The Influence of Emotional Labor and Emotional Intelligence on Job Performance: Does Ownership Type Matter? A Comparison of Public and Private Organizations in China

ABSTRACT Given the limited knowledge about the effect of contextual factors of organizational ownership types on emotional labor, this research addresses two main questions: (1) whether emotional labor varies among organizations with different ownership types; and (2) whether emotional labor and emotional intelligence relate to job performance in different ways in public and private organizations. This paper examines the research questions with 306 self-report questionnaires from the public sector, domestic privately-owned enterprises and foreign-invested firms in China. Significant differences were found in the emotional labor reported in public and private organizations. Overall, emotional labor was found to have a significant effect on in-role performance, and emotional intelligence moderated the link between emotional labor and job performance in public organizations, but not in private organizations.


Introduction
Emotional labor involves the management of feelings to create publicly observable expressions that are in accordance with the display rules established by organizations (Hochschild, 1983). The display rules that shape the expression of emotional labor are associated with occupational norms and organizational contexts (Ashforth & Humphrey, 1993;Grandey & Melloy, 2017). For example, Kelly and Barsade (2001) found that group-level emotion norms may constrain or amplify group members' emotions, while Diefendorff, Erickson, Grandey, and Dahling (2011) reveal the existence of shared emotional display norms in the same work unit, but different levels of display rules in distinct work units. To date, the primary focus of most studies is the examination of emotional labor with the key contextual factors being on the occupation group or work unit (Kelly & Barsade, 2001;Kim, Bhave, & Glomb, 2013). In contrast, there are few studies that have provided an in-depth analysis of the influence on the expression of emotional labor and display rules of larger contextual factors such as by organization type (e.g., public vs. private organization). Nor have many studies moved beyond Western samples of employees.
This study considers the ownership type of organizations in China to understand whether the expression of emotional labor differs in organizations by different ownership types. In addition, this study investigates how emotional labor and emotional intelligence may impact upon job performance in both public and private sectors. The study contributes to the understanding of demands of emotional labour and its consequence among organizations with different ownership types in transitional economies like China as the organizational structures, types and culture evolve more market-driven forms as part of the transition (Ma, Silva, Callan & Trigo, 2016;Tsui, Wang, & Xin, 2006). Moreover, the responsibility for China's successful economic transition rests predominantly with the well-educated managers and professionals. However, little is known about their emotional labor and how their levels of emotional labor impact upon job performance. This study targets this specific group of managers and professionals, turning the focus to a group that has not been studied in detail compared to the numerous investigations on the emotional labor of frontline service workers (Humphrey, 2012).
In summary, the present study contributes to narrowing the gap in the knowledge about the effect of contextual factors on emotional labor through investigating organizations of different ownership types in China. Specifically, this study examines differences on emotional labor among managers and professionals working in Chinese public organizations (CPOs), foreign-invested enterprises (FIEs) and Chinese privately-owned enterprises (POEs), as well as differences in the relationships between emotional labor, emotional intelligence and job performance. In doing so, this paper seeks to extend the emotional labor literature into the domain of ownership type. Furthermore, this study applies the existing theory on emotional labor developed and researched in Western contexts, but focuses on a sample of Chinese managers and professionals from a wide variety of occupations rather than on frontline service workers. In achieving this aim this paper provides further evidence that managers and professionals as "people at work" (Brotheridge & Grandey, 2002) also engage in emotional labor.

