CSR incentives and corporate irresponsibility: evidence from the United States

Corporate social irresponsibility (CSiR) refers to firm behaviors – illegal or legal but unethical – that violate widely accepted norms and harm stakeholders or the environment (Hadani, 2024). Examples include environmental pollution, labor rights violations, and product safety failures (Clark et al., 2022). Unlike earnings management, which manipulates financial reporting within legal boundaries to influence perceptions of performance, CSiR encompasses a broader set of socially harmful actions that can be operational, strategic or governance-related. In addition, CSiR often impacts multiple stakeholder groups simultaneously, such as employees, customers, communities and regulators, making it a multidimensional and far-reaching construct (Lin-Hi and Müller, 2013).

The consequences are substantial: in the USA, CSiR destroyed an estimated $534bn in firm value between 2014 and 2019 (Luo, 2021). Scandals such as Volkswagen’s Dieselgate, Wells Fargo’s mortgage abuse and Boeing’s 737 MAX incidents demonstrate how managerial decision-making failures can have severe societal and financial repercussions. Preventing CSiR is thus a central governance challenge.

One increasingly common governance tool is linking executive pay to environmental and social objectives (hereafter CSR incentives). Adoption has grown rapidly, with about 30% of publicly traded firms using them by 2021 (Cohen et al., 2023). Proponents argue these incentives promote goals such as employee satisfaction (Ikram et al., 2023) and diversity (Bugeja et al., 2016). Critics caution they can enable rent extraction (Bebchuk and Tallarita, 2022) or symbolic compliance (Edmans, 2023). For example, McDonald’s tied bonuses to gender diversity targets yet faced allegations of racial discrimination from Black franchisees (Patton, 2021).

While it may seem intuitive that mechanisms promoting CSR should also reduce CSiR, this relationship is neither automatic nor well established. CSR and CSiR are not simply opposite ends of a single continuum; firms can simultaneously engage in both “doing good” and “doing harm” (Lin-Hi and Müller, 2013). CSR incentives might encourage targeted improvements in measured areas, while leaving unrelated harmful behaviors untouched, or even indirectly encouraging them if resources are diverted. Conversely, they could foster a broader culture of stakeholder orientation that reduces harm across domains. Understanding which dynamic dominates is vital for clarifying the link between positive and negative corporate behaviors and for guiding the design of effective incentive systems.

Existing research leaves two gaps. First, while prior studies link CSR incentives to sustainability and financial performance (e.g. Cohen et al., 2023; Flammer et al., 2019; Ikram et al., 2023), their role in preventing socially harmful corporate behavior remains underexplored. This neglects the possibility that CSR incentives influence not only “doing good” but also “avoiding harm,” potentially via different mechanisms. Second, contingency factors shaping the effectiveness of CSR incentives are poorly understood (Hou et al., 2024). Although research on executive pay (Edmans et al., 2017) and for CSR incentives in particular (Ikram et al., 2023) acknowledges context dependency, little is known about how environmental conditions affect their effectiveness in curbing CSiR.

Drawing on stakeholder–agency theory (Hill and Jones, 1992; Jain and Zaman, 2020), we propose that CSR incentives align managerial decision-making with stakeholder expectations, thereby reducing CSiR. Two reinforcing channels are identified: encouraging deeper engagement with environmental and social issues, which influences behavior beyond specific incentive targets, and prompting investments in management control systems, which improve monitoring and governance (Derchi et al., 2021, 2023). A contingency framework based on resource dependence theory (Pfeffer and Salancik, 1978) examines how three environmental dimensions (i.e. complexity, munificence, and dynamism) moderate the CSR incentives–CSiR link. We expect them to be more effective in dynamic environments, less effective in munificent ones, and to have ambiguous effects in complex environments.

The USA provides an appropriate research context because executive pay disclosure is extensive and standardized under Securities and Exchange Commission rules, enabling accurate identification of CSR incentive adoption. US firms also face high media scrutiny and active stakeholder engagement, both of which increase the salience of CSiR and make reputational consequences more immediate. This makes the USA a fertile context for examining CSR incentives and CSiR.

