International Journal of Economics and Financial

Issues
ISSN: 2146-4138

available at http: www.econjournals.com

International Journal of Economics and Financial Issues, 2024, 14(5), 220-232.

Audit Quality and Financial Statement Manipulation: The

Moderating Effect of Tone at the Top

Alastair Marais*

School of Accounting, Economics and Finance, University of KwaZulu-Natal, South Africa. *Email: [email protected]

Received: 10 March 2024 Accepted: 21 July 2024 DOI: https://doi.org/10.32479/ijefi.16409

ABSTRACT
South Africa has recently fallen victim to numerous major corporate frauds. This has called into question South Africa’s audit quality, as indicated
by its loss of the number one place for auditing and reporting standard strength, according to the World Economic Forum. This study examined the
moderating effect of tone at the top on the relationship between audit quality and financial statement manipulation. Using a South African sample
of 829 firm-year observations from 2011 until 2018, principal component analysis was used to construct two audit quality variables (competence
and independence) and two tone at the top variables (autocratic and pragmatic). Financial statement manipulation was measured using fraud and
discretionary accruals. The findings revealed that while an autocratic tone strengthened the negative relationship between audit quality and fraud, it
also resulted in higher income-increasing discretionary accruals. This study contributes to the audit quality literature in South Africa by revealing
how the tone at the top can affect the auditor’s ability to conduct a quality audit. The results will interest auditors and shareholders seeking to reduce
fraud and earnings management.
Keywords: Audit Quality, Financial Statement Manipulation, Discretionary Accruals, Fraud, Tone at the Top, South Africa
JEL Classifications: G340, M410, M420

1. INTRODUCTION audit quality (DeFond and Zhang, 2014). Despite these reforms,
South Africa has recently been the victim of major corporate
The audit function aims to ensure that financial statements are frauds such as Steinhoff and Tongaat-Hulett. Consequently, as a
free from material misstatement and error (DeFond and Zhang, result of the governance reforms and continued corporate frauds
2014). However, numerous corporate scandals (such as Enron and globally, there has been a surge in academic research focusing on
Worldcom) have occurred despite this fundamental role. Various audit quality (Baatwah et al., 2018).
regulatory reforms have been introduced to address this issue, of
which the Sarbanes-Oxley Act (SOX) in the United States (US) Audit quality is the auditor’s ability to identify and report on
is one of the most well-known and influential worldwide. SOX manipulations of financial statements (DeAngelo, 1981). Prior
marked a shift from a self-regulating audit profession in the US to literature identifies two auditor attributes required to produce high-
a government-regulated one. Similarly, South Africa has recently quality audits: Competence and independence (Harber and Marx,
been affected by large corporate fraud (such as Tongaat-Hulett 2020). Competence is the auditor’s ability to identify violations and
and Steinhoff). South Africa implemented the new Companies is commonly proxied by auditor type (Yasser and Soliman, 2018),
Act, No. 71 of 2008 (effective 2011) and developed a series industry specialisation (Baatwah et al., 2018) and joint auditors
of world-renowned governance codes (referred to as the King (Piot and Janin, 2005). Independence relates to reporting such
codes) (Mokoaleli-Mokoteli and Iatridis, 2017). These regulatory violations and proxies include auditor size (Mokoaleli-Mokoteli
reforms have reshaped the audit environment, intending to improve and Iatridis, 2017), auditor tenure (Mukhlasin, 2018), and the

This Journal is licensed under a Creative Commons Attribution 4.0 International License

220 International Journal of Economics and Financial Issues | Vol 14 • Issue 5 • 2024
 Marais: Audit Quality and Financial Statement Manipulation: The Moderating Effect of Tone at the Top

provision of non-audit services (Hohenfels and Quick, 2020). (PCA). My results identify two composite audit quality variables:
Despite extensive research, empirical evidence on the relationship competence and independence. While competence does not reveal
between audit quality and financial statement manipulation (FSM) any association with FSM, independence is associated with larger
remains mixed. This may be because the proxies are considered absolute discretionary accruals driven by larger income-decreasing
individually. The individual variables may reflect different aspects accruals. A second contribution of the study is its addition to the
of audit quality, providing an incomplete picture (Singh et al., literature by considering the moderating effect of tone at the top
2019). Therefore, a composite variable may better represent the on the relationship between audit quality and FSM. Bicudo de
construct by considering the orthogonal relationships between the Castro et al. (2019) note that the impact of tone on auditor risk
proxies (Sun and Cahan, 2009). assessments requires further exploration, while Makhlouf et al.
(2021) call for including moderator variables within studies. My
Additionally, mixed evidence may result from failing to results indicate that the interaction between an autocratic tone and
consider the moderating effect of tone at the top, which is the auditor competence is negatively associated with fraud. However,
foundation of a company’s internal control environment (Garrett this interaction is associated with more income-increasing
et al., 2021). Management may manipulate the extent and discretionary accruals. Finally, the study corroborates prior
quality of the information provided to the auditors (Lisic et al., literature by showing that tone at the top improves the ability to
2016). Therefore, auditing standards require auditors to assess explain financial statement manipulation.
management characteristics such as tone at the top (Greiner et al.,
2020). An ethical tone would enable the auditor to modify audit The remainder of this study is arranged as follows. The next
procedures nature, timing and extent and rely more on management section presents the literature review. Herein, I define FSM before
representations (Garrett et al., 2021). When assessing tone at reviewing the prior evidence around the audit quality proxies.
the top, Schmidt (2014) notes three biases which work against Following this, I explain the research methodology, present the
auditors. Firstly, prior mental representations of management results and draw conclusions.
(which are more likely to be positive) will impact current
assessments and audit judgements. Secondly, the tone at the top
evidence is more likely to reflect favourably on management rather 2. LITERATURE REVIEW
than unfavourably. Thirdly, audit procedures are designed to find
positive rather than negative evidence. Despite these concerns, 2.1. Defining Financial Statement Manipulation
Zengin-Karaibrahimoglu et al. (2021) showed that auditors Financial statement fraud is “a deliberate attempt by corporations
adjust their risk assessments for top management narcissism. to deceive or mislead users of published financial statements,
However, the tone at the top’s impact on auditor competence and especially investors and creditors, by preparing and disseminating
independence is yet to be explored. materially misstated financial statements” (Rezaee, 2005,
p. 279). Financial statement fraud violates acceptable accounting
This study investigates the moderating effect of tone at the top on frameworks (Dechow and Skinner, 2000). Earnings management
the relationship between audit quality and FSM for non-financial occurs when management uses the flexibility allowed within
companies in South Africa. South Africa poses a unique context the accounting frameworks to mislead financial statement users
for this investigation. Firstly, because as a developing country (Dechow and Skinner, 2000). Thus, both financial statement
with a dual economy, the country shares aspects of both developed fraud and earnings management are intended to deceive the user,
and developing countries (Wesson, 2021). With their predominant but fraud violates the accounting frameworks while earnings
focus on developed economies, existing studies on audit quality management does not. I broadly define FSM to incorporate both
or tone are not generalizable to developing economies with lower concepts.
investor protection (Bicudo de Castro et al., 2019; Makhlouf et al.,
2021). Secondly, South Africa held the top position for auditing 2.2. Agency Theory and Audit Quality
and reporting standard strength from 2010 to 2016 (Wesson, 2021), The separation of ownership and control creates an agency
a ranking that subsequently dropped to 30th following revelations relationship between the shareholder and management (Jensen
of various economic crimes (Mnguni and Subban, 2022). and Meckling, 1976). This relationship creates information
Thirdly, South African literature on auditor quality is limited, asymmetries, allowing management to act in their own best
predominantly due to a lack of data availability (Wesson, 2021). interests (Makhlouf et al., 2021). Consequently, agency costs are
Recent audit quality research has focused on the Independent incurred to align management and shareholder interests. Credible
Regulatory Board for Auditor’s decision to enforce mandatory financial statements reduce information asymmetry, allowing
audit firm rotation (see, for example, Harber and Marx, 2020). shareholders to monitor management (DeFond and Zhang, 2014).
Consequently, this is the first study to investigate the moderating However, they require independent assurance of these financial
effect of tone at the top on audit quality in South Africa. statements, generally provided by external auditors (Mukhlasin,
2018). A high-quality audit, therefore, mitigates opportunistic
This study contributes to the literature in several ways. Firstly, it FSM (Lin and Hwang, 2010). The provision of a high-quality
extends the research on audit quality effect on FSM by considering audit depends on two key auditor attributes discussed below:
a comprehensive set of audit quality measures. Unlike prior competence and independence (Harber and Marx, 2020). Given
research, I built composite variables to account for the relationships the large volume of literature, for brevity, I focus only on selected
between individual proxies using principal component analysis studies.

