This study aims to investigate the relationship between board gender diversity (BGD) and credit ratings, using agency theory, resource dependence theory and critical mass theory as theoretical frameworks.
This paper analyses a sample of 1,037 North American companies from 2008 to 2017. The methodology includes the Arellano–Bond generalized method of moments (GMM), an ordinal extension of the binary logit model and robustness tests to address potential endogeneity and sample selection bias.
The results indicate that increasing female representation on boards significantly affects credit ratings. Specifically, each additional female board member increases the likelihood of obtaining a higher credit rating by up to 17.71. This effect is particularly pronounced for firms transitioning to investment-grade ratings, where the impact of female representation is amplified fourfold. These findings highlight the important role of board gender diversity in improving firms’ credit evaluations.
By examining a crucial period and employing rigorous analytical techniques, this study fills a significant gap in the literature; it offers valuable insights into how BGD affects credit ratings and emphasizes its strategic importance in corporate risk governance.
1. Introduction
Credit ratings are crucial for firms due to their significant impact on capital access (Rahmani et al., 2023; Bauer and Esqueda, 2017), market volatility and valuations (An and Chan, 2008). Given the complexity of the financial system and the diversity of market participants, credit ratings have become the most widely accepted tool for assessing a firm’s creditworthiness (Cantor and Packer, 2000). This has led firms to seek ways to maintain or enhance their ratings, often by altering how rating agencies perceive their credit risk (Ali and Javid, 2015). This issue is heightened by the fact that rating agencies do not publicly disclose their methodologies: They remain confidential. However, it is understood that the assessment of credit risk involves more than financial metrics; qualitative factors play a crucial role, underpinning the quantitative indicators. As noted by Standard and Poor’s (S&P, 2002), “Ratings represent an art as much as a science” (p. 17).
These qualitative aspects include management-related factors such as the role of directors in shaping operational success, establishing credibility and formulating financial risk policies (S&P, 2002). These responsibilities largely fall to the board of directors, who set strategic values and manage company operations (Bordean et al., 2011). Board gender diversity (BGD) is associated with reduced agency problems (Ain et al., 2020) and better monitoring function (Maxfield and Wang, 2024), while offering a range of perspectives, experiences and skills that contribute to more effective decision-making and enhanced risk management (Darmawan, 2024). The inclusion of women on boards is known to foster strong communication channels with external stakeholders, thus addressing agency issues and improving resource access (Khan et al., 2023). There is also a growing societal demand for increased gender equality (e.g. it is the fifth U.N. Sustainable Development Goal) and BGD, but parity remains elusive. A Fortune (2024) report highlights that, in 2024, female executives managed only 52 of the world’s 500 largest companies, underscoring the ongoing disparity in corporate leadership.
Despite the critical importance of BGD, the literature linking female board members to credit ratings remains sparse and inconsistent. Existing studies present contradictory findings. For instance, Grassa (2016) found a positive correlation between the proportion of women directors and credit ratings in Islamic banks, while Kim and Kim (2022) report that female CEOs and executives have a positive effect on the credit ratings among Korean listed firms. Muricken et al. (2024) discovered that the credit ratings of firms in India improved after women were added to their boards following a gender quota mandate. However, Iryanti and Mawardi (2021) found no significant impact in Indonesia. These studies are limited in scope and geographic relevance, which highlights the need for further research. In addition to the limited geographic and sectoral scope, prior research has not sufficiently explored whether BGD influences firms during especially important phases, such as transitions from speculative- to investment-grade ratings. Many existing studies have also relied on single-country (Iryanti and Mawardi, 2021; Kim and Kim, 2022) or narrow sectoral samples (Grassa, 2016; Kim and Kim, 2022), thus limiting the generalizability of their findings. This study addresses these gaps by analysing how BGD influences credit ratings within the context of Western financial markets, focusing on a diverse sample of North American firms over a ten-year period. Our study makes a unique contribution by focusing on credit ratings, which are a critical measure of credit risk. As highlighted in recent systematic literature reviews (Kalia and Gill, 2023), most studies on corporate governance and risk have tended to focus on commercial risks, particularly equity risk, with only 18% examining the relationship between governance attributes and credit risk. By analysing the impact of BGD on credit ratings, this research fills an important gap in the literature.
To this end, we analysed a sample of 1,037 publicly traded U.S. firms covering the period after the 2008 global financial crisis and concluding in 2017. This timeframe was carefully selected to exclude potential distortions caused by regulatory changes introduced during the Trump administration, particularly the modifications and partial rollback of the Dodd–Frank Act. This period also saw ongoing calls to integrate risk management into governance (Stein and Wiedemann, 2016), as well as an increasing recognition of the importance of holistic governance principles, such as stakeholder engagement and resilience, in addressing the challenges posed by financial crises (Myeza et al., 2023). We examine this relationship through several key analyses. Firstly, we investigate whether BGD affects a firm’s leverage, a primary indicator of credit risk. Secondly, we analyse how BGD affects credit ratings and its contribution to the likelihood of obtaining higher ratings. Finally, we evaluate whether reaching a critical mass of women on the board enhances credit ratings.
Our study makes several contributions to the extant literature on credit ratings and corporate governance. Firstly, by analysing the effects of BGD on firms’ credit ratings, it sheds light on the financial implications of substandard BGD. Specifically, it addresses the impact on credit ratings, a relatively unexplored area. Secondly, this research provides new insights into how BGD affects credit ratings, particularly revealing that the most significant effects occur when firms transition from speculative to investment-grade ratings. This nuanced understanding highlights the strategic importance of BGD as a means of enhancing risk governance and mitigating agency problems through more effective monitoring (Post and Byron, 2015; Francoeur et al., 2008), enhancing communication with stakeholders (Jizi and Nehme, 2017) and improving the firm’s market perception. Thirdly, by focusing on U.S. firms, we expand the geographic scope of research on BGD and its influence on credit ratings, thus providing insights highly relevant to Western financial markets, which have been underexplored in this context. Finally, the findings of this research validate ongoing efforts and initiatives to improve corporate governance through enhanced BGD and reinforce the importance of these measures in strengthening firm performance and creditworthiness.
The rest of the paper is organized as follows. Section 2 reviews the most significant literature and establishes the hypotheses. Section 3 describes the data and methodology. Section 4 presents the main results. Section 5 summarizes and concludes the paper.
2. Literature review and hypotheses
2.1 Board gender diversity and leverage level
Behavioural theory suggests that women, who may be less confident and more risk-averse than men, tend to make more conservative decisions (Li et al., 2022; Faccio et al., 2016; Jianakoplos and Bernasek, 1998) [1]. This is reflected in studies like Palvia et al. (2015), which showed that women assess risks more cautiously, thereby reducing the likelihood of bank failure. There is, however, no consensus about the link between women in the boardroom and risk-taking; indeed, several authors have reported a negative relationship between these two variables (Azzim-Gulamhussen and Fonte Santa, 2015; Lenard et al., 2014; Adams and Ferreira, 2009). However, other studies have obtained a positive relationship (Berger et al., 2014; Adams and Funk, 2012), and finally, there are studies that do not find any association (Mathew et al., 2016; Sila et al., 2016; Maxfield et al., 2010; Van Der Walt et al., 2006).