Conceptualizing emotional labor and the research gaps
Three conceptualizations of emotional labor have influenced research to date in the field (see Ashforth & Humphrey, 1993;Grandey, 2000;Hochschild, 1983;Morris & Feldman, 1996), and this study notes some research gaps in these conceptualizations.
First, the organizational or the service context are identified as among the major antecedents of emotional labor. Morgeson, Dierdorff, and Hmurovic (2010, p. 357) propose that "organizational context can act as a cross-level moderator, with positive contexts enhancing the relationships between work characteristics and outcomes and negative contexts intensifying the adverse effect of poor work characteristics". In the model of emotional labor proposed by Grandey (2000) who recently revised it (Grandey & Melloy, 2017), organizational factors that are theorized to contribute to emotional labor include job autonomy, supervisor and colleague support. Prior research has largely focused on the occupational context at the work unit level (Diefendorff et al., 2011;Kim et al., 2013), but few empirical studies have examined the association between emotional labor and the role of ownership type as another feature of the organizational context.
Second, earlier research on emotional labor has focused on customer service characterized by "service with a smile", while studies have been centered on interactions between employees and external customers, again particularly in the service sector. On the other hand, internal organizational contexts have received much less attention, particularly the role of organizational insiders, such as supervisors and co-workers (Grandey, Kern, & Frone, 2007). As several researchers point out (Ashforth & Humphrey, 1993;Grandey, 2000), emotional labor is relevant to the external customer, as well as to internal customers such as co-workers (Ozcelik, 2013). Hence, this paper argues that the demands of emotional labor in the workplace need to consider more fully the overall levels of emotional labor performed with both external and internal customers rather than only with one or the other of these two types of customers.
Third, while target groups in previous studies have typically been frontline service employees in the private sectors, the relevance of emotional labor to other organizational members and sectors has been largely downplayed (Ashforth & Humphrey, 1993). There is a need to extend the understanding of the construct beyond service roles to other organizational roles (Morris & Feldman, 1996), including leadership and management roles that highly influence the everyday performance of any organization (e.g. Gardner, Fischer, & Hunt, 2009). Recent studies have applied the concept of emotional labor to broader interpersonal contexts beyond customer service (Grandey & Gabriel, 2015), moving away from the frontline service provider to more sophisticated relationships (e.g., intimacy, social status) with co-workers (e. Deng, Walter, Lam, & Zhao, 2017), and leaders (Gardner et al., 2009). Tschan, Rochat, and Zapf (2005, p. 210) contend that "emotion work requirements are not restricted to clients but exist for interactions with co-workers as well". Linked to this is growing attention to managers and leaders and their emotional work in the public sector (Guy & Lee, 2015;Potipiroon, Srisuthisa-ard, & Faerman, 2018;Sloan, 2014), with emotional labor an under-researched component of work in the public sector (Rayner & Espinoza, 2016).

