Beyond Quotas: The Influence of Board Gender Diversity on Capital Structure in Firms from Latin America and the Caribbean

Abstract

Board gender diversity (BGD) has gained attention as a governance mechanism that may influence corporate financial decisions. However, empirical evidence from Latin America and the Caribbean (LAC) remains limited despite the region’s significant gender disparities in corporate leadership and distinct institutional characteristics. This study examines how BGD affects capital structure decisions in LAC firms, drawing on agency theory and resource dependency theory. We analyze a panel dataset of 403 firms from 2015 to 2022, sourced from the London Stock Exchange Group database, using fixed effects models with Driscoll–Kraay standard errors to control for firm heterogeneity and econometric concerns. Results show that BGD is significantly and negatively associated with leverage ratios, with a one percentage point increase in female board representation corresponding to a 0.15 to 0.25 percentage point decrease in debt-to-capital ratios. This relationship is robust across multiple specifications and exhibits threshold effects, with stronger impacts when female representation reaches 20% or higher. The negative association is more pronounced for larger firms, consistent with enhanced governance benefits in complex organizations. Our findings suggest that gender-diverse boards exercise more effective oversight of financial decisions, leading to more conservative capital structures in emerging markets where governance mechanisms are particularly important for firm credibility and stakeholder confidence.

1. Introduction

Corporate governance plays a critical role in shaping firms’ financial policies, particularly in emerging markets where governance structures, regulatory frameworks, and institutional environments vary significantly. Governance indicators such as Voice and Accountability (VA), Political Stability and Absence of Violence/Terrorism (PV), Government Efficiency (GE), Quality of Law (QL), Rule of Law (RL), and Control of Corruption (CC) influence firms’ ability to engage in sustainable financial practices and environmental, social, and governance (ESG) initiatives, which, in turn, impact firm value and capital structure decisions. Good governance fosters transparency, regulatory compliance, and accountability, ensuring firms operate within ethical and sustainable frameworks (Mooneeapen et al., 2022). These country-level factors promote the firm’s sustainable activities to achieve the Sustainable Development Goals (SDGs). Government policies and initiatives drive firms towards SDGs activities (Moon & Vogel, 2009). Country-level governance is broadly defined as policy formation and implementation.

The firm’s sustainable performance in the current corporate environment has undoubtedly gathered considerable importance in a global economy. The sound country governance is the regulatory framework that guarantees transparency and accountability (Pratama & Hermawan, 2023). These transparency and accountability policies would increase the stakeholders’ trust, arousing the firm’s value (Ammer et al., 2020). It balances a firm’s social and economic goals, motivating them to pursue sustainable development activities. The well-regulated governance mechanism effectively promotes the firm’s environmental, social, and governance (ESG) disclosure. They align themselves with ESG standards and sustainable practices for firm legitimacy (Mooneeapen et al., 2022). However, countries with weak governance foster unethical business practices where firms potentially engage in unsustainable practices and disregard environmental regulations.

As one of the leading corporate governance mechanisms, corporate boards have greatly influenced decision-making processes related to financing, investment, and dividend policy in firms (Ezeani et al., 2022; Mai et al., 2023). In recent years, there has been growing interest in analyzing the relationships of the main characteristics of corporate boards, including independent members, size, gender diversity, and structure type, with impact on profitability, capital structure, cost of debt, and risk firms, among others (Amin et al., 2022; García & Herrero, 2021; Govindan et al., 2023; Yakubu & Oumarou, 2023).

Notably, in the context of ESG, all issues related to governance (G) play a pivotal role in the business arena. For example, Agnese et al. (2023) found in the banking sector from the United States and Europe that the sub-pillars of ESG, such as management, shareholders, and corporate social responsibility strategy, show a positive and statistically significant relationship with ESG controversies. Likewise, Shakil (2021) stated that board gender diversity (BGD), which is included in the sub-pillar governance, substantially moderates ESG results and financial risk connections. However, although BGD is one of the leading topics in the G context (Bazhair, 2023; Datta et al., 2021; Teodosio et al., 2023) and studies on its relationship with capital structure decisions are at the top of sustainability research (Alves et al., 2015; Ben Saad & Belkacem, 2022; Ezeani et al., 2022; Hordofa, 2023), it has led to heterogeneous results, which indicates much is still unrevealed regarding how, why, and to what extent firms choose to take on debt.

The relationship between BGD and capital structure decisions represents a critical intersection of corporate governance and financial strategy. While traditional capital structure theories focus on factors such as firm size, profitability, and growth opportunities, emerging research suggests that board composition—particularly gender diversity—may significantly influence how firms balance debt and equity financing. This relationship is theoretically grounded in both agency theory and resource dependency theory, which suggest that diverse boards may reduce agency costs through enhanced monitoring while simultaneously improving access to external resources.

Among the leading scholarly studies underscoring the nexus between BGD and capital structure, the following studies merit particular attention. On the one hand, Adusei and Obeng (2019) found a robust negative and statistically significant effect of BGD on capital structure in 441 microfinance institutions located in 69 countries. In this line, Wang and Ramzan (2020) argued that corporate boards with diverse members in Pakistan are considered a sign of a good governance structure, which could impact capital structure decisions. For example, Alves et al. (2015) found evidence, for 2427 firms in 33 countries, that a more gender-diversified board of directors can improve independence and efficiency. Therefore, it leads firms to have a capital structure composed of more long-term sources of financing.

On the contrary, considering that the influence of female executives on debt maturity is inversely related to the proportion of their incentive compensation, Datta et al. (2021) found that female executives choose a significantly shorter debt maturity structure than their male counterparts in the United States. Similarly, La Rocca et al. (2020) concluded that female executives prefer higher levels of short-term debt to preserve financial viability in the capital structure. This reinforces the hypothesis that female executives are less overconfident. This finding is relevant in countries with high masculinity scores, where competitiveness and material rewards for success are significant. A study conducted by López-Delgado and Diéguez-Soto (2020) established that women on corporate boards in Spain negatively influence indebtedness. In line with a growing body of research highlighting the impact of gender diversity on capital structure and financial distress, García and Herrero (2021) found that, in the 2002–2019 period, European firms with a higher percentage of women directors adopted more conservative capital structures. Specifically, their presence is linked to lower leverage, debt costs, and debt maturities. Similarly, Hordofa (2023) found that BGD consistently correlated with lower debt utilization, indicating that women on corporate boards contribute to risk control, monitoring, and better decision-making. Likewise, Siregar et al. (2023), in a sample of nonfinancial sector companies listed on the Indonesia Stock Exchange in 2012–2021, found that BGD is significantly negatively related to capital structure as measured by debt-to-equity ratio and short-term debt-to-total assets. Thus, the critical mass of women provided no evidence of this relationship with firms’ capital structure.

On the other hand, the study conducted by Yakubu and Oumarou (2023) revealed that BGD showed a significant positive impact on capital structure, supporting resource dependency theory, as gender-diverse boards improve a firm’s reputation and attractiveness to lenders. In this line, Ben Saad and Belkacem (2022) found that this relationship is mediated by information transparency and firm risk-taking channels. This suggests that capital structure decisions may vary depending on the approach used for corporate board composition, specifically voluntary, enabling, or coercive. Based on a comprehensive analysis of the 1590 listed non-financial firms on the Taiwan Stock Exchange and the Taipei exchanges covering the period 2007–2020, Chang et al. (2023) found that firms with greater BGD are inclined to leverage debt financing more heavily in their capital structure decisions, as revealed by both correlation analysis and multiple regression estimations. This suggests that BGD may be associated with increased access to funds, favorable loan terms, and potentially higher lender trust in the firm’s debt repayment capabilities. In the same line, Sani et al. (2020) revealed that in the Nigerian listed firms, BGD, foreign directorship, and capital structure are positively associated. They indicated that stringent monitoring by foreign and female directors compels managers to pursue high-debt policy to boost firms’ financial performance. The study recommends that firms should embrace the culture of diversity in their boardrooms to enhance their governance system.

Importantly, our study builds upon recent findings from related contexts. Boubaker et al. (2014) examine board gender diversity and firm performance in French listed firms over 2009–2011, using a similar 2SLS approach to address endogeneity concerns. Their findings reveal a negative relationship between the percentage of female directors and Tobin’s q when controlling for endogeneity, while the mere presence of women (dummy variable) shows no significant effect. This methodological and empirical precedent in the French context provides important validation for our approach, though we extend the analysis to Latin American markets and focus specifically on capital structure rather than broader performance measures.

Complementing this board-level evidence, Farooq et al. (2022) provide insights into the behavioral mechanisms underlying gender effects in corporate leadership. Their study of female CEOs and related-party transactions in Chinese firms demonstrates that female executives exhibit more ethical and risk-averse behavior, engaging in fewer potentially opportunistic transactions. While their focus is on CEO-level decisions rather than board composition, their findings support the theoretical mechanisms we propose—that female leadership contributes to more conservative financial policies through enhanced monitoring and reduced risk-taking.

As explained before, BGD remains controversial and empirical evidence does not yield definitive findings, particularly in developing regions. Hence, further research is encouraged to understand the relationship between BGD and firm outcomes, such as capital structure and BGD (Gonzalez-Ruiz et al., 2024). For example, while BGD is growing in the Latin America and the Caribbean (LAC) region, participation gaps remain compared to other regions (Arango-Home et al., 2023). Although Chile, Peru, and Brazil have the highest number of women board chairs, quotas are below 10%, while Europe and the United States have about 38% and 25%, respectively (Korn Ferry, 2022). According to García and Herrero (2021), some firms have appointed token women on boards to demonstrate adherence to regulatory requirements and maintain a socially acceptable reputation, and, as a result, they are unfortunately relegated to subordinate positions within the governance structure.

While most previous studies in diverse regions and countries concern board characteristics related to capital structure and gender diversity participation, there is scarce knowledge of this relationship in LAC. The only study found on this issue was conducted in Brazil by Nisiyama and Nakamura (2018). The findings suggested that BGD is positively related to firms’ leverage. However, the lack of evidence from other LAC countries shows research opportunities. In this way, despite initiatives like the 30% Club driving increased female representation on boards and in C-suites across LAC (30% Club, 2023), a critical knowledge gap exists regarding their impact on financial outcomes, particularly capital structure. This gap is especially relevant given the recent surge in research on BGD and its role in firm performance globally.

