and dynamic estimators within sufficiently large subsamples.

To analyse whether increasing the proportion of women on boards has an impact on different levels of environmental sensitivity and transparency in different sectors, we use data from 388 businesses from 16 different countries, operating in 11 different sectors and 4 activity groups. Our environmental measure is the Bloomberg Environmental Disclosure Score (E-score). This dataset, which we obtained from the Bloomberg database, covers a 17-year period between 2008 and 2024. The results show that across sectors there is a positive relationship between WOB and environmental transparency (disclosure-based E score), but the magnitude of this association varies. We believe that the results are important for the United Nations’ goals and sectoral targets that aim for a comprehensive improvement. This study has once again demonstrated the importance of sectoral differences in terms of prioritizing and determining sector-based roadmaps for solving climate problems and social equality.

Sectors in Economy

The distinction of sectors in the economy is made in different ways. In the distinction made by Kuznet, three headings stand out: agriculture, industry, and services. While determining these three headings, Kuznet drew attention to the headings in which the three main sectors are significantly different from each other as justification. These differences are (Kuznets, 1973).

1. Differences in the use of natural resources,

2. Differences in the operating scale of production units,

3. Differences in the production process in which they participate,

4. Differences in the last products they contribute to,

5. Differences in the shares of total production and resources used.

Kenessey adheres to the main criteria highlighted by Kuznets and made a four parts distinction, considering the processing stage (Kenessey, 1987).

1. Primary activities: Agriculture, forestry, fishing, mining.

2. Secondary activities: Construction, manufacturing.

3. Tertiary activities: Transportation, electric, gas, sanitary services, wholesale trade retail trade.

4. Quaternary activities: Finance, insurance, real estate, services.

Bloomberg classifies sample firms into 11 sectors (Energy, Materials, Industrials, Utilities, Communication Services, Consumer Staples, Consumer Discretionary, Health Care, Information Technology, Real Estate, and Financials), which we aggregate into Kenessey’s four activity groups (Table 1). Primary and secondary activities are typically more resource- and emissions-intensive and face higher regulatory and stakeholder scrutiny, whereas tertiary and quaternary activities are more service-based with comparatively indirect environmental footprints. Because our objective is to test heterogeneity by environmental intensity while maintaining adequate subsample size for fixed-effects and dynamic models, we report results primarily by activity group rather than estimating separate models for each of the 11 sectors.

Table 1.

Bloomberg Sector, Kenessey’s Classification and Women on Board in the Sample.

No	Kenessey’s classification	Number of sector firms (Pcs)	Number of sector firms (%)	No	Code	Sectors	Number of firms (Pcs)	Number of firms (%)
1	Primary Act.	51	13.1	1	ENG	Energy	18	4.6
				2	MT	Materials	33	8.5
2	Secondary Act.	75	19.3	3	IND	Industrials	75	19.3
3	Tertiary Act.	158	40.7	4	UTI	Utilities	29	7.5
				5	CMS	Communication services	17	4.4
				6	CS	Consumer staples	42	10.8
				7	CD	Consumer discretionary	34	8.8
				8	HC	Health care	36	9.3
4	Quaternary Act.	104	26.8	9	IT	Information technology	31	8.0
				10	RE	Real estate	12	3.1
				11	FIN	Financials	61	15.7

Related Studies

The literature offers several mechanisms through which gender-diverse boards may influence environmental accountability. First, consistent with resource dependence theory, women directors can strengthen legitimacy and stakeholder linkages, increasing incentives and capacity to disclose credible non-financial information (Hillman et al., 2000). Second, gender diversity may improve monitoring and board deliberation quality, which can raise the salience of sustainability oversight and disclosure (Nguyen et al., 2015). Third, social identity and tokenism arguments caution that influence is conditional on boardroom dynamics: in male-dominated boards, minority voices can be marginalized, implying that effects may be nonlinear and depend on critical mass (Kanter, 1977; Torchia et al., 2011).

Empirical findings generally report a positive association between WOB and ESG- or environment-related disclosure and outcomes, but results vary across institutional settings, outcome measures, and sectors. Many studies document positive effects (e.g., Campbell et al., 2008; Farrell & Hersch, 2005; Zahid et al., 2023), while others find weak or insignificant relationships (Çolakoğlu et al., 2021) or context-dependent/negative effects (Cucari et al., 2018). A key source of heterogeneity is that environmental disclosure demands and materiality differ across industries and regulatory regimes, implying that sector intensity and stakeholder scrutiny can condition the WOB—disclosure relationship (Hawn & Ioannou, 2016).

To isolate environmental transparency from the social and governance pillars of ESG, we focus on Bloomberg’s disclosure-based Environmental Score (E score). This choice reduces mechanical overlap between WOB and ESG’s “S” and “G” dimensions and allows a more targeted test of environmental disclosure (Gürol & Lagasio, 2023). Building on emerging evidence that board diversity effects can be stronger in environmentally intensive settings (e.g., Khatri, 2023), we examine whether the WOB–E association differs systematically across Kenessey’s four activity groups, which capture heterogeneity in environmental intensity and natural-resource dependence. Based on the integrated framework, we develop testable hypotheses that predict a positive association overall and stronger effects in higher-intensity activity groups.

Data and Methodology

Hypotheses

Theoretically, women on boards (WOB) can strengthen environmental accountability and transparency through legitimacy and resource-access mechanisms and through improved monitoring and stakeholder responsiveness. Because Kenessey’s primary and secondary activity groups are characterized by higher environmental pressure and higher exposure to regulatory and stakeholder scrutiny than tertiary and quaternary activity groups, we expect the WOB—environmental transparency association to be systematically stronger where environmental salience is higher.

Accordingly, we formulate the following testable hypotheses:

• H₁: Women on boards (WOB) is positively associated with firms’ Environmental Score (E score).

• H₂: The positive association between WOB and E score is stronger for firms in (Kenessey) Primary activities than for firms in Quaternary activities.

• H₃: The positive association between WOB and E score is stronger for firms in (Kenessey) Secondary activities than for firms in Quaternary activities.

Data

The Standard and Poor’s 500 (S&P500) is a stock market index that tracks the stocks of the 500 largest companies listed on U.S. stock exchanges. In addition, STOXX 600 represents 600 companies of different sizes in 17 European countries. Our initial sampling frame consisted of all constituent firms in these two indices. We then extracted annual ESG disclosure metrics and governance/financial variables from the Bloomberg database for the period 2008 to 2024.

Bloomberg does not report ESG disclosure scores for all firms and all years within this window. Accordingly, we first construct an unbalanced raw dataset for 2008 to 2024 and retain firms with at least one observed E-score and the governance/financial variables required for our models. For the main multivariate panel regressions, variable availability (especially BTEN) restricts the usable sample to a balanced panel over 2013 to 2024, yielding 385 unique firms and 4,511 firm-year observations (see Table 9 and Appendix Table A1). Table 2 reports the definitions of all variables.

Table 2.

Definition of Variables.