International Journal of Economics and Financial Issues | Vol 14 • Issue 5 • 2024 221
 Marais: Audit Quality and Financial Statement Manipulation: The Moderating Effect of Tone at the Top

2.2.1. Auditor competence firms. As discussed under auditor type, the empirical evidence on
Auditor competence reflects the auditor’s ability to detect FSM. auditor size is mixed. The literature does not attempt to disentangle
As such, I hypothesise that: the Big N effects between competence and independence (DeFond
H1: Greater auditor competence is associated with lower FSM. and Zhang, 2014).

Auditor competence is proxied using auditor type, industry Audit tenure has become increasingly topical, with several
specialisation and joint auditors. Auditor type is commonly countries introducing mandatory audit firm rotation. One argument
proxied by Big N membership (DeFond and Zhang, 2014). The is that longer audit tenure increases familiarity with the client,
Big N companies, namely Deloitte, EY, KPMG and PWC, are reducing independence (Harber and Marx, 2020). Alternatively,
larger in size and face greater litigation risk. Consequently, they longer audit tenure improves client knowledge, increasing audit
attract greater accounting expertise, have greater exposure to quality (Nugrahanti and Puspitasari, 2018). The empirical evidence
different industries, offer better training and have more resources to on tenure is mixed. Al-Thuneibat et al. (2011) found that longer
monitor audit quality, resulting in higher-quality audits (Baatwah tenure resulted in higher FSM levels. Conversely, El Guindy
et al., 2018). The empirical evidence, however, is mixed. In line and Basuony (2018) and Mukhlasin (2018) found that longer
with expectations, Mokoaleli-Mokoteli and Iatridis (2017) and tenure reduced FSM. Finally, Özcan (2019) and Nugrahanti and
Makhlouf et al. (2021) found that auditor type was associated with Puspitasari (2018) found no association between audit tenure and
lower FSM. However, studies conducted in countries with lower FSM levels.
investor protection found that auditor type was associated with
higher FSM (such as Baatwah et al., 2018; and Bala et al., 2019), Non-audit services threaten independence in two ways. In the
while others found no association between auditor type and FSM first, the auditor may end up making decisions for the client. In
(Magnis and Iatridis, 2017; Selahudin et al., 2018; Sellami and the second, the auditor may become financially dependent on
Slimi, 2016; Yasser and Soliman, 2018). the client (DeFond and Zhang, 2014). Conversely, non-audit
services provide the auditor with greater insight into the client’s
Industry specialists have greater knowledge of a particular operations, allowing for higher-quality audits (DeFond and
industry and its accounting practices (Baatwah et al., 2018) and Zhang, 2014). Hohenfels and Quick (2020) found that non-audit
have more reputational capital at stake. Therefore, they should services increased FSM levels, whereas Svanström (2013) found
provide higher-quality audits (DeFond and Zhang, 2014). While that non-audit services reduced the extent of FSM. It may be that
the empirical evidence is mixed, it favours industry specialisation different non-audit services affect the FSM level differently. In
being associated with lower FSM levels (Balsam et al., 2003; further analysis, Hohenfels and Quick (2020) found that other
Inaam and Khamoussi, 2016; Krishnan, 2003; Mukhlasin, 2018). assurance and consultancy services increased FSM, while tax
However, other studies find no such relationship (Yasser and services had no effect.
Soliman, 2018), or that industry specialisation results in more
FSM (Baatwah et al., 2018). 2.3. Tone at the Top, Audit Quality and Financial
Statement Manipulation
Finally, joint auditors offer reciprocal checks on each other’s Tone at the top is the ethical culture a company’s top management
diligence, improving competence (Piot and Janin, 2005). Collusion creates through their language and actions. It forms the foundation of
between the auditor and management is also less likely to occur, a company’s internal control system (Schmidt, 2014) and influences
increasing independence (Marmousez, 2009). Few studies consider employees’ ethical behaviour (Noviyanti and Winata, 2015). Tone at
joint audits due to their rarity in practice. Marmousez (2009) found the top is apparent in top management language, as it enables them to
that, in France, joint auditors increased FSM. She attributed this create a company’s identity and manage others’ expectations (Patelli
to one auditor’s over-reliance on the other’s work, thus reducing and Pedrini, 2015). Consequently, tone at the top may be conveyed
effort. Alternatively, Bisogno and De Luca (2016) and Zerni et al. through the CEO and chairperson’s letters in the financial statements,
(2012) found that Italian and Swedish companies that voluntarily financial press appearances, and other similar sources (Amernic et
engaged joint auditors had less FSM. al., 2010). Thus, linguistic analysis of these communication types
provides insight into a company’s tone at the top.
2.2.2. Auditor independence
Auditor independence relates to reporting FSM. Consequently, I Amernic et al. (2010) investigated linguistic analysis in measuring
hypothesise that: the tone at the top. Given the severity of recent corporate frauds,
H2: Greater auditor independence is associated with lower FSM. they noted that a more holistic approach to understanding financial
information is necessary, with a greater appreciation of top
Common proxies for auditor independence include auditor management narratives. Following this, increased attention has
size, auditor tenure and the provision of non-audit services. been given to the linguistic analysis of top management narratives
Joint auditors, discussed earlier, also enhance the auditor’s and their relationship to FSM.
independence.
Early studies relied on human coders to perform linguistic
Auditor size, like auditor type, is proxied as Big N membership. analysis. While this has greater precision, it is hampered by
Big N firms have a larger customer base and more financial smaller samples and low replicability (Li, 2010). Technological
resources, resulting in lower dependence on one client than smaller advances have enabled automated approaches. These can be either

222 International Journal of Economics and Financial Issues | Vol 14 • Issue 5 • 2024
 Marais: Audit Quality and Financial Statement Manipulation: The Moderating Effect of Tone at the Top

statistical (such as Naïve Bayesian algorithms used by Li, 2010) 2018, representing 1673 firm years. Financial companies were
or dictionary-based. Initially, the statistical approach was superior, excluded due to different regulations impacting that industry
as no dictionary was explicitly developed for the corporate setting (Makhlouf et al., 2021). The 2011 year represents the 1st year
(Li, 2010). Since then, however, domain-specific wordlists (such that King III was effective in South Africa, ensuring adequate
as Henry, 2008; Loughran and McDonald, 2011) have been disclosure. The sample ended in 2018 to ensure sufficient time
developed and successfully used along with general wordlists for regulators to identify and publish enforcement actions against
such as Harvard General Inquirer (GI), Linguistic Inquiry and fraudulent companies. Karpoff et al. (2017) noted the median time
Word Count (LIWC) and DICTION. Consequently, the dictionary from fraud to first enforcement account in the US was 2.41 years.
approach has gained favour. Given that regulators in developing economies are not as advanced
and well-resourced as their developed country counterparts (Rabin,
Hope and Wang (2018) and Larcker and Zakolyukina (2012) 2016), 5 years (2019 to 2023) was deemed acceptable. This,
applied the LIWC dictionary in the US context. The DICTION however, is shorter than the period that Steinhoff and Tongaat-
software was used in Australia by Alshorman (2016) and in the US Hulett (two companies found guilty of FSM in South Africa) were
by Patelli and Pedrini (2015). Abou-El-Sood and El-Sayed (2022) identified as having committed FSM before they were discovered,
applied Henry’s (2008) wordlist in the Middle East and North Africa being nine and 8 years, respectively.
region. Finally, Huang et al. (2018) and Kayed and Meqbal (2024)
applied the Loughran and McDonald (2011) wordlist to companies I removed 52 firm years due to year-end changes and a further
in the US, UK and US-listed Chinese companies, respectively. 488 firm years that did not contain separate CEO statements. Not
All these studies found that linguistic analysis of top management all companies disclosed the required auditor data; consequently,
narratives provided incremental value in identifying FSM. I removed 279 firm years with missing data. Finally, 25 firm
years were removed for companies with only one observation.
Most linguistic analysis studies have been performed in developed A final sample of 829 firm years, representing 151 unique
economies (Bicudo de Castro et al., 2019). In South Africa, companies, remained. Table 1 presents a breakdown of the
linguistic analysis usage is limited, focusing on analyst stock sample by industry.
recommendations (Caglio et al., 2020), integrated reporting
(Mokoaleli-Mokoteli et al., 2009) and impressions management 3.2. Measuring Financial Statement Manipulation
(Nel et al., 2022; Totowa and Mokoaleli-Mokoteli, 2021) rather DeFond and Zhang (2014) recommend using multiple
than FSM. Thus, there is a gap in South African literature. Based measures of FSM. Therefore, I used two measures: fraud and
on the international findings, I draw the following non-directional discretionary accruals. In South Africa, two bodies investigate
hypothesis: financial non-compliance. Thus, I identified fraud as companies
H3: Tone at the top is associated with FSM. found guilty through an applicable Financial Sector Conduct
Authority (FSCA) enforcement action or forced Financial
Given their position within a company, top management has an Reporting Investigation Panel (FRIP) restatement. In addition,
information advantage over a company’s external stakeholders. I included qualified audit opinions1 within the fraud sample.
Top management, therefore, has the power to determine what Fraud is then measured as a dummy variable assigned the value
information is shared, how it is shared and when it is shared of one if the company was guilty and zero otherwise (Marais
(Zengin-Karaibrahimoglu et al., 2021), consequently they may et al., 2023).
manipulate the evidence provided to the external auditor (Greiner
et al., 2020). This may be achieved by using narratives to justify I measured discretionary accruals using Kothari et al.’s (2005)
manipulated numbers and mitigate concerns (Abou-El-Sood and cross-sectional, performance-adjusted model as defined in
El-Sayed, 2022). To address this risk, the auditor must evaluate a equation (1). This model improves upon the modified Jones
company’s tone at the top (International Auditing and Assurance model by including a performance and constant term, increasing
Standards Board, 2019). Prior studies by Greiner et al. (2020), the model power and reducing misspecification (Kothari et al.,
Rose et al. (2021) and Zengin-Karaibrahimoglu et al. (2021) 2005).
have shown that auditors adjust their risk assessment to account
for inappropriate tone in US and Dutch companies. However,
TACit  1   ∆SALESit − ∆RECit 
internationally, there remains a gap considering how tone at the =α 0 + α1   + α2  
top moderates the external auditors ability to restrain FSM. Thus, TAit −1  TAit −1   TAit −1 
I draw the following non-directional hypothesis:  PPEit 
+α 3   + α 4 ROAit + ∆ i ,t  (1)
H4: Tone at the top moderates the relationship between audit  TAit −1 
quality and FSM.
Where TACit represents total accruals for company i in year
t, TA represents total assets, ΔSALES is the change in sales,
3. METHODOLOGY
and ΔREC is the change in receivables, PPE is the gross book
3.1. Sample 1 I only included qualified audit opinions relating to fraud or an IFRS
The initial sample comprised all 278 non-financial companies violation. Qualifications based on issues such as going concern were not
listed on the Johannesburg Stock Exchange between 2011 and included within the fraud sample.

