Burnout and counterproductive workplace behaviours among frontline hospitality employees: the e ﬀ ect of perceived contract precarity

Purpose – This study aims to investigate affective commitment (ACS) and leader – member exchange (LMX) as resources mitigating against burnout and counterproductive work behaviours (CWBs) in the hospitality sector, andexamines the effect ofzero-hourcontracts ontheserelationships. Design/methodology/approach – Through conservation of resources theory, this study tests a framework exploring ACS and LMX as resources against burnout and CWBs, using a dataset of 260 frontline hospitalityemployees workingin Ireland,considering zero-hourcontracts asa moderator. Findings – Findings indicate that burnout is associated with CWB, and ACS and LMX are resources against burnout and CWB. Furthermore, zero-hour contract perceptions moderate the resource effect of ACS and LMX. Yet, zero-hour contract perceptions do not moderate the relationship between burnout and CWB, indicatingtheseemployees may be caught ina resource-lossspiral. Practical implications – This study proposes mechanisms to enhance resources against burnout, with speci ﬁ c strategies to support young employees who are more likely to experience burnout. As ﬁ ndings suggest unique negative impacts of burnout for employees on zero-hour contracts, this paper also provides guidanceto supportthesevulnerable employees. Originality/value – This study provides unique insights into hospitality employees ’ ability to harness resources against burnout and CWB consequences of burnout. The results indicate that perceived precarity does notmoderatetheserelationships,suggesting that burnoutaffects this cohort differently.


Introduction
Extant research recognises that organisation-specific features of work support life satisfaction (Hassan et al., 2021). Recent studies have examined the impact of organisational factors on the well-being of frontline workers (Selzer et al., 2021) and on hospitality workers, in particular (Ayrom and Tumer, 2021;Kotera et al., 2021). Hospitality workers experience an especially challenging work environment, due to workload, working hours, the emotionally demanding nature of work (Kotera et al., 2021) and customer incivility (Boukis et al., 2020;Kim and Qu, 2019), leading to high levels of burnout (Abubakar et al., 2021;Ayachit and Chitta, 2022).
Compounding these issues, many frontline hospitality employees are considered "casual", where they are available for work, but their hours are not specified under contract and the employer is not obliged to provide minimum or specified hours. This is a situation known as a "zero-hour contract" (Farina et al., 2020). Koumenta and Williams (2019) list hospitality workers among seven of the top ten occupational types where zero-hour contracts are prevalent. When employees perceive that they do not have a say in their hours worked per week, or they cannot negotiate their work schedule, they may perceive lower workplace support (Creed et al., 2020) and insecurity in other aspects of their lives (Campbell and Price, 2016). These feelings may result in burnout (Allan et al., 2021;Khan et al., 2022). Given the prevalence of precarious employment in hospitality, rising concerns about organisational "custom and practice" and social norms and associated vulnerabilities of "zero-hour" workers (Ioannou and Dukes, 2021), it is important to consider these workers in a study of burnout.
Burnout is prevalent among hospitality workers (Bufquin, 2020;Karatepe, 2015;Zhang et al., 2020). Yet, recent papers have demonstrated a paucity of related research in Western Europe, with extant literature predominantly investigating negatively valenced concepts such as customer incivility (Boukis et al., 2020;Kim and Qu, 2019), abusive supervision (Yu et al., 2020a(Yu et al., , 2020b and emotional labour (Chen et al., 2019) as antecedents of burnout (Abubakar et al., 2021;Ayachit and Chitta, 2022). There is a lack of research investigating factors to mitigate burnout and its effects. Furthermore, predominant outcomes studied in extant research are job performance and intention to quit (Ayachit and Chitta, 2022) with less focus on behaviours, including counterproductive work behaviours (CWBs), exhibited by employees due to burnout.
Building on calls for research to investigate hospitality employees' coping strategies for burnout (Ayachit and Chitta, 2022) and the high levels of burnout in hospitality (Ayachit and Chitta, 2022;Bufquin, 2020;Karatape, 2015), this study investigates burnout among hospitality employees in Ireland. The study explores the consequential impact on the level of CWBs. CWBs are deliberate acts that harm the organisation or its members (Carpenter et al., 2021). Zhao et al. (2013, p. 219) cautioned that "of the negative behaviours that employees develop at work, none are more prevalent than CWBs". The nature of the hospitality sector, with uncertainty about working hours and associated fluctuations in financial well-being, and perceived inequities related to workloads, can engender motives for these deviant behaviours (Harris and Ogbonna, 2012;Zhao et al., 2013). CWBs have an undesirable influence on customer evaluations of service, negatively impacting satisfaction and loyalty, and damaging long-term profitability (Harris and Ogbonna, 2006;Vatankhah and Darvishi, 2021).
This study draws on conservation of resources (COR) theory (Hobfoll, 1989), a resourcebased theory of stress and a theoretical lens popular in investigating behaviours in the hospitality sector (Ayachit and Chitta, 2022;Xu et al., 2020;Zhou et al., 2018). COR suggests that psychological strain occurs when individuals invest resources but fail to gain the level of return anticipated (Halbesleben, 2006). Extant literature also proposes resources to "buffer" employees against negative outcomes such as burnout. We investigate two of those "buffers", specifically affective commitment (ACS; in line with Lapointe et al., 2011) and leader-member exchange (LMX; in line with Montani et al., 2017), as resources against burnout for hospitality employees.
Our study makes several important theoretical contributions. Firstly, we answer calls from the literature to investigate CWBs in hospitality (Vatankhah and Darvishi, 2021;Zhou et al., 2018). This paper demonstrates that burnout is connected to both CWB against the individual (co-workers) and CWB against the customer/organisation. Secondly, drawing on COR theory (Hobfoll, 1989) and recognising the importance of organisation-specific features in improving and maintaining life satisfaction (Hassan et al., 2021), we explore LMX and ACS as resources against burnout. We find that LMX and ACS play important roles in helping to reduce employees' perceptions of burnout. Thirdly, we investigate the role of precarious contracts on these relationships, as these contracts are prevalent in the hospitality sector (Koumenta and Williams, 2019). We explore "zero-hour" contract perceptions as a moderator and we find that they reduce the effect of ACS and LMX as resources against burnout, and they do not directly affect CWBs among employees experiencing burnout. We, therefore, caution that burnout affects this vulnerable cohort differently.
This paper unfolds as follows. We begin by reviewing COR theory, and we develop our conceptual framework, building our hypotheses. We then explain our sampling strategy and our measures. Our results are presented and we discuss the implications of our work for theory and practice. Before concluding on our study, we outline the limitations of our approach and we propose future research directions.

