Why and When do Big Four Auditors Engage in more
Questionable Audit Behaviours than Non-Big Four
Auditors?
Sjoerd A. Voetdijk
S3536793Master Thesis MSc Accounting 10644 words
Supervised by dr. D.B. Veltrop 20-1-2020
University of Groningen
Abstract
A crucial question that lies at the heart of what drives audit quality is 'why and when do auditors engage in questionable audit quality threatening behaviours (QTB)? QTB has received a great deal of attention from both academic research as well as regulatory agencies. In combination with the presence of structurally high auditor workload, QTB is an increasingly problematic issue for many audit firms. Overall, the goal of this study is to examine whether Big Four auditors are more likely to engage in QTB in comparison with non-Big Four auditors because they experience higher workload and to study whether auditor resilience moderates the effect of workload on QTB. I expect that auditor workload can be considered as an essential driver of QTB in audit firms. Furthermore, drawing from academic research as well as views from practice I anticipate that QTB may pose a more severe problem in Big Four audit firms because Big Four auditors generally experience higher workload than non-Big Four auditors. It is anticipated that auditor resilience plays a crucial role for auditors in dealing with such high workloads. Analysing survey data from 218 auditors – across Big Four and non-Big Four audit firms – the results of this study demonstrate that Big Four auditors, indeed, do engage in higher levels of QTB than non-Big Four auditors and that this is caused by higher workload. Surprisingly, no support is found for the moderating effect of resilience. In an additional analysis, it is shown that this moderating effect is significant for non-Big Four auditors, but in the opposite direction as anticipated. Finally, the implications, limitations and future research possibilities of this study are discussed.
Keywords: Auditing, Auditors, Audit Quality, Big Four auditors, Non-Big Four auditors, Quality Threatening Behaviour, Resilience, Workload
1
Introduction
On 25th September 2014, the Authority for the Financial Markets (AFM) in the Netherlands published a report stating that the auditing industry should improve governance and behaviour and create a quality-oriented culture to guarantee the quality of statutory audits. This is because the audit quality of multiple audit engagements of Big Four audit firms was found to be insufficient (AFM, 2014). Research showed that there was a lack of sufficient and appropriate audit evidence to substantiate the financial statements as a whole. Reason for AFM's recommendations was that the factors to be improved affect the competence of the individual auditor and the audit process the auditor executes, and therefore the audit quality (Knechel, Krishnan, Pevzner, Shefchik, & Velury, 2013).
The most common and used definition of 'audit quality' in literature is the (technical) ability to discover errors (of material misstatement) in accounting systems and to report these errors (DeAngelo, 1981). An audit and its quality are essential to reduce the information risk between different parties and to ensure that important decisions are based upon the correct and accurate information, such as the financial statements (Knechel, 2016). High audit quality results in high-quality financial reporting and the auditor has a significant role in guaranteeing the financial reporting quality (Schilder, 2011). But, what is high audit quality? Literature can often only describe what high audit quality is not, such as faults or other shortcomings by gathering audit evidence that negatively affects the output of the audit (Knechel et al., 2013).
In many countries, regulators and researchers have an increasing focus on the factors that drive audit quality (Francis, 2011). Academic research regarding audit quality has shown that the competence of individual auditors has a direct impact on the audit quality and that competence is also dependent on regulatory requirements and changes (Favere-Marchesi, 2000). Francis (2011) found that the audit process (thus the audit quality) is mainly affected by inputs of audit-testing procedures and the individual auditors on the engagement team. Knechel (2010) stated that the outcome of an audit (i.e. the quality) is uncertain and unobservable since the residual audit risk is hard to determine. Also, in 2008 the Financial Reporting Council (FRC) suggested that audit firm culture, such as ensuring sufficient time for auditors to complete the audit engagement, is recognised as an important driver of audit quality (FRC, 2008).
In recent years many auditing research has focused on how to increase audit quality by imposing new criteria for auditors. Examples of these criteria or regulations such as the SOX, non-audit services and mandatory audit firm rotation. When analysing regulatory changes such as SOX, Beattie, Fearnley and Hines (2013) find that most of the regulatory measures only moderately or slightly enhanced audit quality. Audit firms that deliver non-audit services to a client that the firms' audit as well, impact the independence of the auditor, resulting in reduced audit quality (Habib, 2012). Also, mandatory audit firm rotation mostly guarantees that the audit quality does not suffer due to the replacement of the incumbent auditor (Cameran, Prencipe, & Trombetta, 2016).
Remarkably enough, none of these studies focuses on the auditors themselves. Therefore, it can be suggested that most studies fail to recognise that the quality of the audit is the result of the work of different individual auditors working on the same audit engagement. However, research relating to audit quality found that perhaps the most important determinants that affect the competence of the auditor, thus the audit quality, are his or her characteristics (Christensen, Glover, Omer, & Shelley, 2016). A reason for lower audit quality is that auditors do not adequately carry out their assigned work.
This is also shown in the AFM research. Factors resulting in low audit quality were based on not adequately carrying out the assigned work by the engagement team (i.e. the auditor applied incorrect audit expertise; the auditor was responsible for too many audit engagements; the audit team had insufficient documentation discipline; insufficient performance reviews and training on
2
the job, cf. AFM, 2014, p.39). Prior research has already shown that auditors showed this kind of behaviour (Otley & Pierce, 1996a; Herrbach, 2005; Pierce & Sweeney, 2006).
The AFM (2017) performed another research and evaluated the shortcomings listed in their 2014 report. They concluded that the 2017 shortcomings were comparable with those listed in the previous report. This conclusion suggests that individual auditors still engage in the same kind of negative behaviours over the years, even though new regulations were implemented. Therefore, this study will focus on the behaviour of individual auditors themselves concerning the question why they do not carry out their assigned work well.
Audit quality threatening behaviour (QTB) includes behaviours such as prematurely signing off tests without completing all the work, for example, when a performance review is not executed (Sweeney & Pierce, 2004). QTB is an important driver of audit quality. It is possible that the consequences of QTB could inflict high costs, such as a future accounting scandal, litigation or new regulation. Therefore, reducing QTB is crucial. QTB occurs more quickly because of the nature of auditing, such as its complexity and the difficulty to measure audit quality (DeAngelo, 1981; Knechel et al., 2013). Auditors are aware of this and are in the position to exploit these weaknesses, definitely when consequences are absent (Pierce & Sweeney, 2006). Herrbach (2005) found that irregular auditing is significant in audit firms. Coram, NG and Woodliff (2004) found that auditors think it is more acceptable to adopt dubious audit evidence instead of re-sample the evidence. Another research found that junior and senior auditors conduct QTB most often because of lesser experience (Pierce & Sweeney, 2006). To ensure audit quality, it is crucial to understand why auditors engage in this type of behaviour completely.
In practice, but also in academic research, it is generally acknowledged that a critical root cause of QTB lies in auditor stress. This is caused by high workload, time budget pressure and deadline pressure (López & Peters, 2012; Bowrin & King, 2010; Ettredge, Bedard, & Johnstone, 2008; Margheim, Kelley, & Pattison, 2005). In recent literature, the focus is mostly on time-budget and time-saving aspects (Coram, Ng, & Woodliff, 2003; Gundry & Liyanarachchi, 2007). Extending this line of reasoning, I suggest that workload (i.e. the perceived workload in terms of pace and volume experienced by an employee, cf. Spector & Jex, 1998, p.3) may be considered as an important driver of QTB (de Vries & Herrijgers, 2018; Svanström, 2016; Persellin, Schmidt, & Wilkins, 2014; López & Peters, 2012).
