Cultural governance and earnings quality: the role of corporate values and managerial ability

Corporate culture is increasingly under the scrutiny of regulators and stakeholders as a critical factor influencing financial reporting and organizational behavior (Hasan et al., 2024b). Unethical corporate disclosure practices — such as overstated earnings, strategic misreporting, or manipulated financial results — have heightened calls to examine how internal cultural values guide managerial decision-making and constrain opportunistic behavior (Public Company Accounting Oversight Board, 2024). Although often intangible and difficult to quantify, corporate culture functions as an informal control system, shaping the ethical norms and incentives that govern managerial actions (Guiso et al., 2015; Arian et al., 2025a).

Despite growing interest in the role of culture in shaping firm outcomes — such as stakeholder violations (Zaman, 2024), environmental performance (Hasan et al., 2024a), credit ratings (Bao et al., 2024), and comparability of financial statements (Afzali, 2023) — its potential to reduce earnings management remains underexplored. While existing research suggests that a strong culture increases transparency and curbs excessive risk-taking, few studies empirically test whether it can effectively constrain discretionary reporting behavior. Prior work often highlights culture’s symbolic importance but lacks rigorous large-sample evidence of its ability to reduce earnings manipulation — a critical concern for investors and regulators alike. This study addresses this gap by examining the relationship between corporate culture and earnings management using a novel machine-learning proxy derived from earnings call transcripts (Li et al., 2021). We hypothesize that firms with stronger internal cultures are less likely to engage in opportunistic accrual-based reporting, resulting in higher earnings quality and improved disclosure. We also explore whether this relationship is mediated by managerial ability — a channel through which cultural values may be embedded into ethical reporting behavior. Lastly, we assess industry-specific variation, showing that culture's effect is most pronounced in sectors where reputation, innovation, and trust are central to long-term value creation. These gaps underscore the need for more robust culture consideration in corporate disclosure quality research. This paper aims to fill this gap by examining how corporate culture can help reduce the prevalence of earnings management. If a strong corporate culture fosters ethical leadership and curbs opportunistic behavior, it is plausible that such culture would discourage earnings management, thereby enhancing the transparency and reliability of financial reporting. To test this proposition, we hypothesize that firms with stronger cultural values are less likely to engage in earnings manipulation, leading to improved disclosure quality and greater financial integrity.

Building on recent literature, we adopt established measures to examine the relationship between corporate culture and earnings management. Following Li et al. (2021), we measure corporate culture using a neural network model that analyzes the unscripted language of earnings call transcripts to identify the presence of core organizational values. To assess earnings management, we employ two widely used proxies of earnings quality: DAQ, the absolute value of residuals from industry-year regressions of accruals on cash flows (McNichols, 2002), and RDCA, a performance-adjusted measure that accounts for firm-level differences in operating outcomes (Kothari et al., 2005). We find that firms with stronger corporate cultures are significantly less likely to manipulate their earnings. These findings suggest that culture serves as a valuable internal control, fostering transparency and ethical behavior in financial reporting. We also show that this effect partially operates through managerial ability—firms with strong cultures are more likely to have capable managers who translate shared values into responsible reporting practices. Finally, we find that the cultural effect varies across industries, with the strongest impact observed in sectors where trust, reputation, and innovation are central to long-term success. These results highlight the broader role of culture—not merely as a symbolic concept, but as a genuine driver of ethics and financial integrity.

The current study makes several important contributions to the literature on corporate culture and financial ethics. First, it extends existing research by establishing a robust connection between corporate culture and earnings management—an area where culture's influence has been conceptually acknowledged but empirically underexplored. While prior studies suggest that culture influences ethical outcomes (Hasan et al., 2024b), this paper offers new evidence showing that stronger cultural values are associated with a lower likelihood of discretionary earnings manipulation, positioning culture as an effective informal control mechanism. Second, the study finds that managerial ability serves as a key mediating channel, demonstrating that firms with strong cultures are more likely to empower capable leaders who uphold ethical norms (Cho et al., 2011), thereby extending the relevance of the managerial efficiency framework developed by Demerjian et al. (2013). Third, the study reveals significant industry-level heterogeneity in culture's impact, indicating that its effects are most pronounced in sectors where reputation, stakeholder trust, and innovation are central to firm value (Guggenmos and Van der Stede, 2020)—a context that has rarely been examined empirically.

