When anger delivers: leader anger expression and employee performance in hospitality

Hospitality and tourism workplaces are characterized by emotionally demanding service environments in which anger forms an inevitable feature of organizational life driven by high customer expectations, service failures, complaints, recovery efforts and operational pressures. Employees, managers and customers routinely encounter anger during service interactions, and the presence of this emotion has consequences for employee well-being, service delivery and organizational performance. The primary investigative focus of hospitality scholars has been the detrimental effects of angry or dissatisfied customers on service employees and business outcomes (e.g. Akarsu et al., 2023), and on the negative consequences of employee anger (e.g. Buick and Thomas, 2001; Min and Joireman, 2021). The literature has largely conceptualized anger as a disruptive emotion that undermines employee well-being, service quality and organizational performance. Although this perspective offers important insights, it is only a partial explanation of anger in hospitality workplaces. Given the ubiquity of anger in service settings, greater attention is needed to explain circumstances where anger generates constructive rather than detrimental consequences.

Recent organizational research has posited that anger is not inherently destructive; rather, its consequences depend on how it is expressed, interpreted and embedded within the applicable social context (Geddes and Callister, 2007). When anger is expressed in a normatively appropriate manner, it may communicate dissatisfaction, urgency and a need for corrective action. This perspective suggests that expressions of anger can function as social signals that encourage individuals to recognize performance deficiencies and make appropriate behavioral adaptions.

Unlike the relatively short-term and episodic nature of customer-related anger, leader anger represents a more enduring and influential − albeit underexplored − emotional dynamic in hospitality organizations. Because leaders occupy positions of authority, their emotional expressions carry symbolic and relational significance that can shape employee interpretations and behavioral responses (Shao et al., 2018; Lindebaum et al., 2016). Whether these expressions produce constructive or destructive outcomes, however, may depend ultimately on the leadership context within which they occur. The interpretive frameworks that employees use to evaluate emotional displays by their leaders are important to advance understanding. Ethical leadership is a salient contextual cue and influence on such interpretations n hospitality settings. Such leadership is reflective of a normative approach that is grounded in fairness, role modeling and in reinforcing appropriate conduct (Brown et al., 2005). It sends a signal to employees that leader behaviors, including the expression of emotions, should be understood within a framework of moral intent and developmental purpose rather than personal hostility or arbitrary dominance. Accordingly, ethical leadership is expected to shape how employees interpret leader anger, influencing whether such expressions are seen as (a) appropriate within organizational norms or (b) as violations of interpersonal conduct.

Since hospitality work is highly personalized and interaction-based, employees are not merely following standardized procedures (SOPs). Instead, they continuously adapt their communication, task priorities, interpersonal coordination and recovery strategies in response to changing customer needs and managerial expectations. This dynamic nature of service delivery helps explain why job crafting is particularly important. Job crafting refers to employees’ self-initiated changes to their tasks, relationships and cognitive understanding of work (Wrzesniewski and Dutton, 2001). When leader anger is expressed, it may be interpreted by employees as signaling that current service delivery is insufficient and that prompt adjustments are required. Accordingly, job crafting serves as a key mechanism through which leader anger is translated into proactive service improvement and enhanced task performance.

Drawing on the emotions as social information (EASI) model (Van Kleef, 2009), the current study examines how different leader expressions of anger influence hospitality employees’ task performance through both affective and behavioral pathways. We distinguish between controlled anger, which is restrained and deliberate, and outward anger, which is expressed more openly and is emotionally intense (Geddes and Callister, 2007; Grecucci et al., 2024). We propose that these two forms of leader anger convey different social signals and trigger different employee responses. Specifically, leader anger may elicit employee anger as an affective response, while also prompting job crafting as a behavioral adjustment. We further examine ethical leadership as a critical boundary condition. Because emotional expressions are interpreted within a social-relational context, ethical leadership provides moral and normative cues that may shape whether leader anger is perceived as legitimate corrective feedback or alternatively as interpersonal mistreatment (Hassan et al., 2023; Wang et al., 2018).

This study contributes to research on hospitality emotion and leadership by advancing a context-dependent understanding of leader anger. First, it contributes to the EASI model by showing how leadership style shapes whether leader anger is interpreted as constructive feedback or as an interpersonal threat. Second, it challenges the dominant view that anger is always destructive, by distinguishing between controlled and outward leader anger and by examining their different effects on employee anger, job crafting and task performance under different levels of ethical leadership.

Developed by Van Kleef (2009), the EASI model provides the core theoretical framework for this study. The EASI model posits that emotions expressed in social contexts serve as social signals that influence observers’ cognitive, emotional and behavioral responses. According to this model, leader anger expressions convey information to employees and can impact their behaviors. The model proposes two mechanisms: inferential processes and affective reactions. Inferential processes involve observers interpreting emotional expressions to draw conclusions about the expresser’s intentions (Keltner and Haidt, 1999). In the case of leader anger, employees may perceive it as a signal of dissatisfaction and infer that change is required, motivating behavioral adjustment. This inferential processing supports proactive responses. Job crafting, involving self-initiated changes to tasks, roles or interactions, can be understood as an inferentially driven response (Van Kleef, 2009; Keltner and Haidt, 1999). Illustrative of this pathway, and through a process of sensemaking, employees conclude that their work approach should be adjusted in line with leader expectations.