We test our arguments on 8,203 firm-year observations from 1,763 US publicly traded firms between 2011 and 2019. Using fixed-effects regressions, we find CSR incentives are associated with significantly lower CSiR, with the effect stronger in dynamic and weaker in munificent environments; complexity has no significant moderating role. The results are robust to two-stage least squares (2SLS) regression and an alternative CSiR measure.

The theoretical contribution is to extend CSR incentive research beyond fostering positive social outcomes to preventing harmful corporate behavior and clarify the CSR–CSiR nexus by showing the relationship is contingent on environmental conditions (e.g. Jain and Zaman, 2020; Kuzey et al., 2024; Zhong et al., 2025). The practical contribution is to provide boards and compensation committees with evidence on when CSR incentives are most likely to reduce misconduct, thereby informing incentive-system design. The policy contribution is to offer empirical support for regulatory debates on CSR-linked incentives and their potential to prevent corporate misconduct, suggesting that disclosure and incentive guidelines should account for contextual contingencies.

Firms have diverse stakeholders, such as communities, customers, employees, and investors, whose interests may conflict (Hill and Jones, 1992). Management is responsible for balancing these interests (Jain and Zaman, 2020), yet may underinvest in areas outside shareholder priorities, such as employee satisfaction (Edmans, 2011), particularly when benefits are uncertain and long-term (Iliev and Roth, 2023).

CSiR refers to noncompliance with laws and ethical standards, breaching accepted societal norms, and potentially harming others and the environment (Clark et al., 2022). Such behavior often stems from managers knowingly engaging in harmful actions for personal gain while externalizing risks to stakeholders (Armstrong, 1977). Consequences can be severe, including brand damage, loss of consumer trust (Rindova et al., 2006) and reduced firm value (Clark et al., 2022). CSiR thus represents an agency cost arising from deviations from stakeholder interests.

Prior research has examined governance mechanisms to curb CSiR, such as board independence, size, gender diversity, shareholder litigation rights and the presence of nomination or CSR committees (e.g. Godfrey et al., 2024; Jain and Zaman, 2020; Treepongkaruna et al., 2024a). However, empirical results are mixed (Kuzey et al., 2024). Boards, while providing oversight, often lack control over daily decisions, and firms may engage in symbolic compliance, adopting formal structures without substantive change (Edmans, 2023). These limitations suggest that aligning managerial and stakeholder interests through executive compensation could be a more effective means of preventing CSiR.

Prior research shows that elements of executive pay influence CSiR. Bonus pay (McGuire et al., 2003), equity incentives (Chircop et al., 2024) and tournament-based pay (Bouslah et al., 2018; Zhong et al., 2025) tend to increase CSiR by rewarding risk-taking and short-term orientation. Long-term pay can mitigate CSiR by aligning horizons with the delayed manifestation of harm (Flammer and Bansal, 2017), but may lose effectiveness when vesting periods exceed managerial tenure (Berrone and Gomez-Mejia, 2009).

We argue that CSR incentives help prevent CSiR. First, by aligning managers with stakeholder interests, they hold executives accountable for non-shareholder concerns and integrate these into decision-making (Hartikainen et al., 2021). Incentives for “doing good” can spill over to avoiding harm, as shown in other contexts – e.g. employers’ environmental CSR increasing employees’ charitable activities (Koch-Bayram and Biemann, 2024). This heightened responsibility can reduce irresponsible actions.

Second, CSR incentives are often adopted as markers of long-term value creation, directing managerial focus toward environmental and social performance (e.g. Eccles et al., 2014; Flammer et al., 2019). To support such strategies, firms invest in management control systems to gather relevant stakeholder and environmental data (Berrone and Gomez-Mejia, 2009). Enhanced information enables better alignment with stakeholder expectations and broader monitoring of corporate actions (Derchi et al., 2023). Accordingly, we expect firms with CSR incentives in executive pay to act more responsibly and avoid CSiR:

Research shows that the impact of executive pay structures varies across firms and industries (Edmans et al., 2017), implying that CSR incentives’ effectiveness in curbing CSiR depends on context. Drawing on RDT, we argue that environmental conditions moderate this relationship by shaping stakeholder pressures and managerial incentives (Flammer et al., 2019).

