ESG investment, as an extension of CSR, underscores a firm’s commitment to environmental, social, and governance issues beyond short-term financial goals (Lee et al, 2021). A considerable body of research grounded primarily in stakeholder theory documents a positive association between firms’ ESG engagement and firm value, suggesting that ESG activities enhance reputation, motivate employees, mitigate litigation risks, and ultimately improve financial performance (Freeman, 2010; Lev et al, 2010; David et al, 2005; Jian and Lee, 2015; Koo and Yang, 2024).

However, it is important to recognize that ESG investments vary in magnitude relative to what firm-specific economic fundamentals, such as firm size, profitability, industry characteristics, and risk, would predict. We refer to the component of ESG investment that exceeds these expected levels as abnormal ESG (Lys et al, 2015). This abnormal ESG represents the unexplained residual from regressions on observable firm traits and, importantly, embodies two competing economic interpretations that must be conceptually disentangled.

From an agency theory perspective (Khanchel and Lassoued, 2025), abnormal ESG represents managerial over-investment driven by self-interest, characterized by inefficient resource allocation and potential value destruction. Managers may leverage ESG expenditures to cultivate personal reputations or social capital at the expense of shareholders, thereby impairing firm value (Masulis and Reza, 2015; Krüger, 2015; Al-Shaer et al, 2023; Raimo et al, 2022). Empirical evidence suggests that such over-investment can signal entrenchment problems or agency costs within firms, and abnormal CSR spending is often interpreted as non-value-adding or even value diminishing (Asogwa et al, 2020).

Conversely, signaling theory (Lys et al, 2015; Khanchel and Lassoued, 2025) conceptualizes abnormal ESG as a strategic signal. Firms with positive private information about future performance may deliberately engage in ESG activities beyond their economic optimum to credibly convey authenticity and long-term orientation to external stakeholders. Such signals can increase investor confidence, reduce information asymmetry, and enhance firm valuation.

Moreover, some studies advance a threshold or neutral perspective (McWilliams and Siegel, 2000; Lambert et al, 2007), proposing that ESG activities increase firm value up to an optimal level but may harm value if investments exceed that threshold, suggesting an inverted U-shaped relationship. Managers engaging in ESG activities beyond this point may face punitive consequences through reduced compensation or reputational risk (Jian and Lee, 2015).

Given the conflicting theoretical perspectives surrounding the impact of abnormal ESG investment on firm value, this study adopts a null hypothesis of no systematic relationship between the two variables. Agency theory suggests that abnormal ESG may represent wasteful managerial over-investment that harms firm value, while signaling theory posits that such investments can credibly convey superior private information about future firm prospects, thus enhancing valuation. Due to these opposing views, it is challenging to predict in advance the net direction of the effect. Therefore, we state the null hypothesis as follows.

Hypothesis 1: There is no relationship between abnormal ESG investment and firm value.

As corporate investment in ESG has become increasingly significant as awareness of environmental issues has grown, public concern over climate change has intensified. Climate change is a major driver of ecological disruption and unprecedented economic damage (Labatt and White, 2007). In response, governments worldwide have implemented regulations and policies aimed at reducing industrial carbon emissions (Jung et al, 2018). According to Harjoto et al (2011), firms are expected to fulfill obligations that extend beyond profit and legal compliance, emphasizing the importance of ethical responsibilities aligned with societal standards. Consequently, many firms have voluntarily disclosed their carbon emissions to demonstrate accountability and address global environmental challenges.

In this context, the Carbon Disclosure Project (CDP), established in 2000, plays a pivotal role in supporting these efforts and enhancing environmental accountability. The CDP collects and analyzes comprehensive carbon emission data from major publicly listed companies, aiming to promote greater environmental responsibility. Participation in the CDP survey is voluntary, and while firms are not mandated to respond, the data provided by the CDP is widely regarded as highly reliable (Matsumura et al, 2014). Stakeholders can evaluate a firm’s environmental performance by examining its disclosed carbon emission data and comparing it to that of industry peers.