Variable name	Notation	Measure
Environmental score	E	Environmental disclosure score of companies
Women on board (%)	WOB	Ratio of women board members
Board average age	BAGE	Average age of board members
Average board tenure	BTEN	Average tenure of board members
Pct independent directors (%)	BIND	Ratio of independent board members
Board size	BS	Total number of board members
Size of firm	lnTA	Natural log of total assets
Return on assets	ROA	Return on assets, a financial ratio that shows profitability
The Blau Diversity Index	BLAUW	The Blau index, which ranges from 0 to 1, where 0 representing complete homogeneity and 1 indicating maximum heterogeneity, was employed in the study to measure the proportion of woman on board (Blau, 1977). In the study, it was used as an alternative to the WOB ratio, as in previous literature
The Shannon Diversity Index	SHNW	The Shannon–Weaver Diversity Index, originally developed for ecological studies to compare biodiversity across different environments (Shannon, 1948), was utilized in this study to assess the diversity of women on corporate boards. In the study, it was used as an alternative to the WOB ratio, as in previous literature

As seen in Table 3, 47% (182) of the firms in the sample are listed in the S&P500 index and 53% (206) in the STOXX 600 index.

Table 3.

Distribution of the Firms.

Index	Number of firms (Pcs)	Number of firms (%)
S&P 500	182	46.9
STOXX 600	206	53.1

Missing-Data Treatment and Potential Sample Selection

Missingness was handled through a complete-case approach at the firm-year level. Firm-year observations with missing values for the dependent variable (Bloomberg E-score) were excluded because the outcome cannot be constructed without E-score coverage, and we do not apply statistical imputation for WOB or controls. For the core E-score regressions, the usable dataset forms a balanced panel over 2013 to 2024; for robustness specifications that rely on alternative outcomes or adjustments, sample size can vary depending on variable availability and missingness.

Because the sample is conditioned on Bloomberg ESG coverage, eliminated firms are those for which Bloomberg does not provide ESG disclosure metrics for the relevant years. Such missingness may not be random: firms with limited ESG data coverage may differ in visibility, disclosure practices, reporting intensity, or other characteristics that can correlate with environmental transparency. Accordingly, our estimates should be interpreted as evidence for the population of large listed firms with observable Bloomberg ESG disclosure data during 2008 to 2024, and we explicitly acknowledge the possibility of sample-selection bias as a limitation.

Table 4 displays the indexes subject to the firms in the research. The 388 firms in the research are in eleven different sectors and sector and activity groups’ average WOB ratios. Table 5 shows countries of sample firms. Dataset consists of 16 different countries’ firms.

Table 4.

Type of Activity Group and Sector in Which the Companies in the Sample are Placed and Their Average WOB Ratio (2008–2024).

No.	Kenessey’s classification	Number of firms (pcs)	WOB ratio in industries (%) for 2008–2024	No.	Code	Bloomberg sectors	Number of firms (pcs)	WOB ratio in sector (%) for 2008–2024
1	Primary Act.	51	25.8	1	ENG	Energy	18	25.5
				2	MT	Materials	33	25.9
2	Secondary Act.	75	25.0	3	IND	Industrials	75	25.0
3	Tertiary Act.	158	27.9	4	UTI	Utilities	29	26.4
				5	CMS	Communication services	17	29.5
				6	CS	Consumer staples	42	29.9
				7	CD	Consumer discretionary	34	27.8
				8	HC	Health care	36	25.9
4	Quaternary Act.	104	27.3	9	IT	Information technology	31	25.8
				10	RE	Real estate	12	28.4
				11	FIN	Financials	61	27.9
	Average	388	26.39

Note. Sector-level WOB ratios are computed as the average WOB (%) within each Bloomberg sector over 2008 to 2024. Activity-group (industry) WOB ratios are computed as weighted averages of the underlying sector means, where weights equal the number of firms in each sector (as reported in Table 1). Therefore, activity-group WOB ratios are not obtained by taking a simple arithmetic mean of sector averages.

Table 5.

Location of the Companies in the Sample.

No.	Countries	Number of firms	Pct (%)
1	USA	178	45.9
2	UK	57	14.7
3	France	35	9.0
4	Germany	11	2.8
5	Sweden	30	7.7
6	Switzerland	18	4.6
7	Spain	7	1.8
8	Denmark	12	3.1
9	Finland	12	3.1
10	Netherlands	9	2.3
11	Italy	4	1.0
12	Norway	9	2.3
13	Austria	1	0.3
14	Belgium	3	0.8
15	Luxembourg	1	0.3
16	South Africa	1	0.3

Dependent Variables

The dependent variable in the study is the E score obtained from the Bloomberg database. The E score is the environmental component of Bloomberg’s ESG scoring framework and is constructed from publicly available, company-reported information. Bloomberg’s methodology explicitly incorporates a disclosure component (a “Disclosure Factor”) that rewards the disclosure of quantitative environmental data and guides the treatment of missing data, which implies that the E score primarily captures the extent and structure of environmental disclosure rather than directly measuring realized environmental outcomes.

In this article, we therefore interpret the E score as an indicator of corporate environmental transparency and disclosure intensity. Bloomberg also applies sector-specific indicators and weighting schemes; as such, the score is designed to support comparisons within peer groups, and cross-sector comparability can be limited. Consistent with this feature, our empirical strategy includes industry and year fixed effects and presents analyses by activity groups that differ in environmental intensity.

Limitations Related to Disclosure-Based E Score and Robustness Approach

Because the E score is disclosure-based, higher scores can reflect more extensive reporting and transparency rather than necessarily stronger underlying environmental performance. This creates two implications for interpretation. First, our findings speak most directly to environmental accountability and transparency (what firms disclose and how consistently), not to physical environmental outcomes (e.g., emissions or resource intensity). Second, disclosure-based metrics may be susceptible to strategic reporting and “greenwashing,” where communications overstate substantive environmental improvements. We therefore frame E score as an environmental transparency measure and explicitly caution against interpreting estimated effects as proof of real environmental impact reductions.

To mitigate concerns about sector-specific scoring and potential reporting-driven effects, we adopt the following robustness approach in the revised manuscript: (i) we re-estimate the baseline models using alternative Bloomberg disclosure outcomes that are available in our dataset (overall ESG disclosure score) to verify that inferences are not unique to the environmental pillar; (ii) we re-estimate models using industry-adjusted E scores (demeaning by sector-year means) to reduce the influence of sector-specific scoring differences; and (iii) we confirm that results are not driven solely by extreme observations by repeating estimations after excluding the top and bottom tails of E scores. The full set of robustness specifications is reported in AppendixTable A1.

Independent Variables

The main motivation of the study is to examine the impact of women directors on environmental transparency (disclosure-based Bloomberg E score) across different activity groups. Therefore, the independent variable of the study is WOB. WOB is the ratio of the number of women directors to the total number of board members. In our dataset, WOB is expressed in percentage points (0–100), so the coefficient on WOB should be interpreted as the change in E score associated with a one percentage-point increase in women’s board representation. The ratio is an indicator of gender diversity in boards of directors and has been used in many studies (Cucari et al., 2018; Nadeem et al., 2020; Nwude et al., 2021; Peng et al., 2022; Wijayanti et al., 2024). The main motivation of the studies is to understand the impact of women on these activities in the boards of directors that determine strategies and monitor executive activities in the company (De Andres & Vallelado, 2008). In this study, unlike the studies in the literature, the effects of the woman board member ratio on environmental transparency and performance are analyzed for different activity groups (Appendix Figure A1).

