Board gender composition and waste management: Cross-country evidence

.


Introduction
Climate change triggered by the release of greenhouse gas emissions and toxic chemicals poses the biggest challenge in generations (IPCC, 2007). The situation is worsening at an alarming rate due to an increase in solid waste generation; this causes the release of pollutants and greenhouse gasses from landfill sites worldwide, resulting in an upsurge in ground-level temperature (Prado-Lorenzo, Rodríguez-Domínguez, Gallego-Álvarez, & García-Sánchez, 2009). According to the World Bank (2019), approximately two billion tons of solid waste were generated worldwide in 2016: this is expected to increase to 3.40 billion tons by 2050 due to enormous growth in the population and as a result of rapid industrialization. Corporations around the globe are playing leading roles in waste generation. Recently, three large US-based manufacturing companies (i.e., Fuchs Lubricants Co., United Industries Corporation, and DCW Casing relationship with waste generation (waste recycling).
Our study contributes to the literature in several ways. First, this is the first study examining the relationship between BGD and corporate waste management activities. Particularly, our study contributes to the sustainability accounting and green finance literature by confirming that an effective board structure with more female directors is desirable for making firms genuinely sustainable, which in turn can bring superior financial performance. The existing literature has generally focused on examining the main pillars of CSR while ignoring the subdimensions of such a pillar. This may cause a CSR policy-practice disconnect and is referred as CSR decoupling (Tashman, Marano, & Kostova, 2019). However, a few studies focus on the association between female directors and subdimensions, such as sustainable investment , environmental disclosure (Liao, Luo, & Tang, 2015), renewable energy consumption , and biodiversity (Haque & Jones, 2020). To heed the recent calls for further examination of sustainability pillars' subdimensions (García-Sánchez, Gallego-Álvarez, & Zafra-Gómez, 2021;Hussain, Rigoni, & Cavezzali, 2018), we provide novel evidence to the existing research on the relationship between BGD and waste management. From a policy perspective, our findings offer new insights to regulators seeking sustainability and urge firms to appoint more female directors by extending the business case of BGD.
Second, our study contributes to BGD literature by empirically examining the assumption of the critical mass of female directors. We find empirical support suggesting that the threshold of two or more female directors is needed to influence corporate decisions, thereby contributing to the extant literature (e.g., Liu, Wei, & Xie, 2014;Owen & Temesvary, 2018). We also distinguish the channel (i. e., independent female vs executive female directors) through which female directors influence waste management activities. We find a more pronounced effect of independent female directors (Liu et al., 2014;Chen, Leung, & Goergen, 2017), suggesting a differential effect of the channel. Moreover, our study sheds new light on the impact of sustainable compensation policies on the relationship between female directors and waste management. Finally, we explore the combined effect of BGD and national culture dimensions, i. e., masculinity vs femininity, on waste management activities. This perspective provides novel insights into the link between corporate environmental initiatives and board composition by revealing that the masculinity dimension of national culture is an important decision-making factor. Therefore, we fill the void in the fragmented gender diversity studies on green finance by considering national culture dimensions and sustainable compensation policies.
The rest of the paper is structured as follows: In Section 2, we present a review of the relevant literature and hypotheses. Section 3 discusses the research design. Section 4 reports the empirical results, robustness checks, and identification strategies. Section 5 comprises additional analyses. Finally, Section 6 concludes the study.

Literature review and hypothesis development
This section reviews the literature related to BGD and its connection to the business case. This section further explores the gender socialization and ethicality theories that provide underpinnings on why females are more likely to be expected to make proenvironmental decisions than their male counterparts. We develop our hypotheses after this discussion.

Business case for gender diversity
A strand of literature empirically shows that BGD improves the quality of corporate decision-making (Kang, Cheng, & Gray, 2007;Croson & Gneezy, 2009). 2 Extant literature explains at least two channels through which BGD improves decision-making and, thus, corporate outcomes. First, female directors enhance the quality of corporate governance practices. For instance, female directors influence the monitoring function of the board, as evidenced by regular board meetings and better attendance at such meetings (Adams & Ferreira, 2009). Female directors have been linked with dividend payout (Chen et al., 2017) for firms with otherwise weaker governance structures.
Second, female directors reduce corporate misconduct and malpractices that may be damaging to the firm's reputation. Female directors also improve governance quality by increasing engaged deliberations regarding strategic decisions such as mergers and acquisitions (Levi, Li, & Zhang, 2014) and lessening male CEO overconfidence (Chen, Leung, Song, & Goergen, 2019). Prior literature concurs that BGD reduces corporate fraud (Cumming et al., 2015), improves earnings quality by reducing earnings management (García Lara, García Osma, Mora, & Scapin, 2017;Gull, Nekhili, Nagati, & Chtioui, 2018), and decreases corporate tax aggressiveness (Lanis, Richardson, & Taylor, 2017). Moreover, female directors enhance a firm reputation through their emphasis on greater stakeholder orientation (Adams & Funk, 2012;Hill & Jones, 1992), are less likely to lay off the workforce , and tend to have a greater environmental orientation (e.g., Hafsi & Turgut, 2013;Al-Shaer & Zaman, 2016). In a nutshell, the above discussion indicates that female directors enhance the quality of governance and decision-making and improve the firm's reputation and environmental orientation.