Hypothesis development 2.1 Theoretical foundations: conservation of resources theory
This study is underpinned by the COR framework (Hobfoll, 1989), a resource-based theory of stress (Hobfoll, 2001). COR has been one of the dominant theories in organisational behaviour for the past 30 years (Hobfoll et al., 2018) and is often used in hospitality studies examining burnout (Abubakar et al., 2021;Ayachit and Chitta, 2022;Hassan et al., 2021). Particular advantages of COR theory in this context are its ability to account for diverse contexts, and its ability to capture different stress-shaping factors, rather than laying the burden of dealing with stress on the beholder (Zhang et al., 2021). Hobfoll (1989) explained that individuals with sufficient resources are more equipped to deal with stresses and to cope with them. COR, therefore, suggests that individuals can gain resources through aspects of work such as supportive relationships with managers (Jolly et al., 2022). COR theory also suggests that if resources are lost, stress increases, and employees will seek ways to replenish those resources (Hobfoll et al., 2018). Otherwise, resource loss leads to an impairment in employee well-being, potential burnout (Halbesleben, 2006) and even maladaptive coping (Alarcon, 2011). We, therefore, investigate burnout, CWB and proposed resource buffers, through the lens of COR.
While COR is a useful lens through which to investigate employee motivation and their coping with job demands (Hobfoll et al., 2018), Halbesleben et al. (2014, p. 15) highlight the challenge associated with the "timeframe in which resource processes play out". It might not be possible for employees to acquire resources indefinitely, for example. Moreover, Hobfoll (1989) suggests that individuals in traumatic states could be less proactive in gaining resources, and could therefore lose resources, which could result in a "loss spiral" (Hassan et al., 2021). This study investigates the buffering effect of resources, among employees in the hospitality sector and uniquely also considers employees who are on precarious contracts that offer less stability. Our framework considers that, due to precarity, those employees might not be able to build up resources, leaving them susceptible to loss-spirals that further inhibit the buffering effect of those resources.
2.1.1 Conceptual framework. Figure 1 sets out our conceptual framework. Drawing on COR, we posit that resource depletion through burnout is associated with CWBs, against both individual co-workers (CWBI) and the organisation (CWBO). In line with COR, we propose resources against burnout, specifically LMX and ACS. We contend that these relationships are moderated by the extent to which the employee perceives they are on a zero-hour contract. We describe these constructs and provide theoretical underpinning for these relationships next.