Workload is an acknowledged and structural problem within the auditing industry (Sweeney & Pierce, 2002). Coincidently, Nyenrode Business University and the Dutch professional body of accountants (NBA) examined the workload which is experienced by auditing professionals in the Netherlands. The results show that their average workweek consists of 60 hours during the busy season (de Vries & Herrijgers, 2018), which is also shown in the academic study of Persellin et al. (2014). López and Peters (2012) investigated the impact of auditor workload compression during the busy season on audit quality. They found that the busy season is characterised by long days and stress, which often resulting in decreased auditor performance. When auditors are not able to conduct all audit work that is expected or planned due to high workload, this could result in QTB such as prematurely signing off the procedures (Pierce & Sweeney, 2004; Otley & Pierce, 1996b; Kelley & Margheim, 1990). According to the results of prior research and because of the structural presence of high workload, it becomes increasingly important for auditors to cope with this problem.
In the end, there are multiple factors driving audit quality even more. One of the key indicators determining audit quality is related to the behaviour of individual auditors taking part in the audit (Knechel et al., 2013; Svanström, 2016; Christensen et al., 2016). Svanström (2016) suggests that the relationship between time pressure and QTB is affected by individual auditor characteristics and that many of these characteristics are still unexplained. One of these barely explained
3
characteristics in auditing literature is resilience. Resilience has all to do with the ability to handle stress-related events and the ability to bounce back. In other professions, such as nursing, resilience is becoming increasingly more important due to their highly demanding work, the significant workload and the staffing shortage (Tusaie & Dyer, 2004). This is similar for auditors since they are confronted with small and large problems, challenges and workload every day. Due to these type of events, the auditor often experiences high amounts of stress (Weick, 1983). Resilience could support auditors to cope with these events. Smith and Emerson (2017) examined resilience in the auditing field and found that the higher the level of resilience the auditor possesses, the less the likelihood of engaging in QTB.
If workload is a crucial driver of QTB, this might also explain why QTB occurs more often within Big Four audit firms in comparison to non-Big Four audit firms, as pointed out in AFM (2014) research. In the study of de Vries and Herrijgers (2018, p.16) it is shown that Big Four auditors face higher volumes of workload than non-Big Four auditors. On average Big Four auditors work 5 hours a week more in comparison to non-Big Four auditors. One of the underlying factors that could affect the workload of auditors could be the relatively high staff outage within the profession (NBA, 2018). Prior research suggests that Big Four audit experience the highest outage of personnel (Hiltebeitel & Leauby, 2001). Francis (2011) states that audit quality is influenced by the audit firm the auditor works in, such as the organizational structure. Also, it has been shown that audit procedures are prematurely signed-off within Big Four audit firms to reduce the workload or time pressure (Otley & Pierce, 1996a; Herrbach, 2001). Not only academic research but also an investigation of regulatory agencies (AFM, 2014) have revealed the presence of QTB within Big Four audit firms. However, the question arises to which extent this behaviour occurs within Big Four audit firms in comparison with non-Big Four audit firms because of differences in workload. Overall, the purpose, but also the relevance, of this study is to gain knowledge of how the (differences in) workload of both Big Four and non-Big Four auditors affect QTB and how the relationship between workload and QTB is modified by resilience. Hence, the research question for this study is split up in four separate, but related questions. 1) How does workload increase QTB? 2) Is workload higher for Big Four auditors in comparison to non-Big Four auditors? 3) Do Big Four or non-Big Four auditors engage in higher levels of QTB because of higher workloads? And finally, 4) Do resilient auditors engage in less QTB when experiencing workload? In order to find the answer to these questions, 218 auditors from many different audit firms through the Netherlands participated in this study by filling in a web-based survey. The outcomes of this study have both academical as well as practical implications. QTB has been pointed out a subject of attention both by regulators (AFM, 2014) as well as by academia (Otley & Pierce, 1996a; Herrbach, 2001). However, QTB research mostly focused on Big Four auditors (Pierce & Sweeney, 2006), while this study also includes non-Big Four auditors. More specifically, the mediating effect of workload between Big Four and non-Big Four auditors on QTB has yet to be examined, as far as I know. Therefore, this study explores the differences between these groups regarding workload, possibly causing a higher level of engagement in QTB. Moreover, this study will also explore resilience, being a barely examined characteristic, for auditors, especially in relation to workload and QTB. Therefore it contributes to academic research (Sweeney & Pierce, 2004; Christensen et al., 2016; Svanström, 2016). If this characteristic appears to be relevant, it might have an impact on the hiring process of auditors. Finally, reducing QTB should be a primary focus for all audit firms, since it could prevent enormous costs for audit firms (i.e. new regulations).
In the next sections, the research will be elaborated. Firstly, the theoretical framework will be substantiated. Secondly, the methods of the research will be described. Thirdly, the results will be analysed. Finally, the conclusion is given, and discussion will be held.
4
Theoretical framework and hypothesis development
Every variable and its relationship will be theoretically substantiated in this chapter. Firstly, the importance of QTB and its connection with Audit Quality will be analysed. Secondly, the effect of workload on QTB will be explained. Thirdly, the difference of workload between Big Four & non-Big Four audit firms will be examined whereafter the mediation effect of workload on non-Big Four and non-Big Four auditors and QTB will be substantiated. In the fourth place, the moderating role of resilience will be described. Finally, all variables of interest are presented in the Conceptual Model of this study, which is shown in figure I.
Audit Quality Threatening Behaviour & Audit Quality
Auditor behaviours reducing the quality of an audit are a critical topic within the practice and audit research (Coram et al., 2004; Sweeney & Pierce, 2004; Herrbach, 2001; Otley & Pierce, 1996; McNair, 1991). These types of behaviours are known as Quality Threatening Behaviour (QTB). QTB is defined in literature as "poor execution of an audit procedure that reduces the level of evidence gathered for the audit so that the collected evidence is unreliable, false or inadequate quantitatively or qualitatively" (Herrbach, 2001, p.5). Examples of these types of behaviour are: not executing the audit procedures properly, false and/or premature sign-offs, failing to re-sample items when needed, inadequate audit trail documentation and lacking review of the audit files (Svanström, 2016; Sweeney & Pierce, 2006; Coram et al., 2004; Herrbach, 2004; McNair, 1991; Kelley & Margheim, 1990). The relevance of preventing QTB is to ensure Audit Quality. QTB results in a decreasing amount of audit work performed, and thus, less audit evidence is collected according to Johansen and Christoffersen (2017). Collecting less appropriate and sufficient audit evidence often result in a less secure view of the financial statements as a whole. By preventing QTB, future accounting scandals, litigation and more regulation could be prevented. Thus, not ensuring an adequate level of audit quality can bring about enormous additional and redundant costs.
When looking at the way QTB impacts audit quality, we must define audit quality first. The definition of audit quality mostly used and recognised has been conceived by DeAngelo (1981). Audit quality is defined as ‘’(a) discovering a breach in the accounting system, which is related to technological capabilities and (b) reporting the breach, which is related to independence’’ (DeAngelo, 1981, p186). Until now, most research has mainly focused on implementing criteria and regulations to ensure audit quality (Cameran et al., 2016; Beattie et al., 2013; Habib, 2012). However, the audit process is tailored to discover these breaches. Eventually, it comes down to the capabilities and competence of the individual auditor to collect sufficient and appropriate audit evidence (Knechel et al., 2013).