BGD, morality, and environmental concerns
Gender socialization and ethicality theories indicate that female directors are more empathetic to societal issues than their male 2 To some extent, there is a disagreement in the literature regarding the impact of board gender diversity on firm value; however, the evidence is inconclusive (e.g., Ali, Ng, & Kulik, 2014;Ben-Amar, Francoeur, Hafsi, & Lebelle, 2013;Campbell & Minguez-Vera, 2008;Erhardt et al., 2003;Owen & Temesvary, 2018). counterparts for at least two reasons. First, women characterize morality and ethics in a different way than men. For instance, morality for women is about responsibilities, including duty of care and concern for others. This concern leads to the mitigation of the "real and recognizable trouble of this world" (Gilligan, 1977, p. 511). On the other hand, morality for men is mainly focused on the right to life, not interfering with the rights of others and self-gratification. Second, helping behavior also varies between the genders (Eagly & Crowley, 1986). For example, men generally focus on gallant and short-term actions, while women's helping behavior is more often caring and long-term in nature.
Of direct relevance to our study is the fact that tackling environmental issues requires long-term actions; thus, we conjecture that women are more likely to take pro-environmental actions, including reducing waste generation and improving recycling activities. Prior literature supports the notion that female managers care more about society compared to their male counterparts (Tormo-Carbó, Oltra, Klimkiewicz, & Seguí-Mas, 2019). Female directors are not inclined to defend malpractices, including matters related to tax evasion and bribery (Chen, Velasquez Tuliao, Cullen, & Chang, 2016), show positive attitudes towards ethical behavior (Ibrahim, Angelidis, & Tomic, 2009), and promote renewable energy consumption  for environmental protection. Given these arguments, female directors are expected to consider long-term societal issues as more important (i.e., environmental concerns) and should therefore be inclined to reduce waste generation and promote the recycling of waste in their firms.
Consistent with the assertions of gender socialization (Dawson, 1997) and ethical theories, Boulouta (2013) supports the compassionate nature of women, as they tend to reduce malpractices in business that result in higher CSRratings and thus increase corporate reputation (Bear, Rahman, & Post, 2010;McGuinness et al., 2017). Moreover, female directors are associated with voluntary disclosures regarding CSR practices valued by the market (Nekhili, Nagati, Chtioui, & Nekhili, 2017). In addition, female directors are more concerned about firms' environmental bearing. For instance, Pearl-Martinez and Stephens (2016) argue that higher workforce gender diversity is crucial for the push toward a more sustainable society. Extant literature supports this assertion, as firms with female directors perform better on environmental performance fronts (see, e.g., Post, Rahman, & Rubow, 2011). Finally, Liu (2018) and  suggest that firms with BGD are less likely to face environmental lawsuits and more likely to promote environmental initiatives.
Based on the above discussion, we conjecture that female directors have a higher tendency to take up environmental concernsone of society's long-term challenges. We anticipate that corporations with female directors are more inclined to address environmental issues by reducing waste generation and increasing recycling activities. Hence, we hypothesize that: H1. Female directors on the board reduce waste generation and enhance recycling activities.
It is evident from the literature review that one woman is considered to be insufficient to affect corporate decision-making. The reason lies in the distorted perceptions of male observers about women's leadership qualities that are the result of gender role stereotypes (Block, 1973;Kanter, 1977;Sherrick, Hoewe, & Waddell, 2014). This kind of bias about female directors impedes them to be influential in corporate decision-making (Terjesen, Sealy, & Singh, 2009). Hence, Kanter (1977) presumes that a single woman is a "token", i.e., an individual representing a specific cohort. This perceived token significance of women underpins the stereotypes that females lack the attributes required to serve in top positions, thus creating a glass ceiling for women's career ladders (Lee & James, 2007). Given the perceived token status attached to women, a critical mass of female directors on the board becomes crucial to ensure their impact on corporate decision-making.
Many prior studies support the fact that more than one female director is required to influence corporate policies (see, for instance, Terjesen et al., 2009;Jia & Zhang, 2013;Owen & Temesvary, 2018;Atif, Liu, & Huang, 2019). Boards are expected to experience significant change when the number of women on the board becomes larger since the women will feel stronger and less constrained (Terjesen et al., 2009). There is an emerging consensus in the extant literature that the increased representation of women results in better environmental decisions, as females are more likely to direct the board's attention toward environmental issues (Al-Shaer & Zaman, 2016).
Critical mass theory suggests that once female directors reach a certain number (i.e., two or more) on the board, they become more influential. Kristie (2011) observes the critical mass as follows: one female director on the board is a token, two are a presence, and three become a voice. This expectation is empirically validated by Post et al. (2011), Liu et al. (2014, and Atif et al. (2019). Therefore, we present our hypothesis as follows: H2. Two or more female directors on the board have a significantly negative impact on waste generation and enhance recycling activities.
The board of directors, being the governing body of the firm, influences decision-making related to waste generation and recycling activities given their supervisory roles (Hillman & Dalziel, 2003;García Lara et al., 2017). However, the role of independent female directors on the board is critical to board efficiency , as it serves as a retort to stakeholders' pressure and expectations of higher board independence. This is because stakeholders may reasonably question the effectiveness of the board if executive directors dominate the board, given their pecuniary relationship with the firm. Extant literature concurs that independent female directors have a pronounced impact on decision-making compared to executive directors, including in relation to firm performance and dividend payouts (e.g., Chen et al., 2017;Liu et al., 2014). The presence of independent female directors on the board is related to the low rate of financial fraud incidence and higher transparency (Erhardt, Werbel, & Shrader, 2003). In a similar context,  and Atif et al. (2020) show that independent female directors have a significant impact on promoting renewable energy and sustainable investment in firms. Therefore, it is plausible to expect that independent female directors are more likely to direct firm resources towards waste management activities than executive female directors. Therefore, we present our next hypothesis as follows: H3. The presence of independent female directors, rather than executive female directors, reduces waste generation and increases recycling.