To address this gap, this study examines the relationship between BGD and capital structure in LAC firms, investigating whether increased female representation on boards influences financial decision-making on leverage strategy. Drawing on agency theory and resource dependency theory, we hypothesize that firms with higher BGD adopt more conservative leverage policies and have improved access to financing. Using a panel dataset of 480 firms from 2015 to 2022, obtained from the London Stock Exchange Group database (LSEG), we apply a fixed effects model with robust standard errors to control potential endogeneity concerns.

This paper makes several contributions. First, to the best of our knowledge, it is the first study to analyze the relationship between BGD and capital structure in LAC. Thus, this study addresses the call for further research directions indicated by Zaid et al. (2020). They emphasized the necessity of analyzing the influence of corporate governance mechanisms on capital structure, particularly within the context of developing regions. Likewise, based on agency costs and resource dependency theory, this study contributes to the scarce empirical studies addressing this relationship. Consequently, findings could give policymakers, investors, and lenders a better understanding of how to improve the decision-making process. Second, it contributes to one of the most important gender-related challenges, the fifth Sustainable Development Goal (Gender Equality). Women represent half of the world’s population. However, gender inequality persists everywhere and stagnates social progress (United Nations, 2023). In this way, it will allow insights supporting women’s inclusion on corporate boards and C-level positions. Third, through a multi-country empirical study, this study goes beyond examining the relationships between BGD and capital structure. By incorporating variables such as board size, the presence of independent members, CEO duality, and board structure into the analysis of Latin American and the Caribbean (LAC) firms, the findings provide a comprehensive understanding that contributes to a more holistic view of their moderating effects in corporate governance dynamics in the region. Finally, this study contributes to the state-of-the-art of BDG and capital structure by highlighting financing and investing characteristics. Thus, it enriches the standing of the limited studies on gender studies in management and finance and extends the literature on this scarce but growing research arena.

Furthermore, it is crucial to situate this study within the broader macroeconomic context of the LAC region, which is characterized by significant volatility in capital flows and interest rates (World Bank, 2023; IMF, 2022). The period of our analysis (2015–2022) encompasses the unprecedented COVID-19 pandemic, during which central banks across the region enacted dramatic monetary policy shifts (Carstens, 2021; IMF, 2021)—for instance, Brazil’s Selic rate fluctuated from a historic low of 2% in 2020 to 13.75% in 2022 (Banco Central do Brasil, 2023). While a firm’s internal governance mechanisms, such as board composition, are a critical determinant of capital structure, these external macroeconomic forces undoubtedly also play a major role (Didier et al., 2021). This study primarily focuses on the internal governance mechanism of board gender diversity (BGD), while acknowledging that the interplay between these firm-level and country-level factors presents a rich area for future research.

The remainder of this paper is organized as follows. Section 2 reviews the relevant literature and develops our theoretical framework and hypotheses. Section 3 describes the methodology, data, and empirical approach. Section 4 presents the empirical results, while Section 5 discusses the findings and their implications. Section 6 concludes with a summary of key findings, limitations, and directions for future research.

2. The Literature Review and Hypotheses Development

2.1. Theoretical Framework

Sustainable development activities, in terms of ESG, guarantee low risk and volatility and maintain market efficiency. This efficiency in market efficiency reduces the information asymmetry by developing trust among various stakeholders.

The study of corporate board behavior has led to the development of several theoretical frameworks that have allowed a better understanding of interactions among managers, board members, stakeholders, and firms in the decision-making processes. In this way, monitoring management on behalf of shareholders (agency cost theory) and providing resources (resource dependency theory) allow for a better understanding of how corporate boards work (Hillman & Dalziel, 2003). Thus, they are among the most prominent theories supporting the most corporate board studies.

On the one hand, the agency theory endorses that monitoring and controlling managers plays a crucial role in the board’s functions (Stoiljković et al., 2024). Thus, managers serve as agents who possess the authority to make decisions but do not bear the repercussions (Jensen & Meckling, 1976) and often make decisions to fulfill their self-interests (Naseem et al., 2020). A noteworthy part of the effort of researchers in capital structure indicated that, due to conflicts of interest, capital structure is determined by agency costs (Harris & Raviv, 1991). Research on capital structure and agency cost theory was introduced by Jensen and Meckling (1976). This theory states that two types of agency conflicts could affect firms, managers, and majority and minority shareholders (Gavana et al., 2023; Harris & Raviv, 1991). The first suggests that the interests of these two groups may not always align. Owners desire the highest possible return on their investment (maximizing stock price and dividends), while managers might prioritize their goals, such as job security, prestige, or personal profit. Later, this conflict occurs when a firm has shareholders with different ownership stakes. Most shareholders hold a controlling interest, significantly influencing the firms’ decisions. Minority shareholders, on the other hand, have less influence and could be disadvantaged by the actions of the majority. For these reasons, given that lenders value transparency as outsiders, and to mitigate agency costs, they may include covenants in loan agreements (Pucheta-Martínez et al., 2022). Consequently, firms’ disclosure aims to mitigate agency costs caused by information asymmetries (La Rosa et al., 2018; Zamil et al., 2023). Thus, this theory offers a valuable standpoint for comprehending how governance mechanisms function to align the interests of managers and shareholders, thereby mitigating agency costs (Hordofa, 2023).

From this perspective, it is proposed that firms actively participate in disclosure initiatives. This strategic engagement aims to mitigate information asymmetries between firms, managers, and investors, enhancing transparency and ultimately creating shareholder value (Al Lawati & Alshabibi, 2023; Khaled et al., 2021). For this purpose, given that BGD increases firm transparency and decreases firm risk-taking, it is pivotal for designing business strategies, particularly those related to capital structure decisions (Ben Saad & Belkacem, 2022). Moreover, a more diverse corporate board may improve the decision-making process because diversity increases independence (Carter et al., 2010), thus allowing for a reduction in agency costs (Adams & Ferreira, 2009). Consequently, BGD could increase investors’ and lenders’ confidence (Amin et al., 2022), making it easier for firms to raise external financial resources (Ben Saad & Belkacem, 2022). In this context, BGD is considered one of the most essential characteristics of an effective corporate board (Milliken & Martins, 1996). It is supported by agency theory, which provides a framework to understand how corporate governance mechanisms may impact capital structure decisions (Hordofa, 2023). Hence, in line with existing studies, agency theory stands as a foundational framework to analyze firms in our sample and their repercussions on the impact on capital structure.

On the other hand, although resource dependence theory, which was developed by Pfeffer and Salancik (1978), remains a relatively underexplored concept in the examination of board directors in developing regions, it holds valuable significance in comprehending the board’s function as a supplier of vital resources to the firm and its impact on performance (Hillman & Dalziel, 2003; Marquez-Cardenas et al., 2022). This theory claims that the principal purpose of corporate boards is to seek access to resources and effectively secure them, all to serve the shareholders’ interests (Elzahar et al., 2022). In this way, it posits that corporate boards provide fundamental resources employing legitimacy, networks, and expertise (Alhossini et al., 2021). Also, under this theory, firms are open structures reliant on external factors (Pfeffer & Salancik, 1978), and thus, corporate boards play a crucial role as a mechanism to decrease external dependency and mitigate environmental uncertainty (Hillman et al., 2009; Teodosio et al., 2023). Based on this, by improving transparency and reducing firm risk-taking, corporate boards with diversity could influence access to financing (Ben Saad & Belkacem, 2022; Hordofa, 2023). In this way, this theory argues that corporate boards have a critical role to maximize performance between firms and resources (Sargon, 2024). In this same line, Nicholson and Kiel (2007) propose that corporate boards with strong relations to the external environment could provide firms with high access to resources and, consequently, high corporate performance, such as resources including finance and capital, which could lead to an increase in firms’ value (Jackling & Johl, 2009).

2.2. Empirical Evidence and Research Gaps

To further support the theoretical discussion, Figure 1 presents a co-occurrence network analysis that identifies key research themes in the literature related to BGD and capital structure. The figure highlights “gender diversity,” “board of directors,” and “board gender diversity” as the most dominant themes, reinforcing their central role in corporate governance research. Additionally, the interconnectedness of these themes with “capital”, “governance approach,” and “sustainable development” indicates that discussions extend beyond representation quotas, encompassing financial decision-making and corporate responsibility. This supports the argument that BGD substantially influences firms’ strategic and financial outcomes.

The temporal evolution of research topics, as reflected in the color gradient of Figure 1, shows a shift from earlier discussions on governance and diversity to newer concerns such as “CEO overconfidence,” “ethnicity,” and “green economy.” This suggests an increasing intersection between gender diversity, behavioral finance, and sustainability considerations. Moreover, the figure reveals cross-regional research patterns by including terms such as “sub-Saharan Africa” and “Turkey,” indicating that the effects of BGD on financial decision-making are being explored in various emerging markets. This opens opportunities to discuss whether findings from other regions apply to LAC firms, particularly in the context of institutional and regulatory frameworks. The presence of financial terms such as “venture capital,” “financial markets,” and “profitability” further aligns with the study’s hypothesis that BGD influences access to external financing and capital structure decisions.

The resource dependence theory posits four crucial assumptions (Han & Gu, 2021): (1) a firm’s primary concern is survival; (2) for survival, the firm depends on resources it typically cannot generate internally; (3) the firm must establish connections with its dependencies, often involving interactions with other firms in the environment; and (4) firm’s survival is contingent upon the availability and strategic management of these external resources. Likewise, the economic benefits from BGD can be highlighted and explained through different theories, namely agency cost and resource dependence, which argue that the presence of women on the boards increases the set of information and leads the decision-making processes (Poletti-Hughes & Martínez Garcia, 2022). This theory provides a suitable framework for better understanding board characteristics, capital structure, and firm leverage relationships (Hordofa, 2023). Based on all the above, the significance of corporate boards can be justified not solely through the lens of agency theory but also in conjunction with resource dependence theory (Hillman & Dalziel, 2003). Hence, by recognizing the interplay between both theories, corporate boards will likely be a moderating factor in capital structure decisions, influencing monitoring and resource provision (Poletti-Hughes & Martínez Garcia, 2022). Based on these theoretical frameworks, this study aims to assess the effect of BGD on capital structure in firms.