High audit quality is hard to define in literature due to the nature of auditing (i.e. its unobservability and its subjectiveness, cf. Knechel 2010). Therefore, literature often describes what high audit quality is not: Quality Threatening Behaviour (Knechel et al., 2013). The chance to discover a breach in an accounting system (a) is reduced by Quality Threatening Behaviour. This is caused by premature sign-offs for example, which result in less sufficient and less appropriate audit evidence collection. Eventually, a breach in the accounting systems may stay undiscovered because of this behaviour. QTB is not considered as high audit quality since it negatively affects the audit process and thus, the output of the audit (Knechel et al., 2013). Therefore, it can be suggested that QTB negatively affects audit quality. However, what causes auditors to engage in such type of behaviour? To find answers to this question, the relationship between workload and QTB will be examined in the first place.
5
The effect of workload on QTB
Workload can be defined as ‘’the perceived amount of work in terms of pace and volume’’ (Spector & Jex, 1998, p.3). Weick (1983) shows that auditing is often associated with a stressful profession because of long working days and high working pressure. In recent years, a high workload has been still very common for auditors. Especially during the busy season auditors often work 60 hours a week, which is considered as exhausting (López & Peters, 2012). It is acknowledged that additional work pressures arise from tight time deadlines and time budgeting, stimulating workload even more (Svanström, 2016). Another reason why auditors face higher workloads is the new laws and regulations for audit firms, which have to be implemented by regulatory agencies to improve audit quality and prevent future accounting scandals (AFM, 2014, 2017; NBA, 2016). Concluding, it is clear that auditing is a profession characterised by long working days and high working pressure, especially during the busy season (Sweeney & Summers, 2002; López & Peters, 2012; Persellin et al., 2014; de Vries & Herrijger, 2018).
De Vries & Herrijgers (2018) examined the workload that auditors experience in the Netherlands. Their results show that auditors work around 60 hours a week during the whole busy season. Persellin et al. (2014) explored the relation between auditors’ workload and audit quality. They identified the optimum threshold of workload after which audit quality would decline (i.e. the audit quality workload threshold). Their optimum threshold to deliver audit quality is between 58 and 60 hours a week. Unfortunately, auditors on all levels exceeded this threshold, which causes audit quality to decline.
Workload is recognised as one of the most critical job stressors for auditors. It has the potential to decrease focus, to cause loss of professional scepticism, to reduce the tolerance for ambiguity and to increase the number of errors made while working (Haskin, Baglioni, & Cooper, 1990). In general, negative consequences of high workload are reduced motivation for the job (Bowling, Alarcon, Bragg, & Hartman, 2015) and reduced performance of the employee him- or herself (Cordes & Dougherty, 1993). All of these factors could lower audit quality directly or indirectly.
Within auditing, a common occurrence is an auditors’ burnout, which is caused by the experience of structurally high workloads (Sweeney & Summers, 2002; Cordes & Doughterty, 1993). Burnouts result in the loss of employees due to being under too high workloads for a too long period. Burnout is known as a dysfunctional stress syndrome for auditors (Sweeney and Summer (2002) and is negatively associated with declines in job performance (Fogarty, Singh, Rhoads, & Moore, 2000). Therefore, it can be concluded that workload has significant negative consequences for the quality of auditing practices (Svanström, 2016; Persellin et al., 2014; Hyatt et al., 2013; Otley & Pierce, 1996a; McNair, 1991).
QTB is considered as one of the most significant adverse effects caused by high workload throughout auditing literature (Bowrin & King, 2010; Ettredge et al., 2008; Margheim et al., 2005; Otley & Pierce, 1996a). López and Peters (2012) examined the effects of high workload during the busy season and found a positive relationship with abnormal accruals. Their study suggests a lower audit quality on the audit team engagement level, which is a result of QTB. Auditors increasingly engage in QTB if high levels of workload are present (Coram, Ng, & Woodliff, 2004; McNamara & Liyanarachchi, 2008; Hyatt & Taylor, 2013). When auditors are exposed to high workload, inappropriate auditing practices such as falsely signing off audit procedures, biasing sample selection or accepting weak client acceptations are more likely to occur (Otley & Pierce, 1996a; DeZoort & Lord, 1997; Pierce & Sweeney, 2003). The reason auditors engage in this type of behaviour is to reduce their workload and therefore meet their deadline (Glover, Hansen & Seidel, 2018). Overall, these arguments point out that when auditors experience high workload, auditors are expected to engage in higher levels of QTB. Therefore, I hypothesise the following:
6
Do Big Four auditors experience higher workload than non-Big Four auditors?
High demands within the auditing profession can arise from different sources, such as audit firms' characteristics and specific characteristics of the auditing profession on its own. Before analysing the underlying factors of these high demands, it is essential to explore the question of whether Big Four auditors or non-Big Four auditors experience higher workloads. De Vries and Herrijgers (2018) examined different volumes of workload auditors face in the Netherlands during the busy season. The study found that Big Four auditors work for 63 hours a week in comparison to 58 hours a week for non-Big Four auditors. This is an average of 5 more working hours a week throughout the busy season for Big Four auditors. Coincidently, these results are in line with the academic study of Persellin et al. (2014). They also found that auditors of Big Four and mid-tier audit firms on average work more hours (66 and 64 hours respectively) during the busy season in comparison to auditors within smaller audit firms (56 hours). These results suggest that auditors working in Big Four audit firms experience higher levels of workload than auditors working in non-Big four audit firms.
Several underlying factors might explain why Big Four audit firms experience higher workloads than non-Big Four audit firms. Causes might be related to the work environment (Bowling & Kirkendall, 2012). Recognised factors in the work environment are organisational culture and varying patterns of work demand. Soeters & Schreuder (1988) examined cultural differences between US-oriented Big Four audit firms and Dutch audit firms. They concluded that culture differs significantly in uncertainty avoidance (i.e. formalisation in the organisation) and masculinity (i.e. preferring income and promotion over job security) for the US-oriented Big Four audit firms in comparison with Dutch audit firms. This reflects a more competitive culture and “up-or-out” mentality to become partner, caused by a strong hierarchical structure of Big Four audit firms (Jeppesen, 2007). A competitive climate is often perceived as stressful (Fletcher, 2008).
A more specific factor that causes a higher workload for audit firms is staff turnover, which is traditionally high in public accounting (Reinstein, Sinason, & Fogarty, 2012). Staff turnover can result in the same amount of work being distributed over fewer auditors, thus in a higher workload. Turnover is caused by auditors wanting to leave the profession. In recent years, the auditing industry has shown increasing numbers of auditors wanting to leave the profession (NBA, 2018). Research suggests that Big Four audit firms experience the largest staff outage within the first three years of employment (Hiltebeitel & Leauby, 2001; Chi, Hughen, Lin, & Lisic, 2013). Auditors’ turnover can result in a staff shortage and decreasing audit experience internally, which threatens audit quality (FSC, 2006; Vance & Stephens, 2010). This is another indicator that Big Four audit firms have to deal with high levels of staff turnover in comparison to non-Big Four audit firms, which causes workload to increase in general (Iqbal, 2010).
