**4. Results**

In the case of factor analysis, the construct Job Demands (JD) was reduced to a single component that explained 35.38% of total variance (the other elements presented eigenvalues below one). The construct Job Control (JC) was composed of two elements: Skill Discretion (SD) and Decision Authority (DA). These are summarized in two factors that explained 38.55% and 46.84% of variance, respectively. The construct Social Support (SS) was composed, as explained above, by two elements: Supervisor Support (SS) and Co-Worker Support (WS). The first is represented by a factor that explained 75.54% of variance, while the second was a construct consisting of a single item.

First, we analyze the mean square error (MSE) of the linear regression, including interactions considered in this research, compared to the model of the trained neural network according to Extreme Learning Machine. The classical model earned an MSE of 31.8862 and a neural network 11.4588 MSE, which justifies that the neural network model summarizes data more effectively than the classic model. The result of the regularization was cross-validated, resulting in a value of 10E3. In addition, the number of hidden layer neurons was fixed at 500.

After verifying that the non-linear model had greater precision than the linear model, its parameters were interpreted by applying the above sensitivity analysis. The first-order analysis qualifies the contribution to the output of the different input variables (job demands, skill discretion, decision authority, supervisor support and co-worker support) in a direct way without interactions. Table 1 presents the contribution to the variance of each of these variables and their signs (which were estimated empirically).


**Table 1.** Analysis of the first order. Contribution to the variance and sign.

JD (Job Demands), SD (Skill Discretion), DA (Decision Authority), SS (Social Support), WS (Co-Worker Support).

The variables that contribute most significantly to the explanation of perceived occupational health of accountants are job demands, with a negative sign, and supervisor support, with a positive sign. These results highlight, on the one hand, that an accountant's health deteriorates as job demands increase, while having the support of immediate superiors contributes positively to perceived occupational health. Since the results of estimation 1 and 2 are close (see Table 1), they can be considered robust. Table 1 also shows that the direct effect of the other variables (skill discretion, decision authority and co-worker support) is irrelevant in the perception of accountants' health.

Then, we analyzed the possible moderating effect of job control and social support in the relationship between job demands and perceived occupational health. This was solved by incorporating the contribution to the variance of the different interactions of the variable two by two.

The results of such interactions, all of them with positive signs, are presented in Table 2. The positive sign of the interactions when job demands had a negative relationship with perceived occupational health confirms the moderating effect that show both social support and job control. Focusing on Table 2, we can appreciate that the possibility of practicing skills as well as having the confidence in support from the top are the two variables that largely reduce the negative effect of job demands in perceived occupational health by the accountants.


**Table 2.** Analysis of the iterations of the variables. Contribution to the variance.

JD (Job Demands), SD (Skill Discretion), DA (Decision Authority), SS (Social Support), WS (Co-Worker Support). Above, the axis values correspond to interactions in Scenario 1, while values below the axis correspond to Scenario 2. In bold, the highest values are highlighted.

#### **5. Discussion**

Accountants play an important role in market economies. Financial information reduces costs of transaction in agency relationship; therefore, accountants are an essential link in the relations between owners and managers [70]. In this context, professional accountants that audit give credibility to the system, constituting an effective signaling mechanism in corporate governance [71] and influence in credit decisions [72]. Likewise, Positive Accounting Theory has shown the influence of accounting figures in taxation, sectoral regulation and executive compensation plans [73].

The significance and relevance of this profession places stress on accountants, who are forced to contend with strong demands that can affect their health. The most significant include professionals who work with tight deadlines, with significant seasonality (since most companies issue financial reports at the same time, coinciding with the calendar year) and in an environment of extreme competition, which reduces prices and increases pressure on resource allocation.

In this context of strain, the accountant must "juggle" to put into practice the independence, judgement, and professional skepticism that international auditing agencies require. Therefore, any deterioration in the health of these professionals, which alter their emotional balance, will decisively affect the judgement of the professional accountant and the quality of their work.

The present paper is relevant in the European context and investigates the effects of job demands in occupational health perceived by the accountants, as well as the moderator role that job control exerts over work and the social support of colleagues and superiors. The results, after the application of a neural network model, confirm some of the hypotheses raised in this research, both with regard to the outcomes and the effects of the moderators.

On the one hand, in relation to the direct effects of the job demands, job control and social support on perceived occupational health (H1 to H5), it has been shown that job demands are the factor that best explains the deterioration of accountants' health, which is in line with the prior literature that has verified the existence of this relationship in other manifestations of loss of health, such as burnout [38]. The other variable that exerts an important influence on perceived occupational health is supervisor support—in this case, in a positive sense. This result is consistent with the model of an organization staffed by professional accountants and is based mainly on professional talent. The implementation of professional talent requires the support and trust of the organization. The results of this study highlight better occupational health among professionals who receive explicit support from their superiors. On the other hand, job control does not exert a significant direct effect on perceived occupational health, possibly because the work of the accountants is structured and formalized, leaving little room for individual initiative.