Agency theory posits that differences in attitudes towards risk between managers and shareholders can lead managers – who are concerned about their careers and short-term stock prices – to be more risk-averse and thus favour conservative projects. This risk aversion can result in agency costs, which need to be mitigated through effective corporate governance mechanisms. Agency theory underscores the importance of these mechanisms in monitoring managerial behaviour (Darmawan, 2024). Complementing this view, Salehi et al. (2024) found that governance mechanisms, including board structure, play a fundamental role in reducing financial costs, which emphasizes the interplay between governance quality and external risk perceptions. Among such mechanisms, BGD is expected to improve management oversight and to balance risk levels within a firm, thereby enhancing shareholder value by encouraging a higher level of risk-taking (Ain et al., 2020). Several studies have attempted to explain this result. For example, Nadeem et al. (2019) found that BGD influences corporate risk-taking by positively affecting board dynamics, as evidenced by increased director attendance and enhanced board independence; to explore this further, Khan et al. (2023) found that women are particularly effective at developing communication channels with external stakeholders, thus helping to overcome agency problems and facilitate resource access.
The impact of BGD remains unclear for credit risk. Faccio et al. (2016) demonstrated that firms led by female CEOs tend to have significantly lower leverage and financial risk compared to those led by male CEOs. Similarly, Perryman et al. (2016) found that the inclusion of women on management teams is associated with reduced leverage levels. However, the findings of Nadeem et al. (2019) differ: they showed a significant positive relationship between the presence of women on boards and leverage. Finally, Matsa and Miller (2013) observed no change in firm leverage following the implementation of a female board representation quota in Norway. Because leverage is a key measure of credit risk, we examine how the percentage of women on the board affects this variable. Given the unclear nature of this relationship, we propose the following hypotheses:
The percentage of women on the board has a positive effect on the level of leverage.
The percentage of women on the board has a negative effect on the level of leverage.
The percentage of women on the board has a null effect on the level of leverage.
2.2 Board gender diversity and credit rating
Research on the link between BGD and credit ratings is limited and inconclusive. While some studies, like Grassa (2016) on Islamic banks and Kim and Kim (2022) on Korean firms, have found a positive relationship, others, such as Iryanti and Mawardi (2021) in Indonesia, have reported no significant impact. Factors like improved corporate social responsibility (CSR) performance, however, may still connect female board representation with higher credit ratings, as highlighted by the S&P (2019) Global Research report, which showed that women CFOs drive better returns and profitability. Additionally, equitable gender representation is a key indicator of diversity in corporate governance (Hafsi and Turgut, 2013; Hillman et al., 2007).
According to agency and resource dependency theories, the board’s composition is critical for CSR development, as board members drive these decisions (Melón-Izco et al., 2020; Cuadrado Ballesteros et al., 2015; Huang, 2010). Similarly, governance mechanisms like risk management play a critical role in shaping firm performance and external credit evaluations, as evidenced by Jallali and Zoghlami (2022), who demonstrate that effective risk governance enhances both financial stability and stakeholder trust in banking institutions. In this vein, research has consistently shown that female directors – known for their social, ethical and participative traits – are particularly effective in enhancing CSR and firm reputation by focusing on a broad range of stakeholders (Bear et al., 2010; Byron and Post, 2016). In their systematic literature review, Ciappei et al. (2023) explore the relationship between women in leadership positions and sustainable performance; they found that female executives often drive greater commitment to sustainability, particularly in the environmental and social aspects of CSR.
From a resource dependency perspective, women directors contribute unique perspectives and values that foster stakeholder engagement and align corporate behaviour with societal expectations. Their participative leadership style promotes awareness, social initiatives and enhanced stakeholder relationships that ultimately improve corporate reputation and social responsibility outcomes (McGuinness et al., 2017; Ramon-Llorens et al., 2020.Women are also found to have a limiting effect on tax avoidance (Kovermann and Velte, 2019), thus reinforcing the importance of greater female representation in senior management as a governance mechanism. Similarly, agency theory suggests that gender-diverse boards can mitigate risks of groupthink and encourage more ethical and transparent decision-making, further strengthening CSR (Marquez-Cardenas et al., 2022; Yahya et al., 2021).
Recent studies have also highlighted the role of managerial traits, such as overconfidence and myopia, in influencing corporate risk-taking and governance decisions. Salehi et al. (2022) found that these traits can exacerbate risk-taking tendencies and undermine effective governance. BGD may counterbalance these tendencies, introducing diverse viewpoints and fostering balanced decision-making processes, which in turn improve governance quality and external assessments like credit ratings.
In addition, previous research have suggested a positive link between CSR and credit ratings. Strong CSR improves stakeholder relationships, boosts long-term sustainability, promotes efficient resource use and reduces the risk of socially irresponsible behaviour, all of which lower financial risk and lead to higher credit ratings (Lin et al., 2020; Jiraporn et al., 2014; Attig et al., 2013). Furthermore, women directors have been found to enhance risk disclosure (Seebeck and Vetter, 2022). Given the established connections between BGD, CSR performance and company reputation, as well as their correlation with credit ratings, we propose the following hypothesis:
The percentage of women on the board has a positive effect on a company’s credit rating.
2.3 Critical mass of women on the board and credit rating
Research has indicated that gender-balanced teams often outperform others. Trinh et al. (2023) found that achieving a critical mass of minority group members, including at least two women on the board, is essential to decision-making effectively. Similarly, Bear et al. (2010) and Boulouta (2013) demonstrated that such teams are more likely to generate alternative solutions and make innovative decisions. Konrad et al. (2008) highlighted additional benefits of including more women on boards, noting that they offer diverse perspectives, broaden the scope of board discussions, address concerns of various stakeholders, pose challenging questions on complex issues and leverage their interpersonal skills to have a positive impact on board dynamics. Kanter’s (1977) critical mass theory suggests that when women are a small fraction of a team, they are seen as gender representatives rather than as individuals; however, once they reach a critical mass – at least three members or 35% – they start to influence the board’s dynamics.
Several studies have highlighted the impact of reaching a critical mass of female board members on various business outcomes. For instance, a critical mass positively influences ESG components (De Masi et al., 2021; Amorelli and García-Sánchez, 2020), enhances firm performance once women account for 30% of the board (Joecks et al., 2013) and boosts innovation (Torchia et al., 2011). Related to our research, Harris (2014) found a negative link between boards with at least 25% women and corporate leverage, while Schwartz-Ziv (2017) showed that boards with at least three directors of each gender exhibit significantly more activity in meetings. Based on the previous arguments, we propose our third hypothesis:
The existence of a critical mass of women on a firm’s board has a positive effect on a company’s credit rating.
3. Data and empirical methodology
3.1 Sample
Our sample includes North American companies with credit ratings available from 2008 to 2017. Company fundamentals were sourced from the Orbis database (Bureau van Dijk), while credit ratings (S&P), board gender, corporate governance and market data were obtained from Eikon (Refinitiv Workspace). After excluding firms with missing or incomplete data, our final data set is an unbalanced panel of 1,037 companies and 5,816 observations.
3.2 Variables
We divided our variables into four groups: dependent variables, gender variables, corporate governance variables and company fundamentals and other control variables (Table 1).