In the study, different variables indicating the character of the board of directors were used as control variables. BAGE is the average age of the board members (De Villiers et al., 2011). BTEN is the average years of experience (tenure) of the board members (Vafeas, 2003). BIND is the ratio of independent directors to total board size (García-Sánchez & Martínez-Ferrero, 2018). BS is the total number of board members (Pfeffer, 1972). lnTA is the natural logarithm of total assets (firm size proxy). ROA is return on assets (Birindelli et al., 2019).

Model Development

$$\begin{array}{cc}\multicolumn{1}{c}{(E{)}_{i,t}=}& \multicolumn{1}{c}{{\alpha }_{\mathrm{ot}}+\beta {\left(\mathrm{board}\mathrm{gender}\mathrm{diversity}\right)}_{i,t}}\\ \multicolumn{1}{c}{}& \multicolumn{1}{c}{+\gamma {\left(\mathrm{corporate}\mathrm{governance}\right)}_{i,t}}\\ \multicolumn{1}{c}{}& \multicolumn{1}{c}{+\omega (\mathrm{firm}\mathrm{characteristics}{)}_{i,t}+\zeta \sum (\mathrm{industry}\mathrm{effect}{)}_i}\\ \multicolumn{1}{c}{}& \multicolumn{1}{c}{+\eta \sum (\mathrm{year}\mathrm{effect}{)}_i+{\epsilon }_{i,t}}\end{array}$$

(1)

In these models, the dependent variable is the E score. Board gender diversity is represented by the WOB variable. There are four variables that show the board structure and are added to the study as control variables: BAGE, BTEN, BIND, BS. In addition, lnTA and ROA variables representing business characteristics are added to the model as control variables. Finally, industry and year fixed effects are also included in the models.

Empirical Strategy and Model Selection

We estimate pooled OLS as a baseline association, firm fixed effects to absorb time-invariant unobserved heterogeneity, and a dynamic system GMM model to address persistence and potential endogeneity (reverse causality and simultaneity) using internal lag instruments (Blundell & Bond, 1998; Wintoki et al., 2012). Standard diagnostics for instrument validity are reported in Table 11 and Figure 1.

Figure 1.

Research workflow and empirical strategy.

Results

Trend Analysis

The changes in the E scores and WOB ratios of activity groups in the sample over the years are shown in Figures 2 and 3.

Figure 2.

The mean of environmental (E) score of the firms between 2008 and 2024.

Figure 3.

Average WOB of companies between 2008 and 2024.

According to the figures, both variables increased in the period between 2008 and 2024. This may be a result of increasing demands and regulations regarding the environmental impacts of businesses and the solutions to the problems experienced by women in working life.

Empirical Findings

Descriptive Analysis and Correlation Matrix

The mean E score is highest in the Primary sector (57.31) and lowest in the Quaternary sector (40.33). WOB ratio across activity groups ranges from 29.99% to 32.13%. BAGE ranges from 60.43 to 61.12, while mean board tenure (BTEN) ranges from 6.32 to 7.32 years. BIND is high for all activity groups and ranges from 73.60% to 78.67%. Boards have 10 to 12 members (Table 6).

Table 6.

Descriptive Statistics of Sectors.

Kenessey's classification	Variables	Mean	Max	Min	SD	Obs.
Primary activities	E	57.31	88.83	1.84	16.55	589
	WOB	29.99	58.33	7.69	10.28	589
	BAGE	61.12	73.17	42.00	3.90	589
	BTEN	6.32	16.10	2.00	2.18	589
	BIND	78.65	100.00	14.29	16.54	589
	BS	10.80	17.00	6.00	1.94	589
	LNTA	4.53	5.66	3.33	.52	589
	ROA	5.03	127.51	−32.20	8.09	589
Secondary activities	E	44.52	81.82	.33	17.54	879
	WOB	29.26	61.54	.00	11.00	879
	BAGE	60.43	71.46	43.75	3.96	879
	BTEN	6.88	20.14	.67	2.49	879
	BIND	73.60	100.00	5.88	19.56	879
	BS	11.17	20.00	5.00	2.41	879
	LNTA	4.36	5.85	2.83	.55	879
	ROA	6.75	95.34	−18.14	5.85	879
Tertiary activities	E	47.18	82.97	.33	16.60	1,818
	WOB	31.70	75.00	.00	11.59	1,818
	BAGE	60.46	72.92	37.00	4.21	1,818
	BTEN	7.32	18.39	.81	2.78	1,818
	BIND	76.29	100.00	16.67	18.68	1,818
	BS	11.49	20.00	5.00	2.49	1,818
	LNTA	4.46	5.74	2.51	.57	1,818
	ROA	6.90	45.55	−35.91	6.76	1,818
Quaternary activities	E	40.33	79.95	.00	16.26	1,200
	WOB	32.13	63.64	.00	11.12	1,200
	BAGE	60.76	73.75	48.70	3.89	1,200
	BTEN	7.01	21.44	1.62	2.93	1,200
	BIND	78.67	100.00	13.33	16.32	1,200
	BS	12.05	24.00	5.00	2.90	1,200
	LNTA	5.05	6.61	2.68	.87	1,200
	ROA	4.91	37.37	−19.27	7.17	1,200

Table 7 shows the Pearson correlation matrix for all activity groups.

Table 7.

Correlation Matrix.

Kenessey's classification	E	WOB	BAGE	BTEN	BIND	BS	LNTA	ROA
Primary activities
E	1
WOB	.405***	1
BAGE	−.150***	−.202***	1
BTEN	−.143***	−.349***	.386***	1
BIND	−.069*	−.018	.534***	.037	1
BS	.010	−.077*	.017	.142***	−.239***	1
LNTA	.067	.148***	.075*	−.179***	−.113***	.465***	1
ROA	.023	.084**	−.078*	−.067	.024	−.074*	−.085**	1
Secondary activities
E	1
WOB	.162***	1
BAGE	.078**	−.266***	1
BTEN	−.045	−.136***	.272***	1
BIND	.112***	−.247***	.614***	.033	1
BS	.111***	.024	.079**	−.074**	−.130***	1
LNTA	.275***	−.015	.172***	.040	−.047	.444***	1
ROA	.001	−.079**	.007	.199***	.043	−.212***	−.109***	1
Tertiary activities
E	1
WOB	.219***	1
BAGE	.083***	−.263***	1
BTEN	−.051**	−.214***	.463***	1
BIND	.132***	−.133***	.495***	.044*	1
BS	.231***	−.021	.218***	.133***	−.109***	1
LNTA	.307***	.094***	.268***	.036	.084***	.452***	1
ROA	−.051**	−.014	−.129***	.092***	−.074***	−.108***	−.195***	1
Quaternary activities
E	1
WOB	.172***	1
BAGE	−.044***	−.241***	1
BTEN	−.201***	−.302***	.562	1
BIND	.166	−.176***	.393	.090*	1
BS	.010***	−.015***	.094*	−.019***	−.304***	1
LNTA	.039**	.233	.184	−.137***	−.042***	.392	1
ROA	.076*	−.060***	−.072***	.058*	.120	−.280***	−.507***	1

Notes. Correlation coefficients are reported with significance levels based on two-tailed tests.

*

, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively.