Burnout and counterproductive workplace behaviour
Burnout has been classified as an "occupational phenomenon" by the World Health Organization (WHO, 2019). It is defined as "a psychological syndrome of emotional exhaustion, depersonalisation, and reduced personal accomplishment that can occur among individuals who work with other people in some capacity" (Maslach et al., 1996, p. 192). Research has typically focused on emotional exhaustion when investigating the phenomenon (Halbesleben and Wheeler, 2011;Tang and Vandenberghe, 2020). Emotional exhaustion refers to "feelings of being emotionally overextended and drained by one's contact with other people" (Leiter and Maslach, 1988, p. 297). The term "burnout" is therefore used throughout this paper to specifically describe emotional exhaustion.
Burnout is a pressing issue in hospitality (Ayachit and Chitta, 2022;Bufquin, 2020). Investigating burnout is important as, when resources are exhausted, individuals may enter a defensive mode to preserve the self. That mode may be defensive or irrational, and aggressive responses may result (Alarcon, 2011;Hobfoll et al., 2018) including CWBs in a hospitality context (Harris and Ogbonna, 2012;Lee and Ok, 2014). These behaviours are intentional and perceived to be justified from the employee's perspective (Smoktunowicz et al., 2015). Drawing on COR theory, employees who experience burnout may conserve resources by engaging in CWBs (Smoktunowicz et al., 2015;Yu et al., 2020a). Lee and Ok (2014) observed a positive relationship between burnout and CWB in hospitality. Bolton et al. (2012) found support for burnout influencing both CWBI and CWBO. In this paper, we consider whether employees who experience burnout engage in CWBs, investigating separately the impact of burnout on CWBI and CWBO, as a means of coping and curtailing resource loss. We hypothesise: H1. The higher the level of burnout, the higher the level of CWBI.

H2.
The higher the level of burnout, the higher the level of CWBO.

Leader-member exchange and affective commitment as resources against burnout
Supervisor support is identified as a driver of frontline employees' coping (Jolly et al., 2022;Vatankhah and Darvishi, 2021), in an environment where employees are often faced with solving customers' problems on their own. Moreover, in hospitality, management style is recognised as a means to decrease turnover intentions and improve service recovery (Ayrom and Tumer, 2021). Drawing on COR theory, Elkhwesky et al. (2022) and Zhou et al. (2018) identified the importance of leadership styles in shaping employee performance. Hobfoll et al. (2018) suggested both an affective connection to work and LMX as examples of routes to "gain spirals" (Salanova et al., 2010), whereby resources trigger the accumulation of further resources. While these findings made a substantial contribution, there remains a dearth of knowledge around factors that may diminish dysfunctional employee behaviours. We, therefore, propose LMX and ACS as resources to reduce burnout and consequently CWB.
2.3.1 Leader-member exchange. Supervisor support, where a supervisor "cares for an employee's well-being and offers work-related assistance to facilitate their job performance" can promote coping, or autonomous problem-solving behaviour (Selzer et al., 2021, p. 387). We investigate LMX as a resource to mitigate against burnout. LMX theory describes the level of support or interaction and trust between subordinates and leaders based on social exchange (Thomas and Lankau, 2009). High-quality social exchanges between managers and employees result in positive outcomes for the individual and the organisation (Graen and Uhl-Bien, 1995). Employees with better social exchanges with supervisors may be associated with increased resources (Hobfoll et al., 2018), and receive more support and resources from their relationship with their leader (Chang et al., 2020). Thomas and Lankau (2009) found that LMX improved organisational socialisation, reducing role stress and curtailing burnout. Boukis et al. (2020, p. 11) emphasised the importance of leadership style in the management of hospitality employees' psychological and behavioural responses. As hospitality is labour intensive, leaders who are difficult to understand, or disrespectful, or fail to communicate lose trust, affecting morale and performance (Chang et al., 2020;Elkhwesky et al., 2022). Yet within hospitality research, the relationship between LMX and other outcomes is largely overlooked, in spite of its potential effects on employees' behaviours and psychological states (Chang et al., 2020). We therefore hypothesise: H3. The higher the quality of LMX, the lower the level of burnout.
2.3.2 Affective commitment. ACS is comprised of positive emotions that relate to well-being and reduced strain (Meyer and Maltin, 2010). Tang and Vandenberghe (2020) suggest that ACS helps to reduce emotional exhaustion, the component of burnout under investigation in this current study. Under COR theory, ACS is identified as a buffer against burnout (Lapointe et al., 2011), and studies have generally reported a negative relationship between ACS and burnout (Hur et al., 2013;Kong et al., 2021). We expect that frontline employees in the hospitality sector can leverage the increased social support that comes with higher levels of ACS (Tang and Vandenberghe, 2020) to reduce their level of burnout (Hur et al., 2013;Kong et al., 2021). We hypothesise: H4. The higher the level of affective commitment, the lower the level of burnout.