The term affective reactions describes immediate emotional responses to the emotions of others and is associated with emotional contagion. Emotional contagion describes the process by which emotions are transmitted through mimicry, mirroring and physiological feedback (Hatfield et al., 1993). This process influences employee emotions and subsequent behaviors (Barsade, 2002). Leader anger may, for example, trigger anger on the part of employees. The EASI model further emphasizes that expressions of emotion depend on contextual factors that shape interpretation and responses (Van Kleef, 2009).

The EASI model conceptualizes social-relational moderators primarily as interaction-specific cues, such as perceived appropriateness and information processing (Van Kleef and Côté, 2022), which guide inferences about intentions. Researchers have shown that leader emotions influence follower attitudes and behaviors (e.g. Hong et al., 2025). Their findings have highlighted the influences of context on employee responses to leader emotions. Building on previous scholarship, the present study extends EASI by incorporating leadership style as a broader social-relational context that extends beyond individual interactions. Rather than an immediate cue, leadership style represents a stable normative environment that is shaping expectations about leaders’ emotional conduct and moral intent, thereby influencing interpretive processes that are triggered by leader anger.

Anger is an emotional response to perceived goal obstruction, intentional harm or blameworthy actions by others (Schmitt et al., 2019; Wang et al., 2018; Ortony et al., 2022). While understanding the nature of anger is important, this study focuses on how anger is expressed. Anger expression refers to how individuals manage and communicate their anger in social interactions. Prior research distinguishes between two primary forms of anger expression: outward anger and controlled anger. Outward anger involves openly expressing anger through overt verbal or behavioral displays, whereas controlled anger refers to expressing anger in a regulated and deliberate manner without allowing it to escalate (Grecucci et al., 2024).

Controlled anger is often considered to be an effective instrumental tool when manifest in the workplace. Researchers have found that leader expressions of controlled anger help to clarify standards, thereby increasing employee compliance and leadership effectiveness (Wang et al., 2018). Because controlled anger is regulated and goal-oriented, it communicates dissatisfaction without personal attack, signaling expectations while maintaining relational appropriateness. Within the EASI framework, such expressions may activate inferential processing, prompting employees to interpret anger as performance-related feedback rather than as interpersonal hostility. In such circumstances, employees are more likely to attribute controlled anger to performance related concerns, rather than personal criticism, thereby reducing the likelihood of responding with anger.

Outward expressions of anger, by contrast, operate primarily through affective mechanisms. Outward anger is often perceived as uncontrolled and destructive, leading to greater employee withdrawal, counterproductive work behaviors and lower team performance (Wang et al., 2018; Yu et al., 2020). Such unregulated anger is frequently associated with abusive supervision and unethical leadership, as it violates social and moral display rules in the workplace (Hassan et al., 2023). According to the EASI model, outward leader anger is more likely than cognitive inferential processing to evoke affective reactions. When leaders display overt anger, employees perceive these affective cues and may proceed to mirror their emotional state. Hatfield et al. (1993) demonstrated that individuals naturally mimic and internalize the emotions of others, particularly in the case of expressions by authority figures. Relative to cognitive sensemaking, outward anger is more likely to elicit employee anger through affective contagion and negative relational judgments (Van Kleef, 2009; Hatfield et al., 1993).

Job crafting may be defined as employee proactive modifications to their tasks and social interactions. Such modifications may reshape the nature and structure of employee roles (Wrzesniewski and Dutton, 2001). Job crafting has also been conceptualized as the enhancement of meaning and satisfaction through physical or cognitive changes to work tasks (Slemp and Vella-Brodrick, 2014). Rather than depending solely on individual traits, job crafting is also shaped by contextual cues that signal a need for adjustment or improvements. Prior researchers have identified several antecedents − individual, relational and contextual, including employees’ personal resources, perceived autonomy, job characteristics and the broader organizational environment (Lyu et al., 2024; Slemp and Vella-Brodrick, 2014; Wrzesniewski and Dutton, 2001).

According to the EASI model, leaders’ emotional expressions serve as social signals that inform employees about situational demands and performance expectations (Van Kleef, 2009). Employees may interpret leader expressions of anger as representations of dissatisfaction that require corrective action, prompting them to align with leader expectations by adjusting their work behaviors. When leaders express controlled anger, it is typically perceived as deliberate and regulated rather than impulsive. As a result, employees are more likely to interpret it as performance-related feedback than as an interpersonal threat, encouraging cognitive appraisal rather than emotional reactions (Hsu et al., 2024).

