The role of gender diversity on corporate boards and audit committees in shaping tax avoidance behaviour

Corporate tax avoidance is commonly discussed across political, academic and public spheres (Kovermann and Velte, 2019), largely due to media coverage of high-profile tax scandals (i.e., the Paradise Papers, Panama Papers and Luxembourg Leaks) revealed in the aftermath of the subprime mortgage crisis (Ruggiero, 2022). It is well-documented that large corporations around the world employ aggressive tax avoidance tactics to minimise their tax liabilities (Ruggiero, 2022). Although there is no universally accepted definition, tax avoidance is broadly understood as any corporate transaction that decreases a company’s tax obligation relative to its pre-tax income (Dyreng et al., 2008). In practice, these activities encompass a wide spectrum of behaviours, from legitimate tax planning to unlawful tax evasion (Li et al., 2020; Li et al., 2025).

A growing number of studies (e.g., Armstrong et al., 2015; Hasan et al., 2024) argue that the effectiveness of corporate governance mechanisms may determine the extent to which managers engage in tax avoidance. Among these mechanisms, board gender diversity has attracted increasing scholarly attention for its potential to enhance board oversight and influence tax-related decision-making (Jarboui et al., 2020; Richardson et al., 2016; Riguen et al., 2020; Hossain et al., 2025b). Such interest is also fuelled by global initiatives such as the UN’s Sustainable Development Goal 5 (Gender Equality) and regulatory measures promoting female board participation. The degree of gender representation on corporate boards, however, varies significantly across countries. Norway introduced a 40% board gender quota in 2003, followed by several European countries, including France, Germany, Italy and The Netherlands, with targets between 30 and 40% (Biswas et al., 2022). In South Asia, India and Pakistan legally mandate the presence of at least one female director on public company boards, whereas Bangladesh Corporate Governance Code (2018) promotes board diversity without enforcing a minimum gender quota (Mazumder, 2025). ^[1] Research suggests that female directors tend to uphold stricter ethical standards (Li et al., 2020; Fourati et al., 2025), exhibit greater risk aversion (Ali et al., 2024), constrain managerial opportunistic behaviour (Ghafoor et al., 2021), demonstrate more independent thinking (Carter et al., 2003), and generally adopt more conservative and rigorous decision-making approaches (Richardson et al., 2016). Some studies contend that these benefits may only be materialised when female representation reaches a “critical mass” (Mazumder, 2025; Torchia et al., 2011). ^[2] Building on this perspective, a number of recent studies (e.g., Garcia-Blandon et al., 2022; Jarboui et al., 2020; Rakia et al., 2024; Riguen et al., 2020; Baatwah et al., 2024; Hossain et al., 2025a) have investigated the association between board gender diversity and tax avoidance, yet the findings remain mixed and inconclusive, particularly in developing economies where institutional contexts differ substantially from those in developed markets.

Against this backdrop, Bangladesh presents a particularly relevant context, given the absence of binding board gender diversity requirements, the relatively low representation of women on boards (Mazumder, 2024; Das and Hossain, 2025), and persistent challenges in tax compliance (The Daily Star, 2023). Examining this issue in Bangladesh offers an opportunity to extend prior evidence by providing insights into how board gender diversity relates to tax avoidance behaviour in a context characterised by weak governance mechanisms and ongoing compliance challenges. In particular, this study complements the existing literature by exploring whether female participation on the audit committee moderates the association between board gender diversity and tax avoidance in the financial sector of Bangladesh.

In developing countries like Bangladesh, tax revenue is pivotal not only for financing social welfare programmes but also for supporting national priorities, including investment in education, healthcare and infrastructure. However, corporate tax avoidance undermines these efforts by restricting government expenditure on critical services (Hossain et al., 2024; Principles of Responsible Investment, 2020). Prior studies indicate that adverse effects of tax avoidance tend to disproportionately affect developing countries, largely due to weak regulatory frameworks and inadequate oversight (Johannesen et al., 2020). Despite having the highest average tax rate in South Asia, Bangladesh loses 30.2% of its Gross Domestic Product (GDP) due to the shadow economy, which is more than 2.6 times its healthcare expenditure, over 100% of its education and technology spending, and more than 300% of its social safety and welfare budget for the year 2021 (Moazzem et al., 2023). According to a report by the Tax Justice Network published in 2023, corporate tax losses through profit shifting and individual tax evasion in Bangladesh amount to $387m every year, which is equivalent to 0.1% of GDP, 1.5% of tax revenue, nearly 33.33% of the health budget, and 6.19% of education spending (The Daily Star, 2023). These significant fiscal losses underscore the significance of understanding the governance factors that may influence corporate tax behaviour. However, there is limited research on corporate tax avoidance and its governance within the Bangladeshi context (e.g., Hassan et al., 2022; Hossain et al., 2025a; Rashid et al., 2024). This study therefore investigates both the direct association between board gender diversity and tax avoidance and the potential moderating role of female participation on the audit committee. Accordingly, this study addresses the following research questions (RQs):