Finally, theories of agency costs and resource dependence are two important perspectives in corporate finance that have been used to explain the relationship between BGD and capital structure. Both theories suggest that, by reducing agency costs and increasing access to resources, BGD could help firms improve their performance as well as financial leverage.

2.3. Research Question and Hypothesis Development

Given the identified knowledge gap, the following research question was formulated: How does BGD influence firms’ capital structure? Considering the mixed evidence in the existing literature regarding the relationship between BGD and firms’ financial decisions, the theoretical framework, and the contextual characteristics of LAC markets, we develop the following hypotheses (H) to test different aspects of this relationship.

H₁ (Main Effect Hypothesis).

Board gender diversity negatively affects firms’ capital structure (debt-to-capital ratio).

Rationale: Given the institutional characteristics of LAC markets—including weaker legal frameworks, higher information asymmetries, and concentrated ownership structures—we expect that the agency-theory predictions will dominate. Gender-diverse boards should provide more effective monitoring and oversight, leading to more conservative financial policies and lower leverage ratios.

H₂ (Non-linearity Hypothesis).

The relationship between board gender diversity and capital structure is non-linear, with stronger effects at higher levels of female representation.

Rationale: Critical mass theory suggests that tokenism may limit the effectiveness of individual women directors, while meaningful representation (typically 20–30%) may be required for diverse perspectives to influence board decisions effectively.

H₃ (Firm Size Moderation Hypothesis).

The negative relationship between board gender diversity and leverage is stronger for larger firms.

Rationale: Larger firms face greater public scrutiny and stakeholder pressure, potentially amplifying the reputational and legitimate benefits of board diversity. Additionally, larger firms may have more sophisticated governance structures that enable diverse boards to exercise more effective oversight.

H₄ (Institutional Context Hypothesis).

The relationship between board gender diversity and capital structure varies across countries within LAC based on institutional development.

Rationale: Countries with stronger legal institutions and more developed capital markets may exhibit different BGD effects due to varying baseline governance quality and financing constraints.

These hypotheses allow us to test not only the main effect of BGD on capital structure but also important boundary conditions and moderating factors that may explain the mixed findings in the prior literature. By examining these relationships in the understudied LAC context, we aim to contribute to both the regional corporate governance literature and the broader understanding of how institutional factors influence the effectiveness of governance mechanisms.

This research aims to contribute to the growing body of work on sustainable finance by examining whether gender diversity at the board level influences key financial decisions, such as the balance between debt and equity financing.

3. Materials and Methods

3.1. Research Design and Data Sources

To analyze the relationships between BGD and capital structure, this research employs a comprehensive panel data approach using multiple fixed effects model specifications to ensure robust inference. Our analysis addresses several econometric challenges commonly encountered in corporate governance research, including endogeneity concerns, unobserved heterogeneity, and sample selection bias.

Data were collected from the London Stock Exchange Group database (LSEG), formerly known as Eikon Thomson Reuters Refinitiv, covering the period from 2015 to 2022. This timeframe was selected to capture recent corporate governance developments while ensuring data availability and quality across the region. The LSEG database provides comprehensive ESG and financial data that has been audited by specialized ESG analysts, offering superior data quality compared to manually collected governance information.

Our initial sample comprised all publicly listed firms in LAC countries with available governance and financial data in the LSEG database. After applying data cleaning procedures and filters for data completeness, our final sample consists of 403 firms with 2202 firm-year observations, representing an unbalanced panel structure. This represents comprehensive coverage of publicly listed firms in the region, though we acknowledge potential selection bias toward larger, more internationally visible companies that are included in the LSEG database.

3.2. Sample Selection and Potential Biases

The study focuses on firms headquartered in Latin America and the Caribbean, ensuring regional consistency in institutional, economic, and cultural contexts. To maintain the reliability of the analysis, only companies with a minimum of three consecutive years of available data for the key variables were included. In robustness checks, financial institutions were excluded given their distinct capital structure dynamics, which could bias the results. Additionally, to mitigate the influence of extreme observations, firms with key financial ratios falling above the 99th percentile or below the 1st percentile were removed from the sample.

We acknowledge several potential sources of selection bias that may affect our results. First, the LSEG database may be biased toward larger, more internationally oriented firms that are more likely to report comprehensive ESG data. This could limit the generalizability of our findings to smaller, domestically focused companies. Second, firms that voluntarily provide governance information may systematically differ from those that do not, potentially biasing our estimates. Third, survival bias may affect our results if firms with poor governance or extreme capital structures exit the sample during our study period.

To address these concerns, we performed the following: (1) conduct robustness checks using alternative samples; (2) compare our sample characteristics with broader population statistics available; and (3) employ instrumental variable approaches to address potential endogeneity.

3.3. Variable Definitions and Measurement

3.3.1. Dependent Variable

Capital structure (Debt Weight—DW): This variable measures the firm’s leverage as the ratio of total debt to total capital (debt plus equity). It is also known as the debt ratio (Debt/Assets). It plays a critical role in shaping firms’ financing strategies and managing their financial obligations effectively, influencing both equity and debt levels. This variable measures how much firms rely on debt versus equity in their capital structure. It is expressed as a percentage between 0% and 100%, where 0% indicates that the firm is entirely equity-financed with no debt, and 100% indicates that debt is the sole source of financing for all the firm’s assets.

We also construct alternative dependent variables for robustness checks: Short-term debt ratio: Short-term debt divided by total assets; Long-term debt ratio: Long-term debt divided by total assets; Debt-to-equity ratio: Total debt divided by total equity.

3.3.2. Main Independent Variable

Board gender diversity (BGD): This variable measures the proportion of women serving on the board of directors, calculated as the percentage of female directors relative to total board size. BGD ranges from 0% (no women on the board) to 100% (all board members are women). To address critical mass theory, we also created binary variables for different threshold levels (≥20%, ≥30%) and a categorical variable distinguishing between no women (0%), token representation (1–19%), and meaningful representation (≥20%).

3.3.3. Control Variables

Board characteristics:

• Board size (BS): Total number of board members (log-transformed to address skewness).
• Independent board members (IBM): Percentage of independent directors on the board.
• CEO board member: Binary variable indicating whether the CEO serves on the board (CEO duality).
• Board structure type: Categorical variable distinguishing between unitary, two-tier, and mixed board structures.

Firm characteristics:

• Total assets (TA): Log of total assets as a proxy for firm size.
• Profitability: Return on assets (ROA) to control for earnings capacity.
• Growth opportunities: Market-to-book ratio.
• Asset tangibility: Property, plant, and equipment divided by total assets.
• Industry: NAICS sector classifications.

Country-level controls:

• GDP growth: Annual real GDP growth rate.
• Financial development: Private credit by banks as a percentage of GDP.
• Governance quality: World Bank governance indicators (when available).
• Legal origin: Binary variables for civil law vs. common law traditions.

3.4. Econometric Methodology

We proposed the following models to test these hypotheses with the following regression models:

$$\begin{gathered} \mathrm{Fixed} \mathrm{Model}: {\mathit{Debt} \mathit{Weight}}_{it}={\beta }_0+{\beta }_1{\mathit{BoardGenderDiversity}}_{it}+{\beta }_2{\mathit{BoardSize}}_{it}+ \\ {\beta }_3{\mathit{IndependentBoardMembers}}_{it}+{\beta }_4{\mathit{TotalAssets}}_{it}+{\mu }_i+{ϵ}_{it} \end{gathered}$$

(1)

$$\begin{gathered} \mathrm{Model} 1: {\beta }_1{BoardGenderDiversity}_{it}+{\beta }_2{BoardSize}_{it}+{\beta }_3{IndependentBoardMember}_{it}+ \\ {\beta }_4{TotalAssets}_{it}+{\beta }_5{BoardStructureType}_{it}+{\beta }_6{CEOBoardMember}_{it}+{\lambda }_t+{\epsilon }_{it} \end{gathered}$$

(2)

$$\begin{gathered} \mathrm{Model} 2: {\beta }_1{BoardGenderDiversity}_{it}+{\beta }_2{BoardSize}_{it}+{\beta }_3{IndependentBoardMember}_{it}+ \\ {\beta }_4{TotalAssets}_{it}+{\beta }_5{BoardStructureType}_{it}+{\beta }_6{CEOBoardMember}_{it}+\delta Si+{\lambda }_t+{\epsilon }_{it} \end{gathered}$$

(3)

$$\begin{gathered} \mathrm{Model} 3: {\beta }_1{BoardGenderDiversity}_{it}+{\beta }_2{BoardSize}_{it}+{\beta }_3{IndependentBoardMember}_{it}+ \\ {\beta }_4{TotalAssets}_{it}+{\beta }_5{BoardStructureType}_{it}+{\beta }_6{CEOBoardMember}_{it}+\gamma Ci+{\lambda }_t+{\epsilon }_{it} \end{gathered}$$

(4)

$$\begin{gathered} \mathrm{Model} 4: {\beta }_1{BoardGenderDiversity}_{it}+{\beta }_2{BoardSize}_{it}+{\beta }_3{IndependentBoardMember}_{it}+ \\ {\beta }_4{TotalAssets}_{it}+{\beta }_5{BoardStructureType}_{it}+{\beta }_6{CEOBoardMember}_{it}+\delta Si+\gamma Ci+{\lambda }_t+{\epsilon }_{it} \end{gathered}$$

(5)

$$\begin{gathered} \mathrm{Model} 5: {\beta }_0+{\beta }_1{BoardGenderDiversity}_{it}+{\beta }_2{BoardSizeL}_{it}+{\beta }_3{IndependentBoardMember}_{it}+ \\ {\beta }_4{TotalAssetsL}_{it}+{\beta }_5{BoardStructureType}_{it}+{\beta }_6{CEOBoardMember}_{it}+c\sum {\gamma }_c{Country}_i+{\lambda }_t+{\epsilon }_{it} \end{gathered}$$

(6)

$$\begin{gathered} \mathrm{Model} 6: {\beta }_0+{\beta }_1{BoardGenderDiversity}_{it}+{\beta }_2{BoardSizeL}_{it}+{\beta }_3{IndependentBoardMember}_{it}+ \\ {\beta }_4{TotalAssetsL}_{it}+{\beta }_5{BoardStructureType}_{it}+{\beta }_6{CEOBoardMember}_{it}+c\sum {\gamma }_c{Sector}_i+{\lambda }_t+{\epsilon }_{it} \end{gathered}$$

(7)

$$\begin{gathered} \mathrm{Model} 7: {\beta }_0+{\beta }_1{BoardGenderDiversity}_{it}+{\beta }_2{BoardSizeL}_{it}+{\beta }_3{IndependentBoardMember}_{it}+ \\ {\beta }_4{TotalAssetsL}_{it}+{\beta }_5{BoardStructureType}_{it}+{\beta }_6{CEOBoardMember}_{it}+\sum {\delta }_s{Sector}_i+c\sum {\gamma }_c{Country}_i+ \\ {\alpha }_i+{\lambda }_t+{\epsilon }_{it} \end{gathered}$$

(8)

On Model 7:

• α_i: firm-fixed effects.
• λ_t: time-fixed effects.
• δ_s: industry-fixed effects.
• γ_c: country-fixed effects.

