Does Emotional Labor Trigger Turnover Intention? The Moderating Effect of Fear of COVID-19

Abstract

Turnover is a costly and time-consuming expense, especially for service industry businesses. To date, little is known about whether and how emotional labor may activate employee turnover intention in the service industry. In order to solve the above problems and fill the gaps, this study aimed to verify how emotional labor can trigger turnover intention during the COVID-19 pandemic. Based on job characteristics theory and job demands–resources theory, this study examined whether emotional display rules and emotional labor strategies affect turnover intention brought on by emotional exhaustion and job dissatisfaction, with fear of COVID-19 as a moderator. After testing our hypotheses using a sample of 623 individuals from China’s service industry, this study found that emotional display rules (positive and negative display rules) are significantly related to emotional labor strategies (deep acting, expression of naturally felt emotions, and surface acting). In particular, positive display rules have a positive impact on deep acting and the expression of naturally felt emotions and are more closely related to the expression of naturally felt emotions. Negative display rules negatively affect surface acting. Moreover, emotional labor strategies correlate significantly with emotional exhaustion, job satisfaction/dissatisfaction, and subsequent turnover intention. Thus, deep acting and the expression of naturally felt emotions are related to low emotional exhaustion and high job satisfaction, while surface acting is related to high emotional exhaustion and low job satisfaction. Emotional exhaustion has a negative effect on job satisfaction and a positive effect on turnover intention. Job satisfaction significantly weakens turnover intention. In addition, fear of COVID-19 has a moderating effect on the relationship between job satisfaction and turnover intention. The group with a high fear of COVID-19 has higher turnover intention even in job satisfaction situations than the group with a low fear of COVID-19. This work advances emotional labor research by combining two dimensions of emotional display rules and three dimensions of emotional labor strategies into a framework, investigating the mechanism through which emotional labor influences turnover intention, and revealing the moderating effect of fear of COVID-19 in the process.

1. Introduction

As an economy matures, its economic focus often shifts from manufacturing to services. Thereafter, as the proportion of service industry firms increases, competition among service firms intensifies. Then, due to globalization and the diversity of consumer needs, it becomes increasingly difficult to maintain a competitive advantage. Thus, service firms must continually endeavor to enhance service quality and organizational performance.

In customer-oriented services (e.g., department stores, discount stores, hotels, hospitals, financial institutions, and airlines), front-line employees play a key role in delivering efficient services and creating/maintaining a competitive advantage [1]. Face-to-face interaction with customers is a form of emotional labor for these front-line employees, and so job-related emotion profoundly affects the performance and long-term success of service firms. Therefore, this study aims to explore the mechanism through which emotional labor triggers the turnover intention of front-line employees.

To guide employees in the expression of appropriate emotions, service firms frequently establish norms for expressing appropriate emotions. These are known as emotional display rules [2]. Studies show that regulated emotional displays support firm outcomes. However, their impact on the psychological and social well-being of front-line employees remains unclear [3]. Unlike previous research that focuses on general emotional display rules [4], this study is based on the model of Diefendorff & Richard (2003) and puts forward the connection between expression rules and emotion regulation strategies, which is based on the control theory of discrepancy monitoring and the reduction process of self-regulation. In other words, if they realize the inconsistency between the expression rules and the actual emotional expression, employees will take necessary actions individually through emotional adjustment. This study looks at emotional display rules in two dimensions (positive and negative display rules) based on Diefendorff et al. (2005) [5]. Employee emotional display rules that are congruent with firm expectations are a key factor affecting emotional regulation strategies [2]. Therefore, this study examines the impacts of positive and negative emotional display rules on emotional labor strategies.

Since Hochschild (1983) introduced the concept of emotional labor strategy, most studies have divided emotional labor strategy into two dimensions: surface acting and deep acting [6,7,8,9]. The expression of naturally felt emotions was proposed as a new emotional labor strategy by Ashforth & Humphrey (1993) and Diefendorff et al. (2005) [10,11]. Since then, a small number of studies have been conducted on these three dimensions of emotional labor strategies [12,13,14]. Therefore, based on the theory of emotion regulation strategies and conservation of resources theory (COR), this study expounds on the relationship between emotional labor strategies and both emotional exhaustion and job satisfaction. That is, staff members experience positive or negative emotional reactions through work activities, but appropriately adjust emotional expression through the emotional adjustment process to master the impact on emotional exhaustion (and job satisfaction). Moreover, through the resource conservation theory, staff members can experience pressure when they realize that their resources are threatened and lost. Therefore, although they want to avoid it, the continuous loss of resources will be emotionally exhausting, which will ultimately reduce job satisfaction. To gain deeper insight into workers’ emotional labor evaluation processes, this study divides emotional labor strategies into three dimensions (deep acting, expression of naturally felt emotions, and surface acting) to further examine the relationship between emotional display rules, emotional exhaustion, and job satisfaction.

Emotional exhaustion refers to a state of mental and physical exhaustion that manifests as emotional exhaustion and eventually negatively impacts job satisfaction [15]. Emotionally exhausted employees are apt to have negative feelings about their firm, their work, and even themselves. Therefore, this study takes job satisfaction/dissatisfaction as the outcome variable of emotional exhaustion.

Turnover intention refers to an employee’s voluntary intention to leave his/her job and/or firm [16]. According to the conservation of resources theory (COR), a loss of resources is the main cause of the negative consequences. Under the condition of increasing pressure like COVID-19, if resources are insufficient, emotional exhaustion will be realized, which will eventually lead to various negative physical and psychological consequences. Therefore, this study applies the conservation of resources theory, assuming that employees will experience job exhaustion and even emotional exhaustion, which will lead to resignation. High employee turnover comes at a great cost to firms, especially turnover among experienced, highly trained employees. Thus, understanding the antecedents of turnover intention is key to helping minimize the negative impact of turnover on firm performance.

Job burnout refers to the feeling that one’s emotional resources are exhausted due to insufficient energy and excessive psychological demands [17]. Researchers have systematically discussed the problem of job burnout [18,19,20,21]. Emotional exhaustion is a component of job burnout that may cause workers to experience tension and fatigue, and eventually lead them to consider quitting or changing jobs [22,23]. Previous studies support the connection between emotional exhaustion and turnover intention [24,25,26]. Therefore, this study investigates the role of emotional exhaustion on turnover intention.

Hackman and Oldman (1976) put forward job characteristics theory (JCT) using a large number of questionnaires [27]. JCT explains how to ensure people have a positive mental state, and then stimulate work motivation and improve job satisfaction. Job satisfaction/dissatisfaction, another antecedent of turnover intention among front-line employees, is another concern for high-contact service firms. Retaining satisfied front-line employees is crucial to firm success. Job satisfaction refers to employees’ positive feelings toward their work [28]. Based on previous studies on job satisfaction [29,30,31], we investigate the role of job satisfaction on turnover intention.

People everywhere are fearful and anxious about the long-term effects of COVID-19. Because the COVID-19 pandemic is a relatively recent phenomenon, there is limited research examining its impact on various aspects of organizational behavior, including the relationship between job satisfaction and turnover intention. While there have been studies on the direct effects of fear of COVID-19 on job satisfaction [32,33,34,35] and turnover intention [36,37,38,39], the literature does not sufficiently address the emotional responses to pandemics such as COVID-19, which could be significant moderators in this relationship. The pandemic has led to changes in work conditions, including changes in job security, and heightened health concerns. However, the existing literature does not fully capture how this impacts their job satisfaction and turnover intentions. Therefore, this study attempts to verify the moderating role of fear of COVID-19 in the relationship between job satisfaction and turnover intention.