International Journal of Economics and Financial Issues | Vol 14 • Issue 5 • 2024 223
 Marais: Audit Quality and Financial Statement Manipulation: The Moderating Effect of Tone at the Top

Table 1: Sample by industry Following Patelli and Pedrini (2015), I applied a dictionary
Industry Number of Number of firm years approach when analysing CEO statements. Although domain-
companies specific wordlists are considered superior when analysing financial
Basic material 43 246 documents (Loughran and McDonald, 2011), they tend to identify
Consumer goods 18 93 only positive, negative or modal words, making them ill-suited to
Consumer services 27 154 measuring tone at the top. Therefore, I used DICTION to measure
Healthcare 6 34
Industrials 45 233
tone at the top, consistent with Patelli and Pedrini (2015) and
Oil and gas 2 9 Totowa and Mokoaleli-Mokoteli (2021).
Technology 10 43
Telecommunications 3 17 I used DICTION’s five themes of activity, certainty, commonality,
Total 1541 829 optimism and realism. Activity (TONEACT) refers to movement,
1
The number of companies per industry (154) exceeds the number of unique companies change and avoiding inertia (Hart and Carroll, 2015). An active
in the sample (151). During the period, three companies changed industry classifications
tone emphasises accomplishments, overconfidence and risk-taking
and have been included in both industries. However, the number of firm years only
represents the years the three companies spent in each sector. (Source: Researcher’s own (Greiner et al., 2020) and undermines credibility (Patelli and
construction) Pedrini, 2015). Certainty (TONECER) indicates determination,
rigidity, completeness and authoritative speech (Hart and Carroll,
value of property, plant and equipment, and ROA is the return 2015). It is a trait of transactional leaders (Patelli and Pedrini,
on assets. Discretionary accruals are measured as the residual 2015) and undermines verifiability and credibility (Greiner et al.,
in the equation, represented by the error term Ɛ. Discretionary 2020). Commonality (TONECOM) relates to group engagement
accruals are calculated by performing separate regressions for and cooperation (Greiner et al., 2020). It creates a sense of
each industry with more than ten observations in a financial year community but may undermine independence (Alshorman,
(Kothari et al., 2005). As a result, I excluded the healthc=are, oil 2016). Optimism (TONEOPT) highlights the positive attributes
and gas, telecommunications and some years of the technology of a person or group (Hart and Carroll, 2015). An optimistic tone
industry from these calculations. Consistent with Singh et al. tends to obscure failure while emphasising success (Griener et al.,
(2019), I measured the magnitude of the discretionary accruals as 2020) and represents a form of impression management, reducing
the absolute value. I further partitioned the discretionary accruals the quality of financial reporting (Alshorman, 2016; Patelli and
into income-increasing and income-decreasing discretionary Pedrini, 2015). An overly optimistic tone may portray the CEO
accruals. as naïve (Wunderley et al., 1998). Finally, realism (TONEREAL)
represents language dealing with real, day-to-day issues (Hart
3.3. Measuring Audit Quality and Carroll, 2015). Increased realism is associated with greater
The prior literature uses a range of variables to measure the transparency (Patelli and Pedrini, 2015), but is also a characteristic
competence and independence attributes of audit quality. The of a pragmatic leader (Alshorman, 2016). A pragmatic leader may
proxies for auditor competence included in this study are auditor be willing to manipulate financial statements to achieve an efficient
type, industry specialisation and joint auditors. Auditor type outcome (Alshorman, 2016).
(AUDTYPE) is a dummy variable assigned the value of one if
the auditor is a Big N auditor and zero otherwise (Baatwah et al., 3.5. Developing Composite Variables for Audit Quality
2018). I measured industry specialisation (AUDSPEC) as the and Tone at the Top
auditor’s market share (based on the number of clients) in each Using individual variables to proxy for audit quality and tone at
industry (Balsam et al., 2003; Chin and Chi, 2009). Finally, joint the top does not capture the orthogonal relationships between the
auditors (AUDJOINT) is a dummy variable assigned the value variables (Tarchouna et al., 2017). I used PCA to create composite
of one if joint auditors audit the company and zero otherwise variables to address this. In addition, PCA reduces measurement
(Bisogno and De Luca, 2016). error and potential multicollinearity, resulting in the composite
variables having a greater impact than the individual variables
I proxied auditor independence using auditor size, audit tenure (Biswas et al., 2022).
and non-audit services. Like auditor type, auditor size is proxied
by Big N membership. Audit tenure (AUDTEN) is the number I retained audit quality and tone at the top components with
of consecutive years a company is audited by the same auditor eigenvalues greater than one (Larcker et al., 2007). After
(Hohenfels, 2016). I measured audit fees (AUDFEES) as the ratio performing a varimax rotation, I used variables with absolute
of audit service fees to total auditor fees (Hohenfels and Quick, loading factors exceeding 0.40 to describe the component (Larcker
2020). et al., 2007). Finally, Bartlett’s sphericity and the Kaiser-Meyer-
Olkin test were used to determine the data’s sufficiency for PCA
3.4. Measuring Tone at the Top (Biswas et al., 2022).
I measured tone at the top by extracting CEO statements from the
company integrated reports. Although such CEO statements are Table 2 presents the development of the PCA components.
carefully planned and not always directly written by the CEO, I retained two components for both audit quality and tone at the
they are heavily involved in the process (Craig and Amernic, top. For audit quality (panel A), the first component comprises
2018). In South Africa, alternatives such as earnings conference auditor type, auditor specialisation and joint auditors. I label this
calls are not available. component as COMPETENCE. A Big N auditor (AUDTYPE)

224 International Journal of Economics and Financial Issues | Vol 14 • Issue 5 • 2024
 Marais: Audit Quality and Financial Statement Manipulation: The Moderating Effect of Tone at the Top

