Employee green behavior (EGB), or employee pro-environmental behavior, refers to workplace actions that benefit the environment (Norton et al., 2015). As a key driver of organizational sustainability, EGB enhances environmental performance and supports sustainable development goals. Scholars often distinguish between two types of EGB: task-oriented green behavior (TGB), which involves environmentally conscious actions embedded in formal job duties such as waste segregation and use of eco-friendly materials (Sibian and Ispas, 2021), and voluntary green behavior (VGB), which encompasses discretionary initiatives such as proposing energy-saving solutions or championing sustainability innovations (Ren et al., 2023). This distinction is critical because TGB is typically compliance-driven, operating within structured control systems and institutional mandates (Norton et al., 2015), whereas VGB is rooted in autonomy and intrinsic motivation. From a managerial perspective, TGB ensures organizational adherence to environmental standards (Zhu et al., 2021), while VGB enables adaptive innovation and long-term sustainability (Zacher et al., 2023). Yet organizations often emphasize compliance through TGB while underutilizing the innovation potential embedded in VGB (Dewiana et al., 2024). Understanding the antecedents of both forms of behavior is therefore essential for balancing compliance enforcement with voluntary engagement.

To theorize the multi-level antecedents of EGB, this study draws on the Pressure–State–Response (PSR) framework, which was first proposed by Canadian researchers in the early 1990s and subsequently developed and widely applied by the Organization for Economic Co-operation and Development (OECD) and the United Nations Environment Program (UNEP) in global environmental assessments. Over the past decades, PSR has been employed in diverse domains such as ecological security evaluation (Zhao et al., 2014), emergency impact assessment (Wei et al., 2012), and sustainable development research (Salemi et al., 2019). Importantly, the PSR framework has not remained static; it has inspired several derivative models, including the Drive–State–Pressure (DSP), Drive–State–Response (DSR), and the Drive–Pressure–State–Impact–Response (DPSIR) frameworks, all of which share a common causal structure for analyzing complex interactions among multiple influencing factors (Jun et al., 2012).

Although PSR originated in environmental science to explain how ecosystems adapt to external pressures (Li, 2004), subsequent studies demonstrate that its logic is not confined to ecological systems. It has also been applied at the individual level—for instance, in explaining psychological or behavioral responses under stress (Hughey et al., 2004)—suggesting its broader generalizability. The underlying logic of “external pressure–internal state–behavioral response” thus provides a transferable framework for analyzing adaptive behavior across domains.

Building on this theoretical commonality, the present study adapts PSR to organizational behavior by proposing the Drive–State–Pressure (DSP) model. This adaptation extends PSR in two key ways. First, the “Response” component is reconceptualized as “Drive,” underscoring that employees are not merely passive reactors but proactive agents whose intrinsic motivation propels environmentally responsible behavior. Second, the DSP model explicitly links institutional pressures (Pressure) with individual capability states (State) and motivational drives (Drive), thereby bridging institutional theory’s macro perspective and organizational behavior’s micro perspective. Within this framework, Drive refers to motivational drivers such as green values (GV), which provide employees with a stable and identity-congruent orientation toward environmental responsibility. State encompasses enabling conditions, including green self-efficacy (GSE)—the confidence to perform environmentally impactful actions—and job autonomy (JA)—the discretion to implement sustainability practices in one’s work. Pressure refers to institutional mechanisms that regulate behavior: green performance management (GPM) as coercive pressure, environment-oriented CSR (ECSR) as normative pressure, and green transformational leadership (GTL) as mimetic pressure. Together, these three dimensions explain how compliance-oriented behaviors emerge under coercive systems, while discretionary innovation is more likely to be fostered when intrinsic motivation and enabling states are strong.

Traditional variable-centered approaches, such as regression or structural equation modeling, assume that each factor contributes independently and additively to outcomes (Howard and Hoffman, 2018). However, organizational behavior is often the result of conjunctural causation, where outcomes arise from specific combinations of conditions rather than isolated variables. The configurational approach addresses this complexity by emphasizing three key principles: equifinality, whereby different pathways of conditions may lead to the same outcome; causal asymmetry, whereby the conditions associated with the presence of an outcome may differ from those associated with its absence; and holistic combination, whereby conditions derive meaning from their configuration rather than from their independent effects (Fiss, 2011).

