The Effect of Role Conflicts on Turnover Intention Auditors of Public Accounting Firms in the Covid-19 Pandemic

Research purposes: This study aims to examine the model of antecedents and consequences of role conflict of auditors working in Big 4 public accounting firms (KAP) in the era of the Covid-19 pandemic. The role conflicts used in the study were work-family conflict (WFC) and family-work conflict (FWC). The antecedent variable of the WFC and FWC used in the study is supervisor support. While the consequence variables of WFC and FWC are job stress, job satisfaction, job performance, and turnover intention. This study used conflict theory and conservation of resource theory to construct the hypotheses. Methods: This research method is a survey method using electronic questionnaires. The respond-ents (sample) of this study are auditors working in KAP Big 4. Data analysis using partial least square structural equation modeling (PLS-SEM). Findings: The results concluded that supervisor support was negatively significant to family-work conflict but not significant to work-family conflict. The results concluded that role conflicts impact job stress, job satisfaction, performance, and turnover intention. Novelty: This article contributes to research in behavioral accounting, primarily related to the topic of job stress due to role conflicts felt by auditors because having to work from home during the Covid-19 pandemic finally impacts turnover intentions. We proposed a comprehensive model approach to the role conflict of auditors.


INTRODUCTION
In early 2020 the world was shocked by the coronavirus (Covid 19) outbreak that infected almost all countries in the world. The Government first announced Covid 19 cases in Indonesia on March 2, 2020. On April 9, 2020, the pandemic spread to 34 provinces in Indonesia. Until now, the Indonesian Government is still focused on overcoming the coronavirus in Indonesia. In anticipation of the spread of Covid 19, the Government first issued a disaster emergency from February 29 to May 29, 2020, related to this virus pandemic with a total of 91 days. The Government has taken steps to solve this extraordinary case: socializing the three movements, washing hands, comprehensive model of antecedents and conflict consequences of the auditor's role in KAP during the Covid 19 pandemic. We used supervisor support as antecedent variable and job stress, job satisfaction, job performance, and turnover intention as consequences variables of role conflict. Based on the problems described above, we propose whether supervisor support negatively affects the WFC and FWC, do WFC and FWC affect job performance and job satisfaction through job stress, does job stress directly affects turnover intention or indirectly affects turnover intention through job satisfaction and job performance.

Role Conflict Theory and Conservation of Resources Theory
Role conflict theory explains that the expectations of two domains, work or family, are not aligned due to differences in norms and obligations (Burke, 1986). Individuals experience conflicts at work due to pressures that co-occur. Conflict occurs because performance in the domain of life (work) reduces the amount of time, energy, and other resources available to other parts, namely families (Hargis et al., 2011). The pressure or conflict of this role causes WFC and FWC where work interferes with family life, and family life interferes with work (family life interfering with work). Greenhaus and Beutell (1985) explained that WFC and FWC are a form of role conflict due to the pressure of roles from the work and family domains that are not somehow aligned with each other. Greenhaus and Beutell (1985) define WFC as a form of role conflict. Some responsibilities of the work and family domains are misaligned, so this form of conflict causes work to interfere with family life. FWC is defined as a form of conflict of roles where family and work responsibilities are not aligned, which causes family life to interfere with work. Conservation of resources (COR) theory developed by Hobfoll (2001). COR theory explains that people have limited resources such as energy, cognitive effort to be used in work and family. WFC and FWC cause stress because these resources are lost in the process of fulfilling their roles in work and family (Hobfoll, 2002). Nonconformity of such role requirements can lead to role conflicts that affect work-related behavior (Lui et al., 2001).

