The fraud process concerning the perpetrators and techniques
A quantitative study on fraud firms in the US
ABSTRACT:
In this thesis is researched whether there are patterns in the involvement of perpetrators and the use of fraud techniques regarding the time span of financial statement fraud. Next to the question whether or not people get involved in fraudulent behaviour and which fraud technique is used the process model of collective corruption is an important guideline in this thesis. The relation between different stages in the fraud process and perpetrators and techniques has been laid in this research, which is not done before. I found that the CEO is a significant initiator in the fraud process, but no relation is found regarding the techniques in the first stage. In the third stage, the institutionalization stage, I found that fictitious revenue as fraud technique is used where the sales force is involved and the finance functions and disclosure failures are significantly together present in the third stage of the fraud. Lastly, the results show that lying to the auditor is significantly started in the institutionalization stage.
Keywords: financial statement fraud, fraud techniques, fraud, fraud process, perpetrators, collective corruption, fraud stages
Name: Thomas Herbrink Student number: S2743205 Address: Vondellaan 304
Postal code: 9721 LM Groningen Phone number: 0611500986 Email: t.herbrink@student.rug.nl Supervisor: dr. K. Linke
Word count: 13.777 (excl. tables and references)
University of Groningen
Faculty of Economics and Business January 2019
2 Table of Contents
I. Introduction --- 3 II. Theory and background --- 6
2.1. Fraud Triangle 6
2.2. Process Model of Collective Corruption 7
2.3. Conceptual Model 9
III. Hypothesis development - --- 9 IV. Research design --- 16
4.1. Research approach 16
4.2. Data collection and sample 17
4.2.1. Source 17
4.2.2. Sample and timing 17
4.3. Variables 18 4.3.1. Research variables 18 4.3.2. Variable scheme 20 4.3.3. Control variables 21 4.3.4. Empirical method 23 V. Results --- 23 5.1. Descriptive statistics 23 5.2. Hypothesis testing 28 5.2.1. Hypothesis 1 28 5.2.2. Hypothesis 2 30 5.2.3. Hypothesis 3 31 5.2.4. Hypothesis 4 32 5.2.5. Hypothesis 5 34 5.2.6. Hypothesis 6 35
5.3. Testing the control variables 36
VI. Discussion of results --- 39 VII. Conclusion --- 41 7.1. Findings 41 7.2. Limitations 42 7.3. Theoretical implications 42 7.4. Practical implications 43 7.5. Future research 43 VIII. References --- 44 IX. Appendix A: Additional tests hypothesis --- 49
3 I. Introduction
A long time ago, in 1948, Edwin Sutherland posited the term white collar crime and claimed that corporate misdeeds have a more negative influence on society compared with street crime, because the trust in organisations and authorities corrodes due to corporate fraud. That corporate fraud is still a current topic in this century appears from the news of the 13th of August 2015 that Royal Imtech B.V. was declared bankrupt (NRC, 2015). Imtech had
to process big losses and moreover, was engaged in financial statement fraud (NRC, 2015). Financial statement fraud is defined as “the intentional material misstatement of financial statements or financial disclosures or the perpetration of an illegal act that has a material direct effect on the financial statements or disclosures” (Beasley et al, 2010, p. 7). At Imtech revenues of a non-existent amusement park were recognized too early (NRC, 2014), losses were shifted by means of accruals and information about setbacks in projects was withheld, implying that incomes gave a false presentation of the economic situation (Financieel Dagblad, 2017). The loss incurring project in Poland was concealed by the German board through understatement of losses. The board members of Royal Imtech B.V. had to monitor the German board members, however, the fraud was concealed for a long time (Financieel Dagblad, 2017). In the end, the German board members were sentenced to imprisonment and a fine (Financieel Dagblad, 2017) and the CEO of Royal Imtech B.V. is sentenced to a one million euro fine (RTL Nieuws, 2018). This is also indicating that executives at Imtech collude together in this comprehensive fraud.