Emotional labor, ownership types and China
Different work environments impact upon the level of emotional demands prescribed by work situations (e.g. Ozcelik, 2013). Two theoretical approaches offer useful lens for understanding how environmental factors influence organizational requirements for emotional expression (Lawler & Thye, 1999;Lively & Powell, 2006): the cultural-normative approach and the structural-relational approach. The cultural-normative approach highlights the role of emotional norms and display rules in a social context, while a structural-relational approach emphasizes the impact of social positions (e.g., power and status) on the emotions that people are likely to feel. The cultural-normative perspective implies that, while emotional norms shape emotional expressions or displays, they vary across social or organizational contexts. Morris and Feldman (1996) propose that, at the organization level, explicitness of display rules, routineness of task, form of interaction and job autonomy are associated with emotional labor behavior. Additionally, considerable evidence exists supporting the relationship between features of organizational cultures and the requirements for emotional labor (Gardner et al., 2009;Martin, Knopoff, & Beckman, 1998).
The structural-relational approach proposes that positional differences lead to divergent felt emotions, which in turn have important effects on exchange relations and networks (Lawler & Thye, 1999). Specifically, low status employees are more likely to experience negative emotions and are more apt to conceal them from group members, thus engaging in emotional labor. For instance, Bono, Foldes, Vinson, and Muros (2007) found that employees experience fewer positive emotions when interacting with their supervisors as compared with coworkers and customers. Martin et al. (1998) further assert that emotional labor is often subject to traditional bureaucratic forms of control, for example, when it is carefully monitored by supervisors. More recently, Grandey and Melloy (2017) argue that status (e.g. power) and relationships (e.g. intimacy) are likely to change the person-and event level emotional labor relationships.
This literature about the influences of structures, cultures and norms upon emotional labor is predominantly based on emotional labor research conducted in the private sector in the West. Looking at countries like China, the transition is from predominantly state-owned enterprises (SOEs) (see Yang, 2007) to a multiple ownership system, including public organizations represented by SOEs and other public organizations funded by government and a growing private sector represented by foreigninvested enterprises (FIEs) and domestic privately-owned enterprises (POEs) (Ma, Silva, Callan, & Trigo, 2016;Tsui et al., 2006). Similar to what have been found in the West (e.g. Andersen & Pedersen, 2013;Boyne, 2002), Chinese public organizations (CPOs) are constrained by their historical heritage, hierarchical structures and systems, and some governmental control (Yang, 2007), while private enterprises have greater flexibility and autonomy to run their business (e.g. Wang, Bruning, & Peng, 2007).
While organization's HR practices can influence emotional labor (Gabriel, Cheshin, Moran, & van Kleef, 2016), studies in China reveal differences in HRM practices of different ownership forms (Ma et al., 2016). Foreignowned enterprises use more formalized and transparent bonus systems (Jaussaud & Liu, 2011) through the transfer of high-performance human resource systems of their home country. Employees are motivated through transparent, open and fair remuneration practices and approaches that promote employee development, as well as a warm and friendly climate (Jaussaud & Liu, 2011).
Compared with FIEs' HRM practices, the systems of Chinese owned firms are much less transparent, with supervisors having stronger levels of personal influence (Jaussaud & Liu, 2011;Yang, 2007). For example, the HRM systems of Chinese privately-owned enterprises are described as pragmatic, nonsystematic and informal (Shen, 2010). Sims and Sun (2012) highlight the problems of overly close supervision of employees in these Chinese enterprises, most of which is attributed to family-based businesses that have a lack of formal rules and regulations.
The different clusters of firms that characterize the three ownership types may result in not only different HRM approaches, but also may lead to different organizational cultures (Tsui et al., 2006). Ralston, Terpstra-Tong, Terpstra, Wang, and Egri (2006) argue that Chinese SOEs (a dominant part of CPOs), POEs and FIEs are each characterized by a different and unique culture: SOEs highlight stress on control, POEs are oriented towards flexibility, and FIEs seek a balance between flexibility and control. Similarly, Tsui et al. (2006) propose that both foreign and domestic private firms have highly integrative cultures that emphasize both internal integration and external adaptation, while public organizations have a hierarchy culture. Thus, differences in HRM approaches and organizational cultures in these three types of organizations may influence emotional labor in different ways.
People working in the public sector in China enjoy a high status and less emotional labor when working with their external customers due to the high-power distance culture in China (Li, 2005). For example, Hu (2017) argues that Chinese civil servants are accountable only to their superiors rather to the citizens. High status people are thought to enjoy more freedom in their emotional expression and hence may put less effort into their emotional management (Lively, 2000). In addition, performance and economic goals are less stressed in the public sector (Bourantas & Papalexandris, 1992) where employees have less pressure to engage in emotional labor. However, as organizational attributes such as a hierarchical organizational structure and bureaucratic controls are often associated with higher emotional labor (Martin et al., 1998), employees working in Chinese public organizations might experience higher emotional labor when interacting with insiders than their counterparts in the private sector. Overall, it is expected that employees working in CPOs will report lower levels of emotional labor than those working in the private sector, since the relatively lower emotional labor received with external customers due to their high status may offset the high emotional labor from managing their emotions with insiders.
Given that employees of both POEs and FIEs are working almost with identical outsiders in a market-driven economy, their emotional labor could be expected with external customers to be at similar levels. However, as Chinese privately-owned enterprises have a less formal and systematic structure, closer supervision, more implicit rules, and focus more on economic goals than foreign firms do (Wang et al., 2007), employees working in domestic private enterprises might devote more effort to the management of their emotional experiences with coworkers and supervisors than employees in FIEs. On the other hand, the open, fair, warm and friendly climate with a good balance of control and commitment management approach (Ma et al., 2016), and formalized and transparent system (Jaussaud & Liu, 2011) of FIEs, may imply that they have more explicit display rules and organizational support for their employees. Therefore, employees working in FIEs are expected to report lower emotional labor than those working in POEs. The organizational characteristics and the effect on emotional labor are summarized in Table 1.
There are differences in organizational culture and structure, job autonomy, explicitness of displays, power of role receiver and closeness of supervision existing in CPOs, POEs and FIEs. Under the lens of culturalnormative approach and structural-relational approach, employees' emotional labor is expected to differ in these three types of organizations and thus the study proposes the following hypotheses.
H1a: The perceived level of emotional labor of employees working in CPOs will be lower than that of employees working in the private sector, including private-owned and foreign-invested enterprises.
H1b: The perceived level of emotional labor of employees working in POEs will be higher than that of employees working in FIEs.

