The Impact of Corporate Governance on Financial Performance: The Mediating Role of Real Earnings Management

Abstract

This study examines the association between corporate governance and financial performance and investigates whether real earnings management (REM) mediates this relationship in an emerging-market context. Using a balanced panel of 434 nonfinancial listed firms in Vietnam from 2020 to 2024, yielding 2170 firm-year observations, the study employs feasible generalized least squares (FGLS) after diagnostic tests indicate heteroskedasticity and autocorrelation. The Durbin–Wu–Hausman test does not indicate significant endogeneity in the current model specification. REM is measured using the Roychowdhury-based approach, and mediation effects are examined through sequential regressions. Tobin’s Q is used for robustness testing, and a two-step System GMM is used as an additional robustness test. The results show that board size, institutional ownership, and state ownership are positively associated with financial performance, while board independence is negatively associated with performance. Board financial expertise has no significant direct relationship with performance. REM is negatively associated with financial performance and serves as a mediating channel in the governance–performance relationship. The study contributes to the corporate governance literature by showing that REM can transmit governance effects to firm performance in an emerging market characterized by evolving enforcement, state ownership, and potential gaps between formal and substantive governance mechanisms.

1. Introduction

Corporate governance has long attracted significant scholarly attention, especially in the wake of accounting and financial scandals and the growing public demand for corporate social responsibility. Effective corporate governance is widely regarded as vital for mitigating managerial misconduct, fostering fairness and transparency, and enhancing the credibility of disclosed information, thereby contributing to improved corporate performance (Asmaranti et al., 2024; Arora & Bodhanwala, 2018; Shivani et al., 2017). Agency theory posits that robust governance mechanisms can reduce agency costs, enhance the quality of managerial decision-making, and, consequently, improve financial performance while reducing earnings management practices (Jensen & Meckling, 1976).

Numerous studies have examined the relationships among corporate governance, earnings management, and firms’ financial performance (Khan et al., 2024; Kirimi et al., 2022; Kyere & Ausloos, 2020; Al-Duais et al., 2022; Hsu & Wen, 2015; Q. Nguyen et al., 2024; Habib, 2024; Lim & Mali, 2023). However, the empirical evidence remains mixed and sometimes inconsistent. Regarding the interaction among corporate governance, earnings management, and financial performance, most prior studies have highlighted the moderating or mediating role of corporate governance in the link between earnings management and financial results; however, they have mainly focused on accrual-based earnings management (Bojan et al., 2024). Additionally, evidence regarding the impact of REM on firms’ financial performance is inconclusive (Kang & Kim, 2011). These findings seem to depend on firm characteristics, institutional contexts, the choice of financial performance measures (e.g., return on assets [ROA], return on equity [ROE], and Tobin’s Q), and the specific models used to estimate REM (Kyere & Ausloos, 2020; Saidat et al., 2019). Furthermore, corporate governance factors—including board characteristics, ownership structures, firm-specific features, and institutional environments—have different effects on earnings management (Al-Duais et al., 2022), especially in emerging markets such as Vietnam (Tuan & Dung, 2024).

Vietnam offers an analytically relevant setting rather than merely a new empirical context. As an emerging market, Vietnam combines concentrated ownership, substantial state participation, evolving disclosure enforcement, and a corporate governance framework that is still moving from formal compliance toward substantive monitoring. These institutional features may alter the effectiveness of governance mechanisms. For example, independent directors may formally satisfy governance requirements yet lack sufficient access to internal information or enforcement power, while state ownership may operate not only as an ownership mechanism but also as a channel for political connections, resource access, and implicit support. In such a setting, REM becomes a particularly relevant transmission mechanism because managers may use real operating decisions to meet short-term performance targets while remaining less visible than accrual-based manipulation. Therefore, examining whether REM mediates the governance–performance relationship in Vietnam can refine existing governance theory by showing when formal governance mechanisms constrain, fail to constrain, or indirectly transmit managerial discretion into firm outcomes.

Accordingly, this study has two primary objectives: (i) to analyze the impact of corporate governance on the financial performance of firms in Vietnam, an emerging market where the corporate governance framework is still developing, and (ii) to investigate the mediating role of REM in the relationship between corporate governance and financial performance.

The results of this research make a significant contribution to the existing body of literature in multiple respects. First, from a theoretical standpoint, the study provides empirical evidence supporting the relationship between corporate governance and financial performance in an emerging economy such as Vietnam. Second, it clarifies how corporate governance influences REM and, consequently, financial performance, thereby offering further support for agency theory within an emerging market context. Third, regarding managerial implications, the study emphasizes that improving corporate governance not only directly enhances financial performance but also helps mitigate REM practices that could lead to long-term negative consequences for firms.

The remaining sections of the document are organized as follows: Section 2 outlines the theoretical framework and articulates the research hypotheses; Section 3 describes the research methodology; Section 4 presents the empirical findings; Section 5 offers concluding remarks, including implications and recommendations for future research; and Section 6 addresses limitations and prospective directions for future investigations.

2. Theoretical Framework and Research Hypotheses

2.1. Theoretical Framework

Corporate governance comprises the mechanisms that coordinate and control relationships among managers, the board of directors, shareholders, and other stakeholders. These mechanisms aim to balance interests, strengthen accountability, enhance transparency, and uphold ethical standards in corporate decision-making (Cadbury, 1992; Velentzas et al., 2017). Effective corporate governance provides the foundation for managerial oversight and strategic decision-making, thereby improving firm performance.

Financial performance is widely regarded as a core measure of a firm’s value-creation capacity. Beyond its close association with market value fluctuations, as reflected in stock prices, financial indicators are effective tools for forecasting a company’s operational performance and future growth prospects (Rizani et al., 2022). They capture a firm’s ability to generate returns on its resources and maintain overall financial health. It is commonly assessed using accounting- and market-based indicators, including profitability, operational efficiency, leverage, liquidity, and market valuation (Mohsin et al., 2020). In governance research, financial performance is often used to evaluate whether internal control and monitoring mechanisms deliver economic benefits to shareholders and other stakeholders.

Agency theory provides the primary explanation for the governance–performance relationship. It holds that agency problems arise when managers pursue private interests rather than maximizing shareholder wealth (Jensen & Meckling, 1976). Weak governance systems may enable managers to engage in earnings management to meet performance targets or secure incentive-based rewards (Hong et al., 2023). By contrast, strong governance mechanisms can reduce information asymmetry, enhance accountability, and limit opportunistic managerial behavior (Shleifer & Vishny, 1997; Bojan et al., 2024).

The board of directors is the primary internal governance body responsible for strategic direction and managerial oversight. Board characteristics are commonly assessed using measures such as board size, the proportion of independent directors, professional expertise, gender diversity, CEO duality, and board involvement in risk management. From an agency perspective, an effective board strengthens monitoring and reduces managerial discretion. From a resource dependence perspective, the board also provides advice, expertise, legitimacy, and access to external resources that support strategic decision-making and firm performance (Pfeffer & Salancik, 1978; Hillman & Dalziel, 2003).