Sample
We construct our sample by extracting data from multiple sources. Our initial sample consists of firms included in the Thomson Reuters Asset4 database, which provides waste generation and recycling data. We collect females' board directorships and other corporate governance data from BoardEx, accounting data from WorldScope, and gross domestic product (GDP) data from the World Bank. We then merge the data from all sources and eliminate firm-year observations with missing data on any of the variables required to perform our analyses. Consistent with prior literature (Chen et al., 2019), we exclude countries with less than ten firm-year observations and winsorize all continuous variables at the bottom 1% and top 99% levels to remove the potential impact of outliers. These data filters yield an unbalanced panel of 8365 firm-year observations for 1581 unique firms between 2002 and 2017 across 37 countries. The definitions of all the variables are given in Table A1 of the appendix. Table A2 in the appendix provides the distribution of the sample and main variables: namely, female directors and waste generation by country and year. Panel A reveals that a large number of firm-year observations are from the US (1,358), UK (1,028), and Japan (862). The top three countries allocating more board seats to women are Norway (45.9%), France (38.7%), and Finland (28.6%). This demonstrates that the introduction of board gender quota laws in Europe has increased the representation of women on corporate boards. Finally, corporations in the three countries producing the most waste (average tons) are the Russian Federation (14.092), Canada (13.370), and Ireland (11.995). Interestingly, Panel B reports an increasing trend in the number of observations and female directors throughout the sample period. However, the level of waste produced by sample firms ranges between 10.484 and 11.178 tons, with a continuous decrease in recent years (2012)(2013)(2014)(2015)(2016)(2017). This indicates to some extent that increasing trends in female board representation may affect waste generation negatively.

Measurement of main variables
The dependent variable of interest in this study is waste management, which is operationalized using multiple proxies. In the recent literature (Benjamin, Regasa, Wellalage, & M Marathamuthu, 2020), waste is measured as the total amount of waste corporations generate and report. Specifically, in the main analysis, we measure waste (WASTE) as the natural logarithm of the total waste (both hazardous and non-hazardous) generated in tons. As an alternative to our primary measure of waste, we also focus on both components of waste, that is, hazardous and non-hazardous waste. Hazardous waste (H_WASTE) and non-hazardous waste (NH_WASTE) are measured as the natural logarithm of the hazardous and non-hazardous waste generated in tons, respectively. We then go beyond the prior literature (Benjamin et al., 2020) and consider another aspect of waste management, namely, waste recycling (R_WASTE), which is measured as the ratio of the recycled waste to total waste generated. Finally, under robustness analysis, we operationalize our measure of waste in two different ways; namely, as the ratio of the total waste in tons to total assets (WASTE/ASSETS) and the ratio of the total waste in tons to total sales (WASTE/SALES) to incorporate size and level of production activity at the firm level.
Our independent variable of interest is BGD. Consistent with extant literature (e.g., Gull et al., 2018;, we measure BGD by the proportion of female directors on the board (F_PRO) in our main analysis (H1) and alternatively by the number of female directors on the board (F_NUM), with a dummy variable capturing the presence of female directors on the board (F_DUMMY). We then employ three dummy variables, F1, F2, and F3, to test the validity of critical mass theory (H2). Following Torchia et al. (2011) and Atif et al. (2019), F1 is set equal to 1 if one female director is present on the board and 0 otherwise; F2 is set equal to 1 if two female directors are present on the board, and 0 otherwise; and F3 is set equal to 1 if three or more female directors are present on the board, and 0 otherwise. Finally, we focus on the specific role assigned to female directors on the board to disentangle the advisory and monitoring effect (H3). This analysis helps illustrate whether female directors have a different impact on waste management based on their role within the board of directors. Following existing literature Nekhili, Gull, Chtioui, & Radhouane, 2020;Nekhili, Bennouri, & Nagati, 2022), we measure independent female directors (FNED_PRO) and executive female directors (FED_PRO) as the proportion of independent female and executive directors on the board, respectively.

Empirical models
To test our hypotheses (H1, H2, and H3) examining the effect of BGD on waste management, we run the following models using the ordinary least squares (OLS) regression technique.
where, Waste Management in three equations represents our proxies for waste management: namely, total waste (WASTE), hazardous waste (H_WASTE), non-hazardous waste (NH_WASTE), and waste recycling (R_WASTE). F_PRO is the proportion of female directors on the board. F1, F2, and F3 are dummy variables capturing the presence of one, two, and three or more female directors on the board, respectively. FNED_PRO and FED_PRO represent the proportion of independent and executive female directors on the board, respectively. Controls refers to the set of control variables that may also impact the level of waste generation and recycling by the sample firms. Our selection of control variables is based on prior studies (e.g., Ahmed, Atif, & Gyapong, 2021;Chen et al., 2017). These variables include board size (B_SIZE), board independence (B_IND), separation of the CEO and board chair roles (SEPARATE), presence of a CSR committee (CSR_COM), third party assurance of CSR reports (CSR_AUDIT), research and development intensity (R&D), profitability (ROA), Tobin's Q (TQ), financial loss (LOSS), the level of cash holdings (CASH), financial leverage (LEVERAGE), annual turnover (SALES), firm size (SIZE), and annual GDP growth (GDP). We also control for year, industry, and country effects. The definitions of all the variables are given in Table A1 in the appendix. Table 1 exhibits the descriptive statistics based on the whole sample. Concerning several measures of waste, the mean value of waste (WASTE), hazardous waste (H_WASTE), non-hazardous waste (NH_WASTE), and recycled waste (R_WASTE) is 10.959, 8.185, 11.074, and 60.6%, respectively. For female directors, the results show that the average proportion and number of female directors (F_PRO & F_NUM) are 15.4% and 1.850, respectively. Furthermore, 76.5% of the sample firms have one female (F_DUMMY) on their board of directors. Regarding the gender balance of the corporate boards, the findings reveal that, on average, 23.7%, 22.5%, and 30.1% of firms appoint one female (F1), two females (F3), and three or more females (F3) on their board of directors, respectively. The mean percentage of independent female directors (FNED_PRO) and executive female directors (FED_PRO) is 14.1% and 1.3%, respectively. Interestingly, these statistics indicate that the sample firms tend to appoint females as independent rather than executive directors. Finally, the mean value of Hofstede's cultural index on masculinity (MAS) is 57.361.