Table 2: Development of audit quality and tone at the top diversity (Hart and Carroll, 2015). Given South Africa’s history,
principal component analysis components a realistic leader is more likely to accommodate diversity, as it
Variable Component 1; Component 2; features highly in South Africa’s governance codes (Institute of
competence independence Directors in Southern Africa, 2016). Realism speaks to day-to-day
Panel A: Audit quality issues, transparency and pragmatic leadership, while optimism’s
AUDTYPE 0.4385 0.4243 inclusion may indicate motivational leadership. Consequently, I
AUDSPEC 0.6825 label component 2 as a PRAGMATIC TONE.
AUDTEN 0.5997
AUDFEES −0.6086
AUDJOINT 0.5808 3.6. Control Variables
Variable Component 1; Component 2; I included various control variables from the literature which
autocratic tone pragmatic tone affect FSM. I measured audit committee independence (ACIND)
Panel B: Tone at the top as the proportion of independent directors on the audit committee
TONEACT 0.4571 (Bicudo de Castro et al., 2019). I included operating cash flows to
TONECER 0.6310 total assets (OPCASH) (Sellami and Slimi, 2016) and a dummy
TONECOM −0.4913 variable assigned the value of one if the company experienced
TONEOPT −0.4835 0.4676 negative earnings in the current year and zero otherwise (LOSS)
TONEREAL 0.7111
(Nugrahanti and Puspitasari, 2018). I measured leverage (LEVER)
Components were determined using PCA with varimax rotation. Only components with
eigenvalues greater than one were retained. The table reports loadings that exceeded
as debt to total assets (Singh et al., 2019), while firm size was
0.4. Statistics for the sufficiency of the audit quality and tone at the top data for PCA calculated as the natural log of market capitalisation (SIZE)
included the Bartlett test of sphericity of 284.152 (P<0.01) and 235.405 (P<0.01), (Sun et al., 2014). Growth opportunities (GROWTH) were
respectively, as well as the Kaiser‑Meyer‑Olkin measure of sampling adequacy of 0.536
and 0.543, respectively. This indicated sufficiency to conduct PCA (Biswas et al., 2022).
measured as the market-to-book ratio (Singh et al., 2019), and,
A rho of 0.5427 was obtained for audit quality while a rho of 0.5468 was obtained for finally, I included a dummy variable assigned the value of one if
tone at the top. (Source: Researcher’s own construction). PCA: Principal component the company had joint CEOs and zero otherwise (JOINTCEO).
analysis

3.7. Regression Models

attracts better expertise and has access to more resources, resulting I used the following pooled logit and panel regression models3 to
in greater competence. A specialist auditor has greater knowledge test the relationship between audit quality, tone at the top and FSM:
of the company’s industry, while joint auditors may bring more
resources. The second component consists of auditor type, audit FSMit = β0 + β1 AUDPCAit + β2 TONEPCAit + β3 ACINDit + β4
tenure and audit fees. I label this component as INDEPENDENCE2. OPCASHit + β5 LOSSit + β6 LEVERit + β7 SIZEit + β8 GROWTHit +
A longer audit tenure creates greater familiarity with management. β9 JOINTCEOit + εit (2)
Audit fees load negatively into this component. A lower proportion
of non-audit services creates less financial dependence on the Where AUDPCA and TONEPCA represent the vector of composite
client. Auditor type also features in component 2 because Big variables developed during the PCA, and all other variables are
N auditors are expected to be more independent due to greater defined above.
reputational capital and a lower likelihood of financial dependence
on clients. However, the positive loading appears to indicate lower I used the following model to test the moderating effect of tone at
independence. This anomaly may arise because the prior literature the top on the relationship between audit quality and FSM:
fails to distinguish between the competence and independence
aspects of Big N auditors (DeFond and Zhang, 2014). Consequently, FSMit = β0 + β1 AUDPCAit + β2 TONEPCAit + β3 AUDPCAit ×
this conundrum needs to be resolved through further research. TONEPCAit + β4 ACINDit + β5 OPCASHit + β6 LOSSit + β7 LEVERit
+ β8 SIZEit + β9 GROWTHit + β10 JOINTCEOit + εit (3)
Table 2 Panel B presents the tone at the top components.
Component 1 comprises activity, certainty and optimism. This All continuous variables were winsorised at the first and ninety-
grouping is consistent with prior studies by Cho et al. (2010) and ninth percentiles to address extreme outliers. I used cluster robust
Totowa and Mokoaleli-Mokoteli (2021), who found that these standard errors to address heteroskedasticity and autocorrelation
tones were indicators of impression management. The negative (Hoechle, 2007).
loading of optimism is consistent with Cho et al. (2010), who found
that optimism and certainty had opposite loadings, and optimism
made a CEO look naïve (Wunderley et al., 1998). Such tones are 4. RESULTS AND DISCUSSION
associated with confidence and authority, so I label component 1
as AUTOCRATIC TONE. Component 2 comprises commonality, 4.1. Descriptive Statistics
optimism and realism. Commonality loads negatively into the Table 3 presents the descriptive statistics. For brevity, I focus on
component, indicating that the component focuses on greater the main variables used in this study (i.e. FSM, audit quality and

2 In component 2, longer audit tenure loaded positively, while higher audit 3 The pooled logit model was used for the fraud measure of audit quality. Due
fees (i.e. lower non-audit fees) loaded negatively. Hence, this component to the lack of variability in the dependent variables within and across firms,
represented less independence. Consequently, component 2 was multiplied panel logit was not feasible. The Hausman test was used to determine the
by negative one to represent independence. appropriateness of fixed or random effects for discretionary accruals.

International Journal of Economics and Financial Issues | Vol 14 • Issue 5 • 2024 225
 Marais: Audit Quality and Financial Statement Manipulation: The Moderating Effect of Tone at the Top

Table 3: Descriptive statistics

Variable Observations Mean Median SD Minimum Maximum
Financial statement manipulation variables
Fraud 829 0.0193 0.0000 0.1377 0.0000 1.0000
Absolute discretionary accruals 7501 0.0539 0.0366 0.0628 0.0000 0.7270
Income‑increasing discretionary accruals 367 0.0543 0.0401 0.0648 0.0000 0.7270
Income‑decreasing discretionary accruals 383 0.0534 0.0352 0.0609 0.0001 0.4752
Audit quality variables
AUDTYPE 829 0.8770 1.0000 0.3287 0.0000 1.0000
AUDSPEC 829 0.2390 0.2308 0.1275 0.0149 0.7500
AUDTEN 829 22.7214 14.0000 23.1444 1.0000 103.0000
AUDFEES 829 0.8720 0.9021 0.1281 0.1954 1.0000
AUDJOINT 829 0.0133 0.0000 0.1145 0.0000 1.0000
Tone at the top variables
Activity (TONEACT) 829 49.5399 49.66 1.9873 38.65 61.90
Certainty (TONECER) 829 46.6869 47.55 4.7665 ‑6.68 63.38
Commonality (TONECOM) 829 49.2946 49.40 1.9990 34.34 61.87
Optimism (TONEOPT) 829 54.8771 54.61 3.3524 45.01 69.84
Realism (TONEREAL) 829 52.3092 52.35 2.9279 41.64 85.51
Control variables
ACIND 829 0.9547 1.0000 0.1520 0.0000 1.0000
OPCASH 829 0.0864 0.0857 0.1016 ‑0.6622 0.7410
LOSS 829 0.1737 0.0000 0.3791 0.0000 1.0000
Leverage (LEVER) 829 0.4813 0.4750 0.1686 0.0286 1.1653
Firm size (SIZE) 829 15.6949 15.8041 2.1915 10.2753 21.4349
Growth (GROWTH) 829 1.0867 0.7377 1.0581 0.0234 8.3679
JOINTCEO 829 0.0205 0.0000 0.1418 0.0000 1.0000
1
The number of observations for the absolute discretionary accruals is less than the actual sample as the healthcare, oil and gas, technology, and telecommunications industries had fewer
than the prerequisite ten observations per industry‑year combination (Kothari et al., 2005). Consequently, I excluded 79 firm‑year observations.(Source: Researcher’s own construction).
AUDTYPE: Auditor type, AUDSPEC: Audit specialization, JOINTCEO: Joint CEOs, AUDTEN: Audit tenure, AUDFEES: Audit fees ratio, AUDJOINT: Joint auditors, ACIND: Audit
committee independence, OPCASH: Operating cash flows, LOSS: Negative earnings, SD: Standard deviation