Supervisor Support and Role Conflict
Supervisory support is the support of supervisors in the organization about work. Supervisory support is considered a common form of social support that can provide socioemotional resources related to the demands of their work (Selvarajan et al., 2013). Research in the work-family conflict study explains that supervisor support is necessary for reducing WFC and FWC (Breaugh & Frye, 2008;Frye & Breaugh, 2004;Hargis et al., 2011;Selvarajan et al., 2013). The supervisory support is support system form from supervisor as a part of organization so can reduce the occurrence of WFC and FWC. Frye dan Breaugh (2004), Breaugh dan Frye (2008), and Selvarajan et al. (2013) found that supervisor support negatively influences the WFC and FWC. Based on the theory and previous research results about antecedents of WFC and FWC, this study proposed the hypothesis: H 1a : Supervisor support negatively affect WFC H 1b : Supervisor support negatively affect FWC

Consequences of WFC and FWC
Job stress is a work-related nervousness or anxiety, which affects an employee's emotions and physical health (Cox, Griffiths, & Rial-González, 2000). Job stress can appear around the workplace or in work itself. Job stress may arise due to role ambiguity, too much work, role conflict, and time pressure (Ji et al., 2020). Netemeyer et al. (2005) found that WFC and FWC positively affect job stress. High role conflict can affect the high job stress. However, Jaramillo et al. (2006) concluded that conflict negatively affects job satisfaction. Job stress has an impact on low job satisfaction. Previous research has found that job stress affects job satisfaction (Jou, Kuo, & Tang, 2013;Mansoor, Fida, & Ali, 2011;Rahman & Zanzi, 1995;Wang et al., 2014;Jou et al. 2013;Turliuc & Buliga, 2014). Based on the theory and previous research results we proposed the hypothesis: H 2a : Job stress mediates the relationship between WFC and job satisfaction H 2b : Job stress mediates the relationship between FWC and job satisfaction Previous research has shown that role conflicts cause job stress in the work environment and ultimately impact the decline in employee job performance (Mete et al., 2014). Job stress causes the decrease job performance because employee less motivation in work. Hoobler et al. (2009) found that job stress mediates the relationship of FWC and job performance. But Mete et al. (2014) found that WFC and FWC did not affect job performance. Based on the theory and previous research results we proposed the hypothesis:

H 3a : Job stress mediates the relationship between WFC and job performance H 3b : Job stress mediates the relationship between FWC and job performance
Low job performance and less job satisfaction leads to voluntary turnover (Iverson & Deery, 2000;McEvoy & Cascio, 1987;Williams & Livingstone, 1994). Jou et al. (2013) found that job satisfaction has a mediating effect on the relationship between job stress and turnover intention. Jaramillo et al. (2006) found that role conflict affects job satisfaction and turnover intention. This result can be concluded that job stress affects turnover intention through job satisfaction and performance. Based on the theory and previous research results about consequences of WFC and FWC, this study proposed the hypothesis:

Respondents and Procedure
Respondents to this study are auditors who work at the Big 4 Public Accounting Firm in Jakarta. Data were collected by survey using electronics questionnaires. Questionnaires were sent to respondents using online questionnaires or web-based questionnaires. The instruments (questionnaires) using Likert scale seven points, which 1 is strongly disagreeable and 7 strongly agree.
To ensure the questionnaire has been appropriately designed before the questionnaire is sent to the respondent pilot test is done. Pilot test questionnaires are conducted by sending to 30 university colleagues who can answer the questionnaire. Of the 30 questionnaires sent, only 26 responded (response rate 86.7%). This pilot test ensures the questionnaire can easily understand and estimated time to answer the questionnaire. The pilot test results concluded that the research instrument was valid, reliable, and easy to understand.

Variables Measurement
Supervisor support was measured by instruments from Frye and Breaugh (2004). The instrument consists of four question items. WFC and FWC were measured by instruments from Netemeyer et al. (1996). The WFC and FWC instrument consists of five questions items each. Job stress was measured by instruments from Cohen, Kamarck, and Mermelstein (1983). The instrument consists of four question items. Job satisfaction was measured by instruments from Rusbult and Farrell (1983). The instrument consists of six question items. Job performance is measured by the Individual Work Performance Questionnaire (IWPQ) adopted from Koopmans et al. (2013). The instrument consists of nine-question items related to the three main dimensions of performance: task performance (three items), contextual performance (three items), and adaptive performance (three items). Turnover intentions were measured by instruments adapted from Collins and Killough (1992), Pasewark and Viator (2006), and Viator (2001). The instrument consists of five question items.