The decision to disclose carbon emissions remains voluntary. However, firms that respond to CDP inquiries often continue to disclose such information in subsequent years (Stanny, 2013). As the number of firms responding to CDP requests increases, the potential cost of providing inaccurate or fraudulent information rises. This implies that companies voluntarily disclosing information to the CDP may face greater reputational damage and loss of credibility if the data they provide is found to be unreliable or misleading. As a result, the information provided by firms reporting to the CDP is considered highly reliable, despite its voluntary nature (Khalid et al, 2024).

Several studies (Jung et al, 2018; Kleimeier and Viehs, 2016; Moussa and Elmarzouky, 2024) have examined the voluntary disclosure of carbon emissions. While firms with high levels of carbon emissions may risk being undervalued in the market, they are still motivated to voluntarily disclose their carbon emission data. This is because such disclosure helps reduce information asymmetry between managers and investors, enabling more efficient allocation of limited resources (Healy and Palepu, 2001), which in turn enhances firm value. Also, voluntary disclosure signals potential future carbon-related costs to the market. Jung et al (2018) examined the relationship between climate change-related risk and financing costs. Using CDP survey data to measure carbon-related risks, the study found that firms failing to respond experienced higher debt-financing costs. Put differently, firms that responded benefited from lower debt costs, suggesting that addressing carbon-related risks can reduce uncertainty and improve financing conditions (Nasih et al, 2024). Kleimeier and Viehs (2016) confirm that firms voluntarily disclosing carbon emission information enjoyed more favorable loan conditions compared to those that did not disclose. This implies that information on carbon emission levels serves as a measure of environmental performance in carbon-constrained environments, resulting in lower debt capital costs. Collectively, voluntary disclosure of carbon emissions plays a positive role in reducing information asymmetry and uncertainty, thereby lowering capital costs and enhancing financial stability for firms (Moussa and Elmarzouky, 2024).

Voluntary carbon disclosure, while often seen in a positive light, serves a critical moderating role in the relationship between abnormal ESG investments and firm value (Diamond and Verrecchia, 1991; Lambert et al, 2007; Armstrong et al, 2010; Beyer et al, 2010; Moussa and Elmarzouky, 2024). This moderation is particularly significant in light of the opposing views presented in Hypothesis 1. Abnormal ESG investments could signal a firm’s dedication to long-term environmental sustainability and corporate social responsibility, aligning with signaling theory’s emphasis on conveying credible information to the market and boosting firm value through enhanced reputation and trust (Lopatta et al, 2022).

On the other hand, consistent with agency theory, abnormal ESG investment may also reflect misaligned managerial incentives, often driven by personal interests that do not necessarily align with shareholder value maximization (Lopatta et al, 2022). Voluntary disclosure of carbon emissions can mitigate the negative perceptions associated with such self-serving ESG investments by enhancing transparency and signaling managerial accountability. Specifically, this disclosure serves not only as a transparency mechanism but also as a means of managerial discipline, indicating to investors that the firm is subject to external scrutiny (Xu et al, 2024). This increased transparency reduces information asymmetry, curbs managerial opportunism, and strengthens monitoring processes in line with stakeholder theory. As a result, investors may interpret voluntary disclosure as a signal that ESG investment aligns strategically with shareholder interests, alleviating agency concerns. Even if abnormal ESG investments stem from motives such as enhancing managerial reputation or personal agendas, the transparency provided by voluntary disclosure helps reassure investors and stakeholders of the firm’s genuine intentions.

Moreover, voluntary disclosure enhances stakeholder engagement by providing transparent and verifiable information that enables investors, customers, and regulators to more accurately assess a firm’s commitment to sustainable practices. By increasing openness and reducing uncertainty, such disclosure fosters collaborative relationships that are crucial to long-term corporate resilience and sustained success (Koo and Yang, 2024).