Univariate results
With regard to the control variables, Table 1 shows that the mean board size (B_SIZE) is 2.413 (natural logarithm value), and the average level of board independence (B_IND) is 77.7%. On average, 34.7% of the firms have separated the CEO, and board chair position (SEPARATE), 82.7% of the firms have set up a sustainability committee (CSR_COM), and 73.0% of the firms have assured their sustainability/CSR reports through third party audit (CSR_AUDIT). The mean value of research and development intensity (R&D) is  Note: This table presents the summary statistics for all variables based on the whole sample and mean differences for the subsample of firms with and without female directors. *, ** and, *** represent significance at the 0.1, 0.05 and 0.01 levels, respectively. All the variables are defined in Appendix Table A1. Gull et al. 0.023; return on assets (ROA) shows a mean percentage of 5.814; and the average value of Tobin's Q (TQ) is 1.608. On average, 8.3% of firms report negative earnings (LOSS), and the mean value of cash holdings is 0.110. On average, 25.9% of the assets are financed by debt (LEVERAGE), and the mean value of turnover is 17.619. On average, the firm size (SIZE) is 16.949, and the mean percentage of GDP growth (GDP) is 1.749. Table 1 also provides a summary of the mean differences in the variables between firm years with and without female directors. It shows that boards of 23.7% of sample firms are solely composed of male directors. In line with our expectations, firms without female directors relatively generate more waste as measured by the mean difference tests based on total waste (WASTE), waste to assets ratio (WASTE/ASSETS), and waste to sales ratio (WASTE/SALES); this difference is significant at the 1% level. Firms with and without female directors do not differ significantly based on the level of hazardous (H_WASTE) and non-hazardous waste (NH_WASTE). However, firms without female directors are more likely to recycle their waste (R_WASTE), and this difference is significant at the 1% level. Interestingly, the mean value of Hofstede's cultural index on masculinity (MAS) is also significantly high for firms without female directors, suggesting that such firms are more likely to be found in highly masculine societies. In short, firms with and without female directors differ significantly based on control variables, and these differences are statistically significant for all variables except research and development intensity (R&D), financial loss (LOSS), and GDP growth (GDP).
We check the correlation among the variables and make sure that our results are not spurious due to the multicollinearity problem; Table A3 in the appendix provides the correlations matrix. Extant literature (Atif & Ali, 2021;Atif et al., 2019;Liu et al., 2014) suggests that a correlation greater than 0.70 or variance inflation factors (VIFs) larger than ten may indicate the issue of multicollinearity. Table A3 shows that the correlation among all the variables is less than 0.7, except for some alternative measures of BGD and waste. We, therefore, do not use these variables in the same model; thus, a high correlation among these variables does not affect our results. To further explore this issue, we calculate the VIFs for all variables. The unreported VIF value for all the variables is within acceptable limits, thus confirming that multicollinearity is not an issue in our study.

Table 2
Female directors and waste management. Note: This table presents regression results for the relationship between female directors and waste management (i.e., waste generated and recycled) as well as the sub-components of waste, i.e., hazardous and non-hazardous waste. T-statistics are given in parenthesis. *, ** and, *** represent significance at the 0.1, 0.05 and 0.01 levels, respectively. All the variables are defined in Appendix Table A1.

BGD and waste management
We start our analysis by analyzing the effect of BGD, measured as the proportion of female directors on the board (F_PRO) and on the level of waste generation (WASTE) and waste recycling (R_WASTE). Table 2 illustrates the results of the OLS regressions by estimating model 1. Columns 1 and 2 present the results using waste generation (WASTE) and waste recycled (R_WASTE) as dependent variables, respectively. As expected, the coefficient on female directors (F_PRO) is negative and significant at the 1% level under Column 1, and positive and significant (at the 1% level) under Column 2, suggesting that female directors not only reduce the level of waste generation but also increase waste recycling. Specifically, a one percentage point increase in the proportion of female directors on the board is associated with a 1.365 (0.132) percentage point decrease (increase) in the level of waste generated (recycled). The economic significance of female directors on the level of waste generation and recycling is also important. For example, an increase in F_PRO by one (sample) standard deviation (e.g., using Table 1)  Additionally, we check the robustness of our main finding by re-estimating model 1 using the individual components of waste Table 3 Females critical mass, independent vs executive female directors and waste management. Note: This table presents regression results for the relationship between female directors and waste management, considering the critical mass and status (i.e., executive or non-executive director) of female directors within the board. T-statistics are given in parenthesis. *, ** and, *** represent significance at the 0.1, 0.05 and 0.01 levels, respectively. All the variables are defined in Appendix Table A1.
generation, namely, hazardous waste (H_WASTE) and non-hazardous waste (NH_WASTE). The results of this analysis, reported under Columns 3 and 4 of Table 2, show that the coefficient on female directors (F_PRO) is still negative and statistically significant at the 5% and 1% levels, respectively. In addition to female directors, R&D, TQ, and CASH have a significantly negative relationship with waste generation. In contrast, B_IND, CSR_COM, and CSR_AUDIT have a positive relationship with waste generation. Overall, these results lend strong support to H1 by providing empirical evidence that females are more sensitive to environmental issues (i.e., waste management) and firms with female directors are more likely to implement environmentally friendly policies Nadeem et al., 2020).