In the end, collusion in fraud cases is an important issue because almost all recent fraud cases (f.e. Enron, WorldCom, Tyco, HealthSouth) resulted in multiple criminal convictions where besides of the CEO and CFO, more employees were involved (Free & Murphy, 2015). This pressure to involve in fraudulent behaviour could arise because leaders can facilitate corruption through not just condoning, but rewarding fraud involvement or punish those who do not participate in fraudulent behaviour (Brief et al., 2001). However, in many cases the initial misstatement reflects an optimistic bias which is not necessarily intentional (Schrand & Zechman, 2012). The problem is that in the upcoming period, when the optimistic performance goal is not reached, managers are compelled to deliberately misstate the financial statements (Schrand & Zechman, 2012). Furthermore, Ashforth & Anand (2003) explain in their model that because a corrupt decision becomes embedded in structures and processes employees can routinely engage in corruption without experiencing conflicts. When the embedded processes are interdependent, a corrupt process can ignite other processes, which make the whole fraud
4 sustainable (Ashforth & Kreiner, 2002). This may indicate that for example an improper revenue recognition process can affect other processes. So, because corrupt processes can ignite each other and people possibly are routinely engaged in corruption it is very interesting to look which actor is engaged and what fraud technique is used during the time span of the fraud. Practically, for auditors, who should discover material fraud during the audit, it is relevant to know if there are patterns in the time span of the fraud process concerning the perpetrators and the fraud techniques used.
Past research on financial statement fraud is often related to determinants, consequences and trends of financial fraud (Hogan et al., 2008). Regarding the fraud triangle of Cressey (1953), researchers in the past decades try to identify risk factors related to the fraud triangle and investigate their relation with the incidence of fraud (Bell & Carcello, 2000; Hernandez & Groot, 2007). They found evidence that factors as high management turnover, rapid industry growth, declining industry conditions and compensation arrangements tied may serve as fraud red flags. Besides, prior research found evidence that the desire to meet analysts forecast (Kaplan, 2001) and financial distress (Huang et al., 2012) are positively significantly related to the occurrence of fraud.
Another aspect of financial statement fraud is the perpetrator. The CEO and/or CFO were already involved in 80 percent of the fraud cases, measured in 2001 by Beasley. Beasley et al. (2010) highlight that in the last couple of years the percentage of CEO’s and CFO’s involved in financial statement fraud increased, but also the commitment of other employees and outsiders (f.e. auditors). In much cases the CEO and CFO possess power to override the internal controls in the organisation, causing this high percentage (Beasley, 2001). Moreover, Cunningham (2003) is surprised how easily reporting rules can be ignored, causing segregation of duties no longer to be effective. He mentioned that conspiracy of the finance function (CFO and controller) with other functions allows too easy bypassing of internal controls. In combination with the research of Ashforth & Anand (2003), who found evidence that people can routinely be engaged in fraud without experiencing conflicts, it seems very relevant to investigate patterns in the time span of the fraud process regarding the actors, which is not done in prior research.
Next to the actors the fraud method applied may change during the fraud process. According to Linke (2012) there is a large variation of methods used and it is a combination of numerous fraudulent acts, on average 4,5 offences of the law per fraud allegation. There is more research related to the fraud methods, researches often make lists of categories of the most common manipulation methods (Beasley, 1999; Bonner et al. 1998), but how it relates to
5 the time span of the fraud process is not investigated yet. This research takes into account the techniques while researching the relation between actors and the time span of the fraud process.
In the end, research in fraud is done in a wide area. Concerning the perpetrators there is research to involvement of certain functions in a company and the opportunities they have (Beasley et al., 2001; Beasley et al., 2010, Cunningham, 2003). Besides, prior research investigated how fraud becomes embedded in the organisation (Schrand & Zechman, 2012) and how people become involved in the slippery slope (Ashforth & Anand, 2003). However, the relation between actors, techniques and the time span of the fraud process is not researched yet. This research will examine whether there are patterns in the contribution of perpetrators, taking into account the moment of occurrence in the fraud process.
Research Question: What are patterns in the contribution of perpetrators, regarding the moment of occurrence in the process of financial statement fraud?