Emotional labor and job performance
Although researchers (e.g. Grandey, 2000) contend that employees' displays of positive emotions result in favorable organizational outcomes including increased positive relationship between employee and customer, and higher employee performance, prior research reveals a mixed relationship between emotional labor and performance (e.g. Grandey & Gabriel, 2015). One reason for mixed findings may be the influence of mediating and moderating variables in the relationship between emotional labor and work performance (Goodwin, Groth, Frenkel, 2011). Overall, the literature tends to support the idea that emotional labor is positively related to work performance (Goodwin et al., 2011). Meta-analyses indicate that deep acting (modifying feelings) is positively associated with employee performance (Hulsheger & Schewe, 2011 However, surface acting (modifying expressions) is not correlated with performance (Kammeyer-Mueller et al., 2013;Mesmer-Magnus et al., 2012) or has a weak negative correlation with overall performance (Hulsheger & Schewe, 2011). In a particular cultural context like China, Allen, Diefendorff, and Ma (2014) contend that Chinese workers use both deep acting and surface acting more often than U.S. workers. More importantly, Chinese workers employ more deep acting than surface acting. The reasons are that in the Chinese collectivistic culture, there are core values around harmony and cooperation within the group (Noon & Lewis, 1992). Chinese people tend to regulate their emotional expressions to maintain harmony and cooperation among groups (Mesquita & Delvaux, 2013). Because deep acting is expected to have a positive association upon performance, more deep acting than surface acting employed by Chinese workers may mean that their emotional labor may be more positively related to performance in more collectivistic countries such as China.
Thus, this study predicts that: H2: Emotional labor will be positively related to job performance.