Within this paradigm, fuzzy-set Qualitative Comparative Analysis (fsQCA) has emerged as a powerful empirical method. Developed by Ragin, fsQCA operationalizes configurational theory by calibrating conditions as fuzzy sets and systematically examining how different combinations are sufficient or necessary for an outcome (Ragin, 2000; Ragin, 2008). Compared with regression-based models, fsQCA is particularly suited to the study of EGB, employees’ behaviors emerge from the interaction of institutional pressures, personal motivations, and capability states. Prior research has highlighted the role of institutional isomorphism in shaping behavior (DiMaggio and Powell, 1983), intrinsic and extrinsic motivations (Ryan and Deci, 2000), and individual efficacy beliefs (Bandura, 1997). Moreover, studies of ecological responsiveness demonstrate that green behaviors often result from the joint influence of external pressures, internal motivations, and organizational capabilities (Bansal and Roth, 2000). By capturing multiple and equally valid pathways, fsQCA allows researchers to account for both compliance-driven and voluntary forms of green behavior.

In summary, this study proposes the following hypotheses:

Taken together, these hypotheses reflect the configurational nature of employee green behaviors within the DSP framework. The overall conceptual model is presented in Figure 1.

From June 13 to June 17, 2024, online questionnaires were distributed to hotel industry employees in China via the Credamo platform. The sample focused on large-chain hotels across 23 provincial-level regions, including Beijing, Shanghai, and Nanjing. Of the 398 questionnaires distributed, 356 valid responses were retained after excluding those that failed attention checks. The final sample consisted of 228 females (64.0%) and 128 males (36.0%), with 67.7% holding a bachelor’s degree. Respondents had an average age of 32.08 years (SD = 8.21) and an average organizational tenure of 5.34 years (SD = 4.07).

Data collection was conducted using a structured questionnaire. Employee green behavior (EGB) was measured with Bissing-Olson et al. (2013) 6-item scale, which differentiates between task-oriented green behavior (TGB; e.g., “I complete assigned tasks using environmentally friendly methods”) and voluntary green behavior (VGB; e.g., “I proactively initiate workplace environmental initiatives”). Responses were recorded on a five-point Likert scale, demonstrating high reliability (α = 0.862 for TGB; α = 0.899 for VGB).

Green values were measured using eight personal norm items adapted from Steg et al. (2005), demonstrating excellent internal consistency (α = 0.916). Green self-efficacy was assessed with Chen et al.’s (2015) 6-item scale (α = 0.889). Job autonomy was measured using Spreitzer’s (1995) 3-item scale (α = 0.834). Green performance management (GPM) was evaluated through four objective-assessment items from Dumont et al. (2017), demonstrating strong reliability (α = 0.900). Environment-oriented CSR (E-CSR) was assessed using the first four items from Farooq et al.’s (2017) scale (α = 0.843). Green transformational leadership (GTL) was operationalized through the green intellectual stimulation and personalized care subscales from Robertson’s (2018) 12-item instrument (α = 0.896). All measurement items are presented in the Appendix A.

Scholarly discourse often conflates necessity and sufficiency, though they are distinct (Chung, 1969). Necessary causality means an outcome cannot occur without a condition (“no X, no Y”), while sufficient causality means a condition alone ensures the outcome (“if X, then Y”; Dul, 2016). This study employs Necessary Condition Analysis (NCA) and Qualitative Comparative Analysis (QCA) to disentangle these logics. NCA assesses whether the six DSP model factors—green values, green self-efficacy, job autonomy, green performance management, environment-oriented CSR, and green transformational leadership—are necessary for task-oriented (TGB) and voluntary green behavior (VGB), while QCA identifies sufficient causal configurations. The NCA was implemented in R (version 4.3.3) using the NCA package (version 4.0.1).

We employed fuzzy-set Qualitative Comparative Analysis (fsQCA) for two main reasons. First, fsQCA enables the exploration of equifinality, uncovering multiple and equally valid pathways that lead to employees’ green behaviors (Ragin, 2008; Fiss, 2011). Second, it accounts for causal asymmetry, recognizing that the conditions sufficient for the presence of task-oriented or voluntary green behaviors may not simply be the inverse of those leading to their absence (Ma et al., 2022). These features make fsQCA especially valuable for studying complex behavioral phenomena in organizational contexts.