These insights are highly relevant to both corporations and policymakers. For managers and boards, the findings underscore the importance of cultivating strong internal values as an informal governance mechanism that complements formal controls (Graham et al., 2022). Investing in cultural development—particularly through leadership quality—can help mitigate unethical financial behavior such as earnings manipulation and foster long-term organizational resilience. For policymakers and regulators, the evidence presented here supports the potential value of encouraging firms to disclose cultural practices related to financial reporting integrity, which could be incorporated into existing ESG or governance reporting frameworks (Chen et al., 2022). By highlighting how culture interacts with managerial capability and industry context, this paper offers a broader perspective on how internal norms influence corporate disclosure practices and lays a foundation for future research and policy development.

The remainder of this paper is organized as follows. Section 2 synthesizes the relevant literature and develops the hypothesis. Section 3 outlines the research design, including data, variable construction, and empirical models. Section 4 presents the main results, additional analyses on potential mechanisms and industry heterogeneity, and a series of robustness checks. Finally, Section 5 discusses the findings and their implications, while Section 6 concludes the study, outlines its limitations, and suggests directions for future research.

Many studies use agency theory to explain conflicts where managers pursue short-term gains at the expense of long-term shareholder or stakeholder interests (Jones, 1995). Traditional solutions emphasize formal governance mechanisms such as management structure or executive compensation (Coles et al., 2001; Dey, 2008). More recent work highlights corporate culture as an informal but equally powerful control, shaping behavior and mitigating agency problems (Van den Steen, 2010). Values such as integrity, respect, and teamwork guide decisions even when rules or monitoring are absent, reducing opportunism and fostering long-term orientation (Li et al., 2021). Accordingly, we view culture as a critical informal governance mechanism.

Earnings management refers to managerial actions—such as altering accounting methods or operations—that influence reported earnings. While sometimes used to smooth performance, it often undermines reporting quality, reliability, and transparency. Performance-driven or “tournament-style” cultures can intensify pressure to meet targets, increasing the likelihood of aggressive accounting (Tayan, 2019; Hadid and Al-Sayed, 2021). In contrast, firms with strong ethical cultures prioritize long-term value and transparency (Yu et al., 2018).

Research shows that firms with weak ethical environments are more prone to earnings manipulation (Liu, 2016). In contrast, cultures emphasizing integrity, teamwork, and respect deter opportunism and reduce reliance on monitoring (Van den Steen, 2010). Culture thus acts as an indirect governance mechanism, filling gaps left by formal controls. Li et al. (2021), using earnings call transcripts, developed a machine learning–based measure of culture and found it correlated with better outcomes, including higher efficiency and lower information asymmetry. Other studies reinforce this role: firms with stronger cultures rely less on bank debt, indicating reduced need for external monitoring (Hasan, 2022), and culture also promotes accounting comparability through consistent, non-opportunistic decisions (Afzali, 2023). Research further links corporate culture to ethical behavior and decision-making. A trust-based culture enhances earnings quality (Garrett et al., 2014), while firms whose employees view top managers as ethical and trustworthy perform better and show less misconduct (Guiso et al., 2015). Survey evidence also indicates that most executives believe stronger cultures increase firm value and reduce unethical practices (Graham et al., 2022).

Further evidence highlights culture’s role in shaping earnings management. Zaman (2024) shows that strong culture reduces stakeholder violations, especially when external monitoring is weak, with information asymmetry as the key channel. Hasan et al. (2024b) find that culturally strong firms report lower carbon emissions, particularly under weak governance, reinforcing culture's disciplinary role. Cumming et al. (2024) demonstrate that such firms secure better IPO pricing, reflecting investor confidence. Bhandari et al. (2022), using the Competing Values Framework and 10-K textual analysis, report that collaboration-oriented cultures improve reporting quality, while competition-oriented cultures reduce it, though their proxy may overlook informal cultural elements and their design limits causal inference.

Taken together, prior studies show that strong corporate culture reduces opportunism, fosters trust, and enhances transparency as an informal control mechanism. Yet, few have directly tested its relationship with earnings management using rigorous cultural measures and accounting-based proxies. This gap motivates our first hypothesis.