The findings show that there is a positive correlation between the E score and WOB for the first 3 activity groups, which is in line with our expectations. The correlation coefficients between the control variables are generally low, meaning that multicollinearity is unlikely to be a significant problem in our analysis. For firms operating in quaternary activities, the bivariate correlation between WOB and environmental transparency is weaker and, in some specifications, statistically insignificant. This pattern is consistent with the relatively lower environmental salience and regulatory pressure in knowledge- and service-oriented sectors, where environmental disclosure is less central to core value creation. As a result, simple correlations may fail to capture the role of board gender diversity, which becomes more visible once firm heterogeneity and governance structures are controlled for in multivariate regressions.

Board Gender Diversity and Environmental Transparency (Bloomberg E Score)

Tables 8 and 9 reports separate regressions estimated for each Kenessey activity group. Presenting the results in a single table facilitates a descriptive comparison of magnitudes because the dependent variable (E score) and the key regressor (WOB) are measured consistently across groups and the same core specification is used. However, coefficients obtained from separate subsample regressions are not automatically statistically comparable across groups. Therefore, any cross-group comparison of coefficient magnitudes in Tables 8 and 9 should be interpreted as descriptive unless formally tested. To formally evaluate whether the WOB coefficient differs across activity groups, we additionally estimate a pooled interaction specification (equivalent to a Chow-type test) and conduct Wald/F tests for equality of the WOB effects across groups.

Table 8.

Effects of Women on Boards on Environmental Transparency by Kenessey Activity Groups.

Independent Variable	P		S		T		Q
	Baseline E	Controls E	Baseline E	Controls E	Baseline E	Controls E	Baseline E	Controls E
(Baseline and controlled models)
WOB	.898*** (17.551)	.647*** (5.868)	.515*** (12.397)	.328*** (3.121)	.478*** (17.961)	.306*** (3.889)	.477*** (15.887)	.249*** (2.766)
BAGE		−.285 (−.700)		−.129 (−.328)		−.086 (−.242)		.018 (.040)
BTEN		.129 (.152)		−.258 (−.409)		−.216 (−.500)		−.986 (−1.638)
BIND		−.018 (−.190)		.175* (1.776)		.157** (2.357)		.231*** (2.770)
BS		.314 (.421)		.079 (.100)		1.160*** (3.321)		.617 (1.382)
LNTA		−.139 (−.044)		9.293*** (3.106)		5.952*** (2.888)		−.232 (−.137)
ROA		−.025 (−.234)		.154 (.831)		.059 (.468)		.211 (.863)
C	26.639*** (18.417)	53.318** (2.369)	26.239*** (22.698)	−10.835 (−.411)***	28.750*** (35.288)	−7.993 (−.464)***	22.433*** (24.695)	12.612 (.536)
Adj. R2	.262	.162	.107	.130	.107	.169	.125	.106
F-stat.	308.035***	17.191***	153.684***	19.674***	2.589***	53.750***	252.396***	21.313***

Note. T values are in parentheses. Baseline models include WOB and fixed effects only. Controlled models additionally include board- and firm-level controls (BAGE, BTEN, BIND, BS, lnTA, ROA). Year and sector fixed effects are included in all specifications. Standard errors are clustered at the firm level.

*

p < .10. **p < .05. ***p < .01.

Table 9.

Cross-Group Coefficient Difference Tests (WOB × Kenessey Activity Groups).

Test	Statistic	Value	p-Value
Primary vs quaternary	Wald χ2 (1)	9.297	.0023
Secondary vs quaternary	Wald χ2 (1)	1.068	.3014
Tertiary vs quaternary	Wald χ2 (1)	1.925	.1653
Primary = secondary	Wald χ2 (1)	2.799	.0943
Primary = tertiary	Wald χ2 (1)	4.329	.0375
Secondary = tertiary	Wald χ2 (1)	.000	.9914
Joint test: All group interactions = 0	Wald χ2 (3)	9.373	.0247

Notes. Tests are based on the pooled interaction model. Dependent variable is Bloomberg E score. WOB is measured in percentage points (0–100). Controls: BAGE, BTEN, BIND, BS, lnTA, ROA. Year and sector fixed effects are included. Standard errors are clustered at the firm level. Sample: 2013 to 2024; Cross-sections (firms): 388; Observations: 4,511.

The results show that there is a positive and significant relationship between WOB and E for all activity groups. This means that increased representation of women on boards is associated with a higher E score (environmental disclosure/transparency), rather than necessarily stronger underlying environmental performance. This finding is in line with existing literature that highlights the positive impact of gender diversity on environmental performance (Adolat et al., 2023; Arora & Aliani, 2024; Buallay & Alhalwachi, 2022; Campbell et al., 2008; Farrell & Hersch, 2005; Ikpor et al., 2024; Zahid et al., 2023). Unlike the existing literature, the motivation of this study is to determine the relationship levels of the variables in question for different activity groups. In particular, businesses in the primary activity group cause the destruction of nature as they directly use natural resources (Rockström et al., 2009). The production processes of businesses in the secondary activity group result in high carbon and waste problems (IPCC, 2021). In this regard, businesses operating within this scope are expected to play a key role in addressing and solving environmental challenges.

Tables 8 and 9 shows the results of the analyses performed for each activity group (Appendix Tables A2–A4).

To illustrate magnitude in practical terms, a 10 percentage-point increase in WOB (e.g., from 10% to 20%) is associated with an increase of approximately 4.39 points in the Bloomberg E score in primary activities (.439 × 10), 2.22 points in secondary activities (.222 × 10), 2.24 points in tertiary activities (.224 × 10), and .94 points in quaternary activities (.094 × 10), based on the pooled interaction model reported in Table 9/Appendix Tables A5 and A6.

Interpreting Cross-Group Differences and Testing Heterogeneity

Because Tables 8 and 9 is based on separate subsample regressions, differences in the estimated WOB coefficients across P/S/T/Q groups should not be interpreted as statistically significant without a formal cross-group test. To test whether the WOB–E association differs across activity groups, we estimate the following pooled interaction model on the full sample:

$$\begin{array}{cc}\multicolumn{1}{c}{{\mathrm{E}}_{\mathrm{it}}=}& \multicolumn{1}{c}{\mathrm{\alpha }+\mathrm{\beta }\mathrm{\_}\mathrm{Q}\mathrm{WO}{\mathrm{B}}_{\mathrm{it}}+\mathrm{\gamma }\mathrm{\_}\mathrm{P}(\mathrm{WO}{\mathrm{B}}_{\mathrm{it}}\times 1{\left[\mathrm{Primary}\right]}_{\mathrm{i}})}\\ \multicolumn{1}{c}{}& \multicolumn{1}{c}{+\mathrm{\gamma }\mathrm{\_}\mathrm{S}(\mathrm{WO}{\mathrm{B}}_{\mathrm{it}}\times 1{\left[\mathrm{Secondary}\right]}_{\mathrm{i}})+\mathrm{\gamma }\mathrm{\_}\mathrm{T}}\\ \multicolumn{1}{c}{}& (\mathrm{WO}{\mathrm{B}}_{\mathrm{it}}\times 1{\left[\mathrm{Tertiary}\right]}_{\mathrm{i}})+\mathrm{Control}{\mathrm{s}}_{\mathrm{it}}+\mathrm{Year}\mathrm{FE}\hfill \\ \multicolumn{1}{c}{}& \multicolumn{1}{c}{+\mathrm{Industry}\mathrm{FE}+{\epsilon }_{\mathrm{it}}}\end{array}$$

In this specification, β_Q represents the WOB effect for the Quaternary group (reference category), while (β_Q + γ_P), (β_Q + γ_S), and (β_Q + γ_T) represent the corresponding effects for Primary, Secondary, and Tertiary groups. We then conduct Wald/F tests of coefficient equality across groups (e.g., H0: γ_P = 0 for Primary vs. Quaternary; H0: γ_S = 0 for Secondary vs. Quaternary; H0: γ_T = 0 for Tertiary vs. Quaternary), and we also test pairwise differences among Primary, Secondary, and Tertiary groups using linear restrictions (e.g., H0: γ_P = γ_S). The results of these interaction-based cross-group tests are reported alongside Tables 8 and 9 (as an additional panel) and confirm whether the apparent magnitude ordering in Tables 8 and 9 is statistically supported.

