Korean Emotional Laborers' Job Stressors and Relievers: Focus on Work Conditions and Emotional Labor Properties.

BACKGROUND: The present study aims to investigate job stressors and stress relievers for Korean emotional laborers, specifically focusing on the effects of work conditions and emotional labor properties. Emotional laborers are asked to hide or distort their real emotions in their interaction with clients. They are exposed to high levels of stress in the emotional labor process, which leads to serious mental health risks including burnout, depression, and even suicide impulse. Exploring job stressors and relieving factors would be the first step in seeking alternatives to protect emotional laborers from those mental health risks.
METHODS: Using the third wave data of Korean Working Conditions Survey, logistic regression analysis was conducted for two purposes: to examine the relations of emotional labor and stress, and to find out job stressors and relievers for emotional laborers.
RESULTS: The chances of stress arousal are 3.5 times higher for emotional laborers; emotional laborers experience double risk-burden for stress arousal. In addition to general job stressors, emotional laborers need to bear burdens related to emotional labor properties. The effect of social support at the workplace is not significant for stress relief, unlike common assumptions, whereas subjective satisfaction (wage satisfaction and work-life balance) is proven to have relieving effects on emotional laborers' job stress.
CONCLUSION: From the results, the importance of a balanced understanding of emotional labor for establishing effective policies for emotional laborer protection is stressed.

Introduction

“Emotional labor (EL)” has become an important concept in discussing work conditions in Korean society as the service industry expands and workers involved in EL increase accordingly. Emotional laborers deal with customers as a “face” of their organizations, and they make efforts to regulate their emotions following “feeling rules”—which defines the type and intensity of emotions the emotional laborers express.

As EL is reported to become more prevalent in Korean society, issues surrounding emotional laborers including their working conditions, human rights, and mental and physical health-related issues have drawn much social attention. Job stress of emotional laborers is especially worthy to study for two reasons: (1) on an organizational level, job stress reduces productivity, causing burnout and job dissatisfaction of workers; and (2) on an individual level, and more importantly regarding health, chronic job stress may lead to critical hazards for workers' mental health including depression, anxiety disorder, and suicidal impulse. An emotional laborer's process of emotion regulation can cause long-term negative effects [1]. A high rate of depression among emotional laborer groups [2], recent media reports on suicides of sales clerks, hardships of call center workers, and court judgments on compensation for mental diseases related to EL have proven the negative effects.

Diverse social discourses have been formed in Korean society to address these related issues from an individual business level to a legislation dimension: e.g., revision of the Occupational Safety and Health Acts and Act on Equal Employment and Support for Work-Family Reconciliation, also known as “emotional laborer protection bill”, is proposed in the National Assembly. However, these efforts to socially regulate the negative effects of EL are still focused on certain industries with visible problems. This approach has a risk of confining EL-related job stress as “someone else's problem”, and thus not understanding it as a new burden at the workplace related to societal work environment change.

In line with the question of the possibility of a more general understanding of EL and its effect on job stress, this paper explores job stressors and relievers of emotional laborers in contrast to general workers using third wave Korean Working Conditions Survey (KWCS) data. This approach questions general views to interpret EL only within certain industries, and attempts to link understandings on EL-related job stress to general job stress discourse. More specifically, analysis of the study encompasses: (1) comparison of stress arousal levels (presented through odds ratio) between emotional laborer groups and general worker groups; (2) examination of the “double risk-burden” effect of general work conditions and EL properties to emotional laborers' job stress; and (3) tests on the stress relieving effect of social support and subjective satisfaction of emotional laborers.

Materials and methods

Data

The KWCS has been carried out by the Occupational Safety and Health Research Institute since 2006, stemming from the European Working Conditions Survey, and has been continued with survey item correction: the second wave data was collected in 2010, and the third wave data in 2011. The KWCS asks questions of workers >15 years of age nationwide. The data has two strengths for EL studies: (1) it contains questions on both the general work environment and EL properties; and (2) the data is standardized with the European Working Conditions Survey so that researchers can broaden the analysis scope to Korea–EU country comparative studies for socio-cultural understanding. The sample size of the third wave KWCS data was 50,032: the number of male respondents was 29,138 (58.2%) and female respondents numbered 20,894 (41.8%).