Corporate culture reflects a firm’s shared values, norms, and expectations that shape decision-making, managerial incentives, and stakeholder relations. It is widely recognized as an internal governance force that fosters ethical conduct, curbs opportunism, and supports sustainable performance (Guiso et al., 2015; Gorton et al., 2022). Traditional proxies such as ESG scores or surveys are often opaque and inconsistent (Demers et al., 2021), prompting the use of machine learning and natural language processing to develop more reliable measures. Following Li et al. (2021), we employ a neural network model that analyzes unscripted segments of earnings call transcripts to identify five core values—integrity, innovation, quality, respect, and teamwork. These spontaneous exchanges provide authentic insights into organizational culture, reducing concerns about impression management. Our main proxy, zCULTURE, is a standardized composite score that benchmarks firm-level cultural strength against peers. Prior studies show this measure is significantly associated with outcomes such as stakeholder relations, emissions performance, audit fees, and earnings comparability (Chen et al., 2022).

Li et al. (2021) further validate the approach, showing that linguistic culture scores capture genuine managerial communication, extend beyond “buzzword” counts through contextual word embeddings, and are linked to operational efficiency, risk-taking, compensation design, and earnings management. Thus, culture scores derived from earnings calls provide a consistent, interpretable proxy suitable for large-sample analysis. Definitions of all variables are provided in Appendices (Appendix A).

The quality of accruals and the nature of earnings management are measured in this study using two conventional proxies for earnings management. First, we calculate Discretionary Accruals Quality (DAQ), which assesses how accruals deviate from predicted values based on firm fundamentals. Adapted from McNichols's (2002) and Francis et al. (2005) work, and following the framework of Dichev and Tang (2009), we estimate a working capital accruals model that relates these accruals to contemporaneous, lagged, leading operating cash flows, sales changes, property, plant, and equipment (PPE). Our DAQ measure is the absolute sum of the model residuals, the portion of accruals that is left unexplained. A higher DAQ value indicates lower earnings quality, as it suggests greater use of discretionary or abnormal accruals.

Following Fama and French (1997), the working capital accruals model is cross-sectionally estimated across 48 industry groups and by year.

Where.

• $ΔWCit=changeinnon−cashworkingcapital$

• with ΔCA = change in current assets,

• ΔCL = change in current liabilities,

• ΔCash = change in cash and equivalents,

• ΔSTD = change in short-term debt.

• $CFOit±1=cashflowsfromoperationsinyearst−1,t,andt+1$

• $ΔSalei,t=changeinnetsales$

• $PPEi,t=netproperty,plant,andequipment$

All variables are scaled by lagged total assets to account for firm size. Here, the absolute value of the regression residuals, ∣ $εit$ ∣, constitutes DAQ as higher (more negative) values imply more deviation from expected accrual behavior, indicating lower earnings quality.

As a separate proxy for earnings management, we compute the firm performance adjusted by RDCA to extract the discretionary component of accruals. Following Kothari et al. (2005), we include return on assets (ROA) to control for this factor to provide a better benchmark for expected accrual behavior.

We estimate the following model.

Where.

• $ΔTAit=totalaccruals,calculatedas:$

• where Dep = depreciation and amortization expense,

• $1/Assetsi,t−1$ accounts for scale effects,

• $ΔSalesi,t$ = change in revenue,

• $PPEi,t$ = capital intensity.

• $ROAi,t−1$ = lagged return on assets.

Like DAQ, all variables are scaled by lagged total assets. RDCA is measured as this model's absolute value of residuals, capturing discretionary current accruals adjusted for firm performance. Higher residuals suggest a higher degree of discretionary accrual activity, which translates into lower-quality earnings. DAQ and RDCA both alert to deviations from expected accrual behavior. In contrast, RDCA improves DAQ by adjusting performance to detect abnormal accruals better. Combining the two measures provides a strong and holistic picture of earnings management practices.

To test the association of corporate culture with firms’ ethical financial behavior, we estimate panel regression models relating our main explanatory variable (corporate culture) to our important outcome: earnings management. These models test our two baseline hypotheses while controlling for firm-level characteristics, governance factors, and unobserved heterogeneity.

We estimate the following baseline model:

Where.