Robustness Tests

The reliability of the results was tested with a series of analyses. In these tests, the independent variable (WOB) was put into the models with the t − 1 and the dependent variable was put into the models with the t value. In other words A 1-year lag was applied in accordance with the literature (Birindelli et al., 2019; De Villiers et al., 2011). The test results are shown in Table 10.

Table 10.

Lagged Effect.

Independent variable	P	S	T	Q
	E	E	E	E
(WOB) t − 1	.632*** (9.803)	.322*** (6.213)	.294*** (9.210)	.271*** (6.353)
BAGE	−.306 (−1.390)	−.156 (−.794)	−.112 (−.916)	.026 (.158)
BTEN	.142 (.405)	−.277 (−1.146)	−.230 (−1.510)	−.985*** (−4.975)
BIND	−.027 (−.555)	.180*** (4.714)	.158*** (6.772)	.228*** (6.938)
BS	.323 (.831)	.078 (.294)	1.163*** (6.970)	.599*** (3.340)
LNTA	−.194 (−.132)	9.084*** (8.025)	5.801*** (7.905)	−.341 (−.507)
ROA	−.041 (−.525)	.151 (1.532)	.055 (1.006)	.211*** (2.919)
C	56.779*** (4.856)	−7.861 (−.770)	−4.898 (−.790)	12.870 (1.556)
Observations	589	879	1,818	1,200
Number of firms	51	75	155	104
Adj. R2	.162	.130	.167	.112
F-stat.	17.222***	19.671***	53.089***	22.655***

Note. T values are in parentheses.

***

p < 0.01.

The results in Table 10 are consistent with the results in Tables 8 and 9. These results indicate that our findings are robust.

Second, we estimate firm fixed-effects (FE) models because unobserved, time-invariant firm characteristics may jointly affect board composition and environmental transparency. The FE estimator removes time-invariant firm heterogeneity and identifies coefficients from within-firm changes over time. Under the standard FE assumption that, conditional on firm effects and time effects, regressors are strictly exogenous with respect to the idiosyncratic error term, the FE estimates mitigate bias from persistent omitted factors. The fixed-effects estimation results are reported in Table 11.

Table 11.

The Result of the Fixed Effect Test.

Independent variable	P	S	T	Q
	E	E	E	E
WOB	.707*** (7.290)	.348*** (3.877)	.301*** (4.841)	.264*** (3.974)
BAGE	.873 (1.582)	.532* (1.930)	−.044 (−.235)	.640** (1.996)
BTEN	.626 (.664)	−.694 (−1.385)	.722** (2.302)	.191 (.580)
BIND	−.032 (−.368)	−.069 (−.829)	.178** (2.269)	.153** (2.447)
BS	−.157 (−.266)	.556 (1.244)	−.070 (−.221)	.462 (1.123)
LNTA	26.186** (2.661)	38.493*** (5.092)	24.703*** (5.554)	12.985** (2.511)
ROA	.023 (.392)	.028 (.351)	.043 (.832)	−.016 (−.195)
C	−135.720*** (−2.913)	−162.291*** (−5.123)***	−88.141*** (−4.064)***	−91.512*** (−3.205)***
Observations	589	879	1,818	1,200
Number of firms	51	75	155	104
Adj. R2	.609	.774	.732	.803
F-stat.	17.092***	38.156***	31.752***	45.422***

Note. T values are in parentheses.

*

p < .10. **p < .05. ***p < .01.

The results presented in Table 11 are consistent with the main findings presented in Tables 8 and 9, indicating that the results are reliable even when potential heterogeneity issues are taken into account. Third, we estimate a dynamic panel model using system GMM to address potential endogeneity more directly. Endogeneity is a concern in this context because WOB may affect E score, but it is also plausible that firms with higher environmental transparency and performance adopt broader diversity policies that increase WOB (reverse causality), and governance and disclosure may be jointly determined (simultaneity). Consistent with a dynamic adjustment process, the model includes the lagged dependent variable, (E)t − 1, which also implies that conventional pooled OLS and FE can be biased in short panels. Therefore, we apply system GMM using internal instruments based on lagged values of the dependent and potentially endogenous regressors, following Blundell and Bond (1998) and Wintoki et al. (2012). We report instrument validity diagnostics (Sargan J-test and its p-value) and instrument count/rank as shown in Table 12.

Table 12.

The Result of the GMM Test.

Independent variable	P	S	T	Q
	E	E	E	E
WOB	.160** (2.272)	.072*** (4.791)	.063** (2.078)	.058*** (13.514)
BAGE	.014 (.047)	−.014 (−.261)	−.323*** (−4.447)	.122*** (6.913)
BTEN	−.277 (−1.002)	.067 (.686)	.096 (1.305)	.245*** (13.448)
BIND	.015 (.298)	.022 (1.461)	.118*** (4.864)	−.002 (−.312)
BS	−.263 (−1.272)	−.211 (−1.565)	.006 (.082)	−.123*** (−3.696)
LNTA	6.837* (1.908)	9.841*** (9.023)	5.529*** (6.917)	3.307*** (11.824)
ROA	−.008 (−.148)	.120*** (5.282)	−.114*** (−5.152)	.033*** (6.389)
(E)t-1	.677*** (31.246)	.709*** (74.912)	.734*** (66.162)	.667*** (220.654)***
Observations	487	729	1,508	992
Number of firms	51	75	155	104
Instrument count/rank	51	75	113	104
Sargan J-test	43.089	67.795	123.884	101.532
Prob (J-test)	.468	.450	.101	.330

Note. T values are in parentheses. The results are estimated by two-step GMM on the orthogonal deviations transformed model (see Arellano & Bover, 1995).

*

p < .10. **p < .05. ***p < .01.

The results in Table 12 are consistent with the results in Tables 8 and 9.

Alternative Measures of WOB: The Blau Index and The Shannon Index

In order to evaluate the sensitivity of the results to different diversity measures, analyses were conducted with two different diversity indicators (Blau Diversity Index and Shannon Index). Blau Index (BLAU) is a measure used to assess group diversity (Blau, 1977). This index takes a value between “0” and “0.5.” The value of “0” indicates no diversity within the group, while the value of “0.5” indicates maximum diversity. This index has been used in various studies to measure gender diversity on boards of directors (Campbell et al., 2008; Gordini et al., 2017).

$$\mathrm{Blau}\mathrm{Diversity}\mathrm{Index}=1-\sum _{i=1}^n{P}_i^2$$

(2)

P_i represents the ratio of male and female members in the company; n represents the number of gender categories (male and female). If men and women are equally represented, the index takes the maximum value of “0.50.” As the value of the index approaches zero, gender inequality increases. In other words, if all board members are female or male, the Blau Index is zero.

