Managerial Overconfidence and ESG Performance: Financial Policy Channels in an Emerging Market

Abstract

This study examines the relationship between managerial overconfidence and environmental, social, and governance (ESG) performance through firm-level financial policy channels in an emerging-market context. Using panel data from non-financial firms listed on the Indonesia Stock Exchange during 2015–2024, this study adopts a multidimensional channel-based perspective in which managerial overconfidence is indirectly reflected through financing, liquidity, and investment decisions. Fixed-effects estimation with Driscoll–Kraay standard errors is employed as the baseline approach and complemented by lagged specifications, system GMM estimation, alternative measurements, and quantile regressions to assess robustness. The findings suggest that managerial overconfidence does not exert a direct and uniform influence on ESG performance but operates indirectly through heterogeneous financial policy behavior. The financing channel provides weak and unstable evidence, whereas the liquidity channel shows a relatively stronger positive association with ESG performance. The investment channel appears most sensitive to measurement and model specification, indicating that different operationalizations may capture distinct dimensions of managerial overconfidence. This study contributes to the behavioral corporate finance and ESG literature by showing that managerial overconfidence influences sustainability outcomes indirectly through heterogeneous financial policy mechanisms in an emerging market setting while highlighting the importance of temporal dynamics, endogeneity, and measurement sensitivity.

1. Introduction

Despite the increasing integration of environmental, social, and governance (ESG) considerations into corporate strategy, substantial variation in ESG performance persists across firms operating within similar institutional environments (Eccles et al., 2014; Friede et al., 2015). While firms face comparable regulatory pressures and stakeholder expectations, their sustainability outcomes differ considerably. This raises a fundamental question: why do firms exposed to similar external conditions adopt different ESG strategies?

Prior research largely attributes these differences to external institutional factors, such as regulatory frameworks, investor demand, and market pressures (Fatemi et al., 2018). Although these perspectives offer important insights, they provide only a partial explanation. Corporate policies ultimately reflect managerial decisions, which are not always fully aligned with rational benchmarks. Evidence from behavioral corporate finance shows that managerial biases, particularly overconfidence, systematically influence corporate decision-making (Ben-David et al., 2013; Malmendier & Tate, 2005, 2008). As a result, external constraints alone cannot fully explain firm behavior, as managerial beliefs and subjective perceptions also play a critical role.

Managerial overconfidence has been widely documented as a behavioral bias that shapes financial decisions. Overconfident managers tend to overestimate future performance and underestimate risks, leading to more aggressive financial policies, including higher leverage, increased investment, and a stronger reliance on internal financing (Huang et al., 2016; Kaplan et al., 2022). While these effects are well established in relation to financial outcomes (Ben-David et al., 2013; Campbell et al., 2011; Deshmukh et al., 2013; Malmendier & Tate, 2005, 2008), their implications for non-financial strategies, particularly ESG engagement, remain theoretically ambiguous and empirically underdeveloped.

The relationship between managerial overconfidence and ESG performance is therefore theoretically ambiguous. On the one hand, optimistic managerial expectations may support sustainability initiatives when managers anticipate long-term value creation and strategic benefits from ESG engagement. On the other hand, the same behavioral bias may induce financial decisions that constrain ESG investment through greater financial pressure or resource reallocation toward projects with more immediate returns. This tension suggests that managerial overconfidence is unlikely to influence ESG outcomes directly but instead operates indirectly through heterogeneous financial policy channels, whereby financing, liquidity, and investment decisions may either facilitate or constrain ESG engagement.

Recent ESG–behavioral studies have increasingly examined the relationship between managerial overconfidence and ESG performance. Prior evidence suggests that managerial overconfidence may directly influence ESG outcomes through behavioral mechanisms such as risk-taking, managerial optimism, and strategic attention allocation (Wang et al., 2023). While this literature provides important insights regarding the direct behavioral effects of managerial overconfidence on ESG performance, limited attention has been given to the indirect financial policy channels through which managerial overconfidence may shape ESG outcomes. In particular, the role of financing, liquidity, and investment decisions as heterogeneous financial policy channels linking managerial overconfidence to ESG performance remains underexplored, especially in emerging market settings characterized by financial constraints and institutional frictions.

To examine this mechanism, this study adopts a financial policy channel framework in which managerial overconfidence influences ESG performance through three interrelated dimensions: financing decisions (captured through leverage, measured by the debt-to-equity ratio), liquidity management (captured through cash holdings), and investment allocation (captured through investment intensity). Rather than serving as direct behavioral measures, these variables are conceptualized as heterogeneous financial policy transmission channels through which managerial overconfidence may be reflected in corporate decision-making and subsequently influence ESG performance. By jointly incorporating these dimensions, the analysis moves beyond single-proxy approaches by conceptualizing financing, liquidity, and investment decisions as heterogeneous financial policy channels through which managerial overconfidence may indirectly influence ESG performance. In doing so, this study responds to concerns in the measurement literature regarding the difficulty of directly observing managerial overconfidence and the limitations of relying on any single behavioral proxy (Kaplan et al., 2022; Sumiyana et al., 2023).

Our empirical analysis focuses on non-financial firms listed on the Indonesia Stock Exchange over the period 2015–2024. Indonesia provides a relevant emerging-market context characterized by financing constraints, evolving ESG disclosure practices, concentrated ownership structures, and institutional frictions that may shape how managerial overconfidence translates into sustainability outcomes. The findings indicate that the relationship between managerial overconfidence and ESG performance varies across financial policy channels and estimation approaches rather than following a uniform pattern. While liquidity-related financial flexibility tends to support ESG engagement, financing and investment channels exhibit weaker and more specification-sensitive relationships, highlighting the importance of temporal dynamics, endogeneity, and measurement choices in explaining ESG outcomes.

Taken together, the evidence shows that ESG performance is influenced not only by external pressures or firm characteristics but also by how managerial biases are reflected in financial policy decisions. In particular, financial capacity and resource allocation are critical in determining whether firms can sustain ESG engagement over time. Managerial overconfidence may influence ESG performance through multiple and potentially competing mechanisms. On the one hand, overconfident managers may overestimate future returns and engage in excessive investment, including ESG-related initiatives. On the other hand, overconfidence may lead to suboptimal financial decisions, increased leverage, and resource misallocation, which can constrain ESG engagement. These competing mechanisms motivate the examination of financial policy channels as the transmission pathway linking managerial overconfidence to ESG outcomes.

This study contributes to the behavioral corporate finance and ESG literature by advancing a channel-based perspective that links managerial overconfidence to ESG performance through firm-level financial decisions. By moving beyond single-proxy approaches, the analysis conceptualizes financing, liquidity, and investment decisions as heterogeneous financial policy channels through which managerial overconfidence may indirectly influence ESG performance. This study further contributes evidence from an emerging-market context, where financial constraints and institutional frictions may amplify the relationship between managerial overconfidence and ESG performance. Moreover, the magnitude and direction of these relationships may vary across firm-specific and institutional conditions, highlighting the importance of examining heterogeneous financial policy channels rather than assuming a uniform behavioral effect.

Unlike prior studies that primarily examine financial policy variables as independent determinants or alternative behavioral proxies, this study conceptualizes leverage, cash holdings, and investment intensity as heterogeneous financial policy transmission channels through which managerial overconfidence may indirectly influence ESG performance.

2. Literature Review and Hypothesis Development

2.1. Managerial Overconfidence

Managerial overconfidence reflects a behavioral tendency in which executives form overly optimistic expectations about their capabilities and the firm’s future performance (Ben-David et al., 2013; Kaplan et al., 2022). Within the upper-echelons framework, managerial characteristics influence strategic choices because decision-makers interpret opportunities through their own cognitive lenses (Hambrick & Mason, 1984). As a result, corporate decisions are shaped not only by external conditions but also by managerial beliefs and perceptions.

Evidence from behavioral corporate finance shows that overconfident managers tend to adopt more aggressive financial policies than would be expected under rational benchmarks (He et al., 2019; Huang et al., 2016; Malmendier & Tate, 2005, 2008). This behavior is characterized by overestimating expected returns while underestimating risks, thereby often leading to higher investment intensity and greater reliance on debt financing. While such strategies may support expansion, they can also expose firms to financial vulnerability through misallocation of resources and underestimation of downside risks.

Because managerial overconfidence cannot be observed directly, empirical studies typically rely on observable financial decisions as proxies (Ben-David et al., 2013; Kaplan et al., 2022; Malmendier & Tate, 2005). Recent research emphasizes that overconfidence is inherently multidimensional and cannot be adequately captured by a single indicator (Kaplan et al., 2022; Sumiyana et al., 2023).