Emotional display rules are important because of the job characteristics of front-line employees who have frequent face-to-face contact with customers. However, research on the impact of specific types of emotional display rules—i.e., positive and negative display rules—is limited. Most previous studies have considered only two dimensions of emotional labor strategies (deep acting and surface acting) and have not looked at the expression of naturally felt emotions as an independent variable. Moreover, few studies have focused on the moderating role of fear of COVID-19.

This study proposes a research model that integrates the two dimensions of emotional display rules (positive and negative display rules) with the three dimensions of emotional labor strategies (deep acting, surface acting, and expression of naturally felt emotions) and empirically investigates the relationship between emotional display rules, emotional labor strategies, emotional exhaustion, job satisfaction/dissatisfaction, and turnover intention of customer contact employees. It also examines whether fear of COVID-19 moderates the relationship between job satisfaction and turnover intention, thereby filling a research gap and providing strategies for reducing employee turnover intention and improving service firm profitability.

2. Theoretical Background and Hypotheses Development

This section mainly discusses the theoretical background of this research and the hypotheses development, which lays the foundation for this study.

2.1. Theoretical Background

The theoretical background section introduces job characteristics theory and job demands–resources theory. The research in this study is based on the above two theories.

2.1.1. Job Characteristics Theory

Hackman and Oldham’s job characteristics theory (JCT) is a widely recognized and influential theory in the field of organizational behavior. Developed by Richard Hackman and Greg Oldham in 1976, the theory provides a framework for understanding how certain job characteristics can influence employees’ motivation, satisfaction, and performance [27].

According to the job characteristics theory, there are five core job characteristics (skill variety, task identity, task significance, autonomy, and feedback) that contribute to the psychological state of an individual in a work setting. These job characteristics interact with three critical psychological states (experienced meaningfulness of work, experienced responsibility for outcomes, and knowledge of results) to influence employees’ outcomes. Together, job characteristics and psychological states influence employees’ motivation, job satisfaction, and other work-related outcomes. When jobs possess high levels of the core characteristics and facilitate positive psychological states, employees are more likely to experience greater satisfaction, motivation, and engagement [40].

Based on JCT, this study assumes that the nature of job characteristics can influence the expectations and requirements for emotional expression in customer interactions. Job characteristics shape the emotional demands placed on employees, which, in turn, affect their adherence to emotional display rules. In other words, the job characteristics of customer-oriented services interact with emotional display rules, shaping employees’ emotional labor strategies and influencing the emotional expressions they present to customers. Emotional exhaustion and job dissatisfaction can be considered psychological states influenced by emotional display rules and emotional labor strategies, which subsequently impact turnover intention. We investigate how emotional exhaustion and job dissatisfaction mediate the relationship between emotional display rules, emotional labor strategies, and turnover intention.

2.1.2. Job Demands–Resources Theory

The job demands–resources (JD-R) theory, initially proposed by Bakker and Demerouti in 2001, provides a framework for understanding the relationship between job characteristics, employee well-being, and work-related outcomes [21]. The theory posits that there are two broad categories of factors in the work environment: job demands and job resources. Job demands refer to the physical, psychological, social, and organizational aspects of a job that require sustained effort and can potentially lead to strain or burnout if resources are lacking. Examples of job demands include high workload, time pressure, emotional demands, and conflicting role expectations. Job resources are the physical, psychological, social, or organizational aspects of a job that are functional in achieving work goals, reducing job demands, and fostering growth, learning, and development. Job resources can include social support, supervisor support, autonomy, feedback, and opportunities for skill development [41].

According to JD-R theory, job demands can lead to negative outcomes such as emotional exhaustion, burnout, and turnover intention, while job resources can lead to positive outcomes such as work engagement, job satisfaction, and lower turnover intention. Job resources are theorized to buffer the impact of job demands on well-being, acting as protective factors that help individuals cope with and manage job stressors [42].

Based on the JD-R theory, emotional display rules represent the expectations and requirements for expressing specific emotions in the workplace, while emotional labor strategies refer to the techniques employees use to manage their emotions during customer interactions. These job demands can contribute to emotional exhaustion and job dissatisfaction, which, in turn, can increase turnover intention. Therefore, we propose that emotional display rules and emotional labor strategies are positively related to emotional exhaustion and job dissatisfaction, which, in turn, increase turnover intention.

2.2. Hypotheses Development

Based on job characteristics theory and job demands–resources theory, we constructed a structural equation model for this study and designed seven main hypotheses, which are as follows.

2.2.1. Emotional Display Rules and Emotional Labor Strategies

Based on Ekman and Friesen’s (1975) concept of social–cultural emotional norms and Hochschild’s (1983) theory of emotional labor, scholars have adopted the term “emotional display rules” to describe expressive expectations for employees inherent to their jobs [43,44,45]. Emotional display rules specify consistent norms for any given occupation or position and do not change dramatically [46,47]. For instance, front-line employees are required to “serve with a smile”; healthcare professionals are supposed to be compassionate toward their patients; and administrative employees are expected to stay calm when dealing with annoying consumers.

Emotional display rules prescribe which emotions are proper and how they should be expressed to other people [5,40]. Employees are always expected to demonstrate positive emotions and hide negative emotions [48]. But in most previous studies, positive and negative emotional display rules were lumped together [49,50,51]. To perform a deeper and more nuanced study, we divided emotional display rules into separate categories.

Positive display rules are demonstrations of positive emotions that convey caring, joy, kindness, and/or pleasure, while negative display rules compel the suppression of negative feelings, such as anger, irritability, or contempt [52]. Employees implement emotional labor strategies to both express positive emotions and suppress negative emotions [53,54].

Emotional labor is the effort needed to form and maintain a healthy emotional attitude according to stated emotional display rules [44]. Xu et al. (2020) posited that emotional display rules could be considered the main antecedents of emotional labor strategies [55], and many studies have examined the impacts of general display rules on emotional labor strategies [56,57]. The relationship between the dimension of emotional expression rules and the dimension of emotional labor strategies has also been extensively studied [52,53,54]. However, prior studies have taken into account only two dimensions of emotional labor strategies (deep and surface acting).

Ashforth & Humphrey (1993) proposed that the expression of naturally felt emotions be included as a third dimension of emotional labor strategies [10]. Expression of naturally felt emotions is a spontaneous behavior that occurs when an internal emotion is consistent with specified service expectations during job execution and little effort is needed to express the emotion. Diefendorff et al. (2005) determined that the expression of naturally felt emotions is different from deep acting or surface acting [11].Therefore, this study considers three dimensions of emotional labor strategies: deep acting, expression of naturally felt emotions, and surface acting [12,13,14]. In addition, this study includes the expression of naturally felt emotions as an independent variable in emotional labor research.

Very little extant research has investigated the relationship between the two emotional display rules and the three emotional labor strategies, and conclusions have been inconsistent. According to the facial feedback hypothesis, facial expression positively links to subjects’ emotional experiences. Negative display rules suppress workers’ actual negative emotions and demonstrate the publicly observable side of positive emotions. Workers’ true feelings are detached from these displays. Such emotional dissonance usually results in negative consequences, for example, surface acting [5]. Therefore, based on Diefendorff et al. (2005) and Kim (2008), the following hypotheses are proposed [11,52]:

H1. 

Emotional display rules are related to emotional labor strategies.

H1a. 

Positive display rules are positively related to deep acting.

H1b. 

Positive display rules are positively related to the expression of naturally felt emotions.

H1c. 

Negative display rules are negatively related to surface acting.