This study focuses on the financial sector in Bangladesh. Unlike non-financial firms, banks and other financial institutions in Bangladesh operate under the direct oversight of the central bank, which imposes strict requirements for financial transparency, corporate governance and regulatory compliance (Mazumder, 2024; Mazumder and Hossain, 2025; Siddiqui, 2010). Given the weak and volatile nature of capital markets in developing economies, financial sectors play a pivotal role in mobilising capital, facilitating financial flows, and supporting economic growth (Rashid et al., 2024). However, the financial sector in Bangladesh continues to face persistent governance challenges and repeated scandals involving non-performing loans (Yesmine et al., 2023). This governance weakness may also create conditions conducive to tax avoidance. This risk is further amplified by the complexity of bank reporting and sophisticated tax strategies embedded in their operations (Folorunso and Lokanan, 2023). Prior empirical studies have shown that board gender diversity enhances governance effectiveness and strengthens the monitoring of management. For example, Mazumder (2024, 2025) find that board gender diversity can play an effective role in minimising information asymmetry through voluntary disclosure. Similarly, Das and Hossain (2025) find that female directorship strengthens the CEO pay-performance link. Therefore, our research underscores an underexamined area in the financial sector of Bangladesh by investigating how women’s dual role in governance, serving both as board directors and committee members, collectively influence tax avoidance.

Based on a sample of 341 firm-year observations from 2016 to 2024, our analysis reveals a negative association between board gender diversity and tax avoidance, suggesting that gender-diverse boards are more attuned to ethical considerations and reputation risks linked to aggressive tax strategies. Further analysis shows that the participation of female directors on the audit committee strengthens the above association, indicating that female representation on the audit committee further encourages transparency and credibility in financial reports, thereby curbing tax avoidance behaviour. Notably, the association between board gender diversity and tax avoidance is positive and significant when there is no or only one female director on the board but turns negative and significant when at least two female directors are present. Consistent with this pattern, the interaction between the presence of at least two female directors on the board and female participation on the audit committee is also negative and significant. Taken together, these findings suggest that once women reach a critical mass, their meaningful participation can help curb tax-avoidance behaviour, consistent with critical mass theory.

We also address several potential concerns. Firstly, we use a lagged model to address potential reverse causality, as firms engaging in aggressive tax avoidance may subsequently appoint women to their boards to enhance their reputation or comply with governance reforms. Secondly, we apply the Heckman two-step selection model to account for potential self-selection bias, as firms that may choose to appoint women to their boards are not random; rather, they often exhibit specific characteristics such as firm size, industry practices and profitability. Finally, we apply the entropy balancing method to address observable heterogeneity among firms. Specifically, firms are classified into treatment and control groups according to female representation on the board and audit committee. Our findings are also robust to various additional tests, including alternative measures of board gender diversity and tax avoidance.

In addition, following prior evidence that female directors limit managerial opportunism (e.g., Hasan et al., 2022; Attia et al., 2024; Lai et al., 2017; Fourati et al., 2025; Ali et al., 2024; Ghafoor et al., 2021), we investigate the mechanism through which managerial opportunism mediates the relationship between board gender diversity and tax avoidance by testing whether female directors reduce earnings management and thereby indirectly constrain tax avoidance. ^[3] We find that earnings management partially mediates this relationship, highlighting its role in linking gender diversity and tax-avoidance behaviour.

This study makes several contributions to the literature while addressing research gaps. Firstly, prior research on tax avoidance yields mixed findings regarding board gender diversity. While some works report a negative association between board gender diversity and tax avoidance (Ali et al., 2024; Baatwah et al., 2024; Hasan et al., 2024; Jarboui et al., 2020), others find no significant relationship (Hossain et al., 2025a; Koay and Sapiei, 2025). These inconsistent findings suggest that the link between board gender diversity and governance effectiveness remains unsettled, highlighting the need to better understand the conditions under which gender-diverse boards influence ethical corporate decisions such as tax avoidance. In this regard, our study contributes by examining whether female directors’ participation on audit committees strengthens or weakens the association between board gender diversity and tax avoidance. Secondly, this study contributes to the theoretical literature by integrating insights from social role theory and social identity theory. Specifically, it connects how the female directors’ communal traits and their shared group identity can collectively enhance corporate accountability and ethical compliance by constraining tax avoidance practices. Thirdly, prior research shows that board gender diversity is associated with lower levels of opportunistic earnings management (Hasan et al., 2022; Attia et al., 2024; Srinidhi et al., 2011) and a reduced likelihood of financial misconduct and fraud (Wahid, 2019; Wang et al., 2022). Building on this evidence, our study further contributes to the literature by examining whether managerial opportunism serves as an underlying mechanism linking board gender diversity to tax avoidance. In particular, we assess whether female directors curb earnings management and thereby indirectly constrain tax avoidance.

The remainder of the paper is organised as follows: Section 2 presents the theoretical framework and literature review, along with the development of hypotheses. Section 3 articulates the research design and methodology, while Section 4 discusses the outcomes of the study. Sections 5 and 6 focus on additional analysis and the issue of endogeneity, respectively. Section 7 presents the mechanism test, while Section 8 concludes the study by outlining policy implications, limitations and directions for future research.