Building on this view, this study adopts a multidimensional channel-based approach in which leverage, cash holdings, and investment intensity are conceptualized not as direct behavioral measures but as heterogeneous financial policy channels through which managerial overconfidence may be reflected in corporate decision-making.

2.2. Managerial Overconfidence and ESG Performance

ESG performance reflects the extent to which firms integrate environmental, social, and governance considerations into their strategic and operational decisions (Eccles et al., 2014; Friede et al., 2015). Compared to conventional investments, ESG initiatives typically involve longer time horizons, higher uncertainty, and delayed financial returns, requiring both financial capacity and sustained managerial commitment.

From a behavioral perspective, managerial overconfidence may influence ESG performance through competing mechanisms. On the one hand, optimistic managerial expectations may support ESG initiatives when managers anticipate long-term value creation, reputational benefits, or strategic advantages from sustainability engagement (Ferrell et al., 2016; Liang & Renneboog, 2017; Wang et al., 2023). On the other hand, the same behavioral bias may induce financial decisions that constrain ESG engagement through greater risk taking, financial pressure, or inefficient resource allocation toward projects with more immediate returns. These opposing effects suggest that managerial overconfidence is unlikely to exert a direct and uniform effect on ESG performance. Instead, its influence operates through financial decision-making channels, particularly those related to risk-taking, financing structure, and resource allocation (Burkhard et al., 2023).

Recent studies have increasingly examined the behavioral determinants of ESG performance, particularly the role of managerial characteristics. However, prior research predominantly focuses on direct behavioral effects and provides limited insight into the financial mechanisms through which managerial overconfidence translates into ESG outcomes. This study addresses this gap by explicitly conceptualizing leverage, cash holdings, and investment intensity as heterogeneous financial policy transmission channels through which managerial overconfidence may indirectly influence ESG performance.

2.3. Financial Policy Channels of Managerial Overconfidence

Managerial overconfidence may influence ESG performance through financial policy decisions that operate as heterogeneous transmission channels of managerial overconfidence. In this framework, leverage, cash holdings, and investment intensity are conceptualized not as direct behavioral proxies but as financial policy channels through which managerial overconfidence is reflected in corporate decisions and subsequently influences ESG outcomes. The magnitude and direction of these relationships may further vary across firm-specific and institutional conditions, particularly in emerging markets characterized by financing constraints, evolving governance structures, and institutional frictions relative to developed economies.

2.3.1. Leverage

Managers with higher levels of overconfidence tend to underestimate financial risks and overestimate expected cash flows, thereby increasing their willingness to rely on debt financing (Huang et al., 2016). From this perspective, debt may be perceived as less costly than it actually is, leading to higher leverage.

While leverage can support firm growth, excessive reliance on debt imposes financial constraints. Firms with higher leverage must prioritize short-term obligations, limiting their ability to invest in long-term projects such as ESG initiatives. Since sustainability investments require sustained commitment and do not generate immediate financial returns, highly leveraged firms are more likely to allocate resources toward debt servicing rather than ESG-related expenditures (Ferrell et al., 2016; Liang & Renneboog, 2017).

These constraints are particularly pronounced in emerging markets, where access to capital is more limited and financing costs are higher (Boubakri et al., 2021). Leverage therefore represents one important financing channel through which managerial overconfidence may constrain ESG performance. Overconfident managers tend to overestimate expected returns and underestimate financing risks, potentially increasing reliance on debt financing. However, greater leverage may constrain ESG investment by increasing financial pressure and prioritizing short-term obligations over long-term sustainability initiatives.

H1. 

Managerial overconfidence, captured through leverage (debt-to-equity ratio, DER), is negatively associated with ESG performance.

2.3.2. Cash Holdings

Overconfident managers often prefer internal financing, driven by the belief that external capital markets undervalue their firms. As a result, they tend to maintain higher cash balances, providing financial flexibility and reducing reliance on external funding.

ESG engagement requires substantial financial resources and long-term commitment. Firms with greater internal liquidity are better positioned to sustain ESG initiatives without facing financial constraints, particularly given the delayed and uncertain nature of sustainability returns (Ferrell et al., 2016; Liang & Renneboog, 2017; Uyar et al., 2023; Wasiuzzaman et al., 2022). This role of internal liquidity is particularly important in emerging markets, where access to external capital may be limited or costly. In such settings, cash holdings serve as an important liquidity channel that enables firms to sustain ESG engagement, absorb financial shocks, and maintain long-term stakeholder commitments.

Managerial overconfidence may influence liquidity management through competing mechanisms. While optimistic managers may maintain lower precautionary cash reserves due to confidence in future cash flows, firms with greater financial flexibility are generally better positioned to sustain ESG-related activities and long-term sustainability commitments. Accordingly, cash holdings may represent one important liquidity channel through which managerial overconfidence indirectly influences ESG performance.

H2. 

Managerial overconfidence, captured through cash holdings, is positively associated with ESG performance.

2.3.3. Investment

Investment intensity represents a financial policy channel through which managerial overconfidence may influence corporate resource allocation toward growth opportunities and long-term initiatives. Overconfident managers are more likely to adopt aggressive investment strategies due to optimistic expectations regarding future returns (Burkhard et al., 2023; Malmendier & Tate, 2005, 2008). However, such investment behavior may not necessarily align with sustainability objectives and may crowd out ESG-related investment through inefficient resource allocation.

Although ESG initiatives can be viewed as long-term strategic investments that enhance firm value (McWilliams & Siegel, 2001), investment decisions do not necessarily align with sustainability objectives. Overconfident managers may prioritize expansion-oriented projects or engage in overinvestment behavior, allocating resources to initiatives that offer more immediate or uncertain returns. This potential misallocation of resources may crowd out investments in ESG activities, particularly in emerging markets where financial constraints and institutional frictions shape corporate priorities (Boubakri et al., 2021).

Investment intensity therefore represents one important investment channel through which managerial overconfidence may indirectly influence ESG performance. Overconfident managers may exhibit stronger investment tendencies due to overly optimistic expectations of future opportunities. However, such investment behavior may not necessarily align with sustainability objectives and may crowd out ESG-related investment through inefficient resource allocation.

H3. 

Managerial overconfidence, captured through investment intensity, is negatively associated with ESG performance.

3. Methodology

3.1. Data

Our empirical analysis is based on panel data from non-financial firms listed on the Indonesia Stock Exchange over the period 2015–2024. Financial institutions are excluded due to their distinct regulatory environment and capital structure, which may not be directly comparable to those of non-financial firms. The data are obtained from the Refinitiv database, which provides firm-level financial and ESG information.

The initial dataset comprises 740 firm-year observations. However, the final estimation sample is unbalanced due to incomplete availability of ESG and financial variables across firms and years. After applying listwise deletion based on model-specific data requirements, the effective sample ranges from 287 to 330 firm-year observations across 52 firms, with variation driven by variable availability and differences in estimation specifications. This variation reflects differences in model requirements rather than systematic sample selection bias. The use of an unbalanced panel is appropriate in this context, as excluding firms with partial observations would substantially reduce sample coverage and potentially omit meaningful ESG variation in an emerging market setting.

Prior to estimation, all variables are converted into numeric format, with non-numeric entries treated as missing values. To reduce the influence of extreme observations, continuous variables are winsorized at the 1st and 99th percentiles. This treatment is applied consistently across the baseline and alternative measurement specifications to improve estimation stability and mitigate the influence of outliers. Nevertheless, negative debt-to-equity ratio (DER) values are intentionally retained after winsorization to preserve economically meaningful observations associated with firms experiencing negative equity. In addition, selected variables are transformed where necessary to address distributional concerns, while variables with greater economic interpretability in their original form are retained accordingly.

Although the sample remains relatively limited due to ESG data availability in an emerging-market context, the use of multiple estimation approaches and robustness checks helps strengthen the reliability of the empirical findings.

3.2. Variable Measurement

3.2.1. Dependent Variable

The dependent variable in this study is ESG performance, measured using the Refinitiv ESG Combined Score which ranges from 0 to 100. This composite index is constructed from three core pillars, namely environmental, social, and governance, based on a broad set of firm-level indicators. In addition, the ESG Combined Score incorporates an ESG controversy overlay, which adjusts the score to account for negative events or corporate controversies. Accordingly, the measure captures both sustainability performance and ESG-related risk exposure at the firm level. Higher ESG scores indicate stronger sustainability performance and broader integration of ESG-related practices into corporate strategy and operations (Eccles et al., 2014).