2.2.2. Emotional Labor Strategies and Emotional Exhaustion

Hochschild (1983) explained that emotional labor is compulsory emotional management in the work setting [44]. Emotions expressed by workers are important because they affect job performance. When engaged in emotional labor, workers must adopt appropriate strategies to adjust their feelings and displays. Based on prior studies [12,13,14], emotional labor strategies are divided into three dimensions: deep acting, expression of naturally felt emotions, and surface acting in this study.

Emotional exhaustion is a reaction related to a specific stressor, i.e., the feeling of exhaustion due to work [58,59]. Previous studies have examined the effects of two dimensions of emotional labor strategies on emotional exhaustion and found that surface acting is positively related to emotional exhaustion, while deep acting depletes valuable mental resources, thus contributing to emotional exhaustion [60,61]. However, little is known about the impact of the expression of naturally felt emotions on emotional exhaustion since few studies have empirically analyzed the relationship between the two. What is more, results from the limited research available are inconsistent [12,14,62,63]. To address the inconsistencies among these findings, it is worthwhile re-examining the effects of emotional labor strategies on emotional exhaustion. So, we hypothesize:

H2. 

Emotional labor strategies are related to emotional exhaustion.

H2a. 

Deep acting is negatively related to emotional exhaustion.

H2b. 

Expression of naturally felt emotions is negatively related to emotional exhaustion.

H2c. 

Surface acting is positively related to emotional exhaustion.

2.2.3. Emotional Labor Strategies and Job Satisfaction

As an economy shifts toward a service orientation, emotional labor takes on greater importance. In service industry jobs, where interpersonal interaction is integral to the work, the emotional labor required for interactions with customers has a big impact on outcome variables, such as emotional exhaustion, job satisfaction, and turnover intention. Although Grandey (2003) considered job satisfaction as the antecedent of emotional labor, most studies regard job satisfaction as an outcome variable of emotional labor, and we concur with this conceptualization [64].

Few scholars have investigated the influence of emotional labor strategies on job satisfaction with the expression of naturally felt emotions included as an independent variable, and the related studies’ conclusions have been inconsistent [12,14,65,66,67]. These conflicting findings indicate a need to re-examine the association between emotional labor strategies and job satisfaction. This study assumes that employees have a sense of personal accomplishment when expressing appropriate emotions, so deep acting and expression of naturally felt emotions are positively related to job satisfaction, while surface acting is negatively related to job satisfaction. In other words, this study takes emotional labor strategies as an antecedent of job satisfaction and offers the following hypotheses:

H3. 

Emotional labor strategies are related to job satisfaction.

H3a. 

Deep acting is positively related to job satisfaction.

H3b. 

Expression of naturally felt emotions is positively related to job satisfaction.

H3c. 

Surface acting is negatively related to job satisfaction.

2.2.4. Emotional Exhaustion and Job Satisfaction

Emotional exhaustion is the perception of being emotionally overextended and exhausted by one’s job [68]. It is strongly related to weakened emotional resources and enhanced fatigue. Workload and time pressure lead to work stress, especially in the service industry and particularly for customer contact employees [69].

Since Demerouti et al. (2001) proposed the job demands–resources (JD-R) model, it has been the main theory for how job demands and resources impact employees’ job stress [70]. As its name indicates, the model has two parts: job demands and job resources. Job demands refer to the additional energy or effort that employees exert to deal with stressful work, employment instability, conflicts, work overload, etc., while job resources are the resources needed to achieve job-related goals. Job demands are usually associated with negative effects, such as exhaustion, complaints, fatigue, etc., while job resources are often associated with positive effects, such as participation, happiness, motivation, etc. [41]. Emotional exhaustion is the result of stress based on the job demands–resources model and is a variable that leads to negative results, such as job dissatisfaction [71].

Extant studies have demonstrated that emotional exhaustion impacts job satisfaction. Allam et al.’s (2021) empirical analysis of employees working in the banking sector in Saudi Arabia showed that emotional exhaustion had a negative impact on job satisfaction [68]. Opoku et al. (2021) studied employees at a public hospital and confirmed the negative effects of emotional exhaustion on job satisfaction [72]. These results are consistent with Lee et al. (2019) and Prajogo (2019) [73,74]. This study also postulates the negative impact of emotional exhaustion on job satisfaction and puts forward the following hypothesis:

H4. 

Emotional exhaustion is negatively related to job satisfaction.

2.2.5. Emotional Exhaustion and Turnover Intention

Emotional exhaustion leads to irritability and fatigue and is considered a key factor affecting turnover intention, especially in the service industry. According to the conservation of resources theory, employees who are emotionally exhausted are unable to perform their duties normally, so they become selective and sensitive when using new resources [75]. Emotional exhaustion caused by the depletion of emotional resources is related to decreased job satisfaction. Studies have found that front-line employees with a high degree of emotional exhaustion show greater turnover intention. For example, Shah et al.’s (2022) research on nurses confirmed that emotional exhaustion had a significant positive impact on turnover intention [25]. Lee et al.’s (2021) research on teachers also found a positive relationship between emotional exhaustion and turnover intention [76]. These results are consistent with Lee et al. (2019), Namin et al. (2022) [73,77]. However, Natalin et al. (2021) found that the influence of emotional exhaustion on turnover intention was not significant [78]. Due to the inconclusive and varied results, it is necessary to re-examine the relationship between emotional burnout and turnover intention. Therefore, we posit the following hypothesis:

H5. 

Emotional exhaustion is positively related to turnover intention.

2.2.6. Job Satisfaction and Turnover Intention

As a core topic of human resource management research, job satisfaction is defined as the degree to which employees like or dislike their work [79]. Brief (1998) pointed out that job satisfaction is linked to employees’ attitudes toward their work, which is usually measured in emotional terms [80]. According to social exchange theory, when the social exchange is successful, employees experience positive emotions and exhibit low turnover intention [81].

In today’s extremely competitive environment, well-trained employees are vital to a business’s success. When skilled employees leave, firm performance suffers, so retaining skilled employees is a high priority for managers. However, although firms strive to reduce employee turnover intention [82], it is increasingly difficult to retain top employees [83]. Thoresen et al. (2003) define turnover intention as an employee’s intention to leave a job [16]. Notably, while there is an intention to leave the job, the actual decision has not yet been made. Therefore, understanding the antecedents of turnover intention can minimize the voluntary turnover of employees who have not yet quit [84]. Because the COVID-19 pandemic led to a prolonged economic recession, many employees temporarily postponed their resignations, although they were considering leaving their jobs.

Some extant research has revealed a significant negative relationship between job satisfaction and turnover intention. In other words, when employees are satisfied with their work, they are less likely to quit [85,86]. However, other studies have found the impact of job satisfaction on turnover intention to be insignificant [87,88,89]. Inconsistencies in findings raise concerns as to whether an association between job satisfaction and turnover intention truly exists. Therefore, we offer the following hypothesis:

H6. 

Job satisfaction is negatively related to turnover intention.

2.2.7. The Moderating Role of Fear of COVID-19

COVID-19 was first detected in humans in 2019. Once the scope of the pandemic became known in 2020, fear of the illness and death associated with it spread globally. Governments focused on developing effective vaccines to stem the outbreak of COVID-19 and restricted movement between countries to limit its spread, and individuals in infected countries also took various preventative measures, including social distancing, self-isolation, and hand washing. Many firms were severely impacted by these measures.

Yet, despite these prophylactic actions, fear of COVID-19 continued to increase [90], especially among front-line employees whose jobs involve face-to-face interactions with customers [91]. Ahorsu et al. (2022) developed a self-reporting one-dimensional scale (FCV-19S) to measure fear of COVID-19 [92]. Grounded in FCV-19S, we offer here a systematic measure scale suitable for China.

