The Effect of Abusive Supervision on Budget Slack: The Mediating Role of Emotional Exhaustion

Past research demonstrates that abuse supervision significantly impacts subordinates’ psychological distress, attitudes, workplace deviant behaviors and performance and thus has strong implications for organization. Most scholars believe that budget slack will lead to waste of resources and inefficient enterprises. How to restrain budget slack has always been one of the focus problems in budget practice and research. However, abuse supervision has not been explored in a budgeting context. Accordingly, we introduce the abusive supervision into the budget context and complement other negative behavior induced by abusive supervision to explore the influence path of abusive supervision on budget slack. Our results imply that the abusive supervision has a positive impact on budget slack of subordinate. In addition, emotional exhaustion played a fully mediating effect between abusive supervision and budget slack.


Introduction
As a kind of negative leadership behavior, abusive supervision exists in all kinds of organizations around the world, and has many negative effects on subordinates' attitude and behavior (Harvey et al., 2007;Huang et al., 2017). Several studies focus on negative consequences of abusive supervision. For example, Tepper (2000) finds that when subordinates perceive superiors' abusive supervision, they usually reduce their emotional commitment to the organization, leading to workplace deviant behavior. Sun (2019) empirically demonstrates that the higher the abusive supervision perceived by subordinates is, the higher the cheating behavior frequency. It further illustrates there is a significant positive effect between the abusive supervision and negative behavior of subordinates.
Budget is an important part of enterprise management and control system.
Most enterprises have realized budget management in China (Yu et al., 2004).
However, opportunistic behaviors such as budget slack are hidden in budget practice (Mao & Zhu, 2009). Budget slack is the behavior in which budget actors deliberately underestimate actual income or capacity, or overestimate the cost or resources needed (Dunk & Nouri, 1998). Most scholars believe that budget slack will lead to waste of resources and inefficient enterprises, thereby eroding enterprise value. It should be controlled (Dunk & Nouri, 1998;Zhang & Zhai, 2009).
Therefore, how to restrain budget slack has always been one of the focus problems in budget research.
In the present research on budget slack, psychology-based researchers focus on employees' mental states and behavior, and propose that the emergence of individual budget slack behavior follows the chain of "stimulating -psychologybehavior" (Covaleski et al., 2003), and in the process, the internal and external pressure felt by individuals is the most commonly used explanatory variable in the study (Libby, 1999;Webb, 2002;Stevens, 2002;Little et al., 2002). Besides, research demonstrates that the abusive supervision carried out by superiors is a main pressure source of subordinates (Ma et al., 2017). However, to our knowledge, the environment where abusive supervision remains unexplored is in a budgeting context.
In particular, psychology-based researchers from the protection motivation theory proposed that external stimulus (such as stress) can induce individuals to open up self-protection mechanism and then produce corresponding self-protection behavior (Rogers, 1975). Studies following this path examine the mediating effect of emotional exhaustion on the relationship between abusive supervision and negative behavior (Chi & Liang, 2013;Whitman et al., 2012). To add to this literature, we draw from the protection motivation theory and research in budgeting to introduce the abusive supervision into budget context and explore the mediating effect of emotional exhaustion between abusive supervision and budget slack.
Our results provide several contributions to research and practice. First, this paper introduces the abusive supervision into the budget context and complements other negative behavior induced by abusive supervision to explore the influence path of abusive supervision on budget slack. Further, from the protection motivation theory, our results indicate empirical evidence on emotional exhaustion as a mediating link between abusive supervision and budget slack.

Development of Hypotheses
Abusive supervision and budget slack Scholars have reached a consensus on the positive correlation between information asymmetry and budget slack (Bourageois, 1981;Kren, 1992), and the information asymmetry between the superior and the subordinate is the prerequisite for the tendency of budget slack (Zhang et al., 2004).
Abusive supervision as a destructive negative leadership style has received widespread attention. The pressure on subordinates in their work is the main cause of negative behavior (Mitchell, Baer, Ambrose, Folger, & Palmer, 2018), and the abusive supervision carried out by superiors is a main source of this pressure (Ma, Xi, Xu, & Zhao, 2017). When subordinates are in a negative emotion for a long time, they gradually reduce the consideration of ethics and morality, and easily induce unethical behavior (Kouchaki & Desai, 2015). The threat of abusive supervision can also lead to employees' cheating behavior (Sun, 2019).
Moreover, the essence of budget slack is a moral adverse selection (Kren, 1992;Gao & Peng, 2009). Follow this logic, in this paper, the process of abusive supervision positively affects the cheating behavior of subordinates is brought into the budget management environment. It is believed that abusive supervision will lead to budget slack.
Thus, the following hypothesis is tested: H 1 : The abusive supervision has a positively impact on budget slack of subordinate.