The critical mass of female directors and waste management
Next, we examine the effect of the critical mass of female directors, measured by dummy variables, indicating one female director (F1), two female directors (F2), and three or more female directors (F3), on the level of waste generation and recycling. Table 3 (Columns 1 and 2) presents the results of this analysis by estimating model 2. This analysis intends to examine whether the influence of female directors on the level of waste generation and recycling increases with an increase in their representation on the board.
The results reported under Column 1 of Table 3 show that F1, F2, and F3 are negatively and significantly (at the 1% level) associated with the level of waste generation (WASTE). However, the coefficient on F3 (− 0.354) is larger than the coefficient on F2 (− 0.310), while the coefficient on F2 (− 0.310) is larger than the coefficient on F1 (− 0.258), thus suggesting that the magnitude of the negative relationship between BGD and the level of waste generation increases with an increase in the number of female directors on the board. We then check the difference in the coefficients by performing the Wald test: the unreported results reveal that the coefficients on F1, F2, and F3 are significantly different from each other.
Column 2 of Table 3 presents the results of equivalent regression for the level of waste recycled (R_WASTE). As predicted, one female director (F1) has a positive but statistically insignificant impact on the level of waste recycled. However, the positive relationship is statistically significant at the 5% and 1% levels with two (F2) and three or more (F3) female directors on the board, respectively. Consistent with prior studies Torchia et al., 2011), these findings lend support to Kristie's (2011, p. 22) review of critical mass theory by showing that "one is token, two is presence, and three is a voice". Specifically, we find the insignificant coefficient on F1 consistent with tokenism, while the coefficient on F2 (0.026) in Column 2 is significant but smaller than the coefficient (0.035) on F3, which is consistent with the level of presence and voice (Kristie, 2013). We also perform differences in the coefficients test, and the unreported results of the Wald test confirm that the coefficients on F1, F2, and F3 are significantly different from each other. Taken together, these results support H2.

Independent and executive female directors and waste management
So far, our findings show that the presence of female directors is negatively (positively) associated with the level of waste generation (recycling). Another important question that needs to be addressed is whether all-female directors behave similarly. In other words, we try to investigate the channel through which female directors influence the level of waste generation and recycling. To do so, we follow existing literature and explore the monitoring and executive power channel (i.e., independent vs executive female directors). Independent female directors are expected to impact strategic decisions, such as waste management activities, via the monitoring channel because of their independent status and advisory role, while executive female directors might affect strategic decisions through their involvement in ongoing managerial issues and policy implementation (Liu et al., 2014;Chen et al., 2017;Atif et al., 2019). We report the results of this analysis under Columns 3 and 4 of Table 3 by estimating model 3: independent female directors (FNED_PRO) have a significantly negative (positive) impact on the level of waste generation (recycling). However, executive female directors (FED_PRO) have neither any significant impact on the level of waste generation (WASTE) nor on the level of waste recycling (R_WASTE). As expected, the impact of BGD on the level of waste generation and recycling is driven by independent female directors supporting H3. These findings are consistent with extant literature Chen et al., 2017).

Robustness tests
In this section, we re-examine the main findings using several robustness tests, including alternative measures of female directors; alternative measures of waste; one-year lagged board variables; excluding board-level variables; excluding dominating countries from the sample; subsample analysis based on corporate governance quality and the environmental orientation of the firms; and dividing the firm-years for the G10 and rest of the sample countries.
First, to test whether our results are sensitive to the choice of BGD measures, we conduct this check using the number (F_NUM) and presence of female directors (F_DUMMY) on the board (Panel A). Second, to test whether our results are sensitive to the measures of waste, we use the ratio of waste to total assets (WASTE/ASSETS) and the ratio of waste to sales (WASTE/SALES) as alternative measures of waste generation (Panel B). Third, we employ one-year lagged board variables to check whether our results are stable (Panel C), as one may argue that the presence of female directors may take time to influence corporate policies. Fourth, our main findings may be driven by the board-level variables: the board of directors is the main decision-making body within the organization and might play an active role in the implementation of waste management activities. We, therefore, re-estimate model 1 and exclude all board-level variables to alleviate this concern (Panel D). Fifth, we check whether our results are driven by the countries with a higher number of observations in the sample. To conduct this check, we re-estimate model 1 after excluding firms from the UK and US (Panel E), given the highest firm years from these countries. Sixth, we check whether our results are contingent on firm-level governance quality. To conduct this check, we re-estimate our model by splitting the sample into strong (High-CG) and weak governance (Low-CG) firms based on the industry-year average of the corporate governance score by country from the Thomson Reuters database (Panel F). Seventh, we check whether our results are contingent on the firms' level of environmental orientation, as some firms may be more environmentally friendly than others. To conduct this check, we re-estimate our model by dividing the sample into high (High-ENV) and low environmental orientation (Low-ENV) based on the industry-year average of the environmental performance score by country from the Thomson Reuters database (Panel G). Finally, we split our sample based on G10 and non-G10 sample countries, as one may argue that firms in advanced economies are more environmentally friendly than those in other economies. Moreover, the large number of firm-years from G10 countries may raise concerns. To address such issues, we re-estimate model 1 based on the split sample and present the results in Panel H. Table 4 reports the regression results for these sensitivity tests, including the control variables, industry, year, and country effects. In line with our main results, we find that female directors reduce (increase) the level of waste generation (recycling) across Panels A to H. Additionally, the results of the subsample analysis (Panels F and G) highlight that the relationship between female directors and the level of waste generated and recycled is relatively more visible in firms with robust governance mechanisms and higher environmental orientation.