Fear of COVID-19 is a psychological state that reflects the perceived threat and anxiety of being infected by the novel coronavirus. It may affect the cognitive and emotional evaluation of one’s job and the intention to quit. Most of the existing studies on fear of COVID-19 focus on its direct effect on job satisfaction [32,33,34,35] and turnover intention [36,37,38,39]. There is a lack of studies that explore how fear of COVID-19 interacts with job satisfaction to influence the turnover intention of employees.

In the face of uncertainty and risk of death due to COVID-19, the level of fear and anxiety among front-line employees has increased [93], causing employees to engage in defensive actions to protect themselves [94]. When employees are fearful due to a pandemic, they may weigh job security more heavily in their satisfaction. This heightened concern can moderate how job satisfaction relates to turnover intention, as individuals might be less willing to leave a stable job despite low satisfaction due to the uncertainty [95]. In addition, employees fearful of COVID-19 may prioritize health and safety over job satisfaction. In such cases, even if an employee is satisfied with their job, the fear of contracting the virus may heighten turnover intentions if they believe the work environment poses a health risk. Conversely, an employee might tolerate lower job satisfaction if they perceive that their current job poses a lower health risk compared with alternatives [92]. What is more, fear of COVID-19 can increase psychological stress, which can affect decision-making processes. Elevated stress levels might cause employees to be more impulsive or more risk-averse, which can moderate how job satisfaction is linked to turnover intention [96]. Furthermore, the pandemic may lead employees to re-evaluate their work values and expectations. Aspects like workplace health policies may become more important. This shift in values can change how job satisfaction is perceived and how it correlates with turnover intention [97]. Thus, fear of COVID-19 can be viewed as a critical factor affecting job satisfaction and turnover intention among front-line employees. We hypothesize:

H7. 

Fear of COVID-19 moderates the correlation between job satisfaction and turnover intention.

Our concept model and theoretical hypotheses are shown in Figure 1. Positive display rules are assumed to affect two emotional labor strategies (deep acting and expression of naturally felt emotions), while negative display rules are assumed to affect one emotional labor strategy (surface acting). Deep acting and expression of naturally felt emotions are presumed to have negative impacts on emotional exhaustion, while surface acting is thought to have a positive impact on emotional exhaustion. Emotional exhaustion is expected to have a negative effect on job satisfaction and a positive effect on turnover intention. Job satisfaction is presumed to have a negative impact on turnover intention. Fear of COVID-19 would moderate the relationship between job satisfaction and turnover intention.

3. Methodology

This section will introduce the data collection, measures, and the source of the investigated question.

3.1. Data Collection

Following a pilot study, our research team conducted an in-person questionnaire survey with front-line service workers between 19 December 2022 and 27 December 2022. We collected data from service staff who face customers directly in the context of the COVID-19 pandemic. During the pilot study, 50 qualified service workers were interviewed in person. We discussed all survey questions in detail with them to ensure the clarity and validity of the questions. The statistical analysis results of the pilot study were generally consistent with expectations. Most respondents affirmed that the questionnaire had readability and understandableness. Based on data analysis and in-depth interview feedback, 52 questions were included in the final questionnaire, of which 46 related to positive display rules, negative display rules, deep acting, expression of naturally felt emotions, surface acting, emotional exhaustion, and job satisfaction. Five questions related to each of these constructs; four questions related to turnover intention; and seven questions related to fear of COVID-19. Respondents’ demographic information was processed as control variables, including gender, age, education, marriage, job tenure, and occupation.

Before conducting the survey, we pre-set a topic: “Have you been engaged in the service industry during the three years (2020–2022) of the COVID-19 pandemic? Yes; no” If the interviewee chose no, the questionnaire ended directly, and if they chose yes, the questionnaire would continue. This step excluded interviewees not involved in this study. Secondly, after collecting the questionnaires, we screened them. If all the answers to the questionnaires were the same option, the questionnaires were determined to be invalid and rejected, because such questionnaires were obviously answered to cope with the situation and had no research significance. Approximately 700 questionnaires were handed out, and 650 were collected. Of these, 27 questionnaires with incomplete or suspect answers were excluded. Ultimately, 623 questionnaires were valid for an 89 percent response rate.

3.2. Measures

To ensure the reliability and validity of data collection, most of the items in the questionnaire were adapted from the existing research and modified on the basis of the research objective. We used a 5-point Likert scale to measure the constructs, with “1” meaning complete disagreement and “5” meaning complete agreement.

3.2.1. Emotional Display Rules

Emotional display rules include positive and negative display rules. The scale for positive display rules was derived from Diefendorff et al. (2005) [11] and Rutner et al. (2015) [98] and was composed of five items: “Part of my work is to make clients feel good (posdis1)”; “I suppress my bad emotions or negative emotional reactions to clients (posdis2)”; “I make a gesture of sympathy to clients (posdis3)”; “I have a friendly attitude towards clients (posdis4)”; and “My firm hopes me to act excited enthusiastic in my contacts with clients (posdis5)”.

The scale for negative display rules was derived from Diefendorff et al. (2005) [11] and Rutner et al. (2015) [98] and was composed of five items: “I don’t express bad emotions to clients (negdis1)”; “I don’t express negative reactions to clients (negdis2)”; “My firm expects me not to show my sadness (negdis3)”; “I pretend not to be angry while at work (negdis4)”; and “I pretend I am not feeling despised while at work (negdis5)”.

3.2.2. Emotional Labor Strategies

Emotional labor strategies include deep acting, the expression of naturally felt emotions, and surface acting. The scale for surface acting was derived from Cossette and Hess (2015) [99], Walsh (2019) [100], and Yang et al. (2019) [14] and was composed of five items: “I put on a show to display the emotions I don’t really have (sur1)”; “I hide negative emotions when interacting with clients (sur2)”; “I try to smile even if it is reluctant (sur3)”; “I pretend to have emotions that I must show to clients (sur4)”; and “I put on an act to express a good mood during customer contact (sur5)”.

The scale for deep acting was derived from Cossette and Hess (2015) [99], Walsh (2019) [100], and Yang et al. (2019) [14] and was composed of five items: “I work hard to develop the feelings inside of me that I need for the work (dep1)”; “I make an effort to actually experience the emotions that I need to show to clients (dep2)”; “I try to make a good impression on clients (dep3)”; “When I help clients, I feel like I’m pretending to be happy (dep4)”; and “When I reflect on the working experience, it is a pleasant memory (dep5)”.

The expression of naturally felt emotions was derived from Cossette and Hess (2015) [99], Walsh (2019) [100], and Yang et al. (2019) [14] and was composed of five items: “The emotions I show clients are genuine (nfe1)”; “The emotions I express to clients come naturally (nfe2)”; “The emotions I express to clients coincide with what I spontaneously feel (nfe3)”; “I execute the task spontaneously for the firm (nfe4)”; and “I treat clients sincerely to show clients a good corporate image (nfe5)”.

3.2.3. Emotional Exhaustion

Emotional exhaustion was derived from Azharudeen et al. (2018) [101] and Sun et al. (2022) [26] and was composed of five items: “I feel emotionally drained from my job (exh1)”; “Interactions with clients give me great stress (exh2)”; “I get discouraged by my work (exh3)”; “Interactions with clients all day is really strain for me (exh4)”; and “I feel exhausted at the time of getting off work (exh5)”.