However, adequacy analysis in QCA often encounters challenges regarding sufficiency, necessity, and completeness, as traditional top-down approaches face limitations such as rigid dichotomization and heavy reliance on counterfactuals. To overcome these issues, Ding (2023) proposed a bottom-up QCA method with four key innovations: (1) bypassing dichotomous truth table constraints, (2) minimizing counterfactual dependence, (3) scaling efficiently without inflating complexity, and (4) reducing omitted variable bias while ensuring validity. By generating parsimonious solutions through algorithmic iteration, this approach enhances rigor, replicability, and causal inference reliability in complex behavioral studies. In line with these methodological advancements, the configurational analyses in this study were conducted in R (version 4.3.3) using the QCA package (version 3.22), which enables robust and transparent implementation of fsQCA procedures.

In fsQCA, the raw data need to be calibrated into fuzzy-set membership scores ranging from 0 to 1, which represent the degree to which a case belongs to a given condition. A score of 1 indicates full membership (the condition is at a high level), 0 indicates full non-membership (the condition is at a low level), and 0.5 represents the point of maximum ambiguity (the crossover point), where it is difficult to determine whether the condition is present or absent. Following prior methodological recommendations (Crilly et al., 2012), values that were exactly calibrated at 0.5 were adjusted slightly to 0.5001 to prevent these cases from being excluded during the truth table analysis.

We adopted an internal calibration strategy, meaning that the thresholds for high and low levels of each condition were defined relative to the distribution of the study sample rather than absolute external standards (Pappas et al., 2017). Specifically, for each condition we computed the sample mean and standard deviation, setting the mean as the crossover point (membership = 0.5). Full membership (membership = 1.0) was anchored at one standard deviation above the mean, while full non-membership (membership = 0.0) was anchored at one standard deviation below the mean. This parametric approach is widely used in organizational behavior research, where defining “high” or “low” levels in relative rather than absolute terms is more meaningful (Aiken et al., 1991). For example, an employee who scores 5 on a 7-point scale may appear high in absolute terms, but in comparison with peers in the same organization, this level could be relatively low. Thus, internal calibration captures the contextual nature of behavioral tendencies more appropriately.

The complete calibration anchor points for all antecedent conditions and outcome variables are presented in Table 1. For ease of interpretation, following prior studies (Xing et al., 2025; Ma et al., 2022), we refer to conditions or outcomes above the crossover point as “high” (presence of condition/outcome) and those below as “low” (absence of condition/outcome) throughout the results section.

All constructs were measured using multi-item scales adapted from prior validated research (e.g., Bissing-Olson et al., 2013). Although no pilot test was conducted prior to the formal survey, the adoption of well-established scales helped ensure measurement reliability and validity. To further assess the psychometric properties of the instruments, internal consistency reliability was examined using Cronbach’s alpha coefficients in SPSS 24.0, with all constructs demonstrating excellent reliability (α > 0.80), exceeding the recommended threshold of 0.70 (Nunnally, 1978). In addition, confirmatory factor analysis (CFA) was conducted in AMOS 24.0 to evaluate construct validity. As shown in Table 2, the CFA results confirmed a good model fit (CFI = 0.95, TLI = 0.93, RMSEA = 0.06), with all factor loadings statistically significant (p < 0.001), thereby establishing measurement robustness. We further examined discriminant validity using the Fornell–Larcker criterion. As shown in Table 3, the square roots of AVEs (on the diagonal) are greater than the correlations between constructs (off-diagonal), indicating satisfactory discriminant validity.

Necessary Condition Analysis (NCA) identifies necessary conditions by assessing the effect size (d) and statistical significance of antecedent variables, with necessity levels further quantified through bottleneck analysis. As shown in Tables 4, 5, the NCA results for task-oriented green behavior (TGB) and voluntary green behavior (VGB) indicate that no individual condition meets the necessity threshold (d > 0.1, p < 0.05) as per Dul et al. (2020) criteria. The analysis utilized two estimation methods: ceiling regression (CR) for continuous variables and ceiling envelope (CE) for discrete variables. Given the absence of necessary conditions for both behavioral types, bottleneck level analysis was deemed unnecessary.

The fsQCA results were reported using the configurational notation system, as established and shown in Tables 6, 7. In these tables, filled circles (●) indicate the presence of a condition, crossed circles (⊗) signify its absence (Fiss, 2011), and blank cells represent logical remainders (i.e., conditions irrelevant to the causal recipe). This symbolic framework facilitates a clear visualization of core and peripheral conditions across sufficient configurations while ensuring methodological consistency with leading QCA studies.

Drawing on the DSP framework and configurational approach, this study investigated how institutional pressures (ECSR, GTL, GPM), capability states (GSE, JA), and motivational drivers (GV) combine to shape task-oriented and voluntary green behaviors. Consistent with our hypotheses, the results confirm that employee green behaviors emerge not from isolated predictors but from distinct configurations of antecedent conditions, thereby demonstrating the principles of equifinality and causal asymmetry.