Ownership structure reflects the distribution of ownership and control rights among different shareholder groups. It is a fundamental governance mechanism because shareholders differ in their monitoring incentives, access to information, investment horizons, and ability to influence management. Institutional shareholders may enhance governance effectiveness because they possess analytical resources and monitoring capacity, whereas state shareholders may affect firm performance through governance oversight and access to politically connected resources (Shleifer & Vishny, 1997; Abedin et al., 2022; Medina et al., 2022; Kubo & Phan, 2019).

REM refers to managerial actions that alter actual business operations to achieve short-term earnings objectives. REM is also more difficult for auditors and regulators to detect. However, REM may generate adverse economic consequences for firms. Specifically, such practices can distort operational decision-making, reduce operating efficiency, negatively affect future cash flows, and ultimately impair the firm’s ability to create sustainable long-term value. Typical practices include reducing discretionary expenses, such as research and development (R&D) and marketing; implementing temporary price discounts to boost sales; or overproducing to reduce unit costs. Empirical measurement of REM uses the abnormal components of operating cash flows (Abn-CFO), production costs (Abn-PROD), and discretionary expenditures (Abn-DISEXP) based on the Roychowdhury model, or consolidates them into a single REM index following the methodology of Cohen and Zarowin (Roychowdhury, 2006; Cohen & Zarowin, 2010).

This study integrates four complementary theoretical perspectives to explain the relationships among corporate governance, real earnings management, and financial performance. First, agency theory provides the core monitoring logic, suggesting that governance mechanisms reduce conflicts of interest between managers and shareholders, thereby limiting managerial opportunism and earnings manipulation (Jensen & Meckling, 1976; Shleifer & Vishny, 1997). Second, resource dependence theory suggests that boards of directors not only monitor management but also provide advice, expertise, legitimacy, and access to external resources that may improve strategic decision-making and firm performance (Pfeffer & Salancik, 1978; Hillman & Dalziel, 2003). This perspective is particularly relevant for explaining why larger or more competent boards may enhance performance in firms operating in resource-constrained emerging markets. Third, institutional theory explains why formally adopted governance mechanisms may not always function effectively in emerging markets, where legal enforcement, transparency, investor protection, and disclosure discipline are still developing (DiMaggio & Powell, 1983; North, 1990). From this perspective, board independence may become more symbolic than substantive if independent directors lack sufficient access to information, authority, or institutional support to challenge managerial decisions. Fourth, political connection theory clarifies the role of state ownership by suggesting that state shareholders may provide firms with access to financing, government-related contracts, legitimacy, and implicit support, while also potentially introducing non-commercial objectives and soft budget constraints (Faccio, 2006; Boubakri et al., 2008; Kubo & Phan, 2019).

Within this integrated theoretical framework, corporate governance can affect financial performance through both direct and indirect channels. Directly, governance mechanisms may enhance monitoring quality, improve access to strategic resources, strengthen legitimacy, and reduce information asymmetry. Indirectly, these mechanisms may influence financial performance by constraining or enabling real earnings management. Effective monitoring by boards, institutional investors, and state shareholders can reduce managers’ incentives to manipulate real operating activities, whereas weak or merely symbolic governance arrangements may fail to prevent such practices. Recent empirical evidence supports this mechanism, showing that institutional ownership and ownership concentration can reduce real earnings management in emerging markets, including China and Vietnam (Ali et al., 2024; Q. Nguyen et al., 2024). Therefore, real earnings management is a theoretically meaningful transmission channel through which corporate governance mechanisms may affect firm performance.

2.2. Research Hypotheses

2.2.1. Board Characteristics and Financial Performance

Board size, measured by the number of directors, is a key governance attribute that may influence firm performance through both monitoring and advisory channels. From the agency theory perspective, a larger board may strengthen managerial oversight by increasing monitoring capacity, reducing information asymmetry, and limiting managerial opportunism (Jensen & Meckling, 1976; Anderson et al., 2004). From the resource dependence theory perspective, a larger board may also provide a broader pool of expertise, external linkages, legitimacy, and access to critical resources, thereby improving strategic decision-making and firm performance (Pfeffer & Salancik, 1978; Hillman & Dalziel, 2003).

Empirical evidence on the relationship between board size and firm performance remains inconclusive. Several studies suggest that board size is positively associated with firm performance because larger boards may offer broader business networks, more diverse professional expertise, and stronger strategic connections that help firms access external financing and expand into new markets (Johl et al., 2015; Detthamrong et al., 2017; Kyere & Ausloos, 2020). However, other studies caution that excessively large boards may reduce governance effectiveness because of coordination problems, slower decision-making, free-riding, and greater susceptibility to managerial influence can weaken the board’s monitoring function (Lipton & Lorsch, 1992; Jensen, 1993). Consistent with this view, Yermack (1996) documents a negative association between board size and firm value, while Saidat et al. (2019) and Mensah and Bein (2023) report a negative relationship between board size and financial performance in emerging-market contexts. Other evidence also suggests that board size may have no significant effect on financial performance in certain institutional settings (Blach et al., 2025).

Although the literature provides mixed evidence, this study expects a positive relationship between board size and financial performance in the Vietnamese listed-firm context. This expectation is based on the argument that the performance effect of board size depends on whether the board is excessively large or remains within a moderate, functional range. In emerging markets, where firms often face higher information asymmetry, weaker external governance mechanisms, and greater dependence on business networks and external resources, a moderately larger board may enhance both monitoring and advisory capacity. Therefore, within the practical range observed among Vietnamese listed firms, additional directors are expected to improve governance effectiveness, broaden expertise, and facilitate access to external resources. Based on this reasoning, the following hypothesis is proposed:

H1a. 

Board size has a positive effect on firms’ financial performance.

Board composition is a key internal corporate governance mechanism for improving monitoring effectiveness and safeguarding stakeholder interests. In particular, independent directors play a critical role in promoting objectivity and transparency in overseeing managerial activities. Board independence is typically measured by the proportion or number of independent (non-executive) directors on the board. According to agency theory, independent directors help reduce agency costs by providing objective oversight and mitigating conflicts of interest between shareholders and management. Additionally, according to resource dependence theory, a higher proportion of independent directors on the board enables firms to gain greater access to critical strategic resources and valuable information. Consequently, increasing the number of independent directors is expected to enhance the firm’s performance through improved monitoring, accountability, and access to external resources.

Nevertheless, an alternative perspective suggests that excessive independence may carry drawbacks. Independent directors who lack comprehensive internal knowledge, familiarity with the organization, or sufficient engagement with management may be less effective in strategic decision-making, potentially diminishing governance quality (Finegold et al., 2007). Evidence on this relationship remains inconclusive. For instance, Khan et al. (2024), examining non-financial listed companies in Pakistan (2003–2018), Lee et al. (2024) in Taiwan, and Saidat et al. (2019) in Jordan all identify a negative correlation between board independence and corporate performance. Conversely, numerous studies conducted in various contexts—such as those by Pearce and Zahra (1992), Aidoo et al. (2024), and Kyere and Ausloos (2020)—demonstrate a positive association, suggesting that increased independence can enhance oversight and augment firm value. Additional research (Bhagat & Black, 1999) reveals no significant relationship, highlighting how factors such as institutional environments, ownership structures, and access to internal information vary across diverse settings.