3.2.4. Job Satisfaction

Job satisfaction was derived from Cheng et al. (2022) [32] and was composed of five items: “I feel pretty satisfied with my current work (jobs1)”; “I like working in my firm (jobs2)”; “My work gives me a sense of achievement (jobs3)”; “I get a lot of enjoyment in my job (jobs4)”; and “I feel proud of my job (jobs5)”.

3.2.5. Turnover Intention

Turnover intention was derived from Sun et al. (2022) [26] and was composed of four items: “I intend to quit my present job (int1)”; “I’ll look for another job if I have a chance (int2)”; “If I could make another choice, I would not choose the current firm (int3)”; and “I would quit my job if I got a suitable offer in another firm (int4)”.

3.2.6. Fear of COVID-19

Fear of COVID-19 was derived from Cheng et al. (2022) [32] and was composed of seven items: “I feel afraid of the coronavirus (COVID-19) (covf1)”; “I feel uncomfortable to think about the coronavirus (COVID-19) (covf2)”; “I feel stressed in my contacts with clients since I worry I’ll get the coronavirus (COVID-19) (covf3)”; “I am anxious about losing my life due to the coronavirus (COVID-19) (covf4)”; “I’m afraid of causing colleagues to be infected with the coronavirus (COVID-19) (covf5)”; “I couldn’t sleep because I’m worried about catching the coronavirus (COVID-19) (covf6)”; and “Thinking about the coronavirus (COVID-19) frightens me (covf7)”.

4. Analysis Results and Hypothesis Testing

In this section, we discuss demographic statistics, normality tests, validity and reliability of measurements, common method bias testing, hypothesis testing, and moderating effect testing. The model in this study was tested, and the corresponding results were obtained.

4.1. Demographic Statistics

Of the 623 respondents, 53.5% (n = 333) were female and 46.5% (n = 290) were male. Overall, 29.7% (n = 185) were aged between 30 and 39, 48% (n = 299) had bachelor’s degrees, 34.3% (n = 214) had 15–20 years of work experience, 72.7% (n = 453) were married, and 24.4% (n = 152) were salespeople working in department stores or discount stores.

4.2. Normality Test

We used a skewness–Kurtosis method to examine the univariate normality of every variable. Skewness values range from −1.655 to 1.002 with an absolute value of less than 2 [102], and kurtosis values range from −0.931 to 2.652 with an absolute value of less than 7. These results support the normality of the univariate distribution [103,104].

4.3. Validity and Reliability of Measurements

To check reliability, Cronbach’s alpha was used. Fear of COVID-19 has the highest Cronbach’s alpha of 0.947, followed by job satisfaction at 0.934, deep acting at 0.929, turnover intention at 0.928, negative display rules at 0.896, the expression of naturally felt emotions at 0.894, surface acting at 0.888, positive display rules at 0.867, and emotional exhaustion with the lowest Cronbach’s alpha of 0.833. All items were accepted on the basis of Cronbach’s Alpha over 0.70, which means adequate reliability.

To evaluate validity, exploratory factor analysis was conducted with exogenous variables and endogenous variables. Based on the analysis, two items of expression of naturally felt emotions (nfe4, nfe5) and one item of fear of COVID-19 (cov4) were excluded. The variance, which can be explained by nine factors, was 74.2%.

Confirmatory factor analysis was used to examine unidimensionality. As exhibited in

Table 1. Results of confirmatory factor analysis.

Variables		Unstandardized Loading	S.E.	t-Value	Standardized Loading	AVE	CR
Positive display rules	posdis1	1.000	-	-	0.812	0.571	0.799
	posdis2	0.844	0.049	17.133	0.703
	posdis5	0.895	0.049	18.244	0.748
Negative display rules	negdis2	1.000	-	-	0.807	0.654	0.850
	negdis3	0.977	0.046	21.404	0.801
	negdis4	0.968	0.045	21.874	0.817
Deep acting	dep3	0.936	0.037	26.245	0.849	0.731	0.890
	dep4	1.000	-	-	0.835
	dep5	0.985	0.038	25.118	0.879
Expression of naturally felt emotions	nfe1	0.962	0.040	24.014	0.826	0.706	0.878
	nfe2	0.947	0.038	24.864	0.848
	nfe3	1.000	-	-	0.847
Surface acting	sur3	0.946	0.042	22.510	0.789	0.686	0.867
	sur4	0.979	0.040	24.421	0.839
	sur5	1.000	-	-	0.854
Emotional exhaustion	exh2	0.987	0.048	20.746	0.836	0.642	0.842
	exh4	1.000	-	-	0.856
	exh5	0.870	0.048	18.055	0.703
Job satisfaction	jobs1	0.993	0.036	27.905	0.870	0.747	0.899
	jobs4	1.000	-	-	0.867
	jobs5	0.968	0.035	27.268	0.857
Turnover intention	int2	1.000	-	-	0.884	0.762	0.906
	int3	0.994	0.035	28.414	0.868
	int4	0.983	0.034	28.313	0.866
Fear of COVID-19	covf1	0.902	0.038	23.894	0.815	0.689	0.899
	covf2	0.893	0.037	24.321	0.825
	covf3	1.000	-	-	0.841
	covf5	0.928	0.037	24.928	0.839

, positive display rules (posdis3, posdis4), negative display rules (negdis1, negdis5), deep acting (dep1, dep2), surface acting (sur1, sur2), emotional exhaustion (exh1, exh3), job satisfaction (jobs2, jobs3), and fear of COVID-19 (covf6, covf7) had two items deleted. In contrast, turnover intention (int5) had one item deleted. Considering the analysis of their adjustment, for the results obtained using the chi-square (χ²/df) ≤ 5; the Tucker–Lewis index (TLI) > 0.90; the goodness-of-fit index (GFI) > 0.90; the comparative fit index (CFI) > 0.90; the root mean square error of approximation (RMSEA) ≤ 0.08; and the root mean square residual (RMSR), a smaller value corresponds to a better adjustment [105]. The data analysis showed that the overall fit index has an acceptable level of fitness: χ² = 517.838 (p = 0.000), df = 314, χ²/df = 1.649, TLI = 0.979, GFI = 0.945, AGFI = 0.928, CFI = 0.982, RMR = 0.029, and RMSEA = 0.032.

Hair et al. (2010) asserted that a measurement model should be assessed grounded on convergent validity and discriminant validity. The assessment of convergent validity should be made by examining the composite reliabilities and the AVE for each construct [104]. As exhibited in Table 1, all the constructs’ composite reliability values were above the 0.7 threshold, and all the values of AVEs surpassed the 0.5 threshold. Thus, it is assured that the proposed model shows an acceptable level of convergent validity.

Discriminant validity is established if the off-diagonal values (the square of inter-construct correlation) are smaller than the diagonal values (AVE). Based on the analysis, the AVE demonstrated in

Table 2. Correlations and square roots of AVE.

Variables	1	2	3	4	5	6	7	8	9
Positive display rules (1)	0.756
Negative display rules (2)	0.745 **	0.808
Deep acting (3)	0.594 **	0.670 *	0.855
Expression of naturally felt emotions (4)	0.604 **	0.664 *	0.671 **	0.840
Surface acting (5)	−0.643 **	−0.753 **	−0.613 **	−0.642 **	0.828
Emotional exhaustion (6)	−0.164 **	−0.329 **	−0.334 **	−0.349 **	0.342 **	0.801
Job satisfaction (7)	0.563 **	0.646 **	0.736 **	0.701 **	−0.631 **	−0.401 **	0.864
Turnover intention (8)	−0.159 **	−0.177 **	−0.234 **	−0.198 **	0.214 **	0.158 **	−0.180 **	0.873
Fear of COVID-19 (9)	0.250 **	0.233 **	0.312 **	0.296 **	−0.291 **	−0.157 **	0.346 **	−0.666 **	0.830

* Values on the diagonal are the square root of the AVE; ** the off diagonals are correlations.

was larger than the squared correlation coefficients. As such, all constructs have sufficient discriminant validity.