Abusive supervision and emotional exhaustion
Several studies in the psychology literature have shown that abusive supervision affects subordinates' work attitude and mental health, including emotional exhaustion, turnover tendency, work-family conflict, job satisfaction, organizational commitment, etc. (Ashforth, 1997;Tepper, 2000Tepper, , 2007Tepper, Duffy, Hoobler, & Ensley, 2001). When employees feel threatened, they create psychological stress, which in turn produces a series of depleted states such as emotional exhaustion (Hobfoll, Freedy, Lane, & Geller, 1990). As a stress, abusive supervision damages the psychological and physiological resources (Whitman et al., 2012) of employees. And studies have confirmed abusive supervision significantly stimulate emotional exhaustion of employees, when employees are subjected to abusive behaviors, such as being ignored, not supported, accumulated some negative emotions, causing emotional exhaustion (Cao, Li, & Zhan, 2015).
When subordinates are in the emotional exhaustion, in order to avoid being punished for the decrease of performance (Golparvar, 2015), it is easy for the subordinates to take false reports or exaggerate his work performance and lead to cheating behavior of subordinates (Chi & Liang, 2013;Whitman et al., 2012;Mitchell et al., 2018). In addition, Lawrence and Michele (2017) examined that emotional exhaustion, as a negative emotional, can reduce employees' self-management ability and relax their high standards, which may lead to some unethical behavior. When subordinates are in the emotional exhaustion, in order to avoid being punished for the decrease of performance, it is easy for the subordinates to take false reports or exaggerate his work performance to avoid punishment (Golparvar, 2015). Kim, Hwang, & Lee (2008) supported when the subordinates is in the emotional exhaustion, its self-control ability will decline, which will affect its control of instinctive impulse, and then leading to the cheating behavior. For example, when subordinates are in the emotional exhaustion, they will give up their commitment to honesty and prefer instant rewards (Mazar et al., 2008).
There is also evidence to support the expectation that emotional exhaustion represents an intervening variable in abusive supervision and budget slack. In order to expand the psychological and behavioral mechanism of employees, many studies have taken emotional exhaustion as a mediating variable between abusive supervision and employees' negative behavior. For example, Whitman et al. (2012) based on conservation of resources theory, found that abusive supervision is significantly negatively correlated with subordinates' feedback avoidance behavior, and the emotional exhaustion plays a mediating role in it. Chi and Liang (2013) also based on conservation of resources theory, found that abusive supervision and negative work behavior of subordinates is significantly positive correlation. Emotional exhaustion plays a moderating role, while cognitive revaluation and expression inhibition play a moderating role. We propose a moderating effect of abusive supervision on budget slack through emotional exhaustion.
Thus, the following hypothesis is tested: H 2 : The emotional exhaustion significantly mediates the relationship between superior abusive supervision and budget slack of subordinate.

Participants
In the selection of samples, at the enterprise level, follow the two restrictions: 1) the enterprise is in the manufacturing industry. Because in China the larger manufacturing enterprises are the first to put into the budget management system, and at present, they are the most extensive and mature enterprises in budget management practice (Yu et al., 2004); 2) the scale of the enterprises is not less than staff of 100. Because the scale of the enterprises is too small budget system is too simple, the budget function is too limited (Libby & Linsay, 2010).
On the basis of the limitation of the research enterprises, the population selected in this paper is the employee of the production department. Because from the point of view of management accounting, the production department is the Y. Tong, M. Prompanyo Open Journal of Accounting most important department of enterprise budget control system, and it is directly related to enterprise performance (Subramaniam & Mia, 2003).
After questionnaire surveys, a total of 600 questionnaires were issued, and 510 were recovered, excluding invalid questionnaires in the statistics, among which 400 were valid, and the effective recovery rate was 78.43%. In 400 research samples, the proportion of respondents with more than 5 years of industrial experience is too high, accounting for more than 60%. From the enterprise point of view, state-owned enterprises were the main respondents, accounting for more than 60%, and the total proportion of enterprises more than 500 people accounted for over 50%.
The Liker's five-point scale is used for the measurement of variables abusive supervision, emotional exhaustion, and budget slack in this survey questionnaire. The question-asking technique of Liker's five-point scale is to design a way for the respondents to judge whether their attitude towards a concept is positive or negative according to the specific score (McDaniel and Gates, 2015). In the Liker's five-point scale, the scale ranging from 1 representing Totally disagree to 5 representing Totally agree. Additional demographic information is presented in Table 1.

Abusive Supervision
The abusive supervision is a kind of feeling perceived by subordinates. At present, the more mature measurement scale in academic research is the 15-item  Tepper (2000). It has been widely used in prior abusive supervision by scholars and verified to good reliability (Aryee et al., 2008;Thau & Mitchell, 2010; was used. The 15-item of an abusive supervision measure are positive score. The higher the score is, the more the abusive supervision is perceived by subordinates.