Consistent with the EASI framework, controlled anger shifts processing from affective contagion to cognitive inference thereby reducing employee anger (Grecucci et al., 2024; Wang et al., 2018). Lower employee anger preserves the cognitive resources and self-regulatory capacity that is needed to engage in proactive behaviors such as job crafting. Because anger can inhibit initiative and flexible adjustment (Blair, 2012), employees experiencing less anger are better positioned to make constructive modifications to their tasks, interactions and work approaches. Although controlled anger can provide a direct communication of the need for improvement, its positive influence on job crafting is expected to operate primarily by reducing employee anger and supporting adaptive regulation. Accordingly, employee anger may be viewed as mediating the relationship between leader-controlled anger and job crafting. Based on the preceding, we hypothesize that:

Outward anger is different with its intensity and emotional expressiveness. This increases its salience as a social signal and amplifies its interpersonal impact. From an EASI perspective, outward leader anger operates through dual pathways (Van Kleef, 2009). First, through an informational pathway, it signals dissatisfaction and urgency, heightening the salience of performance-related cues and encouraging vigilance toward potential shortcomings (Van Kleef, 2009). In service environments where failures are immediately visible and consequential, such signals may prompt employees to use job crafting to avoid future negative encounters by reconsidering and adjusting their work roles (Weseler and Niessen, 2016). Thus, outward anger can motivate behavioral adaptation by conveying corrective informational value.

Second, through an affective pathway, outward anger can trigger employee anger via emotional contagion and perceived interpersonal threat. Empirical evidence shows that intense leader anger displays elevate anger and anxiety among followers (Schwarzmüller et al., 2018). Anger depletes cognitive resources, narrows attentional focus and promotes defensive rather than constructive orientations (Blair, 2012), all of which are critical for the cognitive flexibility and self-regulation required for job crafting. These effects may be particularly pronounced in emotionally demanding hospitality contexts (Haver et al., 2019). Integrating these dual pathways, we argue that although outward anger carries informational value that may encourage job crafting, its affective costs are likely to dominate in shaping employee responses. Specifically, the emotional strain induced by leader anger is expected to constrain employees’ capacity to translate informational cues into proactive behavioral change. Accordingly, we hypothesize that:

Task performance refers to employees’ effectiveness in carrying out core job duties, reflecting proficiency and quality in completing essential tasks (Koopmans et al., 2014). Research has documented the detrimental effects of anger on task performance. Experimental studies show that anger impairs precision, reduces performance accuracy (Zur et al., 2019) and slows task performance (Rivera-Rodriguez et al., 2021), suggesting it disrupts cognitive control and undermines effective task execution when unmanaged. However, emerging research indicates that anger is not inherently dysfunctional but can also function constructively at the within-person level. When experienced at low intensity and processed through inferential mechanisms, anger may promotes problem-focused coping and adaptive regulation rather than impulsive reactions. Cognitively processed anger heightens goal salience, mobilizes effort and facilitates behavioral adjustment, enabling individuals to channel dissatisfaction into corrective action and improved performance (Lench et al., 2023; Van Kleef, 2009). Accordingly, employees who perceive controlled leader anger are likely to experience less personal anger, engage more in job crafting and ultimately enhance task performance.

Empirical evidence shows that employees who engage in job crafting demonstrate sustained improvements in task performance (Tims et al., 2015). Similarly, job crafting leads to significant improvements in task performance, particularly by expanding task boundaries (Weseler and Niessen, 2016). Hospitality researchers also report that hotel employees who proactively adjusted their work tasks showed higher engagement in their job duties (Petrou and Xanthopoulou, 2021). Together, these findings highlight the performance-enhancing role of job crafting across organizational contexts.

Integrating these perspectives, controlled anger functions as a regulated, goal-oriented signal that employees interpret as providing constructive feedback, rather than as a personal threat, consistent with the inferential pathway of the EASI model. This reduces employee anger by shifting responses from affective reactions toward cognitive sensemaking (Van Kleef, 2009; Wang et al., 2018). Lower employee anger preserves cognitive control, while leader anger simultaneously signals the need for improvement, thereby encouraging employees to adjust their work proactively through job crafting (Schwarzmüller et al., 2018). Together, these arguments suggest a sequential process: controlled leader anger reduces employee anger, fosters job crafting and ultimately improves task performance. Accordingly, we propose the following hypothesis:

In contrast, outward leader anger may undermine employees’ task performance by increasing employee anger and reducing job crafting. Extensive evidence shows that high-intensity anger impairs routine task performance. Hospitality research further indicates that anger disrupts work−life balance (Moreo et al., 2020) and contributes to burnout (Weber et al., 2017), both of which have negative implications for performance. According to the EASI model, outward leader anger primarily influences employees through emotional contagion, increasing employees’ own anger (Van Kleef, 2009; Hatfield et al., 1993). Such anger may impair self-regulation and diminishes proactive behaviors (Blair, 2012). By depleting cognitive resources and narrowing attention, employee anger diminishes the initiative required for job crafting, which depends on cognitive resources, intrinsic motivation and proactive effort. Reduced job crafting limits employees’ ability to align tasks with their strengths, thereby undermining task performance (Weseler and Niessen, 2016). Employee anger also directly impairs task performance by disrupting attention and cognitive control (Rivera-Rodriguez et al., 2021). Accordingly, outward leader anger is expected to reduce task performance by increasing employee anger and subsequently decreasing job crafting, leading to the following hypothesis:

Ethical leadership reflects a sustained pattern of moral conduct, fairness and norm enforcement (Brown et al., 2005). This shapes follower expectations over time about appropriate leader behaviors. This normative role of ethical leadership aligns with recent work on organizational moral commitment, showing that shared moral norms and values shape expectations about leaders’ emotional conduct and followers’ interpretations of emotional expressions (Bañón-Gomis et al., 2025). As such, ethical leadership establishes a normative interpretive context for the evaluation of emotional expressions.