Besides that, when examining differences of workload between audit firms from regulation perspectives, it can be suggested that Big Four auditors have a higher workload due to increased regulations. The 2014 AFM report concluded that measures had to be implemented by Big Four audit firms to guarantee audit quality, such as additional procedures within the audit process (AFM, 2014, p. 118). The NBA (2016) expects that these measures result in an even higher workload for Big Four auditors. This is because the more intensive the audit procedures become (i.e. higher audit quality is guaranteed), the more time it takes for an auditor to complete them (Pierce & Sweeney, 2004). Therefore, Big Four audit firms may guarantee high audit quality by spending more time on audit procedures. More intensive audit procedures increase the workload the auditor experiences, which eventually could stimulate higher levels of QTB. Overall, this research expects that these elements could result in an even higher workload for Big Four auditors in comparison with non-Big Four auditors.
H2a: Big Four auditors (in comparison to non-Big Four auditors) have higher levels of workload than non-Big Four auditors.
7
To extend this line of reasoning, if higher workload results in higher levels of QTB (H1) and if Big Four auditors have a higher workload (H2a), Big Four auditors may very well engage in higher levels of QTB. These arguments lead to the following hypothesis:
H2b: Big Four auditors (in comparison to non-Big Four auditors) engage in higher levels of QTB because they have a higher workload.
The moderating effect of resilience on workload and QTB
QTB is also affected by many, still unknown, characteristics of auditors (Kelley & Margheim, 1990; Pierce & Sweeney, 2004; Gul, Wu, & Yang, 2013; Svanstorm, 2016). Individual characteristics of the auditor are perhaps the most important factor affecting audit quality (Christensen et al., 2016). It is acknowledged throughout literature that high workload frequently causes stress, which is often known as occupational stress (Baker, 1985; Akanji, 2013; Ahmad, Hussain, Saleem, Qureshi, & Mufti, 2015). In turn, regarding coping with stress, the characteristic of resilience is an important topic. Resilience is defined as ‘’the resistance to illness, adaptation, and thriving, the ability to bounce back or recover from stress’’ (Smith, Dalen, Wiggins, Tooley, Christopher, & Bernard, 2008, p.1). Positive effects of resilience are success and growth adaption, despite being affected by a stressful environment (Charney, 2004; Haddadi & Besharat, 2010). Resilience is becoming increasingly more important in behaviour science due to the increasing amount and presence of job stressors in different professions (Tusaie & Dyer, 2004). In-depth insights are needed how professionals experience and deal with causes of job stressors.
In the previous paragraphs, it has been argued that auditors experience high workload and that it leads to QTB. Also, workload is already identified as one of the leading job stressors in auditing literature (Haskins et al., 1990). Gaertner and Ruhe (1981) found that when an auditor experiences high workloads, there is a higher chance of failing to meet their job responsibilities since they suffer from stress or depression. Other research concludes that high levels of workload that need to be completed within limited time result in stress (PCAOB, 2014). Sanders, Fulks and Knoblett (1995) found that stress regarding workload can be both quantitative and qualitative. Regarding quantitative workload, stress is caused by the perception of the auditor, which is due to too high volumes of work within the given time. Concerning the qualitative workload, stress is caused by the job requirements of the auditors, which might exceed his competence level. Also, the investigations of the AFM, fines and accounting scandals are likely to result in even higher levels of stress.
In different professions, resilience has been examined in recent years. In elementary education, resilience is recognised as an important characteristic. This profession is also characterised by intense workloads and therefore high levels of stress. Resilience is found to reduce perceived role overload (i.e. perceived workload) for teachers (Richards, Levesque-Bristol, Templin, & Graber, 2016). In the nursing profession, resilience is considered as an essential quality to succeed, as it is also characterised by high workload and stress (Tusaie & Dyer, 2004). Among nurses, Lanz and Bruk-Lee (2017) examined how the relationship between workload and turnover intention is moderated by resilience and found that the moderating effect of resilience mitigates the relation between workload and turnover intention. Since turnover intention is positively related with QTB (Donnely et al., 2013) and the moderating effect is negative in the study of Lanz and Bruk-lee (2017), this research suggests that the moderating effect of resilience will mitigate the relationship between workload and QTB.
Within auditing, resilience has only been examined relating to stress. Smith and Emerson (2017) examined the effects of resilience on QTB. They found a negative and significant relationship between resilience and QTB. Therefore, the higher the level of resilience an auditor possesses, the more likely it is that engagement in QTB declines. Overall, resilience is expected to
8
mitigate the adverse effects of workload and therefore, the negative impact on QTB. This results in the last hypothesis:
H3: Auditor resilience moderates the relationship between auditor workload and QTB, such that this relationship is weaker for resilient auditors.
Figure I
Conceptual Model
Methodology
Sampling and Data Collection
The sample in this study consists of Big Four and non-Big Four auditors. Auditors were personally contacted with the request to fill out the questionnaire. The auditors were sampled by using convenience sampling so that the data is representative of the entire population (Etikan, Musa, & Alkassim, 2016). The sample mostly consists of auditors with at least two years of experience. The web-based survey was distributed through mailings. Between April 2019 and November 2019 239 auditors received the invitation link to the survey. One week after the initial e-mail was sent out, they received a first reminder, and a second reminder was sent out after two weeks. Of 239 auditors who initially agreed to participate, 218 (91%) completed the survey.
Survey instruments
A seven-point Likert scale structured the questions in the surveys relating to the used variables in this study. The seven-point Likert scale provides a more accurate view regarding the information given by the participants for surveys in comparison to the five-point Likert scale (Finstad, 2010). The scales of the questions within the survey were based upon scientific research and therefore, already validated (Otley & Pierce 1996; Specter & Jex, 1998; Smith et al., 2008). The scales of the specific questions that were used were already proven to provide appropriate information such as the reliability, validity and internal consistency of the constructs. The used questions can be found in Appendix A. In the personal message the respondents received before agreeing to participate, information regarding the purpose of the survey was given, including the expected time needed for completion. More in-depth information was given with a cover letter on the first page of the
web-9
based survey. Furthermore, confidentiality regarding the answers of the respondent was secured because the data of the survey was directed and managed by an independent researcher.
Dependent variable
Quality Threatening Behaviour
In this research, the dependent variable Quality Threatening Behaviour (QTB) was measured. To measure QTB, a group of eight questions that are substantiated by literature was asked to auditors (Otley & Pierce, 1996a). These questions are proven to measure QTB, and other studies relating to QTB also use this measure (Herrbach, 2001; Smith & Emerson, 2017). An example of a question is ‘’failing to complete the audit procedure such as signing off on an audit programme step without completing the work’’. The answer to these questions is given on a scale from 1 (never) to 7 (very often). The eight questions are computed into one new variable to measure (the mean of) QTB after testing for internal reliability (Cronbach’s alpha = .795).
Independent variables
Workload
Workload was measured with a validated workload scale (Spector & Jex, 1998). The used scale in this study to measure workload is also known as the Quantitative Workload Inventory scale (QWI). It measures the pace and volume of the perceived amount of work (Spector & Jex, 1998). How workload is perceived gives more insights into workload than just the amount of hours worked, such as how fast someone has to work. For the perception scale of how auditors experience the workload, five additional questions were asked. An example of one question is ‘’the time pressure someone experiences from its’ workload’’. The scale is from 1 (never) to 7 (very often). Other research also used this scale (Cote & Morgan, 2002; French, 2005). The five questions are computed into one variable to measure the mean of workload after testing for internal reliability (Cronbach’s alpha = .821).