3.2.2. Explanatory Variables: Financial Policy Channels

Managerial overconfidence is the primary explanatory construct. As a latent behavioral trait, it cannot be observed directly and is therefore inferred from observable corporate decisions. Consistent with behavioral corporate finance theory, managerial overconfidence is reflected in financial policy choices that capture managerial expectations, risk perceptions, and beliefs about future firm performance (Ben-David et al., 2013; Malmendier & Tate, 2005).

This study adopts a financial policy channel approach to examine managerial overconfidence through three interrelated dimensions: financing, liquidity, and investment decisions. These dimensions are operationalized using leverage (DER), cash holdings (CashH), and investment intensity (Invest), respectively. Rather than serving as direct behavioral measures, these variables are conceptualized as heterogeneous financial policy transmission channels through which managerial overconfidence is indirectly reflected in corporate financial decisions and subsequently influences ESG performance. Although these financial policy variables may also capture broader firm-level conditions, including financial constraints, liquidity availability, and investment capacity, this study conceptualizes them as behavioral transmission channels through which managerial overconfidence may indirectly shape financial decisions and sustainability outcomes. Accordingly, these variables are not interpreted as pure behavioral measures but rather as indirect manifestations of managerial overconfidence embedded in corporate financial policies.

Leverage represents the financing channel and is measured using the debt-to-equity ratio. Equation (1) defines leverage as follows:

$${\mathrm{DER}}_{\mathrm{it}}={\displaystyle \frac{{\mathrm{TotalDebt}}_{\mathrm{it}}}{{\mathrm{TotalEquity}}_{\mathrm{it}}}} ,$$

(1)

where ${\mathrm{DER}}_{\mathrm{it}}$ denotes the leverage of firm i in year t, measured as the debt-to-equity ratio; ${\mathrm{TotalDebt}}_{\mathrm{it}}$ represents total debt, including both short-term and long-term borrowings; and ${\mathrm{TotalEquity}}_{\mathrm{it}}$ denotes total shareholder equity.

From a behavioral perspective, overconfident managers tend to overestimate future cash flows and underestimate financial risks, leading to more aggressive capital structures and higher leverage (Huang et al., 2016). Accordingly, leverage represents a financing channel through which managerial overconfidence may influence ESG performance. Negative DER values are retained to preserve the economic interpretation of firms with negative equity positions.

Cash holdings represent the liquidity channel and are measured as cash and cash equivalents scaled by total assets. Equation (2) defines cash holdings as follows:

$${\mathrm{CashH}}_{\mathrm{it}}={\displaystyle \frac{{\mathrm{CashAndCashEquivalents}}_{\mathrm{it}} }{{\mathrm{TotalAssets}}_{\mathrm{it}}}},$$

(2)

where ${\mathrm{CashH}}_{\mathrm{it}}$ denotes cash holdings of firm i in year t, ${\mathrm{CashAndCashEquivalents}}_{\mathrm{it}}$ represents cash and highly liquid short-term investments, and ${\mathrm{TotalAssets}}_{\mathrm{it}}$ denotes total assets.

Cash holdings capture internal liquidity and financial flexibility. Behavioral theory suggests that overconfident managers prefer internal financing because they believe external markets undervalue the firm (Malmendier & Tate, 2005). As such, cash reserves reflect managerial expectations about future investment opportunities and serve as a liquidity channel that may influence the firm’s capacity to allocate resources to ESG-related activities.

Investment intensity represents the investment channel and is measured as capital expenditure scaled by total assets. Equation (3) defines investment intensity as follows:

$${\mathrm{Invest}}_{\mathrm{it}}={\displaystyle \frac{{\mathrm{CapitalExpenditure}}_{\mathrm{it}}}{{\mathrm{TotalAssets}}_{\mathrm{it}}}},$$

(3)

where ${\mathrm{Invest}}_{\mathrm{it}}$ denotes investment intensity of firm i in year t, ${\mathrm{CapitalExpenditure}}_{\mathrm{it}}$ represents firm investment in long-term assets, and ${\mathrm{TotalAssets}}_{\mathrm{it}}$ denotes total assets.

Investment intensity reflects firms’ capital allocation decisions and is closely associated with managerial optimism and risk-taking behavior. Overconfident managers are more likely to engage in aggressive investment strategies due to overly optimistic expectations of project returns (Ben-David et al., 2013; Malmendier & Tate, 2005).

Unlike financing and liquidity decisions, investment allocation does not necessarily align with sustainability objectives. Therefore, investment intensity represents an important behavioral channel through which managerial overconfidence may influence ESG performance.

3.2.3. Control Variables

Several firm-level characteristics are included as control variables, as summarized in Table 1. Consistent with the prior literature, these variables are commonly associated with ESG performance and are incorporated to reduce omitted variable bias and isolate the effects of financial policy channels.

Table 1. Definition and measurement of control variables.

Variable	Measurement
Firm size	${\mathrm{Fsize}}_{\mathrm{i}\mathrm{t}}=\mathrm{l}\mathrm{n}\left({\mathrm{TotalAssets}}_{\mathrm{i}\mathrm{t}}\right)$
	Firm size is measured as the natural logarithm of total assets.
Firm age	Firm age is measured as the number of years since the firm’s establishment and is retained in its actual form in the empirical analysis.
Profitability	${\mathrm{ROA}}_{\mathrm{i}\mathrm{t}}={\displaystyle \frac{{\mathrm{NetIncome}}_{\mathrm{i}\mathrm{t}}}{{\mathrm{TotalAssets}}_{\mathrm{i}\mathrm{t}}}}$
	Return on assets (ROA) is measured as net income divided by total assets and obtained from Refinitiv using the indicator TR.ROAActValue, expressed in percentage terms.
Growth opportunity	${\mathrm{Growth}}_{\mathrm{it}}={\displaystyle \frac{{\mathrm{MarketValue}}_{\mathrm{it}}+{\mathrm{TotalDebt}}_{\mathrm{it}}}{{\mathrm{TotalAssets}}_{\mathrm{it}}}}$
	Growth opportunities are proxied using Tobin’s Q and transformed using ln(1 + Growth) to reduce skewness and accommodate zero values.
Operating cash flow	${\mathrm{OCF}}_{\mathrm{it}}={\displaystyle \frac{{\mathrm{OperatingCashFlow}}_{\mathrm{it}}}{{\mathrm{TotalAssets}}_{\mathrm{it}}}}$
	Operating cash flow is measured as operating cash flow scaled by total assets.
Board Size	${\mathrm{Bsize}}_{\mathrm{it}}=\ln ({\mathrm{NumberOfDirectors}}_{\mathrm{it}})$
	Board size is measured as the natural logarithm of the number of directors.
CEO Duality	CEO duality (CEOD) is measured as a dummy variable equal to 1 if the CEO simultaneously serves as board chair, and 0 otherwise.

Notes: i and t denote firm and year, respectively. TotalAssets represents total assets; NetIncome represents net income after tax; TotalDebt denotes total liabilities; MarketValue refers to the market value of equity; OperatingCashFlow denotes operating cash flow from operations; and NumberOfDirectors represents the number of board directors.

4. Empirical Model and Estimation

This study employs panel data regression with firm-level fixed effects to examine the relationship between managerial overconfidence and ESG performance. The fixed-effects specification accounts for unobserved, time-invariant firm characteristics that may systematically influence ESG outcomes.

The Hausman specification test indicates that the fixed-effects model is preferred over the random-effects model (χ² = 28.95, p = 0.035), suggesting that unobserved firm-specific effects are correlated with the explanatory variables. Therefore, the fixed-effects estimator is considered more appropriate for the baseline analysis (Hausman, 1978).

Diagnostic tests indicate violations of several classical panel data assumptions. The Wooldridge test reveals significant first-order autocorrelation (F = 33.858, p < 0.01), while the Modified Wald test confirms the presence of groupwise heteroskedasticity (χ² = 9242.72, p < 0.01). The multicollinearity test reports an overall mean VIF of 4.24, indicating that multicollinearity remains within an acceptable range, although firm size and growth variables exhibit relatively high individual VIF values. Accordingly, this study employs fixed-effects estimation with Driscoll–Kraay standard errors, which are robust to heteroskedasticity, serial correlation, and cross-sectional dependence (Driscoll & Kraay, 1998).

Rather than relying on a single behavioral indicator, this study adopts a financial policy-channel approach to examine managerial overconfidence. Separate model specifications are estimated for leverage, cash holdings, and investment intensity to capture the financing, liquidity, and investment channels through which managerial overconfidence may influence ESG performance. The three channels are initially estimated separately to isolate their distinct relationships with ESG performance and preserve firm year observations given differences in data availability across the unbalanced sample. Subsequently, a joint specification is estimated to account for the interdependence among financing, liquidity, and investment decisions. Year fixed effects are included to account for common macroeconomic shocks, regulatory changes, and time-specific ESG-related developments that may simultaneously affect firms over time.