Previous research has drawn on a range of theoretical frameworks to examine how board gender diversity influences corporate decision-making and organisational behaviour. Among these, agency theory and resource dependency theory have been particularly prominent (Boshanna, 2021; Boulouta, 2013; Pucheta‐Martínez et al., 2019). Agency theory suggests that an effective board composition enhances monitoring functions and mitigates agency conflicts by curbing managerial opportunism (Jensen and Meckling, 1976; Mazumder, 2024). In contrast, resource dependency theory underlines that organisations depend on external resources to survive and grow, and that the board of directors plays a pivotal role in securing and managing these resources, which in turn shape organisational behaviour and decision-making (Mazumder, 2025; Pfeffer and Salancik, 2003). Although both theories have been extensively applied in earlier studies to understand the role of gender diversity in shaping governance outcomes, Carter et al. (2010) argue that neither agency nor resource dependency theory can directly and sufficiently explain the association between board gender diversity and firm performance. These theories do not fully account for why women act more (or less) responsibly and ethically than their male counterparts, even when both genders occupy key oversight roles within the organisations. Male and female directors exhibit systematic differences in their personality traits, preferences and ways of thinking (Chen et al., 2016a). As Anglin et al. (2022, p. 1470) state, “individual behaviour can be understood and predicted if one knows the roles occupied and the corresponding behavioural expectation tied to that role”. Building on this perspective, our study draws on social role theory (Eagly, 1987) and social identity theory (Tajfel and Turner, 1986) alongside insights from previous empirical research to predict how board gender diversity might influence corporate tax avoidance behaviour. ^[4]

Social role theory (Eagly, 1987) argues that societal expectations and beliefs about gender roles significantly shape individuals’ behaviour in professional environments. In line with this theory, men and women are socialised into distinct roles. Women are often associated with communal qualities such as empathy, care, ethical sensitivity and a strong sense of responsibility (Boulouta, 2013; Eagly, 2009). On the other hand, men are recognised as possessing agentic traits such as aggressiveness, competitiveness and a higher risk tolerance (Ali et al., 2024; Eagly, 2009). It is argued that these socially constructed expectations influence the way individuals perceive and perform their responsibilities in organisational settings (Ali et al., 2024; Boulouta, 2013; Eagly, 2009; Yarram and Adapa, 2021). As such, female board members are more likely than their male counterparts to prioritise social responsibility (Ali et al., 2024; Ben-Amar et al., 2017), ethical considerations (Briano-Turrent, 2022; Fourati et al., 2025), risk aversion (Menicucci and Paolucci, 2022; Seebeck and Vetter, 2021) and transparency (Gul et al., 2011; Nadeem, 2022) in their corporate decisions. By focusing on ethical compliance (Alfraih, 2016; Isidro and Sobral, 2015) and prioritising the firm’s long-term reputation (Elzahar et al., 2022; Fourati et al., 2025), female board members may be more inclined towards curbing firms’ aggressive tax avoidance strategies.

In contrast, Social Identity Theory (Tajfel and Turner, 1986) shifts the focus to how the group identity of female directors shapes their behaviour and influences group dynamics within organisations. The theory is in line with the notion that collective action and behaviour cannot be fully understood solely through individual differences or personality traits (Chen et al., 2016b). Individuals often align their actions with group norms, particularly when they find themselves as part of an under-represented or minority group (Mathisen et al., 2013). In male-dominated boardrooms, female directors may perceive themselves as part of an out-group or minority. Such perceptions may create a shared identity and inspire them to act in ways that challenge established groupthink (Abbott et al., 2012; Mathisen et al., 2013; Wahid, 2019) and place greater emphasis on ethical standards and social responsibility (Post et al., 2011). As representatives of their gender, they might also feel motivated to champion policies or practices that improve corporate transparency and reduce perceived risks (Bear et al., 2010; Post et al., 2011). Hence, we can argue that female board members may collectively advocate for a more cautious approach to tax avoidance strategies, viewing aggressive tax avoidance as incompatible with their shared ethical principles and risk perceptions.

Previous studies show that board gender diversity is negatively associated with opportunistic earnings management (Attia et al., 2024; Srinidhi et al., 2011) and the likelihood of financial misconduct and fraud (Wahid, 2019; Wang et al., 2022). Moreover, Chen et al. (2016a) provide evidence that female managers are less inclined to defend immoral business practices such as bribery and tax evasion than their male counterparts. In tax avoidance and governance literature, Jarboui et al. (2020) and Riguen et al. (2020) present evidence of a negative relationship between board gender diversity and tax avoidance in UK firms. Similarly, Baatwah et al. (2024) and Salehi et al. (2024) report similar findings in the emerging market context of Oman and Iran, respectively. In contrast, Hossain et al. (2025a) and Koay and Sapiei (2025) find no significant relationship between board gender diversity and corporate tax avoidance in the contexts of emerging economies, specifically Bangladesh and Malaysia, respectively. Overall, empirical evidence of the association between board gender diversity and corporate tax avoidance remains limited and mixed.