RDT highlights that firms depend on external stakeholders for resources such as capital, reputation and legitimacy (Coombs and Gilley, 2005). Stakeholder influence is uneven: customers and employees, bound by formal contracts, exert direct financial impact, while local communities or environmental groups influence indirectly, often without immediate economic consequences (Flammer et al., 2019). In contexts with weaker direct pressures, governance mechanisms like CSR incentives can play a stronger role in aligning managerial actions with stakeholder expectations.

Following Dess and Beard’s (1984) framework, we examine three conditions that are likely to shape the CSR incentives–CSiR link. Environmental complexity reflects heterogeneity in the stakeholder environment (Cannon and John, 2007). High complexity involves multiple stakeholder groups with diverse, sometimes conflicting, expectations (Freeman, 1984; Jain and Zaman, 2020), increasing the need for firms to process large volumes of stakeholder-related information. CSR incentives encourage formal monitoring and control systems for nonfinancial performance (Derchi et al., 2023), enabling managers to integrate diverse social responsibility demands into strategic decisions (Derchi et al., 2021). By explicitly linking executive pay to CSR outcomes, firms in complex industries are better equipped to mitigate the risk of CSiR and to promote more balanced, responsible decision-making:

Environmental munificence denotes the abundance of resources that facilitate firm growth and survival (Richter and Kapteina, 2025). While munificence can ease the financing of CSR initiatives, it can also weaken CSR incentives’ disciplining effect. Resource-rich firms may underinvest in management control systems (Janka and Guenther, 2018) and meet CSR targets by reallocating slack rather than embedding responsibility into core practices, fostering symbolic compliance (Edmans, 2023). In such contexts, CSR incentives may facilitate a “check the box” approach that masks underlying irresponsible practices rather than addressing the root causes of CSiR:

Environmental dynamism captures the rate and unpredictability of environmental change, including technological shifts, market volatility and regulatory fluctuations (Dess and Beard, 1984). Such conditions heighten uncertainty, encouraging opportunism and short-term profit seeking that can increase CSiR (Cicchiello et al., 2023; Richter et al., 2022). CSR incentives can counteract these pressures by embedding stakeholder-oriented objectives into executive pay, anchoring decision-making in long-term ethical standards even amid turbulence (Li and Yang, 2025). These incentives serve as a governance tool, encouraging decision-makers to internalize stakeholder interests and resist pressures to prioritize short-term profits over societal well-being:

Our sample comprises US firms covered by LSEG/Refinitiv. The 2006 SEC compensation disclosure rule requires US firms to report detailed executive pay information (Wang et al., 2020), enabling robust analysis of CSR incentives. Following Cohen et al. (2023), we begin in 2011, when CSR incentive data is consistently available, and end in 2019, to avoid pandemic-related disruptions in economic activity and CSiR media coverage (Beckmann and Rogmann, 2024). We first compiled a list of US firms in LSEG/Refinitiv containing CSR controversy scores (our dependent variable), corporate governance and CSR ratings. Financial and fundamental data from Bureau van Dijk’s Orbis and Compustat were then matched to this list. Our sample comprises 8,203 firm-year observations from 1,763 firms between 2011 and 2019.

We determine our empirical model using specification tests. First, a Hausman test favors the fixed-effects estimator (Hausman, 1978). Second, a pF test against a pooled OLS model is significant (F = 2,390.57, p < 0.001), confirming that fixed effects are preferable due to unobserved firm heterogeneity. We estimate our model using equation (1):

We test whether environmental conditions moderate the CSR incentives–CSR controversies score relationship by adding interaction terms, using equation (2):

where $α$⁠, $β1$–$β7$⁠, and $γ$ denote the regression coefficients, $i$ the respective firm, $j$ the industry, $t$ the respective year, $CSR incentivesit-1$ our independent variable, $Xit-1$ a vector of control variables, and $FE$ the firm, time, and industry-fixed effects. Independent and control variables are lagged one year to mitigate reverse causality (Wooldridge, 2015). Standard errors are clustered at the firm level to address heteroscedasticity and autocorrelation. Table 1 summarizes all variables.