From an EASI perspective, such contexts influence whether observers engage primarily in inferential processing (interpreting anger as purposeful and performance-related) or in affective reactions, such as emotional contagion. The current study extends prior EASI research (e.g. Van Kleef and Côté, 2022) that examined leadership and interpersonal moderators of emotional expressions. Ethical leadership is conceptualized here as a broader normative interpretive context that shapes emotional meaning and appropriateness judgments across repeated interactions. This extension is particularly relevant in hospitality, where ongoing leader–follower relationships and strong emotional display norms heighten the role of moral expectations in shaping responses to leader expressions of anger.

It is typical for ethical leaders who emphasize integrity and moral conduct, to view anger as a negative emotion and that its expression merits regulation (Brown et al., 2005). Over time, such consistent moral conduct shapes follower expectations regarding how ethical leaders should behave emotionally, thereby creating a moral prototype of the leader as composed, fair and norm-adherent. When leaders express anger in a controlled manner, and this display aligns with their moral image, followers are more likely to interpret it as appropriate and performance-related and not as personally threatening. Thus, ethical leadership should attenuate emotional contagion and reduce employees’ anger responses to controlled leader anger. According to the EASI model, controlled expressions under strong ethical leadership are more likely to trigger inferential processing rather than affective contagion, enabling employees to interpret the performance-related message without becoming emotionally overwhelmed. Although subtle cues such as tone shifts or facial tension may still signal restrained displeasure (Geddes and Callister, 2007), the moral credibility of ethical leaders provides interpretive reassurance that limits followers’ anger reactions. Based on this reasoning, we propose the following hypothesis:

When ethical leaders express anger outwardly, the impacts may be particularly disruptive due to the incongruity between this emotional display and their typical conduct (Brown et al., 2005). Employees generally do not expect ethical leaders, who are perceived as being composed and morally grounded, to express anger outwardly. When such expressions occur, they may violate employee expectations, leading to increased emotional dissonance and elevated employee anger. This reaction can be understood through a moral discord mechanism. Because ethical leadership establishes a stable moral prototype of fairness, self-regulation and norm adherence, outward anger represents a deviation from this prototype. Such violation of expectations heightens follower sensitivity to emotional display and intensifies affective reactions, as the anger is interpreted as inconsistent with the leader’s moral identity rather than as a performance-related signal.

In the context of the EASI model, such expressions are likely to activate emotional contagion, especially when they contradict contextual norms, thereby intensifying negative emotional responses. The unexpected nature of anger from an ethical leader may therefore amplify its negative consequences. This perceived breach of the ethical leader’s image can also hinder constructive behaviors such as job crafting. Moreover, the inconsistency may disrupt inferential processing, reducing the likelihood that employees view the anger as purposeful or performance related. Accordingly, we propose the following hypothesis:

The proposed conceptual model, including the hypothesized relationships among leader anger expressions, employee anger, job crafting, task performance and the moderating role of ethical leadership, is presented in Figure 1.

The target population for this study included individuals who are (a) currently employed in tourism and hospitality, (b) residing and working in the USA, (c) are at least 18 years old, (d) demonstrate advanced English proficiency and (e) have an immediate workplace supervisor. All the responses were provided by employees, and there was no participation by supervisors. Employees reported on their own perceptions of anger expressions by their supervisor/s, consistent with prior research showing that subordinates can provide reliable observations and reporting of leaders’ emotional displays (Knight et al., 1988; Forbes et al., 2004). Data was collected online in Spring 2025 via Prolific. Ethical approval for this study was obtained prior to data collection. All participants were informed of their rights before starting the survey, including the option to withdraw at any time and assurance of anonymity. A two-wave time-lagged survey design was deployed to reduce common method bias. This was assured through the temporal separation of key constructs while maintaining feasibility and response quality (Podsakoff et al., 2003; Koopmans et al., 2014). Leader expressions of anger and ethical leadership were measured at Time 1, since they reflect relatively stable perceptions of supervisory behaviors. Constructs such as employee anger and job crafting were measured at Time 2.

In the first survey, 409 responses were collected. Among these, 62 were excluded due to survey timeout (n = 6), exiting before completion (n = 30), failure to pass attention checks (n = 8) or completing the survey too quickly (n = 15). As a result, 347 participants (85.8% completion rate) successfully completed the first survey. These 347 were invited to participate in the second survey one week later. Twenty-two participants dropped out and 29 responses were excluded due to survey timeout (n = 3), exiting before completion (n = 10), failure to pass attention checks (n = 1) and completing too quickly (n = 16). As a result, 296 participants (85.3% completion rate) completed the second survey, forming the final sample.