• $Yi,t$​ is the dependent variable: either DAQ (Discretionary Accruals Quality), RDCA (Return-Adjusted Discretionary Current Accruals),

• $zCULTUREi,t$​ is the standardized corporate culture score for firm i in year t,

• $Controlsi,t$​ includes firm size, leverage, profitability (ROA), market-to-book ratio, capital expenditure, asset tangibility, sales growth, board size, board gender diversity, CEO duality, corporate governance score, audit committee presence, audit tenure, and intangibles intensity,

• $FE$​ represent firm, year, industry, and state fixed effects, respectively, to account for time-invariant and location-based heterogeneity,

• $εi,t$​ is the error term.

The coefficient $β1$​ captures the effect of corporate culture on financial reporting behavior. A negative coefficient in the earnings management models (DAQ, RDCA) would support our main hypothesis (H1) that a stronger culture reduces earnings manipulation.

All regressions reported in the tables are clustered at the firm level to address autocorrelation and heteroscedasticity. In the extended analysis, we examine the mechanisms through which culture shapes firm behavior, focusing on whether managerial ability mediates the culture–outcomes relationship. To assess mediation, we employ the causal steps approach of Baron and Kenny (1986), complemented by the Sobel, Aroian, and Goodman tests to evaluate the significance of indirect effects. These analyses highlight how intangible cultural values can translate into observable accounting outcomes.

To ensure robustness, we include a comprehensive set of control variables widely used in the earnings management literature (Hasan et al., 2024b). These account for firm-level financial characteristics, governance structures, and contextual factors. Controls include firm size (Ln_TA), financial performance (ROA), market-to-book ratio (MTB), leverage, tangibility (TANG), capital expenditures (Capex), and sales growth—factors that capture reporting incentives, financial pressure, and operational flexibility Daradkeh et al. (2023). Governance-related variables include board size, gender diversity, CEO duality, overall governance quality, audit committee presence, and audit tenure (Zaman et al., 2021). We also control for intangibles intensity, given its link to reputation and non-physical assets. Firm, year, industry, and state fixed effects are included to address unobserved heterogeneity. These controls collectively enhance model specification and help isolate the effect of corporate culture on earnings management.

Table 1 summarizes data construction and sample characteristics. We begin with 52,877 firm-year observations for 2001–2021, remove 2,867 cases with missing dependent variables and 15,989 with incomplete controls, yielding a balanced sample of 34,021 firm-years across 21 years. This filtering ensures consistency across model specifications.

Panel B shows the industry distribution (GICS). The largest sectors are IT (23.06%), Industrials (17.85%), Health Care (17.80%), and Consumer Discretionary (17.25%), together exceeding 75% of the sample. Energy (8.44%), Materials (6.30%), and Consumer Staples (4.71%) are also well represented, while Communication Services (2.82%), Utilities (1.41%), Real Estate (0.19%), and Financials (0.18%) are smaller. Broad sector coverage enhances generalizability and enables industry-level analysis.

Panel C reports summary statistics. The main variable, zCULTURE, has a mean of 0.0352 (SD = 1.0004), ranging from −2.1 to +2.4, consistent with Li et al. (2021). Its distribution aligns with prior studies applying this metric to corporate behavior and performance (Hasan et al., 2024a). Earnings management proxies show means of 0.0274 (DAQ) and 0.0513 (RDCA), consistent with McNichols and Stubben (2008), Kothari et al. (2005) and Srinidhi et al. (2011). Managerial ability (M_Ability) averages −0.0045 (SD = 0.1422), following Demerjian et al. (2013). Control variables are in expected ranges: average firm size (Ln_TA) 6.79, ROA 0.0693, leverage 0.259, MTB 6.65, sales growth 0.72, and Capex 0.047. Governance indicators reflect U.S. norms: CEO duality in ∼62% of firms, mean board size 12.66, board gender diversity 28.5%, audit committee presence 99%, and audit tenure 11.3 years. Intangible intensity averages 0.37, underscoring the role of intellectual capital. Overall, descriptive statistics confirm sample robustness and consistency with prior governance and earnings management research. Table 2 reports Pearson correlations among key variables. As expected, zCULTURE is negatively associated with DAQ (r = −0.010) and RDCA (r = −0.016), indicating that stronger cultures correspond to lower earnings management. It is positively correlated with managerial ability (M_Ability), consistent with capable managers fostering stronger values. Firm characteristics show mixed patterns: size, profitability, and leverage correlate negatively with culture, while market-to-book, Capex, and sales growth show positive links, suggesting forward-looking firms emphasize cultural cohesion. Governance variables are also mixed: CEO duality is negatively related to culture, whereas board size, gender diversity, and governance quality show modest positive associations. Although several coefficients are statistically significant, none approach multicollinearity thresholds (|r| < 0.80), supporting the distinctiveness of the independent variables and the stability of regression models.