Overall, these findings suggest that institutional and governance environments significantly affect the effectiveness of independent directors. In emerging markets such as Vietnam, where investor protection and market discipline mechanisms are still evolving, independent directors are expected to enhance financial performance by strengthening oversight and ensuring greater accountability.

H1b. 

Board independence has a positive impact on the financial performance of firms.

The professional expertise of board members is typically assessed based on their educational qualifications in disciplines such as accounting, finance, economics, or law, as well as their professional certifications. According to resource dependence theory, directors with backgrounds in accounting, finance, or executive management provide valuable knowledge and experience that enhance oversight quality, standardize control processes, and facilitate more efficient capital allocation. This expertise serves as a vital strategic resource, strengthening the firm’s governance capabilities and overall performance.

Empirical research robustly corroborates this theoretical assertion. For instance, Guner et al. (2008), Kor and Sundaramurthy (2008), and Johl et al. (2015) demonstrate that having directors with financial or managerial expertise enhances monitoring effectiveness and subsequently improves firm performance. Recent empirical evidence from emerging and developing economies further substantiates these findings: Aidoo et al. (2024) identify a positive correlation between board expertise and firm performance in Ghana; Lee et al. (2024) observe similar results for companies in Taiwan; Danso et al. (2024) affirm this relationship across publicly listed companies in several African nations; and Q. Nguyen et al. (2024) offer consistent evidence within the Vietnamese context.

The convergence of theoretical reasoning and empirical findings suggests that board expertise serves as a strategic resource that enhances financial performance primarily through improved decision-making, heightened monitoring, and more robust governance practices. Accordingly, in Vietnam—an emerging market with developing governance frameworks and an increasing emphasis on board competence—we propose the following hypothesis:

H1c. 

Board expertise has a positive impact on firms’ financial performance.

2.2.2. Ownership Structure and Financial Performance

From the perspective of agency theory, increasing institutional ownership is considered an effective corporate governance mechanism for mitigating the monitoring costs borne by shareholders. However, the empirical literature on the relationship between institutional ownership and corporate financial performance has produced mixed and inconclusive findings. Multiple scholarly studies suggest a positive correlation between institutional ownership and financial success, highlighting the strategic advisory and oversight functions of institutional investors in enhancing operational efficiency and augmenting firm value (Queiri et al., 2021; Abedin et al., 2022). Institutional investors, equipped with expertise and access to superior information, frequently serve as astute monitors, helping align management decisions with shareholder interests and bolstering overall governance.

Conversely, other studies indicate a negative correlation between institutional ownership and firm performance. They propose that institutional investors might sometimes cause inefficiencies due to conflicting objectives, short-term investment horizons, or dependence on business relationships with the companies they invest in (Rostami et al., 2016; Tawfik et al., 2022). For example, institutional investors focused on short-term outcomes, or those characterized as “pressure-sensitive,” might prioritize immediate financial gains over long-term value, putting pressure on management to manipulate earnings or adopt shortsighted strategies.

Furthermore, numerous studies show mixed results—indicating adverse effects on ROA but minimal or no impact on other metrics such as ROE, earnings per share (EPS), or net interest margin (NIM)—suggesting that the relationship depends on the measurement methodology and contextual variables (Kirimi et al., 2022). These differences may stem from distinctions among institutional investors, particularly between “independent, pressure-insensitive” entities and “grey” or affiliated entities that maintain business relationships with the companies they invest in, as well as from variations in market structures and ownership concentration across countries.

Evidence from Vietnam supports this diversity. Hong and Linh (2023) show that independent, long-term institutional investors contribute positively to firms through effective monitoring and strategic guidance, whereas “grey” institutions—characterized by intertwined business interests—do not improve, and may even harm, firm performance. Therefore, in a developing market like Vietnam, where investor protection and governance systems are still evolving, institutional ownership is likely to have a positive effect on firms’ financial performance through enhanced oversight and advisory skills. Based on these theoretical reasons and empirical findings, we propose the following hypothesis:

H1d. 

Institutional ownership has a positive effect on firms’ financial performance.

State ownership creates a distinctive set of corporate objectives in which economic goals are often pursued alongside social and political objectives. Unlike private-sector firms, which primarily focus on maximizing profits and shareholder value, state-owned enterprises are frequently required to fulfill multiple public policy functions, including the provision of public services, the promotion of macroeconomic stability, and the enhancement of social welfare. Consequently, the operational objectives of state-owned firms may not be fully aligned with profit-maximization incentives. The State’s presence in a company’s ownership structure is viewed as an administrative instrument that can mitigate agency costs arising from the separation of ownership and control. It also lays a foundation for long-term strategic planning and protects minority shareholders’ interests. According to agency theory, the State—serving as a substantial shareholder with monitoring capabilities—can oversee management, thereby potentially enhancing corporate performance and financial efficiency (Chhibber & Majumdar, 1998). Empirical evidence suggests a positive correlation between state ownership and firm performance, highlighting the State’s role in providing resources and institutional support in specific contexts.

On the positive side, empirical evidence suggests that firms with higher levels of state ownership tend to outperform their counterparts, likely due to enhanced access to resources, credit, and contractual relationships. This argument is supported by Chhibber and Majumdar (1998), who document that state involvement may enhance firm performance under certain institutional conditions through stronger monitoring and resource support. Similar evidence is reported in Vietnam, where state ownership is associated with improved financial performance, particularly when state capital is managed through professional ownership arrangements and commercially oriented governance structures (Kubo & Phan, 2019). These findings reinforce the notion that the State, as a shareholder, can augment firms’ resource mobilization and mitigate information asymmetry, thereby enhancing overall efficiency.

Conversely, numerous studies indicate that state ownership can harm corporate performance, particularly when political or social objectives are prioritized over commercial interests, or when ownership concentration exceeds a threshold that allows for control. Such circumstances often lead to higher agency costs and the emergence of a “soft budget constraint” issue. For example, Kirimi et al. (2022) identify an inverse relationship between state ownership and NIM, with no significant effect on ROA, ROE, or EPS. Similarly, Blach et al. (2025) report no meaningful correlation between state ownership and financial performance. In line with these findings, B. Nguyen et al. (2022) suggest that elevated levels of state ownership—especially when exceeding controlling thresholds—may impair performance by fostering managerial entrenchment and incorporating non-commercial objectives.

Accordingly, within the context of an emerging economy such as Vietnam—where corporate governance frameworks and mechanisms of state ownership are progressively being reformed toward professionalism—we propose the following hypothesis:

H1e. 

State ownership has a positive effect on firms’ financial performance.