Emotional Exhaustion
The Maslach Burnout Inventory (MBI) developed by Maslach et al. (1986) has been widely used effectively in prior burnout research. MBI contains three dimensions: 1) emotional exhaustion, 2) depersonalization, and 3) diminished personal accomplishment. Among them, emotional exhaustion is the core dimension of burnout, which refers to the negative psychological state (Maslach, Jackson, Leiter, Schaufeli, & Schwab, 1986) in which individuals over-consumption of their own resources (interpersonal relationships, knowledge and skills, etc.) to produce exhaustion, which is mainly manifested in the loss of interest and enthusiasm for something and the inability to raise spirit, etc. The emotional exhaustion scales has considerable psychometric support Li & Li, 2006), and has been used in previous mental state studies alone (Lawrence & Michele, 2017). Therefore, we used a self-rated measure of emotional exhaustion in the MBI (Maslach et al., 1986). The 9-item scale of an emotional exhaustion measure are positive score. That is, the higher the score is, the higher the emotional exhaustion frequency.

Budget Slack (Dependent Variable)
Budget slack is the behavior in which the subordinates to take advantage of the asymmetry between superior and subordinate, deliberately overestimate expenditure, underestimate income, or undervalue performance for the sake of their own interests (Dunk & Nouri, 1998;Yu et al., 2004). A four-item scale of the tendency of budget slack measure employed by Onsi (1973) and used effectively in prior budgeting research was used.

Control Variable
Previous studies have shown that the employee's tenure have an impact on his budget behavior and attitude through experience effects (Schatzberg & Stevens, 2008), and the ownership and scale are related to the budget characteristics of the enterprise and then affect the employee's budget behavior (Onsi, 1973). Therefore, in this study, the employee's tenure, the ownership and the scale of the enterprise are selected as control variables to exclude their interference with the employees' budget slack.
The employees' tenure is measured by interval range, and the empirical analysis is carried out according to interval assignment "1 -5". As far as the ownership of the enterprise is concerned, it is simplified into two categories: state-owned enterprises and non-state-owned enterprises as category variables, and 0 -1 variables are formed by virtualization. The enterprise scale is calculated by the suggestion of Kimberly (1976) and according to the research population and sample determined in this paper, that is manufacturing enterprises with no Y. Tong, M. Prompanyo Open Journal of Accounting less than 100 people. Further according to the large scale, medium and small scale of Chinese manufacturing enterprises, mainly divided into three categories, and assignment "1 -3". Table 2 presents the min, max, means and standard deviations, and Table 3 shows the correlations of the variables. The results in Table 3 show that there is a significant positive correlation between abusive supervision and emotional exhaustion (r = 0.454, p < 0.01), there is a significant positive correlation between abusive supervision and budget slack (r = 0.208, p < 0.01); there is a significant positive correlation between emotional exhaustion and budget slack (r = 0.375, p < 0.01); there is a negative correlation between moral identity and emotional exhaustion (r = −0.173, p < 0.01). The correlation between these variables is consistent with the research hypothesis, which provides preliminary evidence for the verification of theoretical hypothesis.

Analysis and Results
Following Baron and Kenny's (1986) research, we estimated three regression equations to test the hypotheses. 1) Independent variables (abusive supervision) have a significant effect on dependent variables (budget slack); 2) Independent  The results in Table 4 show that: in Equation (1), when Budget Slack is regressed on Abusive Supervision, the regression coefficient for Abusive Supervision is significant, indicating that Abusive Supervision has a significant effect on Budget Slack (β = 0.203, p = 0.000); in Equation (2), when Emotional Exhaustion is regressed on Abusive Supervision, the regression coefficient for Abusive Supervision is significant, indicating that Abusive Supervision has a significant impact on Emotional Exhaustion (β = 0.454, p = 0.000); in Equation (3)

Discussion and Conclusions
In the study of abusive supervision, it is usually connected with count-produce behavior (Ding et al., 2016) and workplace deviation behavior (Tepper, 2000).
Restraining the budget slack has always been one of important issue in organizations. Most of the relevant literatures study how to restrain budget slack from the perspective of agency theory such as budget participation (Huang, 2012), information asymmetry (Chow et al., 1988), budget-based contract (Schif & Lewin, 1970), variance investigation (Webb, 2002) and so on. However, the problem of budget slack is not only the agent problem, but also the violation of social norms and deviant behavior (Deng et al., 2014). To our knowledge, abusive supervision has not been extended into the budget environment. Thus, our results contribute and innovate to research that introduces the abusive supervision into the budget environment and complements other negative behavior induced by abusive supervision to explore the influence path of abusive supervision on budget slack.
There are several limitations. First, the population selected in this paper is the employee of the production department. Second, the individual differences are not taken into account in the study of abusive supervision perceived by subordinates and budget slack.
Future research is needed to take into account individual differences, such as the degree of individual dependence on the organization, and the final budget slack behavior will be different. For example, according to the power/dependence theory, the impact of abusive supervision on budget slack is also affected by the degree of individual dependence on the organization. The study has shown that abusive supervision is more strongly associated with subordinates' workplace deviance behavior when subordinates' intention to quit is higher (Tepper et al., 2009). Therefore, the variables of individual dependence on organization should be taken into account in future research.

Conflicts of Interest
The authors declare no conflicts of interest regarding the publication of this paper.