Research design

The study set up four hypotheses:

H1. Emotional laborers will experience more stress than general workers.

H2. Emotional laborers will feel double stress burden due to general job stressors and EL properties.

H3. Social support will relieve stress of emotional laborers.

H4. Emotional laborers with high satisfaction will experience less stress.

H1 compares the stress arousal odds ratio of emotional laborers with general workers for examining the premise of the study. H2 examines “double risk burden” in emotional laborers' job stress. Generally, EL studies on job stress follow two directions: part of the studies focuses on distinctive properties of EL, and the others analyze the mediating effect of general job stressors such as work autonomy and employment types on emotional laborers' stress levels. This paper aims to show that the two groups of stressors (general work environment and EL property) both have effects on emotional laborer's stress logic. H3 and H4 test the effects of well-known stress relievers for emotional laborers. Social support can be categorized into an individual level (support by peer and/or friend) and organizational level (labor representative organization at work including labor union). In this paper, the effects of three social support variables (perceived peer support, perceived friend support, and existence of labor representative organization at work) on individual emotional laborers' job stress are analyzed H4 examines whether emotional laborers would feel less stress if they think their emotional burden is ‘rewarded’ adequately at work, and/or if their family and social life functions adequately as “back stage.” The analytic frame for this study around the four hypotheses is shown in Fig. 1.

Fig. 1

Analytic frame for the study.

The study opted for logistic regression with stress arousal as a dependent variable. Firstly, logistic regression was conducted for the general sample population (general worker group) to test H1. In this first step, EL is considered as an independent variable along with other stress-related work environment variables, namely employment status, industry sector, scale of workplace, tenure, weekly work hours, work autonomy, social supports, wage satisfaction, and work-life balance. Individual socio-demographic variables such as gender, age (measured by generation), educational level, and subjective health satisfaction are controlled. Secondly, the sample was divided into two subsets, emotional laborers and nonemotional laborers following the study's operational definition, and logistic regression analysis was carried out for each subset. This was to examine the effect of job stressors and relievers for emotional laborers in a comparative and exploratory way. Independent variables were the same in the second analysis, except the EL variable was used for dividing the subsets. SPSS 18.0 (SPSS Inc. Released 2009. PASW Statistics for Windows, Version 18.0. Chicago) was used for the analysis.

Operational definitions

Emotional laborers

The term “emotional labor” was coined by a US sociologist, Arlie Russell Hochschild, in The Managed Heart [3]. EL refers to processes of “managing emotional expressions and gestures in public” to compromise customers' emotions and/or to convey values requested by the company. Theoretically, direct interaction with clients and subjective emotional burden in the interactive labor process are considered two major components in defining EL. According to the Korea Occupational Safety and Health Agency's definition in Guideline for Emotional Labor Related Job Stress Prevention [4], EL is defined as “labor which includes emotional display and expressions under company's request regardless of the worker's own emotional state.” However, former empirical studies often tended to choose only the “contact with clients” aspect in detecting EL. This narrower variable operation seems to be related to the limit of precedent data: the KWCS is the first survey which contains questionnaire items on emotional aspects among nationwide survey data in Korea. Frequency of client contact is closely related to the intensity of EL and, thus, partially adequate for exposure risk to EL, however, demands on ‘emotional management’ in the client contact process and related emotional burden should be considered in the operational definition for better fitting to the theoretical definition of EL. Therefore, in this study, for distinguishing emotional laborers, client contact (portion of work hours for direct contact with people outside the workplace such as clients, passengers, students, patients, etc.) and emotional demand (pressure of hiding emotions in the work process) were operated as discrete variables, and those who both had contact with clients and felt emotional demand in the labor process, were defined as emotional laborers. This can be considered as a way of distinguishing emotional laborers within risk groups of EL.