Equation (4) presents the baseline empirical model for separate channel estimation:

$$\begin{array}{cc}\mathrm{E}\mathrm{S}{\mathrm{G}}_{\mathrm{i}\mathrm{t}}=& \mathsf{\alpha }+{\mathsf{\beta }}_1\mathrm{O}{\mathrm{C}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_2\mathrm{F}\mathrm{s}\mathrm{i}\mathrm{z}{\mathrm{e}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_3\mathrm{F}\mathrm{a}\mathrm{g}{\mathrm{e}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_4\mathrm{R}\mathrm{O}{\mathrm{A}}_{\mathrm{i}\mathrm{t}}\hfill \\ & +{\mathsf{\beta }}_5\mathrm{l}\mathrm{n}\mathrm{G}\mathrm{r}\mathrm{o}\mathrm{w}\mathrm{t}{\mathrm{h}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_6\mathrm{O}\mathrm{C}{\mathrm{F}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_7\mathrm{B}\mathrm{s}\mathrm{i}\mathrm{z}{\mathrm{e}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_8\mathrm{C}\mathrm{E}\mathrm{O}{\mathrm{D}}_{\mathrm{i}\mathrm{t}}\hfill \\ & +{\mathsf{\gamma }}_{\mathrm{i}}+{\mathsf{\delta }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{it}} \hfill \end{array}$$

(4)

where $\mathrm{E}\mathrm{S}{\mathrm{G}}_{\mathrm{i}\mathrm{t}}$ denotes ESG performance, and $\mathrm{O}{\mathrm{C}}_{\mathrm{i}\mathrm{t}}$ represents the financial policy channel variable, estimated separately as $\mathrm{D}\mathrm{E}{\mathrm{R}}_{\mathrm{i}\mathrm{t}}$, $\mathrm{C}\mathrm{a}\mathrm{s}\mathrm{h}{\mathrm{H}}_{\mathrm{i}\mathrm{t}}$, or $\mathrm{I}\mathrm{n}\mathrm{v}\mathrm{e}\mathrm{s}{\mathrm{t}}_{\mathrm{i}\mathrm{t}}$. Control variables are defined in Table 1. The terms ${\mathsf{\gamma }}_{\mathrm{i}}$ and ${\mathsf{\delta }}_{\mathrm{t}}$ denote firm and year fixed effects, respectively, while ${\mathsf{\epsilon }}_{\mathrm{it}}$ represents the idiosyncratic error term.

Equation (5) presents the joint channel specification in which leverage, cash holdings, and investment intensity are estimated simultaneously:

$$\begin{array}{cc}\mathrm{E}\mathrm{S}{\mathrm{G}}_{\mathrm{i}\mathrm{t}}=& \mathsf{\alpha }+{\mathsf{\beta }}_1\mathrm{D}\mathrm{E}{\mathrm{R}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_2\mathrm{C}\mathrm{a}\mathrm{s}\mathrm{h}{\mathrm{H}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_3\mathrm{I}\mathrm{n}\mathrm{v}\mathrm{e}\mathrm{s}{\mathrm{t}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_4\mathrm{F}\mathrm{s}\mathrm{i}\mathrm{z}{\mathrm{e}}_{\mathrm{i}\mathrm{t}}\hfill \\ & +{\mathsf{\beta }}_5\mathrm{F}\mathrm{a}\mathrm{g}{\mathrm{e}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_6\mathrm{R}\mathrm{O}{\mathrm{A}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_7\mathrm{l}\mathrm{n}\mathrm{G}\mathrm{r}\mathrm{o}\mathrm{w}\mathrm{t}{\mathrm{h}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_8\mathrm{O}\mathrm{C}{\mathrm{F}}_{\mathrm{i}\mathrm{t}}\hfill \\ & +{\mathsf{\beta }}_9\mathrm{Bsiz}{\mathrm{e}}_{\mathrm{it}}+{\mathsf{\beta }}_{10}\mathrm{CEO}{\mathrm{D}}_{\mathrm{it}}+{\mathsf{\gamma }}_{\mathrm{i}}+{\mathsf{\delta }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{it}.}\hfill \end{array}$$

(5)

Lagged specifications, system GMM estimation, alternative measurement specifications, and quantile regressions are further employed as complementary robustness approaches to assess the stability of the findings, address potential endogeneity, and account for ESG persistence. All estimations are performed using Stata/MP 18.0.

5. Results and Discussion

5.1. Descriptive Statistics

Table 2 reports the descriptive statistics of the variables used in this study. The average ESG score is 48.108, indicating a moderate level of ESG performance among non-financial firms listed on the Indonesia Stock Exchange (IDX). The substantial variation in ESG scores (SD = 18.388), ranging from 11.276 to 86.015, suggests considerable heterogeneity in corporate sustainability practices across firms, indicating that ESG integration remains uneven in the Indonesian market.

Table 2. Descriptive statistics.

Variable	Observation	Mean	SD	Min	Max
ESG	499	48.108	18.388	11.276	86.015
DER	688	0.895	1.562	−1.955	12.078
CashH	704	0.114	0.087	0.004	0.415
Invest	687	0.057	0.055	0.001	0.258
Fsize	704	14.774	4.011	5.718	19.518
Fage	585	28.383	16.821	1.997	85.099
ROA	521	7.966	8.362	−11.455	39.268
lnGrowth	653	2.819	3.537	0.278	11.161
OCF	704	0.101	0.106	−0.126	0.487
Bsize	499	1.696	0.378	0.693	2.708
CEOD	740	0.027	0.162	0.000	1.000
LEVASSET	688	0.290	0.274	0.000	1.712
CashHAsset	704	0.141	0.132	0.004	0.708
InvestAsset	689	0.127	0.248	−0.203	1.587

Notes: This table reports descriptive statistics for all variables. The number of observations varies due to data availability. All continuous variables are winsorized at the 1st and 99th percentiles. ESG denotes environmental, social, and governance score; DER denotes debt-to-equity ratio; CashH denotes cash holdings; Invest denotes investment intensity; Fsize denotes firm size; Fage denotes firm age; ROA denotes return on assets; lnGrowth denotes growth opportunities; OCF denotes operating cash flow; Bsize denotes board size; CEOD denotes CEO duality; LEVASSET denotes total debt scaled by total assets; CashHAsset denotes cash holdings relative to non-cash assets; and InvestAsset denotes asset growth-based investment intensity.

Leverage exhibits substantial dispersion (mean = 0.895; SD = 1.562), with values ranging from −1.955 to 12.078, reflecting heterogeneity in firms’ capital structures and financial risk exposure. The presence of negative debt-to-equity-ratio values is intentionally retained, as these observations reflect firms with negative equity positions and preserve economically meaningful variation in financing conditions. Cash holdings are relatively stable (mean = 0.114; SD = 0.087), indicating broadly consistent liquidity management across firms. Investment intensity shows moderate variation (mean = 0.057; SD = 0.055), suggesting differences in firms’ capital allocation patterns.

The alternative measurements display broadly comparable distributions. Leverage scaled by total assets reports a mean of 0.290 (SD = 0.274), while cash holdings relative to non-cash assets record a mean of 0.141 (SD = 0.132). Asset growth-based investment intensity exhibits greater dispersion (mean = 0.127; SD = 0.248), indicating wider variation across firms.

Control variables also display meaningful variation. Profitability varies considerably (mean = 7.966; SD = 8.362), while firm size and growth opportunities show substantial dispersion across firms, with mean values of 14.774 and 2.819, respectively. CEO duality remains relatively rare (mean = 0.027), indicating that most firms maintain separate executive and board leadership roles. Overall, the distribution of the variables supports the suitability of panel data estimation and provides sufficient cross-sectional variation for empirical analysis.

5.2. Correlation Analysis

Table 3 presents the Pearson correlation matrix for the baseline variables. Although several correlations are statistically significant, the overall pattern does not indicate severe multicollinearity, consistent with the multicollinearity diagnostics reported in Section 4.

Table 3. Pearson correlation matrix for baseline model.