Description of the variables
| Variable | Definition | Expected sign on rating | Source |
|---|---|---|---|
| Dependent variables | |||
| Leverage1 | Total debt to total assets | Orbis | |
| Rating | Ordinal variable with seven thresholds from 1–7 depending on the rating score given by S&P | Eikon | |
| Gender variables | |||
| Femaleboard | Percentage of women on the board | + | Eikon |
| Bequitable | Dummy variable equal to 1 if there are at least two women on the board | + | Eikon |
| Corporate governance variables | |||
| Boardindep | Percentage of independent directors on the board | + | Eikon |
| Duality | Dummy variable that takes a value of 1 if the CEO also serves as chairperson and 0 otherwise | − | Eikon |
| Company fundamentals and control variables | |||
| Interestcoverage | EBIT to interest expense | + | Orbis |
| Loss | Dummy that takes a value of 1 if ROA is negative in the current and previous year | − | Orbis |
| Tangibility | Fixed to total assets | + | Orbis |
| ROA | EBIT to total assets | + | Orbis |
| Size | Napierian logarithm of net sales | + | Orbis |
| MtB | Market value of the share to its book value | + | Eikon |
| Audit | Dummy that takes a value of 1 for firms with a favourable report | + | Eikon |
| Financial | Dummy equal to 1 for financial companies | Orbis | |
| Grade2 | Dummy that takes a value of 1 for rating values over BB+ and zero otherwise | Eikon | |
| Variable | Definition | Expected sign on rating | Source |
|---|---|---|---|
| Dependent variables | |||
| Leverage1 | Total debt to total assets | Orbis | |
| Rating | Ordinal variable with seven thresholds from 1–7 depending on the rating score given by S&P | Eikon | |
| Gender variables | |||
| Femaleboard | Percentage of women on the board | + | Eikon |
| Bequitable | Dummy variable equal to 1 if there are at least two women on the board | + | Eikon |
| Corporate governance variables | |||
| Boardindep | Percentage of independent directors on the board | + | Eikon |
| Duality | Dummy variable that takes a value of 1 if the CEO also serves as chairperson and 0 otherwise | − | Eikon |
| Company fundamentals and control variables | |||
| Interestcoverage | EBIT to interest expense | + | Orbis |
| Loss | Dummy that takes a value of 1 if ROA is negative in the current and previous year | − | Orbis |
| Tangibility | Fixed to total assets | + | Orbis |
| ROA | EBIT to total assets | + | Orbis |
| Size | Napierian logarithm of net sales | + | Orbis |
| MtB | Market value of the share to its book value | + | Eikon |
| Audit | Dummy that takes a value of 1 for firms with a favourable report | + | Eikon |
| Financial | Dummy equal to 1 for financial companies | Orbis | |
| Grade2 | Dummy that takes a value of 1 for rating values over BB+ and zero otherwise | Eikon | |
Note(s): 1This is also an independent variable for H2 and H3; the expected sign for H2 and H3 is negative. 2Grade is used for H1
For dependent variables, we used the leverage ratio for H1 and credit rating for H2 and H3. The leverage ratio (Leverage) is the ratio of total debt to total assets and reflects capital structure risk; higher leverage indicates greater default probability (Faccio et al., 2016). Leverage also serves as a control in H2 and H3. Credit rating (Rating) is based on S&P scores converted into an ordinal scale from 1 (CCC or below) to 7 (AAA) (Ashbaugh-Skaife et al., 2006; Bhojraj and Sengupta, 2003; Blume et al., 1998).
For gender variables, we measured the percentage of women on the board (Femaleboard) (Bernile et al., 2018; Rodríguez-Ariza et al., 2017; Sila et al., 2016). To test H3, we used a binary variable (Bequitable) indicating the presence of at least two women, given that the average in our sample was 1.15 (Trinh et al., 2023).
In terms of corporate governance, we considered the percentage of independent directors (Boardindep) to enhance management monitoring (Daily et al., 2003; Johnson et al., 1996) and CEO duality (Duality), a dummy variable that takes the value 1 if the CEO is also the chairperson.
For company fundamentals and control variables, we used nine measures. Interest coverage (Interestcoverage) is the ratio of EBIT to interest expenses (Sila et al., 2016; Ashbaugh-Skaife et al., 2006). The negative earnings (Loss) variable takes the value 1 for firms with negative ROA for two consecutive years and 0 otherwise (Ashbaugh-Skaife et al., 2006). Asset structure (Tangibility) is the ratio of fixed to total assets (Faccio et al., 2016). Profitability (ROA) is the EBIT to total assets ratio (Ashbaugh-Skaife et al., 2006). Firm size (Size) is the natural logarithm of total assets (Ashbaugh-Skaife et al., 2006; Bhojraj and Sengupta, 2003). Market to book (MtB) is the ratio of the market to book value (Samaniego-Medina and di Pietro, 2019). Auditor report (Audit) is a dummy variable (1 for unqualified opinion) (Aman and Nguyen, 2013; Ashbaugh-Skaife et al., 2006). Financial sector (Financial) is a dummy variable (1 for banks and financial institutions). Finally, for H1, we used a classification variable Grade, which is 1 for investment-grade firms and 0 for speculative-grade firms.
3.3 Methodology
We used different models to evaluate our hypotheses; this next section is divided into three parts dedicated to the different hypotheses developed.
3.3.1 Board gender diversity and leverage.
To test H1, we used Arellano and Bond’s generalized method of moments (GMM) approach (Arellano and Bover, 1995; Arellano and Bond, 1991). They proposed the use of GMM with lagged values of the original independent variables as instruments to resolve the problem of endogeneity. Unobservable heterogeneity is controlled for through the individual effect ni; δt is the time variable included to control for the effect of macroeconomic factors on leverage. This is represented in equation (1):
We considered possible heteroskedasticity and used robust standard errors, as in Wooldridge (2002). We also applied the Hansen test of overidentifying restrictions to verify the noncorrelation of the instruments and the standard errors. Finally, to test the absence of any secondary-order serial correlation in the regression residuals, the m2 statistic was used.
3.3.2 Board gender diversity and credit rating.
To verify H2, we considered the rating as the dependent variable. Following prior research (Bhojraj and Sengupta, 2003; Blume et al., 1998), we used an ordered logistic model that offers a better fit than does the standard linear regression model (Mansoor et al., 2021; Lin et al., 2020; Greene, 2000). Ordered logistic regression (OLR) considers a continuous but unobserved variable, Yn, which is a linear function of Xs and a stochastic standard logistic variable (Fullerton, 2009). This yields equation (2):
We then estimated the marginal effects (MEs) of the presence of female board members to measure the direct impacts of a unit change in the respective variable when all other variables are held constant. The ME on the probability of choosing alternative j when regressor xr changes is given by equation (3):
3.4 Critical mass of women on the board and credit rating
To test H3, we developed equation (4) through OLR:
In this equation, we used the binary variable Bequitable to indicate if the board has a critical mass of women. MEs were also estimated for this case.