Investigating perceptions of contract precarity as a moderator
The UN's Sustainable Development Goal of Decent Work and Economic Growth targets protecting labour rights and secure working environments for those in precarious employment. Precarious work includes forms of non-standard work (Campbell and Price, 2016) and uncertainty in the quantity of work (Allan et al., 2021). One characteristic of hospitality careers is irregular working hours (Goh and Lee, 2018), and uncertainty around those hours is recognised as a form of labour market insecurity (Farina et al., 2020) that can affect employee resources (Khan et al., 2022). Scholars distinguish between objective precarious work and subjective feelings of work precarity (Creed et al., 2020). This study investigates the impact of one form of precarity: uncertainty about hours worked, which is known as a form of "zero-hour contract" (Koumenta and Williams, 2019).
The rise of precarious work and its impact on employees' psychological experiences is a critical problem (Allan et al., 2021) and highlighted as an area of concern in hospitality (Koumenta and Williams, 2019). As noted earlier, these individuals may be less proactive in gaining resources, and could lose resources, resulting in a "loss spiral" (Hassan et al., 2021) inhibiting the buffering effects of resources for this vulnerable group. In addition, Huang et al. (2017) found that perceptions of job insecurity affected CWBs. Drawing on extant literature, we expect that workplace precarity would have a negative effect on the role of resources in reducing burnout and would increase the effect of burnout on CWBI and CWBO. We hypothesise: H5a. The positive relationship between burnout and CWBI will be increased in the presence of a perception of an employee being on a zero-hour contract.
H5b. The positive relationship between burnout and CWBO will be increased in the presence of a perception of an employee being on a zero-hour contract.
H5c. The negative relationship between the quality of LMX and burnout will be reduced in the presence of a perception of an employee being on a zero-hour contract.
H5d. The negative relationship between affective commitment and burnout will be reduced in the presence of a perception of an employee being on a zero-hour contract.
We next describe the method used to test our hypotheses.

Sample
We conducted a survey of frontline hospitality workers. Students at a large Irish university distributed the questionnaire online to the general public aged over 18, for credit, over a three-week period. Checking for duplicates by email address, and removing those who failed attention checks, did not take adequate time to complete the survey or had missing data on key variables, yielded a usable sample of 805 responses. Criteria for inclusion in this study were that the participants worked in a customer-facing role, in the hospitality sector. Both authors independently analysed the data (job title and organisational type) with inter-rater reliability of 98.08%. The remaining cases were discussed, and the final number of participants that worked in the hospitality sector was 260. There is no missing data in our final sample. Our sample is 56% female, with an average age of 23 years, and a standard deviation of 3.30 years. All respondents work as frontline employees in hospitality in paid employment, with 71.2% working in food and beverage (mainly bars and restaurants), 21.1% in lodging and hotels and 7.7% working in other tourism-related enterprises. Of the participants, 46.9% have less than one year's experience in their current role, 48.8% have 1 to 5 years and the remaining 4.3% have greater than 5 years of such experience. In all, 28.3% worked for their current manager for less than 6 months, 33.4% for between 6 months and one year and the remaining 38.3% for more than one year. In total, 48.4% work for smaller (less than 25 employees) organisations, 40.2% in mid-sized (between 25 and 100 employees) and the remainder (11.4%) work in larger organisations.

Measures
Items were adapted from well-established scales. Scale items are presented in Table 1, with their reliabilities. LMX (Liden and Maslyn, 1998) and ACS (Meyer and Allen, 1991) were measured using 7-point scales with anchors of "Strongly Agree" and "Strongly Disagree". Burnout was measured using the emotional exhaustion items from the Human Services Inventory (Maslach and Jackson, 1981). CWBI and CWBO were measured using the Workplace Deviance Scale by Bennett and Robinson (2000). The burnout, CWBI and CWBO items were measured on a 7-point scale asking respondents to indicate the frequency of occurrence anchored with "Never" and "Every day". Our "zero-hour contract" perception variable was measured using a single item -"Would you consider that you are on a 'zerohour contract'?", in line with extant research (Farina et al., 2020), where a lower score indicated that the worker perceived they were on this type of contract. We also investigated gender, age and organisation size as controls, in line with extant literature (Jolly et al., 2022).