Identification strategies
We acknowledge that our main findings might be subject to endogeneity concerns due to self-selection bias and reverse causality. In line with extant studies (Gull et al., 2018;Hossain, Atif, Ahmed, & Mia, 2020;Nadarajah, Atif, & Gull, 2021), we use four identification strategies; namely, PSM, the Heckman selection model, 2SLS, and DiD to alleviate endogeneity concerns.
Self-selection bias refers to the possibility that effective waste management activities may be in place because of some firm-specific factors (i.e., control variables) rather than BGD. We, therefore, follow recent literature (Alam, Atif, Chien-Chi, & Soytaş, 2019;Nekhili et al., 2022) and apply the PSM technique in two steps to control for firm-level factors that may affect the implementation of waste management activities. First, we create a treatment dummy variable (F_DUMMY), which takes the value of 1 if the firm has at least one female on the board and 0 otherwise. Based on F_DUMMY, we define firm-years with and without female directors as the treatment and control groups, respectively. We then proceed to estimate the probability that a firm has female directors. We run a probit regression to explain F_DUMMY with the same explanatory variables used in model 1, including industry, year, and country effects. Panel A (Column 1) of Table 5 reports the results for the probit regression. The pseudo-R-square for the regression is high (0.468), and most of the independent variables are (highly) significant.
We use the matching procedure without replacement and set the caliper distance at the 1% level to ensure that firms in the treatment and control groups are sufficiently identical. We then estimate the predicted value of having a gender-diverse board by running a probit regression for the dummy variable (F_DUMMY) with the control variables specified in model 1, including industry, year, and country effects. These estimates are the propensity scores for each firm-year observation. In the second step, based on the propensity scores, we form two similar subsamples based on different criteria, such as the treatment and control groups. Finally, we have a matched sample based on which we can examine the association between BGD and the level of waste generation and recycling, respectively.
To verify that the firm-year observations in the treatment and control groups are indistinguishable in terms of observable characteristics, we conduct two diagnostic tests following Chen et al. (2017) and Atif et al. (2019). The first test consists of re-estimating the probit regression for the post-match sample. The results (Column 2 in Table 5) suggest that none of the coefficients on the explanatory variables is statistically significant, indicating that there are no significant differences between the two groups. Moreover, the coefficients in Column 2 are typically smaller in magnitude than those in Column 1, indicating the decline in the degrees of freedom in the restricted sample. The pseudo-R-square declines from 0.468 to 0.017. This suggests that PSM removes all observable differences other than the difference in the presence of female directors. The second test examines the differences in the mean of each observable characteristic between the treatment and control firm-year observations both in the pre-match and post-match samples. Panel B of Table 5 shows significant differences in the pre-match sample; however, none of the differences in the observable characteristics between the treatment and control groups is statistically significant in the post-match sample. Overall, the two diagnostic tests suggest that PSM removes all the observable differences in the explanatory variables other than those relating to BGD. Panel C reports the results of the PSM estimator, which are also aligned with our main findings. Finally, we rerun the regression based on the matched sample of firm-year observations (Columns 3 and 4, Panel A in Table 5). The coefficient on WASTE (R_WASTE) is significantly negative (positive), suggesting that female directors have a strong impact on waste management (i.e., the level of waste generation and recycling).
As discussed earlier (in subsection 3.1), our sample is restricted to firms included in the Thomson Reuters Asset4 database who report waste data (i.e., waste generated and recycled). All firms do not report waste data due to the voluntary nature of such disclosure, suggesting that firms who report waste data may differ from those who do not. For instance, firms that disclose waste data are likely to care more about stakeholder concerns and may also have gender-diverse boards to seek legitimacy from stakeholders. Alternatively, qualified female directors, due to their limited pool, may deliberately choose to sit on the boards of firms that are more sensitive to stakeholder concerns, such as waste management. This suggests that our findings may suffer from bias.
Consistent with García Lara et al. (2017), Nadeem et al. (2020), and Gull, Atif, Issa, Usman, and Siddique (2021), we use Heckman's two-step selection model to address this concern. In the first step, we estimate the likelihood of having a gender-diverse board using F_DUMMY as the dependent variable and include the female to male directors' ratio by headquarter city as an additional control variable, which is expected to influence the tendency of firms to have gender-diverse boards (Usman, Gull, Zalata, Wang, & Yin, 2022) but not necessarily affects the level of waste generation or recycling (our dependent variables); therefore, we do not add it in the second-stage model. We also control for variables, including year, industry, and country effects, in the first step. The results of the first step probit regression are reported under Column 1 of Table 6. We then calculate the Inverse Mills Ratio (MILLS) and re-estimate model  Gull et al. 1 in the second step using MILLS as an additional control variable to correct for sample selection bias. 3 We report these results under Columns 2 and 3 of Table 6, which are consistent with the results reported in Table 2. Hence, this confirms our main findings are not subject to sample-selection bias. Our findings provide evidence that female directors affect waste management. However, firms with poor waste management activities (i.e., more waste generation and less recycling) in place may wish to hire female directors on their boards to seek legitimacy from stakeholders. This suggests that the presence of female directors on the board affects waste management activities or vice versa, causing reverse causality. We corroborate our findings using 2SLS. Following the spirit of extant studies (Nadeem et al., 2020;Usman et al., 2022;Zalata, Ntim, Choudhry, Hassanein, & Elzahar, 2019), we use two IVs, i.e., the industry average of female directors (FD_IA) and the female to male directors' ratio by headquarter city (FM_RATIO). Chen et al. (2017) and  have used female to male workforce participation as an instrument. However, we follow Usman et al. (2022) and use the female to male directors' ratio by headquarters city. The rationale behind the use of this IV is that it is highly unlikely that firms will appoint female directors from the general workforce because all female workers are not qualified to be appointed to the board of directors. However, firms are more likely to appoint female directors from the pool of available female directors. We, therefore, use the female to male directors' ratio over the ratio of female to the male workforce participation rate, as it is more relevant IV. Additionally, both IVs (FD_IA & FM_RATIO) are likely to meet the exclusion criterion by (not) being correlated with the (dependent variables, i.e., WASTE & R_WASTE) female directors on the board (F_PRO).
The results of the first-stage estimation are reported under Column 1 of Table 7, where the proportion of female directors on the board (F_PRO) is the dependent variable. Column 1 shows that the coefficients on both IVs (FD_IA & FM_RATIO) are positive and statistically significant at the 1% level. The value of the F-statistic is also high. The Anderson canon. corr. LM-statistic is statistically Note: This table presents regression results for the relationship between female directors and waste management, using alternative measures of female directors, alternative measures of waste, excluding board characteristics, excluding firm-years from countries dominating the sample, firmlevel corporate governance quality and environmental orientation, and firms operating in G10 (G10 countries include Belgium, Canada, France, Germany, Italy, Japan, the Netherlands, Sweden, Switzerland, the United Kingdom, and the United States). Control variables are included in the regressions; however, for purposes of brevity, results of the main independent variables are provided. T-statistics are given in parenthesis. *, ** and, *** represent significance at the 0.1, 0.05 and 0.01 levels, respectively. All the variables are defined in Appendix Table A1.  Note: Panel A presents regression results for the relationship between female directors and waste management using Propensity score matching (PSM). Panels B and C present the mean differences in firm-level (i.e., control variables) variables and estimators (i.e., main dependent variables) for the treatment and control groups based on the pre-and post-match samples. T-statistics are given in parenthesis. *, ** and, *** represent significance at the 0.1, 0.05 and 0.01 levels, respectively. All the variables are defined in Appendix Table A1.