The findings of the panel regression analyses examining the relationship between corporate culture and earnings management are presented in Table 3. Columns (1) and (2) test the first hypothesis using two commonly applied measures of earnings management: Discretionary Accruals Quality (DAQ) and Return-Adjusted Discretionary Current Accruals (RDCA). As shown in models (1) and (2), the results indicate a negative and statistically significant relationship between corporate culture and earnings management. Specifically, zCULTURE is negatively associated with both DAQ (β = −0.0550, p < 0.01) and RDCA (β = −0.0233, p < 0.01), suggesting that firms with stronger corporate cultures are less likely to engage in discretionary earnings manipulation. These findings support our first hypothesis and extend prior research by reinforcing the notion that corporate culture is a critical determinant of ethical managerial behavior and serves as a viable deterrent to opportunistic financial reporting (Afzali, 2023).

A number of control variables also show significant associations with earnings management. Leverage is positively and significantly correlated with both DAQ and RDCA, suggesting that highly leveraged firms may face greater pressure to manage earnings (Anagnostopoulou and Tsekrekos, 2017). Sales growth and capital expenditures (Capex) are also positively related to earnings management, potentially reflecting incentives to achieve performance targets during periods of investment or expansion. Similarly, firm size (Ln_TA) and board size (B_Size) are negatively associated with DAQ, indicating that larger firms with broader boards are better able to monitor corporate behavior and constrain earnings manipulation. Notably, CEO-chair duality (CEO_Dual) exhibits a negative relationship with both earnings management measures. This finding suggests that in some cases, powerful CEOs may promote a firm-wide cultural vision that discourages earnings manipulation, particularly within firms that maintain strong internal controls (Coles et al., 2008; Sandvik, 2020). In addition, both corporate governance quality (CG) and audit tenure are negatively and significantly associated with earnings management. These results highlight the importance of long-term oversight and robust governance systems in promoting financial reporting integrity (El Diri et al., 2020; Cornett et al., 2009). Finally, intangibles intensity (Int_Intensity) is negatively and significantly related to both DAQ and RDCA, indicating that firms with higher investments in intangible assets, such as intellectual property and brand reputation, may engage less in earnings manipulation, instead relying more on maintaining stakeholder trust. Together, these results strongly reinforce the idea that a stronger corporate culture is significantly associated with reduced earnings management. They also lend further support to the view of corporate culture as an informal governance mechanism that helps align managerial interests with the long-term values and norms of the firm.

We consequently investigate the mediating role of managerial ability (M_Ability) to explore the mechanism through which corporate culture influences earnings management. Following Demerjian et al. (2013), we construct a managerial ability score using Data Envelopment Analysis (DEA), capturing managers’ efficiency in converting firm resources into revenue while controlling for firm-level characteristics.

To formally test the mediation pathway, we employ the causal steps approach by Baron and Kenny (1986), complemented by Sobel (1982), Aroian (1947), and Goodman (1960) tests to assess the significance of the indirect effect. This approach is grounded in prior research suggesting that managerial ability is a credible transmission channel through which internal values influence financial decision-making. Table 4 presents the results of this analysis.

In this framework, zCULTURE is the independent variable, M_Ability the mediator, and DAQ the outcome, Panel A of Table 4 presents the stepwise regression results used to test the mediation pathway. In Column (1), we observe a negative and statistically significant association between zCULTURE and DAQ (β = −0.0550, p < 0.01), indicating that firms with stronger cultures are less likely to engage in earnings management. Column (2) shows that zCULTURE is positively associated with M_Ability (β = 0.0156, p < 0.01), suggesting that strong cultural values are linked to the presence of more capable managers. Column (3) reveals that M_Ability is negatively related to DAQ (β = −0.5783, p < 0.01), implying that managerial ability contributes to reduced earnings manipulation. In Column (4), when both zCULTURE and M_Ability are included in the same model, both remain statistically significant. The coefficient on zCULTURE becomes more negative (β = −0.0647, p < 0.01), while M_Ability remains a strong predictor (β = −0.6211, p < 0.01), consistent with a partial mediation effect. Panel B formally tests the indirect effect. The Sobel test (Z = 5.7201, p < 0.001), Aroian test (Z = 2.1654, p < 0.05), and Goodman test (Z = 2.1459, p < 0.05) all confirm that the mediating path from zCULTURE through M_Ability to DAQ is statistically significant. These results indicate that corporate culture reduces earnings management both directly and indirectly via managerial ability. In this sense, fostering managerial competence may serve to amplify the impact of cultural values on financial transparency (Zhong, 2018).