Furthermore, voluntary carbon emission disclosure creates an institutional environment that incentivizes managerial accountability. By subjecting their environmental performance to public scrutiny, firms reinforce internal controls and governance mechanisms that discourage opportunistic behavior, aligning managerial actions with stakeholder interests (Elamer and Boulhaga, 2024). Consequently, the interaction between abnormal ESG investments and voluntary disclosure functions as an integrated governance mechanism, whereby firms transparently share information about their ESG initiatives—especially when investment levels are abnormal—signaling both a commitment to responsible action and openness to external monitoring. This synergy enhances firm legitimacy and improves market perceptions, ultimately contributing positively to firm value. Given the conflicting theoretical perspectives, it is challenging to predict in advance the net direction of the effect. Therefore, we state the null hypothesis as follows.

Hypothesis 2: Voluntary disclosure of carbon emission information influences the relationship between abnormal ESG investment and firm value.

Table 1 describes the data selection process used in this study. The initial sample consists of 3834 firms that disclosed ESG information and participated in the CDP survey on the Korea Stock Exchange (KSE) and Korean Securities Dealers Automated Quotation (KOSDAQ) from 2017 to 2021. Next, companies with ESG data provided by the Korea Institute of Corporate Governance and Sustainability (KCGS) were selected. Financial data for this study were obtained from the Fn-Guide database. The information on voluntary disclosure of carbon emissions was manually extracted from Carbon Disclosure Project (CDP) reports. The CDP reports are indispensable tools for measuring carbon emissions and monitoring firms’ climate change-related responses. All responses to the CDP reports are survey-based and are used to encourage firms to address climate change. Measuring voluntary participation in the CDP serves as a useful tool for assessing firms’ environmental impacts (Kumar and Dua, 2022). To enhance comparability with other industries under the same conditions, the financial industry was excluded, given its distinct financial reporting characteristics. At the same time, only firms with a December fiscal year-end were selected to ensure homogeneity in reporting periods and comparability across target companies. This selection process minimized potential distortions in financial information, thereby improving the reliability and accuracy of the data under examination. Consequently, the final sample consists of 3047 company-years. The top and bottom 1% of control variables were winsorised to mitigate outlier effects.

In this study, we adopted the methodology proposed by Lys et al (2015) to distinguish ESG investment into optimal and abnormal levels. This approach estimates the expected level of ESG through regression analysis, using economic factors likely to influence a company’s ESG engagement as explanatory variables. The expected level of ESG, derived from these economic factors, is referred to as the “normal ESG level”. Essentially, ESG engagement can be understood as a process of allocating limited resources. The normal ESG level represents optimization, being the portion of the actual ESG level that can be explained or predicted by economic determinants, reflecting efficient resource allocation.

On the other hand, the abnormal ESG level, which is not explained by such economic factors, is measured by the residual ($\epsilon$) from the regression model in Eqn. 1. In this framework, the abnormal ESG level is operationalized as the regression residual, capturing the deviation of the firm’s actual ESG engagement from the predicted, normal level. This represents an unexpected or unexplained component of ESG engagement and reflects potential inefficiency in the firm’s resource allocation.

Based on this framework, the regression model employed to separate optimal from abnormal ESG engagement is specified in Eqn. 1. Key variables such as Sales, Cash, and Lev were selected based on Lys et al (2015), which established their significance in explaining firms’ ESG activities. Utilizing this approach enables a more refined analysis of ESG engagement by isolating its efficient, economically justified component from inefficiencies, thereby facilitating deeper insights into how ESG practices affect corporate performance and value.

(1)$$Es{g}_t=Sale{s}_t+Ib{e}_t+Cas{h}_t+Cf{o}_t+Le{v}_t+Mt{b}_t$$

Variable definition: Esg = log(ESG scores from KCGS); Sales = Net sales/Total assets; Ibe = Income before extraordinary items/Net sales; Cash = Cash/Total assets; Cfo = Cash flow from operations/Total assets; Lev = (Long-term debt + Debt in current liabilities)/Total assets; Mtb = (Market value of equity + Long-term debt + Debt in current liabilities)/Total assets; Size = log(Total assets); R&d = Research and development expense/Total assets; Adv = Advertising expense/Net sales; Ind = Industry dummy; Yr = Year dummy; $\epsilon$= residual term (abnormal ESG level).