Common method bias
We collected our data using survey measures from a single source thus there is potential for common method bias. The use of established scales and proximal separation served to reduce this risk (Podsakoff et al., 2012). Several post hoc tests have been developed to test for common method effect but perhaps the most effective is the marker variable method (Lindell and Whitney, 2001;Williams et al., 2010). The household size of respondents was used as a marker variable, as it was not expected to correlate with any of the key constructs. As per Lindell and Whitney (2001), a discounted correlation matrix was created. For common method variance not to be problematic, it is suggested that the discounted correlations should not change from significant to non-significant or change sign when compared with the original correlation matrix (Lindell and Whitney, 2001;Williams et al., 2010). The highest difference was 0.008, and there were no changes in the directionality or the significance of any of the zero order inter-construct correlations.

Results
We analysed our data using structural equation modelling with MPlus. The measurement model including all constructs had a good level of fit (x 2 = 1,065.754, degrees of freedom (df) = 580, root mean square error of approximation (RMSEA) = 0.052, comparative fit index (CFI) = 0.921, Tucker-Lewis index (TLI) = 0.914, standardized root mean squared residual (SRMR) = 0.062) with LMX modelled as a second-order construct with four first-order  Liden and Maslyn (1998). Table 1 depicts the items, their means, standard deviations and their reliabilities using composite reliability (CR) and average variance extracted (AVE). Table 2 is the correlation matrix with the square root of the AVE on the diagonal. Reliability was good with the exception of CWBO that fell slightly below the 0.5 required for AVE. Discriminant validity was tested, and all inter-construct correlations were below the lowest square root of the AVEs, except the relationship between CWBI and CWBO which, as expected, had a higher correlation, thus indicating that there is adequate discriminant validity. A structural model was set up to test H1 to H4 with controls, and including a correlation between CWBI and CWBO. The results of the structural model indicate that the model fits well (x 2 = 1,142.463, df = 650, RMSEA = 0.054, CFI = 0.912, TLI = 0.905, SRMR = 0.064), with the organisation size control dropped as it had no significant effect. H1 and H2 suggested a positive association between burnout and both CWBI and CWBO. H1 was supported, with an effect of 0.165 (p = 0.016) and H2 with an effect of 0.242 (p = 0.000). H3 suggested a negative relationship between LMX and burnout and the standardised effect of À0.390 was significant (p = 0.000). H4 suggested a negative relationship between ACS and burnout, and the standardised effect was significant though only at the 10% level with a value of À0.133 (p = 0.095). Gender had a significant effect on burnout with a significant effect of À0.147 (p = 0.008), suggesting that females are more likely to be burntout. Younger employees also had higher levels of burnout (effect of À0.196, p = 0.001) and higher levels of CWBI (À0.244, p = 0.000) and CWBO (À0.133, p = 0.047). As expected, there was a high and significant correlation (0.877, p = 0.000) between CWBI and CWBO.
We tested our moderating hypotheses (H5a-H5d) with our zero-hour contract variable as the moderator using the latent moderated structural method available in Mplus, a relatively robust method for assessing the effects of latent variables interacting with other variables all embedded in a structural model (Cheung et al., 2021). The benefit of this method is that it uses the latent variable calculated within the model to estimate the effect rather than the creation of a summary score as is more common in other estimation methods (Cheung et al., 2021). No support was found for the moderating effects of zero-hour contracts on the path between burnout and both forms of CWB (H5a and H5b). The standardised effect for H5c relating to the association between ACS and burnout was À0.630 (p = 0.000), demonstrating that being on a zero-hour contract reduces the buffering effect of ACS against burnout. The standardised effect for H5d relating to the association between LMX and burnout was À0.432 (p = 0.029), demonstrating that being on a zero-hour contract reduces the effect of LMX as a resource against burnout. Notes: * p < 0.10; ** p < 0.05; *** p < 0.01; ns = not significant; ACS = affective commitment; LMX = quality of the LMX; EE = burnout; CWBI = counterproductive work behaviours towards the individual; CWBO = counterproductive work behaviours towards the organisation To test if the moderators add value to the model, we use the Akaike information criterion (AIC) value (Akaike, 1974). AIC is a good measure of fit, as it prefers more parsimonious models (Cheung et al., 2021). When comparing nested models, a lower AIC is preferred (Akaike, 1974). The AIC of the base model (without any moderating effects) was 31,829.519, whereas the AIC of the model including the significant moderating effects only was 31,827.716; including the two non-significant effects increased the AIC to 31,829.576. As the model with the significant moderating effects had a lower AIC value than the base model, this model is said to have a better level of fit and the addition of the moderation improves the model.