Overall, these findings support the notion that managerial ability links corporate culture to managerial behaviors such as earnings management. Firms with strong internal values may foster ethical conduct and transparency by enabling capable managers to act as stewards of those cultural norms (Godos-Díez et al., 2011). This aligns with prior research (Graham et al., 2022) suggesting that effective leadership is a critical mechanism through which culture is translated into measurable financial outcomes. The results underscore the importance of human capital as a driver of ethical and sustainable corporate conduct.

Table 5 presents industry-specific regressions examining how the relationship between corporate culture and earnings management varies across sectors. Using a consistent specification across industries, the results reveal considerable heterogeneity. A significantly negative association is observed in Consumer Discretionary (β = −0.0292, p < 0.05) and Information Technology (β = −0.0205, p < 0.10), suggesting that strong cultures are particularly effective in curbing earnings manipulation in these sectors. A large negative coefficient is also found in Financials (β = −0.4454, p < 0.10), though the limited sample size warrants caution. Conversely, the Materials sector shows a positive association (β = 0.0690, p < 0.05), possibly due to capital intensity or lower disclosure transparency weakening culture's disciplinary effect. These findings underscore the contextual nature of culture's influence. In sectors where intangible assets—such as trust, innovation, or brand value—are central to performance, culture plays a stronger role in deterring opportunistic behavior, aligning with prior literature on reputational risk and stakeholder sensitivity (Lewellyn, 2017).

More broadly, the findings confirm that the ethical impact of corporate culture is context dependent. Industry-specific factors—such as regulatory pressure, stakeholder expectations, and reputational risk—shape how cultural norms are internalized and enforced. Thus, the influence of culture on financial behavior is not uniform but contingent on the strategic and operational environment in which firms operate (Han et al., 2010).

Table 6 presents regression results examining how individual dimensions of corporate culture are associated with earnings management, using Discretionary Accruals Quality (DAQ) as the dependent variable. This analysis decomposes the aggregate zCULTURE measure into its five core components—Integrity, Teamwork, Innovation, Respect, and Quality—as defined by Li et al. (2021), to identify which specific values are most effective in deterring opportunistic financial reporting.

The results show that all five cultural dimensions are negatively and significantly associated with earnings management. Among them, Innovation (β = −0.0586, p < 0.01) and Quality (β = −0.0610, p < 0.01) have the strongest effects, suggesting that a forward-looking strategic orientation and an emphasis on high performance standards are particularly effective in reducing earnings manipulation (Chen et al., 2022; Dechow et al., 1996). Integrity (β = −0.0270, p < 0.01) and Respect (β = −0.0181, p < 0.01) also show meaningful negative relationships, reinforcing the role of ethical norms and inclusive values in promoting transparent reporting practices (Easley and O'Hara, 2023; Chun et al., 2013). Although Teamwork (β = −0.0132, p < 0.05) has a smaller coefficient, its significance indicates that collaborative organizational environments may still exert downward pressure on individual incentives to manipulate financial outcomes (Mitchell and Singh, 1996).

Overall, these findings emphasize that not all cultural dimensions contribute equally to shaping financial reporting behavior. While each plays a role, innovation and quality appear to be the most potent cultural levers for constraining earnings management. These insights have practical implications for firms aiming to strengthen internal controls and ethical conduct through targeted cultural initiatives.

