The impact of the Financial Crisis on the commitment of Financial Statement Fraud
A quantitative study among US fraud firms
ABSTRACT
This study examines whether the (global) financial crisis of 2007 – 2010 had a significant impact on the commitment of financial statement fraud. Using hand-collected data of 134 US fraud firms and a data sample period of 2002 – 2012, this study was able to identify a significant association between the occurrence of the financial crisis and the frequency of financial statement fraud. Moreover, this study found a significant relation with respect to the commitment of financial statement fraud of firms with low profitability numbers. In addition, a significant association was found regarding the commitment of financial statement fraud of firms which are/were operating in capital-intensive industries (e.g. real estate).
Nevertheless, this study found some associations concerning the control variable ‘firm size’, which could have an important impact on the commitment of financial statement fraud. Overall, this study contributes to the current literature in accounting, law, and regulation after the implementation of SOX (in 2002), especially by investigating the consequences of the financial crisis.
KEYWORDS: financial statement fraud; financial crisis; (fraud) risk assessment model.
Name: Frank Tuinder (S2759802) Address: Tweede Willemstraat 58A Postal code: 9725 JM Groningen Phone number: 06 230 220 88
Email: frank-tuinder@hotmail.com Coordinator: dr. K. (Kristina) Linke
Word count: 11.688 (excl. tables and references)
University of Groningen Faculty of Economics & Business June 2018
Table of contents
Ⅰ. Introduction ... 2
Ⅱ. Theory and background ... 4
2.1 Main theory: risk assessment model ... 4
2.1.1 Conditions ... 4
2.1.2 Motivations... 5
2.1.3 Attitudes ... 5
2.2 Literature review and hypotheses ... 6
2.2.1 Financial crisis and financial statement fraud ... 6
2.2.2 Consequences of the financial crisis ... 7
ⅡI. Methodology ... 11
3.1 SEC database ... 11
3.2 Data collection ... 11
3.2.1 Fraud case definition ... 11
3.2.2 Data sample ... 12
3.2.3 Other data sources ... 12
3.3 Research variables ... 13
3.4 Control variables ... 14
IV. Results ... 16
4.1 Descriptive statistics ... 16
4.2 Testing for normality ... 19
4.3 Testing for homogeneity of variance ... 20
4.4 Hypotheses testing ... 20
V. Discussion and conclusion ... 25
5.1 Discussion of the results ... 25
5.2 Conclusion ... 26
5.3 Theoretical and practical implications ... 27
5.4 Limitations and future research ... 27
References ... 28
“What we know about the global financial crisis is that we don’t know very much.”
- Paul Samuelson
Ⅰ. Introduction
“The financial crisis of 2007 - 2010 could have been avoided” (The Guardian, 2011). According to the Financial Crisis Inquiry Commission (FCIC) Report (2011), the financial crisis was (to a great extent) caused by governance mismanagement and aggressive expansion of US banks. The Federal Reserve also failed in its duty to establish more sufficient rules: in fact, institutional investors were able to buy and sell mortgage securities and other investments without examination and/or verification by the US government (FCIC, 2011). Moreover, institutional investors also played a central role in enabling the financial crisis by investing ‘blindly’ in potentially flawed securities to gain major short-term profits (FCIC, 2011).
The (global) financial crisis began with a major financial breakdown in the US subprime mortgage sector.
Besides, the crisis was triggered by its rapidly spread to the rest of the world due to the global trading of bonds and – so-called – credit default swaps (CDSs) on subprime mortgages (Fisher et al., 2011).
The first indications became visible when the US housing bubble collapsed due to an increase in interest rates and a reduction in economic growth. These developments caused a major increase in defaults rates on subprime mortgages. Subsequently, credit-rating agencies, like Standard & Poor’s and Fitch Ratings, immediately downgraded their AAA-ratings (Fisher et al., 2011). The issuers of the aforementioned CDSs, like the American International Group (AIG), were not able to cover the major losses of these downfalls.
Due to these sudden downgrades of credit rating agencies, banks were forced to post more collateral under the – so-called – Basel II agreements and, to realize the required liquidity targets, numerous banks were forced to rely on governmental bailouts or to go into liquidation (Fisher et al., 2011).
According to the FCIC Report (2011), 26 million US citizens were out of work at the time of the (global) financial crisis and nearly 11 trillion dollar in household wealth has completely vanished (FCIC, 2011).
Next to the major financial losses of households and the liquidity problems at several banks, the financial crisis also affected ‘normal’ businesses (Polat et al., 2013). Firms have been forced to cut costs, as reflected in particular measures such as the reduction of staffing levels (Polat et al., 2013). In addition, due to the high magnitude of financial distress, firms experienced delays in payments to suppliers and tax authorities.
Moreover, many organizations were forced to find new ways to be profitable (again). As a consequence, decision-making was primarily focused on generating profit (at any cost). Finally, risky strategies were implemented and organizations massively ‘destructed’ jobs to reduce salary costs (Polat et al., 2013).
The US subprime mortgage sector can be seen as a significant contributor to the (global) financial crisis and the certain problems in this sector can be related to (mortgage) fraud and corruption (Button, 2011 ed.).
The relation between (mortgage) fraud and the occurrence of the financial crisis has been studied many times before (Fisher et al., 2011; Allen et al., 2012). However, according to Levi et al. (2011), analysts and researchers are not in (complete) agreement that fraud will increase as a result of the financial crisis.
Ernst & Young (2009) found that 55% of the respondents in their study expected that corporate fraud
will increase after and during the financial crisis and Levi et al. (2011) showed that the risks regarding the
falsification of documents will increase due to the consequences of the financial crisis (Levi et al., 2011).
Reviews of several studies expose that there are a variety of arguments which imply that the (global) financial crisis of 2007 – 2010 will lead to either a higher or lower level of corporate fraud (Gill, 2011).