Resilience
The measure that was used for resilience is based on the research of Smith et al. (2008). In this research, questions such as ‘’I have a hard time making it through stressful events’’ are asked. This is known as the brief resilience scale (BRS) and consists of 6 questions. The scale exists of 1 (strongly disagree) to 7 (strongly agree). In a methodological review of resilience measurement scales, the brief resilience scale received one of the best psychometric ratings and is there validated (Windle, Bennet, & Noyes, 2011). Furthermore, question four, five and six were reversed, because the questioning was negative instead of positive. Therefore, the answers to these three questions were recoded by 1 (strongly agree) to 7 (strongly disagree). Next, the six questions are computed into one variable to measure the mean of resilience after three questions were reversed, and after testing for internal reliability (Cronbach's alpha = .848).
Big Four auditors & non-Big Four auditors
A dummy variable is used to include Big Four and non-Big Four auditors. Big four audit firms are Deloitte, Ernst and Young, KPMG and Pricewaterhouse Coopers. The non-Big Four audit firms are all audit firms in the Netherlands except the previously mentioned four audit firms. Auditors from Big Four audit firms are assigned the value of 1 and auditors from non-Big Four audit firms the value of 0.
10
Control variables
Variables such as organisational identification (OrgID), age (AGE), gender (GEN), tenure (TEN) and seniority (SEN) are controlled. Previous research suggests that organisational commitment mitigates the effect of QTB (Otley & Pierce, 1996b). The authors suggest that when auditors have a high commitment towards their organisation, they will engage in QTB in lesser amounts. Organisational identification consists out of six questions which are validated by previous research (Bamber & Iyer, 2002). An example from a question is ‘’ The audit firm’s successes are my successes’’. The five questions are computed into one variable to measure the mean of Organizational identification after testing for internal reliability (Cronbach alpha = .880). Furthermore, auditors’ are controlled by their seniority and tenure. It is expected that the amount of experience in the audit affects quality threatening behaviour. For example, a junior auditor with 1-2 years of work experience has lesser experience with auditing in comparison with a senior auditor with 5+ years of work experience. Therefore, the chance that juniors auditors engage in QTB is higher (Pierce & Sweeney, 2006). Finally, I control auditors’ age and gender, which are regular control variables when studying behaviours such as workload and QTB (Otley & Pierce, 1996a; McNamara & Liyanarachchi, 2008).
Data Analysis
All data analytic steps were performed in SPSS 26. Hypothesis 1 and 2a were tested through simple linear regression analysis. For hypothesis 2b, the mediation effect was examined (Baron & Kenny, 1986). The PROCESS tool, model 4, of Hayes (2012) was used to test the mediation effect. Finally, the moderating effect of hypothesis 3 was tested by estimating a moderated OLS regression model (Aiken & West, 1991). To run the regression analysis, standardised values for resilience (Resilience) and Workload (WL) were created. Hereafter an interaction term (Resilience*Workload) was created by transforming the variables by multiplying the moderator times the independent variable. This was executed to test the moderating effect of resilience on the relationship between workload and Quality Threatening Behaviour.
Results
Demographic Sample Characteristics
Table 1 presents the details regarding the demographic characteristics of the survey respondents. The sample consists of 126 auditors from Big Four audit firms (57.3%) and 90 auditors from non-Big Four audit firms (41.3%). 158 auditors are male (72.5%) and 60 are female (27.5%). The tenure of respondents ranges from less than two years to more than 26 years. The average tenure is 7.04 year, and most respondents have more than three years of experience within an audit firm. Most auditors in the sample have a medior position or higher within their audit firm. Also, most auditors are under 30: 20.6% of the auditors are younger than 25 years, and 42.2% are between 26 and 30 years old. The sample size per group auditors is sufficient (>50) for organisational or social studies to give unbiased and accurate insights in multilevel modelling (Maas & Hox, 2005). Overall, the ratio between Big Four and non-Big Four auditors is distributed well. Also, since most auditors have more than two years of experience, it is expected that the auditors within the survey sample have enough working experience to form properly substantiated opinions about aspects such as QTB and how they perceive workload.
11
Table 1: Demographic sample characteristics
Descriptive statistics and correlation analysis
The descriptive statistics of the survey respondents are shown in Table 2. As already mentioned, 58.3% of the auditors are active in a Big Four audit firm. The average tenure of an auditor in this sample is 7.04 years (SD = 16.38). The average results of the amount of workload in hours (#WL) and workload (Workload) are 44.97 hours (SD = 6.14) and 4,92 (SD = .88), respectively. This indicates that auditors have above average workweeks and also perceive workload as relatively high. The mean of resilience (Resilience) is 5.05, which suggest that auditors are resilient. Furthermore, the mean of Quality Threatening Behaviour (QTB) is 2.53, which shows that auditors do not engage in QTB very often. Finally, the mean of Organizational Identification (OrgID) is 4.85, which suggest that auditors are above average committed towards their firm. In Table 2, the correlations between the dependent, independent and control variables are given. The correlation coefficients report the linear association between two variables and its significance (Taylor, 1990). Firstly, Workload (r = .28, p < .01) and QTB are significantly positively correlated, which indicates that when the workload increases, QTB also increases. Remarkable is that this effect is also positively correlated but not significant for the amount of workload. However, workload and the amount of workload correlate significantly (r = .29, p < .01) Furthermore, Big Four and non-Big Four auditors are positively significant correlated with both workload (r = .21, p < .01) as with the amount of workload (r = 0.30, p < .01).
Characteristic Frequency Percentage Characteristic Frequency Percentage
Big Four & non-Big Four auditors Gender
Big Four 126 58.3 Male 158 72.5
non-Big Four 90 41.7 Female 60 27.5
Tenure Seniority
< 2 years 32 14.7 Starter 4 1.8
3 -5 years 99 45.4 Junior 27 12.4
6 - 9 years 37 17.0 Medior 80 36.7
10 - 15 years 28 12.8 Senior 55 25.2
16 - 25 years 13 6.0 Very senior 25 11.5
26 > years 7 3.2 Partner 27 12.4
Characteristic Frequency Percentage Characteristic Frequency Percentage
Age
< 25 years 45 20.6 36 – 45 years 30 13.8
26 – 30 years 92 42.2 46 – 55 years 12 5.5
12
Table 2: Mean, Median, Standard Deviation and Correlation Analysis
**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
Regression analysis
After completing the standard procedures, regression analyses were executed to test the relationship between the variables. Firstly, heteroscedasticity was diagnosed to exclude a violation of the assumption of homoscedasticity when executing linear regressions (Cook & Weisberg; 1983). The linear regression was scatter plotted, with on the Y-axis *ZRESID and the X-axis *ZPRED. Little evidence for heteroscedasticity was found after plotting a scatterplot while running the regression analyses. In Table 4, the results of the regression analyses on QTB are shown. The table contains four models, one for each group of variables. The first model represents the standardised coefficient β and its significance of the control variables. In the second model, the main variables were added. In the third model, the moderator was added. Finally, information regarding the R², Δ R², F for change in R² and the variance inflation factors (VIF) were included for each model. The R² has low explanatory power (.13), which is in line with other behavioural models (Peterson et al., 1985). VIF shows in which level multi-collinearity is present within the (multiple) regression analysis. VIF values between 5 and 10 indicate excessive multicollinearity and cause for concern (Neter, Wasserman, & Kutner, 1989). Since the VIF values of the models range between 1.65 and 2.00, no multi-collinearity is present. Because the VIF values are low, the model has a high tolerance and reflects stable regression coefficients (O’brien, 2007).