Table 6
Two-step Heckman model. Note: This table presents regression results for the relationship between female directors and waste management using the Two-step Heckman model. FM_RATIO is the average female to male directors' ratio by headquarters city used as an instrument in the probit regression of two-step Heckman. T-statistics are given in parenthesis. *, ** and, *** represent significance at the 0.1, 0.05 and 0.01 levels, respectively. All the variables are defined in Appendix Table A1.
significant (under-identification test), and the Cragg-Donald Wald F-statistic is also higher than the Stock-Yogo weak ID test critical value at the 10% IV size. All the diagnostic tests suggest that our IVs are valid. Columns 2 and 3 of Table 7 report the results of the second-stage estimates using the predicted proportion of female directors on the board from the first-stage regression to estimate the level of waste generation and recycling. These results also validate a negative (positive) and statistically significant relationship between female directors and the level of waste generation (recycling). Therefore, we can reliably infer that female directors not only reduce the level of waste generation but also increase waste recycling. Finally, we test the influence of BGD on waste management around an exogenous event, the introduction of board gender balance reforms between 2002 and 2017, 4 to address the potential issue of endogeneity. 5 We employ a DiD analysis around the board gender Note: This table presents regression results for the relationship between female directors and waste management using two-stage least squares (2SLS). FD_IA (the industry average of female directors) and FM_RATIO (the average of female to male directors by headquarter city) are used as instruments in the 1st stage. T-statistics are given in parenthesis. *, ** and, *** represent significance at the 0.1, 0.05 and 0.01 levels, respectively. All the variables are defined in Appendix Table A1.
balance reforms during our sample period. We define our treatment group based on firms that are subject to board gender balance reforms (i.e., 3561 firm-year observations), while the control group consists of firms not subject to such reforms (i.e., 4804 firm-year observations). The control firms are matched using the same procedures as noted in PSM. Table 8 presents the results of the DiD analysis. We start with a univariate analysis to ensure that the treatment and control groups are comparable in terms of observable characteristics, except regarding their association with female directors. Panel A reports the results of mean differences between the treatment and control groups' characteristics and shows that both groups are identical. We follow Zaman, Atawnah, Baghdadi, and Liu (2021) and Atif, Liu, and Nadarajah (2022) to run the DiD estimator and evaluate the impact of board gender balance reforms on waste management. In Panels B and C, we calculate the change in waste generation (WASTE) and recycling (R_WASTE) from the pre-event period (i.e., the three years before the introduction of board gender balance reforms) to the post-event period (i.e., the three years after the introduction of board gender balance reforms) for both the treatment and control groups. The average change during the three-year period in WASTE (R_WASTE) for the treatment group is − 0.375 (0.020), while the change for the control group is 0.006 (− 0.041). The DiD estimator for WASTE (R_WASTE) is − 0.381 (0.061) and significant at the 10% (1%) level. This confirms our main findings.

Moderating role of national culture
In line with critical mass theory, extant studies Jia & Zhang, 2013;Owen & Temesvary, 2018;Torchia et al., 2011) suggest that an inclusive and gender-balanced board is required to garner the real benefits of BGD. By performing a meta-analysis of 140 articles published on the nexus of female directors and firm performance, Post and Byron (2015) show a negative association between female directors and firm performance in countries with less gender equality in terms of human capital and vice versa, suggesting that women find it difficult to work in male-dominated societies. Inspired by these conjectures, we examine whether the association between female directors and waste management (i.e., the level of waste generation and recycling) is also contingent on male dominance in society (i.e., masculinity vs femininitya dimension of national culture). To perform this analysis, we mainly rely on Hofstede's cultural index of the masculinity and femininity dimension due to its direct relevance to our research. 6 However, we control other dimensions of Hofstede's cultural index, such as power distance (PD), individualism (IND), and uncertainty avoidance (UNCERT), in our empirical model. To model the relationship between female directors (F_PRO), masculinity (MAS), and waste management (WASTE & R_WASTE), we use the following model.
where the dependent variable is waste management, measured by WASTE and R_WASTE. The independent and control variables are similar to those in model 1 except for the national culture dimensions; namely, masculinity (MAS), power distance (PD), individualism (IND), uncertainty avoidance (UNCERT), and the interaction between female directors and masculinity (F_PRO X MAS), which is our main variable of interest. We estimate the model using OLS regressions and report the results under Columns 1 and 2 of Table 9. The results show that the coefficients on the interaction term for waste management (i.e., WASTE & R_WASTE) are statistically insignificant, suggesting poor waste management by female directors in countries high on masculinity. This confirms that the relationship between BGD and waste management is moderated by the masculinity dimension of national culture.