While poor economic circumstances result in less credit being available to individuals (incentive for fraud), it can also indicate, for example, that individuals spend more time at home which will lead to relatively more ‘guardianship’ (pull-away). Moreover, due to the (global) financial crisis, there are multiple concerns that less emphasis has been laid on fraud prevention (Gill, 2011). Employees and managers are able to take advantage of this situation to manipulate the firms’ financial statements.
Furthermore, Gill (2011) also suggests that fraud (in general) increased, because firms were suffering and – consequently – they will be more desperate for credit. Three out of four US executives who participated in KPMG’s Fraud Survey (2009) also mentioned that financial statement fraud will either increase or stay the same in the first twelve months of the recovery after the (global) financial crisis of 2007 – 2010.
The results of the KPMG Fraud Survey (2009) entail that employees and managers were pressured to find new ways to commit fraud, especially when organizations implement (new) changes into their systems as a result of the financial crisis. Besides, employees and managers experienced more pressure to hit their financial targets and goals, which are harder to achieve in adverse/poor economic conditions (Svare, 2009).
In addition, the increasing magnitude of financial distress might have forced employees and managers to take more drastic measures in order to receive more credit: ”Everybody has survival instincts. People need a minimum (amount of money/credit) to live in accordance with their lifestyle and whenever this minimum is threatened, a ‘survival mode’ comes out” (Gill, 2011, p. 210).
According to Gill (2011), more research is needed in this area to examine what types of financial fraud – or what characteristics of them – are most likely to be affected by poor economic conditions, like the (global) financial crisis. Fraud perpetrators, who generally believed that financial fraud would increase because of the financial crisis, didn’t feel that it would influence their own situations (Gill, 2011).
During adverse economic circumstances, it seems to be clear that organizations will look more precisely at what impacts on the bottom line and, subsequently, more fraudulent behaviour may be exposed (Gill, 2011). To conclude, there still exists a lot of uncertainty regarding the impact of the financial crisis on the commitment of financial statement fraud.
The goal of this study is to understand the aforementioned impact of the (global) financial crisis of 2007 – 2010 on the commitment of financial statement fraud. The relationship between the occurrence of financial statement fraud and – for example – the characteristics of the perpetrator(s) is often described (Robison et al., 2011). The individual motivation and the role of the CEO in financial statement fraud schemes is also subject to continuous interest for research (McAnally et al., 2008; Efendi et al., 2007;
Zhang et al., 2008). In addition, Beasley et al. (2010) concluded that problems regarding the commitment of financial statement fraud still exist and requires further consideration. Like mentioned before, the literature regarding this impact is still scarce. However, this study is intended to give additional (quantitative) insights with respect to the results of the qualitative study performed by Gill (2011).
As a result, the research question that will be guiding this study will be the following:
“Did the (global) financial crisis of 2007 – 2010 impact the commitment of financial statement fraud and how?”
In the remainder of this study, the theory and background, methodology, results, and – finally –
the discussion and conclusion will be described.
Ⅱ. Theory and background
This section describes the underlying theory and literature which will be the departure point of this study.
In the first place, the main theory of this research paper will be described. Besides, the important concepts will be described by means of a review of existing literature. Finally, the hypotheses are formulated.
2.1 Main theory: risk assessment model
Fraudulent behaviour is an important global problem which affects organizations of all types and sizes.
According to the 2016 Global Fraud Study of the Association of Certified Fraud Examiners (ACFE), an organization loses nearly 5% of annual revenues as a result of fraud (in general). The total loss caused by fraud exceeded 6.3 billion dollar in their most recent study (ACFE, 2016). Firms are continuously trying to recognize the techniques and incentives of fraud perpetrators in order to design sufficient internal controls to prevent, detect, and correct certain fraudulent activities (Murdock, 2008). Statement on Auditing Standards (SAS) No. 99 identifies three factors which are present when fraud (in general) occurs.
In the first place, there is the motivation for committing fraud. Second, certain conditions provide an opportunity for fraud to be perpetrated. Finally, there may be a form of rationalization which helps the fraud perpetrator to deal with his/her own personal disagreement with respect to their individual behaviour (Huang et al., 2017). The model that brings these three factors together is known as the fraud triangle.
These three main concepts of (financial statement) fraud also form the basis for the theoretical risk assessment model of Loebbecke et al. (1989). This model is established for auditors to lead an assessment regarding the likelihood of financial statement fraud commitment and refers to the following categories:
conditions, motivations, and attitudes. These categories are quite similar to the aforementioned factors of the fraud triangle. However, Loebbecke et al. (1989) state that there are 55 fraud indicators (‘red flags’) which could potentially indicate financial statement fraud, like weak internal control measures, industry decline, rapid growth, high leverage, etc. Based on their conclusions, Loebbecke et al. (1989) labelled the more important indicators as ‘primary’ and all other relevant fraud indicators as ‘secondary’.
These 55 fraud indicators don’t solitarily possess a fraud prediction character, but they can be used to describe the commitment of financial statement fraud (Linke, 2012). Therefore, the risk assessment model of Loebbecke et al. (1989) will be used as the departure point for this study. In the following sections, the aforementioned categories of the risk assessment model will be described with special consideration to the commitment of financial statement fraud.
2.1.1 Conditions
The first condition of the risk assessment model is concerned with ‘the degree to which conditions are such that a material management (financial statement) fraud could be committed’ (Loebbecke et al., 1989, p. 5).
Loebbecke et al. (1989) found that dominant decision-making and weak internal controls are the primary
indicators or ‘red flags’ which are present and relevant in over 75% of the examined fraud cases. Moreover,
other (secondary) ‘red flags’ include periods of rapid growth, decentralized organizations, assets subject
to misappropriation, and new clients (Loebbecke et al., 1989). SAS No. 99 also provides several examples
of risk indicators that increase the likelihood of the commitment of financial statement fraud. These risk
factors include, for example, the nature of the industry in which the organization operates, like significant,
complex, or related party transactions; ineffective monitoring measures by management; a complex
organizational structure; and, of course, inadequate internal controls (Huang et al., 2017). To conclude,
a sufficient and effective internal control environment is possibly the most important measure to minimize
(or eliminate) the aforementioned conditions for committing financial statement fraud (Hogan et al., 2008).