Emotional labor, emotional intelligence and job performance
Although researchers often extol the virtues of deep acting over surface acting in enhancing performance outcomes, empirical work has demonstrated that this relationship is subject to boundary conditions (Gabriel, Daniels, Diefendorff, & Greguras, 2015;Prati, Liu, Perrewé, & Ferris, 2009). Individual difference variables are found to affect emotional labor (Ashforth & Humphrey, 1993), while emotional intelligence is one of those factors that may explain how emotional labor impacts on performance (Kim, Yoo, Lee, & Kim, 2012). Emotional intelligence involves the ability to perceive, understand, appraise and express emotion, coupled with the ability to generate and regulate feelings (Mayer & Salovey, 1997). Emotional intelligence has an important influence on the relationship between emotional labor and job performance. Many scholars believe that those high in emotional intelligence are better able to discern the emotional demands prescribed by the situation, so they choose better emotional labor strategies than their counterparts with low emotional intelligence (Prati et al., 2009). For example, Brotheridge (2006) found that the availability of emotional intelligence may help employees to detect more accurately interpersonal emotional cues. Furthermore, emotional intelligence is likely to moderate the relationship between emotional labor and job performance because of the social influence of emotions within the emotional labor process (Grandey, Fisk, Mattila, Jansen, & Sideman, 2005). Researchers argue that emotionally intelligent employees are more likely to engage in deep acting rather than surface acting (Mesmer-Magnus et al., 2012). Deep acting is generally perceived as authentic for more deeply felt emotions, invoking positive reactions in the audience, and building more favorable interpersonal relationships at work which are central to job performance (Groth, Hennig-Thurau, & Walsh, 2009).
A moderating role of emotional intelligence is theoretically supported by the conservation of resources (COR) theory (Hobfoll, 1989), which postulates that personal resources (e.g., emotional intelligence) may mitigate efforts to display the appropriate emotions (Newton, Teo, Pick, Ho, & Thomas, 2016). Wong and Law (2002) and Joseph and Newman (2010) contend that high emotionally intelligent employees perform better than low emotionally intelligent employees at jobs that demand emotional labor.
Based on the review of the earlier literature, Chineseowned companies tend to have HRM systems that are low in transparency, and the situational cues are more evasive. These factors make the performance of emotional labor activities more dependent on idiosyncratic competencies and, consequently more dependent on personal resources (Liu & Dong, 2012). In both POEs and CPOs, emotional intelligence may play a critical role in influencing the relationship between emotional labor and job performance. For example, in the bureaucratic public sector, subordinates are evaluated more by the superior's personal judgement and their personal and political fortunes are determined by how their superiors judge their performance (Hu, 2017). Therefore, it is natural to expect those working in Chinese public organizations to leverage emotional intelligence to "please" supervisors for favorable job performance. Conversely, in FIEs employees experience a more formalized and transparent system, and a warm and friendly atmosphere, and enjoy a higher level of autonomy than their counterparts in domestic employers (Jaussaud & Liu, 2011). In other words, the organizational context of FIEs with more explicit and objective performance criteria, and less controls from supervisors, may mitigate people's tendency to rely on their emotional intelligence to leverage on emotional labor to achieve job performance. Hence, although employees with high emotional intelligence may have a stronger emotional labor-performance relationship than those with lower emotional intelligence, the differences between the two groups are expected to be less in FIEs. Thus, this study proposes that: H3a: In CPOs and POEs, emotional intelligence will moderate the effect of employees' emotional labor on their job performance. The relationship between emotional labor and job performance will be stronger among employees with higher levels of emotional intelligence than among those with lower levels of emotional intelligence.
H3b: In FIEs, there will be no difference in the job performance of individuals with high or low levels of emotional intelligence when emotional labor is high or low.

Sample and procedure
To understand the emotional labor of non-frontline service providers, this study targeted a specific group of managers and professionals working for public organizations, domestic private enterprises and foreign firms in China. Respondents indicated whether their organizations were public, foreign-invested or private owned. Five hundred questionnaires were administered to MBA graduates, students and their colleagues. A total of 306 questionnaires were returned with valid responses, representing a response rate of 61%. Respondents were on average 31 years old, 59% were male and the majority of them (67%) held managerial positions. The sampling of ownership type consisted of public sector with 107 (35%), and private sector with 199 (65%), including Chinese POEs 20% and FIEs 45%.