As predicted in H1a, the positive display rules are positively correlated with deep acting (γ = 0.594, p < 0.01), the positive display rules in H1b are positively correlated with the expression of naturally felt emotions (γ = 0.604, p < 0.01), and the negative display rules in H1c are negatively correlated with surface acting (γ = −0.753, p < 0.01). As predicted by H2a, deep acting is negatively correlated with emotional exhaustion (γ = −0.334, p < 0.01), the expression of naturally felt emotions in H2b is negatively correlated with emotional exhaustion (γ = −0.349, p < 0.01), and surface acting in H2c is positively correlated with emotional exhaustion (γ = 0.342, p < 0.01). As predicted by H3a, deep acting is positively correlated with job satisfaction (γ = 0.736, p < 0.01), the expression of naturally felt emotions in H3b is positively correlated with job satisfaction (γ = 0.701, p < 0.01), and surface acting in H3c is negatively correlated with job satisfaction (γ = −0.631, p < 0.01). As predicted by H4, emotional exhaustion is negatively correlated with job satisfaction (γ = −0.401, p < 0.01). As predicted by H5, emotional exhaustion is positively correlated with turnover intention (γ = 0.158, p < 0.01). As predicted by H6, job satisfaction is negatively correlated with turnover intention (γ = −0.180, p < 0.01).

The Heterotrait–Monotrait ratio (HTMT) methodology was also used to examine discriminant validity. As seen in

Table 3. Discriminant validity (HTMT criteria).

Variables	1	2	3	4	5	6	7	8
Positive display rules (1)
Negative display rules (2)	0.751
Deep acting (3)	0.597	0.673
Expression of naturally felt emotions (4)	0.606	0.664	0.671
Surface acting (5)	0.646	0.753	0.615	0.640
Emotional exhaustion (6)	0.172	0.324	0.334	0.346	0.343
Job satisfaction (7)	0.557	0.644	0.703	0.703	0.631	0.404
Turnover intention (8)	0.157	0.179	0.236	0.196	0.204	0.169	0.180
Fear of COVID-19 (9)	0.256	0.237	0.319	0.298	0.298	0.167	0.349	0.663

, the HTMT values of each construct are below the 0.85 threshold, signifying that all constructs used in the proposed model show satisfactory discriminant validity [106].

The initial data set (full sample) was separated into two parts based on the median value. As displayed in

Table 4. Comparison of unconstrained and constrained models.

Model	χ2	df	GFI	CFI	RMR	RMSEA	Δχ2 (Δdf)	sig.
Unconstrained model	719.229	448	0.917	0.970	0.030	0.031	20.238 (Δ16)	0.210
Constrained model	739.467	464	0.914	0.970	0.033	0.031

, the cross-validation for model selection achieved good results. The fit indices exceeded the recommended threshold (CFI was 0.970, RMSEA was 0.031, GFI ranged from 0.914 to 0.917, and RMR ranged from 0.030 to 0.033). Furthermore, the constrained model was χ² = 739.467 (df = 464), while the unconstrained model was χ² = 719.229 (df = 448). The constrained model’s variation in the difference in the chi-square values with 16 degrees of freedom Δχ² (16) was 20.238 (p = 0.210). The insignificant result indicates similar explained variances, thereby establishing the validity of the model.

We also performed a multi-collinearity test. The problem that the model estimation is distorted or difficult to estimate accurately due to a high correlation between variables in the model was prevented. If the general tolerance (Tol) is less than 0.1 or the variance expansion factor (VIF) is greater than 10, collinearity exists. However, the tolerances in this study range from 0.568 to 1, all of which are greater than 0.1. The VIF ranges from 1 to 1.76, which is less than 10. This indicates that there is no multi-collinearity problem in this study [107].

4.4. Common Method Bias Testing

Harman’s one-factor test was performed to evaluate common method bias. To detect common method variance, we entered all measurement scales into a principal component analysis and interpreted the unrotated factor solution [108]. The factor analysis results produced nine distinct factors, the largest of which accounted for 27.7% variance explained by a single factor. This means that common method bias was not serious in this research (less than the 50% cut-off point).

In addition, confirmatory factor analysis was conducted for further verification. As exhibited in

Table 5. Results of common method bias test.

Model	χ2	df	Δχ2	Δdf	p
Single-factor model	838.144	322	320.306	8	0.000
Multiple-factor model	517.838	314

, fit indices of the multiple-factor model (chi-square = 517.838, df = 314, chi-square/df = 1.649, RMR = 0.029, GFI = 0.945, AGFI = 0.928, CFA = 0.981, RMSEA = 0.032) were significantly better than the single-factor model (chi-square = 838.144, df = 322, chi-square/df = 2.603, RMR = 0.109, GFI = 0.918, AGFI = 0.897, CFA = 0.955, RMSEA = 0.051), revealing that common method bias was not a major threat in this research.

In a third approach, the common latent factor (CLF) analysis, was carried out to assess the common variance among the constructs used. We added a first-order factor to all observed variables in the model and compared the standardized regression weights from this CFA model to the standardized regression weights of a model without the CLF. The results indicated that all values are similar because the difference is less than 0.2 [109]. Therefore, common method bias was not a big concern in our data set.

4.5. Hypothesis Testing

The hypothesis model was tested using the structural equation modeling method. According to

Table 6. Analysis results of hypothesis testing.

Hypothesis	Coefficient	t-Value	Support
H1a: PDR → DA	0.814	15.197 *	Yes
H1b: PDR → NFE	0.837	15.355 *	Yes
H1c: NDR → SA	−0.803	−17.807 *	Yes
H2a: DA → EE	−0.112	−2.246 **	Yes
H2b: NFE → EE	−0.133	−2.677 *	Yes
H2c: SA → EE	0.161	2.988 *	Yes
H3a: DA → JS	0.436	10.443 *	Yes
H3b: NFE → JS	0.254	6.355 *	Yes
H3c: SA → JS	−0.188	−4.374 *	Yes
H4: EE → JS	−0.106	−2.738 *	Yes
H5: EE → TI	0.149	1.982 **	Yes
H6: JS → TI	−0.220	−3.295 *	Yes

Fit statistics: χ² = 514.770 (p = 0.000), df = 239, χ²/df = 2.154, TLI = 0.966 GFI = 0.932, AGFI = 0.915, CFI = 0.971, RMSEA = 0.043. * p < 0.01, ** p < 0.05. PDRs, positive display rules; NDRs, negative display rules; DA, deep acting; NFE, expression of naturally felt emotions; SA, surface acting; EE, emotional exhaustion; JS, job satisfaction; TI, turnover intention.

, which shows the results of the hypotheses testing, the overall fit index reaches a satisfactory level of fitness: χ² = 514.770 (p = 0.000), df = 239, χ²/df = 2.154, TLI = 0.966, GFI = 0.932, AGFI = 0.915, CFI = 0.971, and RMSEA = 0.043.

The path diagrams and causal relationships are shown in Figure 2. Hypotheses H1a, H1b, and H1c investigated correlations between emotional display rules and emotional labor strategies. After accounting for the control variables, positive display rules demonstrated a significant positive impact on deep acting (β = 0.814, t = 15.197, p < 0.01) and the expression of naturally felt emotions (β = 0.837, t = 15.355, p < 0.01), while negative display rules demonstrated a significant negative impact on surface acting (β = −0.803, t = −17.807, p < 0.01). In other words, H1a, H1b, and H1c were statistically supported, as hypothesized.