All measurements were adapted from previous studies. Leader expressions of anger were measured using the Spielberger anger expression scale (Knight et al., 1988), encompassing controlled anger (four items) and outward anger (eight items). Employee anger was evaluated using the state-trait anger expression inventory, consisting of seven items (Forbes et al., 2004). Job crafting was assessed using the job crafting scale (15 items) developed by Slemp and Vella-Brodrick (2014), including three dimensions: task crafting (TC), cognitive crafting (CC) and relational crafting (RC), each with five items. Task performance was measured using the scale (five items) developed by Koopmans et al. (2014). Additionally, ethical leadership style was assessed using the ethical leadership scale (ten items) (Brown et al., 2005). All variables were measured using a seven-point Likert scale. Employee anger and leader’s anger expression (controlled and outward) were rated from 1 (Not at all) to 7 (A lot). Task performance and ethical leadership were rated from 1 (Strongly disagree) to 7 (Strongly agree), while job crafting was rated from 1 (Hardly ever) to 7 (Very often). The variation in wording reflects the original phrasing of the respective validated scales, which was retained without modification.

This study deployed partial least squares structural equation modeling (PLS-SEM) to analyze the complex interrelationships among the constructs. Although the study is theoretically grounded and confirmatory in nature, several factors justified the use of PLS-SEM over CB-SEM. First, while the main paths are derived from an established framework (EASI), the moderating role of leadership style, particularly in potentially reversing or amplifying relationships, was not fully established in prior research. Second, the sample size of 296 participants, though adequate, is relatively modest for a complex CB-SEM model with multiple latent constructs and 49 indicators. PLS-SEM offers greater flexibility for estimation under these conditions without compromising statistical reliability. Finally, PLS-SEM emphasizes prediction and variance explanation (R²), which aligns with the study’s interest in mediating and moderating mechanisms in employee behavior. Therefore, PLS-SEM was selected as the most appropriate technique for the present study (Hair et al., 2019). In line with conventional practice, effects are considered statistically significant at p < 0.05. Results with p-values between 0.05 and 0.10 are reported as marginally significant and interpreted cautiously.

The demographic information of the sample is presented in Table 1. Most participants were female (67.9%), followed by male (30.4%) and with a small proportion identifying as other genders (1.7%). Participant ages ranged from 18 to 84 years, with the largest proportion in the 35–44 age group (33%); overall, 79% were aged 25–54. Regarding education, the largest portion reported completion of some college education (42.2%), while 38.8% held a bachelor’s or graduate degree. Nearly half of the participants (47%) reported annual incomes between $30,000 and $59,999, indicating a predominantly middle-income sample. The largest share of participants worked in the food and beverage sector (49.3%), followed by accommodation (26%) and transportation or tourism services (9.1%), with smaller proportions in parks, entertainment and recreation and other sectors. Most participants (54.7%) worked in front-of-house roles, 24.7% in both front and back-of-house roles and 16.2% in back-of-house positions. Regarding job levels, 46% held entry-level roles, followed by 24.7% in mid-level management and 22.6% in supervisory positions. Most respondents (68.2%) reported over ten years of work experience, while 27.4% had between three and ten years. With respect to their current supervisor, over half (52.4%) had fewer than three years of experience with their current supervisor, and 42.2% had between three and ten years. Slightly more participants reported having female supervisors (52%) than male (48%).

To mitigate common method bias (CMB), a time-lagged data collection approach was implemented. Additionally, Harman’s single-factor test was conducted to evaluate the presence of CMB (Podsakoff and Organ, 1986). The results indicate that in the second (broader) model, the largest proportion of variance explained by a single variable is 25.74%, less than the 50% threshold (Podsakoff et al., 2003). Based on the research design and the results of Harman’s single-factor test, it may be concluded that this study remains unaffected by CMB.

Several steps were undertaken to examine the outer model. The first was to assess the reliability of individual items, with the corresponding factor loadings presented in Table 2. Among the 49 items analyzed, 9 exhibited loadings below the threshold (0.70), with the lowest loading at 0.551. Nonetheless, these items were retained, as the overall construct reliability, convergent validity and discriminant validity remained within acceptable limits (Hair et al., 2019). To evaluate the internal reliability, Cronbach’s alpha (α), rho_A and composite reliability (CR) were computed, with threshold values of 0.60, 0.70 and 0.80, respectively (Suneera et al., 2025). All constructs meet the requirements, confirming satisfactory internal reliability. Convergent validity was also assessed through the average variance extracted (AVE). Following Fornell and Larcker’s (1981) guideline, an AVE value of at least 0.50 is considered acceptable. All constructs in this study met this criterion, demonstrating adequate convergent validity.

Finally, the Fornell−Larcker criterion was satisfied (Fornell and Larcker, 1981), confirming the discriminant validity of the model (Table 3). In addition, the Heterotrait-Monotrait ratio of correlations (HTMT) is checked (Henseler et al., 2015). As shown in Table 4, all HTMT values are below the conservative threshold of 0.85, indicating sufficient discriminant validity among the constructs. This analysis further confirms that the constructs are empirically distinct.

The grounded hypotheses in the model are tested in a second-order inner model analysis, and the results are presented in Figure 2. Age, gender, education, income and front-/back-of-house role were controlled in the model for task performance to ensure that the path effects under investigation are not influenced by any confounding variables (Podsakoff et al., 2003). The results showed insignificant impacts from all controlled variables. In this model, it is confirmed that while leader expression of controlled anger significantly decreases employee anger (β = −0.209, p = 0.002), leader expression of outward anger significantly increases employee anger (β = 0.159, p = 0.042). On the other hand, leader anger expression, whether controlled or outward, increases job crafting (β = 0.224, p = 0.002 and β = 0.261, p = 0.001, respectively). Meanwhile, employee anger significantly decreases both job crafting and task performance (β = −0.258, p = 0.000 and β = −0.306, p = 0.000, respectively). Finally, job crafting significantly increases task performance (β = 0.239, p = 0.000).