Researchers measure CSiR via media reports on socially irresponsible issues such as bribery, pollution, human rights abuse, and tax evasion (Jain and Zaman, 2020). Following prior work, we use the ESG controversies score from LSEG/Refinitiv to capture media coverage of CSiR (e.g. Agnese et al., 2023; Beckmann and Rogmann, 2024; Jain and Zaman, 2020; Kuzey et al., 2024). Alternative measures include the RepRisk score (e.g. Kölbel et al., 2017), event-based scores from Covalence EthicalQuote (Capelle-Blancard and Petit, 2019) and SEC penalty records (Chen et al., 2020).

The ESG controversies score reflects the annual number of news articles on a firm’s alleged CSiR, sourced from major English-language media (Refinitiv, 2022). It goes beyond traditional ratings by covering negative stories from global outlets and ranks firms percentually across 23 topics, such as environmental incidents and labor conditions, benchmarked to industry peers (Dorfleitner et al., 2020).

Scandals can affect reputation beyond the fiscal year, as legislative disputes, lawsuits and fines persist, influencing future ratings and long-term perception. The index adjusts for potential size bias, as media attention may vary with firm size (Refinitiv, 2022). The original score is negatively coded; we reverse-code it for interpretability, following prior studies (e.g. Kuzey et al., 2024), so our CSR controversies score ranges from zero (not at all irresponsible) to 100 (highly irresponsible).

We measure CSR incentives with a dummy variable equal to one if a CSR-related target is linked to the management compensation scheme and zero otherwise. Following prior studies (e.g. Cohen et al., 2023; Derchi et al., 2021; Jain and Zaman, 2020), we use the Compensation Policy/Sustainability Compensation Incentives variable from LSEG/Refinitiv. Because this variable is based on mandatory disclosure items for US-listed firms, coding is objective and consistent across the sample.

Complexity. We measure industry complexity using the Herfindahl–Hirschman Index (HHI), defined as $∑iIsij2$⁠, where $sij2$ is the sales share of firm i in industry j. Lower HHI values indicate greater heterogeneity in firm size within an industry, reflecting a more complex competitive and stakeholder environment (Cannon and John, 2007; Hou and Robinson, 2006). To reduce short-term fluctuations and data errors, we average the HHI over three years for each firm in the four-digit SIC industry code.

Munificence. Following Price and Sun (2017), we capture munificence as the average annual sales growth rate at the two-digit SIC industry level over a rolling five-year window. This reflects sustained resource availability rather than short-term volatility.

Dynamism. We adopt an objective, variation-based approach to environmental dynamism (Li and Liu, 2014), avoiding potential biases in perception-based measures. Consistent with Aksin and Masini (2008), we use annual Compustat data at the two-digit SIC level for total sales, employment and value added (k = 1, 2, 3). For each industry j, we regress each dependent variable y_jk on year [see equation (3)]:

Dynamism is calculated over a rolling five-year window, meaning that the value for year t is based on data from years t − 4 to t (e.g. the 2015 value is based on 2011–2015 data). Then, we compute dynamism as the mean of the standard errors from these regressions, averaged across the three variables, and divide by their respective means:

where $εjk$ is the standard error from equation (3) for variable k in industry j, $mjk$ the mean of variable k over the same five-year period, and K = 3 (total sales, employment and value added).

We include board- and firm-level controls identified in prior research as relevant predictors of CSiR. At the board level, we control for CEO duality, coded as one if the CEO also serves as board chair and zero otherwise, to capture CEO power, which is negatively associated with CSiR (Ghafoor and Gull, 2024). Board size is measured as the natural logarithm of the number of directors in year t, as larger boards may improve oversight and reduce CSiR (Treepongkaruna et al., 2024a). The sustainability committee is coded as one if such a committee exists and zero otherwise, as it facilitates CSR strategies and lowers exposure to CSiR (Jain and Zaman, 2020).