Conclusions
Drawing on COR, this study tested a model of burnout buffers, specifically LMX and ACS, and burnout consequences, specifically CWBs, and investigating frontline hospitality employees' perceptions of being on zero-hour contracts as a moderator. Findings supported our hypotheses, indicating that both LMX and ACS are negatively associated with burnout, and burnout is positively associated with CWBs. In addition, our findings reveal that, among frontline hospitality employees who perceive they have a zero-hour contract, the "buffering" effect of both LMX and ACS against burnout are reduced. Yet, zero-hour contract perceptions have no moderating effect on the relationship between burnout and CWBs. We discussed these implications below.

Implications for theory
Our findings offer several theoretical contributions. Firstly, findings provide support for the consideration of the COR perspective in hospitality research, in line with recent studies (Ayachit and Chitta, 2022;Boukis et al., 2020;Xu et al., 2020). Specifically, our study proposed LMX and ACS as resources against burnout. Boukis et al. (2020) revealed that an empowering leadership style reduced role stress. We add to this research by showing a negative relationship between LMX and burnout. Building on substantial LMX research by Chang et al. (2020), we show that when employees perceive a positive relationship with their line manager, this helps to protect them against burnout.
Drawing on COR theory, we also reveal ACS as a buffer against burnout. While our result was significant only at a 10% level, we nevertheless show that employees who have greater emotional attachment to their organisation are less likely to be burnt out, in line with extant literature (Kong et al., 2021;Tang and Vandenberghe, 2020). We offer a suggestion for the weaker buffering effect of ACS, relative to LMX, as resources in our study. Tang and Vandenberghe (2020) asserted that the relationship between ACS and burnout may be affected by the length of employee service, with long-tenured employees less likely to invest resources to widen their social network. Although length of service was not significant as a control in our study, our findings indicated more burnout among younger employees. Our study also indicates that, as burnout is greater among the younger age cohort, additional resources may be considered as potential "buffers", as ACS may not yet have stabilised and therefore its role as a buffer could be supplemented by other resource buffers. Furthermore, we find support for higher rates of burnout in female employees, supporting recent work by Jolly et al. (2022).
Secondly, we provide valuable insights into the relationship between burnout and CWBs in the hospitality sector. Findings show that burnout mediates the relationship between quality of LMX and CWBs, as when employee's relationships with their managers are weaker, burnout is greater and associated with CWBs. We also find that burnout mediates the relationship between ACS and CWBs, such that when ACS is lower, burnout is greater and associated with greater CWBs. Therefore, we contribute to the understanding of burnout and its impact on CWBs (Bolton et al., 2012;Lee and ok, 2014) in the hospitality sector. Furthermore, our study provides insights into a target-based conceptualisation of negative behaviours, as we show that burnout is positively associated with both CWBs against individuals, and CWBs against the organisation, highlighting the damaging effect of burnout.
A further interesting finding is that younger employees are more likely to engage in CWBs when they experience burnout. In hospitality, Harris and Ogbonna (2006) found that sabotage was associated with greater employee self-esteem and team spirit. We suggest that younger employees who experience burnout engage in CWBs partly to "regain" some of their resources, by fitting in with peers and thereby enhancing their self-esteem. We caution that CWB could therefore be used by young employees to achieve an esprit de corps when they experience burnout, and we advocate managers should put supports in place to curtail these CWBs.
Thirdly, our study investigates zero-hour contract perceptions as a moderator in our conceptual framework. The zero-hour contract has received much attention recently, with media coverage voicing concerns about its effect on employee well-being and identity, contending that, rather than providing greater flexibility, these contracts are leaving workers in low pay and insecure work and are often unable to pay their bills (Mohdin, 2021). Yet, to our knowledge, the specific study of zero-hour contracts and their impact on hospitality employees has received little attention in the literature, other than official statistics-based studies of its prevalence (Farina et al., 2020). By investigating zero-hour contract perception as a moderator in our model, we show that this perception reduces the buffering effect of both LMX and ACS as resources against burnout.
Our results therefore suggest that employees who work under perceived contract precarity do not benefit in the same way from these "resources" as other employees. There are two important theoretical implications arising from this finding. Firstly, we noted earlier that a challenge with COR is the time frame during which the resource processes occur (Halbesleben et al., 2014). While this concern often relates to employees' ability to acquire enough resources to reach a "ceiling" of resources (Halbesleben et al., 2014), we find that, for precarious workers, their ACS and LMX are not as sufficient to buffer them against burnout as compared with those with less precarity. This interesting result may be a consequence of these employees' uncertainty about potential future work opportunities with the organisation, or their perceptions of their own identities as outsiders in relation to the organisation, which may leave them less able to develop and harness ACS and LMX to buffer them against the challenges of their work. Secondly, under COR, people with insufficient resources or those who have lost resources may be more likely to experience further losses, as they are caught in a "resource loss cycle" (Zhang et al., 2021 p. 3). Findings indicate that those who work under perceived contract precarity may be experiencing a "resource loss cycle", as resources that benefit other frontline workers do not benefit them in the same way, due to their precarity. Although these employees may have good relationships with their immediate line managers and experience ACS, they may still experience burnout because of the uncertainty around their working hours.
Perhaps our most surprising finding is that zero-hour contract perceptions had no moderating effect on the relationship between burnout and CWB. While finding is reassuring, as it suggests that these employees are not more likely to "act out" through CWB, it also raises questions about burnout's effects for employees who are on zero-hour contracts, and in particular the potential well-being implications for these individuals.
Previously, Creed et al. (2020) cautioned that precarious work can disrupt a healthy identity development, diminishing life satisfaction. We caution that an inability to harness ACS and LMX as buffers may result in greater burnout for those on zero-hour contracts, and they do not subsequently retaliate via CWBs so instead, they may turn their response inwards.
Finally, we were also surprised to find that younger employees are more likely to be suffering from burnout, and more likely to engage in CWBs against both individuals and the organisation. The lower emotionality of older employees may lead them to maximise positive and minimise negative emotional experiences (Carstensen, 1992), and therefore younger employees may be more likely to feel more burntout although they share the same work experiences, and they may consequently engage in more CWBs. Our finding is in line with recent research that suggests older employees engage less in CWBs in general, partly due to age-related changes in personality traits such as honesty-humility, emotionality and conscientiousness (Pletzer, 2021). As hospitality businesses rely heavily on younger employees (Frye et al., 2020), and young people are the future hospitality workforce (Goh and Lee, 2018), these findings warrant further study. We discuss their implications for managers below.