4. Results
4.1 Descriptive statistics
Table 2 shows the descriptive statistics of the variables. The average percentage of women on boards is 15.84%, with values ranging from 0 to 66.67%. Independent boards account for 50.36%, whereas duality exists in 64.49% of firms. The average leverage ratio is 67.00%. The mean interest coverage is 26.25%, and only 3% of the sample firms registered losses in the previous two years.
Descriptive statistics
| Variable | Mean | Median | SD | Minimum | Maximum |
|---|---|---|---|---|---|
| Leverage | 67.00443 | 64.68084 | 28.08109 | 0 | 742.3808 |
| Rating | 2.641096 | 3 | 1.362492 | 1 | 7 |
| Femaleboard | 15.84137 | 1 | 0.4951254 | 0 | 1 |
| Bequitable | 0.6569358 | 1 | 0.4747489 | 0 | 1 |
| Boardindep | 50.36043 | 50 | 18.40379 | 0 | 93.33 |
| Duality | 0.6488454 | 1 | 0.4773611 | 0 | 1 |
| Interestcoverage | 26.2557 | 4.05188 | 509.58690 | −4544.125 | 34,760 |
| Loss | 0.03753 | 0 | 0.19007 | 0 | 1 |
| Tangibility | 0.30366 | 0.20057 | 0.27872 | 0 | 0.91783 |
| ROA | 0.0682313 | 0.0660674 | 0.1330499 | −5.537217 | 1.951352 |
| Size | 7.826457 | 7.785287 | 1.629001 | −5.115996 | 13.08885 |
| MtB | 4,089.692 | 202.9389 | 3,55879.6 | −89,2602.1 | 1,020,719 |
| Audit | 0.8575342 | 1 | 0.3495392 | 0 | 1 |
| Financial | 0.21096 | 0 | 0.40800 | 0 | 1 |
| Grade | 0.314726 | 0 | 0.4644226 | 0 | 1 |
| Variable | Mean | Median | SD | Minimum | Maximum |
|---|---|---|---|---|---|
| Leverage | 67.00443 | 64.68084 | 28.08109 | 0 | 742.3808 |
| Rating | 2.641096 | 3 | 1.362492 | 1 | 7 |
| Femaleboard | 15.84137 | 1 | 0.4951254 | 0 | 1 |
| Bequitable | 0.6569358 | 1 | 0.4747489 | 0 | 1 |
| Boardindep | 50.36043 | 50 | 18.40379 | 0 | 93.33 |
| Duality | 0.6488454 | 1 | 0.4773611 | 0 | 1 |
| Interestcoverage | 26.2557 | 4.05188 | 509.58690 | −4544.125 | 34,760 |
| Loss | 0.03753 | 0 | 0.19007 | 0 | 1 |
| Tangibility | 0.30366 | 0.20057 | 0.27872 | 0 | 0.91783 |
| ROA | 0.0682313 | 0.0660674 | 0.1330499 | −5.537217 | 1.951352 |
| Size | 7.826457 | 7.785287 | 1.629001 | −5.115996 | 13.08885 |
| MtB | 4,089.692 | 202.9389 | 3,55879.6 | −89,2602.1 | 1,020,719 |
| Audit | 0.8575342 | 1 | 0.3495392 | 0 | 1 |
| Financial | 0.21096 | 0 | 0.40800 | 0 | 1 |
| Grade | 0.314726 | 0 | 0.4644226 | 0 | 1 |
Source(s): Authors’ own work
4.2 Results of the base regression model
Regarding H1, our analysis found this relationship nonsignificant, as shown in Table 3, Model 1 (β = −0.0018, p > 0.1). With respect to the other variables, firms with an unqualified audit report are associated with less leverage, and better interest coverage is positively linked to the use of debt; firms with higher fixed asset investments also present a higher debt ratio. We can therefore confirm H1c that the leverage level is not gender specific. These results are in line with those obtained by Sila et al. (2016), Nelson (2016), Maxfield et al. (2010) and Van Der Walt et al. (2006).
GMM estimation of the effect of female governance on leverage
| Variable | Model 1 |
|---|---|
| Leveraget−1 | 0.9514 *** |
| Femaleboard | −0.0018 |
| Boardindep | −0.0103 |
| Duality | 0.0075 |
| Interestcoverage | 0.0451*** |
| Loss | −0.0415 |
| Tangibility | 0.0890* |
| ROA | 0.0590 |
| Size | −0.0404 |
| MtB | −0.0355 |
| Audit | −0.0791** |
| Financial | 0.0281 |
| Grade | 0.0207 |
| m1 | −2.08** |
| m2 | −0.69 |
| Hansen | 266.98 (0.160) |
| Number of observations | 5015 |
| Number of instruments | 266 |
| Number of groups | 907 |
| Variable | Model 1 |
|---|---|
| Leveraget−1 | 0.9514 *** |
| Femaleboard | −0.0018 |
| Boardindep | −0.0103 |
| Duality | 0.0075 |
| Interestcoverage | 0.0451*** |
| Loss | −0.0415 |
| Tangibility | 0.0890* |
| ROA | 0.0590 |
| Size | −0.0404 |
| MtB | −0.0355 |
| Audit | −0.0791** |
| Financial | 0.0281 |
| Grade | 0.0207 |
| m1 | −2.08** |
| m2 | −0.69 |
| Hansen | 266.98 (0.160) |
| Number of observations | 5015 |
| Number of instruments | 266 |
| Number of groups | 907 |
Note(s): Determinants of the leverage ratio using the difference-GMM estimator. Significance levels are indicated as follows: ***1 level, **5 level and *10% level
H2 is examined in Model 2 (Table 4, OLR of the rating, and Table 5, MEs). The positive and highly significant β value of Femaleboard indicates a positive effect of Femaleboard on rating. Likewise, Boardindep, Duality, Tangibility, ROA, Size and Financial are positively significant. In contrast, Interestcoverage and Leverage are also significant, although they present negative signs. The adjusted R-squared value (24.6%) is reasonably acceptable.