Table 7 presents robustness checks using alternative operationalizations of corporate culture to ensure our findings are not dependent on a single specification. Consistent with prior research (e.g. Li et al. (2021), Zaman (2024)), we re-estimate the baseline models with three related proxies: (1) mCULTURE, the standardized composite score applied in our main analysis; (2) sCULTURE, the unstandardized raw sum of the five cultural dimensions; and (3) dCULTURE, a binary indicator equal to 1 if the firm’s culture score exceeds the sample median. These proxies represent different functional forms of the same linguistic construct rather than distinct dimensions but applying them helps verify the stability of our results across alternative transformations.

The results in Panel A show that all three measures are negatively and significantly associated with earnings management (measured by DAQ). Specifically, mCULTURE (β = −0.0450, p < 0.01), sCULTURE (β = −0.0090, p < 0.01), and dCULTURE (β = −0.0769, p < 0.01) consistently suggest that firms with stronger cultural values are less likely to manipulate earnings. The robustness of these findings across standardized, raw, and binary measures indicates that the relationship is not an artifact of measurement design. Overall, the alternative specifications strengthen confidence in our baseline conclusion that corporate culture functions as a meaningful informal governance mechanism. Regardless of how it is operationalized, stronger culture remains reliably associated with reduced discretionary reporting, providing further support for our main hypothesis.

To further address potential sample selection bias, we implement entropy balancing as an additional robustness check. This method, commonly used in applied research (e.g. Hainmueller (2012), Hasan et al. (2024a)), reweights observations in the control group such that the covariate distributions (means, variances, and skewness) match those of the treated group, while retaining the full sample. We define the treatment group as firms with a corporate culture score above the sample median. The balancing process covers key firm characteristics, including size, leverage, profitability, growth, investment, tangibility, governance quality, and board attributes. Panel A of Table 8 shows that balancing effectively reduces differences in means, variances, and skewness across groups. Panel B reports the reweighted regression results, where the coefficient on zCULTURE remains significantly negative with respect to DAQ (β = −0.0590, p < 0.01), confirming that stronger culture is associated with lower earnings management. The adjusted R² remains stable, indicating that model fit is unaffected by reweighting.

Overall, the entropy balancing results reinforce the robustness of our conclusions, suggesting that the observed effects of corporate culture reflect genuine differences in ethical orientation—not merely artifacts of firm fundamentals or sample selection.

To address potential sample selection bias, we implement a Propensity Score Matching (PSM) procedure, which compares firms with similar observable characteristics to isolate the effect of culture on earnings management. Following standard practice, we adopt a two-stage approach. In the first stage, we estimate the likelihood of being a high-culture firm using firm-level and governance covariates. In the second stage, we examine the culture–earnings management relation within the matched sample. Panel A of Table 9 reports the first-stage logistic regression. High-culture firms are more likely to be smaller, less leveraged, and less profitable, but they exhibit higher tangibility, faster sales growth, and greater board gender diversity. Governance factors also play a role: CEO duality, board size, audit tenure, and audit committee presence are all significant predictors. The adjusted R² of 0.1968 indicates reasonable model fit. Panel B shows the second-stage regression results. Consistent with baseline findings, the fitted culture score (fitted_zCULTURE) is negatively related to earnings management (β = −0.0014, p < 0.05). This confirms that cultural strength continues to mitigate opportunistic reporting even after accounting for observable firm characteristics.

Overall, the PSM analysis reinforces our main conclusion: corporate culture serves as a distinct and meaningful governance mechanism. Its influence on ethical reporting extends beyond structural firm attributes, reflecting intrinsic organizational values that shape managerial behavior among otherwise comparable firms.

To address potential endogeneity concerns—such as omitted variable bias, reverse causality, and simultaneity—we implement a two-stage least squares (2SLS) instrumental variable (IV) approach. Following prior studies (Attig et al., 2016; Arian, 2024, 2025b), we instrument zCULTURE with two exogenous proxies: (1) Alt_zCULTURE, standardized within the full sample, and (2) Alt_zCULTURE_In, standardized within each industry.

The IV regressions, not tabulated here for brevity, show that both instruments are negatively and significantly associated with discretionary accruals quality (DAQ), consistent with our baseline findings. Control variables behave as expected, and standard IV diagnostics confirm instrument validity: the underidentification test rejects the null, weak identification tests indicate strong instruments, and overidentification tests are not applicable given the single excluded instrument.

Overall, the IV analysis reinforces a causal interpretation: stronger corporate culture independently curbs earnings management, underscoring the role of intangible organizational values in promoting ethical financial conduct.