Measures
Emotional intelligence was measured using a short version of Wong, Law, and Wong (2004) scale which includes 20 scenarios. We used this scale partly because emotional intelligence is conceptualized as a set of abilities pertaining to emotions (Mayer, Salovey, & Caruso, 2000), and partly because the scale was developed for Chinese respondents. Respondents were asked to choose one option out of two that best reflects their most likely reaction in each scenario. For example, the respondents were asked to choose one of the options in the following scenario: when you are very down, you will: (a) try to do something to make yourself feel better; (b) just ignore it because you know your emotion will be back to normal naturally. The scale has acceptable reliability and validity (Peng, Wong, & Che, 2010;Wong, Wong, & Law, 2007).
Emotional labor was measured using the 5-item scale adapted from Wong and Law (2002). The scale reflects the emotional interactions with internal and external constituents. A sample item is "To perform my job well, it is necessary for me to spend most of my work time interacting with people (e.g., customers, colleagues, and other workers in this organization". Ratings were made on a 7-point scale, ranging from 1 "strongly disagree" to 7 "strongly agree." Job performance was measured using Farh and Cheng (1997) four-item scale. It consists of items such as "I make a significant contribution to the overall performance of our work unit" and "I can always complete my task on time". Ratings were made on a 7-point scale, ranging from 1 being "strongly disagree" to 7 being "strongly agree." This study controlled for the effects of demographic variables on performance by measuring participants' age (in years), gender, academic area, job position and job tenure (in years). Table 2 displays the descriptive statistics, internal reliability coefficients and intercorrelations among the variables used in this study. Table 3 reports the ANOVA analysis of emotional labor differences across ownership types. The analysis revealed the presence of statistically significant differences between the three groups of organizations. Univariate results indicated that CPOs employees were relatively less prone to emotional labor than their private sector counterparts. Though respondents in POEs reported slightly higher levels of emotional labor than those in FIEs, the difference  was not significant. Therefore, these results supported hypothesis 1a, but not hypothesis 1b. The independent variables were mean centered prior to entering the regression analyses, and interaction terms were calculated using mean-centered variables (Aiken & West, 1991). This study hypothesized that emotional labor would influence job performance (hypothesis 2). The overall sample results showed that, after controlling for the demographic variables, emotional labor made a significant contribution in explaining the performance of an employee. Specifically, emotional labor explained an additional 5% of the variance in job performance (F(6, 253) = 3.917, p < 0.01), thus demonstrating support for this hypothesis.

Hypothesis testing
This study also hypothesized that, in CPOs and POEs but not in FIEs, emotional labor would relate more positively to job performance among employees high in emotional intelligence. The results supported a main effect for emotional labor (Model 2, Table 4) and a significant interaction between emotional labor and emotional intelligence in predicting job performance (Model 3, Table 4) in CPOs. Following Aiken and West (1991), this study plotted the interaction (Figure 1) and found that it was consistent with Hypothesis 3a. The simple slope representing the association between emotional labor and job performance was positive and significant at one standard deviation above the mean of emotional intelligence (β = .31, t = 34.02, p < .001), and negative and significant at one standard deviation below the mean (β = −.03, t = −3.29, p < .05). However, in POEs and FIEs, the interaction term between emotional labor Note: N FIEs = 114, N POEs = 47, N CPOs = 91; a Gender was coded as 1, male and 2, female; b Academic area was coded 1, art; 2, science; and 3, engineering; c Job position was coded 1, employee, 2, supervisor; 3, manager; and 4, director or above; + p < .1 *p ≤ .05, **p < .01. Figure 1. Emotional labor and emotional intelligence interaction effects on job performance. and emotional intelligence on performance was not significant. Therefore, hypothesis 3a was partially supported and hypothesis 3b was supported.

Discussion
The current study examined three hypotheses regarding emotional labor, its relationship with emotional intelligence and job performance. First, this study compared the emotional demands of the respondents in organizations with different ownership types. Second, this study tested the impact of emotional labor on job performance. Third, this study examined the moderating effect of emotional intelligence in the relationship between emotional labor and job performance. Results are discussed below.

Different emotional labor across organizations of different ownership type
This study reported a significant difference in emotional labor demands in the case of public and private organizations. Specifically, professionals working in POEs and FIEs reported significantly higher levels of emotional labor than those working in CPOs. Two explanations are offered. First, most private organizations are market-driven and pursue the philosophy that the "customer is king". Employees working in private sectors experience more emotional labor when interacting with external customers (see Bolton, 2002). Moreover, as past studies (e.g. Bourantas & Papalexandris, 1992) reveal, private sectors stress the achievement of performance and economic goals more than the public sector, which could be an important impetus for emotional labor. On the other hand, public organizations are less performance-driven and employees may have less pressure to engage in emotional labor for inrole performance. The data from this study do reveal that job performance in the public sector was significantly lower than in the private sector (F = 4.16, p < .05).
In addition, employees working in the public sector in China are more likely to enjoy more power and status, and thus face fewer pressures to "please" their external customers. As others report (Brotheridge & Grandey, 2002), occupational types contribute to the reported levels of emotional labor. This finding suggests ownership type as a potentially new antecedent of emotional labor.