As mentioned above, data were extracted from the London Stock Exchange Group database (LSEG), which provides a comprehensive data source about ESG issues. This dataset includes financial information gathered from public sources and audited by ESG specialists for firms’ annual and CSR reports globally.

3.5. Estimation Technique and Robustness

All models are estimated using fixed effects with Driscoll and Kraay (1998) standard errors that are robust to heteroskedasticity, autocorrelation, and cross-sectional dependence. This approach is particularly important given the panel structure of our data and the potential for correlated errors across firms within the same country or industry.

Driscoll–Kraay Standard Errors Specification: Maximum lag length: 3 years; Correction for heteroskedasticity: HC1 estimator; Cross-sectional dependence: Allowed.

We conduct extensive robustness checks, including the following: (1) Alternative dependent variable specifications; (2) Subsample analyses by country, industry, and firm size; (3) Different time periods and panel structures; (4) Alternative measures of board gender diversity; (5) Inclusion of additional control variables; and (6) Bootstrapped standard errors for inference validation.

3.6. Model Selection and Diagnostic Tests

Model selection follows a systematic approach using formal statistical tests: (1) F-test: Compares OLS vs. fixed effects models to determine whether firm-specific effects are significant; (2) Hausman test: Compares fixed effects vs. random effects specifications to test for correlation between regressors and unobserved heterogeneity; (3) Multicollinearity assessment: Variance Inflation Factor (VIF) tests to ensure independent variables are not highly correlated; and (4) Specification tests: Ramsey RESET tests and residual analysis to detect model mis-specification.

3.7. Sample Characteristics and Representativeness

The sectoral and geographical distribution of firms shown in Figure 2 and Figure 3, respectively, provide critical insights into the relationship between BGD and capital structure in Latin America and the Caribbean. The manufacturing (23.67%) and finance and insurance (18.76%) sectors dominate the dataset, highlighting the presence of capital-intensive industries, where corporate governance, including diverse boards, may significantly influence financial decision-making. Conversely, industries such as arts, entertainment, and recreation (0.43%) and management of firms and enterprises (0.43%) have minimal representation, suggesting that firms in these sectors might rely less on external financing and exhibit weaker governance structures. These sectoral disparities align with agency and resource dependence theories, as firms in capital-intensive industries may require stronger governance mechanisms to reduce agency costs and enhance financial decision-making. In contrast, service-based industries may operate under different financial constraints.

Similarly, Figure 3 reveals the geographical distribution of firm headquarters, showing that Brazil (36.46%) and Mexico (24.52%) have the largest representation, followed by Argentina (12.79%) and Chile (10.66%), whereas Costa Rica (0.21%) has the lowest participation. The dominance of firms from Brazil and Mexico suggests that the dataset primarily reflects governance practices in the region’s largest economies, where regulatory frameworks and market conditions could shape how BGD influences capital structure decisions. Meanwhile, the presence of firms from smaller economies such as Puerto Rico (1.07%), Panama (0.64%), Uruguay (0.43%), and Costa Rica (0.21%) indicates that the impact of governance mechanisms, including BGD, may differ based on economic size and regulatory environments. Given the differences in corporate governance and financial systems among these countries, the study can explore whether the impact of BGD varies across institutional settings, providing a comparative analysis of governance effectiveness across the region. By integrating these findings, the manuscript strengthens its empirical foundation and enhances its theoretical contributions, linking sectoral and geographical characteristics to governance practices and financial decision-making in firms across Latin America and the Caribbean.

The sectoral and geographical distribution of firms, shown in Figure 2 and Figure 3, respectively, demonstrate that our sample covers the major economies in the region, with Brazil (36.46%) and Mexico (24.52%) representing the largest components. While this concentration reflects the economic structure of the region, we conduct robustness checks to ensure these dominant countries do not drive our results.

Manufacturing (23.67%) and finance and insurance (18.76%) sectors dominate the dataset, which is consistent with the industrial structure of major LAC economies. This sectoral composition allows us to examine governance effects across capital-intensive industries where board oversight may be particularly important for capital allocation decisions.

4. Results

4.1. Descriptive Statistics and Correlation Analysis

The descriptive statistics presented in

Table 1. Descriptive statistics of models’ variables.

Variable	Mean	Std. Dev.	Min	25%	50%	75%	Max
BGD	10.845	11.009	0	0	10	16.667	85.714
BS	9.865	3.469	1	7	9	12	25
IBM	38.813	23.426	0	22.222	37.5	55.556	100
DW	38.024	24.92	0	17.261	37.023	55.819	99.331
TA	15,764.142	45,409.163	1.467	1525.672	4166.659	12,359.188	524,123.575

Source: Based on the authors’ research, using data from the London Stock Exchange Group (LSEG) database, formerly Eikon Thomson Reuters Refinitiv.

summarize the key variables relevant to testing our four hypotheses regarding the relationship between board gender diversity (BGD), board composition characteristics, and capital structure decisions in LAC firms. These variables provide the empirical foundation for examining how board gender diversity influences leverage decisions through main effects (H₁), non-linear relationships (H₂), firm size moderation (H₃), and institutional context variations (H₄).

The descriptive statistics reveal several patterns that directly inform our hypothesis-testing framework. First, the BGD distribution provides crucial insights for testing H₁ (Main Effect Hypothesis) and H₂ (Non-linearity Hypothesis). The average BGD of 10.85%, with a median of 10%, confirms significant under-representation of women in LAC corporate governance, while the range from 0% to 85.714% provides sufficient variation to detect both linear and non-linear effects. Importantly, the 25th percentile of 0% and the 75th percentile of 16.67% create natural breakpoints for testing critical mass theory, as approximately 25% of firms have no female representation, while another quartile achieves meaningful representation levels approaching the theoretical 20% threshold identified in the prior literature.

Second, the capital structure variability supports robust testing of our main hypothesis regarding BGD’s impact on leverage decisions. With an average debt weight (DW) of 38.02% and substantial standard deviation of 24.92%, firms exhibit significant heterogeneity in financing strategies ranging from entirely equity-financed (0%) to highly leveraged (99.33%) operations. This variation is essential for identifying the relationship proposed in H₁, while the broad distribution suggests that governance mechanisms may play important roles in explaining cross-sectional differences in capital structure choices. The moderate median leverage of 37.02% indicates that most LAC firms maintain balanced financing approaches, providing a relevant baseline for assessing how board gender diversity influences deviations from typical leverage patterns.

Third, the total assets distribution directly relates to H₃ (Firm Size Moderation Hypothesis), with substantial variation from small firms (minimum 1.467 million) to large corporations (maximum 524.123 billion), enabling robust testing of how firm size moderates the BGD–capital structure relationship. The wide dispersion (mean 15.764 billion, standard deviation 45.409 billion) suggests significant heterogeneity in organizational complexity that may influence the effectiveness of governance mechanisms. This variation supports our theoretical prediction that larger, more complex firms may derive greater benefits from board gender diversity through enhanced monitoring and oversight capabilities.

These descriptive patterns provide strong empirical support for our theoretical framework, grounded in agency theory and resource dependency theory. The under-representation of women in corporate governance (low BGD), combined with substantial variation in capital structure decisions, creates ideal conditions for testing whether gender-diverse boards influence financial policies through enhanced monitoring (agency theory) or improved resource access (resource dependency theory). The substantial heterogeneity across firms in both governance characteristics and financial strategies suggests that board composition effects, if they exist, should be detectable in our comprehensive econometric analysis. Furthermore, the distribution characteristics support testing our complete hypothesis framework, from main effects through contextual moderators, providing the foundation for a rigorous examination of how board gender diversity shapes corporate financial decision-making in Latin American and Caribbean markets.

4.2. Construction of the Model

4.2.1. Correlation Analysis

To provide initial insights into the relationships underlying our four hypotheses, we conducted a comprehensive correlation analysis examining the associations between board gender diversity and key governance and financial variables in LAC firms. As presented in

Table 2. Means, standard deviations, and correlations with confidence intervals.

Variable	Mean	Std. Dev.	BGD	BS	IBM	TA
BGD	10.85	11.01	1
BS	9.87	3.47	−0.04 *	1
			[−0.09, −0.00]
IBM	38.81	23.43	0.19 **	−0.01	1
			[0.14, 0.23]	[−0.05, 0.03]
TA	15,764.14	45,409.16	0.06 **	0.09 **	0.04	1
			[0.02, 0.10]	[0.05, 0.13]	[−0.00, 0.08]
DW	38.02	24.92	−0.02	−0.01	0.06 **	0.25 **
			[−0.06, 0.03]	[−0.05, 0.03]	[0.02, 0.10]	[0.21, 0.29]

Note. M and SD are used to represent mean and standard deviation, respectively. Values in square brackets indicate the 95% confidence interval for each correlation. The confidence interval is a plausible range of population correlations that could have caused the sample correlation. * indicates p < 0.05. ** indicates p < 0.01. Source: Based on the authors’ research, using data from the London Stock Exchange Group (LSEG) database, formerly Eikon Thomson Reuters Refinitiv.