1 Big Four auditor = 1, non-Big Four auditor = 0
Mean Median SD 1 2 3 4 5 6 7 8 9 QTB 2.53 2.375 .82 Workload 4.92 5.00 .88 .28** # WL 44.97 45.00 6.14 .05 .29** Big 4 / Non-big41 .58 1.00 .49 .02 .21** .30** Resilience 5.05 5.17 .94 -.11 -.11 -.01 -.00 OrgID 4.85 5.00 1.13 -.10 -.23** .12 -.06 -.02 GEN .28 .00 .45 .02 .08 -.23** .04 -.15* -.10 AGE 31.00 28.00 7.74 -.15* -.20** .17* -.04 .07 .24** -.17* TEN 8.10 5.00 16.3 -.09 -.12 .22** .17* .10 .27** -.14* .56** SEN 3.69 3.00 1.25 -.20** -.17* .21** -.03 .14* .21** -.17* .81** .61**
13
Table 3: Linear Regression Analyses on Quality Threatening Behaviour
N = 214, *p < .05, **p < .01, ***p < .001
Control variables
Model 1 represents the five control variable (gender, age, tenure, seniority and organisational identification). The most significant control variable is seniority in model 1 (β = -.27, p < .05). The significant and negative relation of seniority on QTB reflects the fact that the higher the function of the auditor, he or she lowers the chances of engaging in QTB.
The direct link with Workload on QTB
H1: Auditors’ workload is positively related to QTB.
In the second model, a simple linear regression was used to predict QTB from workload2. Other
main variables, such as Big Four and non-Big Four auditors and resilience, are also included. This model answers the first hypothesis. The model shows that workload has a significant positive effect
2 The scale of workload is relating to the perception of how auditors experience workload (cf. Appendix A).
Model 1 Model 2 Model 3
Variables β β β Intercept 3.28*** 2.17*** 2.17*** Control variables GEN -.02 -.03 -.02 AGE .06 .09 .07 TEN .05 .07 .06 SEN -.27* -.26* -.24* OrgID -.09 -.05 -.03 Main effects Big Four & non-Big Four -.08 -.09 Workload .26*** .25*** Resilience -.05 -.04 Two-way interaction Workload x Resilience .12 R2 .05 .12 .13 Δ R² .05 .07** .01 F for change in R² 2.19 3.37*** 3.36*** VIF 2.00 1.71 1.65
14
on QTB (β = .26, p < .001). This means that when an auditor experiences high workload, he or she will engage in higher levels of QTB. Next to the answer regarding hypothesis 1, other relevant information is given. Big Four and non-Big Four auditors are only negatively, but not significantly related to QTB (β = -.08, p = .35) on itself. Also, resilience negatively impacts QTB (β = -.05, p = .33). Overall, the evidence is found that proves workload does positively affect QTB so that hypothesis 1 is supported.
The direct link with Big Four auditors on Workload
H2a: Big Four auditors (in comparison to non-Big Four auditors) have higher levels of workload than non-Big Four auditors.
To test hypothesis 2a, another regression analysis was executed to measure the effect of Big Four and non-Big Four auditors on workload. The regression model is given in Table 4 below.
Table 4: Linear Regression Analyses on Workload
Model 1 Model 2 Variables β β Intercept 5.88*** 5.62*** Control variables GEN .04 .13 AGE -.16 .08 TEN .03 .06 SEN -.01 .08 OrgID -.19** .05** Main effects Big 4 / non-big 4 .12** Resilience .06 R2 .08 .12 Δ R² .08** .05** F for change in R² 3.46** 5.49*** VIF 2.00 1.77 N = 214, *p < .05, **p < .01, ***p < .001
In Table 4, information is given that the independent variable Big Four and non-Big Four auditors have a significant positive effect on the dependent variable workload (β = .12, p < .01). It can be stated that Big Four auditors, in comparison to non-Big Four auditors, experience higher levels of workload. Therefore, hypothesis 2a is supported.
15
Mediation effect of Workload on Big Four auditors and QTB
H2b: Big Four auditors (in comparison to non-Big Four auditors) engage in higher levels of QTB because they have a higher workload.
After hypotheses 1 and h2a are confirmed it can already be suggested that Big Four (in comparison to non-Big Four auditors) engage in higher levels of QTB because they experience higher workload. After executing the PROCESS model 4 of Hayes (2012) with 5000 bootstrap samples and a 99% confidence interval to test h2b, the mediation model (Table 5) shows the indirect mediation effect of workload between Big Four and non-Big Four auditors on QTB.
Table 5: Mediation analysis of the Indirect Effect with Big Four and non-Big Four auditors as Independent variable, Workload as Mediator and QTB as Dependent Variable. Variable Indirect effect SE 99% Confidence Interval
Lower boundary Upper boundary
Workload .09 .04 .01 .22
The output of the mediation from the PROCESS model of Hayes (2002) is the following. The a path (Big Four auditors predicting workload): b = .35, t(207)= 2.84, p = .005, 99% CI [0.03, 0.66]. The b path (workload predicting QTB): b = .25, t(206)= 3.82, p = .0002, 99% CI [0.08, 0.42]. The c path (Big Four auditors predicting QTB): b = -.02, t(207) = -.21, p = .83, 99% CI [-0.33, 0.28]. The result of the indirect effect of Big Four auditors on QTB, mediated by workload is: b = .09, SE = .04, 99% CI [0.01, 0.22]. The mediation model shows a significant and positive indirect effect of workload on Big Four and non-Big Four auditors in relation to QTB. Both the lower- and upper boundary is positive on a 99% confidence interval. Interestingly, these results suggest that when Big Four auditors experience higher workload, they engage in higher levels of QTB in comparison to non-Big Four auditors (99% CI [.01, .22]). Thus, I find support for hypothesis 2b. When examining the direct non-significant and the indirect significant effect in the mediation model, it can be stated that the mediation effect is a partial mediation effect (Baron & Kenny, 1986). The direct effect is reduced when mediating with workload.
Moderating effect of Resilience on Workload and QTB
H3: Auditor resilience moderates the relationship between auditor workload and QTB, such that this relationship is weaker for resilient auditors.
Finally, hypothesis 3 was tested. Model 3 in Table 3, the moderating effect of resilience (Workload x Resilience) on workload and QTB is shown (β = .12, p > .05), which is positive and non-significant. Surprisingly, the results suggest that auditors’ resilience moderates the relationship between auditor workload and QTB, such that this relationship is stronger for resilient auditors. This means that resilient auditors engage in even higher levels of QTB when experiencing high workload. Therefore, support for hypothesis 3 was not found.