2.2.3. The Mediating Role of Real Earnings Management

Regarding the correlation between REM and financial performance, extant empirical studies indicate that although REM can help firms achieve short-term earnings objectives, it often entails substantial operational adjustment costs and leads to reduced long-term efficiency as REM intensity increases (Gunny, 2010; Cohen & Zarowin, 2010; Al-Shattarat et al., 2022). These findings suggest that REM offers immediate benefits at the expense of long-term financial stability.

Evidence on the mediating role of REM in the relationship between corporate governance and financial performance reveals complex dynamics. Several empirical studies have examined this framework and confirmed that REM serves as a mediator in specific contexts (Kang & Kim, 2011). Research on corporate governance and firm performance through earnings management indicates that effective governance mechanisms constrain earnings management, thereby enhancing financial performance (Manan & Amin, 2023; Rizani et al., 2022). For instance, managerial ownership influences financial performance, and REM either reinforces or alters the governance–performance relationship during critical periods such as the pre-IPO phase (Kang & Kim, 2011; Semsomboon et al., 2024).

According to Bojan et al. (2024), corporate governance serves as both a direct driver of firm performance and a moderator that mitigates the adverse effects of earnings management on financial outcomes. Board independence and audit committee size are particularly critical governance factors that constrain earnings management and promote improved performance (Manan & Amin, 2023; Bojan et al., 2024). In the context of REM, institutional ownership serves as an effective governance instrument that curtails such practices (Ali et al., 2024), although the influence of board characteristics remains inconsistent. Accounting for earnings management effects markedly enhances the positive influence of corporate governance on financial performance, underscoring earnings management as a detrimental mediator between governance and firm outcomes. Similarly, evidence from Pakistani organizations suggests that increased board independence enhances financial performance by reducing earnings management (Manan & Amin, 2023). Further evidence from Iran demonstrates that having non-executive board members and separating the roles of Chief Executive Officer and Chairman diminish the relationship between managerial incentives and REM (Mansoori & Al-Abdallah, 2024).

In Vietnam, no empirical research has yet examined whether REM serves as a mediator in the relationship between corporate governance and financial performance. Given the institutional characteristics of emerging markets such as Vietnam, where investor protection mechanisms are still developing and transparency remains relatively low, REM’s role as a transmission channel is expected to be particularly significant. Consequently, this study proposes the following hypothesis.

H2. 

Real earnings management mediates the relationship between corporate governance and a firm’s financial performance.

3. Methodology

This study uses a panel dataset of non-financial firms listed on the Vietnamese stock market from 2020 to 2024. Financial firms were excluded because their capital structures and accounting standards differ markedly from those of non-financial companies, which could affect data comparability (Firth et al., 2007). Additionally, firms with missing or incomplete data were excluded to ensure data consistency and reliability.

The final sample comprises 434 firms across ten industry groups (47 in real estate, 22 in healthcare, 5 in information technology, 79 in industrials, 25 in utilities, 22 in telecommunications services, 23 in energy, 109 in materials, 53 in consumer discretionary, and 49 in consumer staples), yielding 2170 firm-year observations (

Table 1. Distribution of sample firms by industry.

Industry	Number of Firms	Firm-Year Observations	Percentage (%)
Real estate	47	235	10.83
Healthcare	22	110	5.07
Information technology	5	25	1.15
Industrials	79	395	18.20
Utilities	25	125	5.76
Telecommunications services	22	110	5.07
Energy	23	115	5.30
Materials	109	545	25.12
Consumer discretionary	53	265	12.21
Consumer staples	49	245	11.29
Total	434	2170	100.00

Note: The sample includes 434 non-financial listed firms across ten industry groups during the 2020–2024 period, yielding 2170 firm-year observations. Firm-year observations are calculated by multiplying the number of firms in each industry by five years.

). All continuous variables were winsorized annually at the 1st and 99th percentiles to reduce outlier effects. Data were sourced from audited financial statements, annual reports, and official disclosures issued by the firms. Market-related indicators were obtained from the Vietnam Stock Exchange’s official publications.

The dependent variable is the company’s financial performance. According to Mwangi and Murigu (2015), two indicators are used: ROA, which reflects accounting-based performance, and Tobin’s Q, which reflects market-based performance.

The independent variables relate to corporate governance and are grouped into two categories: (i) board characteristics and (ii) ownership structure. Board characteristics are assessed using three indicators: (i) board size, measured by the number of directors (Kyere & Ausloos, 2020); (ii) board independence, quantified by the proportion of independent directors (Kyere & Ausloos, 2020); and (iii) board financial expertise (BE), measured as the proportion of board members with formal academic qualifications or professional backgrounds in accounting, auditing, finance, or closely related financial fields. This measure is derived from publicly available information disclosed in annual and corporate governance reports. Because the study relies on standardized public disclosures, BE captures formal financial and accounting expertise rather than the broader managerial, legal, or industry-specific expertise of board members. Accordingly, the variable is interpreted as board financial expertise rather than a comprehensive measure of overall board competence (Johl et al., 2015).

The ownership structure comprises two components: (i) institutional ownership, representing the share of shares held by institutional investors (Kirimi et al., 2022; Tawfik et al., 2022), and (ii) state ownership, representing the share of shares owned by government or state-affiliated entities (Blach et al., 2025; Kirimi et al., 2022).

The mediating variable is real earnings management, estimated using the Roychowdhury (2006) model, which measures abnormal levels of cash flow from operations, production costs, and discretionary expenses.

Following Roychowdhury (2006), REM is estimated using abnormal cash flows from operations, abnormal production costs, and abnormal discretionary expenses. Normal levels of these components are estimated by industry using pooled firm-year observations over the sample period. This industry-based approach accounts for differences in operating structures across industries while preserving a sufficient number of observations for residual estimation. Because the sample covers 434 non-financial listed firms over five years, some narrowly defined industry-year cells contain a limited number of observations. Therefore, estimating the Roychowdhury models separately for each industry-year cell may produce unstable residuals. To improve the reliability of estimates, the study uses industry-based estimates and reports the distribution of firms across industries to demonstrate the adequacy of industry-level observations. The residuals from these models are used as abnormal components, and the aggregate REM index is calculated as REM = (−1 × Abn_CFO) + Abn_PROD + (−1 × Abn_DISEXP), so that higher REM values indicate a greater extent of real earnings management. The measurement procedures for REM are outlined in

Table 2. Measurement of REM.