Based on the above discussion, we propose the following hypothesis in its alternative form:

Audit committee plays a crucial role in overseeing financial reporting, ensuring compliance, and maintaining robust internal controls (Ghafran and O’Sullivan, 2013; Rashid et al., 2026). As Spira (1999) points out, it acts as a practical safeguard against unethical behaviours within business. However, how well the audit committee can perform these responsibilities mostly depends on its composition (Malik, 2014; Rashid et al., 2026).

In recent years, female representation on the audit committee has received increasing attention due to its potential to enhance effective and ethical oversight. For example, Thiruvadi and Huang (2011) find that the presence of a female member on the audit committee constrains earnings management, while Ghafoor et al. (2021) and Hasan et al. (2024) show that their presence on the audit committee hinders managers’ risk-taking and opportunistic behaviour. Similarly, Thiruvadi (2012) also finds that audit committees with at least one female director tend to convene more frequently than all-male committees, which leads to better diligence and oversight. Therefore, female directors on audit committees not only enhance audit quality (Ben‐Amar et al., 2013) but also improve the committee’s effectiveness in curbing tax avoidance (Dang and Nguyen, 2022; Hasan et al., 2024). Similarly, Ittonen et al. (2010) argue that female members on audit committees can reduce audit fees by enhancing internal controls and fostering financial integrity, thereby lowering perceived audit risk. In contrast, Peterson and Philpot (2007) argue that males are likely to serve on key committees (e.g., compensation, finance), while females are often on the less important public affairs committee, which Bilimoria and Piderit (1994) attribute to gender bias, viewing female directors as figureheads rather than active contributors.

Green and Homroy (2018) highlight that one avenue through which female directors can more effectively influence firm performance is by participating in key board committees, such as audit committee. They argue that female directors’ involvement in key committees may open opportunities for them to actively and directly shape the company’s core practices. Ararat and Yurtoglu (2021) find that female directors are associated with higher firm value, particularly when they take on more active roles in board governance through committee memberships. Similarly, Rashid et al. (2026) underscore that female directors on the audit committee tend to promote greater transparency in corporate social responsibility, which in turn strengthens corporate accountability and ethical compliance. Therefore, when women hold roles in both governance areas (i.e., board directorship and committee membership), their combined presence may strengthen and enhance their collective influence on corporate behaviour and performance. In line with social identity theory (Tajfel and Turner, 1986), such dual presence may also empower female directors to adopt a more unified and effective stance against opportunistic corporate behaviours. Building on this insight, we can argue that although board gender diversity sets the tone for governance, their influence on functional execution of board decisions and corporate practices such as tax avoidance is likely to be boosted by the female presence on the audit committee that oversees financial reporting, internal control and risk management.

Hence, we frame the second hypothesis as:

The study sample includes all firms in the financial sector, including banking and non-bank financial institutions listed on the Dhaka Stock Exchange (DSE) during the period 2016–2024. The sample period begins in 2016, which aligns with the Bangladesh government’s policy shift following the adoption of the Sustainable Development Goals (notably, SDG 5: Gender Equality) and the launch of the 7^th Five-Year Plan (2016–2020), which formally integrated women empowerment and participation in political and economic activities as national priorities (Mazumder and Hossain, 2025; Ministry of Women and Children Affairs, 2016). The year 2024 represents the most recent period for which published annual reports are available. At present, there are 36 banks and 23 non-bank financial institutions listed on the DSE. After dropping firms listed after 2021, missing control variables, and loss-making years, the final sample consists of 341 firm-year observations, including 29 banks and 17 non-bank financial institutions. Table 1 outlines the sample selection process. All data were manually sourced from annual reports, which are widely regarded as a primary source of financial and governance information.

Tax avoidance is difficult to quantify directly due to the confidential nature of tax filings. Therefore, prior research often derives proxy measures from financial statements (Desai and Dharmapala, 2006). One common proxy is the effective tax rate (ETR), which reflects the relative corporate tax burden (Lanis and Richardson, 2018; Rashid et al., 2024). Since a higher ETR indicates lower tax aggressiveness, the measure is multiplied by −1 so that higher (lower) values represent greater (lower) levels of tax avoidance (Rashid et al., 2024).

Consistent with earlier studies (Abbasi et al., 2020; Ali et al., 2024; Das and Hossain, 2025), we measure board gender diversity (FPBOD) as the proportion of female directors to the total number of directors on the board.

Female representation on the audit committee serves as the moderating variable in our study. This is measured as the proportion of female directors on the audit committee (FPAC) relative to the total number of directors on audit committee (Abbasi et al., 2020; Dang and Nguyen, 2022).

The study uses a comprehensive set of control variables to enhance its validity. We control for board size (BS) as larger boards can improve monitoring through diverse knowledge and skills (Salehi et al., 2024). Board independence (INDP) is expected to be associated with reduced tax avoidance because independent directors oversee management impartially and prioritise maintaining their reputation (Ali et al., 2024). The study also considers three ownership variables: directors’ and/or sponsors’ ownership (DO), foreign ownership (FO) and institutional ownership (IO), consistent with Mazumder (2024) and Das et al. (2025), as these variables may improve firms’ oversight and transparency. The size of audit committee (SAC) is controlled, as larger audit committees may be more effective at detecting aggressive tax avoidance practices (Salehi et al., 2024).