Variable	ESG	DER	CashH	Invest	Fsize	Fage	ROA	lnGrowth	OCF	Bsize	CEOD
ESG	1.0000
DER	−0.1474 ***	1.0000
CashH	0.2788 ***	−0.2509 ***	1.0000
Invest	−0.0655	0.0528	−0.1326 ***	1.0000
Fsize	−0.1991 ***	0.0545	−0.0930 **	0.0772 **	1.0000
Fage	−0.1597 ***	−0.0007	−0.1399 ***	0.2086 ***	0.1384 ***	1.0000
ROA	0.1450 ***	−0.2522 ***	0.2865 ***	0.0523	0.0053	−0.0834 *	1.0000
lnGrowth	0.2492 ***	−0.0493	0.1665 ***	−0.0283	−0.9534 ***	−0.1374 ***	−0.0021	1.0000
OCF	0.1812 ***	−0.1233 ***	0.2686 ***	0.1824 ***	0.0499	0.0211	0.7837 ***	0.0046	1.0000
Bsize	0.3300 ***	0.0030	0.0743 *	0.1191 ***	−0.0352	−0.2555 ***	−0.0381	0.1391 ***	0.0953 **	1.0000
CEOD	−0.1801 ***	−0.0279	0.0019	0.0270	0.0729 *	0.0790 *	0.2089 ***	−0.0811 **	0.1596 ***	−0.0680	1.0000

Notes: This table reports Pearson correlation coefficients for the baseline model variables. Alternative measurements are excluded to maintain table readability and avoid redundancy. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. ESG denotes environmental, social, and governance score; DER denotes debt-to-equity ratio; CashH denotes cash holdings; Invest denotes investment intensity; Fsize denotes firm size; Fage denotes firm age; ROA denotes return on assets; lnGrowth denotes growth opportunities; OCF denotes operating cash flow; Bsize denotes board size; and CEOD denotes CEO duality.

ESG performance is negatively correlated with leverage (DER; r = −0.1474, p < 0.01), firm age (r = −0.1597, p < 0.01), and CEO duality (r = −0.1801, p < 0.01), while it is positively correlated with cash holdings (CashH; r = 0.2788, p < 0.01), profitability (r = 0.1450, p < 0.01), growth opportunities (r = 0.2492, p < 0.01), operating cash flow (r = 0.1812, p < 0.01), and board size (r = 0.3300, p < 0.01). The correlation between ESG and investment intensity (Invest) is negative but statistically insignificant (r = −0.0655, p > 0.10), suggesting no statistically meaningful bivariate association with ESG performance.

Overall, the observed correlations provide preliminary evidence of heterogeneous associations across financial policy channels and ESG performance, which are further evaluated through multivariate panel regression analysis.

5.3. Empirical Results: Financial Policy Channels and ESG Performance

Table 4 presents the baseline fixed-effects estimation results using Driscoll–Kraay standard errors. The analysis considers leverage (DER), cash holdings (CashH), and investment intensity (Invest) as financing, liquidity, and investment channels associated with managerial overconfidence. Each channel is initially estimated separately to isolate its relationship with ESG performance, followed by a joint specification incorporating all channels simultaneously.

Table 4. Fixed-effects estimation results for financial policy channels.

Variables	(1)	(2)	(3)	(4)
	DER	CashH	Invest	Joint Channels
DER	−0.777	–	–	−0.303
	(0.430)			(0.471)
CashH	–	26.145 **	–	23.053 *
		(10.441)		(11.460)
Invest	–	–	23.908 **	15.774 *
			(8.661)	(8.068)
Control Variables
Fsize	8.085 ***	7.462 ***	7.430 ***	7.188 ***
	(1.199)	(1.229)	(1.038)	(1.093)
Fage	1.216 **	0.782	1.938 ***	1.425 ***
	(0.434)	(0.634)	(0.390)	(0.424)
ROA	−0.268 ***	−0.232 ***	−0.201 **	−0.265 ***
	(0.058)	(0.068)	(0.065)	(0.062)
lnGrowth	0.083	0.305	0.072	0.052
	(1.545)	(1.559)	(1.698)	(1.509)
OCF	16.197 **	8.978	10.738	8.524
	(6.732)	(5.975)	(7.887)	(6.141)
Bsize	−4.337 ***	−4.989 ***	−4.326 ***	−4.724 ***
	(1.247)	(1.299)	(1.127)	(1.102)
CEOD	−15.646 ***	−14.607 ***	−12.153 **	−14.267 ***
	(3.378)	(3.524)	(3.926)	(3.497)
Fixed Effects
Firm FE	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
Observations	330	330	322	322
Number of firms	52	52	52	52
R2 (within)	0.5648	0.5758	0.5332	0.5506

Notes: Driscoll–Kraay standard errors are reported in parentheses. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. All models include firm fixed effects and year fixed effects. The joint channel estimation includes leverage (DER), cash holdings (CashH), and investment intensity (Invest) simultaneously.

Leverage exhibits a negative but statistically insignificant association with ESG performance in both the separate and joint baseline estimations (β = −0.777, p > 0.10; β = −0.303, p > 0.10, respectively), suggesting no statistically meaningful relationship within the baseline specification. Although the negative direction is broadly consistent with prior studies arguing that financially constrained firms may allocate fewer resources to long-term sustainability initiatives (Ferrell et al., 2016; Liang & Renneboog, 2017), the lack of statistical significance indicates limited empirical support for H1.

Cash holdings show a positive and statistically significant association with ESG performance in both the separate and joint baseline estimations (β = 26.145, p < 0.05; β = 23.053, p < 0.10, respectively), supporting the financial flexibility argument. Firms with greater internal liquidity appear better positioned to sustain ESG-related engagement by reducing dependence on external financing and accommodating the long-term nature of sustainability investments (Cheng et al., 2014; Ferrell et al., 2016). Accordingly, the findings provide partial support for H2.

Investment intensity exhibits a positive and statistically significant association with ESG performance in both the separate and joint baseline estimations (β = 23.908, p < 0.05; β = 15.774, p < 0.10, respectively). The positive coefficient suggests that higher investment activity is associated with stronger ESG performance within the baseline specification. Accordingly, the baseline findings do not support H3.

Taken together, the baseline results indicate that the relationship between managerial overconfidence and ESG performance is channel-dependent. Liquidity and investment channels are positively associated with ESG performance, whereas the financing channel represented by leverage remains negative but statistically insignificant. These patterns are further examined through lagged specifications, system GMM estimation, alternative measurement specifications, and quantile regressions.

5.4. Robustness Tests

To assess the robustness of the baseline findings, this study employs several complementary approaches, including lagged specifications, system GMM estimation, alternative measurements, and quantile regressions. Lagged specifications are used to capture temporal dynamics, while system GMM addresses potential endogeneity and ESG persistence. Quantile regressions examine whether the effects of financial policy variables vary across ESG distribution levels. In addition, alternative measurements are employed by defining leverage as total debt scaled by total assets, cash holdings relative to non-cash assets, and investment intensity using asset growth.

Overall, the robustness results indicate that the baseline findings are sensitive to model specification, dynamic adjustments, and measurement choices. While several associations observed in the baseline estimations weaken under lagged and system GMM specifications, the overall evidence suggests that the relationship between managerial overconfidence and ESG performance remains indirect, channel dependent, and context specific, with varying degrees of persistence across estimation approaches.

5.4.1. Lagged Specification

The lagged specification provides additional insight into the temporal persistence of financial policy effects on ESG performance by introducing a one period lag of the explanatory variables. Table 5 reports the corresponding estimation results.

Table 5. Lagged specification results.

Variables	(1)	(2)	(3)	(4)
	L.DER	L.CashH	L.Invest	Joint Lagged Channels
L.DER	−0.858 *	–	–	−0.366
	(0.412)			(0.374)
L.CashH	–	19.076	–	10.126
		(11.856)		(7.274)
L.Invest	–	–	23.131 ***	18.045 **
			(5.938)	(7.780)
Control Variables
Fsize	6.399 ***	6.721 ***	6.360 ***	6.301 ***
	(1.179)	(1.284)	(1.069)	(1.090)
Fage	1.811 ***	1.505 ***	1.559 ***	1.443 **
	(0.318)	(0.403)	(0.379)	(0.519)
ROA	−0.158 **	−0.214 **	−0.160 **	−0.205 ***
	(0.068)	(0.079)	(0.059)	(0.059)
lnGrowth	−0.892	0.317	0.099	0.030
	(1.703)	(2.148)	(1.787)	(1.705)
OCF	14.006	14.122	11.067	15.470 *
	(8.121)	(9.754)	(7.120)	(7.437)
Bsize	−4.540 ***	−4.742 ***	−4.634 ***	−4.628 ***
	(1.250)	(0.957)	(0.875)	(0.975)
CEOD	−14.386 **	−13.834 **	−6.224 ***	−9.085 ***
	(6.083)	(5.685)	(1.086)	(2.189)
Fixed Effects
Firm FE	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
Observations	305	307	298	297
Number of firms	52	52	52	52
R2 (within)	0.5139	0.5126	0.4873	0.4936

Notes: Driscoll–Kraay standard errors are reported in parentheses. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. All specifications include firm and year fixed effects.