Measurement of REM Following Roychowdhury (2006)
The level of abnormal cash flow from operating activities (REM_CFO) is estimated using the residual from the following model: ${\displaystyle \frac{CF{O}_{it}}{A_{i,t-1}}}=\alpha \left({\displaystyle \frac{1}{A_{i,t-1}}}\right)+{\beta }_1\left({\displaystyle \frac{SALE{S}_{it}}{A_{i,t-1}}}\right)+{\beta }_2\left({\displaystyle \frac{\Delta SALE{S}_{it}}{A_{i,t-1}}}\right)+{\epsilon }_{it}$	- CFOit: Cash flow from operating activities of firm i in year t - DISCit: Total selling and administrative expenses of firm i in year t - PRODit: Production cost of firm i in year t, measured as the sum of cost of goods sold and the change in net inventory - Ai,t − 1: Total assets of firm i in year t − 1 - SALESit: Net sales revenue of firm i in year t - SALESit − 1: Net sales revenue of firm i in year t − 1 - ΔSALESit: Change in sales between years t and t − 1 - ΔSALESit − 1: Change in sales between years t − 1 and t − 2
The level of abnormal discretionary expenses (REM_DISC) is estimated as the residual from the following model: ${\displaystyle \frac{DIS{C}_{it}}{A_{i,t-1}}}=\alpha \left({\displaystyle \frac{1}{A_{i,t-1}}}\right)+{\beta }_1\left({\displaystyle \frac{SALE{S}_{i,t-1}}{A_{i,t-1}}}\right)+{\epsilon }_{it}$
The level of abnormal production costs (REM_PROD) is estimated as the residual from the following model: ${\displaystyle \frac{PRO{D}_{it}}{A_{i,t-1}}}=\alpha \left({\displaystyle \frac{1}{A_{i,t-1}}}\right)+{\beta }_1\left({\displaystyle \frac{SALE{S}_{it}}{A_{i,t-1}}}\right)+{\beta }_2\left({\displaystyle \frac{\Delta SALE{S}_{it}}{A_{i,t-1}}}\right)+{\beta }_3\left({\displaystyle \frac{\Delta SALE{S}_{i,t-1}}{A_{i,t-1}}}\right)+{\epsilon }_{it}$
The aggregate REM index is calculated as follows: REM = (−1) × REM_CFO + REM_PROD + (−1) × REM_DISC

.

The control variables include firm size and financial leverage, commonly used in studies of corporate governance, earnings management, and firm performance. Firm size controls for differences in organizational scale, resource availability, and disclosure capacity, while financial leverage captures financial risk and creditor pressure (Johl et al., 2015; Kyere & Ausloos, 2020; Tawfik et al., 2022). Although other variables, such as ownership concentration, audit quality, firm age, growth opportunities, and industry fixed effects, may also influence financial performance and real earnings management, the present study does not include these additional controls because consistent, comparable data for all 434 firms over the 2020–2024 balanced panel are not fully available. To preserve sample consistency and avoid further sample reduction, the empirical models retain firm size and leverage as the primary control variables. Accordingly, the estimated results are interpreted as conditional associations rather than definitive causal effects.

To evaluate the hypotheses concerning the influence of corporate governance on financial performance (H1), a panel data estimation model with firm and year fixed effects is used, along with robust standard errors to mitigate heteroskedasticity and autocorrelation.

FP_i,t = γ₀ + γ₁CG_i,t + γ₂Controls_i,t + e_i,t

(1)

where CG represents the vector of corporate governance variables (such as board size, board independence, board expertise, institutional ownership, and state ownership), and controls represents the vector of control variables (including firm size and financial leverage).

To examine the mediating role of REM within the relationship between corporate governance and firm financial performance (H2), we have developed and tested the following two models.

REM_i,t = α₀ + α₁CG_i,t + α₂Controls_i,t + ε_i,t

(2)

FP_i,t = β₀ + β₁REM_i,t + β₂CG_i,t + β₃Controls_i,t + u_i,t

(3)

Panel regression techniques are used to examine the proposed relationships. FGLS is used as the baseline estimator because diagnostic tests indicate heteroskedasticity and first-order autocorrelation. To assess potential endogeneity, the Durbin–Wu–Hausman test is conducted. The test results do not indicate statistically significant endogeneity in the current model specification. Nevertheless, because the relationship between corporate governance and firm performance may still be affected by reverse causality, omitted-variable bias, and dynamic persistence, the study further conducts a robustness analysis using the two-step System GMM estimator.

In the System GMM specification, the lagged dependent variable is included to capture the dynamic persistence of firm performance and REM. Governance variables are treated as potentially endogenous and instrumented with their appropriate lagged values. To reduce the risk of instrument proliferation, the GMM instruments are collapsed. The validity of the instruments is assessed using the Hansen test and the Difference-in-Hansen test, while the Arellano–Bond AR(1) and AR(2) tests assess serial correlation in the differenced residuals. The System GMM results are reported as an additional robustness and endogeneity check rather than as a replacement for the baseline FGLS estimates.

To further assess the robustness of the findings, this study uses Tobin’s Q as an alternative measure of financial performance. Tobin’s Q is a market-based indicator that reflects investors’ expectations of firms’ future prospects. Figure 1 presents the conceptual research model, illustrating how corporate governance is associated with financial performance through the mediating role of REM.

4. Results and Discussion

4.1. Results

Table 3. Descriptive Statistics.

Variable	Obs	Mean	SD	Min	Max
ROA	2170	0.056	0.071	−0.302	0.603
TQ	2170	1.195	0.731	0.180	17.173
BS	2170	5.498	1.353	2.000	11.000
BI	2170	0.731	0.163	0.200	1.000
BE	2170	0.026	0.071	0.000	0.667
IN	2170	0.559	0.232	0.000	0.999
ST	2170	0.196	0.277	0.000	0.993
REM	2170	−0.206	0.363	−2.052	3.450
FS	2170	27.965	1.736	23.603	34.360
LE	2170	0.461	0.211	0.013	1.295

presents descriptive statistics for 2170 firm-year observations, with all continuous variables winsorized at the 1% level to mitigate the influence of outliers. The mean return on assets (ROA) is 0.056 (approximately 5.6%), with a standard deviation of 0.071 and a range from −0.302 to 0.603, indicating substantial variability in firms’ financial performance. The market-based metric, Tobin’s Q, has a mean of 1.195, a standard deviation of 0.731, and values ranging from 0.180 to 17.173, reflecting the typically right-skewed distribution characteristic of market valuation indicators.

Regarding board characteristics, the mean board size (BS) is 5.5 members (SD = 1.35), with a range of 2 to 11, indicating that Vietnamese listed companies generally maintain medium-sized boards. The average board independence (BI) is 0.731 (73.1%), with a standard deviation of 0.163. In contrast, the mean BE is 0.026 (2.6%), indicating that formally disclosed financial and accounting expertise among board members remains limited in the sample. This low value should be interpreted in light of the restrictive definition of BE, which captures only formal financial or accounting qualifications reported in public disclosures rather than broader managerial or industry expertise.

Regarding ownership structure, institutional ownership (IN) averages 55.9% (SD = 0.232), and state ownership (ST) averages 19.6% (SD = 0.277), indicating moderate ownership concentration. The mediating variable, real earnings management (REM), has a mean of −0.206 (SD = 0.363), with a range from −2.052 to 3.450. Control variables show an average firm size (FS) of 27.97 (SD = 1.74) and financial leverage (LE) of 46.1% (SD = 0.21).