The Shannon Index (SHNN) is a statistical method used to measure the diversity of a group (Shannon, 1948). This method, which is used to measure diversity in boards of directors, is expressed with the “H” function (Campbell et al., 2008; Gordini et al., 2017).

$$\mathrm{Shannon}\mathrm{Diversity}\mathrm{Index}(\mathrm{H})=-\sum _{n=1}^n(\mathrm{pi}*\ln p_i)$$

(3)

The Shannon Equivalence Index (EH) is calculated as follows:

$$E_H=\frac{H}{\ln (n)}$$

(4)

This index is obtained by dividing the Shannon diversity index by the maximum diversity value and is normalized between 0 and 1. In cases where men and women are equally represented, the index takes the value of “1.” As gender inequality increases, the index approaches the value of “0.”Table 13 shows the regression results obtained using the Blau Index (BLAU) and the Shannon Index (SHNN). Both indices show a positive relationship with E for all activity groups. These findings are consistent with the results obtained earlier.

Table 13.

The Result of The Blua Index and The Shannon Diversity Index.

Independent variable	P		S		T		Q
	E	E	E	E	E	E	E	E
BLAUW	85.891*** (10.506)		48.197*** (6.291)		40.228*** (9.421)		38.570*** (6.751)
SHNW		51.740*** (10.412)		23.401*** (6.793)		22.294*** (9.020)		20.135*** (5.786)
BAGE	−.259 (−1.187)	−.261 (−1.194)	−.141 (−.718)	−.164 (−.831)	−.117 (−.963)	−.140 (−1.151)	.005 (.027)	.003 (.020)
BTEN	.187 (.540)	.152 (.440)	−.265 (−1.098)	−.283 (−1.169)	−.165 (−1.080)	−.174 (−1.138)	−.913*** (−4.588)	−.922*** (−4.639)
BIND	−.031 (−.653)	−.032 (−.675)	.171*** (4.496)	.168*** (4.428)	.144*** (6.172)	.144*** (6.146)	.223*** (6.813)	.221*** (6.742)
BS	.478 (1.240)	.467 (1.209)	.067 (.254)	.061 (.230)	1.110*** (6.666)	1.101*** (6.601)	.542*** (3.046)	.530*** (2.982)
LNTA	−.305 (−.209)	−.307 (−.210)	8.775*** (7.738)	8.740*** (7.674)	5.713*** (7.780)	5.735*** (7.787)	−.311 (−.465)	−.296 (−.444)
ROA	−.023 (−.303)	−.021 (−.278)	.145 (1.468)	.143 (1.444)	.048 (.879)	.048 (.887)	.195*** (2.690)	.194*** (2.673)
C	36.401*** (3.003)	27.643** (2.212)	−12.630 (−1.211)	−12.463 (−1.180)	−10.414 (−1.634)	−11.595* (−1.792)	7.074 (0.836)	3.136 (.362)
Adj. R2	.179	.177	.131	.125	.169	.166	.116	.116
F-stat.	19.335***	19.044***	19.826***	18.859***	53.736***	52.521***	23.472***	23.560***

Note. T values are in parentheses.

*

p < 0.10. **p < 0.05. ***p < 0.01.

Sub-Sample Analyses

We further examine whether the WOB–E association differs across governance environments by splitting the sample at the mean governance score (G = 85.24) into “weak” and “strong” governance subgroups; results are reported in Table 14.

Table 14.

Sub-Sample Analyse According to the Governance Score.

Independent variable	P		S		T		Q
	Low G	High G	Low G	High G	Low G	High G	Low G	High G
	E	E	E	E	E	E	E	E
WOB	55.341*** (3.330)	66.414*** (9.349)	10.739 (1.021)	36.016*** (6.060)	21.764*** (3.549)	26.574*** (7.106)	25.549*** (2.745)	15.942*** (3.274)
BAGE	−1.137** (−2.272)	−.245 (−1.056)	−.991** (−2.059)	−.200 (−.995)	−.136 (−.647)	−.129 (−.875)	.679** (2.081)	−.612*** (−3.123)
BTEN	.096 (.103)	.140 (.382)	−.339 (−.679)	−.259 (−.993)	−.806*** (−2.614)	.054 (.324)	−1.679*** (−4.223)	−.610*** (−2.733)
BIND	−24.736*** (−3.218)	7.564 (1.400)	2.345 (.279)	17.700*** (4.096)	−3.642 (−1.163)	16.495*** (5.645)	2.937 (.475)	21.498*** (6.018)
BS	.506 (.552)	.100 (.246)	−.884* (−1.745)	−.222 (−.701)	.312 (.848)	1.085*** (5.851)	.185 (.566)	.555*** (2.655)
LNTA	.739 (.207)	−.831 (−.532)	12.162*** (6.012)	7.464*** (5.269)	7.162*** (4.772)	5.074*** (6.282)	−.816 (−.611)	−.372 (−.487)
ROA	−.121 (−.976)	.052 (.513)	.164 (1.011)	.061 (.509)	−.043 (−.389)	.126** (2.114)	−.410*** (−2.691)	.440*** (5.657)
C	111.878*** (3.368)	48.453*** (3.760)	47.672** (1.980)	6.873 (.592)	13.902 (1.266)	−.181 (−.024)	−1.452 (−.082)	55.015*** (5.450)
Adj. R2	.307	.166	.149	.111	.163	.117	.137	.112
F-stat.	8.460***	14.429***	7.558***	11.920***	13.522***	27.215***	7.135***	17.840***

Note. T values are in parentheses. The average of G per capita is 85.24.

*

p < 0.10. **p < 0.05. ***p < 0.01.

Table 14 shows that WOB is positively and significantly related to E score in the strong-governance subgroup across all activity groups, consistent with better oversight and accountability amplifying directors’ influence on disclosure practices. In primary and tertiary activities, the estimated WOB effects are larger under strong governance. In quaternary activities, where many environmental initiatives are comparatively less capital-intensive, women’s influence appears less dependent on strong governance structures.

The income level of the country where the business is located can affect the relationship between board gender diversity and environmental performance (Lv et al., 2020). To analyze this effect of the country’s income level, we calculated the average of GDP per capita values (72,609 USD/GDP per capita). We divided the companies below this value into two groups as weak (GDP < Mean = 72,609 USD); and companies above this value (G > Mean = 72,609 USD) as strong income countries.

According to the analysis results (Table 15), the impact of WOB on the environment is positive and significant in both subgroups for all activity groups. However, the contribution of women on the boards of directors of companies located in countries with high GDP per capita to environmental transparency and performance is higher. This result can be explained by the fact that in more economically developed countries, corporate governance standards, regulatory frameworks and stakeholder expectations are generally more developed, and therefore the influence of women managers on shaping and implementing environmental policies increases (Eccles et al., 2014). In addition, since women in developed economies generally have higher levels of education and professional qualifications, their capacity to contribute to environmental management and sustainability strategies may also increase (Galbreath, 2013; Post et al., 2011).

Table 15.

Sub-Sample Analyse According to the GDP Per Capita.