Robustness checks
The first robustness check applied was dummy coding (MacCallum, Zhang, Preacher & Rucker, 2002). A dichotomised dummy variable of the independent variable workload was made. Firstly, the independent variable workload was dichotomised at its median of 5.00 by recoding it. Auditors who had a score between 0 to 5.00 were recoded with a ‘’0’’ (i.e. low workload), and auditors with a score between 5.001 to 7 were recoded with a ‘’1’’ (i.e. high workload). After these steps, another
16
linear regression was executed with the dummy variable of workload on QTB. Both the coefficient and the significance of the dummy variable declined slightly (β = .19, p = .008). Also, another regression model was to robustness check hypothesis 2a. The effect of the independent variable Big Four and non-Big Four auditors had a significant positive effect on the (dichotomous) dependent variable workload (β = .244, p = .000). Unfortunately, when attempting to run a mediation analysis for h2b with the PROCESS model, it gave an error that the PROCESS model does not allow dichotomous mediators.
For a second robustness check, the effect coding method was applied (Alkharusi, 2012). Instead of splitting the variable at its median into two groups, now three groups were computed. The first group of auditors received ‘’-1’’ when their score was between 0 to 4.999. The second group of auditors were recoded with a ‘’0’’ when their score was 5. Finally, auditors with a score between 5.001 to 7 were recoded with a ‘’1’’. After these steps, another linear regression was executed with the dummy variable of workload on QTB. The coefficient of the dummy variable also declined, and significance slightly reduced (β = .20, p = .005). Finally, another regression model was executed to robustness check hypothesis 2a. The effect of the independent variable Big Four and non-Big Four auditors had a significant positive effect on the (trichotomous) dependent variable workload (β = .226, p = .001). Fortunately, the PROCESS model could run this kind of mediator. The mediation effect of the variable workload is the following.
Table 6: Robustness Check of the Mediation analysis of the Indirect Effect with Big Four and non-Big Four auditors as Independent variable, Workload (effect coding method) as
Mediator and QTB as Dependent Variable.
The a path (Big Four auditors predicting workload): b = .44, t(207)= 3,34, p = .001, 99% CI[0.09, 0.79]. The b path (Workload predicting QTB): b = .17, t(206)= 2,90, p = .004, 99% CI [0.02, 0.33]. The c path (Big Four auditors predicting QTB): b= -.02, t(207) = -.21, p = .83, 99% CI [-0.33, 0.28]. The indirect effect of Big Four auditors on QTB: b = .08, SE = .04, 99% CI [0.01, 0.19]. These results of the PROCESS model were similar with the initial results. Therefore, I can conclude that the results of this study are robust.
Additional analysis
Results separated by Big Four and non-Big Four auditors
As an additional analysis, the data was separated by isolating both Big Four and non-Big Four auditors. Firstly, the descriptives and correlation analysis for each separate group are given for explanatory information (Appendix B1). The goal is to flag potential differences between the groups and also for possible comparison with the descriptives and correlations of the total sample (Table 2). The most remarkable differences between Big Four and non-Big Four auditors is regarding the variables workload (Mean = 5.02 and 4.70 respectively) and the amount of workload ( Mean = 46.51 and 42.72 hours a week respectively). Therefore, it can be noted that Big Four auditors reported higher workload, in both terms of volume and pace. Auditors report similar resilience means and medians. The mean of QTB is also quite similar for both groups of auditors. Finally, it is noticeable that the means of organisational identification and seniority of non-Big Four auditors are higher. This indicates that non-Big Four auditors are more identified with their firms and are also longer active within the auditing industry in comparison to Big Four auditors.
Variable Indirect effect SE 99% Confidence Interval
Lower boundary Upper boundary
17
Next, two regression analyses were run to separate Big Four and non-Big Four auditors. In one regression analysis, all Big Four auditors from the sample were included. In the other regression analysis, only non-Big Four auditors from the sample were included. The results are recorded in Appendix B2. When comparing and analysing the results of the separate regression analyses regarding Big Four and non-Big Four auditors a few changes between variables attract attention. Regarding Big Four auditors, in model 1, age becomes significant. In model 2 and 3, almost no changes occur in comparison to the original regression output in Table 4. Regarding non-Big Four auditors, the most surprising changes in model 3 were that workload became non-significant and therefore, it can be stated that workload has a bigger impact on Big Four auditors than for non-Big Four auditors. Reason is that the more positive coefficient (β = .26 and β = .23 respectively) and workloads ‘significance is higher (p = .002 and p = .028 respectively). Furthermore, another difference between Big Four and non-Big Four auditors is the change of the moderating effect of resilience (β = .27, p < .05). Since the moderating effect is significant for non-Big Four auditors, it is shown in figure II. The coefficients of the independent variable (Workload), moderator (Resilience) and the interaction term (Workload * Resilience) are plotted in the figure below.
Figure II
The plot of the Moderating Effect of Resilience on non-Big Four
Auditors
The figure shows that highly resilient auditors engage in higher levels of QTB when experiencing high workload.
18
Changing Workload with the Amount of Workload in all models
As a second additional analysis, the workload (perception) variable was changed to the amount of workload (volume) variable in all models. The regression table is included in Appendix C. This change was made because workload is defined as both the pace and volume of the amount of work. The variable workload indicates the pace of workload, while the amount of workload gives information regarding the volume. Also, both variables correlate significantly with each other (Table 2). Therefore, I suggest that the amount of working hours could give additional insights regarding the effects of workload since it differs from the perceived workload. For example, one auditor can experience ‘’hard-working’’ at 50 hours a week, while another auditor experiences ‘’hard-working’’ at 60 hours a week, which could differently affect QTB.
Changing the variable workload with the amount of workload to check if this results in changes in the regression model has led to additional insights. This changes will be given for each hypothesis. Regarding hypothesis 1, the amount of workload is positively related to QTB, but not significant. The effect of resilience on QTB also did not change, which is still negative and non-significant. When testing the effect of Big Four and non-Big Four auditors on the amount of workload (h2a), the results became more significant (β = .316 p < .001) regarding the amount of workload. Therefore, the relationship is more significant (p = .000 in comparison to p = .005) and also the coefficient (β = .316 in comparison to .192) is higher. When testing the (indirect) mediation effect of the amount of workload on Big Four auditors and QTB, the results showed that the mediation effect declined. The moderating effect of hypothesis 3 slightly changed to a more neutral effect, but overall it was still positive and not significant.
Conclusion and Discussion
In recent years many studies regarding audit quality have been elaborated. Academic researchers suggest that audit firm size is positively related to audit quality (DeAngelo, 1981), while regulators found that the audit quality is below standards within Big Four audit firms (AFM, 2014; AFM, 2017). Practitioners state that the audit quality is affected by external factors, such as high workloads, which increases even more because of the measures to guarantee audit quality imposed by regulators (NBA, 2016). High audit quality is hard to determine because of its unobservability and subjectiveness (Knechel, 2010). However, low audit quality is easier to examine through Quality Threatening Behaviour (Knechel et al., 2013). Besides, recent studies focused on the effects of measures and regulations to improve audit quality (Cameran et al., 2016; Beattie et al., 2014; Habib, 2012). Therefore, the goal of this study was to focus on the individual auditors who eventually perform the audit and ensure the audit quality (threatening behaviour). Individual auditor behaviour was investigated to examine if workload affects the level of engagement in quality threatening behaviour and if this relationship would be mitigated by resilience. Also, by comparing workloads of Big Four or non-Big Four auditors in this study, new insights are given to the questions which auditors and why they engage in higher levels of QTB when high workload is experienced.