At the firm level, we control for firm age (log of current year minus incorporation year) and firm size (log of total employees), as older and larger firms attract more public scrutiny, increasing the likelihood of CSiR (Agnese et al., 2023). Tobin’s Q (log of the ratio of market value of assets to book value) captures market valuation, which may influence investor and media attention (Kölbel et al., 2017). ROA (log of EBIT over total assets) reflects profitability, which can enable CSR engagement and potentially reduce CSiR (Jain and Zaman, 2020). Leverage (log of total debt over total assets) measures financial risk, which is linked to higher CSiR likelihood (Treepongkaruna et al., 2024b).

We also include the CSR rating score (0–100) from LSEG/Refinitiv, constructed from 23 environmental, social, and governance indicators, as higher CSR performance is generally associated with lower CSiR (Kotzian, 2024). Tangibility is the log of property, plant, and equipment over total assets, capturing cross-industry differences affecting CSR-related metrics (Treepongkaruna et al., 2024b). Sales growth is measured as the firm’s annual sales growth rate, as higher growth can attract stakeholder attention and media scrutiny (Kölbel et al., 2017).

We winsorize all variables at the 1st and 99th percentiles to exclude outliers. Table 2 presents the descriptive statistics. The average CSR controversies score is 10.53, indicating that most firms have relatively low exposure to negative media coverage regarding CSiR. CSR incentives are adopted by 28.6% of sample firms, corroborating with prior evidence (Cohen et al., 2023; Flammer et al., 2019).

Table 3 presents the variance inflation factors (VIF) and Pearson correlation coefficients. CSR incentives and the CSR controversies score are negatively and significantly correlated (r = −0.149, p < 0.01), indicating a moderately strong inverse relationship between CSR incentives and CSiR. All correlation coefficients are below 0.7, and VIF values range from 1.13 (sales growth) to 3.77 (log employees), suggesting that multicollinearity is not a concern.

Table 4 reports the results from the fixed-effects regressions. Across all specifications, the coefficient on CSR incentives is negative and statistically significant, consistent with H1. In Model 2, the coefficient is −0.002 (p < 0.01), indicating that linking executive pay to CSR-related targets is associated with fewer incidents of socially irresponsible behavior. This supports stakeholder–agency theory (Hill and Jones, 1992): when managerial incentives are explicitly tied to stakeholder-oriented objectives, agency problems between managers and non-shareholder constituencies are mitigated, reducing decisions with social costs. While prior research shows that CSR-linked pay can promote socially beneficial activities (e.g. Cohen et al., 2023; Flammer et al., 2019; Ikram et al., 2023), our evidence extends this literature by showing that such incentives can also avoid harmful outcomes. This clarifies the CSR–CSiR nexus (Strike et al., 2006) by demonstrating that the same governance tool can encourage positive behaviors and deter negative ones.

These results also inform debates about the substantive versus symbolic value of CSR incentives. Critics argue that such contracts may serve as cosmetic devices, fostering symbolic compliance without changing core practices (Edmans, 2023) or facilitating rent extraction (Bebchuk and Tallarita, 2022). Our evidence points the other way: when CSR metrics are embedded in pay structures, they shift decision-making at the point of action in ways monitoring alone has not achieved (Jain and Zaman, 2020). This is consistent with prior evidence showing that incentive alignment can be a proactive governance device, shaping managerial behavior before misconduct occurs rather than relying solely on ex post oversight.

Several control variables yield insights. CEO duality is positively associated with CSiR (β = 0.063, p < 0.05), consistent with the argument that concentrated CEO power can weaken board oversight (Kuzey et al., 2024). Firm size (log employees) is positively associated with CSiR (β = 0.260, p < 0.01), reflecting visibility and complexity effects (Agnese et al., 2023). CSR ratings are positively linked to CSiR (β = 0.084, p < 0.05), echoing findings that public CSR commitments may invite greater scrutiny and thus more controversy coverage (Kotzian, 2024). ROA is negatively associated with CSiR (β = −0.046, p < 0.05), suggesting profitability provides slack resources to maintain compliance and stakeholder engagement (Chen et al., 2020).