The study also includes other firm-specific control variables such as profitability (ROE), leverage (LEV), firm age (FAGE), the frequency of board meetings (LBM), audit quality (BIG4), the expenditure of corporate social responsibility (CSR) and firm size (FSIZE) based on prior research (e.g., Abbasi et al., 2020; Ali et al., 2024; Hossain et al., 2025a; Salehi et al., 2024; Dang and Nguyen, 2022). The measurement of all variables is detailed in  Appendix A.

To test the hypotheses framed in section 2, we apply the following regression models:

The first model examines the association between board gender diversity (FPBOD) and tax avoidance (ETR). The second model investigates the moderating role of female presence on the audit committee (FPAC) in this relationship. The study includes several control variables: BS (board size), INDP (board independence), DO (directors’ and/or sponsors’ ownership), IO (institutional ownership), FO (foreign ownership), SAC (audit committee size), ROE (profitability), LEV (leverage), LBM (board meetings), FAGE (firm age), FSIZE (firm size), BIG4 (audit quality) and CSR (CSR expenditure intensity).

We run ordinary least squares regression to test our hypotheses. To address potential heteroskedasticity in the data, we incorporate robust specifications in our regressions. In addition, we include year fixed effects to account for variations over time. To address self-selection bias, the Heckman two-step method is employed, while a lagged regression model mitigates potential reverse causality, and entropy balancing is used to reduce observable heterogeneity among firms. We also incorporate one-way clustering (firm-level) and two-way clustering (firm and year-level), using a cluster-robust covariance matrix to account for error correlations within firms and across years.

Table 2 presents the summary statistics for all variables included in the study. The mean values of tax avoidance in the financial sector, measured by BTD and ETR, are −0.002 and −0.497, respectively. Rashid et al. (2024) also document that BTD and ETR averages are −0.005 and −0.442, respectively, when considering only banking companies. The proportion of female participation (FPBOD) on the board is 0.139, meaning that about 14% of board members are women. This finding aligns with Mazumder (2024), who emphasises ongoing concerns regarding gender inequality in the corporate environment in Bangladesh. Regarding the presence of female board members, 48.7% have at least two women on the board, while female participation on the audit committees is 11.7%, indicating greater concern for gender inequality within audit committees in financial firms. All continuous variables are winsorised at the 5th and 95th percentiles to minimise the impact of outliers.

Table 3 presents the Pearson correlation coefficients, along with their corresponding significance levels. ETR is found to be positively and significantly associated with SAC and ROE, while exhibiting negative and statistically significant correlations with BS, IO, LEV, LBM, FAGE, FSIZE and CSR. None of the pairwise correlation coefficients among the explanatory variables exceed the threshold of 0.80, suggesting that multicollinearity is unlikely to compromise the validity of the subsequent regression estimates (Gujarati and Porter, 2009). Furthermore, the variance inflation factor (VIF) values (untabulated) range from 1.28 (BIG4) to 3.05 (FSIZE), indicating no issues with multicollinearity.

Table 4 presents the baseline regression result. The models exhibit satisfactory overall fit, as indicated by a significant F-statistic (p < 0.000). The adjusted R-squared values are 31.9%, 31.2% and 32.7%, for Columns 1–3, respectively, suggesting that the explanatory variables collectively explain a reasonable proportion of the variation in the dependent variable. Column 1 shows that FPBOD is negatively and significantly associated with ETR (β = −0.148, p < 0.05), thereby supporting H1. This finding suggests that higher female representation on corporate boards in the financial sector of Bangladesh is associated with lower levels of tax avoidance. One possible explanation is that female directors may place greater emphasis on ethical conduct and reputational considerations, thereby discouraging aggressive tax avoidance practices. This interpretation is consistent with the predictions of social role theory (Eagly, 1987), which suggests that women adopt more cautious and ethically oriented decision-making behaviours. The result is also consistent with prior empirical evidence. For example, Chen et al. (2019) and Hossain et al. (2025b) find that firms with gender-diverse boards tend to be more cautious regarding the reputational risks associated with aggressive tax avoidance. Similarly, Ali et al. (2024) find that female directors, driven by their risk-averse nature and effective monitoring of management, help to limit aggressive tax practices of managers. However, the evidence in the literature is not always consistent. In the Bangladeshi context, Hossain et al. (2025a) find no significant relationship between board gender diversity and tax avoidance. The authors argue that the presence of female representation on corporate boards may be largely symbolic or tokenistic rather than substantively influential in corporate decision-making.

In Column 2 of Table 4, we examine the direct effect of female representation on the audit committee, while Column 3 analyses the moderating role of female presence on the audit committee in the association between board gender diversity and tax avoidance. The direct effect of FPAC is negative and marginally significant (Column 2), while the interaction term (FPBOD × FPAC) (Column 3) is found to be negative and significant in relation to tax avoidance, thereby supporting H2. We would also like to note that, in a corporate governance context like Bangladesh, audit committee members are drawn from the board of directors. Accordingly, the standalone FPAC coefficient in Column 3 represents a theoretical case (where FPAC > 0 while FPBOD = 0). Therefore, the meaningful effect is captured by the negative interaction term (i.e., when both the board and the audit committee are gender-diverse, tax avoidance is lower).