To test the first hypothesis regarding the relationship between abnormal ESG investment and firm value, we specify the regression model outlined in Eqn. 2. In this specification, firm value is proxied by Tobin’s Q, and abnormal ESG investment, calculated as the residuals from the regression in Eqn. 2, is employed as the main independent variable. Control variables reflecting firm characteristics and financial performance are also included to isolate the effect of abnormal ESG investment.

Eqn. 2 allows us to empirically investigate whether deviations from optimal ESG investment are associated with increase or decrease in firm value. This analysis is crucial, as theoretical perspectives are ambiguous. Agency theory predicts that abnormal ESG investment may signal managerial overinvestment detrimental to shareholder wealth, while signaling theory suggests it could convey positive private information about future prospects. Accordingly, our first hypothesis posits the presence of a systematic relationship between abnormal ESG investment and firm value.

(2)$${\text{TobinQ}}_t={\alpha }_1+{\beta }_1{\text{AbEsg1}}_t+{\beta }_2{\text{Size}}_t+{\beta }_3{\text{Lev}}_t$$ $$+{\beta }_4{\text{Roa}}_t+{\beta }_5{\text{Ocf}}_t+{\beta }_6{\text{Invrec}}_t+{\beta }_7{\text{Da}}_t$$ $$+{\beta }_8{\text{Loss}}_t+{\beta }_9{\text{Growth}}_t+{\displaystyle \sum \text{Ind}}+{\displaystyle \sum \mathrm{Yr}}+\epsilon$$

Variable definition: TobinQ = (Market value of common stocks and preferred stocks + Total debt)/Total assets; AbEsg1 = abnormal ESG from Eqn. 1; Size = log(total assets); Lev = total liabilities/total assets; Roa = net income/total assets; Ocf = cash flow from operation/total asset; Invrec = ratio of accounts receivables; Da = Discretionary accruals measured by the model in Kothari et al (2005); Loss = 1 if the firm-year reports net loss, and 0 otherwise; Growth = (total assets in the current year – total assets in the previous year)/total assets in the current year; Ind = industry dummy variables; Yr = year dummy variables.

To elaborate Eqn. 2, the dependent variable, TobinQ, is used as a proxy for firm value, calculated as the sum of the market value of common stock and preferred stock plus total debt, divided by total assets. The key independent variable, AbEsg1, represents abnormal ESG investment, defined as the residual from the ESG prediction model described in Eqn. 1.

Control variables are selected based on prior research. Among the control variables, firm size (Size) is anticipated to positively influence firm value due to the potential benefits of economies of scale and scope associated with larger firms. However, an increase in firm size may also lead to higher political costs, which could negatively impact firm value (Lee and Jeon, 2019). The leverage ratio can affect firm value by mitigating moral hazard through stronger managerial discipline (Kim, 2025) while simultaneously increasing financial risk. The leverage ratio (Lev) is calculated as the ratio of total liabilities to total assets. Profitability (Roa) is expected to have a significant positive (+) relationship with firm value, as higher profits lead to higher firm value (Kim, 2025). According to Jensen and Meckling (1976), an increase in a firm’s cash flow reduces the risk of default while simultaneously exacerbating managerial moral hazard. Based on this perspective, operating cash flow (Ocf) is employed as a proxy to represent the agency problem. Hayn (1995) reported that firms reporting losses exhibit significantly lower relevance between accounting earnings and firm value compared to other firms. Taking these factors into account, the variable Loss was included in the model to assess firm value. Higher sales growth rates are associated with an increase in firm value, indicating that firms with higher sales growth rates may possess inherent characteristics that positively impact firm value. Consequently, it is anticipated that sales growth rates (Growth) will demonstrate a positive (+) correlation with firm value (Jeon et al, 2012; Kim and Yoo, 2023). Juan García-Teruel and Martínez-Solano (2007) argued that reducing working capital through proactive inventory management can lead to cost reductions and ultimately enhance firm value, which is why Ocf was included in the model. Firms with higher investment opportunities or growth potential are more likely to have managers with opportunistic incentives for earnings management, which can increase market volatility (Dhaliwal et al, 2017). Therefore, discretionary accruals (Da) are included as a control variable. The calculation of Da was derived from discretionary accruals as suggested by Kothari et al (2005), as shown in Eqn. 3. It was estimated for every industry year in the target data.