I expected that workload positively affects QTB, Big Four auditors have higher workloads in comparison to non-Big Four auditors, that when Big Four auditors experience higher workloads, they engage in higher levels of QTB (in comparison to non-Big Four auditors) and that resilience weakens the relationship between workload and QTB. After 218 auditors participated in this web-based survey, I found that the results of regression analyses supported three of the four hypotheses. Workload is indeed positive and also significantly related to QTB. Next, a significant positive relationship was also found for Big Four auditors, who experience higher workloads than non-Big Four auditors. The mediating effect was also confirmed: Big Four auditors engage in higher levels of QTB because they have higher workloads in comparison to non-Big Four auditors. Surprisingly,
19
hypothesis 3 was not confirmed since resilience does not weaken the relationship between workload and QTB.
More fully, first, I found that workload is significant and positive related to Quality Threatening Behaviour. This result suggests that when workload of auditors increases, the auditor also engages in higher levels of QTB. Quality Threatening Behaviour reduces audit quality (i.e. Quality Threatening Behaviour equals low audit quality). Therefore, it can be stated that a higher workload is a significant impediment for auditors and their audit quality. This result is in line with other studies involving workload and QTB (Pierce & Otley, 1996a; Summers & Sweeney, 2002; López & Peters, 2002).
Secondly, I found that the relation between Big Four auditors and workload is positive and is also significant. This means that Big Four auditors experience higher workloads than non-Big Four auditors. This result is in line with the studies of Persellin et al. (2014) and de Vries and Herrijgers (2018) as expected beforehand. However, within these studies, the relationship between Big Four or non-Big Four auditors and workload was not (statistically) investigated. Therefore, this study gives additional insights by proving the significant positive relation between Big Four auditors and workload.
Thirdly, I found, using the mediation model of Hayes (2012), that the indirect relation between Big Four auditors and QTB is significant and positive. Therefore, it can be stated that Big Four auditors engage in higher levels of QTB because they experience higher workloads in comparison to non-Big Four auditors. This is new evidence in auditing literature since this mediation effect of workload on Big Four and non-Big Four auditors and QTB was not yet examined (as far as I know). Concluding that, this research proves and emphasises the negative consequences of high workloads for (Big Four) auditors.
Fourthly, surprisingly I found a positive and non-significant moderating effect of resilience between workload and QTB. This indicates that when auditors are resilient, they engage in higher levels of QTB when experiencing higher workloads, which is counter-intuitive. Next to the moderating effect of resilience, resilience by itself is negatively non-significantly related to QTB in both models 2 and 3. This suggests that resilience weakens QTB slightly. The results of this study regarding resilience differ from the study of Smiths and Emerson (2017), which suggest that resilience is both negatively and significantly related to QTB. These findings provide opportunities to explore resilience further and with different constructs in future research.
Next to the initial study, two additional analyses were executed regarding the same theoretical model. Firstly, by separating Big Four and non-Big Four auditors in the regression analysis, differences in the relationships of variables between both groups were shown. One surprising result is that the resilience gives opposite results for Big Four and non-Big Four auditors (Appendix B2), which indicate that resilience could be more important for Big Four auditors than for non-Big Four auditors. The fact that these results are opposite is quite remarkable. I expect that these opposite results arise due to the limitations of this study (see the last limitation). Another result is that the moderating effect of resilience between workload and QTB turned significant for non-Big Four auditors. Since the moderating effect of resilience strengthens the relationship between workload and QTB in all cases, it can be stated that the moderating effect of resilience is not important for auditors to reduce QTB whenever high workload is present.
Secondly, to give additional insights into the relation of workload on QTB, the perception scale of workload (pace) is replaced by the amount of workload (volume) in the theoretical model. After running the regression analyses with the amount of workload instead of workload on QTB, all relations declined. Therefore, it can be stated that the amount of workload is less informative and provides a less complete view regarding QTB in comparison with workload.
20
Implications
The findings of this study benefit academic research. High workload is a significant and structural problem within the auditing profession, and therefore an in-depth understanding of the causes and consequences is crucial. Previous research has explored the topics of workload and QTB. For example, de Vries & Herreijger (2018), provided an overview of the workload and the negative consequences, such as QTB, whereas Persellin et al. (2014) examined the effects of high workload on audit quality. However, both studies have not (statistically) researched the relationship between workload and QTB. Furthermore, both studies distinguish workloads of Big Four and non-Big four auditors, but not which group experiences more negative consequences of workload and which underlying factors are responsible for this difference.
Therefore, one significant implication of this study is that new and more in-depth insights are given in the differences in QTB, caused by workload, of Big Four and non-Big Four auditors and in the underlying drivers behind this difference. In this study, the most critical underlying factors that cause a higher workload for Big Four auditors are the organisational structure, higher turnover and regulatory measures. The findings of this study enrich auditing literature and provide suggestions for new research for both academia and regulators. According to DeAngelo (1981), audit firm size positively affects audit quality. But what are the implications when Big Four auditors structurally experience too high workloads causing QTB? Is there a possibility that Big Four auditors structurally engage in QTB due to this kind of reasons?
Workload is one of the leading causes of stress, resilience being an important mitigator for stress. Even though this study shows that resilience directly mitigates QTB, it also indicates that resilience stimulates QTB when auditors experience high workloads, which is counter-intuitive. Therefore, resilience should be further investigated in combination with different constructs, such as turnover (intention) or stress. The reason is that turnover and stress are prevalent and structural problems within the auditing profession and that resilience could mitigate these problems (Weick, 1983; Reinstein et al., 2012; Smith & Emerson, 2017; Lanz & Bruk-Lee, 2017).
The outcomes of this study offer substantial implications for audit firm managers. Due to the current and structural high workloads, the audit quality of audit firms is endangered because of increasing engagement in QTB. Increasingly engagement in QTB negatively affects audit quality, which could have negative consequences for the reputation of the auditing profession. The costs of QTB could be enormous, caused by, e.g. reputation loss, more regulatory measures or even accounting scandals. To illustrate, when QTB increases significantly during the busy season, auditors may be failing to detect breaches in the accounting systems of clients. The results of this study emphasise the necessity to reduce workload within the auditing profession to prevent QTB, especially at Big Four audit firms. All audit firms should monitor periodically how their auditors perceive the workload, especially during the busy season, to prevent their auditors engaging in QTB caused by too high workloads. This monitoring could be executed by analysing registered hours of audit staff and perform in-depth data analytics to signal potential cases of QTB prematurely (i.e. when the audit quality workload threshold is surpassed). This could prevent engagement in QTB and stimulate audit quality in return.
This study also has implications for regulators. Most recently, on 14 January 2020, the Monitoring Commission Accountancy (MCA) reported structural problems (i.e. Wicked Problems) in the auditing profession, such as quality defects due to a performance gap (MCA, 2020). Technical aspects are suggested to be the problem, while in reality, high workload is the root cause. New mandatory changes are announced through legislation and regulation by the commission. It is therefore strongly advised that regulators should assess the impact of their announced regulatory measures on workload before implementing them. This is important to prevent adverse effects of regulations. The goal of regulators is to ensure audit quality by implementing measures (AFM, 2014, 2017), but if those measures increase workload (NBA, 2016), the implementation can cause QTB and have an opposite negative effect in turn.