Association between emotional labor and job performance
The overall results show that emotional labor is an important predictor of job performance. Individuals who are skilled at managing and displaying emotions are likely to have better job performance. After controlling for the demographic characteristics, job performance was predicted by emotional labor (R 2 = 0.09, p ≤ 0.001) in the overall sample. However, when examining the samples of private and public sectors separately, emotional labor was a valid predictor of job performance only in CPOs, and not in FIEs or POEs.
Again, as discussed above, the different results for the association of emotional labor with job performance further illustrate that organizational environments in the public and private sectors in China shape employee emotional labor behaviors in different ways. First, though employees in both FIEs and POEs report higher levels of emotional labor than their counterparts in CPOs (Table 3), their job performance is not affected by their emotional labor (Table 4). This is interesting, because it suggests that the high levels of emotional labor employees perform in FIEs and POEs do not influence their job performance. Organizational HRM practices, culture and norms that are associated with ownership type might be the key factors that impact upon the role of emotional labor upon job performance.
For example, Martin et al. (1998) contend that emotional labor is subject to bureaucratic forms of control such as the supervision from superiors. This is the case in the CPOs in this study, which are characterized by a more hierarchical structure, bureaucratic culture and significant status differences (i.e. power distance). Therefore, in public organizations, on the one hand "getting-the-job-done" on a daily basis depends very much on "guanxi" (relationships) with colleagues and particularly with supervisors, as job autonomy is much more constrained in the public sector (Wang et al., 2007). On the other hand, performance appraisal in Chinese public organizations is subject to supervisor's personal judgement and favoritism as objective performance criteria are absent or vague (Hu, 2017;Liu & Dong, 2012;Yang, 2007). Therefore, developing and maintaining good "guanxi" with the supervisor and colleagues are critical in achieving positive performance evaluations for job performance, but such high quality relationships involve emotional exchange (Graen & Uhl-Bien, 1995).
Researchers suggest that employees actively invest resources to develop relationships that might foster future social supports to facilitate more positive job and performance outcomes (Halbesleben & Bowler, 2007). Given the hierarchical difference in social status and the vague display rules in such Chinese organizations, employees in CPOs need to engage in emotional labor to develop personal relationships with colleagues, and particularly with their supervisors. Moreover, the traditional influence of collectivistic cultural values around harmony and cooperation within group (Noon & Lewis, 1992) is more prominent in CPOs than in FIEs and POEs (Wang et al., 2007). Employees in CPOs thus may use more deep acting than surface acting to maintain harmony in their working relationship with colleagues and their supervisors (see Allen et al., 2014), while the greater use of deep acting leads to a positive influence on job performance.
On the contrary, as discussed earlier, in private enterprises, job performance is more merit-based with explicit and objective criteria and less influenced by the personal influences of supervisor. Therefore, despite employees in the private organizations performing more emotional labor with external customers, the overall influence of emotional labor on job performance is constrained.
Moreover, the finding that emotional labor has no relationship with job performance in the private sector is similar to previous studies in the West (Hulsheger & Schewe, 2011;Mesmer-Magnus et al., 2012). This finding may imply that employees working in FIEs and POEs may experience demands around emotional labor in their organizations identical to their counterparts in Western countries, suggesting that HRM practices in the POEs are converging toward those practices of FIEs in the West (see Ma et al., 2016).

Moderating effect of emotional intelligence on the relationship between emotional labor and job performance
Though the overall sample shows a moderating effect of emotional intelligence on the relationship between emotional labor and job performance, the effect does not exist in private organizations and only in public organizations (Table 4). In Chinese public organizations, emotional labor is more strongly related to job performance in the case of high emotionally intelligent employees rather than in low emotionally intelligent employees (Figure 1). For the different results in public and private organizations, the interpretation lies in the differences of emotional norms, display rules and social positions in the two types of organizations. The explicitness of rules (e.g. objective performance indicators), and the highly integrative cultures in private organizations, encourage their employees to rely less on performing emotional labor and leveraging emotional intelligence to achieve job performance. However, the greater ambiguity of display rules, more hierarchical cultures, higher levels of position power and significant personal influence on job performance from supervisors in public organizations (Hu, 2017), encourage employees to engage in displays of emotional labor with insiders by leveraging their emotional intelligence to gain desirable job performance.