tone at the top). I found that 1.93% of observations were fraudulent. means that, on average, each company’s auditor audited 23.9% of
Although this is higher than the 0.98% identified by Marais et al. the companies in the related industry. Audit fees average 87.2%
(2023), the current sample covers a shorter period in which more of total fees, leaving, on average, 12.8% for non-audit fees. This
corporate scandals were identified. Absolute discretionary accruals finding aligns with South African regulations that do not prohibit
amount to 5.39% of total assets. Income-increasing discretionary all non-audit services.
accruals are 5.43% of total assets, which is similar in value to the
income-decreasing discretionary accruals of 5.34%. This closeness For the tone at the top variables, certainty has the lowest mean
of value aligns with the reversing nature of accruals. Concernedly, of 46.69, while optimism has the highest of 54.88. The means
the highest income-increasing discretionary accruals were reported of activity, certainty and commonality all fall within the normal
at 72.7% of total assets (slightly lower than the 76.7% reported ranges provided by DICTION. The means of optimism and realism
by Eloff and Steenkamp (2022) in South Africa over a similar exceed the DICTION ranges, indicating that South African CEO
period), while the maximum income-decreasing is only 47.52%. statements are more optimistic and realistic than the Fortune
500 companies used to develop the DICTION ranges. However,
Considering the audit quality variables, 87.7% of firm-year given their different context, Fortune 500 companies may not be
observations were audited by big-4 companies. This exceeds the comparable to JSE-listed companies. Few South African studies
concentration Wesson (2021) reported from 2010 to 2018, which have used linguistic analysis, DICTION in particular. Despite this,
ranged from 62.66% to 70.04%. My study excludes financial the means of activity, optimism and certainty are comparable to
companies, while Wesson (2021) excludes non-South African those obtained by Totowa and Mokoaleli-Mokoteli (2021), while
engagement partners and joint auditors. Moreover, this study the mean for commonality aligns with that obtained by Nel et al.
excludes companies with missing data. Had such companies (2022). Of the South African studies identified as using DICTION,
been included, the proportion of observations audited by the big- only Nel et al. (2022) reported a score for realism. While their
4 would decrease to 76.45%, indicating that non-big-4 auditors score was lower than the current study, it may have been driven
are less likely to disclose all auditor information. Audit tenure by a smaller sample size and shorter study period.
indicates a mean of 22.72 years. This is longer than the 17 to
17.6 years reported by Wesson (2021). However, Wesson (2021) 4.2. Audit Quality and Tone at the Top Results
only reports audit tenure from 2016 until 2018. I found that 1.33% Table 4 presents the regression results for audit quality and tone
of companies have joint auditors, slightly more than the 0.98% of at the top. All models have significant explanatory power for
companies reported by Wesson (2021). Her proportion is based FSM. Auditor competence reveals no relationship with the fraud
on the number of companies, while mine is based on firm-year or discretionary accrual measures of FSM. While contrary to the
observations. The auditor specialisation mean is 23.9%. This theoretical expectations, this aligns with studies by Magnis and

226 International Journal of Economics and Financial Issues | Vol 14 • Issue 5 • 2024
 Marais: Audit Quality and Financial Statement Manipulation: The Moderating Effect of Tone at the Top

Table 4: Regression estimates of audit quality and tone at the top

Variable (1) (2) (3) (4)
Fraud1 Absolute discretionary Income‑increasing Income‑decreasing
accruals2 discretionary accruals3 discretionary accruals2
Auditor competence −0.4045 (0.5782) 0.0059 (0.0036) 0.0007 (0.0038) 0.0057 (0.0042)
Auditor independence 0.3620 (0.4364) 0.0111** (0.0050) −0.0033 (0.0027) 0.0203** (0.0079)
Autocratic tone −0.1079 (0.1730) 0.0024 (0.0016) −0.0003 (0.0016) 0.0033 (0.0025)
Pragmatic tone −0.1970 (0.2006) −0.0016 (0.0020) 0.0016 (0.0022) −0.0047* (0.0028)
ACIND4 −0.0093 (0.0179) 0.0294* (0.0167) −0.0158 (0.0178)
OPCASH −6.9934** (2.9876) 0.0593* (0.0337) −0.1682*** (0.0558) 0.2171*** (0.0527)
LOSS −1.9396 (1.2533) 0.0169** (0.0067) −0.0011 (0.0078) 0.0222** (0.0107)
LEVER 2.0397** (0.9960) 0.0354 (0.0266) −0.0205 (0.0192) 0.0561* (0.0329)
SIZE 0.3426* (0.1997) −0.0065 (0.0054) −0.0017 (0.0015) −0.0180*** (0.0069)
GROWTH −1.0992 (0.7424) 0.0015 (0.0041) 0.0011 (0.0029) 0.0008 (0.0063)
JOINTCEO5 0.0003 (0.0108) 0.0032 (0.0065)
Constant −9.0363*** (2.7141) 0.1350* (0.0796) 0.0703*** (0.0268) 0.2947*** (0.1012)
Observations 829 750 367 383
Number of companies 151 141 132 128
χ2 69.67*** 27.70***
F 2.377*** 5.457***
Pseudo R2 0.139
R2 0.3657 0.5197
Adjusted R2 0.2055 0.2510
Between R2 0.127
***P<0.01, **P<0.05, *P<0.1, 1Pooled logit regression with clustered standard errors was applied as the dependent variable was binary, 2A fixed effects panel regression model was
applied based on the outcome of the Hausman test, 3A random effects panel regression model was applied based on the Hausman test outcome, 4ACIND was excluded from the fraud
regression due to perfect prediction, 5JOINTCEO was excluded from the fraud regression due to perfect prediction. It was excluded from the income‑decreasing discretionary accruals
because it was non‑zero for only one cluster, Cluster robust standard errors are presented in parenthesis to address heteroskedasticity and autocorrelation. Based on the correlation matrix,
I identified no issues with multicollinearity. (Source: Researcher’s own construction)

Iatridis (2017), Sellami and Slimi (2016) and Yasser and Soliman display no relationship with discretionary accruals. However,
(2018), who found that the underlying variables of auditor certainty and optimism have positive, significant associations
competence are unrelated to FSM. Thus, I found no support for with fraud (at the 1% and 5% levels, respectively). As these load
H1, that greater auditor competence is associated with lower FSM in opposite directions in the PCA, this confirms that their effects
levels. offset, resulting in the composite autocratic tone variable revealing
no relationship with fraud.
Auditor independence also reveals no statistically significant
relationship with fraud. However, a positive association (P < 0.05) The pragmatic tone also shows no relationship with fraud.
was found with absolute discretionary accruals. This appears to be However, a negative association with income-decreasing
driven by independence’s association with higher levels of income- discretionary accruals was identified, indicating that the pragmatic
decreasing discretionary accruals (P < 0.05). To protect their tone is associated with lower discretionary accruals. Alshorman
independence, auditors may favour management decisions which (2016) found that only optimism is positively associated with FSM,
decrease earnings (Bédard et al., 2004). Thus, I found no support while commonality and realism are not. Alternatively, Patelli and
for H2 that greater auditor independence is associated with lower Pedrini (2015) found that commonality and realism are negatively
FSM levels. Instead, the evidence supports that independence is associated with FSM. Untabulated results show no significant
associated with higher income-decreasing discretionary accruals. relationship between the individual variables with discretionary
accruals. However, realism shows a negative association with
The autocratic tone shows no association with any measure of fraud, while optimism shows a positive relationship. As these
FSM. This is contrary to the majority of studies, which found loaded in the same direction in the PCA, their effects would offset.
that tone at the top provided incremental value in detecting FSM
(Hope and Wang, 2018; Kayed and Meqbal, 2024; Patelli and Generally, the tone at the top variables show no relationship
Pedrini, 2015). This contrary finding may result from most studies with FSM. Thus, these results predominantly do not support H3,
considering individual tone variables, ignoring the relationships that there is a relationship between tone at the top and FSM. An
between such tones. Of the DICTION-specific studies, Alshorman exception is that a pragmatic tone is associated with lower income-
(2016) found that activity, certainty and optimism are all positively decreasing discretionary accruals. The findings for H1 to H3 are
associated with FSM, while Patelli and Pedrini (2015) only found robust for different panel regression models.
certainty positively associated with FSM. I note that optimism
loaded negatively in the PCA while activity and certainty loaded Finally, I consider the control variables. Audit committee
positively, indicating a potential offsetting of effects which was not independence shows no relationship with absolute discretionary
considered by prior studies. In untabulated results considering the accruals and income-decreasing discretionary accruals. However,
individual variables, I found that activity, certainty and optimism a positive relationship (P < 0.1) exists with income-increasing

International Journal of Economics and Financial Issues | Vol 14 • Issue 5 • 2024 227
 Marais: Audit Quality and Financial Statement Manipulation: The Moderating Effect of Tone at the Top

discretionary accruals. This positive relationship may result holders. Firm size is positively associated with fraud. I also found
from information disadvantages due to the audit committee’s a statistically significant negative relationship between size and
independence from the company (Chen et al., 2020). Operating income-decreasing accruals, indicating that larger firms are less
cash shows statistically negative relationships with fraud and likely to manage earnings downwards. Growth opportunities reveal
income-increasing discretionary accruals and statistically no significant association with any measure of FSM, consistent
significant positive associations with absolute and income- with Sellami and Slimi (2016). Finally, the presence of joint CEOs
decreasing discretionary accruals. Although the discretionary also indicates no relationships.
accruals appear to have different signs, the overall indication is
that higher operating cash is associated with downward earnings 4.3. The Moderating effect of Tone at the Top on Audit
management (either through lower income-increasing or higher Quality
income-decreasing discretionary accruals). These findings Table 5 presents the regression results that include the interaction
contradict Sellami and Slimi (2016), who found no relationship. between audit quality and tone at the top. The significance
Companies incurring losses are associated with larger absolute of the individual audit quality and tone at the top variables
discretionary accruals, driven by higher income-decreasing remain unchanged from Table 4, except for auditor competence.
discretionary accrual levels. Companies experiencing losses may Competence now shows a statistically significant positive
take big baths by managing their earnings downwards to reflect relationship with absolute discretionary accruals, indicating that
improved performance in future years (Jordan and Clark, 2004). the standalone auditor competence is associated with higher
Higher leverage is an alternative monitoring mechanism (Sellami discretionary accruals when the interaction terms are included.
and Slimi, 2016). Surprisingly, I found that higher leverage is Although the income-increasing and decreasing components
positively associated with fraud. Liu et al. (2010) found that do not reveal significant relationships, the direction of the
companies issuing new debt manage earnings upwards to achieve coefficient signs (negative for income-increasing and positive for
lower borrowing costs. Companies with higher debt levels may income-decreasing) may suggest the companies audited by more
commit fraud for similar gains and to ensure covenants are met. competent auditors manage their earnings downwards. There are
Higher leverage is also positively associated with greater income- three changes in the control variable results. Audit committee
decreasing discretionary accruals levels. Companies with more independence no longer shows a significant relationship with
debt may act more conservatively to avoid the attention of debt income-increasing discretionary accruals, while operating cash no