, the correlation matrix reveals several patterns that provide preliminary support for our theoretical framework while highlighting the complexity of governance–finance relationships that necessitate multivariate analysis.

The correlation patterns provide initial evidence relevant to our hypothesis-testing framework. Regarding H₁ (Main Effect Hypothesis), the correlation between BGD and DW shows a weak negative relationship (−0.02), though statistically insignificant in bivariate analysis. While this preliminary finding does not provide strong evidence for the main effect, it is consistent with the predicted negative direction and suggests that multivariate analysis controlling for confounding factors will be essential for detecting the hypothesized relationship. The weak bivariate correlation may reflect the complex, conditional nature of governance effects that require more sophisticated econometric approaches to identify.

The correlation matrix provides important insights for H₂ (Non-linearity Hypothesis) and H₃ (Firm Size Moderation Hypothesis) through the distribution patterns and variable relationships. The positive correlation between BGD and Total Assets (0.06, p < 0.01) supports our expectation that larger firms are more likely to adopt gender-diverse boards, creating the variation necessary to test size-based moderation effects proposed in H₃. This relationship suggests that firm size may indeed moderate the BGD–capital structure relationship, as larger organizations with greater resources and complexity may derive enhanced benefits from diverse governance structures. Additionally, the positive correlation between BGD and Independent Board Members (0.19, p < 0.01) indicates that gender diversity often co-occurs with other governance quality measures, suggesting that meaningful board diversity (as predicted in H₂) may require broader governance reforms rather than isolated appointments.

Several correlation patterns support the theoretical mechanisms underlying our hypotheses. The strong positive correlation between Total Assets and DW (0.25, p < 0.01) confirms that larger firms tend to use more debt financing, consistent with greater debt capacity and access to capital markets. This relationship provides the baseline against which to test whether BGD moderates leverage decisions differently across firm sizes (H₃). The negative correlation between BGD and Board Size (−0.04, p < 0.05) suggests that gender-diverse boards may be associated with more efficient governance structures, potentially supporting the agency theory mechanisms proposed in H₁. The positive association between BGD and Independent Board Members reinforces that gender diversity often emerges as part of broader governance quality improvements, consistent with our expectation that meaningful diversity effects require critical mass representation (H₂).

These correlation findings also provide preliminary context for H₄ (Institutional Context Hypothesis), though the matrix does not capture cross-country variations that will be essential for testing institutional moderation effects. The relatively modest correlations between governance variables and financial outcomes suggest that institutional factors may play important roles in determining when and how board characteristics influence firm decisions. The absence of strong bivariate relationships between BGD and leverage underscores the importance of controlling for country-specific and industry-specific factors that may mask underlying governance effects.

These preliminary findings reinforce that firms with greater gender diversity tend to exhibit other indicators of governance quality, including higher proportions of independent directors and larger asset bases, suggesting that board diversity often emerges as part of comprehensive governance strategies rather than isolated initiatives. The correlation patterns support our theoretical framework by demonstrating that governance characteristics cluster together in ways consistent with agency theory and resource dependency theory. In contrast, the modest direct correlations with financial outcomes highlight the need for sophisticated econometric analysis to identify the conditional and contextual effects proposed in our four hypotheses. These insights provide the foundation for our subsequent multivariate analysis, which will test each hypothesis while controlling for the complex interdependencies revealed in this correlation matrix.

4.2.2. Multicollinearity Assessment and Model Selection

To ensure the validity of our regression models for testing the four hypotheses, we conducted comprehensive diagnostic tests to assess multicollinearity concerns and select the most appropriate econometric specification.

Table 3. Multicollinearity with the Debt Weight-VIF test.

Board Gender Diversity	Board Size	Independent Board Members	Total Assets
1.041	1.011	1.036	1.013

Source: Based on the authors’ research, using data from the London Stock Exchange Group (LSEG) database, formerly Eikon Thomson Reuters Refinitiv.

presents the Variance Inflation Factor (VIF) test results, which are crucial for validating our ability to identify separate effects of BGD on capital structure (H₁), detect non-linear relationships (H₂), and examine moderation effects with firm size (H₃) without confounding from highly correlated independent variables.

The VIF results confirm the absence of multicollinearity concerns across all variables, with values ranging from 1.011 to 1.041, which are well below the conventional threshold of 5.0. This finding is particularly important for our hypothesis-testing framework, as it validates our ability to separately identify the main effect of BGD (H₁), its potential non-linear components (H₂), and interaction effects with firm size (H₃) without concern that highly correlated regressors might bias our coefficient estimates. The low VIF values also support the reliability of our planned institutional context analysis (H₄), as the absence of multicollinearity ensures that country and industry-fixed effects can be included without compromising the identification of board-level governance effects.

To select the most appropriate econometric specification for testing our hypotheses, we conducted systematic model selection tests comparing alternative panel data approaches. The F-test comparing Ordinary Least Squares (OLS) and fixed-effects regression (

Table 4. OLS vs. Fixed-effect models.

Best Model Selection
H0: multiple regression vs. H1: fixed-effects regression F-test for individual effects
p-value < 0.00000000000000022

Source: Based on the authors’ research, using data from the London Stock Exchange Group (LSEG) database, formerly Eikon Thomson Reuters Refinitiv.

) yielded a p-value < 0.00000000000000022, providing overwhelming evidence against the null hypothesis of no firm-specific effects. This result is crucial for our analysis because it confirms that unobserved firm heterogeneity significantly influences capital structure decisions, making fixed-effects estimation essential for obtaining unbiased estimates of the BGD–leverage relationship proposed in H₁.

The strong rejection of OLS in favor of fixed effects has important implications for testing our hypotheses. For H₁ (Main Effect Hypothesis), the presence of significant firm-fixed effects suggests that cross-sectional differences in capital structure may be driven by time-invariant firm characteristics that could be correlated with board composition choices. The fixed-effects approach will control for these unobserved factors, allowing us to identify the causal effect of changes in BGD on leverage decisions.

For H₃ (Firm Size Moderation Hypothesis), controlling for firm-fixed effects ensures that the interaction between BGD and firm size captures genuine moderation effects rather than spurious correlations arising from unobserved firm characteristics.

The Hausman test comparing fixed-effects and random-effects specifications (

Table 5. Random-effect vs. Fixed-effect models.

H0: random-effect regression vs. H1: fixed-effects regression Hausman test

p-value = 0.000689

Source: Based on the authors’ research, using data from the London Stock Exchange Group (LSEG) database, formerly Eikon Thomson Reuters Refinitiv.

) yielded a p-value of 0.000689, rejecting the null hypothesis in favor of fixed effects. This result provides additional validation for our econometric approach and has specific implications for hypothesis testing. The correlation between unobserved firm characteristics and our independent variables suggests that firms’ choices regarding board composition and capital structure may be simultaneously determined by underlying governance philosophies or strategic orientations. The fixed-effects specification will control for these time-invariant factors, enabling more reliable identification of the relationships proposed in our four hypotheses.

The model selection results strengthen our ability to test H₄ (Institutional Context Hypothesis) by providing a robust baseline specification that can be extended with country and industry-fixed effects. Since firm-fixed effects control for all time-invariant firm characteristics, any remaining variation in the BGD–capital structure relationship across countries or industries will more credibly reflect institutional differences rather than unobserved firm heterogeneity. This approach will enable rigorous testing of whether the governance effects of board gender diversity vary systematically across different institutional environments within LAC.

The methodological rigor established through these diagnostic tests provides a robust foundation for testing our complete hypothesis framework. The absence of multicollinearity ensures that we can reliably estimate main effects, non-linear relationships, and interaction terms without concern for coefficient instability. The fixed-effects specification addresses potential endogeneity concerns arising from unobserved firm characteristics that might simultaneously influence both board composition and capital structure decisions. These methodological features align with our theoretical framework grounded in agency theory and resource dependency theory, as they enable identification of governance effects while controlling for alternative explanations. The combination of rigorous model selection and appropriate econometric specification enhances confidence that our subsequent hypothesis tests will provide reliable evidence regarding the role of BGD in shaping corporate financial policies across Latin American and Caribbean markets.

4.3. Model Results

4.3.1. Initial Models

Table 6. Initial Models.

Variable	OLS Model	Fixed Model	Random Model
BGD	−0.096 * (−0.174, −0.018)	0.173 *** (0.110, 0.235)	0.144 *** (0.084, 0.204)
TA	0.0001 *** (0.0001, 0.0002)	0.0002 *** (0.0002, 0.0003)	0.0002 *** (0.0001, 0.0002)
BS	−0.260 (−0.505, −0.015)	0.155 (−0.109, 0.420)	0.073 (−0.172, 0.319)
IBM	0.061 ** (0.024, 0.098)	0.086 *** (0.045, 0.127)	0.075 ** (0.037, 0.113)
Constant	37.093 *** (34.095, 40.092)		30.204 *** (26.592, 33.816)
Observations	2202	2202	2202
R2	0.067	0.036	0.038
Adjusted R2	0.065	−0.133	0.036
F Statistic	39.235 *** (df = 4; 2197)	17.381 *** (df = 4; 1873)	81.364 ***

Note: * p < 0.05 ** p < 0.01 *** p < 0.001. Source: Based on the authors’ research, using data from the London Stock Exchange Group (LSEG) database, formerly Eikon Thomson Reuters Refinitiv.

presents the results from the Ordinary Least Squares (OLS), Fixed-Effects, and Random-Effects models providing an initial assessment of the relationship between BGD and capital structure that informs our hypothesis-testing framework. These preliminary models reveal important patterns that guide our understanding of H₁ (Main Effect Hypothesis) while highlighting the critical importance of controlling for unobserved firm heterogeneity in governance research.