Overall, the results indicate that the presence of females on the audit committee strengthens the negative association between board gender diversity and tax avoidance, reinforcing their prioritisation of responsible and credible corporate conduct as critical factors. Female directors in male-dominated boardrooms may perceive themselves as part of an out-group, leading them to challenge groupthink and emphasise ethical standards and social responsibility. As argued in social identity theory (Tajfel and Turner, 1986), individuals’ behaviour is shaped by their group identity. As a minority on boards and audit committees, female directors may be more likely to question dominant norms and promote more responsible decision-making. This, in turn, may result in advocating for more cautious tax avoidance strategies. These findings are consistent with prior studies (e.g., Hasan et al., 2024; Dang and Nguyen, 2022; Ittonen et al., 2010; Ghafoor et al., 2021), which suggest that female representation on the audit committee enhances corporate governance effectiveness by constraining tax avoidance, strengthening internal controls, and limiting managerial risk-taking and opportunistic behaviour.

Among the control variables, BS, DO, LEV, FAGE and CSR show a negative and significant association with ETR across all columns. IO is significantly negative in Column 2, while LBM is negative and significant in Column 1. Larger boards may curb tax avoidance through diverse expertise, while higher director ownership can promote tax compliance by protecting reputational interests (Salehi et al., 2024). Firms subject to higher monitoring by debt holders and institutional owners (Rashid et al., 2024; Salehi et al., 2024) and more frequent board meetings indicating higher director interaction are more likely to restrain tax aggressiveness (Ali et al., 2024). Older and socially responsible firms are less prone to tax avoidance due to their emphasis on ethical conduct and stakeholders’ trust (Rashid et al., 2024). In contrast, SAC and ROE are positively and significantly associated with ETR across all columns. Profitable firms may engage in tax avoidance to retain earnings for internal use, despite potential reduction of reported profits (Hossain et al., 2025a), while larger audit committees with diverse risk preferences may foster greater tax aggressiveness.

To assess the robustness of our key findings, we employ an alternative measure of tax avoidance, namely the Book-tax difference (BTD) which captures discrepancies between income tax based on accounting earnings before tax and taxable income required by tax law (Dang and Nguyen, 2022; Desai and Dharmapala, 2006; Rashid et al., 2024). To adjust for firm size, BTD is scaled as a proportion of the total assets of the firm. Table 5 shows the regression results using this measure, which is qualitatively similar to the baseline outcome reported in Table 4.

To further validate our initial findings, we employ two alternative measures of board gender diversity, namely, the Blau Index (BLAU) and the Shannon Index (SHANNON), in alignment with prior studies (Das and Hossain, 2025; Das et al., 2025) (see  Appendix for variable definitions). As shown in Table 6, the results remain qualitatively similar to the key findings reported in Table 4.

Critical Mass Theory (Kanter, 1977a, 1977b) postulates that female directors are unlikely to play a meaningful role in shaping corporate decision-making until their number reaches a certain threshold (i.e., critical mass). When women are significantly under-represented (i.e., tokenistic representation) in governance positions, their ability to influence key decisions becomes limited, and their voices often go unheard. However, when women reach a critical mass, they move beyond being seen as mere representatives of diversity and start to make a real impact on shaping policies and practices (Ben-Amar et al., 2017; Kanter, 1977b; Mazumder, 2025; Torchia et al., 2011; Hossain et al., 2025a; Das and Hossain, 2025; Das et al., 2025). Consistent with Mazumder (2025) and Usman et al. (2019), we measure the token representation of females (0/1WOMAN) using a dummy variable, assigning a value of “1” when there is at most one female director on the board and “0” otherwise. Similarly, we measure female representation beyond the token level (2 / 2 + WOMEN) with a dummy variable assigned a value of “1” when there are at least two female directors on the board and “0” otherwise.

In Column 1 of Table 7, the coefficient for 0/1WOMAN is positive and significant at the 1% level, while in Column 2, the coefficient for 2 / 2 + WOMEN is negative and significant at the 1% level with ETR. These results suggest a tokenism effect when there is at most one female board member, whereas achieving a critical mass of at least two females is necessary for curbing tax avoidance. The moderating role of female participation in audit committees is investigated using both a tokenism measure (at most one female, Column 3) and a critical mass measure (at least two females, Column 4). The interaction term 0/1WOMAN × FPAC is positive and insignificant, whereas 2 / 2 + WOMEN × FPAC is negative and marginally significant with ETR. These outcomes are consistent with Richardson et al. (2016), indicating that boards with at least two females significantly reduce tax aggressiveness. The results are also aligned with Mazumder (2025), who emphasises the importance of achieving a critical mass of female directors in strengthening governance in the banking sector of Bangladesh.