Hypotheses H2a, H2b, and H2c investigated correlations between emotional labor strategies and emotional exhaustion. Deep acting (β = −0.112, t = −2.246, p < 0.05) and the expression of naturally felt emotions (β = −0.133, t = −2.677, p < 0.01) displayed a significant negative impact on emotional exhaustion, while surface acting (β = 0.161, t = 2.988, p < 0.01) displayed a significant positive impact on emotional exhaustion. Thus, H2a, H2b, and H2c were validated.

Hypotheses H3a, H3b, and H3c investigated correlations between emotional labor strategies and job satisfaction. Deep acting (β = 0.436, t = 10.443, p < 0.01) and the expression of naturally felt emotions (β = 0.254, t = 6.355, p < 0.01) positively affected job satisfaction, while surface acting negatively affected job satisfaction (β = −0.188, t = −4.374, p < 0.01). These results supported H3a, H3b, and H3c.

Hypotheses H4 and H5 investigated the effects of emotional exhaustion on job satisfaction and turnover intention. Emotional exhaustion had a significant negative relationship with job satisfaction (β = −0.106, t = −2.738, p < 0.01) and a significant positive relationship with turnover intention (β = 0.149, t = 1.982, p < 0.05), thus supporting H4 and H5 respectively.

In addition, the hypothesized correlation between job satisfaction and turnover intention was significant, thus supporting H6.

4.6. Moderating Effect Testing

In the second testing phase, a multiple-group analysis was performed to verify the moderating effect of fear of COVID-19 in the correlation between job satisfaction and turnover intention (H7). To perform the multiple-group analysis, two subsamples (high fear of COVID-19 and low fear of COVID-19) were created based on the median value. The high fear of COVID-19 group contained 300 respondents, while the low fear of COVID-19 group had 323 respondents.

Table 7. Comparison of path coefficients between the low and high fear of COVID-19 groups.

Hypothesis	Low	High	Support
	Coefficient	Coefficient
H7: Job Satisfaction → Turnover Intention	−0.031	−0.435 *	Yes

* p < 0.01.

and

Table 8. Comparison of unconstrained and constrained model of fear of COVID-19.

Model	χ2	df	Δχ2 (Δdf)	p
Unconstrained model	940.129	478	8.091 (1)	0.004
Constrained model	912.220	479

show an analysis of the moderating role of fear of COVID-19. According to Table 7, the path coefficient of the group with a low fear of COVID-19 is −0.031, and the path coefficient of the group with a high fear of COVID-19 is −0.435, which indicates that the influence of job dissatisfaction on turnover intention increases with increasing fear of COVID-19. In addition, it is believed that if the constrained model’s variation in the difference in the chi-square values is statistically significant to a higher degree than the chi-square criteria threshold, the hypothesis is validated [110]. According to Table 8, the value for the constrained model was 912.220 (df = 479), while the unconstrained model was χ² = 940.129 (df = 478). The constrained model’s variation in the difference in the chi-square values with one degree of freedom Δχ² (1) was 8.091 (p = 0.004). The statistically significant results supported H7.

Since the control effect of fear of COVID-19 was statistically significant, the control effect to confirm its shape is shown in Figure 3. As a result of examining the slope by dividing the control variable fear of COVID-19 into low and high groups to confirm a significant interaction pattern, it can be seen that the group with a high fear of COVID-19 has higher turnover intention even in job satisfaction situations than the group with a low fear of COVID-19.

5. Discussion

In this section, the findings of this study are explained, and the theoretical and practical implications are expounded. In addition, the limitations and future research of this study are also discussed.

5.1. Findings

This study aimed to verify how emotional labor can trigger turnover intention. For this purpose, we explored the relationship between emotional display rules, emotional labor strategies, emotional exhaustion, job satisfaction, and turnover intention and verified the moderating effect of fear of COVID-19. We surveyed service employees who had face-to-face contact with customers in banks, department stores, discount stores, hospitals, etc., during the COVID-19 pandemic. Our findings show that positive display rules positively affect deep acting and the expression of naturally felt emotions, while negative display rules negatively affect surface acting; this is consistent with the research results of Kim (2008) [52] and Cheung et al. (2018) [111]. Deep acting and the expression of naturally felt emotions are related to low emotional exhaustion and high job satisfaction, while surface acting is related to high emotional exhaustion and low job satisfaction; this finding is consistent with the research results of Jin et al. (2020) [7], Lee et al. (2019) [73], Muparangi et al. (2021) [61], and Noreen et al. (2021) [13]. However, our results are slightly different from those of Anafarta (2015), whose results show that the expression of naturally felt emotions is not related to job satisfaction. The inconsistency between our findings and those of Anafarta (2015) regarding the relationship between the expression of naturally felt emotions and job satisfaction might be because of the context in which the studies were conducted [65]. Anafarta’s study was published in 2015, and our study was conducted in the context of the COVID-19 pandemic. The pandemic significantly altered work environments and may have changed how individuals perceive the expression of emotions in the workplace and its relation to job satisfaction. Emotional exhaustion has a negative effect on job satisfaction and a positive effect on turnover intention, this is consistent with the conclusions of Allam et al. (2021) [68] and Amissah et al. (2022) [12]. Job satisfaction significantly weakens turnover intention, which is consistent with the research of Abd-ellatif et al. (2021) [36], Berber et al. (2022) [29], and Hu et al. (2022) [30]. Fear of COVID-19 moderates the relationship between job satisfaction and turnover intention, and the results show that the group with a high fear of COVID-19 has higher turnover intention, even in job satisfaction situations, than the group with a low fear of COVID-19.

5.2. Implications for Theory

This research makes three important contributions to human resource management literature. First, the body of empirical research on emotional display rules is insufficient. Most prior research focuses on the influence of general rules. This study deepens the understanding of emotional display rules by dividing them into positive and negative display rules and empirically investigating the correlation between emotional display rule dimensions and emotional labor strategy dimensions.

Second, the existing research on emotional labor strategies has considered only two dimensions (deep and surface acting). Very little extant research has empirically investigated the impact of emotional labor strategies by including the expression of naturally felt emotions as an independent variable [112,113]. This study provides greater insight into emotional labor strategies by empirically examining a research model including a third dimension of emotional labor strategies (expression of naturally felt emotions) and explores how dimensions of emotional labor strategies influence emotional exhaustion and job satisfaction.

Third, the COVID-19 pandemic has created an unprecedented context in the workplace, and an accurate scale to measure fear of COVID-19 has not been well-developed. In addition, studies about its impacts are limited. Prior research put more emphasis on the effect of fear of COVID-19 as an antecedent [114,115,116]. There is insufficient attention to its moderating role. This study deepens the existing literature by developing a systematic measurement scale for fear of COVID-19 suitable for China and verifying whether fear of COVID-19 is a moderator of the relationship between job satisfaction and turnover intention, which has been relatively under-emphasized in the literature. Although the relationship between job satisfaction and turnover is a well-researched area, the pandemic has added a new layer of complexity. By examining the moderating role of fear of COVID-19, this research contributes to the development of more comprehensive models that may interpret the decision of employees to leave or stay in their jobs during a pandemic situation. It can also help to understand how the psychological impact of COVID-19 affects the work attitudes and behaviors of employees, especially those who are exposed to high-risk environments or have high emotional demands in their jobs. Furthermore, it can provide insights into how global health crises can affect traditional organizational behavior models. Understanding this moderation helps in adapting existing theories to novel and challenging circumstances.