In addition to the path coefficients, the effect size (f²) and predictive relevance (Q²) were assessed to provide a more comprehensive evaluation of the model. The f² values indicate that controlled anger has a small effect on employee anger (f² = 0.032), whereas anger out shows only a negligible effect (f² = 0.019). Leader anger expression, both controlled and outward, exerts small effects on job crafting (f² = 0.037 and 0.050, respectively). Employee anger also demonstrates small effects on both job crafting and task performance (f² = 0.067 and 0.106). Finally, job crafting has a small effect on task performance (f² = 0.063). The blindfolding procedure revealed that all endogenous constructs exhibit Q² values above zero (employee anger = 0.085; job crafting = 0.041; task performance = 0.043), confirming the model’s predictive relevance (Hair et al., 2019).

The analysis of the indirect effects (see Table 5) revealed consistent and meaningful patterns across the base model. The path-controlled anger → employee anger → job crafting was positive and statistically significant (β = 0.054, p = 0.019) and sequential mediation to task performance was also positive and statistically significant (β = 0.013, p = 0.043). In contrast, the path outward anger → employee anger → job crafting was negative and marginally significant at the 10% level (β = −0.041, p = 0.082), whereas the sequential path to task performance was not statistically significant. These results supported H1 and H3, provided marginal support for H2 and failed to support H4.

After testing the baseline mediating model, ethical leadership was introduced as a moderator, and moderation analysis was conducted. The interaction between ethical leadership and controlled anger was positive and statistically significant (β = 0.149, p = 0.042). However, because H5 predicted that ethical leadership would amplify the negative effect of controlled anger on employee anger, the positive interaction coefficient indicates the opposite pattern. Specifically, ethical leadership appears to weaken rather than strengthen the negative relationship. Therefore, H5 is not supported. In contrast, the interaction between ethical leadership and outward anger was positive and highly significant (β = 0.268, p < 0.001), indicating that ethical leadership strengthens the positive relationship between outward anger and employee anger. Thus, H6 is supported (Table 6). Moreover, when moderated mediation analyses were undertaken, they demonstrated that ethical leadership significantly moderated several indirect effects. The analysis revealed that ethical leadership strengthened the negative indirect effect of outward anger on job crafting (β = −0.032; p < 0.05) and task performance via employee anger (β = −0.035; p < 0.05), and the positive indirect effect of controlled anger on task performance via job crafting (β = 0.019; p < 0.01).

In addition, several demographic and job-related variables, such as age, gender, education, income, front-/back-of-house role, etc. were included as control variables in the analysis. As shown in  Appendix 1Table A1, none of these controls had a statistically significant effect on the final endogenous variable in the model, task performance and their inclusion did not meaningfully alter the magnitude or significance of the hypothesized relationships, consistent with established practice in organizational research (Podsakoff et al., 2003). We also conducted supplementary analyses examining the three job crafting dimensions separately (see  Appendix 2Table A2). The results showed that both controlled and outward leader anger were positively associated with task, cognitive and relational crafting. Moreover, when moderated mediation analyses ( Appendix 3Table A3) were undertaken, they demonstrated that ethical leadership significantly moderated several indirect effects. The analysis revealed that ethical leadership strengthened the negative indirect effect of outward anger on job crafting (β = −0.032; p < 0.05) and task performance via employee anger (β = −0.035; p < 0.05), and the positive indirect effect of controlled anger on task performance via job crafting (β = 0.019; p < 0.05).

The current study sought empirical evidence for the potentially constructive effects of leader expressions of anger in hospitality and tourism. The findings indicate that leader anger is not uniformly constructive or destructive; rather, its consequences depend on how it is expressed. Controlled anger was found to reduce employee anger and increase job crafting, thereby improving task performance. In contrast, outward anger increased employee anger while also encouraging job crafting, resulting in more mixed outcomes. Although outward anger showed a marginally negative indirect effect on job crafting through employee anger, its indirect effect on task performance was insignificant. The findings further demonstrate that ethical leadership shapes employee emotional responses to leader anger. Specifically, ethical leadership weakened the anger-reducing effect of controlled anger while strengthening the positive relationship between outward anger and employee anger. Together, these findings challenge the dominant view that anger is uniformly destructive to employee outcomes (e.g. Moreo et al., 2020), suggesting instead that its effects depend on both the form of anger expression and the leadership context in which it occurs. Recent hospitality research has suggested similarly that reprimands and other negative interpersonal communications are not inherently counterproductive; rather, their effectiveness depends on how they are delivered and interpreted (Silvestro and Silvestro, 2026).