To further validate our findings, we conduct additional robustness checks (not tabulated for brevity), all of which support our main results. First, we stratify firms by corporate culture strength, classifying them into high (≥75th percentile) and low (≤25th percentile) groups based on zCULTURE. The negative association between culture and earnings management, are significantly stronger in the high-culture group, while effects are weaker or insignificant among low-culture firms—highlighting the role of strong culture in promoting ethical behavior. We also test alternative thresholds for defining culture strength and re-estimate models using lagged independent variables to address simultaneity concerns. Across all specifications, the direction and significance of the results remain consistent, reinforcing the robustness and generalizability of our conclusions.

Our findings have implications for corporate leaders, investors, policymakers, and future research. For managers and boards, cultivating a strong culture is strategically important, especially in industries where reputation and trust underpin long-term value. Culture complements formal governance by curbing opportunism, while the mediating role of managerial ability highlights the need for leadership development and succession planning. Selecting and retaining capable managers who act as stewards of cultural norms enhances transparency. For investors, culture represents a critical but often overlooked intangible asset. Firms with strong cultural foundations are more resilient to uncertainty, regulatory change, and reputational shocks. Incorporating cultural indicators into investment analysis may improve assessments of sustainability and governance risk. For policymakers, the results suggest that culture should be integrated into governance frameworks. This could involve embedding cultural indicators within ESG reporting, encouraging voluntary disclosure (e.g. the Lumen Principles), or using soft-law mechanisms such as codes of conduct to promote ethical behavior. Finally, for researchers, opportunities remain to extend the analysis beyond U.S. public firms to other ownership forms and institutional contexts. Future work might triangulate cultural proxies with surveys or ESG ratings, exploit policy shocks to strengthen causal inference, and explore how culture interacts with technological capabilities, ESG initiatives, and diversity to influence ethical and performance outcomes.

This study examines how corporate culture influences earnings management, drawing on a large panel of U.S. firms from 2001 to 2021. Using a linguistic measure of culture, we find consistent evidence that firms with stronger cultural values engage in less discretionary earnings manipulation. These results underscore culture's role as an informal governance mechanism that promotes transparency and constrains opportunism. We also identify managerial ability as a key channel. Using the Demerjian et al. (2013) score, we show that managerial quality partially mediates the culture–earnings management link. Capable managers act as stewards of cultural norms, reinforcing ethical and transparent practices (Zhong, 2018; Cho and Ringquist, 2011). This highlights the interplay between organizational values and human capital, suggesting that culture operates both through shared norms and through the individuals empowered to uphold them.

Our study also shows that culture’s influence is context-dependent rather than uniform. Cultural constraints on earnings management are stronger in sectors such as Consumer Discretionary and Information Technology, where reputation, innovation, and stakeholder trust are central to value creation (Guggenmos and Van der Stede, 2020). By contrast, the effect is weaker or mixed in more standardized or heavily regulated industries like Materials and Utilities. These results highlight the need to contextualize culture within firms’ strategic and operational environments. Overall, we extend research on the governance role of intangible organizational factors and underscore the importance of cultivating strong cultures to enhance financial integrity and ethical compliance.

While this study offers valuable insights, several limitations remain. First, although we mitigate endogeneity through instrumental variables and robustness checks, our design does not fully establish causality; future work could exploit natural experiments or policy shocks to strengthen causal claims. Second, the focus on U.S. public firms limits generalizability to SMEs, private firms, or those in emerging markets, where applying this framework could yield richer comparisons. Third, beyond managerial ability, other organizational factors—such as leadership style, board dynamics, or corporate identity—may also mediate the culture–ethics link. Fourth, although our culture measure is based on validated linguistic analysis of unscripted earnings calls, managerial language may still reflect strategic signaling rather than lived norms. Fifth, our robustness tests rely on functional variants (mCULTURE, sCULTURE, dCULTURE) of the same linguistic construct, which confirm stability but do not provide independent validation. Finally, future research could explore how technological capabilities and ESG initiatives interact with culture and leadership in shaping ethical and performance outcomes. Overall, our findings provide robust evidence that strong cultural values, reinforced by capable leadership, enhance financial integrity—especially in industries where trust and reputation are central to firm value.

The supplementary material for this article can be found online.