2.1.2 Motivations
The second category of the risk assessment model is concerned with ’the degree to which the person or persons in positions of authority and responsibility in the organization have a reason or motivation for committing financial statement fraud’ (Loebbecke et al., 1989, p.5). The primary indicators or
‘red flags’ regarding this category include industry decline, inadequate profits, and placing undue emphasis on meeting earning projections. Besides, secondary indicators are, for example, rapid industry changes, sensitive operating results, and adverse legal circumstances (Loebbecke et al., 1989). The desire to meet internal or external earning expectations, the pressure to obtain external financing, and the desire to increase management compensations based on financial performances can potentially cause pressures and motivations for managers and employees to commit financial statement fraud (Beasley et al., 2010).
These pressures can be seen as unsolvable through legal measures and non-shareable with others who could offer assistance to the offender’s personal crisis, which motivates the perpetrator to behave illegally in order to (re)solve his/her own problem (Cressey, 1971). Previous studies showed that the desire to meet external analysts’ forecasts is, generally, the most frequent motivation for managers to commit fraud (Kaplan, 2001).
In addition, the motivational factor for committing financial statement fraud is in 43% of the examined financial fraud cases – indeed – related to the desire to meet analysts’ forecasts (Huang et al., 2017).
Moreover, other incentives or pressures may arise when internal compensation structures, which are generally based on financial performances, cause employees or managers to illegally misreport earnings (Goldman et al., 2006). Furthermore, financial distress can be a motivational factor to commit fraud.
Previous studies showed that the level of financial distress is higher at organizations where fraud has been committed in the past than at non-fraud organizations (Huang et al., 2017).
2.1.3 Attitudes
The last category of the risk assessment model is concerned with ’the degree to which the person or persons in positions of authority and responsibility in the organization have an attitude or a set of ethical standards such that they would allow themselves to commit financial statement fraud’ (Loebbecke et al., 1989, p. 5).
The primary indicators or ‘red flags’ regarding this category include, for example, lies or evasiveness, dishonest management, and personality anomalies. The secondary indicators include conflicts of interest, frequent disputes with the auditor, and weak internal controls (Loebbecke et al., 1989). These indicators are related to the rationalization of fraudulent behaviour by fraud perpetrators who – eventually – think that they are being consistent with their personal code of ethics. “The fraud perpetrators generally admit the wrongdoing, but deny that is was wrong” (Dellaportas, 2013, p. 32). This rationalization allows fraud perpetrators to continue engaging in their fraudulent behaviour without the guilt that arises from their illegal actions. Rationalisation with respect to financial statement fraud can take many forms:
“We will fix the books”, “It’s for a good purpose”, or “There were no internal controls so I wanted to show them how easy it was.” Therefore, due to these rationalisations, fraud perpetrators are able to view themselves as moral and ethical individuals, simply forced to act in an illegal manner (Anand et al., 2004).
This study is intended to primarily focus on the categories regarding the conditions and motivations of the risk assessment model of Loebbecke et al. (1989). The underlying attitudes of fraud perpetrators include variables which are less relevant for this study. Besides, these variables are (nearly) impossible to measure by means of a quantitative research approach. This study is not concerned about understanding the underlying perceptions, feelings, and opinions of perpetrators who committed financial statement fraud.
However, this study is intended to give quantitative insights regarding the impact of the financial crisis on
the commitment of financial statement fraud. In the following section, these concepts are described in order
to develop our hypotheses.
2.2 Literature review and hypotheses
This section describes the important concepts related to the financial crisis and financial statement fraud.
By means of the following literature review, the hypotheses of this study will be formulated.
2.2.1 Financial crisis and financial statement fraud
“When the US sneezes, the world catches a cold.” This statement is referring to the bursting of the US housing bubble and its rapidly spread to the rest of the world by means of the global trading of bonds and – so-called – credit default swaps (CDSs) on subprime mortgages (Fisher et al., 2011). In 2001, the Federal Reserve (Fed) ‘suddenly’ decreased its interest rate from 6.5% to 1% until July 2004.
However, the inflation rate over this period was around 2%, which stimulated both business and households to engage in ‘favourable’ credit lending. Due to the monetary stimulus implemented by the Fed concerning favourable credit lending for households, the demand in the housing market increased rapidly (Scott, 2010).
A constant increase in US housing prices, since the burst of the internet bubble, was the result.
During the US housing boom, lenders engaged in ‘greedy’ lending of mortgages as they were unaffected by the possible defaults on these loans (due to lax Federal laws). Once these loans were issued, the issuers passed them on to a bundler who then securitised them and sold them as collateralized debt obligations (CDOs) to investors (Fisher et al., 2011). However, the owners of these CDOs had very little information about the underlying quality of these loans and were dependent on the credit agencies for guidance.
In order to compensate for the risks of the defaults on mortgages, the holders of CDOs purchased (more) CDSs, as a form of insurance (Fisher et al., 2011). Moreover, since the US housing prices were constantly increasing in value, and since mortgages were collateralized, mortgage lenders added more risks to new mortgages: not requiring down payments, no proof of monthly income, etc. These new mortgages were also known as ‘subprime mortgages’. Investment banks entered this market in a big way through major investments in order to gain more leverage (Scott, 2010). However, there was not a great demand for the ‘risky investments’ due to lower credit ratings (Scott, 2010). Yet, in order to sell these investments, banks combined them into a new pool of home mortgages, and combined them with many other pools of other home mortgages (Beltran et al., 2017). In addition, banks added some other forms of consumer debts to these pools, like credit card loans, student loans, and also commercial loans. As a result, the underlying investments became further removed from the securities which were held by the investor (Scott, 2010).
After approximately six years of constantly increasing US housing prices, the housing bubble collapsed due to an increase in interest rates and a significant decrease in economic growth (Fisher et al., 2011).