The contribution of this research is that it investigates if there are patterns in an actor’s contribution on financial statement fraud taking into account the moment of occurrence in the fraud process. The relation with the moment of occurrence in the fraud process is not laid before.
The remainder of this study is organized as follows: In chapter II the theory and background are described, chapter III contains the hypothesis development, chapter IV explains the research methodology, and chapter V provides an overview of the results. At the end of the paper, in chapter VI and VII, you can find respectively the discussion and conclusion.
6
II. Theory and background
2.1. Fraud Triangle
One of the most used theoretical concepts in fraud literature is the fraud triangle of Cressey (1953). Almost all researchers that investigated factors which may influence fraud commitment are referring to the fraud triangle (Farahmand & Spafford 2013). The triangle consists of three important factors that are needed for someone to commit occupational fraud. These three factors are motivation, opportunity and rationalization. Motivation (or incentive) could arise from the perceived pressure of a person for example the pressure to meet analysts forecast (Kaplan, 2001) and / or due to financial distress (Huang et al., 2012). On the other hand, pressures such as gambling habit, drug addiction or the need to sustain a lavish lifestyle could be an incentive to engage in fraud (Roth & Espersen, 2008).
Opportunity is the second factor of the fraud triangle, resulting from circumstances that provide chances to commit fraud. Weaker internal control systems give a perpetrator the opportunity to commit and conceal the fraud (Trompeter et al., 2013). Furthermore, the strength of corporate governance and management oversight from the board or the audit committee are factors that determine the opportunity to commit and conceal fraud (Hogan et al., 2008; Trompeter et al., 2013). Besides, the degree of complicated transactions in an organisation influences the opportunities to commit fraud (Lou & Wang, 2011). This is because there is a high inherent risk when the involvement of management judgement and subjectivity is high.
The third and last factor of the fraud triangle is rationalization, where people rationalize their fraudulent behaviour consistent with their personal beliefs (Suyanto, 2009). Management characteristics and personal ethics are important determinants for this factor (Apostolou et al., 2001). Furthermore, management integrity and attitude are major determinants for financial statement quality (Lou & Wang, 2011). They suggest that when more queries about management integrity arise, a firm has a greater probability of fraud.
When one or more of these fraud determinants related to the fraud triangle are present in a fraud organisation, it might influence the choice of a perpetrator to use a specific technique to commit this fraud. Therefore, the fraud triangle provides an explanation whether a perpetrator has the power and motivation to contribute in the fraud process.
However, according to Morales et al. (2013), the pressure part of the fraud triangle is to much focused on the individual. They suggest to not underestimate the influence of group pressure. In the fraud process this seems very important to take into account. Moreover, this finding is endorsed by some researchers. Lou & Wang (2011) found support for their
7 hypothesis that when directors and supervisors experience more financial pressure, lower management have more incentives to manipulate the earnings, resulting in fraud. This implies that when a director or supervisor feels pressure to reach a goal, the other subordinate managers are influenced by the supervisors and directors to misstate earnings. Besides, Cooper et al. (2013) mention that the decision-making process is important in the fraud process. They suggest that organisational or governmental fraud is often a result of the decision of a dominant decision maker (e.g. board, CEO or executive), which is also endorsing the finding of Morales et al. (2013) that group pressure should not be underestimated.
Another approach of the decision-making process is the garbage can model (Cohen et al., 1972). This model suggests that fraud is an outcome of not making conscious decisions and tends to emphasis the role of impulse, error and organisational routines and structures. Furthermore, Cooper et al. (2013) suggest that this connects to approaches that fraud and wrongdoing can be unintentional, inadvertent or based on the complexity of technical solutions. For example, Vaughan (1999) found that accidental fraud can evolve over time as normalization of deviance.
2.2. The Process Model of Collective Corruption
Next to the fraud triangle of Cressey (1953), which is mostly individual based (Morales et al., 2013), the theory of Palmer & Maher (2006) is used, which is based on another insight of organisational crime, namely collective corruption.