(3)$${\displaystyle \frac{{\mathrm{Ta}}_{\mathrm{t}}}{{\mathrm{A}}_{\mathrm{t}}}}={\alpha }_0+{\beta }_1{\displaystyle \frac{1}{{\mathrm{A}}_{\mathrm{t}}}}+{\beta }_2\left({\displaystyle \frac{\mathrm{\Delta }{\text{Sales}}_{\mathrm{t}}-\mathrm{\Delta }{\mathrm{Ar}}_{\mathrm{t}}}{\mathrm{\Delta }{\mathrm{Ar}}_{\mathrm{t}}}}\right)$$ $$+{\beta }_3{\displaystyle \frac{{\mathrm{Ppe}}_{\mathrm{t}}}{{\mathrm{A}}_{\mathrm{t}}}}+{\beta }_2{\mathrm{Roa}}_{\mathrm{t}}+\epsilon$$

Where, Ta = Net income – cash flow from operations; A = Total assets; Sales = Sales revenue; Ar = Accounts receivable; Ppe = Plant, property, and equipment; Roa = Return on assets, Net income/total assets.

Additionally, Yr and Ind denote year and industry dummy variables, respectively, to control for time- and industry-specific effects.

The second hypothesis examines the moderating effect of voluntary carbon emission disclosure on the relationship between abnormal ESG investment and firm value. To test this hypothesis, an interaction term is introduced in Eqn. 4, which captures how voluntary carbon emission disclosure influences the impact of abnormal ESG investment on firm value. This allows us to assess whether the effect of abnormal ESG investment on firm value differs depending on a company’s choice to disclose carbon emission information voluntarily. By including this interaction term, we explore how increased transparency regarding carbon emissions may condition the value implications of ESG investment decisions.

(4)$${\text{TobinQ}}_t={\alpha }_1+{\beta }_1{\text{AbEsg1}}_t+{\beta }_2{\text{Dis}}_t$$ $$+{\beta }_3\text{AbEsg}1{\text{ X Dis}}_t+{\beta }_4{\text{Size}}_t+{\beta }_5{\text{Lev}}_t$$ $$+{\beta }_6{\text{Roa}}_t+{\beta }_7{\text{Ocf}}_t+{\beta }_8{\text{Invrec}}_t+{\beta }_9{\text{Da}}_t$$ $$+{\beta }_{10}{\text{Loss}}_t+{\beta }_{11}{\text{Growth}}_t+{\displaystyle \sum \text{Yr}}$$ $$+{\displaystyle \sum \text{Ind}}+\epsilon$$

Variable definition: Dis = Voluntary disclosure of carbon emission information; $AbEsg1XDis$ = interaction term between abnormal ESG investments and voluntary disclosure of carbon emission information; see other variable definitions in Eqns. 1,2.

To address potential endogeneity between abnormal ESG investment and carbon emission disclosure, this study employs a propensity score matching (PSM) approach. Following Lawrence et al (2011), a logistic regression based solely on the control variables was conducted to calculate the probability of voluntary carbon emission disclosure. Treated firms (Dis = 1) and control firms (Dis = 0) were subsequently matched one-to-one using the nearest neighbor method with a 3% caliper to ensure high-quality matches. After excluding observations with missing values, the final matched sample comprised 1546 firm-year observations, forming the basis for the subsequent regression analyses.