The first indications appeared when more households defaulted on their (monthly) mortgage payments.
Due to the collateralization of mortgages, and the major volume of defaults on mortgage payments, the supply in the housing market increased excessively. As a result, housing prices immediately fell and (institutional) investors were, in fact, holding several CDO’s which contained ‘worthless’ houses and other types of valueless securities (Scott, 2010). Besides, issuers of the CDSs (e.g. AIG) didn’t have enough cash flows to honor all the swaps. Due to the devaluations of CDOs and the major corporate bankruptcies, like the Lehman Brothers, the US mortgage market froze and collapsed (Andersan et al., 2011).
Subsequently, people started to sell their securities and consumers withdrew their bank accounts due to
the lack of trust in the US banking system (Andersan et al., 2011). As a consequence, the borrowing costs
of US banks increased excessively, as did the degree of mistrust between US banks: ‘no one wanted to be
left holding the bag of worthless CDO’s and other investments’ (Van der Cruijsen et al., 2016). As a result,
numerous individuals lost their confidence in the US financial system, which caused a major reduction
in consumption, and – subsequently – firms’ profitability. A severe financial crisis was well underway.
At times of a financial recession, employees and managers experience more pressure to hit their financial targets and goals, which are also harder to achieve in adverse/poor economic conditions (Svare, 2009).
This pressure to meet, for example, external analysts’ expectations and particular earnings projections, can be seen as a possible indication for the (potential) commitment of financial statement fraud by managers and/or employees (Loebbecke et al., 1989). According to the results of Beasley et al. (2010), senior management is involved for some level in the examined fraud cases. However, in 89% of the cases, the CEO and/or CFO are primarily being named in the examined AAER cases for the leading involvement in financial statement fraud schemes (Beasley et al., 2010). According to the study of Khanna et al. (2015), CEOs have a significant influence on the (legal) organizational behaviour. This has to do with the ‘close’
connection with other top executives and directors. Furthermore, CEOs often have powerful positions within an organization that allows them to be dominant during decision-making processes (Ge et al., 2010).
However, CFOs are often involved in financial statement fraud schemes because they submit to the pressure of their CEO (Boyle et al., 2010). In addition, Linke (2012) concluded that the involvement of financial middle managers in financial statement fraud schemes is associated with the particular fraud size.
Subsequently, the aforementioned executives/managers (CFOs, CEO, etc.) were experiencing relatively more financial pressure due to the consequences of the financial crisis. As a result, they are (more) inclined to commit financial statement fraud in order to ‘keep the good times rolling’ (Finnerty et al., 2016).
During the financial crisis, profits were declining due to thinner margins. Therefore, top executives and managers were pressured to manipulate earnings, especially when increasing revenues growth was expected by external analysts (Dechow et al., 2011). In addition, Moore et al. (2016) argue that the financial crisis was the cause for a sharp downfall in nearly all industry performance indicators. The industries that relied heavily on external financing were associated with major declining industry rates (Moore et al., 2016).
To conclude, the occurrence of the (global) financial crisis is associated with a decline in industry growth, which causes (more) financial distress/pressure for top executives and managers to hit their financial goals.
As a consequence of the financial crisis, the aforementioned actors/agents were (more) pressured to commit fraud in order to meet external (and internal) expectations. Hence, the following hypothesis is formulated:
H1: During the (global) financial crisis of 2007 – 2010, the frequency of financial statement fraud commitment increased
2.2.2 Consequences of the financial crisis
Like mentioned before (section 2.1), Loebbecke et al. (1989) identified several ‘red flags’ which could potentially indicate financial statement fraud. This study is intended to analyse whether these indicators were being influenced by the occurrence of the (global) financial crisis. In order to determine the relevance of the particular fraud indicators for this study, three evaluation criteria have been established. First of all, the fraud indicator should be related to financial statement fraud. For example, the fraud indicator that is concerned with weak internal controls is related to a different type of fraud, namely embezzlement or theft.
Second, the fraud indicator should be vulnerable to (drastic) economic changes, like the financial crisis.
The situation in which an officer of the organization is a former member of the audit firm, for example, is expected not to be influenced by the occurrence of the financial crisis.
Finally, the fraud indicator must be measurable in terms of quantitative data. Like mentioned before,
the underlying perceptions, feelings, and opinions of fraud perpetrators are nearly impossible to measure
by means of a quantitative research approach. Therefore, in the following sections, the most relevant fraud
indicators of the risk assessment model (Loebbecke et al., 1989) are being described and explained.
2.2.2.1 Leverage
During the financial crisis, especially firms with established lending relationship were able to increase leverage ratios significantly more than firms without such lending relationships (Dewally et al., 2014).
As a result, for each dollar of equity, these high leveraged firms could have 40 dollar of assets.
Exploiting this way of gaining leverage left these firms highly vulnerable to fluctuations in asset values, which led to major financial instabilities and distress (especially during the financial crisis).
According to Grammatikos (2012), extremely high leveraged firms were experiencing significantly lower stock-market performances during the financial crisis than firms with lower financial leverage ratios.
Besides, Tsuruta (2015) concluded that high leveraged firms have a high(er) probability of bankruptcy, especially when asset values (e.g. of derivatives) massively decreased a result of the (global) financial crisis.
As a consequence, high leveraged firms were experiencing relatively more financial distress/problems.
These financially distressed firms may have greater incentives for committing financial statement fraud than firms that are not financially distressed, especially during the (global) financial crisis of 2007 – 2010 (Brazel et al., 2009). Previous studies showed that the cost of high financial leverage ratios can lead to poor firm performances and financial distress (Tsuruta, 2014), which could be considered as an important indicator for the potential commitment of financial statement fraud.