Our findings align with prior research in showing that institutional pressures are critical in shaping compliance-oriented behaviors, while personal values and efficacy underpin discretionary engagement (Wang et al., 2018). At the same time, the results deviate from traditional perspectives that portray external pressures solely as restrictive. We find that under certain configurations, the selective absence of pressure can encourage agency-driven innovation, offering a more nuanced understanding of how organizations can foster voluntary green behaviors (Zafar et al., 2022).

By mapping multiple viable pathways to both TGB and VGB, this study extends configurational theorizing in organizational behavior. It demonstrates that sustainability engagement cannot be reduced to single “critical factors” but instead arises from diverse pressure–state–drive combinations. This insight contributes theoretically by bridging institutional and motivational perspectives, and practically by suggesting that managers should design adaptive systems: structured pressure regimes to secure task compliance and enabling environments that cultivate autonomy and intrinsic motivation for voluntary engagement.

In conclusion, this study highlights that organizational green behavior is not governed by one dominant mechanism but emerges through multiple, equally effective pathways. Recognizing this diversity helps move beyond linear models toward a more realistic understanding of how institutions and individuals interact to advance sustainability.

This study advances environmental behavior research through three key theoretical contributions, each critically engaging with existing literature while offering novel insights into the interplay between organizational systems and individual agency. First, this research bridges the macro–micro divide in institutional theory by clarifying how organizational pressures shape individual green behaviors. While institutional theory emphasizes coercive, normative, and mimetic pressures as drivers of isomorphism (Jaja et al., 2019), it rarely explores their employee-level impact. The DSP model fills this gap by mapping pressures to behavioral pathways: coercive mechanisms like green performance management (GPM) standardize task-oriented green behaviors (TGB), while normative pressures, such as environment-oriented CSR (ECSR), foster voluntary green behaviors (VGB) by embedding sustainability into corporate identity. Mimetic influences, exemplified by green transformational leadership (GTL), legitimize green experimentation through role modeling. Crucially, the study highlights negotiated institutionalization, where employees reinterpret organizational pressures through intrinsic green values (GV) and self-efficacy (GSE). TGB aligns with structured mandates, reinforcing compliance through coercive and normative pressures, whereas VGB emerges from agentic negotiation, as employees internalize normative CSR cues or mimetic leadership influences to integrate sustainability into their role identity (Steg and Vlek, 2009). This refines institutional theory by demonstrating that organizational pressures are not static constraints but dynamically shaped through employee agency.

Second, this study addresses methodological limitations in green behavior research by demonstrating the value of configurational thinking. Traditional symmetric methods, such as regression analysis, assume linear, additive relationships between variables, which can obscure the complexity of behavioral causation (Breiman, 2001). In contrast, our fuzzy-set Qualitative Comparative Analysis (fsQCA) reveals equifinality and substitution effects that challenge conventional wisdom. For instance, the identification of eight distinct pathways to high task-oriented green behavior (TGB) dismantles the myth of a universal “best practice” showing that compliance can emerge from moral alignment through green values and transformational leadership, institutional control through green performance management and environmental CSR, or capability scaffolding through green self-efficacy and performance management. Equally significant is the finding that cultural pressures, such as environmental CSR, can substitute for intrinsic green values in driving voluntary green behaviors (VGB). In contrast, task-oriented green behaviors (TGB) require a structured pressure-agency alignment, where leadership modeling plays a critical role in amplifying the effects of green values. When transformational leadership is present, green values function as a reinforcing rather than an indispensable factor, aligning with Bissing-Olson et al.’s (2013) argument that values alone are insufficient but gain influence through social validation. This explains why certain TGB pathways rely on leadership to enhance compliance-driven engagement, ensuring that green practices are embedded within employees’ professional norms. These insights validate Fiss (2011) advocacy for set-theoretic methods in behavioral research, shifting the paradigm from linear causality to context-dependent behavioral configurations.