Because the scope of this study focuses more on experience of EL itself than on its intensity, level of client contacts and emotional demand, originally measured in ordinal scales, were compressed to dichotomy. In KWCS data, the level of client contact is measured on a 7-score scale, from ‘never exposed’ to ‘always.’ Answers were coded from ‘1/4 of time’ to ‘always’ as 1; ‘never’ and ‘rarely’ as 0. Regarding emotional demand, the KWCS asks whether the respondent “should hide emotions during work” using a 5-score scale. Answers of ‘always,’ ‘frequently,’ and ‘sometimes’ are coded as 1, and ‘rarely’ and ‘never’ as 0.

Table 1 is a cross-table of client contact and emotional demand. Emotional laborers according to the operational definition number 19,472 of 50,032 persons (estimated 37%), corroborate the general estimation of emotional laborers as 30–40% of total workers [5]. Socio-demographic information of emotional laborers defined in such a way is presented in Table 2.

Table 1

Cross-table for sorting out emotional laborers

		Emotional demand: hiding emotion during work		Total(unit: persons)
Yes	No
Client contact	Yes	19,472	9,073	28,545
No	12,102	9,385	21,487
Total		31,574	18,458	50,032

Table 2

Socio-demographic information of emotional laborers

		Frequency(unit: persons)	%
Gender	Male	10,138	52.1
Female	9,334	47.9
Education level	No Education	123	0.6
Primary	646	3.3
Preprimary	1,407	7.2
Secondary	8,368	43.0
College	3,704	19.0
University	4,772	24.5
Tertiary	452	2.3
Generation (y)	Teenagers (15–19)	136	0.7
Young (20–39)	7,108	36.5
Middle-aged (40–59)	10,416	53.5
Old (60+)	1,812	9.3
Employment status	Unpaid workers	1,127	5.8
Irregular workers	2,072	10.6
Regular workers	8,640	44.4
Self-employed	7,149	36.7
Employers	484	2.5
Industry sector	Nonservice	2,566	13.2
Production service	2,572	13.2
Distribution service	6,973	35.8
Personal service	4,559	23.4
Public service	2,802	14.4

As seen in Table 2, more than one half of emotional laborers are men, in contrast to the common stereotype that EL is a trait of female workers. Educational distribution of EL is similar to the general social distribution shape, with a high frequency of high school and tertiary graduates.

In a generational perspective, the middle-aged (aged 40–60 years) population occupies more than one half of emotional laborers, followed by the youth (20–39 years). Within each generation group, around 43% engage in EL. The fact that >40% of the active working age group (the youth and the middle-aged) are categorized as emotional laborers may imply that EL is becoming more pervasive in the Korean work environment, however, the finding requires further elaborations. Among the elderly, 1,812 of 8,102 persons (22%) are categorized into emotional laborers. This not-so-little portion of emotional laborers in the aged work population can be interpreted along with the trend that the elderly get jobs in care work or service work which does not require much physical labor, working at a grocery as a cashier, at a gas station or in subway parcel services, after their retirement.

Many emotional laborers are working as regular workers or as the self-employed, whereas irregular workers make up 10.6% of total emotional laborers. The result does not match general discourses on EL, which centers around irregular-contract workers in outsource call centers or big discount stores, and legislation efforts pursuing employment status transition for emotional laborer protection. This mismatch can be explained further by actual business systems in Korea. The high proportion of self-employed can be ascribed to the fact that many small-scale self-employed businesses in Korea are restaurants or small retailers.

Distribution of emotional laborers by service sectors also assures the expansion of EL in Korean society. Based on Castells' [6] argument on the service industry, the study divided service sectors into nonservice (manufacturing works); production service (finance and insurances, real estate and leasing services, and other production-related services); distribution service related to transportation of goods, services, knowledge, or humans (wholesale and retailers, transportation and communication service businesses); personal services for individuals (accommodation businesses, restaurants, leisure and culture, health, household domestic services, etc.); and public services in nonmarket sectors (public administration, education, public health, and social welfare).