The particular fraud indicator that is concerned with high leveraged firms is one of the 55 ‘red flags’
in the risk assessment model of Loebbecke et al. (1989). According to this model, high leveraged firms are (more) motivated/pressured to commit financial statement fraud. Managers and/or employees might commit financial statement fraud to avoid financial distress and, eventually, to avoid bankruptcy problems (Johnson et al., 2009). Moreover, due to high financial leverage, firms are (more) motivated/pressured to commit financial statement fraud in order to prevent the violation of certain debt covenants or other significant contractual commitments (Loebbecke et al., 1989). To put it briefly, especially during the (global) financial crisis, high leveraged firms were (more) motivated to commit financial statement fraud.
Hence, the following hypothesis is formulated:
H2: During the (global) financial crisis of 2007 – 2010, the commitment of financial statement fraud of high leveraged firms increased
2.2.2.2 Profitability
Despite the emerging organizational motivations for corporate social responsibility, sustainable businesses, environmentally friendly policies, etc., the maximization of profits will still be a leading factor in management and executive decision-making; especially during the (global) financial crisis (Yap, 2014).
According to Jones (2010), firms paid relatively more attention to earnings management and profitability, as a result of the financial crisis. In times of a financial downturn, organizational profitability is declining due to thinner margins. This has to do with a decline in household and individual spending/consumption.
Firms are dependent on the consumption of households and individuals to guarantee their continuity.
Throughout the financial recession, 60% of household wealth declined massively between 2007 and 2009, and nearly 25% lost more than half of their household wealth (Deaton, 2012). Many individuals reduced their spending even without facing financial losses due to the (global) financial crisis of 2007 – 2010.
Therefore, the dynamics of spending/consumption may change during financial recessions when
individuals and households find it more difficult to obtain credit due to increasing risk-aversion of
(investment) banks and tightened credit standards (Alegre, 2016). Due to this reduction in consumption
of households and individuals, firms’ profitability significantly declined.
As a consequence, firms might have been more motivated (or pressured) to manipulate earnings in order to meet profitability projections or other financial targets/goals. This potential fraud indicator (‘red flag’) has been recognized by the risk assessment model of Loebbecke et al. (1989). Furthermore, Linke (2012) argues that revenue is mainly manipulated through either the creation of false revenues or too early recognition of revenues. Therefore, as a result of the (global) financial crisis, earnings management and revenue recognition techniques were used massively to hide poor financial profits in order to place undue emphasis on meeting profit projections and meeting compensation arrangements which are based on recorded performance targets (Jones, 2010).
To put it briefly, the resulting combinations of tightened credit standards, reduced consumer spending, and reduced business-to-business transactions, caused a significant decline in firms’ profitability numbers (Yap, 2014). This decline in profitability will cause, amongst other things, organizational financial distress, solvency and liquidity problems, and – eventually – bankruptcy problems. Accordingly, these problems could be an incentive to commit financial statement fraud, especially during the (global) financial crisis.
Hence, the following hypothesis is formulated:
H3: During the (global) financial crisis of 2007 – 2010, the commitment of financial statement fraud of firms with low profitability increased
2.2.2.3 Liquidity
The financial crisis of 2007 – 2010 was a ‘big shock’ to the US banking system, which raised important questions about liquidity and credit risks. Due to the major financial losses, US (investment) banks had to start selling their assets, hoarding cash, and tightening risk/credit management. As a result, liquidity
‘dried up’ during the financial crisis of 2007 – 2010 (Cornett et al., 2011). Some particular firms, however, relied extensively on the credit lines provided by banks (Campello et al., 2011). Subsequently, the firms that relied heavily on the availability of (investment) bank’s credit experienced major liquidity constraints.
This created uncertainty about the ease of future lending and, therefore, about the financial health of many US firms on the long-term (Dewally et al., 2014).
In addition, according to Campello et al. (2011), the lack of liquidity created managerial uncertainty and fear regarding the continuity of the firm’s operations in the future. Consequently, firms could experience relatively more financial distress during the financial crisis, because of the aforementioned uncertainty and anxiety for liquidity problems. As a consequence of the (global) financial crisis, liquidity constraints have been mentioned as the primary cause of many decisions to file in for bankruptcy and/or for the outcome of many bankruptcy cases (Ayotte et al., 2013).
Furthermore, Ericsson et al. (2006) concluded that the extent of financial distress is affected by the illiquidity of the general market for distressed debt. In conclusion, low liquidity numbers will lead to a high extent of financial distress. Subsequently, managers and/or executives might commit financial statement fraud to avoid these liquidity constrains, and, therefore, to avoid financial distress and – eventually – to avoid bankruptcy problems (Johnson et al., 2009). According to the risk assessment model of Loebbecke et al. (1989), firms are (more) pressured/motivated to commit fraud when adverse conditions are present, such as liquidity constraints and/or tightened credit standards. In brief, especially during the (global) financial crisis, low liquidity firms are/were (more) pressured to commit financial statement fraud in order to ‘survive’ the financial downfall. Hence, the following hypothesis is formulated:
H4: During the (global) financial crisis of 2007 – 2010, the commitment of financial statement fraud
of firms with low liquidity increased
2.2.2.4 Type of industry
In times of the financial crisis, (investment) banks reduce lending to their customers and new lending opportunities declined substantially across various types of loans. This drop in the supply of credit had important implications for several industries (Ivashina et al., 2010). However, the financial crisis had a negative impact mainly on those industries which are more reliant on external financing and on industries which are capital-intensive (Ivashina et al., 2010). These capital-intensive industries (e.g. construction and real estate) experienced major adverse conditions during the crisis in terms of many business failures.
Due to the aforementioned ‘credit crunch’, many firms in capital-intensive industries were not able to perform their operational activities due to liquidity problems. In addition, the financial crisis changed the business assumptions in capital-intensive industries due to thinner margins (Jones, 2010). Therefore, more emphasis has been laid on meeting several earnings projections in these particular industries.
In addition, particular industries are highly sensitive to changing economic factors which could have an impact on asset values, financial leverage, earnings, etc. Subsequently, the pressure to commit financial statement fraud could be higher in particular (capital-intensive) industries (Loebbecke et al., 1989).