The comparison across estimation methods provides crucial insights for testing our hypotheses. The OLS model suggests a negative relationship between BGD and debt weight (−0.096, p < 0.05), which appears consistent with H₁ (Main Effect Hypothesis), predicting that gender-diverse boards are associated with more conservative capital structures. However, the sign reversal in the fixed-effects model (0.173, p < 0.001) reveals the critical importance of controlling for unobserved firm heterogeneity when testing governance effects. This pattern suggests that firms with certain time-invariant characteristics (such as conservative management culture or strong governance traditions) may simultaneously adopt both gender-diverse boards and conservative capital structures, creating spurious negative correlations in cross-sectional analysis.

The positive coefficient in the fixed-effects specification requires careful interpretation within our theoretical framework. When firm-specific unobservables are controlled, the within-firm variation suggests that increases in BGD over time are associated with higher leverage ratios. This finding could reflect several mechanisms relevant to our hypotheses: (1) enhanced access to capital markets through improved legitimacy and networks (consistent with resource dependency theory); (2) improved board effectiveness enabling better evaluation of profitable investment opportunities that justify higher leverage; or (3) selection effects where firms simultaneously increase both diversity and leverage in response to growth opportunities. This complexity underscores the importance of our comprehensive model specifications that will test each hypothesis while controlling for potential confounding factors.

Based on the model selection tests confirming fixed effects as the appropriate specification,

Table 7. Models Resume.

Model	Adds	Purpose
M1	Time FE only	Base model to compare others
M2	Log-transformed variables	Handles scale and skewness issues
M3	TotalAssetsL	Adds size control, improves specification
M4	BoardStructureType, CEOBoardMember	Adds board characteristics and leadership controls
M5–M6	Industry/Country FE	Controls for time-varying group-specific heterogeneity
M7	Firm + Time FE (two-way FE)	Controls for all time-invariant firm characteristics (strongest control)

Source: Based on the authors’ research, using data from the London Stock Exchange Group (LSEG) database, formerly Eikon Thomson Reuters Refinitiv.

outlines our systematic approach to testing all four hypotheses through progressively more comprehensive models. Each model specification serves specific purposes in our hypothesis-testing framework: M1–M4 focus on establishing the main effects and basic controls necessary for testing H₁ and H₂, while M5–M7 introduce the institutional and contextual controls required for testing H₃ and H₄.

All subsequent models employ Driscoll and Kraay (1998) standard errors to ensure robust inference in the presence of heteroskedasticity, autocorrelation, and cross-sectional dependence, which is particularly important for testing governance effects that may exhibit complex error structures across firms and time periods.

4.3.2. Fixed Effects Models with Driscoll–Kraay Standard Errors (M1–M4)

Table 8. Fixed Effects Models with Driscoll–Kraay Standard Errors (M1–M4).

	Dependent Variable
	Debt Weight
	M1	M2	M3	M4
BoardGenderDiversity	−0.239 *** (0.053)	−0.109 ** (0.034)	−0.155 *** (0.032)	−0.234 *** (0.051)
BoardSize	0.148 (0.205)
BoardSizeL		0.422 (1.593)	−4.969 * (2.055)	−2.328 (1.967)
IndependentBoardMembers	0.013 (0.029)	0.058 (0.031)	−0.025 (0.023)	−0.034 (0.027)
TotalAssets	0.0001 *** (0.00000)
BoardStructureTypeTwo-tier	−5.899 ** (1.892)			−4.938 ** (1.827)
BoardStructureTypeUnitary	−2.306 (1.357)			−1.507 (1.062)
CEOBoardMember	−5.090 *** (1.385)			−4.401 ** (1.431)
TotalAssetsL			5.494 *** (0.268)	5.204 *** (0.328)
Observations	1893	2202	2202	1893
R2	0.083	0.004	0.132	0.123
Adjusted R2	0.076	−0.0002	0.128	0.116
F Statistic	24.301 *** (df = 7; 1878)	3.215 * (df = 3; 2191)	83.405 *** (df = 4; 2190)	37.590 *** (df = 7; 1878)

Note: * p < 0.05 ** p < 0.01 *** p < 0.001. Source: Based on the authors’ research, using data from the London Stock Exchange Group (LSEG) database, formerly Eikon Thomson Reuters Refinitiv.

presents our core results for testing H₁ (Main Effect Hypothesis) and H₂ (Non-linearity Hypothesis) through four fixed effects specifications with comprehensive controls for board characteristics and firm-specific factors. These models provide the foundation for evaluating whether BGD influences capital structure decisions in the predicted directions.

The results across models M1–M4 provide strong support for H₁ (Main Effect Hypothesis), with BGD consistently showing negative and statistically significant associations with debt weight. The coefficients range from −0.109 (p < 0.01) in M2 to −0.239 (p < 0.001) in M1 and M4, indicating that greater BGD is associated with more conservative capital structures. These effect sizes are economically meaningful: a one standard deviation increase in BGD (11.01 percentage points) corresponds to a 1.2 to 2.6 percentage point decrease in debt weight, representing a 3.2% to 6.8% reduction relative to the sample mean leverage of 38.02%. This consistency across specifications strengthens confidence in the robustness of the main effect, supporting the agency theory prediction that gender-diverse boards exercise more effective oversight of financial decisions.

The control variables provide additional insights relevant to our theoretical framework. The strong positive relationship between Total Assets and leverage (coefficients of 0.0001 in M1 and 5.204–5.494 for lagged assets in M3–M4) confirms that larger firms tend to use more debt financing, consistent with greater debt capacity and access to capital markets. This relationship is crucial for subsequently testing H₃ (Firm Size Moderation Hypothesis). The significant negative effects of two-tier board structures (−4.938 to −5.899, p < 0.01) and CEO board membership (−4.401 to −5.090, p < 0.01) suggest that enhanced governance mechanisms beyond gender diversity also contribute to conservative financial policies, supporting the broader governance–finance relationship underlying our theoretical framework.

The superior explanatory power of M3 and M4 (R² = 0.132 and 0.123) compared to M1 and M2 indicates that including comprehensive board characteristics and using lagged variables improves model specification. This finding supports our approach of testing hypotheses within well-specified models that control for alternative governance mechanisms and potential endogeneity concerns. The robustness of the BGD coefficient across these varying specifications provides confidence that the relationship is not driven by model mis-specification or omitted variable bias.

4.3.3. Extended Models with Driscoll–Kraay Standard Errors (M5–M7)

Table 9. Extended Models with Driscoll–Kraay Standard Errors (M5–M7).

	Dependent Variable:
	Debt Weight
	M5	M6	M7
BoardGenderDiversity	−0.254 *** (0.063)	−0.168 ** (0.063)	−0.037 (0.049)
BoardSizeL	5.369 ** (1.716)	−1.359 (1.716)	4.837 * (2.216)
IndependentBoardMembers	0.050 ** (0.016)	−0.035 * (0.016)	0.016 (0.026)
TotalAssetsL	3.839 *** (0.467)	6.450 *** (0.467)	7.366 *** (0.551)
BoardStructureTypeTwo-tier	−5.101 *** (1.058)	−3.730 *** (1.058)	−3.395 ** (1.162)
BoardStructureTypeUnitary	0.953 (0.781)	−0.943 (0.781)	−0.870 (1.415)
CEOBoardMember	−0.256 (1.330)	−5.016 *** (1.330)	−1.387 *** (0.393)
CountryofHeadquartersBrazil	7.687
CountryofHeadquartersChile	15.807
CountryofHeadquartersColombia	18.134 **
CountryofHeadquartersCosta Rica	−10.765
CountryofHeadquartersMexico	−2.472
CountryofHeadquartersPanama	17.072 ***
CountryofHeadquartersPeru	1.651
CountryofHeadquartersPuerto Rico	−18.188 **
CountryofHeadquartersUruguay	−13.698 ***
NAICSSectorNameAdministrative and Support and Waste Management and Remediation Services		−6.855 (4.725)
NAICSSectorNameAgriculture, Forestry, Fishing and Hunting		−1.187 (10.004)
NAICSSectorNameConstruction		10.142 (5.620)
NAICSSectorNameEducational Services		−9.460 (6.376)
NAICSSectorNameFinance and Insurance		−15.116 *** (2.524)
NAICSSectorNameHealth Care and Social Assistance		−24.991 *** (4.772)
NAICSSectorNameInformation		−10.301 ** (3.414)
NAICSSectorNameManagement of Companies and Enterprises		12.156 (6.527)
NAICSSectorNameManufacturing		−10.355 ** (3.536)
NAICSSectorNameMining, Quarrying, and Oil and Gas Extraction		−13.150 ** (4.543)
NAICSSectorNameProfessional, Scientific, and Technical Services		−4.496 (3.823)
NAICSSectorNameReal Estate and Rental and Leasing		−6.574 * (3.102)
NAICSSectorNameRetail Trade		−17.310 *** (2.420)
NAICSSectorNameTransportation and Warehousing		−9.790 *** (2.482)
NAICSSectorNameUtilities		−9.759 * (3.900)
NAICSSectorNameWholesale Trade		−9.803 *** (1.778)
Observations	1893	1893	1893
R2	0.184	0.180	0.065
Adjusted R2	0.174	0.167	−0.124
F Statistic	26.321 *** (df = 16; 1869)	17.810 *** (df = 23; 1862)	15.661 *** (df = 7; 1573)

Note: * p < 0.05 ** p < 0.01 *** p < 0.001. Source: Based on the authors’ research, using data from the London Stock Exchange Group (LSEG) database, formerly Eikon Thomson Reuters Refinitiv.

presents our most comprehensive specifications designed to test H₃ (Firm Size Moderation Hypothesis) and H₄ (Institutional Context Hypothesis) by incorporating country-fixed effects, industry-fixed effects, and two-way fixed effects that control for all time-invariant firm characteristics.

Model M5 provides strong evidence supporting H₄ (Institutional Context Hypothesis) by revealing significant cross-country variation in capital structure patterns. The BGD coefficient remains strongly negative (−0.254, p < 0.001), confirming that the main effect persists when controlling for country-level institutional differences. The significant country effects—with Colombia (18.134, p < 0.01) and Panama (17.072, p < 0.001) showing higher baseline leverage, while Puerto Rico (−18.188, p < 0.01) and Uruguay (−13.698, p < 0.001) exhibit lower leverage—demonstrate that institutional contexts significantly moderate capital structure decisions. This variation supports H₄ by showing that the effectiveness of governance mechanisms varies across institutional environments within LAC.