At a closer glance, emotional laborers are distributed in every industry sector. Interpreted with caution, the result may show emotional demands are expanded beyond traditional EL industries. For example, the proportion of public service workers marks 14% of total emotional laborers, which implies increasing government officials' EL burden due to privatization and subsequent emphasis on service minded attitude [7], [8].

Job stress for emotional laborers

Job stress is defined as harmful physical and mental reactions occurring when on-the-job demands do not match workers' capability, resources, or own desires. Job stress variables covered in related research are diverse but tend to be standardized around the Korean Occupational Stress Scale (KOSS), developed by the Korea Occupational Safety and Health Agency in 2006 and categorized into eight dimensions: physical environment, job demands, autonomy, relational conflict, job insecurity, organizational system, reward inadequacy, and workplace culture. Among those dimensions, EL seems to be closely associated with job demands. Karasek's [9] job strain model (1979) argues that high job demands and low self-control cause job strain and consequently high levels of stress. Karasek [9] did not elaborate on EL directly in his argument, but his frame can be applied to EL in that the extra demand of putting emotional effort in the client interaction process is added, and that emotional laborers feel less self-control because of feeling rules imposed by the company and client's demands.

Many former researches, as discussed earlier, studied specific service industries, or more narrowly, a specific business type or workplace and focused more on the effect of EL-specified properties to job stress such as emotional dissonance, burnout, fatigue due to interaction with clients, and psychological trauma [10], [11]. The author tried to link Korean emotional laborer's job stress in line with the general job stress argument in another study with coauthors using the first wave KWCS data [12], but faced opposite bias, omitting EL properties due to data limitation at that time. On considering their experience, this study includes both general work environment variables and EL property variables as job stressors for emotional laborers.

Dependent variables of the study are odds ratios of stress arousal. The KWCS originally asked stress levels using a 5-score scale, however, the study converted the variable as a discrete to control individual differences in stress awareness. Other individual level socio-demographic variables (sex, age as generation, educational level, and subjective overall health condition) are also controlled in the analysis.

Independent variables related to general work conditions include: industry type, employment status, scale of workplace, tenure (measured by year), weekly work hours (measure by hour), average monthly wage (log), job insecurity, and job autonomy (on rest; worktime scheduling; and work method). Among EL properties, emotional involvement, work setting, and extra physical burden are analyzed. Variables that continuously failed to obtain statistical significance such as workplace scale and tenure, are omitted in the explanation.

Employment status is divided into regular workers/irregular workers for the employed workers, and the self-employed (hiring 4 employees or less) and employers (hiring >5 employees). Workplace scale is categorized as small (with 1–4 persons working), small-medium (with 5–49 persons working), medium (with 50–249 persons working), and large (with >250 persons working). Work tenure is an adequate variable to see whether EL contains a trait of skilled labor. Tenure was surveyed using both year and month as units, however, the study rounded the measure by year unit. Weekly work hours are collected and analyzed using an hourly unit. Average monthly wage is a reward for the labor: log was taken to control its skewed distribution.

Perceived job insecurity and job autonomy are powerful variables in general job stress. Regarding job insecurity, the KWCS asks whether the respondent agrees with the statement that “I might lose this job in next 6 months” with a yes-no answer. Because job insecurity is expected to have a strong correlation with employment status, the analysis included the interaction term of irregular worker (dummy variable) and perceived job insecurity as well.

Autonomy can affect a worker's feeling of self-control. Dimensions of autonomy are specified as autonomies on rest, work schedule, and work method. Both rest autonomy and work schedule autonomy are measured using a 5-score scale, whereas work method autonomy is measured using a discrete measure in the data: thus, three variables were modified as discrete. Work method autonomy can be more important in understanding emotional laborers' job stress because lack of autonomy in work method implies the company's regulation plays a strong role in emotional laborers' labor process, which includes emotional display and expressions [13].