In conclusion, the extent in which the crisis had an impact on the commitment of financial statement fraud could be context-specific. The fraud indicator that is concerned with declining industries with many business failures could be more relevant for certain industries than for others (Loebbecke et al., 1989).
This has to do with the fact that some industries are more sensitive to business/economic cycles than others.
According to the risk assessment model of Loebbecke (1989), major industry declines and business failures are related to the commitment of financial statement fraud. Moreover, the fraud indicator regarding adverse conditions in the industry is also related to financial statement fraud (Loebbecke et al., 1989).
Therefore, the impact of the financial crisis of on the commitment of financial statement fraud could be seen as context-specific. Therefore, during the financial crisis of 2007 - 2010, some particular industries were (more) pressured to commit financial statement fraud. Hence, the following hypothesis is formulated:
H5: During the (global) financial crisis of 2007 – 2010, the commitment of financial statement fraud
in capital-intensive industries (e.g. like real estate) increased
ⅡI. Methodology
This section describes the research methodology in terms of used databases, data collection methods, research variables, and statistical methods.
3.1 SEC database
The database concerning ‘Accounting and Auditing Enforcement Releases’ (AAERs), which is published by the US Securities and Exchange Commission (SEC), will be used as the primary database for this study.
The SEC published all (fraudulent) reporting related enforcement actions in terms of civil lawsuits, criminal charges, and/or settlement of administrative proceedings. Additionally, the AAERs are often used by other studies when investigating financial statement fraud (Beasley et al. 2010; Farber, 2005). Besides, according to Beasley et al. (2010), the SEC database is the most reliable and comprehensive database that is concerned with the commitment of financial statement fraud in the US. The AAERS are frequently accompanied by certain litigation releases and/or SEC complaints. These publications contain specific information about the alleged fraud cases and – therefore – the use of the litigation releases and/or SEC complaints are preferred ever since the SEC guarantees its completeness (Linke, 2012).
However, the SEC database has its limitations. First of all, the majority of the fraud cases are settled by the defendant without admitting or denying the allegations. Furthermore, there are certain resource limitations regarding the SEC enforcement staff; therefore, it is possible that not all fraud cases are being identified.
Thirdly, the possibility exists that the SEC is focusing solely on particular industries or transactions due to a higher probability of detection (Beasley et al., 2010).
The reason why this study is related to US firms has to do with the fact that US data is publicly available and (relatively) more financial statement cases occurred in the US than in other countries (Linke, 2012).
Besides, the (global) financial crisis started with the collapse of the US housing market. As a result, the aforementioned consequences (section 2.2.2) are to a greater extent noticeable within US firms.
The particular information concerning firm characteristics (e.g. CIK-code, type of industry, etc.) will be collected from the EDGAR database.
3.2 Data collection
In the next section, the fraud case definition, sample period, data collection aspects, research variables, and – finally – the control variables will be described.
3.2.1 Fraud case definition
The alleged cases of financial statement fraud that are available in the AAER database have to correspond to the following definition of Beasley et al. (2010):
“… the intentional material misstatement of financial statements of financial statements/disclosures or the perpetration of an illegal act that has a material direct effect on the financial statements or financial disclosures” (p. 7).
This definition focuses primarily on financial statement fraud and excludes forms of earnings management
that is based on GAAP or real operating actions (Linke, 2012). The published allegations regarding this
definition relate to the Sections 10(b), 13(a), and 13(b)(5) of the 1934 Securities Exchange Act,
along with section 17(a) of the 1933 Securities Act (Linke, 2012).
3.2.2 Data sample
In order to correspond with the aforementioned definition of Beasley et al. (2010), it is important to mention that the fraud case should have been committed by an US firm in order to be part of the analyses. Moreover, the possibility exists that certain fraud cases violated the aforementioned sections, but do not include a manipulation of the financial statements, like insider trading, bribery and/or embezzlement (Linke, 2012).
These cases are (more) related to self-enrichment and are, therefore, excluded from the data sample.
Finally, errors and/or failures that could result in material misstatements of the financial statements are not included since these cases are not related to an ‘intentional act’.
The AAERs archives (period of 1999 – 2018) were reviewed according to the aforementioned requirements.
Furthermore, the identified fraud cases were arranged on the basis of their starting fraud year/quarter.
The fraud cases that started before 2002 are excluded from the data sample and, therefore, the sample period of this study is concerned with the years 2002 – 2018 (beginning year). The starting date July 30, 2002 was selected because of the implementation of the Sarbanes-Oxley (SOX) Act by the US Congress.
According to the results of Coates et al. (2014), financial reporting quality has been improved after the implementation of SOX. This law extended and set new rules and requirements for US public firms in terms of improving the accuracy and reliability of financial statements.
The direct effects of the financial crisis impacted the US economy in the last quarter of the year 2007 (Lang et al., 2012). However, the actual financial distress in the US economy was noticeable in the beginning of 2007. Furthermore, Clinch et al. (2011) also argue that the financial crisis began in 2007, and, therefore, the consequences of the global financial crisis should have been noticeable in the financial statements of 2007 of US firms. In addition, Choi (2013) argue that the consequences of the (global) financial crisis were affecting the US economy and – simultaneously – US firms until 2010.
Shahrokhi (2011) also points out that the financial crisis (and distress) began in 2007 and ended in 2010, which was related to slow/stagnant economic growth and high unemployment rates. As a consequence, the years (and quarters) from 2007 (Q1) through 2010 (Q4) will be used to categorize the impact of the (global) financial crisis on the commitment of financial statement fraud.
3.2.3 Other data sources
Next to the SEC databases regarding the AAERs and EDGAR, COMPUSTAT will be used as an additional database to collect financial and accounting information of US fraud firms. Together, the identified fraud cases are being analysed in terms of the following aspects:
• Firm name;
• Beginning year (quarter) and ending year (quarter) fraud;
• Specific financial information, like total liabilities (LT), total stockholders’ equity (SEQ), total assets (AT), total current assets (ACT), total inventories (INV), total current liabilities (LCT), gross profit/loss (GP), total amount of revenue (REVT), and the type of industry.