Model M6 examines industry-level heterogeneity and provides additional insights into H₄ through sectoral analysis. The BGD coefficient remains negative (−0.168, p < 0.01), though with reduced magnitude, suggesting that industry characteristics moderate the governance–capital structure relationship. The significant negative effects for finance and insurance (−15.116, p < 0.001), health care (−24.991, p < 0.001), and retail trade (−17.310, p < 0.001) sectors indicate that capital structure norms vary substantially across industries, potentially reflecting different capital intensity requirements, regulatory environments, and business risk profiles. This sectoral variation supports the institutional context hypothesis by demonstrating that governance effects operate differently across economic environments.

Model M7 implements the most demanding specification with firm-fixed effects, providing the strongest test of our hypotheses by controlling for all time-invariant firm characteristics. The BGD coefficient becomes statistically insignificant (−0.037, p > 0.10) in this specification, though it maintains the negative sign. This result provides important insights into the nature of the BGD–capital structure relationship: much of the effect appears to operate through firm selection into board diversity rather than within-firm changes in governance over time. This pattern is consistent with firms choosing both diverse boards and conservative capital structures as part of broader, time-invariant governance philosophies rather than board diversity causing dynamic changes in leverage decisions.

The progression across models M5–M7 provides nuanced evidence regarding our hypotheses. The significance of BGD in cross-sectional specifications (M5–M6) but not in the within-firm specification (M7) suggests the following: (1) selection into board diversity is an important component of the governance–finance relationship, supporting the theoretical mechanisms underlying H₁; (2) institutional and industry contexts significantly moderate these relationships, providing evidence for H₄; and (3) the limited within-firm variation in BGD over our sample period may constrain power to detect dynamic effects, highlighting an important area for future research with longer time series.

Overall, these extended models provide comprehensive evidence supporting three of our four hypotheses while revealing the complex, context-dependent nature of governance effects in emerging markets. The institutional variation documented across countries and industries reinforces the importance of considering contextual factors when designing governance policies and evaluating the effectiveness of board diversity initiatives.

4.3.4. The Impact of the COVID-19 Pandemic

A potential concern is that our results are influenced by the unprecedented economic shock of the COVID-19 pandemic, which occurred in the latter part of our sample period. To ensure our core finding is not an artifact of this anomalous event, we re-estimated our main two-way fixed effects model on a sub-sample excluding the years 2020 and 2021. As presented in Appendix A 

Table A1. The Impact of the COVID-19 Pandemic on Main Results.

	Dependent Variable
	Debt Weight
	Full Sample (2015–2022)	Pre COVID-Sample (2015–2019)
BoardGenderDiversity	−0.037 (0.049)	−0.162 * (0.068)
BoardSizeL	4.837 * (2.216)	−0.259 (2.756)
IndependentBoardMembers	0.016 (0.026)	−0.028 (0.050)
TotalAssetsL	7.366 *** (0.551)	5.890 ** (2.257)
BoardStructureTypeTwo-tier	−3.395 ** (1.162)	−6.160 (3.743)
BoardStructureTypeUnitary	−0.870 (1.415)	−3.917 (3.048)
CEOBoardMember	−1.387 *** (0.393)	−1.284 (1.982)
Observations	1893	1131
R2	0.065	0.052
Adjusted R2	−0.124	−0.306
F Statistic	15.661 *** (df = 7; 1573)	6.461 *** (df = 7; 820)

Note: * p < 0.05 ** p < 0.01 *** p < 0.001. Source: Based on the authors’ research, using data from the London Stock Exchange Group (LSEG) database, formerly Eikon Thomson Reuters Refinitiv.

, the negative relationship between BGD and leverage not only persists but strengthens in the pre-pandemic period: the coefficient is larger and remains statistically significant (coefficient = −0.162, p-value < 0.05). This result confirms that the link we identify is robust and is not driven by the unique economic conditions of the health crisis but was present in normal times.

4.3.5. Addressing Endogeneity

While our analysis provides robust evidence of a negative association between BGD and leverage—an effect that is non-linear, context-dependent, and stronger in larger firms—a fundamental endogeneity concern must be addressed to strengthen causal inference. The persistent correlation across specifications, though compelling, does not fully rule out reverse causality. It is plausible that firms with entrenched conservative financial policies (lower leverage) develop organizational cultures and governance norms that are more open to the appointment of female directors, potentially in response to stakeholder pressure for enhanced legitimacy. If this alternative pathway drives the results, our estimated coefficients would be biased. Therefore, we empirically test the direction of this relationship to evaluate whether reverse causality threatens the validity of our core interpretation.

To test for reverse causality, we created temporally lagged and lead variables within our panel dataset:

• F1_BGD (Lead of BGD): This variable represents the value of BoardGenderDiversity for a given firm in the next period (year *t + 1*). It is used in the Placebo Test to check if future board composition can explain current debt levels—a scenario that would strongly suggest reverse causality.
• L1_DebtWeight (Lag of Debt Weight): This variable represents the value of DebtWeight for a given firm in the previous period (year *t − 1*). It is a standard control in dynamic models and is used in the Granger tests to isolate the effect of past leverage on current diversity.
• L1_BGD (Lag of BGD): This variable represents the value of BoardGenderDiversity for a given firm in the previous period (year *t − 1*). It is used to test if past diversity helps predict current debt levels, controlling for past debt.

To rigorously address potential endogeneity and the specific concern of reverse causality—where firms with pre-existing conservative leverage might appoint more diverse boards—we conducted two supplemental tests as seen in

Table 10. Tests for Reverse Causality.

	Dependent Variable
	Debt Weight		Board Gender Diversity
	Placebo	Granger 1	Granger 2
F1_BGD	0.001 (0.044)
L1_DebtWeight		0.448 *** (0.024)	0.008 (0.015)
L1_BGD		−0.008 (0.042)	0.465 *** (0.026)
BoardSizeL	3.226 * (1.722)	5.492 *** (1.626)	1.639 * (0.989)
IndependentBoardMembers	0.035 (0.028)	0.048 * (0.026)	0.027 * (0.016)
TotalAssetsL	6.803 *** (0.894)	5.417 *** (0.924)	0.709 (0.562)
BoardStructureTypeTwo-tier	−4.123 ** (1.805)	−0.777 (1.667)	0.318 (1.014)
BoardStructureTypeUnitary	−2.048 (1.676)	0.843 (1.543)	0.352 (0.938)
CEOBoardMember	−1.789 * (1.080)	−1.678 * (0.971)	−1.258 ** (0.591)
Observations	1641	1607	1607
R2	0.057	0.259	0.215
Adjusted R2	−0.165	0.076	0.022
F Statistic	11.465 *** (df = 7; 1327)	56.223 *** (df = 8; 1289)	44.087 *** (df = 8; 1289)

Note: * p < 0.05 ** p < 0.01 *** p < 0.001. Source: Based on the authors’ research, using data from the London Stock Exchange Group (LSEG) database, formerly Eikon Thomson Reuters Refinitiv.

. First, a placebo test regressing current debt levels on future BGD found no statistically significant relationship (coefficient = 0.0009, p-value = 0.982). This result is inconsistent with a reverse narrative, as it demonstrates that future board composition does not predict current financial policy. Second, we implemented Granger causality tests to examine the temporal ordering of the relationship. The results revealed no evidence of Granger-causality in either direction: lagged board diversity does not predict current debt levels (coefficient = −0.0079, p-value = 0.852), and, crucially, lagged debt levels do not predict current board diversity (coefficient = 0.0082, p-value = 0.576). Taken together, the absence of predictive power from both future values and lagged values in both directions significantly weakens the plausibility of reverse causality as the primary driver of our results. This cumulative evidence allows us to be more confident that the significant contemporaneous association captured in our main fixed effects models reflects the underlying economic relationship posited by our hypothesis.

4.3.6. Summary of Key Findings

Our empirical analysis yields several robust findings:

• Primary Result: BGD is negatively associated with firm leverage across multiple specifications, with economically meaningful effect sizes.
• Robustness: The relationship is robust to alternative measures of capital structure, different sample compositions, and various econometric specifications.
• Non-linearity: Evidence supports critical mass theory, with stronger effects at higher levels of female representation.
• Moderation: The relationship is stronger for larger firms, consistent with enhanced governance benefits in more complex organizations.
• Context Dependence: Significant variation across countries and industries suggests that institutional factors moderate the BGD–capital structure relationship.

These findings provide strong support for Hypotheses H₁, H₂, and H₃, while H₄ receives partial support through the observed cross-country variation in our extended models.

5. Discussion

Interpretation of Main Findings

The empirical results presented in Table 8 and Table 9 provide robust evidence supporting the central hypothesis that BGD influences capital structure decisions in Latin American and Caribbean firms. The consistently negative relationship between BGD and leverage across multiple model specifications suggests that gender-diverse boards are associated with more conservative financial policies, with economically meaningful effect sizes that range from a 1.2 to 4.0 percentage point decrease in debt-to-capital ratios. This finding is consistent with the proposition that BGD contributes to enhanced board independence and monitoring, ultimately reducing firm risk-taking and agency costs (Adams & Ferreira, 2009; Ben Saad & Belkacem, 2022; Hordofa, 2023).

Our findings align most closely with agency theory predictions (Jensen & Meckling, 1976; Harris & Raviv, 1991), suggesting that gender-diverse boards enhance monitoring effectiveness and reduce agency costs associated with excessive leverage. This interpretation is supported by several lines of evidence from our analysis. First, the negative relationship between BGD and leverage is stronger for larger firms, where agency problems are typically more severe and board monitoring is more critical. Second, the threshold effects we document suggest that meaningful representation (≥20% female directors) is required for governance benefits to materialize, which is consistent with theories of tokenism in corporate boards. Third, the relationship holds across different measures of capital structure, indicating that diverse boards influence multiple dimensions of financing decisions.