The low mean of BE suggests that formally disclosed financial and accounting expertise among board members is limited. However, this result should be interpreted cautiously because BE captures only observable financial and accounting qualifications, not broader forms of board expertise. Similarly, the dispersion of REM reflects variation in abnormal real operating activities across firms and industries. To improve transparency in REM estimation, the revised manuscript reports the distribution of sample firms by industry. Overall, the data show balanced variability and are suitable for subsequent regression and mediation analyses.

Table 4. Correlation Matrix.

	ROA	TQ	BS	BI	BE	IN	ST	REM	FS	LE
ROA	1.000
TQ	0.404 ***	1.000
BS	0.084 ***	0.187 ***	1.000
BI	−0.066 ***	0.020	0.086 ***	1.000
BE	−0.034	−0.015	0.004	−0.017	1.000
IN	0.126 ***	0.115	−0.033	0.041 *	0.048 **	1.000
ST	0.078 ***	0.083 ***	−0.018	−0.101 ***	0.057 ***	0.461 ***	1.000
REM	−0.454 ***	−0.300 ***	−0.130 ***	0.037 *	0.057 ***	−0.168 ***	−0.159 ***	1.000
FS	−0.087 ***	0.100 ***	0.354 ***	0.056 ***	−0.033	−0.118 ***	−0.030	0.107 ***	1.000
LE	−0.392 ***	−0.092 ***	0.046 **	−0.024	−0.026	−0.080 ***	−0.015	0.198 ***	0.364 ***	1.000

Note: ***, **, * represents p < 0.01, p < 0.05, and p < 0.1 respectively.

presents the correlation matrix for variables assessing financial performance, corporate governance characteristics, REM, and control variables. The majority of the correlation coefficients among these variables are relatively modest. The most notable value is 0.461 between institutional ownership (IN) and state ownership (ST), signifying a moderate association. As all coefficients are below 0.8, there is no indication of multicollinearity among the variables.

Moreover, the multicollinearity tests presented in

Table 5. Results of the Multicollinearity Test.

Variables	Model 1 (ROA)		Model 2 (REM)		Model 3 (ROA)
	VIF	1/VIF	VIF	1/VIF	VIF	1/VIF
REM					1.12	0.893
BS	1.16	0.862	1.16	0.862	1.20	0.836
BI	1.03	0.970	1.03	0.970	1.03	0.967
BE	1.01	0.995	1.01	0.995	1.01	0.988
IN	1.30	0.767	1.30	0.767	1.32	0.760
ST	1.30	0.771	1.30	0.771	1.31	0.763
FS	1.34	0.744	1.34	0.744	1.35	0.739
LE	1.17	0.857	1.17	0.857	1.20	0.835
Mean VIF	1.19		1.19		1.19

indicate that all variance inflation factor (VIF) values are below 10, confirming the absence of multicollinearity in the research models.

Diagnostic tests affirm the existence of heteroskedasticity and first-order autocorrelation; consequently, the study employs Feasible Generalized Least Squares (FGLS) estimation to generate efficient and dependable results. No indication of endogeneity is detected, thereby guaranteeing the credibility of the estimates. Model 1 (

Table 6. Regression Results on the Mediating Role of REM.

Variables	Model 1 (ROA)		Model 2 (REM)		Model 3 (ROA)
	Coef.	p (Sig)	Coef.	p (Sig)	Coef.	p (Sig)
REM					−0.0440	0.000
BS	0.0031	0.000	−0.0284	0.000	0.0027	0.000
BI	−0.0162	0.000	0.1311	0.000	−0.0137	0.000
BE	0.0004	0.961	0.1448	0.011	0.0009	0.917
IN	0.0193	0.000	−0.1305	0.000	0.0174	0.000
ST	0.0064	0.014	−0.0627	0.000	0.0017	0.477
FS	0.0017	0.001	0.0175	0.000	0.0026	0.000
LE	−0.1057	0.000	0.2620	0.000	−0.0998	0.000
_cons	0.0343	0.009	−0.6620	0.000	−0.0004	0.965
Observations	2170		2170		2170
Prob > chi2	0.0000		0.0000		0.0000

) demonstrates statistical significance according to the Wald χ² test (p < 0.000), corroborating the comprehensive model fit.

The findings demonstrate that board size (BS) has a positive and significant effect on firm performance (ROA), supporting hypothesis H1a. Larger boards appear to enhance advisory and monitoring functions, which in turn contribute to higher profitability. Conversely, board independence (BI) shows a negative and significant relationship with ROA, leading to the rejection of hypothesis H1b. Additionally, board expertise (BE) is statistically insignificant, providing no support for hypothesis H1c.

Regarding ownership structure, institutional ownership (IN) has a positive and significant effect on ROA, confirming H1d. This underscores the role of institutional investors in strengthening oversight and corporate discipline. Likewise, state ownership (ST) shows a positive and significant relationship with ROA, supporting H1e and indicating that state participation—when well-managed—can boost firm performance.

In conclusion, three hypotheses (H1a, H1d, H1e) are supported, one (H1b) is refuted, and one (H1c) remains unsupported. These findings provide a robust foundation for investigating REM’s mediating role in subsequent models.

The Wald χ² test (Prob > χ² = 0.000) confirms that Models 2 and 3 are well-fitted. In Model 2, with REM as the dependent variable, board size (BS) has a negative and significant effect on REM. In contrast, board independence (BI) and board expertise (BE) have positive and significant effects. Regarding ownership, both institutional (IN) and state ownership (ST) have negative and significant effects on REM, confirming the necessary condition for mediation—that governance variables substantially affect the extent of real earnings management.

Model 3, using ROA as the dependent variable, indicates that REM has a negative effect on financial performance, implying that higher REM levels are associated with lower profitability. When REM is included, the direct effects of governance factors diminish: BS remains positive but smaller, BI remains negative but less significant, IN remains positive but less pronounced, and ST loses statistical significance.

These patterns indicate that REM partially mediates the effects of board size, board independence, and institutional ownership on ROA, and fully mediates the relationship between state ownership and ROA. Therefore, hypothesis H2 regarding REM’s mediating role is supported. Practically, strengthening governance mechanisms that reduce REM—particularly through board structure, institutional oversight, and state ownership discipline—can indirectly improve firm performance.

The Durbin–Wu–Hausman test yields a p-value of 0.4127, which does not reject the null hypothesis of exogeneity in the current specification. This result suggests that statistically detectable endogeneity is not evident in the estimated model. However, because corporate governance and firm performance may be subject to reverse causality and dynamic persistence, the findings should be interpreted as evidence of associations rather than definitive causal effects.

4.2. Discussion

This study had two primary objectives: to assess the impact of corporate governance on financial performance and to examine the mediating role of REM.

Regarding the initial objective, the estimation models indicate that board- and ownership-related factors have varying effects on ROA. Specifically, board size has a positive and statistically significant impact, whereas board independence shows a negative and significant relationship. Conversely, board expertise is not statistically significant. Regarding ownership, institutional ownership has a positive and significant influence on ROA, and state ownership also shows a positive effect in the baseline model.