Implication for emotional labor theory
The current study contributes to the literature in the following ways. First, ownership type appears to matter in what concerns employees' emotional labor and in the relationships between emotional labor and job performance. In particular, respondents from public and private organizations reported significant differences in their emotional labor demands (Table 3), revealing ownership type as a new antecedent of emotional labor. This study extends the set of contextual factors such as supervisor support, job autonomy, explicitness of display rules (Grandey, 2000;Morris & Feldman, 1996) to include ownership type or organization type as more macro factors that influence the relationship between emotional labor and job related outcomes.
Second, this study is the first to compare emotional labor and its consequences between public organizations and private ones. Grandey and Gabriel (2015) call for more study on "social groups", but most of the existing literature is restricted to gender, age, race and nation. The current research found that significant differences exist in the emotional labor across public and private organizations, and in the relationships between emotional labor, emotional intelligence and job performance. This finding identifies an interesting potential for future research on organizational types as a "social group".
Third, the study also extends emotional labor research beyond the typical frontline service workers to cover a composite sample of Chinese managers and professionals. Therefore, it also provides evidence for the argument that managers perform emotional labor in their jobs (Brotheridge & Grandey, 2002;Humphrey, Pollack, & Hawver, 2008). Such findings enrich the under researched area of emotional labor in the case of managerial workers, bringing evidence from Chinese organizational settings.

Implications for practice
The current study has several noteworthy implications for employers and employees. First, given the different HRM approaches employed by organizations with different ownership types (see Ma et al., 2016), and the different levels of emotional labor reported in this study, organizations need to be aware of the associations between emotional labor and the organizational context created by their HRM and other practices. For example, more explicit display rules and autonomy-supportive policies may make employees engage in emotion regulation in a more positive way (Chi & Grandey, 2016).
In this regard, public organizations are encouraged to benchmark the management approach followed by private organizations by having more transparent HRM practices and through creating more positive work climates in their workforces (Grandey, Rupp, & Brice, 2015). This is in line with the New Public Management (NPM) rhetoric and transference of management practices from the private to the public sector (Hodgkinson, Hughes, Radnor, & Glennon, 2018). In addition, while many studies on emotional labor have highlighted the importance of person-job fit (e.g. Humphrey, Ashforth, & Diefendorff, 2015), this study suggests the importance of person-organization fit. For example, given the significant roles of emotional labor and emotional intelligence in public organizations, these organizations may need to consider assessing the emotional intelligence of candidates in the recruitment of employees. Moreover, given the significant levels of emotional labor reported by the sample of managers and professionals, employers should consider and weigh up the costs of emotional labor on employees. In addition to provision of emotional management training, the employment of more humanistic practices to replace emotional display rules might be one strategy .

Limitations and future research
This study uses a cross-sectional design and the current study is potentially subject to common method bias. The direction of causality cannot be tested, and more longitudinal studies are needed. Second, the different sample sizes of each ownership type present a limitation. In particular, the number of respondents in domestic private enterprises is much lower than for the other groups.
In terms of future research, as ownership types are increasingly more diverse, researchers might explore how different emotional regulation strategies (e.g. deep acting, surface acting) are used in organizations with different types of ownership type, and how employees engage in emotional labor with both clients and organizational members (insiders) in different organizational types. Future research may also investigate the association between HRM practices and emotional labor. For example, there is a need to better understand how control and commitment HRM approaches (e.g. Ma et al., 2016) may influence levels of emotional labor and job-related outcomes. As Grandey and Gabriel (2015, p. 341) conclude there is a need for a better understanding of "how practices (e.g. recruitment, training, performance management) function as a system to create environments conducive to more beneficial forms of emotion regulation".