The moderation analysis shows when CSR incentives are most effective. For H2, the interaction between CSR incentives and complexity is statistically insignificant (β = 0.034, p > 0.1). This suggests that in industries with heterogeneous demands, firms may already have institutionalized broad stakeholder-management routines and internal control systems, reducing the marginal benefit of extra alignment (Freeman, 1984; Jain and Zaman, 2020). This result refines the boundary conditions of stakeholder–agency theory: where stakeholder engagement processes are deeply embedded, the incremental disciplinary effect of CSR-linked pay appears muted.

In line with H3, the coefficient on CSR incentives × Munificence is negative and significant (β = −0.0163, p < 0.05). This finding is consistent with resource-dependence theory (Pfeffer and Salancik, 1978), which suggests that in resource-rich environments, managers can meet CSR-related performance targets by redeploying slack resources rather than altering underlying routines or decision logics. In such cases, the disciplinary role of incentives is weakened, and the potential for symbolic compliance increases (Edmans, 2023). This aligns with findings that abundant slack can reduce the need for strategic change, thereby dampening the behavioral impact of performance-contingent pay (Ittner et al., 2003).

Supporting H4, the interaction between CSR incentives and dynamism is positive and significant (β = 0.065, p < 0.1). High environmental dynamism heightens uncertainty and shortens managerial decision horizons, increasing the temptation to prioritize immediate performance over long-term stakeholder interests (Holmstrom, 1982). This suggests that CSR-linked pay can act as a counterweight, anchoring managerial decision-making in broader stakeholder objectives even when external conditions are volatile. This aligns with research that implies that non-financial CSR incentives can mitigate myopic decision-making in uncertain environments (Flammer and Bansal, 2017). CSR-linked pay thus functions as a stabilizing governance mechanism, reinforcing stakeholder alignment.

From a theoretical perspective, these findings extend CSR incentive research in two ways. First, they move the conversation beyond the effectiveness of such incentives to foster positive outcomes, showing that they can also prevent socially harmful conduct. This dual role reconciles CSR and CSiR research by demonstrating that a single governance tool can influence both ends of the corporate social performance spectrum. Second, by identifying environmental munificence and dynamism as contingencies, the results refine stakeholder–agency theory’s boundary conditions. They specify when incentive alignment is likely to be most – and least – effective as a disciplinary device in context.

The practical implications are equally important. For boards and compensation committees, the evidence indicates that CSR-linked pay is most effective in dynamic and resource-constrained environments, where it can meaningfully shape managerial behavior and reduce the incidence of socially irresponsible actions such as environmental violations, labor disputes or product safety failures. In contrast, in highly munificent industries, its marginal effect is lower, underscoring the need for complementary governance practices to verify and reinforce substantive engagement. Designing incentives without regard to environmental context risks paying for symbolic compliance.

From a policy standpoint, the findings contribute to regulatory debates on the role of CSR-linked pay in preventing corporate misconduct. Empirical evidence that such incentives can reduce CSiR supports the case for encouraging their adoption through disclosure requirements or incentive design guidelines. However, our results caution against a one-size-fits-all approach: environmental conditions materially shape effectiveness. Regulators seeking to promote CSR-linked pay as a governance tool should therefore account for industry-specific context to ensure incentives drive substantive, not symbolic, improvements.

To mitigate concerns about endogeneity from unobserved firm characteristics or reverse causality, we estimate a 2SLS regression (de Villiers et al., 2011), using firm-level instruments (e.g. capital intensity and R&D intensity) that are theoretically related to the adoption of CSR incentives but plausibly unrelated to CSiR except through this channel (Larcker and Rusticus, 2010). In the first stage, we regress CSR incentives on these instruments; in the second stage, we replace CSR incentives with their fitted values (⁠$CSR incentivesit-1^$⁠) in equation (5):

The instruments are relevant (Wald F = 58.901; under-identification test χ² = 97.26, p < 0.05) and valid (Sargan test χ² = 6.092, p > 0.1). The second-stage results (Table 5) confirm a negative and significant effect of CSR incentives on CSiR, consistent with the main findings. Under this specification, the CSR incentives × Dynamism interaction becomes significant, lending partial support to H2 and suggesting that unobserved heterogeneity may have attenuated this relationship in the fixed-effects model.