Neglecting clustering in panel data may distort standard errors and inflate statistical significance (Petersen, 2009). Following Mazumder (2025), one-way and two-way cluster-robust standard errors are applied to ensure the robustness of our findings. As shown in Table 8, the results with firm-wise clustering (Columns 1 and 2) and both firm and year-wise clustering (Columns 3 and 4) remain consistent with the outcomes reported in Table 4.

Researchers studying corporate governance, particularly those focusing on board structures and their role in corporate outcomes, often encounter the challenge of addressing endogeneity due to reverse causality. To mitigate this concern, lagged independent variables are employed following Usman et al. (2019) and Das and Hossain (2025). Regression results using the lagged model (Table 9) remain qualitatively consistent with those in Table 4.

Moreover, the impact of female directors on tax avoidance may be influenced by firm characteristics. To address any such potential self-selection bias, the two-stage Heckman (1979) method is employed. In the first stage, a probit regression uses a dummy variable (coded “1” for firms with at least one female director serving on the board, “0” otherwise) along with all control variables from the baseline model to predict female presence on boards. In addition, the industry average gender diversity (INDGEN) is included as an explanatory variable to satisfy the exclusion restriction, based on the assumption that industries with higher female representation are more likely to have gender-diverse boards (Usman et al., 2019). We then calculate the Inverse Mills Ratio (IMR) using a probit model similar to those in Richardson et al. (2016), Baatwah et al. (2024) and Usman et al.(2019) and include IMR in the second-stage analysis. Table 10, Column 1 reports the first-stage results, showing that INDGEN is significant, thereby validating the exclusion restriction. Overall, these findings are qualitatively similar to the baseline results in Table 4.

Finally, we apply the entropy balancing method to address the issue of observable heterogeneity among firms (Hainmueller and Xu, 2013). We classify the treatment group based on female representation on the board and audit committee. For equation (1), firms are assigned to the treatment group if the board includes at least one female director; otherwise, they are classified as control firms. For equation (2), firms are assigned to the treatment group if they have at least one female member not only on the board but also on the audit committee; firms not meeting this criterion constitute the control group. Panels A and B of Table 11 report descriptive statistics before and after entropy balancing, illustrating that initial differences in covariate means between treatment and control groups are substantially mitigated post-balancing. Subsequently, regressions are re-estimated using the entropy-balanced sample. Panel C presents these results, which confirm the robustness of our findings in Table 4 after controlling for observable heterogeneity.

Our main analysis shows that greater female representation on corporate boards is associated with reduced corporate tax avoidance. From a governance and agency theory perspective, gender-diverse boards may strengthen monitoring effectiveness and enhance accountability and transparency, thereby restricting managerial opportunism (Adams and Ferreira, 2009). Prior research provides evidence that gender diversity on boards improves the quality of financial reporting and constrains earnings management, which is commonly viewed as an empirical proxy for managerial opportunism. For example, Srinidhi et al. (2011), Hasan et al. (2022) and Attia et al. (2024) show that higher board gender diversity is associated with lower levels of earnings manipulation. Moreover, Wahid (2019) and Wang et al. (2022) find that it lowers the likelihood of fraudulent reporting and financial misconduct. Taken together, this evidence suggests that earnings management can serve as an observable manifestation of managerial opportunism in financial reporting. This literature therefore raises an important question: does earnings management serve as an underlying mechanism through which board gender diversity affects corporate tax avoidance?

This study, therefore, further investigates the mediating role of earnings management in the association between board gender diversity and tax avoidance. Earnings management is commonly examined using accrual-based models such as the Jones and Modified Jones models (Jones, 1991; Dechow et al., 1995). However, these models are generally unsuitable for financial firms because their regulatory framework and business operation differ fundamentally from those of non-financial firms (Chaity and Islam, 2021). Consistent with prior banking research, we employ loan loss provision-based measures of earnings management. Loan loss provisions are bank-specific, directly influence reported net income, and offer a clear proxy for managerial discretion in financial reporting (Kanagaretnam et al., 2004; Alam et al., 2020; Janahi et al., 2022). Accordingly, we employ the empirical specification used by Mersni and Ben Othman (2016) and Alam et al. (2020).

Firstly, we estimate the determinants of the two components of loan loss provisions (LLP): discretionary (DLLP) and non-discretionary (NLLP). NLLP captures default risk and expected credit losses beyond managerial control, whereas DLLP involves managerial judgement and discretion (Mersni and Ben Othman, 2016). We use the absolute residuals as proxies for DLLP. In the second stage, these residuals are regressed on explanatory variables.

where LLP denotes loan loss provisions, NPL represents non-performing loans, △NPL indicates changes in non-performing loans, and △LOAN refers to changes in total loans. To mitigate potential heteroscedasticity, all variables are scaled by the lagged beginning balance of loans (Alam et al., 2020).

We apply Baron and Kenny’s (1986) four-step regression framework to examine whether earnings management mediates the relationship between board gender diversity and corporate tax avoidance. The procedure involves:

1. establishing the direct effect by regressing the outcome (ETR) on the main independent variable (FPBOD);

2. regressing the mediator (DLLP) on the independent variable (FPBOD) to confirm their dependence;

3. regressing the outcome (ETR) on the mediator (DLLP) to confirm its explanatory role; and

4. jointly including the independent variable (FPBOD) and the mediator (DLLP) to determine whether the mediator attenuates or eliminates the original relationship, indicating partial or full mediation.