Kenessey's classification	P		S		T		Q
GDP per capita group	GDP < mean	GDP > mean	GDP < mean	GDP > mean	GDP < mean	GDP > mean	GDP < mean	GDP > mean
Dependent Variable	E	E	E	E	E	E	E	E
WOB	34.229*** (4.501)	90.271*** (8.456)	26.763*** (3.554)	35.454*** (4.607)	31.738*** (7.021)	23.035*** (4.805)	25.616*** (5.429)	30.095*** (12.737)
BAGE	−.292 (−1.240)	−.022 (−.063)	−.553** (−2.194)	.382 (1.190)	.304 (1.570)	−.059 (−.378)	.187 (1.063)	.237*** (2.769)
BTEN	−.027 (−.054)	.678 (1.304)	.523 (1.293)	−.529 (−1.498)	−.404 (−1.582)	.068 (.323)	−.213 (−.738)	−.683*** (−6.343)
BIND	5.779 (1.025)	1.917 (.245)	20.622*** (4.092)	6.517 (.663)	15.625*** (4.365)	12.077*** (3.247)	20.910*** (6.744)	10.738*** (6.691)
BS	.142 (.273)	.494 (.820)	.856*** (2.661)	−2.550*** (−4.655)	1.021*** (4.497)	1.091*** (4.352)	.817*** (5.002)	.574*** (5.122)
LNTA	2.916* (1.654)	−3.074 (−1.301)	5.985*** (4.305)	18.739*** (7.023)	6.286*** (5.943)	6.075*** (6.069)	−2.257*** (−3.324)	.505 (1.208)
ROA	−.016 (−.193)	−.060 (−.449)	−.035 (−.272)	.476*** (3.090)	−.051 (−.718)	.212** (2.580)	−.652*** (−6.842)	.128*** (3.368)
C	48.828*** (3.750)	32.885 (1.497)	16.503 (1.112)	−45.840*** (−3.029)	−27.509*** (2−.657)	−8.090 (−0.891)	10.264 (1.086)	9.349** (2.077)
Adj. R2	.116	.178	.113	.222	.258	.104	.220	.069
F-stat.	4.875***	10.964***	9.139***	18.569***	37.424***	19.246***	22.303***	48.906***

Notes. T values are in parentheses. The average of GDP per capita is 72,609 USD.

*

p < 0.10. **p < 0.05. ***p < 0.01.

The Climate Change Policy in companies covers the policies and practices adopted by companies to combat climate change. These policies include both compliance with legal requirements and voluntary environmental responsibilities (Eccles et al., 2014; Liesen et al., 2015). Table 16 presents the analysis results regarding the environmental impact of women on the boards of directors of companies with a valid Climate Change Policy.

Table 16.

Sub-Sample Analyse for the companies which has a Climate Change Policy.

	P		S		T		Q
Independent variable	CPOL = 0	CPOL = 1	CPOL = 0	CPOL = 1	CPOL = 0	CPOL = 1	CPOL = 0	CPOL = 1
	E	E	E	E	E	E	E	E
WOB	26.286 (1.029)	50.524*** (8.571)	16.812 (1.375)	12.901** (2.361)	−8.396 (−1.025)	24.190*** (7.020)	6.826 (0.773)	9.505** (2.039)
BAGE	.568 (.667)	−.086 (−.473)	.423 (.922)	−.157 (−.798)	−.644** (−2.087)	.148 (1.202)	−.227 (−.730)	−.003 (−.017)
BTEN	−1.048 (−.835)	.055 (.180)	−1.045** (−2.424)	.513* (1.934)	.277 (.765)	−.098 (−.618)	−1.267*** (−3.512)	−.784*** (−3.681)
BIND	−28.305** (−2.053)	−3.547 (−.944)	17.523* (1.773)	14.434*** (3.853)	22.801*** (3.891)	7.564*** (3.479)	13.744** (2.026)	15.390*** (4.713)
BS	2.808* (1.843)	.113 (.339)	−1.094 (−1.175)	.200 (.786)	1.711*** (3.585)	.778*** (4.647)	−.134 (−.368)	.491*** (2.597)
LNTA	−10.606* (−1.727)	−4.027*** (−3.081)	16.097*** (5.847)	3.777*** (3.218)	2.510 (1.447)	6.115*** (8.072)	3.490** (2.407)	−2.255*** (−3.279)
ROA	−.541* (−1.761)	−.049 (−.731)	.857*** (3.042)	.140 (1.482)	.223* (1.737)	.076 (1.332)	.047 (.276)	.266*** (3.680)
C	38.199 (.902)	70.015*** (6.768)	−67.936*** (−2.820)	20.191* (1.942)	22.230 (1.430)	−9.300 (−1.429)	20.873 (1.319)	38.101*** (4.269)
Adj. R2	.048	.141	.262	.042	.110	.128	.170	.076
F-stat.	1.486***	13.297***	9.766***	5.463***	5.116***	34.577***	6.809***	12.787***

Notes. T values are in parentheses. If the company has not a Climate Change Policy that takes CPOL = 0, CPOL = 1 otherwise.

*

p < 0.10. **p < 0.05. ***p < 0.01.

The companies in the sample were divided into two groups: those without a climate policy (CPOL = 0) and those with a climate policy (CPOL = 1). According to the analysis results, the environmental impact of women on the boards of directors of companies with a Climate Change Policy is positive and significant. On the other hand, this relationship is insignificant in all companies without this policy. This result indicates how environmental corporate policies are a supporting factor in increasing women’s contributions to sustainability and the environment (Galbreath, 2013; Post et al., 2011).

Audit Financial Experts are individuals who serve on the audit committees of companies and have expertise in the fields of accounting, auditing or finance. These experts undertake tasks such as ensuring the accuracy of the company’s financial reporting processes, evaluating internal control systems and ensuring the transparency of especially complex financial transactions (Krishnan et al., 2008). The presence of financial experts in audit committees ensures that board decisions are based on more solid foundations at the technical level, while creating a platform where female members can implement their environmental and social sensitivities more effectively (Bear et al., 2010; Fernandez-Feijoo et al., 2014; Pucheta-Martínez et al., 2015). Today, non-financial results of business activities (environmental, social and governance) are also intertwined with financial reporting. ESG studies have become an integral part of the finance field. Stakeholders want to see environmental results, especially those that are likely to affect financial results, in reports. Therefore, the presence of financial experts in companies ensures that environmental issues are treated more sensitively. In our study, we divided the companies in the sample into two groups: those without a climate change policy (FEXP = 0) and those with a climate change policy (FEXP = 1). Table 17 shows the results regarding the environmental impact of women on the boards of directors of companies with Audit Financial Expert.

Table 17.

Sub-Sample Analyse for the Companies Which has an Audit Financial Expert.