These findings align with the EASI model by showing that individuals respond to expressions of dissatisfaction by adjusting their behaviors. Outward leader anger increased employee anger, highlighting its negative emotional consequences. By contrast, controlled anger increased job crafting and task performance while reducing employee anger. This finding is consistent with Andrus et al. (2025), who highlighted that employees respond to contextual cues. Thus, controlled leader anger produced the most favorable overall outcomes. The results also showed that employee anger significantly decreased job crafting and task performance, consistent with previous research (Hwang et al., 2021).

The mediation analysis revealed that leader expressions of controlled anger enhanced both job crafting and task performance indirectly by reducing employee anger (supporting H1 and H3). This finding highlights the constructive role of regulated anger, enabling employees to manage their emotions while remaining engaged and proactive. In contrast, the mediation effect of outward anger on job crafting through employee anger was negative and marginally significant at the 10% level (H2 received marginal support), whereas the sequential mediation effect on task performance remained non-significant (H4 not supported). These findings suggest that openly expressed leader anger exerts a weak but detrimental indirect effect on job crafting through an increase in employee anger. Although outward anger stimulated job crafting and task performance directly, it also increased employee anger, which undermined both outcomes. These opposing effects likely offset one another, producing only a marginal indirect effect on job crafting and no significant indirect effect on task performance.

The moderation analysis revealed that ethical leadership plays a complex role in shaping how leader anger expression influences employee emotions. Contrary to H5, ethical leadership weakened rather than amplified the negative relationship between controlled anger and employee anger. Although controlled anger generally reduced employee anger, this calming effect diminished when leaders were perceived as highly ethical. One possible explanation is that followers hold stronger expectations of emotional composure and moral consistency from ethical leaders. Even when anger is expressed in a controlled manner, it may violate these expectations, reducing its effectiveness as a regulatory signal. We may conclude that ethical leadership may constrain rather than strengthen the benefits of controlled anger. In contrast, and in support of H6, ethical leadership strengthened the positive relationship between outward anger and employee anger. When ethical leaders display outward anger, followers may experience stronger emotional reactions because of the incongruity between the leader’s moral image and intense expression of emotions. This incongruity may amplify employees’ emotional arousal, consistent with arguments regarding ethical leadership and moral expectations (e.g. Brown et al., 2005).

This pattern was further clarified by the simple slope analyses. For controlled anger, the relationship with employee anger was significantly negative at low (−1 SD) levels of ethical leadership, slightly negative at the mean and positive at high (+1 SD) levels (see Figure 3), indicating that ethical leadership not only attenuated but reversed this relationship. When leaders were perceived as low in ethical leadership, controlled anger appeared to signal emotional regulation and professionalism, reducing emotional contagion and employee anger. Conversely, among highly ethical leaders, controlled anger may have been perceived as inconsistent with expectations of transparency and authenticity, thereby increasing emotional sensitivity and anger amongst employees.

For outward anger, the relationship with employee anger was strongly positive at high (+1 SD) levels of ethical leadership, modestly positive at the mean and slightly negative at low (−1 SD) levels (see Figure 4). These findings suggest that ethical leadership amplifies the emotional transmission of outward anger, whereas at low levels of ethical leadership the relationship weakens and reverses. When highly ethical leaders expressed outward anger, employees appeared to react more strongly because such behavior conflicted with their positive moral image. By contrast, outward anger from leaders perceived as less ethical may have been less surprising and therefore elicited weaker emotional reactions.

Taken together, these results underscore the complex and context-dependent nature of leader anger, emphasizing that leadership style serves as a key interpretive lens through which employees evaluate and respond to leaders’ emotional displays.

Drawing on the EASI model, this study has examined how leader expressions of anger influence task performance by hospitality and tourism employees through employee anger and job crafting. The findings demonstrate that leader anger is not universally detrimental. Job crafting can enhance task performance by promoting both controlled and outward anger. However, the two forms of anger produced divergent emotional consequences. Controlled anger reduced employee anger and indirectly supported constructive behavioral and performance outcomes, whereas outward anger increased employee anger, thereby weakening its indirect benefits. Interestingly, the findings suggest that ethical leadership may not always be the most compatible leadership context in environments where anger expression is commonplace and is functionally embedded within daily operations. Ethical leadership weakened the anger-reducing effect of controlled anger and intensified employee emotional reactions to outward anger, suggesting that the moral expectations associated with ethical leadership may increase employee sensitivity to expressions of anger by leaders. Overall, this study advances understanding of leader emotions in hospitality by demonstrating that the consequences of anger depend not only on how it is expressed but also on the leadership context through which it is interpreted by employees.

This study makes two primary contributions to knowledge on leadership, emotions and service management. First, the study extends the EASI model by providing a broader understanding of the social-relational context that shapes employee responses to leader expressions of emotion. The EASI model proposes that the effects of emotional expressions go beyond the emotions themselves and extend to the social-relational context in which they are interpreted (Van Kleef, 2009). Prior research on EASI and on leadership has shown that interpersonal and relational contexts influence employee reactions to leader emotions (Wang et al., 2018). Leadership-related factors may also condition the effects of emotional displays (Connelly and Ruark, 2010). However, EASI related research has focused primarily on interaction-specific moderators, including appropriateness judgments, display rules, power differences and information-processing conditions (Van Kleef and Côté, 2022). The current findings extend this perspective by demonstrating that leadership style functions as a stable leadership characteristic that shapes employees’ interpretations of leader anger. By showing that ethical leadership has a significant impact on employees’ emotional responses to both controlled and outward anger, this study broadens the EASI framework beyond interaction-specific conditions and identifies leadership style as an enduring social-relational condition that influences how leader emotional expressions are interpreted. This emphasis on contextual interpretation is consistent with broader hospitality research highlighting the importance of organizational environments and social contexts in shaping individuals’ interpretations of interactions and subsequent behavioral responses (Bordbar et al., 2026).