The first mediation condition is met, as FPBOD is statistically significant in explaining ETR at the 5% level (Column 2, Table 12). In Column 1 of Table 12, FPBOD is negatively and significantly associated with DLLP at the 10% level ^[5], indicating that board gender diversity reduces discretionary loan loss provisions and satisfies the second condition. Column 3 shows a significant negative association between DLLP and ETR, satisfying the third condition. In Column 4, both FPBOD and DLLP remain negative and significant, and the magnitude of the FPBOD coefficient weakens relative to Column 2, consistent with the proposed mediation mechanism and indicative of partial mediation. Moreover, the Sobel and Goodman tests further validate the causal interpretation of the mediation effect.

In sum, these findings provide evidence of a partial mediating role, suggesting that female directors help constrain earnings-management practices, which in turn contributes to lower levels of tax avoidance.

This study investigates the association between board gender diversity and tax avoidance in the financial sector of Bangladesh. It further examines whether female representation on the audit committee moderates this relationship. The empirical findings suggest that a higher proportion of female directors is associated with lower tax avoidance. The interaction between board gender diversity and female representation on the audit committee is negative and statistically significant, indicating that female directors on the audit committee strengthen the negative link between board gender diversity and tax avoidance. Our key findings are consistent with the notions of social role theory (Eagly, 1987) and social identity theory (Tajfel and Turner, 1986), which suggest that socially perceived communal traits associated with female directors (e.g. ethical awareness, risk aversion) and the shared group identity of female directors collectively foster accountability and ethical compliance by discouraging aggressive tax avoidance practices.

Further analysis reveals that this relationship is contingent on the number of female directors on the board. Specifically, the relationship between board gender diversity and tax avoidance is positive and significant when there is no or only one female director on the board but turns negative and significant once the board includes at least two female directors. Moreover, the interaction term involving two or more female directors and their representation on the audit committee is also negative and significant. This pattern aligns with the critical mass theory (Kanter, 1977a, 1977b), which suggests that female directors can exert meaningful influence only when their representation reaches a substantive threshold, thereby contributing to lower levels of tax avoidance.

To assess the robustness of the findings, we employ alternative measures of both tax avoidance and board gender diversity. We further mitigate endogeneity concerns by using lagged ordinary least squares regression, the Heckman two-step selection method, and entropy balancing. The main findings remain consistent across these robustness analyses. Finally, the mechanism analysis indicates that earnings management partially mediates the relationship between board gender diversity and tax avoidance.

The findings of this research have important managerial and policy implications. Firstly, the results suggest that greater female representation on corporate boards may serve as an effective governance mechanism for curbing tax avoidance. Boards of directors and nomination committees should therefore consider improving gender diversity when making board appointment decisions. Secondly, the moderation results highlight the importance of promoting gender diversity not only at the board level but also within key subcommittees, particularly the audit committee, where oversight of financial reporting and compliance is most critical. Therefore, financial firms should also ensure that female directors are represented in key board subcommittees. Thirdly, regulatory authorities such as the National Board of Revenue, Bangladesh Bank and the Bangladesh Securities and Exchange Commission may consider policies that promote greater female participation on corporate boards. In particular, introducing a minimum requirement of at least two female directors may help achieve the critical mass necessary for women directors to exert meaningful influence on corporate decision-making and governance outcomes like tax avoidance. However, implementing such policies may face practical challenges in Bangladesh due to the relatively limited pipeline of qualified and experienced female candidates for board positions (The Business Standard, 2025). Therefore, alongside mandating a minimum number of female representatives on boards, policymakers may also consider measures aimed at developing a stronger pipeline of qualified female directors, such as leadership development programmes, professional training and mentorship opportunities for women in corporate governance roles. Regulators may collaborate with professional bodies such as ICAB, ICMAB and ICSB to design and implement capacity-building programmes that expand the pool of qualified female candidates for board positions.

Like any empirical study, this study has several limitations. Firstly, the current study focuses on a single-country context, which may limit the generalisability of the findings to other countries with different institutional and cultural contexts. Secondly, although the study incorporates a comprehensive set of firm-level and governance control variables, it does not include macroeconomic variables (e.g. Inflation and GDP) or other political and behaviour factors that may also affect tax avoidance. Thirdly, although this study considers the numerical representation of females on corporate boards, it does not consider their educational background, professional expertise and workplace behaviour, which may also influence board effectiveness. Moreover, the potential contribution of female directors in Bangladesh may be hindered by institutional and socio-economic factors, including token appointments, limited professional networks and the dominance of family-controlled firms. Further research could explore how female directors navigate these structural constraints and endeavour to influence policy decisions, particularly through qualitative or a mixed-method approach. Finally, measuring tax avoidance and earnings management, particularly for firms in the financial sector, remains challenging due to their complex business models and business practices. Further research may therefore incorporate developing and employing more nuanced measures to assess the robustness of these findings.