Table 6 presents the regression results for the PSM-matched sample. In Eqn. 1, which includes only the abnormal ESG investment variable (AbEsg1), the coefficient was negative and statistically significant at the 1% level, indicating that firms engaging in abnormal ESG investment exhibited lower firm value even after controlling for selection bias. In Eqn. 2, both voluntary carbon emission disclosure (Dis) and the interaction term (AbEsg1 $\times$ Dis) were included. The coefficient for AbEsg1 remained negative and significant, while Dis was likewise negative and significant. Importantly, the interaction term was positive and significant at the 5% level, suggesting that voluntary carbon emission disclosure attenuates the adverse impact of abnormal ESG investment on firm value. In other words, greater transparency in carbon emission reporting mitigates the potential negative valuation effects associated with abnormal ESG engagement.

Overall, these findings are consistent with the primary regression results and confirm that the moderating effect of carbon emission disclosure on the relationship between abnormal ESG investment and firm value remains robust, after controlling for endogeneity concerns through PSM. This underscores the validity of the main findings and suggests that the results are not merely attributable to unobserved firm characteristics.

Table 7 reports the results of the two-stage instrumental variable regression applied to examine the relationship between abnormal ESG investment and firm value. To address potential endogeneity between ESG and firm value, we use the industry-year average ESG composite score (AbEsg_IV1) as the instrumental variable, following the approaches of Attig et al (2013) and Yu and Xiao (2022).

Panel A reports the results of testing the first hypothesis. In the first stage, the instrument (AbEsg_IV1) exhibits a strong, positive, and statistically significant association with firm-level abnormal ESG (AbEsg1), confirming its validity and relevance as an instrument. The second stage estimates the effect of the predicted abnormal ESG investment (AbEsg2) on firm value, proxied by Tobin’s Q. The coefficient on AbEsg2 is negative and significant at the 1% level, indicating that elevated abnormal ESG investment reduces firm value. These findings support the agency theory perspective, suggesting that abnormal ESG investment may represent managerial overinvestment or inefficiency that reduces firm valuation.

Panel B extends the analysis by investigating the moderating effect of voluntary carbon emission disclosure on the relationship between abnormal ESG investment and firm value. In the first stage, the instrument (AbEsg_IV2) continues to exhibit strong predictive power for firm-level abnormal ESG investment, as indicated by the highly significant coefficient of 1.261. Notably, the positive and statistically significant interaction between the instrument and disclosure status (AbEsg_IV2 $\times$ Dis) suggests that firms’ voluntary disclosure is systematically associated with abnormal ESG investment.

In the second stage, the coefficient on predicted abnormal ESG investment (AbEsg3) remains significantly negative, reaffirming that abnormal ESG investment negatively affects firm value in the absence of disclosure. Conversely, the interaction term between abnormal ESG investment and voluntary disclosure (AbEsg3 $\times$ Dis) is positive and statistically significant, indicating that voluntary disclosure attenuates the adverse impact of abnormal ESG investment on firm value.

These findings underscore the critical role of transparent carbon emissions disclosure as a mitigating mechanism that reduces agency-related inefficiencies and information asymmetries associated with abnormal ESG investment. Consequently, voluntary disclosure serves as an important governance tool to alleviate the negative valuation consequences of abnormal ESG investment.

Table 8 presents robustness tests employing an alternative proxy for firm value to confirm the stability of the primary results. Following the methodology of Yu and Xiao (2022), the market-to-book ratio is selected as an alternative dependent variable in the analysis. Abnormal ESG investment shows a consistently negative and statistically significant relationship with these alternative measures of firm value, with coefficients of –0.249 and –0.535 for each model. The adverse effect of abnormal ESG investment on firm value remains consistent regardless of the value metric used. Additionally, the interaction term between abnormal ESG investment and voluntary disclosure of carbon emissions (AbEsg1 $\times$ Dis) is positive and statistically significant in the alternative specification. This finding reinforces the overall evidence by showing that voluntary disclosure helps alleviate the negative valuation impact of abnormal ESG investment. These findings align with empirical research emphasizing the essential influence of high-quality disclosure and transparent communication with stakeholders in enhancing the effectiveness of ESG investments, particularly in contexts where firms undertake substantial ESG investments (Moussa and Elmarzouky, 2024; Koo and Yang, 2024).