OLR model: impact of BGD on the credit rating
| Variable | Model 2 | Standard error |
|---|---|---|
| Femaleboard | 0.5108*** | 0.0722 |
| Boardindep | 0.1256** | 0.0536 |
| Duality | 0.4108*** | 0.1329 |
| Leverage | −0.5314*** | 0.1058 |
| Interestcoverage | −1.1523*** | 0.3485 |
| Loss | −0.1361 | 0.2232 |
| Tangibility | 0.3324** | 0.1429 |
| ROA | 0.1582** | 0.0718 |
| Size | 3.7386*** | 0.1688 |
| MtB | 0.8690 | 0.7564 |
| Audit | −0.4434 | 0.3521 |
| Financial | 2.2801*** | 0.5305 |
| N | 5,816 | |
| sigma2_u | 24.5982 | |
| Variable | Model 2 | Standard error |
|---|---|---|
| Femaleboard | 0.5108*** | 0.0722 |
| Boardindep | 0.1256** | 0.0536 |
| Duality | 0.4108*** | 0.1329 |
| Leverage | −0.5314*** | 0.1058 |
| Interestcoverage | −1.1523*** | 0.3485 |
| Loss | −0.1361 | 0.2232 |
| Tangibility | 0.3324** | 0.1429 |
| ROA | 0.1582** | 0.0718 |
| Size | 3.7386*** | 0.1688 |
| MtB | 0.8690 | 0.7564 |
| Audit | −0.4434 | 0.3521 |
| Financial | 2.2801*** | 0.5305 |
| N | 5,816 | |
| sigma2_u | 24.5982 | |
Note(s): Determinants of the rating using OLR. The models include year dummies. Significance levels are indicated as follows: ***1 level, **5 level and *10% level
Assessment of the MEs of female board representation on rating categories
| Rating | AMEs | Sig. |
|---|---|---|
| AAA | 0.0002077*** | 0.006 |
| AA+ to AA− | 0.0011658*** | 0.000 |
| A+ to A− | 0.0131536*** | 0.000 |
| BBB+ to BBB− | 0.0186087*** | 0.000 |
| BB+ to BB− | −0.0055564*** | 0.000 |
| B+ to B− | −0.0082796*** | 0.000 |
| CCC+ to D/SD | −0.0192997*** | 0.000 |
| Rating | AMEs | Sig. |
|---|---|---|
| AAA | 0.0002077*** | 0.006 |
| AA+ to AA− | 0.0011658*** | 0.000 |
| A+ to A− | 0.0131536*** | 0.000 |
| BBB+ to BBB− | 0.0186087*** | 0.000 |
| BB+ to BB− | −0.0055564*** | 0.000 |
| B+ to B− | −0.0082796*** | 0.000 |
| CCC+ to D/SD | −0.0192997*** | 0.000 |
Note(s): Average MEs of female representation on the board on the probability of receiving each rating category. Significance levels are indicated as follows: ***1 level, **5 level and *10% level
The MEs in Table 5 show positive values for investment-grade and negative for speculative-grade categories: women increase the likelihood of investment-grade ratings and decrease the chances of speculative-grade ratings. Analysing AMEs by rating category, the positive and significant value of Femaleboard in groups with higher ratings (AAA to BBB) indicates that each additional percentage point of female board representation raises the probability of achieving these ratings by 0.02% (AAA), 0.12% (AA+ to AA−), 1.3% (A+ to A−) and 1.86% (BBB+ to BBB−). Notably, the Femaleboard coefficient quadruples (from −0.0056 to +0.018) when moving from speculative to investment grade. We can thus conclude that an increase in female board representation fosters better ratings and that this promotion is especially important for obtaining an investment-grade rating, confirming H2. Specifically, each additional percentage point of female board presence increases the likelihood of obtaining an investment-grade credit rating by 1.86%. Given that the average board size in our sample is 10.5 members (with a standard deviation of 2.47), the addition of one woman on the board increases, on average, the probability of obtaining an investment-grade rating by approximately 17.71%, with the effect ranging from 13.54% to 21.88%. These results are in line with the positive effect that BGD has on variables such as CSR (Bear et al., 2010) and company reputation (Bernardi et al., 2006).
We tested H3 in Model 3 using OLR (Table 6). The results support our hypothesis for Bequitable (β = 0.4145, p < 0.01), with MEs (Table 7) showing significant positive betas for investment-grade and negative for speculative-grade categories. AMEs indicate that each additional unit increase in female board representation raises the likelihood of an investment-grade rating by 0.02% (AAA), 0.10% (AA+ to AA−), 1.06% (A+ to A−) and 1.49% (BBB+ to BBB−), while reducing the probability of a speculative rating. Again, the greatest increase in the coefficient (from −0.004 to +0.015) occurs during the transition from speculative to investment grade, representing a 4.38-fold increase compared to the prior value. This confirms H3 and aligns with critical mass theory (Kanter, 1977).
OLR model of the effect of the equitable distribution of gender on boards and rating
| Variable | Model 3 | Standard error |
|---|---|---|
| Bequitable | 0.4145*** | 0.1605 |
| Boardindep | 0.1197** | 0.0524 |
| Duality | 0.4028** | 0.1322 |
| Leverage | −0. 4420*** | 0.1027 |
| Interestcoverage | −1.1328*** | 0. 3474 |
| Loss | −0.1427 | 0.2211 |
| Tangibility | 0.3370** | 0.5324 |
| ROA | 0.1159* | 0.0698 |
| Size | 3.8893* | 0.1657 |
| MtB | 0.8620 | 0.7437 |
| Audit | −0.3669 | 0.3497 |
| N | 5,816 | |
| sigma2_u | 24.77 | |
| Variable | Model 3 | Standard error |
|---|---|---|
| Bequitable | 0.4145*** | 0.1605 |
| Boardindep | 0.1197** | 0.0524 |
| Duality | 0.4028** | 0.1322 |
| Leverage | −0. 4420*** | 0.1027 |
| Interestcoverage | −1.1328*** | 0. 3474 |
| Loss | −0.1427 | 0.2211 |
| Tangibility | 0.3370** | 0.5324 |
| ROA | 0.1159* | 0.0698 |
| Size | 3.8893* | 0.1657 |
| MtB | 0.8620 | 0.7437 |
| Audit | −0.3669 | 0.3497 |
| N | 5,816 | |
| sigma2_u | 24.77 | |
Note(s): Determinants of the rating considering a critical mass of women on the boards using OLR. The models include year dummies. Significance levels are indicated as follows: ***1 level, **5 level and *10% level
MEs of gender-equitable boards on firm ratings
| Rating categories | AMEs | Sig. |
|---|---|---|
| AAA | 0.0001739** | 0.020 |
| AA+ to AA− | 0.0009538 *** | 0.001 |
| A+ to A− | 0.0106423 *** | 0.000 |
| BBB+ to BBB− | 0.0149432 *** | 0.000 |
| BB+ to BB− | −0.0044191 *** | 0.000 |
| B+ to B− | −0.0066536*** | 0.000 |
| CCC+ to D/SD | −0.0156405*** | 0.000 |
| Rating categories | AMEs | Sig. |
|---|---|---|
| AAA | 0.0001739** | 0.020 |
| AA+ to AA− | 0.0009538 *** | 0.001 |
| A+ to A− | 0.0106423 *** | 0.000 |
| BBB+ to BBB− | 0.0149432 *** | 0.000 |
| BB+ to BB− | −0.0044191 *** | 0.000 |
| B+ to B− | −0.0066536*** | 0.000 |
| CCC+ to D/SD | −0.0156405*** | 0.000 |
Note(s): Average MEs of a critical mass of women on the board on the probability of obtaining each rating category. Significance levels are indicated as follows: ***1 level, **5 level and *10% level
4.3 Robustness checks
Addressing potential endogeneity is crucial (Coles et al., 2012; Fama and Jensen, 1983). Omitted variables might influence both firm rating and board gender diversity. A higher rating correlates with less risk and could result in a better ESG score (Sila et al., 2016), while greater social responsibility may affect female board selection. Reverse causality could also occur if women self-select into better-rated firms (Farrell and Hersch, 2005). To address this, we transformed the dependent variable (Pérez-Calero Sánchez et al., 2015; Kor and Sundaramurthy, 2009; Wincent et al., 2009) and used GMM estimation. While we used rating as a credit risk measure, other studies have used default probability (García et al., 2021; Moreno et al., 2020; Lingnan, 2019; Baselga-Pascual et al., 2015). We applied the StarMine® Combined Credit Risk Model (Abid et al., 2020; Yan et al., 2014), integrating three credit risk models for a unified estimate. Model 4 estimates H2 and Model 5 estimates H3 (Table 8). The beta coefficients for Femaleboard (β = −0.0362; p = 0.001) and Bequitable (β = −0.1140; p = 0.023) confirm our previous findings regarding H2 and H3.
