**4. Data Analysis and Results**

SPSS (IBM, Armonk, NY, USA) was used in this study. First, we calculate the correlations among core variables of this study. The results are shown in Table 1. The coefficients are not high. Second, the descriptive statistics are calculated. We can see that there is no significant problem with the mean and S.D. (standard deviation) of the core variable. All the correlation coefficients do not exceed 0.7. Third, the multicollinearity issue may impact the results. Consequently, we calculate the coefficients of variance inflation factors (VIFs). The results show that there is no VIF that exceeds 10. This indicates that there is no significant multicollinearity based on the view of Hair et al. [68].


**Table 1.** The results of the correlation matrix and descriptive statistics.

Note: \*\*\* *p* < 0.001; \* *p* < 0.05.

Fourth, we apply hierarchical linear analysis (HLA) to test the hypotheses proposed in this study. Seven models are created and the results are shown in Tables 2 and 3. We control the firm age, the number of firm employees, the educational background of employees, and the work experience of employees. Model 1 showed the results of the impact of three control variables on job performance. The results of model 1 verified that the influence of control variables is not significant. Then, the results of model 2 indicated that H1 was verified. The coefficient for employee mental health is 0.256, which is significant at *p* < 0.01. (Model 2). Therefore, the influence of employee mental health is positive. Model 3 was built to test H2a and H2b. From the results of model 3, both coefficients for work engagement and employee innovative behavior were positive and significant. The results show that work engagement and innovative behavior are positively related to job performance.

**Table 2.** The results of regression analysis (models 1–3).


Note: \*\*\* *p* < 0.001; \*\* *p* < 0.01; \* *p* < 0.05.



Note: \*\*\* *p* < 0.001; \*\* *p* < 0.01.

In order to test H3a and H3b, we built models 6 and 7 (Table 3). The results show that the impact of employee mental health on work engagement is positive (Model 6: β = 0.284; *p* < 0.001). Additionally, the impact of employee mental health on innovative behavior is positive (Model 7: β = 0.335; *p* < 0.001). The results indicated that both H3a and H3b were verified by the samples.

Model 4 was built based on model 2, which was applied to test the mediating role of work engagement. The results of model 4 indicated that the coefficient for work engagement is significant (model 4, β = 0.447; *p* < 0.001). However, the coefficient for employee mental health was not significant (model 4, β = 0.129; *ns*). From the results of model 2, model 4 and model 6, we could see that the positive mediating effect of work engagement on the relationship between employee mental health and job performance is significant. Therefore, hypothesis 4a is supported by the samples.

Model 5 was built based on model 2, which was applied to test the mediating role of innovative behavior. The results of model 5 indicated that the coefficient for innovative behavior was significant (model 5, β = 0.444; *p* < 0.001). However, the coefficient for employee mental health is also not significant (model 5, β = 0.108; *ns*). From the results of model 2, model 5 and model 7, we could see that the positive mediating effect of innovative behavior is significant. Therefore, hypothesis 4b is supported by the samples.

The endogeneity problem may be driven by some unobservable characteristics of the firm and employee [69]. Therefore, we consider several control variables in the model. We control the firm age, the number of firm employees, the educational background of employees, and the work experience of employees. Moreover, job satisfaction is considered as a proxy variable of employee mental health. We conduct the hierarchical linear analysis (HLA) and find that Hypotheses 1–3 are supported by data. Therefore, the results show there is not a significant endogeneity problem.

## **5. Discussion**

The present study pursued three goals in extending the extant knowledge on the relationship between employee mental health and job performance. First, we set out to investigate how employee' mental health influences job performance in an emerging economy context. In line with previous theories and research in developed economies [25], we predicted and found that employee mental health also exerts a positive influence on job performance in China. This finding indicates that the mental health of employees is an important factor to predict job performance. Moreover, the result that employee mental health positively affects job performance is robust and valid.

Second, to enrich our insights into the antecedents of job performance, we predicted and found that employee innovation behavior and work engagement positively affect job performance. It is plausible that job performance is enhanced because employees who are more dedicated to work and exhibit more innovative behavior are more effective in meeting the demands of firms, thereby leads to better development of the firm. Importantly, this result extends the findings on factors that promote job performance.

Third, we found that employee mental health is indirectly associated with job performance via innovative behavior and work engagement, which addresses tasks associated with work effectiveness. The positive affect state inherent to mental health is conveyed through innovative behavior and work engagement that are important for work demands. In turn, these two behaviors are positively associated with job performance. These results concerning indirect effects suggest an important nomological chain that begins with the mental health of the employee, which is positively connected with investing more energy and resources via innovative behavior and work engagement to create positive, productive work conditions, leading to better job performance. This indirect connection leads to useful insights into the mental health–job performance relationship.