Emotional involvement is measured using a 5-score scale. In the case of EL, more emotional involvement can be interpreted as deep acting, workers' efforts to change perceptions and/or emotions in the interactive situations. The other direction, surface acting, means efforts only to regulate outward expressions and gestures. Both strategies have negative effects on emotional laborers' mental health, because excessive effort in deep acting can cause burnout and surface acting can lead to emotional dissonance [14].

The work setting can be divided into worker-familiar locations (workers' home or workplace) and client-familiar locations (visiting client's home or workplace). Because EL is closely associated with situational context in which emotional display occurs, familiarity of work setting would influence emotional laborers' job stress. Physical burden refers to how much they should endure certain physical positions such as standing up and repeated performing of certain gestures.

Interest in emotional laborers' stress may expand to interest in coping strategies. Recent legislating efforts suggest such ideas as secured time and space for rest, enhancement of the Employee Assistance Program for stress management, work environment advancement, and standardized protection manual establishment. Among various factors considered as stress relievers, the study analyzes effects of social support and subjective satisfaction. Peer support, existence of good friend at work, and existence of a worker-representative organization at work are measured as social supports. Together with social supports, subjective satisfaction level for wages and work-life balance are opted for analysis.

Results

Emotional labor and job stress

To see whether EL functions as a significant stressor for Korean workers in general, a logistic regression was conducted using the whole sample. Given other job stress variables, the possibility of stress arousal for emotional laborers is 3.5 times higher than workers whose work does not contain EL. The result verifies that workers doing EL feel much more stress than those who do not need to perform EL (H1 accepted).

Table 3 is a logistic regression result for the overall worker sample. Stress odds ratios, or simply relative probability of stress arousal differs according to sex, educational level, and subjective health satisfaction level with statistical significance, whereas generational difference is not statistically significant. In general, women feel more stress than men, and people with high education feel more stress. The effect of subjective health satisfaction was relatively strong among control variables, in negative correlation.

Table 3

Logistic regression on general workers' job stressors

Variables	B	S.E.	Exp(B)
Demographic (control)
Gender (ref: male)	−0.083*	0.032	0.920
Education	0.047†	0.015	1.048
Health satisfaction	−0.271‡	0.023	0.763
Work conditions
Employment status (ref: regular work)
Irregular work	0.143†	0.042	1.154
Irregular work*, job instability	−0.298†	0.087	0.742
Industry sector (ref: nonservice)
Production service	0.224‡	0.038	1.251
Weekly work h	0.011‡	0.001	1.011
Autonomy
Rest	0.180‡	0.014	1.197
Work method	0.062*	0.030	1.064
Social support
Union in workplace	0.288‡	0.043	1.334
Peer support	0.060‡	0.016	1.062
Friends at work	0.191‡	0.034	1.210
Reward
ln(wage)	0.561‡	0.036	1.753
ln(wage)* satisfaction	−0.101‡	0.007	0.904
Work-life balance	−0.0169‡	0.025	0.845
Emotional labor	1.261‡	0.034	3.530
Constant	−2.147‡	0.186	0.117
N	50,032
−2 log likelihood	29,364.808‡

*p < 0.05.

†p < 0.01.

‡p < 0.001.

B, Beta value; Exp(B), exponential value of the Beta; S.E., ​standard error.

Common job stressors' effects on the probability of stress arousal are proven in the overall sample. Irregular workers tend to experience stress more easily than regular workers, though the probability drops considering the interaction term of employment status as an irregular worker and job insecurity. Stress odds ratios are rather equivalent among individual service sectors, with an exceptional difference between production service and nonservice. The result also shows an interesting paradox that social support raises the possibility of stress arousal, as opposed to common argument, which means that workers who perceive they can get social supports at the workplace may have a higher possibility of feeling stress at work.

Emotional laborer's job stressors and relievers

For the next step, to identify stressors and stress relievers especially for emotional laborers and to test their effects in a comparable sense, the sample was divided into two subsets, emotional laborers (n = 19,472) and a second group who do not necessarily involve EL at work (hereafter referred as nonemotional laborers for convenience purpose, n = 30,560), and the two subsets' job stress model compared. Comparable logistic regression results on the two groups' job stress are presented in Table 4.