GMM estimation of the effect of female governance on the default probability
| Variable | Model 4 | Model 5 |
|---|---|---|
| DPt−1 | 0.3951*** | 0.4364*** |
| Femaleboard | −0.0362*** | |
| Bequitable | −0.1140** | |
| Boardindep | 0.0153 | 0.0085 |
| Duality | −0.0154 | 0.0114 |
| Leverage | 0.0998*** | 0.0803*** |
| Interestcoverage | −0.0220 | −0.0057 |
| Loss | 0.2188*** | −0.1252 |
| Tangibility | 0.0019 | 0.0181* |
| ROA | 0.0094 | −0.0152 |
| Size | −0.1029* | −0.1301** |
| MtB | 0.0039 | 0.1081** |
| Audit | −0.0057 | −0.0102 |
| Financial | −0.0738* | −0.0837** |
| m1 | −1.68 (sig 0.093) | −1.77 (sig 0.077) |
| m2 | −0.54 (sig 0.590) | −0.60 (sig 0.551) |
| Hansen test | 28.33 (0.846) | 166.07 (0.295) |
| Number of observations | 5,303 | 5,740 |
| Number of instruments | 140 | 179 |
| Number of groups | 1,037 | 1,033 |
| Variable | Model 4 | Model 5 |
|---|---|---|
| DPt−1 | 0.3951*** | 0.4364*** |
| Femaleboard | −0.0362*** | |
| Bequitable | −0.1140** | |
| Boardindep | 0.0153 | 0.0085 |
| Duality | −0.0154 | 0.0114 |
| Leverage | 0.0998*** | 0.0803*** |
| Interestcoverage | −0.0220 | −0.0057 |
| Loss | 0.2188*** | −0.1252 |
| Tangibility | 0.0019 | 0.0181* |
| ROA | 0.0094 | −0.0152 |
| Size | −0.1029* | −0.1301** |
| MtB | 0.0039 | 0.1081** |
| Audit | −0.0057 | −0.0102 |
| Financial | −0.0738* | −0.0837** |
| m1 | −1.68 (sig 0.093) | −1.77 (sig 0.077) |
| m2 | −0.54 (sig 0.590) | −0.60 (sig 0.551) |
| Hansen test | 28.33 (0.846) | 166.07 (0.295) |
| Number of observations | 5,303 | 5,740 |
| Number of instruments | 140 | 179 |
| Number of groups | 1,037 | 1,033 |
Note(s): Determinants of the default probability using the difference-GMM estimator. The models include year dummies. Significance levels are indicated as follows: ***1% level, ** 5% level and *10% level
Secondly, we modified the dependent variable by identifying instances where firms transitioned between speculative- and investment-grade ratings. We then regressed this variable on the changes in female board participation. The marginal effects of the logistic regression in Table 9 are consistent with the base model. An increase in female board participation raises the probability of a firm improving its credit rating by approximately 0.08 percentage points and reduces the likelihood of a downgrade by around 0.12 percentage points. On the other hand, while the effect on firms maintaining their current credit rating is positive, it is not statistically significant.
MEs of an increase in female board participation on changes in firm rating
| Rating changes | AMEs | Sig. |
|---|---|---|
| Downgrade to speculative grade | −0.00117* | 0.075 |
| No change | 0.00038 | 0.171 |
| Upgrade to investment grade rating | 0.000787* | 0.081 |
| Rating changes | AMEs | Sig. |
|---|---|---|
| Downgrade to speculative grade | −0.00117* | 0.075 |
| No change | 0.00038 | 0.171 |
| Upgrade to investment grade rating | 0.000787* | 0.081 |
Note(s): Average MEs of the increase in female board participation on the probability of a transitioning between speculative-grade and investment-grade ratings. Significance levels are indicated as follows: ***1% level, ** 5% level and * 10% level
We also tested the effect of changes in female board participation by considering cases where at least one female member was either added or removed from the board. The sample including only firms with increased (or decreased) women’s participation on board comprised 913 firms with 4,915 observations. In Table 10, the marginal effects of the ordered logistic model confirm our previous results for H2: the AMEs are positive for companies in the investment grade and negative for speculative-grade firms with the highest impact for the BBB step.
MEs of increase in female board participation on firm ratings
| Rating | AMEs | Sig. |
|---|---|---|
| AAA | 0.00000583 | 0.176 |
| AA+ to AA− | 0.0000446* | 0.061 |
| A+ to A− | 0.0004576** | 0.034 |
| BBB+ to BBB− | 0.0002805** | 0.032 |
| BB+ to BB− | 0.0001132** | 0.035 |
| B+ to B− | −0.0002659** | 0.033 |
| CCC+ to D/SD | −0.0006059** | 0.033 |
| Rating | AMEs | Sig. |
|---|---|---|
| AAA | 0.00000583 | 0.176 |
| AA+ to AA− | 0.0000446* | 0.061 |
| A+ to A− | 0.0004576** | 0.034 |
| BBB+ to BBB− | 0.0002805** | 0.032 |
| BB+ to BB− | 0.0001132** | 0.035 |
| B+ to B− | −0.0002659** | 0.033 |
| CCC+ to D/SD | −0.0006059** | 0.033 |
Note(s): Average MEs of the increase in female board participation on the probability of obtaining each rating category. Significance levels are indicated as follows: *** 1% level, ** 5% level and * 10% level
Furthermore, to address any sample selection bias, we applied propensity score matching (Rosenbaum and Rubin, 1983) to study empirically any changes in credit ratings due to the existence of a critical mass of women on a firm’s board. The sample was thus divided into two groups: the treatment group of firms with gender-balanced boards (Bequitable = 1) and the control group of firms that did not meet this requirement (Bequitable = 0). In all regressions, we used cluster-robust standard errors.
The untabulated results of the MEs indicate that the likelihood of having a gender-balanced board is higher for larger, more leveraged and more profitable firms, while this probability is lower for companies based on more tangible assets. Moreover, more independent boards and firms with better audit reports also present a higher probability of having a gender-balanced board. Above all, we found evidence that there are significant differences between the treatment and control groups that could influence their boards’ gender composition. To build the counterfactual group, we used nearest neighbour matching with the psmatch2 command (Leuven and Sianesi, 2003). After the matching process, we found no significance in the mean differences between the matched treatment and control firms (Table 11), which indicates that they are equivalent in their observable characteristics. The pseudo R2 value after matching is reduced to just 0.004, while the mean and median biases fall from 17.2 to 3.5 and from 12.6 to 3.5, respectively. We found a significant average treatment effect for the treated group, which is 0.4323 higher for firms with gender-balanced boards; the credit ratings of such firms are thus positively affected. The consistency of these results continues to support H3.