Table 5: Regression estimates of the moderating effect of tone at the top on audit quality
Variable (1) (2) (3) (4)
Fraud1 Absolute discretionary Income‑increasing Income‑decreasing
accruals2 discretionary accruals3 discretionary accruals2
Auditor competence −0.4223 (0.4903) 0.0074** (0.0034) −0.0010 (0.0036) 0.0058 (0.0040)
Auditor independence 0.4419 (0.4017) 0.0111** (0.0049) −0.0036 (0.0025) 0.0192** (0.0082)
Autocratic tone −0.1496 (0.1675) 0.0025 (0.0016) 0.0005 (0.0016) 0.0032 (0.0028)
Pragmatic tone −0.2122 (0.2476) −0.0013 (0.0020) 0.0019 (0.0023) −0.0053* (0.0031)
Interaction of autocratic tone with
Auditor competence −0.2487* (0.1366) 0.0039 (0.0027) 0.0077*** (0.0024) −0.0006 (0.0044)
Auditor independence 0.0816 (0.0849) 0.0014 (0.0018) −0.0004 (0.0018) 0.0033 (0.0034)
Interaction of pragmatic tone with:
Auditor competence 0.3641 (0.2673) −0.0019 (0.0025) −0.0001 (0.0025) −0.0041 (0.0040)
Auditor independence 0.0347 (0.0936) −0.0018 (0.0013) −0.0015 (0.0018) 0.0017 (0.0020)
ACIND4 −0.0062 (0.0183) 0.0246 (0.0160) −0.0145 (0.0181)
OPCASH −7.5011** (3.1613) 0.0552 (0.0342) −0.1632*** (0.0563) 0.2162*** (0.0521)
LOSS −1.9725* (1.1633) 0.0162** (0.0065) −0.0018 (0.0081) 0.0207* (0.0106)
LEVER 2.4504** (0.9869) 0.0353 (0.0256) −0.0243 (0.0188) 0.0644** (0.0323)
SIZE 0.3816* (0.2175) −0.0059 (0.0053) −0.0020 (0.0015) −0.0180** (0.0069)
GROWTH −1.1686 (0.7821) 0.0013 (0.0041) 0.0003 (0.0028) 0.0007 (0.0066)
JOINTCEO5 0.0015 (0.0111) 0.0036 (0.0070)
Constant −9.9550*** (3.0565) 0.1238 (0.0786) 0.0829*** (0.0266) 0.2904*** (0.1030)
Observations 829 750 367 383
Number of companies 151 141 132 128
χ2 90.95*** 40.61***
F 2.386*** 4.470***
Pseudo R2 0.1585
R2 0.3720 0.5251
Adjusted R2 0.2081 0.2473
Between R2 0.1654
χ2 for joint test of tone 18.74*** 1.92* 19.31*** 1.52
***P<0.01, **P<0.05, *P<0.1. 1Pooled logit regression with clustered standard errors was applied as the dependent variable was binary, 2A fixed effects panel regression model was
applied based on the outcome of the Hausman test, 3A random effects panel regression model was applied based on the Hausman test outcome, 4ACIND was excluded from the fraud
regression due to perfect prediction, 5JOINTCEO was excluded from the fraud regression due to perfect prediction. It was excluded from the income‑decreasing discretionary accruals
because it was non‑zero for only one cluster.Cluster robust standard errors are presented in parenthesis to address heteroskedasticity and autocorrelation. Based on the correlation matrix, I
identified no issues with multicollinearity.(Source: Researcher’s own construction)

228 International Journal of Economics and Financial Issues | Vol 14 • Issue 5 • 2024
 Marais: Audit Quality and Financial Statement Manipulation: The Moderating Effect of Tone at the Top

longer reveals a significant relationship with absolute discretionary FSM. Specifically, this partial support applies to the interaction
accruals. These were previously statistically significant, but only between an autocratic tone and auditor competence. This finding
at the 10% level. The last change in the control variables is that for H4 is robust for different panel regression models. Given
companies experiencing losses reflect a statistically significant the lack of support for H4, I ran a Chi-squared test of the joint
negative relationship with fraud (P < 0.1). significance of the tone and interaction variables. The findings
(presented in Table 5) reveal that the explanatory power of the
The interaction between an autocratic tone and auditor competence models (excluding income-decreasing discretionary accruals)
indicates a negative association with fraud and a positive is significantly improved by including the tone variables. This
relationship with income-increasing discretionary accruals. confirms the findings of prior studies that inclusion of tone at the
The components of the autocratic tone (activity, certainty and top provides incremental value in identifying FSM (Hope and
pessimism) are associated with authority, self-confidence and Wang, 2018; Patelli and Pedrini, 2015).
rigid thinking, which would increase audit risk (Alshorman, 2016;
Patelli and Pedrini, 2015). The negative relationship with fraud
5. CONCLUSION
confirms Zengin-Karaibrahimoglu et al.’s (2021) findings that
competent auditors would increase their audit risk assessment,
Given recent corporate scandals in South Africa (such as Steinhoff
resulting in a commensurate adjustment to audit procedures.
and Tongaat-Hulett) and the mixed evidence on the auditors’
This adjustment outweighs the higher risk associated with the
ability to detect and report on FSM, this study aimed to identify
autocratic tone, thereby being associated with lower fraud.
the moderating effect that tone at the top has on the relationship
However, discretionary accruals are discretionary and fall within
between audit quality and FSM. Using a sample of 829 firm-
the bounds of accepted accounting frameworks (Dechow and
year observations, I used PCA to develop two comprehensive
Skinner, 2000). Therefore, auditors may be less able to challenge
audit quality measures (competence and independence) and two
or influence management’s judgements. An autocratic leader
with rigid thinking would be less willing to consider alternatives comprehensive tone measures (autocratic and pragmatic). FSM
offered by the auditor. was measured using identified fraud cases and discretionary
accruals.
The remaining interaction between an autocratic tone and
independence, as well as a pragmatic tone and both competence and As individual variables, I found that auditor independence results
independence, reveal no statistically significant relationships with in higher absolute discretionary accruals. This is driven by larger
any measure of FSM. The lack of significant relationships for the income-decreasing discretionary accruals indicating that firms who
remaining interaction terms has several possible explanations. On use more independent auditors tend to manage their earnings down.
the one hand, South African auditors may not consider tone at the Stakeholders should, therefore, be aware that more independent
top as a material risk related to fraud and discretionary accruals and auditors are likely to act more conservatively. Auditor competence
thus do not adjust their risk assessment. This appears to be justified, failed to reveal any relationship with FSM. Independently, neither
given the lack of statistically significant relationships found in this the autocratic nor pragmatic tones reveal an association with FSM.
study between the tone at the top variables and FSM. Alternatively, Regarding the interactions, an autocratic tone strengthens the
auditors may not consider management’s overall tone but only negative association between fraud and audit competence, but also
focus on certain tone aspects. Such an argument could explain why results in a more positive relationship between income-increasing
a relationship is only found with the autocratic tone, as the few discretionary accruals and auditor competence. Thus, stakeholders
prior South African studies using DICTION have focused on the in a company with an autocratic tone should feel comforted by
activity, certainty and optimism components while excluding the greater auditor competence, as it is associated with reduced fraud,
commonality and realism tones (Caglio et al., 2020; Mokoaleli- but they should be aware that management may attempt to push
Mokoteli et al., 2009; Totowa and Mokoaleli-Mokoteli, 2021). the boundaries of acceptable accounting standards. Other than
Furthermore, the major domain-specific wordlists for financial this, tone at the top was shown to have no moderating effect on
language (such as Henry, 2008; Loughran and McDonald, 2011) the relationship between audit quality and FSM.
focus predominantly on the positive-negative tone relationship.
Given the extensive research on this positive-negative tone, auditors This study contributes to the existing literature in several ways.
may concentrate more on it while neglecting other tones. A final Firstly, it corroborates the critical traits of auditor quality used in the
explanation may be that auditors do adjust their risk assessments literature: Competence and independence. Using PCA to develop
based on the tone at the top (as found by Greiner et al., 2020; composite measures from an array of observable characteristics
Rose et al., 2021; and Zengin-Karaibrahimoglu et al., 2021). allows an understanding of how these characteristics work together
This risk adjustment then almost perfectly offsets the risk effects and either complement or act as substitutes. Secondly, the study
(either positive or negative) of management’s tone, resulting in an reveals how tone at the top moderates the audit quality. Although
insignificant relationship to FSM. This explanation is less likely three of the four interaction terms show insignificant relationships,
in South Africa, given that the individual tone variables already the findings still reveal areas where management could push the
do not show any significant relationship with the FSM measures. boundaries with competent auditors to inflate earnings within the
scope of acceptable accounting frameworks. Finally, the study
Consequently, I found only partial support for H4, that tone at confirms the findings of the international literature that accounting
the top moderates the relationship between audit quality and for tone improves the ability to explain FSM in South Africa.