Independent variable	P		S		T		Q
	FEXP = 0	FEXP = 1	FEXP = 0	FEXP = 1	FEXP = 0	FEXP = 1	FEXP = 0	FEXP = 1
	E	E	E	E	E	E	E	E
WOB	14.771 (1.251)	48.924*** (6.454)	−.387 (−.034)	36.675*** (5.526)	10.267 (1.017)	22.398*** (5.852)	−28.077* (−1.785)	16.009*** (2.981)
BAGE	−0.277 (−1.269)	0.146 (0.496)	−0.321 (−0.552)	0.160 (0.752)	0.040 (0.110)	−0.225 (−1.565)	1.549*** (3.168)	−0.216 (−1.082)
BTEN	−0.046 (−0.095)	0.035 (0.088)	−2.060*** (−3.058)	0.075 (0.274)	2.644*** (4.310)	−0.264 (−1.583)	−1.131 (−1.346)	−0.923*** (−4.021)
BIND	17.787*** (4.972)	−13.902** (−2.125)	31.290*** (3.150)	20.248*** (4.525)	3.284 (0.747)	17.622*** (6.296)	15.065** (2.019)	21.796*** (5.444)
BS	1.880*** (3.396)	−.226 (−.531)	−.454 (−.703)	1.333*** (3.880)	−1.111*** (−2.786)	1.440*** (7.377)	.435 (1.151)	.539** (2.462)
LNTA	−8.916*** (−4.141)	−1.323 (−.777)	5.407** (2.179)	8.123*** (5.564)	10.163*** (5.473)	5.519*** (6.629)	−1.926 (−.746)	−.010 (−.013)
ROA	−.149*** (−3.086)	.181* (1.798)	.687** (2.347)	−.022 (−.215)	−.371** (−2.199)	.083 (1.370)	.096 (.171)	.206*** (2.677)
C	87.342*** (4.951)	53.662*** (3.717)	38.918 (1.317)	−42.621*** (−3.478)	−8.691 (−.414)	2.266 (.299)	−35.537 (−1.335)	31.779*** (3.170)
Adj. R2	.422	.107	.145	.183	.332	.140	.214	.086
F-stat.	7.358***	8.484***	4.755***	19.642***	10.784***	33.502***	4.546***	13.480***

According to the results, in the groups with Audit Financial Expert, the effects of women on the boards of directors of companies on environmental outcomes and transparency are positive and significant. However, while the environmental effects of women in the management of companies in the Q industry with Audit Financial Expert are positive and significant, in the same industry with no Audit Financial Expert, the environmental effects of women are negative and significant. The negative environmental impact of women managers in companies in the Q industry with no financial audit expert shows the importance of the role of the audit mechanism on the expected positive effect between WOB and E score in this activity group.

The Effect of the EU Non-Financial Reporting Directive (NFRD)

The EU Non-financial Reporting Directive (NFRD) includes regulations on non-financial reporting in the EU. According to the regulation, businesses are obliged to disclose their impacts on sustainability and how they manage these impacts (European Parliament & Council of the European Union, 2014). The regulation is important because it has made sustainability practices a necessity rather than an intention (Christensen et al., 2021). In this part of the study, we created a subgroup from the data of EU companies in the sample. Our aim is to understand whether the NFRD, which was adopted by the EU Parliament and Council in 2014 and started to be implemented in 2017, has an impact on the relationship between WOB and E. For this purpose, we conducted a DID (difference in differences) test on our data set. The Difference in Differences method is a quasi-experimental econometric technique used to measure the impact of policy changes or interventions (Bertrand et al., 2004). In the model, the (Post × Treat) variable captures the immediate effect of the NFRD implementation, while the (Post × Treat)_t + 1 term is included to account for any lagged policy impact that may manifest in the following year. The D_Post2017 dummy controls for overall post-policy temporal effects, and the D_EU dummy identifies EU firms subject to the regulation. These specifications allow the model to isolate the causal effect of the policy and distinguish it from general time trends or group-specific characteristics. The test results are shown in Table 18.

Table 18.

DID Analysis Results.

Independent variable	P	S	T	Q
	E	E	E	E
(Post × Treat)t + 1	4.773* (1.701)	7.264*** (2.884)	5.989*** (3.264)	3.624 (1.572)
Post × Treat	−11.777*** (−3.441)	−12.990*** (−4.058)	−7.726*** (−3.577)	−4.348 (−1.585)
Post_2017	19.560*** (10.500)	14.777*** (7.774)	10.179*** (9.915)	8.259*** (6.077)
In EU	15.527*** (6.644)	3.484 (1.499)	8.768*** (6.124)	−4.808*** (−2.605)
WOB	21.101*** (3.178)	10.823* (1.887)	11.104*** (3.290)	12.384** (2.476)
BAGE	−.169 (−.872)	−.262 (−1.347)	.153 (1.297)	−.240 (−1.454)
BTEN	.412 (1.295)	−.316 (−1.292)	.148 (.954)	−1.311*** (−6.411)
BIND	8.902** (2.154)	13.145*** (3.218)	17.080*** (7.344)	12.051*** (3.522)
BS	.890** (2.557)	.156 (.584)	1.222*** (7.718)	.409** (2.305)
LNTA	−1.070 (−.832)	7.613*** (6.904)	5.747*** (8.228)	.242 (0.368)
ROA	−.098 (−1.413)	.131 (1.382)	.098* (1.893)	.146** (2.019)
C	28.540** (2.319)	4.521 (0.370)	−30.064*** (−4.291)	40.884*** (4.395)
Adj. R2	.355	.204	.236	.154
F-stat.	30.564	21.401	52.519	21.002

The results show that the impact of the policy was seen in the following year, although not in the same year, and the relationship between WOB and E was strengthened in the P, S, and T activity groups. This positive situation, especially in the P and S sectors where expectations regarding environmental sensitivity are high, shows that women can undertake strategic roles that increase environmental transparency (Abd Majid & Jaffaar, 2023).

Discussion

Taken together, the results support a robust positive association between women’s board representation and disclosure-based environmental transparency. Across Kenessey’s activity groups, higher WOB is linked to higher Bloomberg E scores, and interaction/Wald tests indicate that the effect is significantly stronger in primary activities relative to quaternary activities. This pattern is consistent with higher environmental salience and stakeholder scrutiny in resource-intensive industries, where boards face stronger incentives to legitimize operations through transparent reporting.

The broader set of analyses strengthens interpretation. Lagged specifications, firm fixed effects, and system GMM yield consistent inferences, suggesting that the main finding is not driven solely by simultaneity or time-invariant unobservables. Alternative diversity measures (Blau and Shannon indices) produce similar results, indicating that the inference does not hinge on a single operationalization of board gender diversity. Conditioning analyses further suggest that complementary governance and policy environments—such as strong governance structures, formal climate policies, and effective audit oversight—can reinforce how board composition translates into transparency outcomes.

Because E score is disclosure-based, the evidence should be interpreted as improved environmental transparency rather than direct proof of physical environmental performance improvements. Nevertheless, the results imply that board gender diversity, especially in environmentally intensive activities, can be an important governance channel for strengthening accountability and disclosure practices.

Conclusion

This study finds that higher women’s board representation is associated with greater disclosure-based environmental transparency across Kenessey activity groups, with the strongest association in environmentally intensive industries. The evidence also indicates that complementary governance and policy arrangements—strong oversight, climate commitments, audit expertise, and supportive institutional environments—amplify this relationship.

Implications

• Sector-sensitive diversity policies may be most consequential in primary/secondary activities, where environmental pressure and stakeholder scrutiny are highest.

• Diversity initiatives are likely to be most effective when paired with credible monitoring and reporting infrastructures that strengthen accountability.

Because Bloomberg’s E-score reflects disclosure intensity, future research can apply the same sector-intensity framework to performance-based environmental measures and to broader firm populations beyond large listed firms.