Test of the balancing hypothesis
| Variables | Unmatched (U)/matched (M) | Mean | % reduct | p-value | ||
|---|---|---|---|---|---|---|
| Treated | Control | % bias | |bias| | |||
| Boardindep | U | 0.1466 | −0.0750 | 22.5 | 0.000 | |
| M | 0.1466 | 0.1858 | −4.0 | 82.3 | 0.107 | |
| Leverage | U | −0.0000 | −0.1696 | 20.9 | 0.000 | |
| M | −0.0000 | 0.1064 | −13.1 | 37.2 | 0.000 | |
| Interestcoverage | U | −0.0188 | −0.0155 | −1.6 | 0.537 | |
| M | −0.0188 | −0.0182 | −0.3 | 79.7 | 0.857 | |
| Loss | U | 0.1811 | 0.0465 | −16.1 | 0.000 | |
| M | 0.1811 | 0.0197 | −0.9 | 94.4 | 0.644 | |
| Tangibility | U | −0.0699 | 0.0514 | −12.6 | 0.000 | |
| M | −0.0699 | −0.0264 | −4.5 | 64.2 | 0.058 | |
| ROA | U | 0.2166 | 0.0726 | 17.2 | 0.000 | |
| M | 0.2166 | 0.2652 | −5.8 | 66.3 | 0.012 | |
| Size | U | 0.2166 | 0.0726 | 17.2 | 0.000 | |
| M | 0.2166 | 0.2651 | −5.8 | 66.3 | 0.012 | |
| MtB | U | −0.0072 | −0.0099 | 4.9 | 0.066 | |
| M | −0.0072 | −0.0095 | 4.1 | 17.5 | 0.212 | |
| Audit | U | 0.9155 | 0.8992 | 5.6 | 0.032 | |
| M | 0.9155 | 0.9082 | 2.5 | 55.2 | 0.307 | |
| Financial | U | 0.1258 | 0.1293 | −1.1 | 0.689 | |
| M | 0.1258 | 0.1255 | 0.1 | 91.0 | 0.970 | |
| Variables | Unmatched (U)/matched (M) | Mean | % reduct | p-value | ||
|---|---|---|---|---|---|---|
| Treated | Control | % bias | |bias| | |||
| Boardindep | U | 0.1466 | −0.0750 | 22.5 | 0.000 | |
| M | 0.1466 | 0.1858 | −4.0 | 82.3 | 0.107 | |
| Leverage | U | −0.0000 | −0.1696 | 20.9 | 0.000 | |
| M | −0.0000 | 0.1064 | −13.1 | 37.2 | 0.000 | |
| Interestcoverage | U | −0.0188 | −0.0155 | −1.6 | 0.537 | |
| M | −0.0188 | −0.0182 | −0.3 | 79.7 | 0.857 | |
| Loss | U | 0.1811 | 0.0465 | −16.1 | 0.000 | |
| M | 0.1811 | 0.0197 | −0.9 | 94.4 | 0.644 | |
| Tangibility | U | −0.0699 | 0.0514 | −12.6 | 0.000 | |
| M | −0.0699 | −0.0264 | −4.5 | 64.2 | 0.058 | |
| ROA | U | 0.2166 | 0.0726 | 17.2 | 0.000 | |
| M | 0.2166 | 0.2652 | −5.8 | 66.3 | 0.012 | |
| Size | U | 0.2166 | 0.0726 | 17.2 | 0.000 | |
| M | 0.2166 | 0.2651 | −5.8 | 66.3 | 0.012 | |
| MtB | U | −0.0072 | −0.0099 | 4.9 | 0.066 | |
| M | −0.0072 | −0.0095 | 4.1 | 17.5 | 0.212 | |
| Audit | U | 0.9155 | 0.8992 | 5.6 | 0.032 | |
| M | 0.9155 | 0.9082 | 2.5 | 55.2 | 0.307 | |
| Financial | U | 0.1258 | 0.1293 | −1.1 | 0.689 | |
| M | 0.1258 | 0.1255 | 0.1 | 91.0 | 0.970 | |
Source(s): Authors’ own work
In our sample, we observed a clear difference in female board participation between firms with higher and lower credit ratings. Specifically, 91.5% of AAA-rated firms have women on their boards, while 81.4% of CCC-rated firms have exclusively male boards. This pattern suggests that firms with better ratings, which tend to be larger (as evidenced by a correlation of 0.55), have more gender-diverse boards compared to those with lower ratings, which are often smaller. To investigate whether the effect of female board participation differs between large and small firms, we divided the sample at the median and re-estimated the initial regressions (Table 12).
MEs of female board participation on firm ratings by firm size
| Rating | Small firms | Sig. | Big firms | Sig. |
|---|---|---|---|---|
| AAA | 0.0002077* | 0.068 | 0.00039* | 0.079 |
| AA+ to AA− | 0.0011658*** | 0.000 | 0.001861*** | 0.002 |
| A+ to A− | 0.0131536*** | 0.000 | 0.016402*** | 0.000 |
| BBB+ to BBB− | 0.0186087*** | 0.000 | 0.011255*** | 0.001 |
| BB+ to BB− | −0.0055564*** | 0.000 | −0.010551*** | 0.000 |
| B+ to B− | −0.0082796*** | 0.001 | −0.007515*** | 0.000 |
| CCC+ to D/SD | −0.0192997*** | 0.000 | −0.011843*** | 0.000 |
| Rating | Small firms | Sig. | Big firms | Sig. |
|---|---|---|---|---|
| AAA | 0.0002077* | 0.068 | 0.00039* | 0.079 |
| AA+ to AA− | 0.0011658*** | 0.000 | 0.001861*** | 0.002 |
| A+ to A− | 0.0131536*** | 0.000 | 0.016402*** | 0.000 |
| BBB+ to BBB− | 0.0186087*** | 0.000 | 0.011255*** | 0.001 |
| BB+ to BB− | −0.0055564*** | 0.000 | −0.010551*** | 0.000 |
| B+ to B− | −0.0082796*** | 0.001 | −0.007515*** | 0.000 |
| CCC+ to D/SD | −0.0192997*** | 0.000 | −0.011843*** | 0.000 |
Note(s): Average MEs of female board participation on the probability of obtaining each rating category by firm size. Significance levels are indicated as follows: *** 1% level, ** 5% level and * 10% level
The results in Table 13 highlight an interesting trend when comparing small and large firms across different credit ratings. The most significant leap in credit rating occurs between the BB and BBB categories, particularly for smaller firms, where the change is notably larger than for larger firms. In smaller firms, the coefficient for BBB more than quadruples that of BB, whereas in larger firms, the jump is only double. Nevertheless, the patterns for the effect of female board participation are quite similar in both cases, showing a decrease in impact starting from the BBB category.