Female directors, waste management, and firm performance
In this section, we examine whether waste management and female directors affect firm performance, measured by Tobin's Q (TQ). Given the positive effect of female directors on waste management, we expect that there might be a spillover effect on firm performance. To model the relationship between female directors (F_PRO), waste management (WASTE & R_WASTE), and firm performance (TQ), we use the following model.
where the dependent variable is firm performance, measured by Tobin's Q (TQ). The independent and control variables are similar to those in model 1, except for the interaction terms between female directors and waste management (F_PRO X WASTE & F_PRO X R_WASTE), which are our main variables of interest. We estimate model 5 using the OLS regressions and report results under Columns 1 and 2 of Table 10. The coefficients on the interaction terms under both columns are positive and statistically significant at the 10% level, suggesting better firm performance because of the effective waste management activities implemented by female directors. Collectively, our results signal a positive impact on the firm performance. 7

Female directors, sustainable compensation, and waste management
Finally, prior literature (e.g., Haque, 2017) suggests that board characteristics, including gender diversity, focus on carbon emission reduction initiatives to improve the environment and environment, social, and governance (ESG) based compensation policies are also considered favorable in this regard. Likewise, we examine whether BGD influences environmentally friendly practices through the design of executive compensation contracts. To do so, we first run a probit regression using SUS_COMP (a dummy variable set equal to 1 if the company has a sustainable compensation policy for executives and 0 otherwise) as the dependent variable and F_PRO (the proportion of female directors) as the independent variable, along with all the control variables used in the main analysis. The results of this analysis, reported under Column 1 of Table 11, show that the coefficient on F_PRO is positive and statistically significant at the 1% level, thus suggesting that firms with female directors are more likely to consider sustainability issues while designing executive compensation contracts. Next, we examine whether BGD influences waste management through the design of executive compensation. To do so, we create a new variable, F_PRO X SUS_COMP, which is the interaction term between female directors and the likelihood of having a sustainable compensation policy for the executives. We then include F_PRO X SUS_COMP and SUS_COMP, along with F_PRO and control variables, and re-estimate model 1. The results of this analysis, reported under Columns 2 and 3 of Table 11, demonstrate that the coefficient on F_PRO X SUS_COMP is significantly negative (positive) for waste generation Note: This table presents the difference-in-differences (DiD) regression results of waste management around the introduction of board gender balance reforms during the sample period (2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017). The treatment group consists of firms that are subject to the board gender balance reform, and control firms are those which are not subject to such reform. Panel A shows the post-matching mean differences for treatment and control firms' characteristics. Panels B and C present the DiD estimators for waste generation and recycling, respectively. T-statistics are given in parenthesis. *, ** and, *** represent significance at the 0.1, 0.05 and 0.01 levels, respectively. All the variables are defined in Appendix Table A1.
(recycling), suggesting that female directors are likely to promote environmentally friendly initiatives by aligning executive compensation with sustainable business practices.

Conclusion
Climate change is a matter of concern for the entire world. Enormous waste generation due to rapid industrialization is exacerbating the situation. Extant literature (e.g., Shaukat et al., 2016) looking at the impact of BGD on corporate environmentally friendly initiatives has largely ignored the consequences of the waste management dimension towards sustainable growth. In addition, national culture and sustainable compensation policies for executives are relevant decision-making factors related to sustainability (e.g., Haque, 2017). The recent literature confirms that firms may focus on a single dimension of sustainability at the expense of other dimensions, leading toward policy-practice decoupling (Sauerwald & Su, 2019;Tashman et al., 2019). Our study, therefore, fills the void in corporate governance and sustainability accounting literature by providing robust empirical evidence on how and through which channels BGD affects waste management activities in different countries. Note: This table presents regression results for the relationship between female directors and waste management after controlling for the moderating effect of the masculinity dimension of national culture. T-statistics are given in parenthesis. *, ** and, *** represent significance at the 0.1, 0.05 and 0.01 levels, respectively. All the variables are defined in Appendix Table A1.
This study, employing a panel of 8365 firm-year observations from 37 countries between 2002 and 2017, finds a negative (positive) impact of BGD on waste generation (waste recycling), and this effect is mainly observed if there are two or more female directors on the board, consistent with critical mass theory. The impact is primarily driven by independent female directors rather than executive directors. Moreover, we document that the relationship is moderated by a national culture dimension, i.e., masculinity vs femininity, and observe the positive impact of sustainable compensation policies on waste management. We find improved financial performance for firms engaged in waste management activities, and this finding contributes to the open question, "Does it pay to be green?" Our results are robust to the use of alternative econometric specifications, alternative measures of BGD and waste management, the exclusion of dominating countries in terms of firm years, exclusion of board characteristics, and subsample analysis. We also employ several identification strategies, including PSM, Heckman's selection model, 2SLS, and DiD and document consistent results. Overall, our analysis indicates that female directors lead firms towards better waste management policies to outperform their counterparts. This makes waste management activities a pathway for creating sustainable business models.
Our study offers essential guidelines for policy formulation and academic research. Given the global drive in fostering sustainable development goals, our study provides important empirical evidence to policymakers, which can inform their decision-making on developing environmentally friendly initiatives (i.e., waste management activities) by promoting gender diversity on corporate boards. Future research benefitting from extended disclosure may consider the association between BGD and various sources of corporate controllable and non-controllable waste. Such a study may also be able to distinguish the trade-off between waste generation and firms' production of goods and services. Future studies may also consider other research methods (i.e., interviews or Note: This table presents regression results for the female directors, waste management and financial performance nexus. T-statistics are given in parenthesis. *, ** and, *** represent significance at the 0.1, 0.05 and 0.01 levels, respectively. All the variables are defined in Appendix Table A1. questionnaires) to offer direct evidence on the mechanisms that female directors employ to shape waste management activities. This would substantially advance research on gender diversity and corporate sustainability.

Declaration of competing interest
The authors, Gull, Atif, and Hussain, declare that they have no conflict of interest. Note: This table presents regression results for the relationship between female directors and waste management after controlling for the moderating effect of a sustainable compensation policy. SUS_COMP is a dummy variable set equal to 1 if the company has a sustainable compensation policy for executives and 0 otherwise. T-statistics are given in parenthesis. *, ** and, *** represent significance at the 0.1, 0.05 and 0.01 levels, respectively. All the variables are defined in Appendix Table A1.

Table A2
Sample distribution by country and year Note: This table presents regression results for the relationship between female directors and waste management (i.e., waste generated and recycled) using subsamples of profitable and loss-making firms. T-statistics are given in parenthesis. *, ** and, *** represent significance at the 0.1, 0.05 and 0.01 levels, respectively. All the variables are defined in Appendix Table A1.