Table 4

Logistic regression result between emotional laborers and nonemotional laborers

	Emotional laborers						Nonemotional laborers
Job-stressor model			Stressor -reliever model			Stressor-reliever model
B	S.E.	Exp(B)	B	S.E.	Exp(B)	B	S.E.	Exp(B)
Demographic (control)
Gender	−0.124	0.066	0.884	−0.134*	0.066	0.874	−0.051	0.041	0.950
Education	−0.032	0.037	0.969	−0.037	0.038	0.963	0.057†	0.020	1.059
Health satisfaction	−0.220‡	0.049	0.803	−0.208‡	0.050	0.812	-0.244‡	0.027	0.783
Work conditions
employment status (ref: regular work)
Irregular work	0.293†	0.088	1.341	0.311†	0.089	1.364	0.130*	0.051	1.138
Irregular work * Job instability	−0.560*	0.266	0.571	−0.481	0.269	0.618	0.041	0.151	1.041
Service sector (ref: nonservice)
Production service	0.182	0.126	1.200	0.187	0.127	1.205	0.120*	0.053	1.127
Distribution service	0.189	0.114	1.208	0.184	0.115	1.202	−0.144*	0.056	0.866
Personal service	−0.025	0.119	0.975	−0.013	0.120	0.987	−0.041	0.054	0.960
Public service	0.179	0.128	1.196	0.170	0.130	1.186	−0.101	0.071	0.904
Weekly work h	0.010‡	0.003	1.010	0.008†	0.003	1.008	0.012‡	0.002	1.012
Autonomy
Work schedule	−0.087	0.046	0.917	−0.067	0.047	0.935	−0.047	0.030	0.954
Rest	0.098†	0.031	1.103	0.075*	0.032	1.078	0.078‡	0.018	1.081
Work method	0.350‡	0.064	1.419	0.328‡	0.065	1.389	−0.016	0.036	0.984
EL properties
Emotional Involvement	0.421‡	0.034	1.523	0.419‡	0.034	1.521	0.761‡	0.021	2.139
Work setting (ref: worker-side)	0.219*	0.097	1.244	0.239*	0.098	1.270	−0.025	0.069	0.975
Social support
Union in workplace				0.237*	0.107	1.268	0.305‡	0.055	1.357
Peer support				0.270‡	0.035	1.311	0.103‡	0.020	1.108
Friends at work				0.113	0.069	1.120	0.221‡	0.041	1.248
Satisfaction reward
ln(wage)	4.61‡	0.078	1.585	0.685‡	0.116	1.984	0.458‡	0.065	1.581
wage satisfaction				0.922	0.580	2.514	−0.225	0.334	0.799
ln(wage)*satisfaction				−0.264*	0.116	0.768	−0.045	0.066	0.956
Work-life balance				−0.137†	0.052	0.872	−0.143‡	0.030	0.867
Constant	−1.445‡	0.414	0.236	−2.439‡	0.606	0.087	−2.549‡	0.343	0.078
N	19,472						30,560
−2 Log likelihood	7,780.460‡			7,670.287‡			19,864.422‡

*p < 0.05.

†p < 0.01.

‡p < 0.001.

EL, emotional labor; Exp, ; S.E.,

The job stress model includes variables expected to have positive influences on emotional laborer's stress arousal. Result shows that general stress-related work environment variables also have positive effects on job stress arousal for emotional laborers. EL properties, namely emotional involvement and work setting also render positive effects to emotional laborers' stress arousal at a significant level. The effect of these EL properties are found in the stressor-reliever model as well, with relative strength to other variables. These findings conclude H2 on the double risk burden of stress arousal for emotional laborers is accepted.