Second, this study contributes to research on emotions in hospitality by offering a more differentiated understanding of leader anger in service settings. Previous hospitality studies have primarily emphasized the detrimental consequences of anger, including burnout, emotional exhaustion, workplace mistreatment and reduced well-being (e.g. Moreo et al., 2020). Our findings demonstrate that leader anger is not inherently dysfunctional and that its consequences depend on how it is expressed. Job crafting is positively associated with both controlled and outward anger which suggests that anger can motivate employees to make proactive adjustments to their work behaviors as a response to performance concerns. At the same time, the two forms of anger produced different emotional consequences: controlled anger reduced employee anger, whereas it was increased by outward anger. By distinguishing between controlled and outward anger, this study advances hospitality scholarship beyond a predominantly negative view of leader anger and demonstrates that different forms of anger trigger distinct emotional and behavioral responses in service environments.

Anger is a frequent and often unavoidable emotion in hospitality and tourism (Tuerlan et al., 2021) due to high customer demands, intense workloads (Mansour and Tremblay, 2016) and constant interpersonal interactions (Zhao et al., 2024). Because anger is commonplace, hospitality organizations should maximize its constructive potential while minimizing its negative consequences. The current findings indicate that both controlled and outward leader anger can directly enhance job crafting and indirectly improve task performance, consistent with Lindebaum et al. (2016). However, outward anger also increases anger amongst employees, thereby undermining psychological safety and increasing turnover intentions over time (Akarsu et al., 2023), whereas controlled anger reduces employee anger. Importantly, these effects are not uniform across contexts, since leadership style shapes how employees interpret and respond to leader anger, thereby altering its emotional and behavioral consequences. Based on these findings, hospitality managers may contemplate adoption of the following strategies.

Based on these findings, hospitality organizations should encourage leaders to express anger constructively by directing feedback toward employees’ behaviors rather than personal characteristics and providing clear, goal-oriented, actionable guidance. For example, after a service mistake, corrective feedback should be delivered promptly and focus on solutions that improve performance. When anger is perceived as deliberate and performance-focused, employees are more likely to respond constructively rather than defensively (Shao et al., 2018). Leaders should also regulate emotional outbursts to minimize emotional contagion and conflict escalation, particularly during customer-facing interactions, by using simple self-regulation strategies to maintain composure in high-pressure situations. Organizations should promote fairness, transparency and integrity to foster more positive interpretations of leaders’ emotional expressions and greater receptiveness to corrective feedback. Finally, leadership style should align with emotional expression because it shapes how employees interpret leader anger. The present findings show that ethical leadership does not function as an emotional shield; instead, it weakens the calming effects of controlled anger while strengthening employee reactions to outward anger, suggesting that higher moral expectations increase employees’ sensitivity to leaders’ emotional displays. Accordingly, organizations should train leaders to align their emotional expressions with their leadership approach and situational demands.

By applying these strategies, hospitality organizations may channel leader anger more constructively to encourage proactive work behaviors and stronger task performance while minimizing the negative consequences of poorly regulated emotional expressions in high-pressure service environments.

Although this study provides valuable insights, several limitations should be noted. First, although a two-wave time-lagged design introduced temporal separation between the various constructs, it cannot provide a full confirmation of causality or capture dynamic changes. Because the variables in the proposed mediation chain were measured within the same wave, the sequential mediation should be interpreted with caution. Future researchers are urged to deploy experimental designs, multi-wave longitudinal studies and within-person daily diaries. Such approaches may reduce common method bias, strengthen causal inference and distinguish more clearly between disentangle antecedents, mediators and outcomes (Podsakoff et al., 2003). It is acknowledged that self-reported anger measures may be susceptible to social desirability bias. Consequently, future studies may incorporate multi-source data, behavioral observations and/or implicit measures and thereby enhance the assessment of emotional responses.

The exclusive use of a US sample limits the generalizability of the findings across multiple cultural and organizational contexts. Future studies should examine a range of industry sub-sectors, regions and cultures to explore how contextual factors shape the relationships between leader anger, employee responses and job outcomes. Beyond ethical leadership, future researchers may investigate alternative leadership styles, such as servant and paradoxical leadership, as well as broader organizational ethical culture (Ruiz-Palomino et al., 2025), to explain the circumstances where leader anger leads to constructive outcomes. Prior research has suggested that self-directed and other forms of anger, may promote positive outcomes (Lench et al., 2023). In particular conditions, leader anger can motivate employees rather than provoke anger (Lindebaum et al., 2016; Shao et al., 2018). Future researchers may engage in the development and experimental testing of models to identify the leadership approaches and situational conditions under which leader anger motivates employees while minimizing adverse emotional reactions.