Data Analysis
Data analysis this study uses Partial Least Square Structural Equation Modeling (PLS-SEM). This approach allows the researchers to analyses the relationships simultaneously of the complex model (Ramli et al., 2018). The PLS-SEM measure two models: outer and inner models. Outer model testing aims to test indicator reliability tests, internal reliability tests, convergent validity, and discriminant validity. While inner models for measuring intervariable relationships, i.e., the relationship of exogenous and endogenous variables. The inner model showed the R-Square for endogenous models, the estimate for path coefficients, effect size or f square to find out the fitness model, and prediction relevance.

RESULTS AND DISCUSSION
Participants involved in the study were 118 accountants/auditors working at public accounting firms ( Deloitte in Jakarta. Average respondents have not stated the origin of KAP; we can be understood this option because it is related to maintaining the privacy of respondents. Of the 118 respondents, 71 were female (60.20%), and 47 were male (39.82%). The gender sample showed that female auditors were more numerous than male auditors. The average age of respondents is a relatively young age, where the age range of 20-29 years as many as 81 people (68.60%) and the age range of 30-39 years as many as 37 people (31.40%). The data shows that the average education of respondents is accounting undergraduate (S1) as many as 72 people (61.0%), while postgraduate (S2) as many as 36 people (39.0%). The mean of respondent's experience ranges from 1-5 years as many as 73 people (61.9%), 6-10 years as much as 35.6%, and 11-15 years as many as three people (2.5%). Table 1 describes descriptive statistics of participant demographic data. Table 2 describes a statistical description of research variables related to the theoretical range, actual range, mean, and standard deviation of the variables. Based on the data (mean) that supervisor support is relatively high, work-family conflict is relatively high but family-work conflict is relative low, job stress is relative low, job satisfaction is relative low, job performance is relative high, and turnover intention is relative low.
The test of the outer model aims to test the validity (the convergent and discriminant validity) and the reliability (instrument reliability and indicator reliability). The results of the validity test using outer loading show indicator values above 0.70 so that they can be concluded valid, except for indicators SS1 (0.494), JSA3 (0.602), and TI4 (0.301). Indicators less than 0.70 are removed from the model so that three items of the question are dropped. After removing the three indicator items from the model, the outer model test is carried out to ensure the instrument's validity. The outer model results show the value of indicator value is above 0.70, so it is concluded  Figure 2 describes the outer model structural model results after removing three indicators because the outer loading value is less than 0.70. This study tests the validity using outer loadings value and average variance extracted (AVE). Reliability test using Cronbach's Alpha and composite reliability. Table 3 shows the outer model of validity and reliability test.
The inner model showed the R-Square values and path coefficients. R-Square measures the degree of variation in the influence of independent variables on dependent variables, while path coefficients are the significance and direction of the research hypothesis. Figure 3 describes the results of the inner model. The hypothesis test uses a path coefficient to see the significance level and the direction between variables. Path Coefficient values show the original sample values, t-statistics, and p-values that are bootstrapping results of Smart-PLS. Table 4 describes the path coefficient.
HIa and H1b stated that supervisor support negatively affects WFC and FWC. The results of the path coefficient concluded that the influence of supervisor support on WFC was insignificant (p-value 0.922 > 0.05), but supervisor support negatively affects FWC (-0.260, p-value 0.029 < 0.05). These results concluded that H1a was not supported, while H1b was supported. This result confirmed conservation of resources (COR) theory which explains that individuals have limited resources (energy, cognitive effort) to be used in work and family. This study supports the results of Frye and Breaugh (2004); Breaugh and Frye (2008); and Selvarajan et al. (2013), who  H2a stated that job stress mediates the relationship between WFC and job satisfaction. H2b stated that job stress mediates the relationship between FWC and job satisfaction. Table 4 showed that job stress mediates the relationship WFC and job satisfaction (0.520, p-value 0.000 < 0.05; -0.541, p-value 0.000 < 0.05) and FWC and job satisfaction (0.184, p-value 0.020 < 0.05; -0.541, p-value 0.000 < 0.05). The results concluded that H2a and H2b were supported, which WFC and FWC affect job satisfaction through job stress. These results explain that job stress is partially mediated. Role conflicts (WFC and FWC) impact stress because these resources are lost in the process of fulfilling their role in work and family (Hobfoll, 2002). This study confirmed the role conflict theory and previous research (Jaramillo et al., 2006;Netemeyer et al., 2005;Turliuc & Buliga, 2014;Y. Wang et al., 2014).
H3a stated that job stress mediates the relationship between WFC and job performance. H3b stated that job stress mediates the relationship between FWC and job performance. Table  4 showed that job stress mediates the relationship WFC and job performance (0.520, p-value 0.000 < 0.05; -0.206, p-value 0.018 < 0.05) and FWC and job performance (0.184, p-value 0.020 < 0.05; -0.206, p-value 0.000 < 0.05). The results of path analysis showed that H3a and H3b were supported, which WFC and FWC affect job performance through job stress.
H4 (H5) stated that job satisfaction (job performance) mediates the relationship between job stress and turnover intention. Table 4 showed that job satisfaction mediates the job stress and auditor's turnover intention (-0.541, p-value 0.000 < 0.05; -0.632, p-value 0.000 < 0.05) and that job performance mediates the job stress and auditor's turnover intention (-0.206, p-value 0.018 < 0.05; -0.129, p-value 0.026 < 0.05). These results concluded that H4 and H5 are supported. High job stress leads to high turnover intentions. In addition, the results of path coefficients found that job stress affects the auditor's turnover intention through job performance and job satisfaction. High stress impacts low job performance and job satisfaction, thus causing high turnover intentions. Job performance and job satisfaction are partial mediation variables (complementary) of the relationship between job stress and turnover intention. The results of this study supported the research of Iverson and Deery (2000), Pasewark and Viator (2006), Mete et al. (2014), Jaramillo et al. (2006), and Jou et al. (2013).