We further test robustness by replacing the CSR controversies score with a reputation-based measure: receipt of a sustainability award in year t (e.g. Yoo and Pae, 2016), using LSEG/Refinitiv data supplemented with hand collection. Sustainability awards, based on independent third-party assessments, capture outstanding environmental or social performance and thus proxy for low levels of CSiR. We use the model in equation (6):

The award measure is strongly correlated with CSR controversies (r = −0.418, p < 0.05). Estimating equation (6) with firm fixed effects (Table 5, Model 6), we find CSR incentives are positively associated with award receipt (β = 0.033, p < 0.1), indicating that such incentives not only reduce irresponsibility but also enhance reputation. As before, dynamism strengthens the relationship, munificence weakens it, and complexity has no moderating effect. Together, these tests confirm the robustness of our main results.

CSR incentives have emerged as a prominent governance tool for aligning executive behavior with stakeholder interests. Yet, their role in preventing CSiR remains underexplored. This study addresses that gap, offering evidence on whether and under what conditions CSR-linked pay can serve as an effective disciplinary mechanism.

Using panel data on 8,203 firm-year observations for US firms (2011–2019), we find that CSR incentives are associated with lower levels of CSiR. This effect is stronger in dynamic environments, where volatility heightens agency conflicts, and weaker in munificent environments, where slack resources facilitate symbolic compliance. Environmental complexity does not significantly moderate the relationship, suggesting diverse stakeholder demands may dilute the impact of CSR-linked pay.

Our findings extend the CSiR literature by showing that CSR incentives can curb harmful corporate behavior, complementing prior work focusing on monitoring-based governance mechanisms (e.g. Godfrey et al., 2024; Kuzey et al., 2024). We also contribute to governance research by identifying environmental contingencies shaping the effectiveness of CSR incentives, and to executive compensation research by shifting the focus from generating positive social outcomes to preventing irresponsible conduct (e.g. Cohen et al., 2023; Flammer et al., 2019; Hou et al., 2024).

For boards and policymakers, the results highlight CSR incentives as a potentially powerful complement to board oversight in promoting ethical decision-making. However, in resource-rich industries, they may foster symbolic rather than substantive change, underscoring the need for complementary monitoring or third-party verification. Regulators should also consider contextual factors when issuing guidance on CSR-linked pay design and disclosure.

The study has some limitations. While we address endogeneity concerns, future research could exploit natural experiments (e.g. regulatory shocks) to strengthen causal claims. Our US-focused sample calls for replication in other institutional contexts, and our binary measure of CSR incentives leaves open questions about the effects of design features such as time horizon and target type. Although our sample ends in 2019 and CSR practices have evolved since then, it captures the formative years of CSR pay adoption in the U.S., when design choices, implementation challenges, and early effectiveness patterns emerged. These early adoption years remain relevant for theory development because they reveal the foundational mechanisms underlying CSR incentive effects – mechanisms that still shape their role today despite practice shifts. Moreover, using this historical window avoids conflating effects with the extraordinary disruptions caused by the COVID-19 pandemic, ensuring cleaner identification of baseline relationships.

This study moves the debate on CSR incentives beyond their role in fostering “doing good” toward their capacity to prevent “doing harm.” As firms face growing scrutiny over their social and environmental conduct, well-designed CSR-linked pay has the potential, if tailored to context, to become a meaningful lever for sustainable and responsible corporate behavior.

Philipp Clemens Richter is also affiliated with the ESCP Europe Business School, Berlin, Germany.

The author sincerely thanks Prof Khaled Hussainey, Guest Editor of this Special Issue, for his valuable guidance and support throughout the editorial process. The author also appreciates the constructive feedback from the two anonymous reviewers. Special thanks go to Johannes Jahn and Peter Schäfer for their helpful comments on earlier versions of this study. The author is also grateful to the participants of his session at the European Accounting Association Conference in Rome, 2025, whose engaging discussions and suggestions enriched the paper. Any remaining errors are, of course, the author’s own.