Lagged leverage retains a negative association with ESG performance in both the separate and joint estimations (β = −0.858, p < 0.10; β = −0.366, p > 0.10, respectively), broadly consistent with the baseline findings. However, the loss of significance in the joint specification suggests that the financing effect weakens once other financial policy channels are jointly considered. This result is broadly consistent with studies suggesting that financing constraints may influence sustainability investment only gradually (Flammer, 2021; Krueger et al., 2020). Accordingly, the lagged specification provides only limited support for H1.

Lagged cash holdings remain positively associated with ESG performance but become statistically insignificant in both the separate and joint estimations (β = 19.076, p > 0.10; β = 10.126, p > 0.10, respectively). Relative to the baseline results, the weakening significance suggests that liquidity effects may not persist over time, although the positive coefficients remain consistent with the financial flexibility argument that internal liquidity may facilitate ESG engagement (Cheng et al., 2014; Ferrell et al., 2016). Thus, support for H2 weakens after incorporating temporal dynamics.

In contrast, lagged investment intensity remains positively and statistically significant in both the separate and joint estimations (β = 23.131, p < 0.01; β = 18.045, p < 0.05, respectively), closely resembling the baseline findings. This persistence suggests that investment-related decisions may influence ESG performance over time. While the behavioral finance literature often associates managerial overconfidence with excessive investment (Ben-David et al., 2013; Malmendier & Tate, 2005), the findings indicate that investment expansion may remain aligned with ESG engagement. Accordingly, the lagged specification does not support H3.

Overall, the lagged estimations indicate heterogeneous persistence across financial policy channels. While the leverage and liquidity effects weaken after incorporating temporal dynamics, the investment channel remains positive and statistically significant.

5.4.2. System GMM Estimation

The system GMM estimator is employed to address potential endogeneity, reverse causality, and the dynamic persistence of ESG performance by incorporating the lagged dependent variable, L.ESG, into the model specification. Following Roodman (2009), the instrument matrix is collapsed and restricted to lag levels 2–3 to mitigate instrument proliferation. All estimations are conducted using two-step robust-system GMM with finite sample correction. In the dynamic specification, the lagged ESG variable and financial policy variables are treated as endogenous to address reverse causality and simultaneity concerns, while control variables are treated as predetermined, consistent with the study’s identification strategy.

The diagnostic results reported in Table 6 indicate that the models are generally well specified. The coefficient on L.ESG remains positive and significant across all specifications (β ranging from 0.598 to 0.813), indicating persistence in ESG performance over time. The AR(2) tests remain statistically insignificant (p = 0.144–0.420), suggesting no evidence of second-order serial correlation, while the Hansen (p = 0.080–0.811) and Sargan (p = 0.064–0.852) tests indicate broadly valid instruments. Moreover, the number of instruments (22–28) remains below the number of firms (52), reducing concerns regarding instrument proliferation.

Table 6. System GMM results.

Variables	(1)	(2)	(3)	(4)
	L.DER GMM	L.CashH GMM	L.Invest GMM	Joint GMM
L.ESG	0.776 ***	0.659 **	0.598 **	0.813 **
	(0.214)	(0.301)	(0.262)	(0.371)
DER	0.142	–	–	−0.945
	(1.167)			(1.235)
CashH	–	−45.235	–	−30.681
		(36.233)		(35.160)
Invest	–	–	23.645	−10.095
			(46.852)	(61.908)
Control Variables
Fsize	0.602	0.173	0.351	−0.272
	(0.812)	(1.299)	(1.172)	(1.385)
Fage	0.009	−0.019	0.006	0.013
	(0.020)	(0.067)	(0.042)	(0.049)
ROA	0.012	0.234	−0.002	0.037
	(0.170)	(0.222)	(0.163)	(0.274)
lnGrowth	1.023	1.036	1.005	0.299
	(0.968)	(1.376)	(1.292)	(1.945)
OCF	10.037	10.443	11.870	17.306
	(10.135)	(12.974)	(18.752)	(22.516)
Bsize	0.719	2.794	3.467	1.844
	(3.374)	(5.986)	(4.650)	(5.376)
CEOD	−4.334	−6.895	−6.490 *	−5.091
	(3.360)	(5.265)	(3.568)	(5.207)
Model Diagnostics
Observations	288	288	288	288
Number of firms	52	52	52	52
Instruments	22	22	22	28
Wald χ2	1570.03 ***	530.17 ***	929.22 ***	729.26 ***
AR(1) p-value	0.013	0.083	0.050	0.074
AR(2) p-value	0.144	0.420	0.216	0.249
Hansen p-value	0.811	0.237	0.409	0.080
Sargan p-value	0.852	0.064	0.429	0.210

Notes: Two-step system GMM estimation with collapsed instruments and lag limits 2–3. Robust standard errors are reported in parentheses. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

After accounting for dynamic persistence and potential endogeneity, the financial policy channels become statistically insignificant. Leverage remains insignificant in both the separate and joint estimations (β = 0.142, p > 0.10; β = −0.945, p > 0.10, respectively), indicating that the negative association observed in the baseline and lagged specifications is not sustained once endogeneity concerns are addressed. Thus, support for H1 is not maintained under the dynamic specification.

Cash holdings also become negative and statistically insignificant in both the separate and joint estimations (β = −45.235, p > 0.10; β = −30.681, p > 0.10, respectively). This finding contrasts with the positive baseline association and suggests that the liquidity effect may be sensitive to model specification and dynamic adjustments. Although prior studies argue that internal liquidity facilitates ESG engagement through financial flexibility (Cheng et al., 2014; Ferrell et al., 2016), the evidence does not support a persistent relationship once endogeneity is considered. Accordingly, H2 is not sustained under system GMM.

A similar pattern is observed for investment intensity. Although the coefficients remain directionally mixed, investment intensity becomes statistically insignificant in both the separate and joint estimations (β = 23.645, p > 0.10; β = −10.095, p > 0.10, respectively). This finding contrasts with the positive association reported in the baseline and lagged specifications and suggests that investment-related behavior does not systematically influence ESG performance after controlling for persistence and endogeneity. Contrary to the behavioral overinvestment argument (Ben-David et al., 2013; Malmendier & Tate, 2005, 2008), the results provide no evidence that investment-related managerial tendencies persistently shape ESG outcomes. Therefore, H3 is not supported under the dynamic specification.

Overall, the system GMM results indicate that the relationship between managerial overconfidence and ESG performance is not stable across estimation approaches. While the baseline and lagged estimations suggest channel-specific associations, these effects weaken once persistence and endogeneity are incorporated, indicating that managerial overconfidence may influence ESG performance in a conditional and context dependent manner rather than through persistent financial policy channels.

5.4.3. Alternative Measurements

To further assess robustness, alternative measurements are employed for the main financial policy variables to evaluate whether the baseline findings are sensitive to measurement choices. Specifically, leverage is alternatively measured as total debt scaled by total assets (LEVASSET, Equation (6)), cash holdings as cash and cash equivalents relative to non-cash assets (CashHAsset, Equation (7)), and investment intensity using asset growth (InvestAsset, Equation (8)). The corresponding estimation results are reported in Table 7 and Table 8.

$${\mathrm{L}\mathrm{E}\mathrm{V}\mathrm{A}\mathrm{S}\mathrm{S}\mathrm{E}\mathrm{T}}_{\mathrm{i}\mathrm{t}}={\displaystyle \frac{\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{ }\mathrm{D}\mathrm{e}\mathrm{b}{\mathrm{t}}_{\mathrm{i}\mathrm{t}}}{\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{ }\mathrm{A}\mathrm{s}\mathrm{s}\mathrm{e}\mathrm{t}{\mathrm{s}}_{\mathrm{i}\mathrm{t}}}}$$