Regarding the second objective, the mediation analysis shows that corporate governance variables significantly affect REM. Additionally, higher REM levels are associated with lower ROA. When REM is included in the model, the direct effect of governance variables on ROA diminishes, and the influence of state ownership shifts from statistically significant to insignificant, indicating full mediation through this pathway.

Compared with the existing literature, the findings on board size are consistent with prior research indicating a positive influence on financial performance, underscoring the advisory benefits and enhanced resource connectivity associated with larger boards in publicly listed companies (Kyere & Ausloos, 2020). Regarding institutional ownership, the identified positive effect aligns with evidence that institutional investors enhance internal monitoring and managerial discipline, thereby improving firm performance (Queiri et al., 2021). State ownership also shows a positive relationship in the baseline model, consistent with earlier research highlighting advantages related to resource access and institutional support in specific contexts (Chhibber & Majumdar, 1998). In the Vietnamese context, this finding aligns with evidence suggesting that “pressure-insensitive” institutional investors are correlated with higher firm valuation (Hong & Linh, 2023), as well as observations that state ownership can enhance performance when capital representation mechanisms are effectively designed (Kubo & Phan, 2019). In emerging economies where the state plays a major role in resource allocation and market regulation, state ownership may reflect not only a governance mechanism but also a form of political connection. State shareholders may confer advantages on firms, including access to capital, policy-related information, government contracts, land, and institutional networks. These advantages can ease financing constraints and support firm performance. However, this positive association should not be taken as evidence that state ownership is universally beneficial. State ownership may also introduce non-commercial objectives, soft budget constraints, and political intervention. Therefore, the positive coefficient observed in this study is better understood as reflecting the resource-access and political-connection advantages associated with state shareholding in the Vietnamese institutional context, rather than a purely monitoring-based governance effect.

Conversely, the findings on board independence show a negative effect on ROA, which contradicts agency theory but aligns with several studies in emerging markets where institutional frameworks and access to internal information are constrained (Saidat et al., 2019). This result is among the study’s most theoretically important findings. Agency theory generally assumes that independent directors can improve monitoring and enhance firm performance; however, this assumption depends on whether independent directors have sufficient access to information, enforcement support, and real autonomy from controlling shareholders and executives. From an institutional theory perspective, increasing board independence does not necessarily enhance corporate performance when independence is adopted mainly for formal compliance rather than substantive oversight. In emerging economies such as Vietnam, where de facto power is often concentrated in controlling shareholders and where monitoring systems, regulatory enforcement, and information transparency remain underdeveloped, the effectiveness of independent board members may be constrained. In this setting, board independence may become more symbolic than substantive if independent directors lack sufficient authority, information access, or institutional support to challenge managerial decisions. Previous research has also yielded mixed results depending on performance indicators and contextual factors (Kyere & Ausloos, 2020), indicating that the effectiveness of independent directors depends on information quality, access rights, enforcement mechanisms, and the actual power of outside directors within specific institutional settings. Thus, the negative relationship observed in this study does not imply that independence is inherently harmful; rather, it suggests that formal independence may not improve performance unless it is supported by genuine authority, sufficient information access, and effective enforcement. Regarding board expertise, the lack of statistical significance suggests that its direct influence on ROA may be limited or context-dependent, potentially operating through indirect pathways rather than exerting a direct effect within the baseline model.

Taken together, these findings refine the conventional agency theory view of corporate governance. They suggest that the effectiveness of governance mechanisms depends not only on their formal presence but also on the institutional environment in which they operate. Board independence may fail to deliver performance benefits when it remains formal rather than substantive, while state ownership may improve performance through political-resource channels rather than through monitoring alone. This evidence underscores the importance of integrating institutional theory and political connection theory into the analysis of governance mechanisms in emerging markets.

Regarding the mediating mechanism, the results demonstrate that corporate governance influences REM as follows: larger board sizes and higher levels of ownership—both institutional and state—are associated with lower REM intensity, whereas board independence and expertise are correlated with higher REM levels. These findings substantiate prior research suggesting that enhanced governance mechanisms can effectively mitigate real activity manipulation (Chouaibi et al., 2018; Al-Duais et al., 2022; Hsu & Wen, 2015). Concurrently, they corroborate evidence indicating that, in the absence of supportive institutional conditions, certain board characteristics may not function as intended and may instead be linked to heightened REM (Al-Haddad & Whittington, 2019; Asmaranti et al., 2024).

The negative relationship between REM and ROA supports the “short-term benefits–long-term costs” hypothesis of real activity manipulation (Habib, 2024). Overall, these findings support an agency-theory interpretation: robust corporate governance effectively diminishes managers’ incentives to manipulate earnings through real activities, thereby enhancing financial performance. In the domestic context, evidence that corporate governance indices are negatively correlated with both AEM and REM among non-financial firms in Vietnam (Q. Nguyen et al., 2024) corroborates that this study’s findings are consistent with broader trends and elucidate how REM specifically transmits these effects.

The implications indicate that companies should focus on corporate governance practices that reduce REM, including adjusting board size to balance advisory and monitoring roles, increasing institutional investor involvement, and strengthening mechanisms for representing state ownership. These measures can improve ROA both directly and indirectly by reducing REM. Boards of directors should enhance the expertise, authority, and information access of independent members to mitigate the risk that independence remains merely formal. Regulators ought to continue refining disclosure standards, enforcement mechanisms, and requirements regarding the substantive role of independent directors. For state ownership, the findings suggest the need to professionalize state capital representation, separate commercial objectives from administrative intervention, and enhance transparency in the exercise of state shareholder rights. These measures can help preserve the resource-access benefits associated with state ownership while reducing the risks of political intervention and non-commercial objectives.

4.3. Robustness Tests

To assess the robustness of the findings, the study replaces the accounting-based measure (ROA) with the market-based indicator Tobin’s Q. The regression results in

Table 7. Robustness Test Using Tobin’s Q as an Alternative Measure.

Variables	Model 1 (TQ)		Model 2 (REM)		Model 3 (TQ)
	Coef.	p (Sig)	Coef.	p (Sig)	Coef.	p (Sig)
REM					−0.1536	0.000
BS	0.0473	0.000	−0.0284	0.000	0.0477	0.000
BI	−0.0425	0.184	0.1311	0.000	−0.0920	0.003
BE	−0.0393	0.587	0.1448	0.011	0.0588	0.470
IN	0.1557	0.000	−0.1305	0.000	0.1305	0.000
ST	0.0513	0.046	−0.0627	0.000	0.0396	0.111
FS	0.0182	0.001	0.0175	0.000	0.0300	0.000
LE	−0.1472	0.000	0.2620	0.000	−0.1431	0.000
_cons	0.3356	0.020	−0.6620	0.000	0.0277	0.834
Observations	2170		2170		2170
Prob > chi2	0.0000		0.0000		0.0000

show that all models remain statistically significant (Wald χ², p < 0.000), thereby affirming the overall model’s validity.