Only the beginning year/quarter of the particular financial statement fraud scheme is relevant for this study.
The ending year, however, is not relevant for this study but will be collected for future study purposes.
The beginning year in which the perpetrator(s) committed fraud is related to the particular year in which
the perpetrator(s) was/were pressured due to the financial crisis (and distress). Therefore, to measure the
impact of the financial crisis on the commitment of financial statement fraud, solitarily the beginning year
(and quarter) of the fraud case will be used for analyses.
3.3 Research variables
In the next section, the several research variables are being described. In order to measure the impact of the financial crisis on the commitment of financial statement fraud, this study will use one predictor variable, five outcome variables, and two control variables.
3.3.1 The financial crisis
The predictor variable (or independent variable) of this study is concerned with the financial crisis (FC).
This variable will be measured in terms of a dummy variable. Like mentioned before, only the beginning year/quarter of the alleged financial statement fraud cases will be used. A short explanation concerning the occurrence of the (global) financial crisis is given in section 2.2.1. The fraud cases that started in the years before 2007 are labelled with a dummy value of 0 as well as the cases that started after 2010.
The fraud cases that started in the years 2007 (Q1) – 2010 (Q4) are labelled with a dummy value of 1.
3.3.2 Frequency of financial statement fraud
The first outcome/dependent variable is concerned with the yearly (and quarterly) frequency (FREQ) of the commitment of financial statement fraud. However, the possibility exists that there are more AAERs available which are related to the same fraud case/firm (e.g. civil actions, administrative procedures, etc.).
These releases are included all at once under the same firm name/case.
3.3.3 Leverage
The second outcome variable (or dependent variable) of this study is concerned with leverage (LEV).
This variable will be measured in terms of the debt-to-equity ratio (total liabilities divided by total equity) of the particular fraud firms. This ratio indicates the long-term financial soundness and correlation between the liabilities of the firm and the stockholders’ equity. This leverage formula will be conducted in the two years prior to the particular year in which the financial statement fraud has been committed. This has to do with the fact that leverage numbers do not have an immediate effect on the firms’ conduct of business (‘lagged effect’). In addition, the particular firm size has been taken into account when calculating the debt-to-equity ratio, and – therefore – this ratio will be divided (once again) by the total amount of assets.
Finally, in order to test the hypothesis, the average of the two leverage ratios will be calculated.
3.3.4 Profitability
The third outcome variable is concerned with profitability (PROF), which will be measured in terms of the gross profit margin (gross profit divided by total revenue). The gross profit margin is used because it is relatively simple to calculate and the outcome indicates an ‘overall picture’ of the firm’s profitability.
This margin is calculated for the year prior to the particular year in which the firm committed fraud.
3.3.5 Liquidity
The fourth outcome variable of this study is concerned with liquidity (LIQ). This variable will be measured
in terms of the quick ratio (current assets minus inventory, divided by inventories). The quick ratio is
preferred in this study because it doesn’t include inventory and/or other current assets that are relatively
(more) complex to turn into cash. By using this ratio, this formula focuses primarily on the firm’s more
liquid assets. Moreover, this liquidity formula will be conducted in the two years prior to the particular year
in which the firm committed the fraud. Finally, the average of the two liquidity ratios will be calculated.
3.3.6 Industry
The last outcome variable of this study is concerned with the type of industry (INDU) in which the particular fraud firm conducts (or conducted) its operations. The Standard Industrial Classicisation (SIC) codes will be determined by means of reviewing the AAERs and, subsequently, will be collected from the aforementioned EDGAR database. In this study, primarily the first two digits that are related to the general industry sectors will be used for analyses. Farber (2005) and Linke (2012) also used the two-digit SIC-codes to measure the type of industry. These SIC-codes are presented in table 1.
SIC-code Industry
01 – 09 Agriculture – Forestry – Fishing
10 – 14 Mining
15 – 17 Construction
20 – 39 Manufacturing
40 – 49 Transportation – Public Utilities
50 – 51 Wholesale Trade
52 – 59 Retail Trade
60 – 67 Finance – Insurance – Real Estate
70 – 89 Services
91 – 99 Public Administration
Table 1: Industry (SIC) codes (link: https://siccode.com/)
The above industries are measured separately in terms of a dummy value. If the fraud firm is operating in one of the 10 industries, the variable will be separately coded in terms of a binary value of 1 and 0 otherwise.
In addition, the particular industries that concerns Finance (60 – 62) and Insurance (63 – 64) are excluded from the data sample. The financial statements of banks, for example, often indicate high leverage ratios that undoubtedly do not have the same meaning as for certain non-financial firms, where high leverage ratios indicate financial distress (section 2.2.2.1).
3.4 Control variables
In order to clarify the relation between the dependent and independent variables, this study will use two different control variables. Based on prior research, the control variable regarding firm size is often used when studying the commitment of financial statement fraud (Farber, 2005; Bonner et al., 1998).
According to Ozcan (2016), firm size is highly associated with the commitment of financial statement fraud.
In addition, prior studies indicated that firm size could be a predictor of the (potential) commitment of financial statement fraud (Baucus et al., 1991). Furthermore, top executives (e.g. CFOs) of larger firms are more likely to report in a (more) fraudulent manner (Gillet et al., 2005).
In addition, the (potential) violation of debt covenants will be used as a second control variable. According to Demerjian et al. (2016), there are different types of debt covenants which are used as restrictions in certain debt agreements to protect the lender/investor. These debt covenants may include the following:
maintaining certain minimum financial ratios, pay taxes and other liabilities when due, provide audited
financial statements, etc. (Demerjian et al., 2016). The two control variables that are concerned with
firm size and debt covenant violation will be described in the next two sections.