(6)

$${\mathrm{C}\mathrm{a}\mathrm{s}\mathrm{h}\mathrm{H}\mathrm{A}\mathrm{s}\mathrm{s}\mathrm{e}\mathrm{t}}_{\mathrm{i}\mathrm{t}}={\displaystyle \frac{\mathrm{C}\mathrm{a}\mathrm{s}\mathrm{h}\mathrm{ }\mathrm{a}\mathrm{n}\mathrm{d}\mathrm{ }\mathrm{C}\mathrm{a}\mathrm{s}\mathrm{h}\mathrm{ }\mathrm{E}\mathrm{q}\mathrm{u}\mathrm{i}\mathrm{v}\mathrm{a}\mathrm{l}\mathrm{e}\mathrm{n}\mathrm{t}{\mathrm{s}}_{\mathrm{i}\mathrm{t}}}{\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{ }\mathrm{A}\mathrm{s}\mathrm{s}\mathrm{e}\mathrm{t}{\mathrm{s}}_{\mathrm{i}\mathrm{t}}-\mathrm{C}\mathrm{a}\mathrm{s}\mathrm{h}\mathrm{ }\mathrm{a}\mathrm{n}\mathrm{d}\mathrm{ }\mathrm{C}\mathrm{a}\mathrm{s}\mathrm{h}\mathrm{ }\mathrm{E}\mathrm{q}\mathrm{u}\mathrm{i}\mathrm{v}\mathrm{a}\mathrm{l}\mathrm{e}\mathrm{n}\mathrm{t}{\mathrm{s}}_{\mathrm{i}\mathrm{t}}}}$$

(7)

$${\mathrm{I}\mathrm{n}\mathrm{v}\mathrm{e}\mathrm{s}\mathrm{t}\mathrm{A}\mathrm{s}\mathrm{s}\mathrm{e}\mathrm{t}}_{\mathrm{i}\mathrm{t}}={\displaystyle \frac{\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{ }\mathrm{A}\mathrm{s}\mathrm{s}\mathrm{e}\mathrm{t}{\mathrm{s}}_{\mathrm{i}\mathrm{t}}-\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{ }\mathrm{A}\mathrm{s}\mathrm{s}\mathrm{e}\mathrm{t}{\mathrm{s}}_{\mathrm{i}\mathrm{t}-1}}{\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{ }\mathrm{A}\mathrm{s}\mathrm{s}\mathrm{e}\mathrm{t}{\mathrm{s}}_{\mathrm{i}\mathrm{t}-1}}}$$

(8)

Table 7. Alternative proxy baseline and lagged specifications for financial policy channels results.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	LEVASSET	CashHAsset	InvestAsset	Joint Channels	L.LEVASSET	L.CashHAsset	L.InvestAsset	Joint Lagged
LEVASSET	−2.924	–	–	−2.314	–	–	–	–
	(2.140)			(1.796)
CashHAsset	–	16.542 **	–	17.925 **	–	–	–	–
		(6.220)		(6.400)
InvestAsset	–	–	−3.681	−4.800	–	–	–	–
			(3.924)	(3.444)
L.LEVASSET	–	–	–	–	−2.738	–	–	−1.925
					(1.908)			(2.514)
L.CashHAsset	–	–	–	–	–	10.747	–	12.534
						(6.730)		(7.046)
L.InvestAsset	–	–	–	–	–	–	−4.427 *	−4.802 *
							(2.315)	(2.315)
Control Variables
Fsize	7.986 ***	7.498 ***	8.308 ***	8.499 ***	6.905 ***	6.756 ***	8.531 ***	8.996 ***
	(0.908)	(1.218)	(0.662)	(0.544)	(1.282)	(1.283)	(1.441)	(1.776)
Fage	1.178 **	0.767	1.264 **	0.622	1.731 ***	1.530 ***	1.777 ***	1.565 **
	(0.413)	(0.640)	(0.456)	(0.703)	(0.300)	(0.383)	(0.283)	(0.495)
ROA	−0.230 ***	−0.243 ***	−0.141	−0.193 **	−0.137 *	−0.209 **	−0.055	−0.113
	(0.060)	(0.072)	(0.088)	(0.079)	(0.073)	(0.079)	(0.113)	(0.105)
lnGrowth	0.223	0.367	−0.228	0.212	−0.364	0.240	−0.752	−0.422
	(1.630)	(1.568)	(1.713)	(1.620)	(1.853)	(2.148)	(1.634)	(1.769)
OCF	15.240 *	9.197	12.440 *	5.588	10.855	13.736	3.458	9.833
	(7.238)	(5.969)	(6.634)	(4.774)	(7.883)	(9.555)	(11.499)	(12.343)
Bsize	−4.661 ***	−5.132 ***	−5.053 ***	−5.802 ***	−4.611 ***	−4.904 ***	−5.234 ***	−5.492 ***
	(1.343)	(1.373)	(1.116)	(1.149)	(1.103)	(0.979)	(1.190)	(1.130)
CEOD	−14.271 ***	−14.326 ***	−13.858 ***	−15.062 ***	−13.051 *	−13.519 **	−11.804 *	−14.039 **
	(3.983)	(3.443)	(4.206)	(3.122)	(6.125)	(5.729)	(5.953)	(5.917)
Fixed Effects
Firm FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Model Diagnostics
Observations	330	330	329	329	305	307	306	304
Number of firms	52	52	52	52	52	52	52	52
R2 (within)	0.5586	0.5761	0.5593	0.5821	0.5042	0.5107	0.5123	0.5291

Notes: Driscoll–Kraay standard errors are reported in parentheses. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. All specifications include firm and year fixed effects. The joint channel estimations include leverage (LEVASSET), cash holdings (CashHAsset), and investment intensity (InvestAsset) simultaneously.

Table 8. Alternative proxy system GMM for financial policy channel results.

Variables	(1)	(2)	(3)	(4)
	LEVASSET	CashHAsset	InvestAsset	Joint Channels
L.ESG_w	0.933 ***	0.605 **	0.435 *	0.434
	(0.313)	(0.284)	(0.257)	(0.333)
LEVASSET_w	−9.546	−	−	−23.335
	(16.844)			(21.054)
CashHAsset_w	−	−25.037	−	−36.733 *
		(16.518)		(18.834)
InvestAsset_w	−	−	−8.793	−7.362
			(14.809)	(12.454)
Control Variables
Fsize_w	0.591	0.147	1.129	1.859
	(1.038)	(1.269)	(1.634)	(2.470)
FirmAge_raw_w	0.013	−0.009	−0.004	0.046
	(0.025)	(0.060)	(0.066)	(0.057)
ROA_w	−0.140	0.213	0.034	0.147
	(0.298)	(0.199)	(0.272)	(0.386)
lnGrowth_w	0.791	1.049	2.006	3.233
	(1.393)	(1.281)	(1.868)	(3.013)
OCF_w	7.206	11.015	16.189	5.512
	(16.016)	(12.763)	(18.725)	(27.422)
Bsize_w	−1.741	4.296	4.293	5.122
	(4.190)	(6.122)	(5.343)	(5.234)
CEOD	−2.052	−7.280	−8.577	−11.871
	(4.716)	(4.780)	(4.643)	(9.515)
Fixed Effects
Firm FE	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
Observations	288	288	287	287
Number of firms	52	52	52	52
Instruments	22	22	22	28
AR(1) p-value	0.023	0.087	0.114	0.163
AR(2) p-value	0.131	0.437	0.443	0.970
Hansen p-value	0.481	0.385	0.286	0.529

Notes: Two-step system GMM estimation with collapsed instruments and lag limits 2–3. Robust standard errors are reported in parentheses. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

For the financing channel, the alternative leverage measurement (LEVASSET) remains negatively associated with ESG performance but statistically insignificant across both contemporaneous and lagged estimations (β = −2.924, p > 0.10; β = −2.738, p > 0.10, respectively), consistent with the baseline results using DER. Similar results are observed under system GMM, where the coefficients remain negative but insignificant in both the separate and joint specifications. These findings suggest that the interpretation of the financing channel is not materially altered by leverage measurement choices.

For the liquidity channel, CashHAsset remains positively associated with ESG performance in the contemporaneous fixed-effects estimations (β = 16.542, p < 0.05; β = 17.925, p < 0.05), broadly consistent with the baseline findings using CashH. However, similar to the lagged baseline specification, the coefficients become statistically insignificant once temporal dynamics are introduced. Under system GMM, the coefficients turn negative, with the joint specification becoming marginally significant (β = −36.733, p < 0.10). While prior studies suggest that internal liquidity facilitates ESG engagement through financial flexibility (Cheng et al., 2014; Ferrell et al., 2016) the findings indicate that liquidity effects remain sensitive to model specification and measurement choices.

Greater variation emerges for the investment channel. Unlike the baseline investment measure, which shows a positive association with ESG performance, InvestAsset becomes negative in both contemporaneous and dynamic estimations. Although the contemporaneous specification remains insignificant (β = −3.681, p > 0.10), the lagged specification becomes negative and marginally significant (β = −4.427, p < 0.10). Under system GMM, the coefficients remain negative but insignificant. These findings suggest that asset growth may capture broader expansion behavior, potentially reflecting investment decisions that do not necessarily align with ESG engagement, consistent with behavioral overinvestment arguments (Ben-David et al., 2013; Malmendier & Tate, 2005, 2008).