In the baseline model (Model 1), board size and institutional ownership have statistically significant positive effects on Tobin’s Q (β = 0.0473, p < 0.01; β = 0.1557, p < 0.01). State ownership also has a positive effect at the 5% level (β = 0.0513, p = 0.046). By contrast, neither board independence nor board expertise shows a statistically significant effect. These results are directionally consistent with those obtained using ROA, although the significance of board independence diminishes when examining Tobin’s Q.

Model 2, with REM as the dependent variable, yields consistent results: an increase in board size has an adverse effect on REM, whereas board independence and expertise have positive effects; both institutional and state ownership also reduce REM. This observation affirms the fundamental prerequisite for mediation.

Model 3, including REM, shows a negative correlation between REM and Tobin’s Q (β = −0.1536, p < 0.01), indicating that higher REM levels are associated with lower market valuation. Including REM reduces the direct effect of governance—particularly for institutional and state ownership—while board independence becomes statistically significant and negatively correlated (β = −0.0920, p = 0.003). These results suggest partial mediation by board size and institutional ownership and complete mediation by state ownership.

The Durbin–Wu–Hausman test (p = 0.6357 > 0.05) does not reject the null hypothesis of exogeneity for this specification. Overall, replacing ROA with Tobin’s Q yields consistent results, supporting the robustness and stability of the main findings.

To further assess the robustness of the main findings and address concerns about dynamic endogeneity and reverse causality, this study re-estimates the models using the two-step System GMM estimator. As reported in

Table 8. Robustness Two-step System GMM estimation results.

Variables	Model 1 (ROA)	Model 2 (REM)	Model 3 (ROA)
L.ROA	0.300 ***		0.275 ***
	(0.085)		(0.070)
L.REM		0.074
		(0.059)
REM			−0.057 ***
			(0.009)
BS	0.003 *	−0.044 ***	0.000
	(0.001)	(0.009)	(0.001)
BI	−0.029 ***	0.101 *	−0.023 **
	(0.011)	(0.055)	(0.010)
BE	−0.042 *	0.284 *	−0.028
	(0.023)	(0.166)	(0.021)
IN	0.021 **	−0.095 *	0.013 *
	(0.009)	(0.049)	(0.008)
ST	0.010	−0.157 ***	0.002
	(0.008)	(0.047)	(0.007)
FS	0.002 *	0.016 *	0.003 ***
	(0.001)	(0.009)	(0.001)
LE	−0.096 ***	0.229 ***	−0.085 ***
	(0.014)	(0.056)	(0.012)
Year fixed effects	Yes	Yes	Yes
Industry fixed effects	Yes	Yes	Yes
Observations	1736	1736	1736
Firms	434	434	434
Instruments	24	24	25
AR(1) p-value	0.000	0.000	0.000
AR(2) p-value	0.397	0.628	0.306
Hansen p-value	0.322	0.171	0.514
Diff-in-Hansen p-value	0.244	0.093	0.315

Notes: Robust Windmeijer-corrected standard errors are reported in parentheses. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. All models are estimated using a two-step System GMM with collapsed GMM instruments. Year and industry dummies are included but not reported. The number of observations decreases because the dynamic specification uses lagged variables.

, the AR(1) tests are significant, whereas the AR(2) tests are insignificant, indicating no evidence of second-order serial correlation. In addition, the Hansen and Difference-in-Hansen tests generally do not reject the validity and exogeneity of the instrument sets. The System GMM results are broadly consistent with the baseline FGLS findings, suggesting that the main conclusions are not driven solely by the baseline model specification and remain robust after accounting for potential dynamic effects and endogeneity concerns.

REM remains negatively associated with financial performance (β = −0.057, p < 0.01), indicating that firms engaging in higher levels of real earnings management tend to have lower profitability. Furthermore, including REM reduces the magnitude of several governance coefficients, supporting the mediating role identified in the baseline models. The number of observations in the System GMM models decreases because the dynamic specification uses lagged variables.

Overall, the consistency between the FGLS and System GMM estimates suggest that the main conclusions are not driven by model specification and remain robust to potential dynamic effects and endogeneity.

5. Conclusions

This study examines the association between corporate governance and financial performance and explores the mediating role of REM using a sample of 434 non-financial listed firms in Vietnam from 2020 to 2024. Using FGLS as the baseline estimator, the findings indicate that board size, institutional ownership, and state ownership are positively associated with ROA, whereas board independence is negatively associated with ROA, and board financial expertise shows no significant direct relationship. The results also show that corporate governance variables are associated with REM and that REM is negatively associated with financial performance. When REM is included in the model, the direct effects of several governance variables decline, suggesting that REM serves as a mediator in the governance–performance relationship. Robustness analyses using Tobin’s Q and two-step System GMM provide broadly consistent evidence, strengthening the credibility of the main findings.

Regarding contributions, the study provides empirical evidence of the transmission mechanism linking corporate governance, REM, and financial performance in an emerging market context. It also clarifies the specific roles of individual governance components—such as board size, independence, and expertise; institutional ownership; and state ownership. From a practical standpoint, the findings indicate that firms should adjust their board size to an optimal level that maximizes both advisory and monitoring functions. Additionally, they should promote greater participation by institutional investors and professionalize the representation of state ownership to limit REM and improve financial performance. Regulatory authorities and stock exchanges might also consider strengthening disclosure standards and enforcement measures to reduce incentives for manipulating real activities.

6. Limitations and Future Research Directions

This study has several limitations. First, REM is measured using the Roychowdhury (2006) model. Although widely used, this approach may be constrained by the five-year panel and uneven industry distribution, limiting the feasibility of estimating abnormal components at a narrowly defined industry-year level. Future studies with longer panels and larger industry-year samples could estimate REM more granularly and compare alternative REM proxies.

Second, board financial expertise is measured by formal accounting, auditing, and finance-related qualifications disclosed in annual and corporate governance reports. While this improves observability and consistency, it may not capture broader managerial, legal, technological, strategic, or industry-specific expertise. Future research could develop a multidimensional board expertise index.

Third, the study relies on observational data from Vietnamese non-financial listed firms from 2020 to 2024, which limits causal inference and generalizability. Although the Durbin–Wu–Hausman test does not indicate statistically significant endogeneity and the additional two-step System GMM analysis helps mitigate concerns about dynamic endogeneity and reverse causality, these procedures cannot fully rule out omitted-variable bias or establish definitive causal effects. Therefore, the findings should be interpreted as robust conditional associations rather than conclusive causal evidence. Future research should use longer panels, richer identification strategies, and alternative approaches such as instrumental-variable estimation, propensity score matching, impact thresholds for confounding variables, or other selection-on-observables methods to further examine the causal direction of the governance–REM–performance relationship.

Fourth, the baseline models include firm size and financial leverage as primary controls but do not fully account for other potentially relevant firm-level and governance-related factors, such as ownership concentration, audit quality, firm age, and growth opportunities. Although year and industry effects are included in the System GMM robustness analysis, future studies should use richer datasets and broader sets of control variables to further validate the governance–REM–performance relationship.