GMM estimation of the effect of female governance on the cost of debt
| Variable | Model 6 |
|---|---|
| Cost of debt−1 | 0.3069** |
| Femaleboard | −0.0002** |
| Boardindep | 0.0015** |
| Duality | 0.0011 |
| Leverage | −0.0003 |
| Interestcoverage | −0.0391* |
| Loss | 0.0105** |
| Tangibility | 0.0012 |
| ROA | −0.0019 |
| Size | −0.0404 |
| MtB | −0.0239 |
| Audit | −0.0791** |
| Financial | −0.0034 |
| m1 | −2.08** |
| m2 | −0.69 |
| Hansen | 266.98 (0.160) |
| Number of observations | 5,284 |
| Number of instruments | 116 |
| Number of groups | 1,036 |
| Variable | Model 6 |
|---|---|
| Cost of debt−1 | 0.3069** |
| Femaleboard | −0.0002** |
| Boardindep | 0.0015** |
| Duality | 0.0011 |
| Leverage | −0.0003 |
| Interestcoverage | −0.0391* |
| Loss | 0.0105** |
| Tangibility | 0.0012 |
| ROA | −0.0019 |
| Size | −0.0404 |
| MtB | −0.0239 |
| Audit | −0.0791** |
| Financial | −0.0034 |
| m1 | −2.08** |
| m2 | −0.69 |
| Hansen | 266.98 (0.160) |
| Number of observations | 5,284 |
| Number of instruments | 116 |
| Number of groups | 1,036 |
Note(s): Determinants of the default probability using the difference-GMM estimator. The models include year dummies. Significance levels are indicated as follows: ***1% level, ** 5% level and *10% level
Finally, we examined whether a more gender-diverse board affects a firm’s borrowing costs, offering a complementary perspective to the existing understanding of leverage dynamics. Specifically, we investigated the effect of the percentage of women on the board on a firm’s financial structure by examining its influence on the effective cost of debt (Table 13).
The results presented in Table 13 suggest that, despite the lack of a significant effect on leverage, a higher percentage of women on the board is associated with a slight reduction in a firm’s cost of debt. On average, for each additional woman on the board, the cost of debt decreases by 0.002%. This statistically significant finding implies that gender-diverse boards may contribute to lower borrowing costs, potentially due to improved governance or a more favourable risk perception by lenders.
5. Discussion
Previous studies have explored the role BGD in various aspects of corporate governance and performance, with mixed findings regarding its influence on financial outcomes (Ain et al., 2020; Maxfield and Wang, 2024). Drawing on agency theory and resource dependence theory, this study examined the impact of BGD on leverage, credit ratings and the presence of a critical mass of women on boards. Our results provide nuanced insights that contribute to these theoretical frameworks.
Firstly, our analysis confirmed that there is no significant relationship between BGD and leverage, thus supporting the notion that gender diversity on boards does not directly influence a firm’s capital structure. This finding is consistent with prior research that has not established a definitive link between board composition and leverage (Sila et al., 2016; Van Der Walt et al., 2006).
Secondly, our findings revealed that female participation on boards has a positive and significant impact on credit ratings, particularly in the transition from speculative- to investment-grade ratings. This result aligns with agency theory, as it indicates that women directors enhance monitoring and oversight functions, thus reducing agency conflicts and improving managerial accountability (Ain et al., 2020; Maxfield and Wang, 2024). From a resource-dependence perspective, this relationship may also reflect the unique skills, perspectives and stakeholder engagement that women bring to boardrooms, which contribute to improved decision-making and risk management (Darmawan, 2024; McGuinness et al., 2017). These attributes appear to be recognized by credit rating agencies, which likely perceive gender-diverse boards as a signal of robust governance and lower financial risk. This aligns with broader findings on the role of governance in reducing risk and improving corporate evaluations (Jallali and Zoghlami, 2022), particularly in contexts where internal governance mechanisms, such as board composition and risk oversight, are key determinants of external ratings.
Thirdly, our results suggest that achieving a critical mass of women on boards positively influences credit ratings by acting as an independent indicator of governance quality. This aligns with Kanter’s (1977) critical mass theory, which suggests that reaching a threshold of minority representation fosters more collaborative and effective board dynamics. A critical mass of women strengthens governance processes and enhances external perceptions of a firm’s stability by fostering inclusivity and encouraging balanced decision-making. These results are consistent with prior research highlighting the importance of stakeholder engagement and transparency in improving creditworthiness (McGuinness et al., 2017; Ramon-Llorens et al., 2020).
While these findings indicate that the benefits of BGD may vary across firms, they provide robust evidence that board diversity initiatives are not merely symbolic but carry tangible governance and financial benefits. These contributions reinforce the strategic value of gender diversity on boards, especially for firms seeking to improve external perceptions of governance and stability. Consistent with Salehi et al. (2024), our findings highlight the broader implications of governance mechanisms in reducing financial risk and improving credit perceptions.
Overall, this study contributes to the ongoing debate on the role of BGD in corporate governance by demonstrating its significant impact on credit ratings, particularly during critical transitions. These results underscore the importance of gender diversity as a strategic asset for enhancing governance and strengthening external evaluations of firm performance.
6. Concluding remarks
In this study, we examined BGD as a mechanism for corporate risk governance. Specifically, we investigated whether the percentage of women on the board is associated with a company’s leverage level and credit rating. We also explored whether the presence of a critical mass of women on the board increases the likelihood of achieving a higher credit rating.
Our findings indicate that while the percentage of women on the board is not linked to a firm’s leverage level, it does have a significant and positive impact on credit ratings. Specifically, we conclude that, on average, each additional woman on the board increases the probability of a firm transitioning from speculative- to investment-grade ratings by 17.71%. Finally, having a critical mass of women on the board enhances credit ratings and raises the probability of achieving better ratings. These results suggest that rating agencies view female board members as valuable intangible assets. This underscores the importance of gender diversity as a key element in risk governance strategies. For firms – especially those with poor credit ratings or no female representation – achieving gender balance on boards is not only an issue of equity but also a strategic approach to enhance credit ratings and improve risk management. This positive effect underscores the role of BGD in improving corporate risk governance, as diverse boards are better equipped to manage and mitigate risks. Enhanced risk governance through gender diversity not only strengthens credit evaluations but also contributes to more robust and resilient corporate strategies.
7. Limitations and future research
While this study provides valuable insights, it has certain limitations. The focus on North American firms from 2008 to 2017, while useful in controlling for specific regulatory and economic factors, restricts the generalizability of findings to other regions and time periods. Additionally, although sectoral controls were applied, the lack of detailed industry-level analysis means the results may not fully capture sector-specific dynamics. Future research could explore these variations more deeply. Finally, this study examines BGD primarily through the proportion of women and the presence of a critical mass, without considering differences in specific roles or levels of influence within the board. Investigating these internal dynamics could offer a richer understanding of how BGD affects governance and financial outcomes.
Note
Croson and Gneezy (2009) review the literature on gender differences in economic experiments.
The authors are especially grateful for the insightful comments and suggestions provided by Arnd Wiedemann, Martin R.W. Hiebl, Volker Stein, and the other participants of the 13th Annual Risk Governance Conference at the University of Siegen. Additionally, the authors would like to thank Proof-Reading-Service.com for their assistance in editing our manuscript.
Funding: The authors acknowledge the financial support of the research project Ref. PID2021-128420OB-I00 funded by MCIN/AEI/10.13039/501100011033 and by “ERDF A way of making Europe”, a research project funded by Consejería de Universidad, Investigación e Innovación (Junta de Andalucia, PAIDI 2021) Ref. “ProyExcel_00934” and the Regional Government of Andalusia, Spain (Research Group SEJ-555).