Overall, the alternative measurement results indicate that the financing and liquidity channels remain broadly consistent across specifications, whereas the investment channel appears more sensitive to measurement choices. This suggests that the relationship between managerial overconfidence and ESG performance depends not only on temporal dynamics but also on how financial policy behavior is operationalized empirically.

5.4.4. Quantile Regression Analysis

Table 9 reports the quantile regression results estimated at the 25th percentile (Q25), median (Q50), and 75th percentile (Q75) of the ESG distribution. This analysis examines whether the relationship between financial policy channels associated with managerial overconfidence and ESG performance varies across firms with different levels of ESG engagement. By moving beyond mean based estimation, quantile regression captures heterogeneous effects across the ESG distribution.

Table 9. Quantile regression analysis result.

Variables	DER			CashH			Invest
	Q25	Q50	Q75	Q25	Q50	Q75	Q25	Q50	Q75
DER	0.075	−0.038	−0.039	−	−	−	−	−	−
	(0.666)	(0.842)	(0.827)
CashH	−	−	−	33.401 **	37.799 **	38.847 ***
				(0.015)	(0.012)	(0.009)
Invest							2.892	21.156	36.068
							(0.917)	(0.480)	(0.208)
Fsize	1.831	1.185	2.152	2.860 **	1.173	1.562	1.673	1.055	2.072
FirmAge_raw	−0.007	0.030	0.017	−0.011	0.036	0.019	−0.019	0.023	−0.003
ROA	−0.003	0.020	−0.187	−0.126	−0.352	−0.411	−0.063	−0.050	−0.032
lnGrowth	3.347 **	3.028 *	4.564 ***	4.041 ***	2.653	3.401 **	3.154 *	2.800	4.259 **
OCF	36.310 *	41.199 *	28.590	36.696 **	53.764 ***	30.184	39.474 *	44.911 **	19.851
Bsize	12.595 ***	12.334 ***	8.490 **	9.009 **	9.751 **	10.390 ***	12.739 ***	13.049 ***	8.222 **
CEOD	−6.184	−15.828 ***	−16.229 ***	−5.640	−14.344 ***	−17.341 ***	−5.882	−15.510 ***	−16.345 ***
Observations	330	330	330	330	330	330	322	322	322
Pseudo R2	0.231	0.218	0.213	0.243	0.234	0.229	0.231	0.209	0.201

Notes: p-values are reported in parentheses. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

Leverage (DER) remains statistically insignificant across all quantiles, indicating that the financing channel does not systematically influence ESG performance regardless of firms’ ESG levels. This finding is consistent with the fixed effects, lagged, and system GMM estimations, which similarly indicate limited support for a persistent financing effect.

In contrast, cash holdings (CashH) exhibit a positive and statistically significant association with ESG performance across all quantiles, with coefficient magnitudes increasing toward higher ESG levels (β = 33.401 at Q25; β = 37.799 at Q50; β = 38.847 at Q75). This pattern suggests that internal liquidity becomes increasingly important among firms with stronger ESG engagement, consistent with the financial flexibility argument that internally available resources facilitate long-term sustainability investments (Cheng et al., 2014; Ferrell et al., 2016).

Investment intensity (Invest) remains statistically insignificant across all quantiles; however, the coefficients become increasingly positive toward higher ESG levels. Consistent with the system GMM findings, the results suggest that investment-related financial behavior does not systematically explain ESG performance across firms with different ESG profiles.

Overall, the quantile regression results indicate heterogeneous effects across financial policy channels. While the financing and investment channels remain weak across the ESG distribution, the liquidity channel appears more robust, particularly among firms with stronger ESG engagement.

5.5. Interpreting Financial Policy Channels and ESG Performance

Taken together, the findings indicate that the relationship between managerial overconfidence and ESG performance is heterogeneous across financial policy channels and sensitive to both measurement choices and temporal specification. Across the baseline, lagged, system GMM, alternative measurement, and quantile estimations, financing, liquidity, and investment decisions do not influence ESG performance in a uniform manner.

The financing channel, represented by leverage, provides relatively weak and unstable evidence. Both DER and the alternative leverage measurement (LEVASSET) remain statistically insignificant across most specifications, including the dynamic system GMM estimation. While leverage may constrain sustainability-related investment under financially constrained conditions (Ferrell et al., 2016; Liang & Renneboog, 2017), the findings do not support a stable financing mechanism linking managerial overconfidence to ESG performance. The consistency of insignificant results across alternative leverage measurements further suggests that the findings are not materially driven by leverage operationalization.

In contrast, the liquidity channel provides relatively stronger evidence of a positive association with ESG performance. Cash holdings remain positively associated with ESG engagement in the baseline and quantile estimations, supporting the financial flexibility argument that internally available resources facilitate sustainability-related investment and stakeholder-oriented activities (Cheng et al., 2014; Ferrell et al., 2016). However, the relationship weakens once temporal dynamics and endogeneity are incorporated, as the lagged and system GMM estimations become statistically insignificant. This pattern suggests that liquidity may support ESG engagement in the short term but does not appear to generate persistently stable ESG effects over time.

The investment channel produces the most specification-sensitive results. While the baseline and lagged estimations suggest a positive association between investment intensity and ESG performance, the alternative investment measurement (InvestAsset) generates a negative lagged effect, implying that broader expansion-oriented investment behavior may reduce subsequent ESG performance. Consistent with behavioral finance arguments, managerial overconfidence may encourage aggressive investment and resource-allocation strategies that weaken long-term sustainability orientation (Kaplan et al., 2022; Malmendier & Tate, 2005, 2008). The contrasting results across investment measurements suggest that different operationalizations may capture distinct dimensions of managerial overconfidence.

Overall, the findings reinforce the importance of treating financial policy variables as heterogeneous transmission mechanisms rather than direct behavioral proxies. The results suggest that managerial overconfidence influences ESG performance indirectly through financing, liquidity, and investment decisions, with effects that vary across measurement approaches, temporal dynamics, and estimation methods. More broadly, the study contributes to the behavioral corporate finance and ESG literature by highlighting the conditional and context-dependent nature of managerial influences on sustainability outcomes in emerging market settings characterized by financial constraints and institutional frictions (Boubakri et al., 2021).

6. Conclusions

This study examines the relationship between managerial overconfidence and ESG performance through firm-level financial policy channels in an emerging-market context. Using panel data from non-financial firms listed on the Indonesia Stock Exchange during 2015–2024, the findings suggest that managerial overconfidence does not exert a direct and uniform influence on ESG performance. Instead, its effects operate indirectly through heterogeneous financial policy behavior related to financing, liquidity, and investment decisions.

The findings reveal substantial heterogeneity across financial policy channels and estimation approaches. The financing channel, represented by leverage, provides weak and unstable evidence across specifications, suggesting that debt-related financial behavior does not constitute a consistent mechanism linking managerial overconfidence to ESG performance. By contrast, the liquidity channel provides relatively stronger evidence, as firms with greater internal financial flexibility tend to exhibit higher ESG engagement; however, the relationship weakens after accounting for temporal dynamics and endogeneity. The investment channel appears the most sensitive to measurement and model specification, indicating that different operationalizations may capture distinct dimensions of managerial behavior.

Taken together, the results suggest that managerial overconfidence influences ESG performance indirectly through heterogeneous financial policy behavior rather than through a single stable transmission mechanism. More broadly, the findings highlight the importance of accounting for temporal dynamics, endogeneity, and measurement sensitivity when evaluating behavioral determinants of ESG performance, particularly in emerging market environments characterized by financial constraints and institutional frictions.

This study contributes to the behavioral corporate finance and ESG literature by adopting a multidimensional channel-based perspective that conceptualizes financial policy decisions as heterogeneous mechanisms through which managerial behavior may shape sustainability outcomes. Rather than treating managerial overconfidence as a uniform behavioral trait, the findings demonstrate that its ESG implications depend on how managerial decision-making is reflected in corporate financial behavior. From a practical perspective, the results suggest that managers and policymakers should consider how financing, liquidity, and investment decisions may differently support or constrain firms’ long-term sustainability engagement.

Several limitations should be acknowledged. First, the study relies on financial policy behavior as an indirect representation of managerial overconfidence rather than direct behavioral measures. Second, the relatively limited and unbalanced sample due to ESG data availability constraints may affect statistical power and external validity. Third, several relationships remain sensitive across estimation techniques, suggesting that the observed effects may be context dependent rather than universally stable. Future research may therefore employ direct overconfidence measures, explore nonlinear dynamics, and examine how institutional quality, governance structures, and industry heterogeneity shape the relationship between managerial behavior and ESG performance.