Implications for managers
We provide practical suggestions for hospitality managers seeking to help prevent burnout and its negative outcomes for co-workers, customers and the organisation. We reveal ACS and LMX are effective buffers against burnout in hospitality. We advocate that firms would seek to enhance frontline employees' ACS, for example, by encouraging greater work-life balance through initiatives including managing overtime for a better reconciliation of work and family life (Hofmann and Stokburger-Sauer, 2017). Hassan et al. (2021) advocated support for work-life balance through alternative work arrangements, cafeteria benefits and on-site childcare and specialised workshops for workers and their families where feedback could be provided by family members about how work affects family life. We echo these calls, and we suggest that managers include precarious workers in work-life balance policies, to minimise burnout and support these employees to feel more included. Internal branding techniques (Ayrom and Tumer, 2021) can promote these initiatives.
Findings highlight the importance of job-related social support provided by LMX. Extant research suggests that role ambiguity can impact employee performance and well-being with frontline employees often solving customers' problems on their own, with little supervisor input during customer contact (Selzer et al., 2021). Therefore, supervisors exhibiting high-quality LMX skills could provide vital support to convey concern and offer guidance to employees (Thomas and Lankau, 2009). This manager-employee interaction would also help to curtail negative outcomes including burnout. By fostering ACS and LMX to reduce burnout, this would ultimately help to reduce CWB. Our findings show that younger employees are more likely to experience burnout, and therefore it may become difficult to attract young employees to work in the hospitality sector. We advocate that managers emphasise the positive aspects of the sector already perceived by young employees, such as interesting/exciting career, or travel opportunities (Goh and Lee (2018), in their recruitment campaigns. We also advocate for greater supports for young employees in the hospitality sector, to help them to cope with work pressures. For example, training could be tailored to address concerns of younger employees, and dramaturgical approaches such as role-playing could train employees in coping mechanisms to curtail burnout arising from interacting with co-workers and customers in high-contact service roles.
Moreover, earlier we noted that age was associated with CWBs, with younger employees engaging in these actions, perhaps to regain some resources lost through burnout. In line with extant research (Harris and Ogbonna, 2006), we caution that such behaviour may lead to a culture where younger employees develop an esprit de corps by acting out against the organisation and individuals. We advocate that hospitality managers might adopt a mentoring programme (Tulucu et al., 2022), whereby younger employees are supported by peers and managers to minimise the potential for CWBs and to create high performance norms for young hospitality workers. We also advocate internal branding mechanisms to train, brief and reward employees to foster adoption of organisational values, and support career growth (Ayrom and Tumer, 2021). These actions would support younger employees to feel a sense of "buy in", to internalise the organisation's values and help to mitigate against burnout.
Seasonality in the hospitality industry (Goh and Lee, 2018;Yu et al., 2020b) means companies may operate with a skeleton permanent workforce, with casual staff during peak times. This issue poses problems for hospitality managers. We highlight that the casual nature of work is problematic in its impact on burnout, as the buffering effect of LMX and ACS is lower for employees who perceive that they are on a zero-hour contract. In addition to legal protections for zero-hour employees, we advocate that organisations would be clearer about working hours, so that fewer employees would perceive that they are on a zero-hour contract. For example, while seeking to optimise workforce flexibility, giving employees more details of working rotas may provide them with greater certainty about their work commitments, even on a weekly basis.
Notably, the relationship between burnout and CWB was not moderated by zero-hour contract perceptions. Recent commentary on the casual worker has suggested that individuals who work in a gig economy may work in isolation without co-workers or supportive mentors, and experience a loss of community (Lillington, 2019). Other recent research cautions that precarious employment may be causing mental distress, as long and antisocial hours, unstable seasonal employment and low job status can compromise individuals' confidence, leading to self-criticism (Kotera et al., 2021). We caution that employees who perceive they are on a zero-hour contract may perceive that they are outsiders, and they may feel there is little point in engaging in CWB, yet they may internalise the effects of burnout. We advocate that management ensure to extend their duty of care towards these workers, so that all workers are protected from burnout and its negative effects. Moreover, managers should take specific actions to support those employees who experience burnout. Vatankhah and Darvishi (2021) propose practical strategies to mitigate against CWBs. For example, training schemes to improve social interactions, promote a healthy environment and destigmatise those experiencing challenges at work, could help to curtail burnout and issues such as counterproductive behaviours and facilitate positive behaviours in hospitality (Kotera et al., 2021).