The stress gap between irregular workers and regular workers was bigger among emotional laborers than nonemotional laborers. Regarding emotional laborers, irregular workers have a 36% higher probability of feeling stress than regular workers, whereas the gap is 14% for the nonemotional laborers group. The interaction effect of irregular work status and job insecurity also occurred in the EL subset. Given that the employment status of employees is linked to their sense of social identity, the finding seems to have associations with loyalty to the company, however, this interpretation calls for further supplements for generalization. The finding implies that some company's strategy of outsourcing EL centering on temporary contract employment is rather inefficient in the long-term, because it reduces company loyalty and increases costs. With some actual supporting cases that employment transition to permanent status led to increased organizational loyalty and satisfaction, and subsequently to cost savings, the finding conveys a message that status transition to permanent work would be key for mutual advancement of the company-emotional laborer.

The possibility of stress arousal is not significantly different among emotional laborers in different industry sectors. In the case of nonemotional laborers, stress odds ratios of workers in the production service sector and distribution service sector differed significantly from that of workers in the nonservice sector when considering EL properties. It roughly suggests a high relation between stress and EL itself, rather than industry or sector.

The effect of autonomy is proven to increase stress arousal of emotional laborers as in general workers. The finding of an estimated 40% higher possibility of stress arousal among respondents of perceived work method autonomy, opposes precedent arguments that following manuals in emotional expression, or decreased work method autonomy, would increase stress arousal probability.

The result proves that the more workers emotionally get involved in work, the more stress they experience. The effect of emotional involvement was stronger for nonemotional laborers, which assures that emotional involvement functioning as an extra job demand is a strong stressor. The finding suggests that ‘deep acting’ strategies, which many organizations pursue, may be more harmful to emotional laborers. In terms of work setting, a client-familiar setting is more stress-intriguing, as expected. Physical burden also has a positive correlation with emotional laborers' stress arousal at an insignificant level.

Social support variables render an increasing effect on emotional laborers' stress arousal, rejecting H3. Reward satisfaction shows a decreasing effect through its interaction effect with wage itself, and work-life balance was proven to decrease the possibility of stress arousal, so consequently, H4 is accepted.

In the stressor-reliever model, gender difference obtains statistical significance unlike in the job stressor model. Gender difference in stress arousal odds ratios suggests a gender-biased effect of stress relievers. Male emotional laborers outnumber women in this study, as presented earlier. Gender gap in emotional laborers' stress arousal lost its statistical significance after considering EL properties. The finding opens up a possible interpretation that the qualitative difference in EL in men and women may lead to a gender gap in stress arousal. The relationship between EL and gender is not so simple, as contradictory evidence has been provided in the field. However, it is argued that ELs for men are qualitatively different from ELs performed by women, and women face stronger demands for EL as emotional management in the private sphere has been perceived as a female task; the other side argues that gender difference is not significant [15], [16]. At present, regarding the finding, it is insufficient to draw a concrete conclusion on gender difference in stress arousal related to EL because it did not gain significance or effect of significant strength.

Discussion

This paper is of exploratory interest with an analytical approach to defining EL and understanding emotional laborers' job stress in line with job stress research for general workers, carried out using logistic regression analysis using KWCS data. Both universal and particular traits of EL are accounted for, by using work environment and EL properties as important variables in explaining emotional laborer's stress arousal. Through the analysis, it was attempted to see whether and how conventional job stressors apply, or not, to the understanding of emotional laborers' job stress.

The study has some inevitable analytical limitations. Firstly, it does not encompass self-employed workers in an analytic scope, although 36.7% of emotional laborers in the sample are self-employed. This was mainly because other independent variables in the data are fitted better to organized workplaces with employer(s) and a number of employees. However, self-employed worker's EL deserves more attention, especially in Korean society because a considerable number of domestic self-employed workers are clustered in restaurants and other small service businesses and subsequently exposed to risk of EL. Hence, more a detailed and comparative understanding of EL of self-employed workers should be investigated. Secondly, this study's macro-approach using data on overall industry sectors in Korean society has strengths in drawing a broad map of various factors, however, this approach also has the limitation of offering an explorative and somewhat vague explanation. This being so, the results of this study are not sufficient for a comprehensive understanding of Korean emotional laborers' job stress as a whole. However, given that microlevel research on EL is expanding through various fact-finding reports and related studies, this paper's approach may contribute to balance the scope of understanding and expanding knowledge in the field.