CONCLUSION
The study aims to propose a comprehensive model of antecedents and the consequences of WFC and FWC. The antecedent variable used is supervisor support, and the consequences are job stress, job satisfaction, job performance, and auditor turnover in KAP. The results concluded several conclusions. First, supervisor support affects the FWC but does not affect the WFC. These results explain that supervisor support reduces conflict roles because the family interferes with work when auditors work at home during the Covid 19 pandemic. Second, job stress has partially mediated WFC and FWC with job performance. The heightened conflict of roles, whether work interferes with family or family interferes work, affects high job stress and finally impacts job performance. The role conflict is caused by competing demands from family and works over the resources owned, thus causing stress because they have to fulfill functions in work and family. Third, WFC and FWC affect job satisfaction auditors through job stress. These results explain an indirect relationship between WFC and FWC with job satisfaction and job performance but through job stress. High role conflict causes high job stress and impacts low job satisfaction and job performance. Job stress has partially mediated the relationship between WFC and FWC with job satisfaction. Fourth, job stress affects the turnover intention of auditors through job performance and job satisfaction. High-stress jobs impact low job performance and job satisfaction, thus causing high turnover intentions. Job performance and job satisfaction have partially mediated the relationship between job stress and turnover intention.
This study implies accounting firms manage the workforce. The accounting firms to consider how work activities interfere with family duties. This study limited to one factor of the antecedents of the role conflict. For future research, the researchers have opportunities to propose other antecedent variables that may affect role conflict or other factors that can mitigate role conflicts. Researchers can use other social support factors such as family-friendly policies, perceived organizational support, and family supportive organizational climate (Frone et al., 1992). In addition, research can further test other factors other than support in the workplace, such as family support, that is, perceived spouse/partner support (Selvarajan et al., 2013).