5.3. Implications for Practice

This study also has practical implications for employee training, support, and monitoring and can serve as a guide for service firms seeking to develop relationships with customers, improve the health and well-being of front-line employees, and increase corporate profitability and competitiveness.

First, emotional display rules were found to affect emotional labor strategies. In other words, emotional display rules, both positive and negative display rules, are important determinants of emotional labor strategies (deep acting, expression of naturally felt emotions, and surface acting). These results are consistent with the findings of previous studies [5,11,52]. Specifically, positive display rules are shown to have a positive impact on deep acting and the expression of naturally felt emotions. What is more, positive display rules were found to be more strongly related to the expression of naturally felt emotions than to deep acting. On the other hand, negative display rules were shown to have a negative impact on surface acting, so it is prudent to establish guidelines for suppressing negative emotional displays (such as coldness, anger, frustration, and depression). In addition, to encourage the expression of naturally felt emotions, service firms should encourage front-line employees to display positive emotions with appropriate education, training, and rewards.

Second, emotional labor strategies (deep acting, expression of naturally felt emotions, and surface acting) were seen to influence emotional exhaustion. Specifically, deep acting and expression of naturally felt emotions negatively impact emotional exhaustion, while surface acting positively impacts emotional exhaustion. These results are consistent with the findings of previous studies [12,14,62]. The relationship between emotional exhaustion and surface acting is stronger than that with deep acting and the expression of naturally felt emotions. In other words, the experience of emotional disharmony caused by surface acting may have a negative impact on front-line employees when servicing customers and thus impact service firm profitability. Therefore, to alleviate the emotional exhaustion of front-line employees, measures should be taken to reduce the negative effects of surface acting. In addition to specific programs to protect the emotional and psychological health of employees, it is advisable to establish systematic emotional management training and emotional control strategy support systems, provide communication opportunities for employees to discuss stress, and engage professional consultants to address workplace problems.

Third, emotional labor strategies (deep acting, expression of naturally felt emotions, and surface acting) were shown to influence job satisfaction such that deep acting and the expression of naturally felt emotions have positive impacts on job satisfaction, while surface acting has a negative impact on job satisfaction. These results are consistent with the findings of previous studies [12,14]. Among the three dimensions of emotional labor strategies, deep acting was found to have the greatest impact on job satisfaction. To improve front-line employees’ job satisfaction, service firms should encourage employees to have more genuine and heartfelt positive feelings about their work. Given the growing proportion of service industry businesses in the industrial structure of China’s economy, the importance of employee emotional management merits attention, and service firms should focus on strengthening the ability of employees to apply deep-acting strategies with training programs. In addition, because front-line employees continue to expend emotional and psychological resources in emotional labor, service firms should set up reasonable compensation, support, and welfare systems.

Fourth, emotional exhaustion was shown to have a negative effect on job satisfaction and a positive impact on turnover intention. According to the conservation of resources (COR) theory, emotionally exhausted employees cannot competently do their jobs. This situation reduces employees’ job satisfaction and further increases their turnover intention. Therefore, service firms should take proactive steps to reduce work pressure for customer-facing staff by ensuring that the working environment is fair and pleasant and fostering transparent communication and respectful interpersonal relationships.

Fifth, consistent with the findings of previous studies [30,31], we confirmed that job satisfaction negatively affects turnover intention. Anxiety brought on by the COVID-19 pandemic remains, and workers still face risks as a result. Therefore, to reduce latent employee turnover intent, service firms need to ensure that working environments are conducive to stimulating and maintaining employees’ job satisfaction. Ensuring the safety, health, and well-being of employees is always important, especially given the existence of COVID-19.

Finally, fear of COVID-19 was shown to moderate the effects of job satisfaction on turnover intention. Front-line employees are highly aware of and highly susceptible to the risk of COVID-19, and those with a high fear of COVID-19 were more affected by the psychological pressure associated with COVID-19 than those with a low fear of COVID-19. To reduce turnover intention caused by fear of COVID-19, service firms should design effective interventions and policies to reduce the fear of COVID-19 among employees and enhance their job satisfaction and retention. It is essential for service firms to pay attention to the psychological impact of COVID-19 on their employees, especially those who work in high-risk sectors.

Specifically, service firms should establish and communicate clear safety measures. This can include providing personal protective equipment, enforcing social distancing, and implementing regular sanitization procedures to alleviate fears and retain satisfied employees. Fear of COVID-19 can be psychologically draining for employees. Managers should consider offering mental health support, such as counseling services or employee assistance programs, and maintaining open communication about organizational policies and updates regarding COVID-19. Also, providing employees with training on how to handle stressful situations and uncertainties brought about by the pandemic can be beneficial. It can empower them to better manage their fears and make more informed decisions regarding their employment. What is more, employees’ priorities may shift during a pandemic. Service firms need to reassess and, if necessary, restructure their benefit packages to address the current needs and concerns of their employees, which may now lean more toward health and safety. Furthermore, managers need to be trained to understand and empathize with the fears of their subordinates during such times. Leadership during crises requires different skills, and management strategies may need to be adapted to focus more on human-centric approaches. In addition, how an organization handles the fear and safety of its employees during a pandemic can have lasting effects on its reputation and employer branding. Taking proactive steps to address fears and concerns can position an organization as a caring and responsible employer. In addition, understanding that external factors like a pandemic can influence employee turnover intentions, service firms should incorporate this understanding into their long-term strategic planning. This may include diversification of skills within the workforce, succession planning, and creating adaptive organizational structures.

5.4. Limitations and Future Research

Although this research yielded interesting results, there are limitations to this study. Firstly, this study was not aimed at any specific sector of the service industry. Therefore, future research might focus on a specific service sector. Secondly, the questionnaire survey was conducted during COVID-19. Given the abnormal circumstances of the pandemic, the results of a similar study conducted before or after the pandemic might differ. Therefore, in the future, researchers might verify these results under different circumstances. Thirdly, this study used fear of COVID-19 as a moderator. However, fear of COVID-19 is not only a moderator but also a mediator, so studying the mediating effect of fear of COVID-19 will be worthwhile in the future. In addition, future research could consider other moderators, such as emotion, resilience, organizational support, and commitment. In particular, the pressure caused by COVID-19 is also an important concept affecting performance, so its role should be considered.

6. Conclusions

In the context of the COVID-19 epidemic, this study explored the relationship among emotional expression rules, emotional labor strategies, emotional exhaustion, job satisfaction, and turnover intention and verified the moderating role of fear of COVID-19 between job satisfaction and turnover intention. Different from most previous studies, this study divided the rules of emotional expression into positive and negative dimensions and subdivided the development of emotional labor strategies into three dimensions: deep, surface, and expression of naturally felt emotions. It also explored the fear of COVID-19 as a regulatory variable.

The results show that positive emotional expression rules have a positive impact on deep acting and the expression of naturally felt emotions and are more closely related to the expression of naturally felt emotions. Negative display rules negatively affect surface acting. Deep acting and the expression of naturally felt emotions are related to low emotional exhaustion and high job satisfaction, while surface acting is related to high emotional exhaustion and low job satisfaction. Emotional exhaustion has a negative effect on job satisfaction and a positive effect on turnover intention. Job satisfaction significantly weakens turnover intention. Fear of COVID-19 moderates the relationship between job satisfaction and turnover intention. In particular, the group with a high fear of COVID-19 has higher turnover intention even in job satisfaction situations than the group with a low fear of COVID-19.

However, because this study has not conducted extensive questionnaire surveys in all parts of China and the sample size is relatively small, there may be some problems in generalization. Therefore, future research needs to expand the sample area and sample size.