International Journal of Economics and Financial Issues | Vol 14 • Issue 5 • 2024 229
 Marais: Audit Quality and Financial Statement Manipulation: The Moderating Effect of Tone at the Top

To conclude, the study has some limitations. Despite making an International Journal of Auditing, 23(2019), 352-364.
interesting contribution to the limited literature on tone at the Bisogno, M., De Luca, R. (2016), Voluntary joint audit and earnings
top in developing countries, this may not be generalisable due to quality: Evidence from Italian SMEs. International Journal of
the focus on South Africa. Future studies should consider further Business Research & Development, 5(1), 1-22.
research in alternative geographical locations or across country Biswas, S., Bhattacharya, M., Sadarangani, P.H., Jin, J.Y. (2022),
Corporate governance and earnings management in banks: An
boundaries. Moreover, I only considered fraud and discretionary
empirical evidence from India. Cogent Economics and Finance,
accruals as measures of FSM. The unbalanced nature of fraud and
10(1), 1-20.
non-fraud observations may make it difficult to identify predictors Caglio, A., Melloni, G., Perego, P. (2020), Informational content and
of fraud. Discretionary accruals are also only one manner in which assurance of textual disclosures: Evidence on integrated reporting.
management may manipulate earnings. Future research should European Accounting Review, 29(1), 55-83.
consider alternative measures of fraud risk, including real-earnings Chen, J.Z., Cussatt, M., Gunny, K.A. (2020), When are outside directors
management measures and impression management measures. more effective monitors? Evidence from real activities manipulation.
Another consideration is that tone at the top was measured using Journal of Accounting, Auditing and Finance, 35(1), 26-52.
an automated bag-of-words approach using a general dictionary on Chin, C.L., Chi, H.Y. (2009), Reducing restatements with increased
planned CEO statements. While this method has various strengths, industry expertise. Contemporary Accounting Research, 26(3),
it does not consider domain-specific language and the carefully 729-765.
planned nature of CEO statements in the integrated reports. In Cho, C.H., Roberts, R.W., Patten, D.M. (2010), The language of US
corporate environmental disclosure. Accounting, Organizations and
the future, spontaneous communication methods should be used
Society, 35(2010), 431-443.
together with manual analysis of tone or domain-specific wordlists.
Craig, R., Amernic, J. (2018), Are there language markers of hubris in
Finally, audit quality is a complex construct. The observable CEO letters to shareholders? Journal of Business Ethics, 149(2018),
characteristics used may not present a complete picture. Therefore, 973-986.
future research should identify more nuanced methods to proxy DeAngelo, L.E. (1981), Auditor size and audit quality. Journal of
for audit quality. Accounting and Economics, 3(1981), 183-199.
Dechow, P.M., Skinner, D.J. (2000), Earnings management: Reconciling
6. FUNDING the views of accounting academics, practitioners, and regulators.
Accounting Horizons, 14(2), 235-250.
DeFond, M., Zhang, J. (2014), A review of archival auditing research.
This work was supported by the University of KwaZulu-Natal
Journal of Accounting and Economics, 38(2014), 275-326.
University Capacity Development Programme. El Guindy, M.N., Basuony, M.A.K. (2018), Audit firm tenure and earnings
management: The impact of changing accounting standards in UK
REFERENCES firms. The Journal of Developing Areas, 52(4), 167-181.
Eloff, A.M., Steenkamp, S. (2022), Integrated report quality and earnings
Abou-El-Sood, H., El-Sayed, D. (2022), Abnormal disclosure tone, management - Evidence from South Africa, South African Journal
earnings management and earnings quality. Journal of Applied of Economic and Management Sciences, 25(1), 1-10.
Accounting Research, 23(1), 402-433. Garrett, J.B., Hoitash, R., Prawitt, D.F. (2022), Perceptions of tone at the
Alshorman, S. (2016), CEO Tone, Earnings Management, and the top from the inside: Insights into audit pricing. Auditing: A Journal
Moderating Effect of CEO Power. Unpublished PhD thesis, of Practice and Theory, 41(1), 115-141.
University of South Australia, Australia. Greiner, A., Patelli, L., Pedrini, M. (2020), Characteristics of managerial
Al-Thuneibat, A.A., Issa, R.T.I., Baker, A.A. (2011), Do audit tenure and tone priced by auditors: Evidence based on annual letters to
firm size contribute to audit quality? Empirical evidence from Jordan. shareholders of large U.S. firms. Auditing: A Journal of Practice and
Managerial Auditing Journal, 26(2011), 317-334. Theory, 39(2), 139-161.
Amernic, J., Craig, R., Tourish, D. (2010), Measuring and Assessing Harber, M., Marx, B. (2020), Auditor independence and professional
Tone at the Top Using Annual Report CEO Letters. Available from: scepticism in South Africa: Is regulatory reform needed? South
https://researchportal.port.ac.uk/en/publications/measuring-and- African Journal of Economic and Management Sciences, 23(1), 1-12.
assessing-tone-at-the-top-using-annual-report-ceo-l [Last accessed Hart, R.P., Carroll, C.E. (2015), DICTION 7.1 Help Manual. Available
on 2018 Oct 04]. from: https://www.dictionsoftware.com/download.php?file=wp-
Baatwah, S.R., Salleh, Z., Ahmad, N. (2018), High-quality auditors content/uploads/2015/07/DICTION-7.1Manual.pdf [Last accessed
vs. high-quality audit: The reality in Oman. Afro-Asian Journal of on 2018 Nov 26].
Finance and Accounting, 8(3), 209-236. Henry, E. (2008), Are investors influenced by how earnings press releases
Bala, H., Amran, N.A., Shaari, H. (2020), Audit committee attributes are written? Journal of Business Communication, 45(4), 363-407.
and cosmetic accounting in Nigeria. Managerial Auditing Journal, Hoechle, D. (2007), Robust standard errors for panel regressions with
35(2), 177-206. cross sectional dependence. The Stata Journal, 7(3), 281-312.
Balsam, S., Krishnan, J., Yang, J.S. (2003), Auditor industry specialisation Hohenfels, D. (2016), Audit tenure and perceived earnings quality.
and earnings quality. Auditing: A Journal of Practice and Theory, International Journal of Auditing, 20(2016), 224-238.
22(2), 71-97. Hohenfels, D., Quick, R. (2020), Non-audit services and audit quality:
Bédard, J., Chtourou, S.M., Courteau, L. (2004), The effect of audit Evidence from Germany. Review of Managerial Science, 14(2020),
committee expertise, independence, and activity on aggressive 959-1007.
earnings management. Auditing: A Journal of Practice and Theory, Hope, O.K., Wang, J. (2018), Management deception, big-bath
23(2), 13-35. accounting, and information asymmetry: Evidence from linguistic
Bicudo de Castro, V., Gul, F.A., Muttakin, M.B., Mihret, D.G. (2019), analysis. Accounting, Organizations and Society, 70(2018), 33-51.
Optimistic tone and audit fees: Some Australian evidence. Huang, X., Krishnan, S., Lin, P. (2018), Tone analysis and earnings