Limitations and future research directions
As with all studies, this research has limitations. Firstly, as the measures were self-reported, our study may have been affected by common method bias (Podsakoff et al., 2012). However, the analysis undertaken provides reassurance that common method variance does not pose a threat to the interpretation of our results. Given the nature of our research, there is also a potential risk of social desirability bias where highly sensitive topics such as CWBs are investigated. To mitigate this, all respondents were assured of anonymity, and no specific details about workplaces were collected. Extant studies on frontline employee performance in hospitality considered the psychological contract (Morrison and Robinson, 1997), with contract fulfilment shaping a chain of positive reciprocity through extra-role behaviour (Ayrom and Tumer, 2021). We advocate research to explore the role of the psychological contract as a potential resource against burnout and its consequences.
Also, we considered the relationship between burnout and CWBs. Recent research has suggested that CWB can be aggregated to the team level (Carpenter et al., 2021) and while our dataset does not capture this effect, it would be an interesting phenomenon to investigate in a hospitality context where team working is prevalent. Earlier we noted that the relationship between burnout and CWBs was not moderated by zero-hour contract perceptions. We cautioned that, if those on zero-hour contract were burnt out, this could instead result in self-destructive outcomes. Murray-Gibbons and Gibbons (2007) found that some hospitality workers experiencing stress engaged in externalisers such as not exercising, under/overeating, drinking, smoking or ignoring stress, as coping mechanisms. These externalisers are counterproductive to individual well-being. We advocate further study on burnout and zero-hour work to investigate the impact of burnout on these behaviours and other negative consequences for the well-being of employees. Furthermore, extant research considers employee shame and self-criticism as factors negatively impacting employees' well-being at work (Kotera et al., 2021). We suggest further study might consider employee shame and self-criticism in a model of burnout, where perceptions of zero-hour contracts are investigated.
Moreover, we measured the perception of zero-hour contract by asking employees whether they knew their working hours in advance, in line with extant methods (Farina et al., 2020;Koumenta and Williams, 2019). We did not consider the specific hours offered to these employees, and so we did not consider whether those working hours were also antisocial. We advocate future studies of zero-hour contract workers would adapt our model, and investigate whether the antisocial nature of hours offered moderates the relationships. Finally, our study was conducted among frontline employees working in the hospitality sector in Ireland. We suggest that our model would be explored in other cultural contexts in line with the gaps in geographical coverage of burnout set out by Abubakar et al. (2021). Investigating our model in other cultural contexts may reveal new insights into these relationships. Nevertheless, we hope that the findings will provide helpful insights for hospitality managers seeking to curtail burnout and its consequences, and to support employees who perceive they work under zero-hour contract conditions.