Third, this research clarifies paradoxes in environmental psychology by distinguishing task-oriented and voluntary green behaviors. Prior studies often conflate these, leading to contradictory findings (Wiernik et al., 2016). For instance, green performance management reinforces task-oriented green behaviors (TGB) through clear expectations and accountability (Ruepert et al., 2016) but may hinder voluntary green behaviors (VGB) by dampening intrinsic motivation (Davis et al., 2020). This reflects the compliance-innovation paradox, where institutional pressures ensure consistency in structured tasks but constrain discretionary initiatives requiring autonomy. In contrast, VGB relies on facilitative mechanisms, with normative pressures like environmental CSR acting as cultural anchors rather than rigid mandates (Matten and Moon, 2008). This fosters personal agency through green values, self-efficacy, and job autonomy, aligning with self-determination theory (Deci and Ryan, 2012): coercive pressures enhance compliance but undermine autonomy essential for innovation-driven sustainability. Similarly, the values-action gap—where green values alone fail to predict environmental behaviors (Steg and Vlek, 2009)—is reframed as a configurational mismatch. Green values require activation through leadership modeling or cultural reinforcement via CSR to translate pro-environmental attitudes into action.

This behavioral typology enhances theoretical precision by outlining the specific pathways through which green behaviors emerge. It also offers practical clarity, explaining why standardized sustainability initiatives often fail to accommodate the diverse motivational structures that drive different types of green behaviors.

For organizations navigating sustainability transitions, our findings advocate a tripartite strategic realignment that integrates behavioral science principles with HR architecture redesign. First and foremost, organizations must tailor interventions to the distinct motivations behind task-oriented (TGB) and voluntary green behaviors (VGB). For TGB, a pressure-capacity synergy should integrate green performance management (GPM) with quantifiable targets and green transformational leadership (GTL) training to reinforce compliance. In contrast, VGB requires autonomy-driven motivation. Environment-oriented CSR should go beyond compliance, embedding sustainability into daily experiences (Adu-Gyamfi et al., 2021)—e.g., linking hotel waste reduction to local ecosystem restoration. Participatory initiatives like gamified eco-challenges, employee-led green innovation labs, and “green experimentation zones” can further enhance intrinsic motivation and job autonomy, fostering engagement and creativity.

Secondly, leadership development programs must align with organizational work design (Hanson, 2013) by tailoring green influence strategies to different workplace contexts. In high-autonomy departments, where employees have greater behavioral discretion, managers should act as cultural facilitators (Teräväinen and Junnonen, 2019), fostering a green innovation culture through shared vision-building and participatory decision-making. In contrast, in standardized operational units with rigid procedural constraints, managers should function as process optimizers, embedding sustainability seamlessly into workflow efficiency through structured task alignment and behavioral nudges.

Finally, a systemic HR architecture must institutionalize a dual-reinforcement approach, ensuring that institutional pressures—coercive, normative, and mimetic—are operationalized through structured HR mechanisms. First, competency models should prioritize green self-efficacy (GSE) development through VR-enabled sustainability simulations, transforming abstract environmental principles into hands-on operational skills. Second, organizations must integrate green behaviors into formal appraisal systems (coercive pressure; Saeed et al., 2019) while fostering peer-driven sustainability incentives (mimetic pressure), such as interdepartmental eco-challenges and public recognition programs. This structured yet flexible approach ensures that green behaviors are not merely encouraged but become deeply ingrained in employees’ professional identity.

While this study provides an in-depth understanding of employee green behavior, it has certain limitations. First, the study is based on a cross-sectional dataset of hotel employees in mainland China. This specific cultural and industry context may shape how employees respond to institutional pressures. For example, employees in China may react differently to coercive pressures compared to those in Western contexts. Thus, the generalizability of our findings is limited. Future studies could conduct cross-cultural and cross-industry comparisons to validate and extend our conclusions.

Second, although the Drive-State-Pressure (DSP) model effectively captures the interaction between intra-organizational institutional and individual factors, it does not directly incorporate external pressures such as customer environmental expectations and government regulations. Future research should integrate these external influences to provide a more holistic understanding of the drivers of green behavior.

Third, the study employs fuzzy-set Qualitative Comparative Analysis (fsQCA), which identifies multiple behavioral pathways but is sensitive to threshold calibration and sample composition. To mitigate these issues, future research could combine fsQCA with longitudinal case studies, enabling the dynamic evolution of green behaviors to be more fully captured.

Fourth, organizational and individual contextual factors—such as company-level green norms and employees’ career stages—were not fully addressed. For instance, early-career employees may be more influenced by mimetic pressures, whereas senior employees may rely more on intrinsic motivation and job autonomy. Exploring these dynamics could further enrich our understanding of green behavior across diverse contexts.

Fifth, the imbalance between female and male participants in the sample may affect the extent to which the results can be generalized to a broader workforce. Future studies should aim to include a more balanced gender distribution to validate and extend the findings.