230 International Journal of Economics and Financial Issues | Vol 14 • Issue 5 • 2024
 Marais: Audit Quality and Financial Statement Manipulation: The Moderating Effect of Tone at the Top

management. Journal of Accounting and Finance, 18(8), 46-61. government: The case of three metropolitan municipalities in South
Inaam, Z., Khamoussi, H. (2016), Audit committee effectiveness, audit Africa. African Journal of Public Affairs, 13(1), 144-172.
quality and earnings management: A meta-analysis. International Mokoaleli-Mokoteli, T., Iatridis, G.E. (2017), Big 4 auditing companies,
Journal of Law and Management, 58(2), 179-196. earnings management and earnings conservatism: Evidence from
Institute of Directors in Southern Africa. (2016), King IV Report on an emerging market. Investment Management and Financial
Governance for South Africa 2016. South Africa, IoDSA. Innovations, 14(1), 35-45.
International Auditing and Assurance Standards Board. (2019), Mokoaleli-Mokoteli, T., Taffler, R.J., Agarwal, V. (2009), Behavioural bias
International Standard on Auditing 315: Identifying and Assessing and conflicts of interest in analyst stock recommendations. Journal
the Risks of Material Misstatement. New York, IAASB. of Business Finance and Accounting, 36(3), 384-418.
Jensen, M.C., Meckling, W.H. (1976), Theory of the firm: Managerial Mukhlasin, M. (2018) Auditor tenure and auditor industry specialisation
behavior, agency costs and ownership structure. Journal of Financial as a signal to detect fraudulent financial reporting. Academy of
Economics, 3(4), 305-360. Accounting and Financail Studies Journal, 22(5), 1-10.
Jordan, C.E., Clark, S.J. (2004), Big bath earnings management: The case Nel, G., Arendse-Fourie, S., Ontong, J.M. (2022), The association
of goodwill impairment under SFAS No. 142. Journal of Applied between optimism and future performance: Evidence of impression
Business Research, 20(2), 63-70. management from chief executive officer and chairperson letters.
Karpoff, J.M., Koester, A., Lee, D.S., Martin, G.S. (2017), Proxies and South African Journal of Business Management, 53(1), 1-12.
databases in financial misconduct research. The Accounting Review, Noviyanti, S., Winata, L. (2015), The role of “tone at the top” and
92(6), 129-163. knowledge of fraud on auditors’ professional skeptical behavior.
Kayed, S., Meqbel, R. (2024), Earnings management and tone Contemporary Management Research, 11(1), 55-74.
management: Evidence from FTSE350 companies. Journal of Nugrahanti, Y.W., Puspitasari, A. (2018), Do audit quality, political
Financial Reporting and Accounting, 22(4), 842-867. connection, and institutional ownership increase real earnings
Kothari, S.P., Leone, A.J., Wasley, C.E. (2005), Performance matched management? Evidence from Indonesia. Afro-Asian Journal of
discretionary accrual measures. Journal of Accounting and Finance and Accounting, 8(4), 413-430.
Economics, 39(2005), 163-197. Özcan, A. (2019), Audit quality and earnings management: Evidence
Krishnan, G.V. (2003), Does big 6 auditor industry expertise constrain from Turkey. International Journal of Economic and Administrative
earnings management. Accounting Horizons, 17(2003), 1-16. Studies, 23(2019), 67-78.
Larcker, D.F., Richardson, S.A., Tuna, I. (2007), Corporate governance, Patelli, L., Pedrini, M. (2015), Is tone at the top associated with
accounting outcomes, and organisational performance. The financial reporting aggressiveness. Journal of Business Ethics,
Accounting Review, 82(4), 963-1008. 126(1), 3-19.
Larcker, D.F., Zakolyukina, A.A (2012), Detecting deceptive discussions Piot, C., Janin, R. (2005), Audit Quality and Earnings Management
in conference calls. Journal of Accounting Research, 50(2), 495-540. in France. Available from: https://papers.ssrn.com/sol3/papers.
Li, F. (2010), The information content of forward-looking statements cfm?abstract_id=830484 [Last accessed on 2023 Apr 06].
in corporate filings - A naive Bayesian machine learning approach. Rabin, C.E. (2016), Earnings Management in South Africa: Evidence
Journal of Accounting Research, 48(5), 1049-1102. and Implications. Unpublished PhD Thesis, Univeristy of the
Lin, J.W., Hwang, M.I. (2010), Audit quality, corporate governance, and Witwatersrand, Johannesburg, South Africa.
earnings management: A meta-analysis. International Journal of Rezaee, Z. (2005), Causes, consequences, and deterence of financial
Auditing, 14(2010), 57-77. statement fraud. Critical Perspectives on Accounting, 16(2005),
Lisic, L.L., Neal, T.L., Zhang, I.X., Zhang, Y. (2016), CEO power, internal 277-298.
control quality, and audit committee effectiveness in substance versus Rose, A.M., Rose, J.M., Suh, I., Thibodeau, J., Linke, K., Norman, C.S.
in form. Contemporary Accounting Research, 33(3), 1199-1237. (2021), Why financial executives do bad things: The effect of the
Liu, Y., Ning, Y., Davidson, W.N.I. (2010), Earnings management slippery slope and tone at the top on misreporting behavior. Journal
surrounding new debt issues. The Financial Review, 45(2010), of Business Ethics, 174(2021), 291-309.
659-681. Schmidt, R.N. (2014), The effects of auditors’ accessibility to “tone at
Loughran, T., McDonald, B. (2011), When is a liability not a liability? the top” knowledge on audit judgments. Behavioral Research in
Textual analysis, dictionaries, and 10-Ks. The Journal of Finance, Accounting, 26(2), 73-96.
66(1), 35-65. Selahudin, N.F., Azman. N.A.N.N., Suhaimi, A.N.F., Ahmad, M.F.M.,
Magnis, C., Iatridis, G.E. (2017), The relationship between auditor Rahman, N.N.A., Sushela, R., Ramesh, M.R. (2018), Do female
reputation, earnings and capital management in the banking sector: director, audit quality and audit committee characteristics influence
An international investigation. Research in International Business the earnings management? Global Business and Management
and Finance, 39(2017), 338-357. Research: An International Journal, 19(3), 130-139.
Makhlouf, M.H.S., Oroud, Y., Soda, M.Z., Ramadam, A.H. (2021), Sellami, Y.M., Slimi, I. (2016), The effect of the mandatory adoption
The moderating effect of audit quality on the relationsip between of IAS/IFRS on earnings management: Empirical evidence from
information asymmetry and earnings management: Evidence from South Africa. International Journal of Accounting and Economic
Jordan. Afro-Asian Journal of Finance and Accounting, 12(2), Studies, 4(2), 87-95.
165-177. Singh, A., Singh, H., Sultana, N., Evans, J. (2019), Independent and joint
Marais, A., Vermaak, C., Shewell, P. (2023), Predicting financial statement effects of audit partner tenure and non-audit fees on audit quality.
manipulation in South Africa: A comparison of the Beneish and Journal of Contemporary Accounting and Economics, 15(2019),
Dechow models. Cogent Economics and Finance, 11(1), 1-33. 186-205.
Marmousez, S. (2009), The Choice of Joint-Auditors and Earnings Sun, J., Cahan, S. (2009), The effect of compensation committee quality
Quality: Evidence from French Listed Companies. In: 2009 on the association between CEO cash compensation and accounting
CAAA Annual Conference. April 2009. Montreal, Canada. http:// performance. Corporate Governance: An International Review,
ssrn.1330061 [Last accessed on 2023 Feb 07]. 17(2), 193-207.
Mnguni, S., Subban, M. (2022), Audit outcome challenges in local Sun, J., Lan, G., Liu, G. (2014), Independent audit committee

International Journal of Economics and Financial Issues | Vol 14 • Issue 5 • 2024 231
 Marais: Audit Quality and Financial Statement Manipulation: The Moderating Effect of Tone at the Top

characteristics and real earnings management. Managerial Auditing Management, 52(1), 1-13.
Journal, 29(2), 153-172. Wunderley, L.J., Reddy, W.B., Dember, W.N. (1998), Optimism and
Svanström, T. (2013), Non-audit services and audit quality: Evidence from pessimism in business leaders. Journal of Applied Social Psychology,
private firms. European Accounting Review, 22(2013), 337-366. 28(9), 751-760.
Tarchouna, A., Jarraya, B., Bouri, A. (2017), How to explain non- Yasser, S., Soliman, M. (2018), The effect of audit quality on earnings
performing loans by many corporate governance variables management in developing countries: The case of Egypt. International
simultaneously? A corporate governance index is built to US Research Journal of Applied Finance, 9(4), 216-231.
commercial banks. Research in International Business and Finance, Zengin-Karaibrahimoglu, Y., Emanuels, J., Gold, A., Wallage, P.
42(2017), 645-657. (2021), Audit committee strength and auditors’ risk assessments:
Totowa, J., Mokoaleli-Mokoteli, T. (2021), Chairman’s letter, impression The moderating role of CEO narcissism. International Journal of
management and governance mechanisms: A case of South African Auditing, 25(2021), 661-674.
listed firms. African Finance Journal, 23(1), 1-19. Zerni, M., Haapamaki, E., Jarvinen, T., Niemi, L. (2012), Do joint
Wesson, N. (2021), Will mandatory audit firm rotation reduce audit market audits improve audit quality? Evidence from voluntary joint audits.
concentration in South Africa? South African Journal of Business European Accounting Review, 21(4), 731-765.

232 International Journal of Economics and Financial Issues | Vol 14 • Issue 5 • 2024