Otherwise, the JDCS model discusses a number of mechanisms that contribute to cushioning the pernicious effect of the job demands on perceived occupational health. The results derived from the neural network model suggest that these mechanisms indeed improve the perception of one's own occupational health—specifically, the implementation of vocational skills and perception of greater support of hierarchical superiors. Accounting firms are characterized by their high level of hierarchy and fierce pyramidal structure. This design requires a high degree of staff turnover, which allows the channels of promotion to be open [74]. Since accountants' immediate superiors evaluate them, support from superiors is one of the mechanisms for recognition with greater impact on the development of a healthy work environment [41]. In a qualitative study, the importance of superiors in the self-esteem of subordinates is underlined: "To hear my senior say I've done a good job is a real boost to my morale!". The comment of this assistant illustrates the process governing her approach to auditing: She was congratulated for her work, and she takes the compliment personally ("I'm doing a good job"), thus strengthening her identity, and this prospect is precisely what motivates her to do her best [31]. To fulfil their expectations in terms of self-achievement, some auditors go further. Undoubtedly, support of one's superiors and a feeling of justice decrease the feeling of burnout among accountants [19,22]. In addition, the confidence of being supported by the organization and its managers in a competitive environment where lawsuits are frequent contributes to a vital balance [18].

The other variable with a significant moderating effect in the relationship between job demand and perceived occupational health is the possibility of practicing personal skills (skill discretion). The analysis of this effect, which has much to do with the concept of hardiness, is especially relevant in the profession discussed in this research. People with hardy personalities perceive stimulating and challenging situations as stressful (commitment dimension), probably because they believe that stress factors are controllable thanks to their professional skills (skill discretion), thus transforming risk into an opportunity for personal growth (challenge dimension). Commitment and challenge show a significant relationship with exhaustion, one of the fundamental dimensions of burnout, as point out Reference [26]. This result offers the possibility of aligning the needs of firms to respond

to a competitive market with the professional interests of the accountants to the point where many professionals consider joining these companies as a promising start to a career or as a highly stimulating professional activity [47].

In short (see Figure 3), the direct effects on perceived occupational health come primarily from job demands (Hypothesis 1) and the recognition of superiors (Hypothesis 4), while the incidence of the two dimensions of job control (Hypothesis 2 and 3) and co-worker support (Hypothesis 5) is not relevant. For the moderating effects, the relationship between job demand and perceived occupational health is reducing among professionals who can practice their professional skills (Hypothesis 6) and have the support of their superiors (Hypothesis 8). The rest of the moderator effects scarcely explain the variance of the model: Decision authority (Hypothesis 7) and co-worker support (Hypothesis 9).

**Figure 3.** Model hypotheses confirmed.

Two important practical implications for organizations and their human resource managers can be drawn from this research. On the one hand, recruitment and selection processes should pay special attention to candidates with hardiness, i.e., those who are capable of transforming stressful situations into opportunities for growth. Professionals who have this competence profile better bear the pressure inherent in the work, and hardiness is especially valuable for developing professional careers with long spans, such as accountancy firms. On the other hand, in an environment that is as demanding as the object of this paper in which one is subjected to intense pressures of time and deadlines, superiors play a central role in the promotion of a healthy work environment. Therefore, superiors should receive practical training in managing people so that they are aware of and develop the skills needed to provide the social support that accounting professionals need and demand. Leading practices and training programmes implemented by large accountancy firms to support the mental health of employees are displayed by Reference [13].

As noted in the introduction to this research, Total Worker Health promotes integral intervention in health that considers both the individual perspective and the organizational context. In this sense, JDCS model is one of the theoretical approaches more suited to understanding and interpreting the relationships between work features and the health and well-being of employees [8]. As a conclusion, we can say that this research confirms the basic postulates of the JDCS model (professional development and supervisor support mainly constitute the basic mechanisms for moderating the high pressures of the accounting work), thus contributes to explaining the perceived occupational health of accounting profession from an original and novel theoretical framework on this field.

#### **6. Limitations and Future Lines of Research**

Finally, it is necessary to mention the major methodological limitations of this study. First, perceived occupational health was measured through individual self-perceptions. Second, the problem

of social desirability is a setback of studies that ask about how labor conditions affect employees. Self-perception and the social desirability may cause bias in responses. Third, the relationship between the variables being investigated cannot be considered causally since we studied cross-sectional data, not experimental data. Finally, the study is limited to the scope of the European Union. Future studies should investigate the influence of the JDCS model in professional accountants who develop their activities in other cultural contexts.