The four hypotheses of this study, except H3 on the effect of social support, are supported by the analysis result: emotional laborers show 3.5 times higher possibility of stress arousal compared with other workers (H1); general work environment and EL properties both have an increasing effect on stress for emotional laborers, and comparison with nonemotional laborer groups enhanced the finding that EL and related properties, and emotional involvement and work setting in particular, function as strong stressors (H2); the analysis also found that the reward of an optimally satisfying level and work-life balance are significant stress relievers (H4).

Social support is usually considered a significant moderating variable in job stress modeling. However, it does not show a significant effect in this study. This contradicting finding can be partially ascribed to insufficient elaboration of the concept or other technical issues, but it is suggested that more ‘tricky’ reasons may apply here. Firstly, the fact that social support raises the probability of stress arousal for general workers implies that there may be an unrevealed context in Korea which makes social support a hidden pressure on job stress rather than a realistic support. Secondly, especially for emotional laborers, social support may not be well-established yet. In reality, worker-representative organizations for services workers, especially one for emotional laborers, are not well-established yet in Korean society. Also, considering that many EL jobs are newly created, chances are that social supports do not exist, or existing supports are not that helpful for individual stressful situations. A third finding of this study is that supports from peers or good friends at work may ease the aftermath of EL, but not emotional pressure itself during client contact situations. Stress coping at the “backstage” with informal networks among emotional laborers at work may not have a preventative effect on stress arousal. This seemingly paradoxical finding should be further elaborated with consideration to the expectation-reality gap in social support, along with more qualitative study on actual workplaces.

For both emotional laborers and nonemotional laborers, higher income is related to higher stress, however, the effect of subjective wage satisfaction and high income occurs only in emotional laborers, ceteris paribus. The relation of the wage system and work demands should also be considered to understand the positive correlation of income, satisfaction level, and stress. Regarding the fact that wage is often a reward for quantitative and/or qualitative work demands, a higher wage would infer the effect of incentive on more work achievement or position effect as the respondent occupies a higher position at work, which also requires more responsibilities. In these situations subjective reward satisfaction would follow the rule of optimum rather than that of maximum. In a similar vein, high wage earners who are subjectively satisfied with their wages are less likely to feel stress at work also implies that agreement on optimal wage has relieving effects. People who think their work and life are well-balanced showed less possibility of stress arousal. These findings bring up the point that rewards should reach a level that emotional laborers who are ‘in the actual fields’ are satisfied with.

The study also found an interaction effect of irregular employment status and perceived job insecurity, and interesting findings include the negative effect of work method autonomy and rest autonomy to emotional laborers' stress, contradictory to common understanding. These findings call for further analytic attention in this field, because implying practical and microlevel factors such as latent social norms in the field may intervene with the process of institutional protection to emotional laborers. Findings that ‘deep acting’ strategy, or emotional involvement in other words, and EL on a client-visiting basis increases stress arousal also suggest these work traits should be considered along with general guidelines in protection policies.

With further efforts to connect the findings of this study to other studies, EL should and can be understood within the context of broader labor market changes in Korea, and not seen as a separate sphere of attention. Such efforts will provide an opportunity for more detailed investigation of practical protection strategies of emotional laborers, from recovery solutions to the socio-psychological trauma that they are exposed daily, and more generally to modifications on reward systems and employment status transition based on the perception that EL results in intangible burdens for both workers and organizations.

Conflicts of interest

I certify that I have NO affiliations with or involvement in any organization or entity with any financial interest (such as honoraria; educational grants; participation in speakers’ bureaus; membership, employment, consultancies, stock ownership, or other equity interest; and expert testimony or patent-licensing arrangements), or non-financial interest (such as personal or professional relationships, affiliations, knowledge or beliefs) in the subject matter or materials discussed in this manuscript.