3.4.1 Firm size
The first control variable will be measured in terms of the total amount of revenues (SIZE) based on prior research (Coles et al., 2006; Dalton et al., 1988). This type of accounting information will be collected from the EDGAR and/or the COMPUSTAT database. Furthermore, only the total amount of revenue (USD) of the year prior to the particular year in which the firm committed fraud will be used in this study. However, the possibility exists that there will be a relationship between the commitment of financial statement fraud and firm size based on prior research (Bonner et al., 1998; Baucus et al., 1991). Finally, in order to correct for possible outliers and normality problems, this study is intended to use a logarithm scale.
3.4.2 Debt covenant violation
The second control variable will be measured in terms of the potential violation of debt covenants (COV).
According to the risk assessment model of Loebbecke et al. (1989), the subjection to particular significant contractual commitments could be an indicator (or motivation) to commit financial statement fraud.
The (potential) violation of these contractual commitments, like debt covenants, are (sometimes) mentioned in the SEC complaints and/or in the litigation releases of the identified financial statement fraud cases.
The variable that is concerned with the (potential) violation of debt covenants will be labelled with a dummy value of 1. If this is not the case, this variable will be labelled with a dummy value of 0.
3.5 Research methodology
The five hypotheses that are formulated in section 2.2 will be tested by means of SPSS Statistics.
This study is intended to used two different statistical analyses. Like mentioned before, the financial crisis is the independent variable with 1 = during the financial crisis, and 0 = before/after the financial crisis.
The dependent variables of this study are concerned with the frequency of financial statement fraud, leverage, profitability, liquidity, and the type of industry. The variable that is concerned with the type of industry is also measured by means of a dummy value. The other four dependent variables are measured through continuous variables. When analysing two dichotomous variables by means of SPSS, this study is intended to use the Pearson Chi² Test. In addition, when analysing dichotomous and continuous variables, the Independent Samples Test will be used. A summary of the research variables is given in table 2.
Variable Type Measurement Test method
1. FC Independent Dummy variable -
2. FREQ Dependent Yearly frequency Independent Samples Test 3. LEV Dependent Debt-to-equity ratio Independent Samples Test 4. PROF Dependent Gross profit margin Independent Samples Test
5. LIQ Dependent Quick ratio Independent Samples Test
6. INDU Dependent Dummy variable Pearson Chi² Test
Table 2: Variable summary
IV. Results
This section describes the results of this study in terms of descriptive statistics and hypotheses testing.
The section that is related to the descriptive statistics will start with a specification of the identified fraud cases according to their time-periods/yearly classifications. In addition, the mean, the standard deviation, and the correlations between the variables are presented. The hypotheses are tested in section 4.4.
4.1 Descriptive statistics
The following table is given for the yearly categorization of the identified fraud cases:
Fraud beginning year Total amount of fraud cases Change (%)
2002 14 NA
2003 14 0%
2004 13 -7%
2005 6 -54%
2006 10 67%
2007 15 50%
2008 15 0%
2009 13 -13%
2010 17 31%
2011 7 -59%
2012 10 43%
2013 2 NA
2014 2 NA
Table 3: Yearly classification of fraud cases
The financial statement fraud cases that began in 2013 and 2014 are excluded from the data analyses.
Due to the aforementioned time lag between the discovery and the actual commitment of the fraud cases, the possibility exists that there are potential financial statement fraud cases committed in 2013/2014 (and not yet discovered). As a consequence, these years will (possibly) distort the results of the analyses, and – therefore – solitarily the years 2002 – 2012 will be included.
As a result, the total amount of identified financial statement fraud cases within the selected data sample period concerns 134 cases, from which 60 cases occurred during the financial crisis and 74 cases occurred before/after the financial crisis. The following table is given:
Time-periods Total amount of fraud cases Average per year
Before the financial crisis 57 11,40
During the financial crisis 60 15,00
After the financial crisis 17 8,50
Total amount of fraud cases 134 12,20
Table 4: Specification of fraud cases according to time-periods
Total sample N Minimum Maximum Mean Standard deviation 1. Research variables:
FC 60 - - -
FREQ 134 - - - -
LEV LOG
134 134
0,00 -2,30
5,56 8,62
0,22 2,02
0,81 2,57 PROF
LOG
134 134
0,00 2,30
1,00 9,88
0,39 5,73
0,25 1,07 LIQ
LOG
134 134
0,01 2,30
15,38 10,95
2,07 7,15
2,29 1,24
INDU 134 - - - -
2. Control variables:
SIZE LOG
134 134
0,00 3,40
87.471,90 18,29
3.508,20 11,99
11.983,54 2,85
COV 16 - - - -
Table 5: Descriptive statistics for total sample
Before/after FC N Minimum Maximum Mean Standard
deviation 1. Research variables
FC 0 - - - -
FREQ 74 - - - -
LEV LOG
74 74
0,00 -2,30
3,46 8,15
0,14 1,99
0,54 2,39 PROF
LOG
74 74
0,00 3,40
0,94 6,85
0,43 5,87
0,24 0,74 LIQ
LOG
74 74
0,01 2,30
15,38 9,64
1,96 7,12
2,25 1,10
INDU 74 - - - -
2. Control variables:
SIZE LOG
74 74
0,00 5,30
31.168,00 17,25
1.748,87 12,09
5.190,51 2,36
COV 8 - - - -
Table 6: Descriptive statistics before/after the financial crisis
During FC N Minimum Maximum Mean Standard
deviation 1. Research variables
FC 60 - - - -
FREQ 60 - - - -
LEV LOG
60 60
0,00 -2,30
5,56 8,62
0,33 2,06
1,04 2,81 PROF
LOG
60 60
0,00 3,00
1,00 6,91
0,34 5,67
0,24 0,75 LIQ
LOG
60 60
0,01 2,30
12,84 9,46
2,21 7,13
2,34 1,31
INDU 60 - - - -
2. Control variables:
SIZE LOG
60 60
0,00 3,40
87.471,90 18,29
5.678,03 11,86
16.784,39 3,38
COV 8 - - - -