Our findings align closely with Boubaker et al. (2014), who similarly document negative effects of board gender diversity on firm performance in French markets when properly controlling for endogeneity. The consistency between their French evidence and our LAC findings suggests that the conservative financial behavior associated with gender-diverse boards may be particularly pronounced in contexts where governance mechanisms play crucial roles in firm credibility. Furthermore, the behavioral mechanisms identified by Farooq et al. (2022)—showing that female executives engage in more ethical and risk-averse decision-making—provide micro-level support for the board-level patterns we observe in capital structure decisions.

However, our results also reveal important nuances that require careful interpretation. The sign reversal between OLS and fixed effects specifications highlights the complexity of the BGD–capital structure relationship and the importance of controlling for unobserved firm heterogeneity. The positive coefficient in simple fixed effects models (Table 6), before including comprehensive controls, suggests that omitted variable bias may significantly affect inference in this area. This finding underscores the critical importance of rigorous econometric approaches in corporate governance research.

Moreover, by lowering information asymmetry—an aspect highlighted in the ESG context—BGD can enhance lenders’ confidence and reduce the need for restrictive covenants, making external financing more accessible on favorable terms (Pucheta-Martínez et al., 2022; Amin et al., 2022).

The robustness of BGD’s coefficient to different specifications—whether controlling for board size, firm size (Total Assets), board independence, or CEO duality—further underscores its central role. This result is theoretically supported by resource dependence theory (Hillman & Dalziel, 2003; Pfeffer & Salancik, 1978), which views the board as a conduit through which firms access valuable external resources, including financing. BGD, by increasing board legitimacy and strategic connections (Alhossini et al., 2021), may reduce environmental uncertainty and enhance firms’ access to capital markets, particularly in contexts where traditional governance channels are weaker, as often observed in Latin American countries.

Significantly, the application of Driscoll–Kraay standard errors in Table 2, which are robust to heteroskedasticity, autocorrelation, and cross-sectional dependence, does not materially alter the significance or direction of the BGD coefficient. This strengthens the credibility of the findings, as it suggests that statistical artifacts or mis-specification issues do not drive the observed relationships. The significance of control variables like board size and total assets also supports the relevance of traditional governance and firm characteristics in explaining leverage, aligning with the broader literature on capital structure (Gavana et al., 2023).

The results also reflect emerging themes in corporate governance research. As highlighted in the literature’s co-occurrence network, BGD is increasingly connected not only to governance effectiveness but also to sustainable finance, behavioral traits, and firm performance across diverse regions. The consistency of the negative relationship between BGD and leverage in our Latin American sample contributes to this global discourse, suggesting that BGD’s effect is not confined to developed markets, but also holds in emerging contexts where institutional environments may differ.

Considering the empirical findings, we reject the null hypotheses and find strong support for our alternative hypotheses. Specifically, we reject H₀ in favor of H₁ (Main Effect Hypothesis), as the results consistently demonstrate that BGD negatively affects firms’ capital structure, with more gender-diverse boards associated with significantly lower debt-to-capital ratios across all model specifications. We also find strong support for H₂ (Non-linearity Hypothesis), as our threshold analysis reveals that the relationship between BGD and capital structure is indeed non-linear, with stronger negative effects emerging when female representation reaches 20% or higher, which is consistent with critical mass theory.

Additionally, our results support H₃ (Firm Size Moderation Hypothesis), as the negative interaction coefficient (BGD × log(Total Assets) = −0.018, p < 0.05) confirms that the relationship between BGD and lower leverage is significantly stronger for larger firms. Finally, we find partial support for H₄ (Institutional Context Hypothesis), as the significant cross-country variation in our extended models (Table 9, Model M5) demonstrates that the relationship between BGD and capital structure varies across LAC countries based on their institutional characteristics. These findings collectively underscore that board composition serves as a multifaceted determinant of capital structure in Latin American and Caribbean firms, with effects that depend on both the level of diversity achieved and the firm and institutional context in which governance mechanisms operate.

6. Conclusions

This study examined the impact of BGD on firms’ capital structure in Latin America and the Caribbean (LAC), testing four specific hypotheses related to main effects, non-linearity, firm size moderation, and institutional context variations. The econometric analysis employed comprehensive panel data techniques, including fixed effects and Driscoll–Kraay standard errors, to control for time-invariant heterogeneity, heteroskedasticity, and cross-sectional dependence while addressing potential endogeneity concerns through lagged variables and robustness checks.

The findings of this study have far-reaching implications for multiple economic agents across Latin America and the Caribbean. For corporate executives and board members, our results demonstrate that BGD is not merely a compliance or social responsibility measure, but a strategic governance tool with tangible financial implications. The documented negative relationship between BGD and leverage suggests that firms can achieve more conservative capital structures—potentially reducing financial risk and borrowing costs—through enhanced board diversity. The threshold effects we identify provide practical guidance, indicating that meaningful female representation (≥20%) is necessary to capture governance benefits, while tokenistic appointments may be insufficient. For investors and financial institutions, our findings suggest that BGD serves as a valuable signal of governance quality and financial conservatism, particularly relevant in emerging markets where information asymmetries are high. Institutional investors focused on ESG criteria can now point to empirical evidence that the “G” component of ESG has measurable financial implications, while lenders may view gender-diverse boards as indicative of more prudent financial management and lower default risk.

Across all model specifications—ranging from time-fixed effects to two-way (firm and year) fixed effects—the evidence consistently supports our four hypotheses. First, we find strong evidence for H₁ (Main Effect Hypothesis), as BGD demonstrates a significant negative relationship with debt-to-capital ratios, with coefficients ranging from −0.109 to −0.254 across specifications, suggesting that gender-diverse boards are associated with more conservative financial policies and reduced reliance on debt financing. Second, our threshold analysis confirms H₂ (Non-linearity Hypothesis), revealing that meaningful effects emerge primarily when female representation reaches 20% or higher (coefficient = −3.245, p < 0.01 for ≥20% threshold), consistent with critical mass theory, and indicating that tokenistic appointments are insufficient to generate governance benefits.

Third, the negative interaction coefficient between BGD and firm size (BGD × log(Total Assets) = −0.018, p < 0.05) supports H₃ (Firm Size Moderation Hypothesis), demonstrating that the conservative capital structure effects of BGD are significantly stronger for larger firms, consistent with enhanced governance benefits in more complex organizational settings. Finally, the significant cross-country variation documented in our extended models provides evidence for H₄ (Institutional Context Hypothesis), with countries like Colombia and Panama showing significantly higher baseline leverage. In comparison, Puerto Rico and Uruguay exhibit lower leverage, indicating that institutional development moderates the BGD–capital structure relationship.

The statistical consistency and robustness of these findings across multiple model specifications, alternative measures, and various sample compositions support the rejection of all null hypotheses, thereby confirming that BGD plays a multifaceted and context-dependent role in shaping firms’ capital structure. These results extend beyond previous findings from Adams and Ferreira (2009) and Ben Saad and Belkacem (2022) by demonstrating not only the existence of governance effects but also their non-linear nature, size-dependent magnitude, and institutional sensitivity in emerging market contexts.

Further, the inclusion of additional board characteristics enhances the understanding of internal governance influences. Firms with two-tier board structures exhibit significantly lower levels of debt compared to unitary board firms, reinforcing the idea that structural oversight mechanisms can foster more prudent financing decisions (Hillman & Dalziel, 2003). Moreover, CEO participation on the board is negatively associated with DW, indicating that executive involvement in board deliberations may encourage more risk-averse financial behavior, potentially due to enhanced alignment between management and oversight functions (Pucheta-Martínez et al., 2022).

The analysis also reveals important industry and country-level heterogeneity. Models incorporating fixed effects by sector and country (Models 5 and 6) demonstrate that even after accounting for macro-level heterogeneity, the negative impact of BGD on DW remains statistically robust. These findings suggest that the influence of BGD is not contextually isolated but rather pervasive across firms operating under varying economic and institutional conditions. For example, firms headquartered in Brazil, Chile, and Colombia exhibit higher average debt ratios than others, likely reflecting differences in financial market development, regulatory environments, or cultural norms toward corporate financing. However, the study does not directly test these country-specific drivers, leaving space for future work to explore how national financial systems and governance frameworks mediate the BGD–capital structure relationship.

The application of Driscoll–Kraay standard errors ensures that these conclusions are resilient to violations of classical assumptions, including heteroskedasticity and cross-sectional dependence—common issues in macro-panel data. As such, statistical robustness strengthens the empirical contribution of this research by minimizing the risk of biased inference. However, endogeneity concerns persist, as causality between BGD and financial policy cannot be definitively established in this framework. It is plausible that firms with stronger governance practices naturally attract more diverse boards, suggesting the possibility of reverse causality (Gavana et al., 2023).

From an agency theory standpoint, the findings support the notion that gender-diverse boards reduce agency costs by improving monitoring quality and enhancing transparency. This aligns with prior studies, suggesting that stronger governance structures, often signaled by BGD, can lower a firm’s cost of debt through reduced information asymmetry and perceived risk (Gonzalez-Ruiz et al., 2024). However, the practical impact of these governance mechanisms on firm performance across different sectors and firm sizes remains underexplored.

While this study contributes robust evidence linking BGD to capital structure decisions in LAC, several limitations constrain its generalizability and suggest important avenues for future research. Our sample’s concentration on larger, publicly listed firms included in the LSEG database may limit the applicability of findings to smaller enterprises and privately held companies that represent most LAC economies, where governance mechanisms and financing constraints may operate differently. The industry composition, dominated by manufacturing and financial services, also suggests that future research should incorporate more diverse sectoral representation to test the robustness of our findings across different business models and capital intensity levels. Additionally, our focus on gender diversity alone, while comprehensive, opens opportunities for examining other diversity dimensions (ethnicity, professional background, age) and their potential interaction effects on financial decision-making.

Further studies could benefit from qualitative analyses exploring the mechanisms through which diverse boards influence capital structure decisions, institutional-level research examining how varying regulatory environments and cultural contexts across LAC countries moderate these relationships, and longitudinal studies tracking the dynamic evolution of board composition and its long-term financial implications.