The dominant approach to explain organizational crime contains two related assumptions. The first one is that people make discrete considerations to embark on wrongful courses of action and second, that people develop positive insertions toward wrongful courses of action before embarking on them. These positive insertions imply that people make both conscious and unconscious considerations regarding the costs and benefits of engaging in wrongful courses of action. However, some theorists think that people engage in normative assessments, conscious or unconscious, before they decide to embark on wrongful actions. The normative assessments are influenced by internalized norms, values, and beliefs.
On the other hand, the theory of Palmer & Maher (2006) in explaining organisational crime, is based on the alternative approach. It assumes that wrongdoers sometimes engage in wrongful courses of actions without making single discrete decisions based on cost-benefits calculations or normative assessments. Instead of this, they engage in wrongful courses of action as a result of a process.
8 One expression of the alternative approach is based on Brief et al. (2000) and Ashforth & Anand (2003), it is an implicitly elaborated model, independent from each other. In the
initiation stage top managers make mindful and rational cost benefit calculations and/or
normative assessments, which is in line with the dominant approach. After these considerations they decide to authorize wrongdoing or not. The second stage is the proliferation stage. The initiators who made the mindful decision to start wrongdoing are using their formal authority to direct subordinates to implement wrongdoing. Ashforth & Anand (2003) suggest that commitment of subordinates to the top managers is achieved by doing small requests and through providing them a rational. However, the rational is often developed after the course of action is initiated (Staw, 1976). That implies that organizational participants who are subject to the influence of formal authority and are engaged in the commitment process can be said to have acted mindlessly (Cialdini, 2001).
The third stage is the institutionalization stage, in which wrongful courses of action are embedded in the structures of the organisation and cultures. The last stage is the socialization
stage, where new organization members are automatically involved in the techniques and
attitudes that result in wrong behaviour. Organisational structures consist among other things of division of labour, routines that guide task performance and incentives. Organisational cultures imply norms, values and beliefs which influence normative assessments. Therefore, in these stages lower-lever participants rationality is limited regarding wrongful actions. Lastly, the situation no development is added to this model. No development is defined as the fraud process where only c-level managers are engaged, where the fraudulent behaviour is not diffused downwards. However, this does not impact the time span of the fraud. This is in line with the process model of collective corruption (Palmer & Maher, 2006), who argue that the initiator is in the top of the organisation (table 1).
9 As mentioned before, the fraud triangle theory can be used for specific individual-based explanations. As appears from the conceptual model, it provides an explanation
whether or not a perpetrator has the power and the opportunity to contribute with a technique in a specific fraud phase.
Second, the process model of collective corruption (Ashforth & Anand, 2003; Brief et al., 2000; Palmer & Maher (2006)) provides the subdivision of the fraud stages. Besides, the social processes where the model is based on explain the group processes related to stages. These two theories combined, provides substantiation for the expectations of a contribution of a perpetrator in a specific stage.
2.3. Conceptual Model
III. Hypothesis development
According to research of the Association Certified Fraud Examiners (ACFE) (2018) the longer the fraud scheme is, the bigger the median loss. Furthermore, when the median loss is bigger more people are involved and higher functions in the hierarchy of the organization are involved (ACFE, 2018). However, the results are based on various types of fraud in organizations and it indicates the importance of the function and number of actors in fraud
10 schemes. They found that in 89 percent of the financial statement frauds, the CEO and/or CFO were engaged in the fraud process.
To start the slippery slope executives need an opportunity (Cressey, 1953). Cunningham (2003) concludes that it is very easy to circumvent internal financial reporting rules. Leaders have the right to shape the environment lower in the hierarchy via punishing and rewarding (Brief et al., 2001) and obedience pressure may lead managers lower in the hierarchy to make inappropriate choices and decisions (Morales et al., 2013; Davis et al. 2006). However, Linke (2012) found that the CEO is not dependent on the CFO or Financial Middle Managers, implying that the CEO is powerful enough to choose and influence required fraud participants. Linke (2012) found that the CFO and other c-level managers significantly work together with middle management. It may imply that the CFO and other c-level managers use their power on middle managers to reach the finance department. However, it suggests that they are not as powerful as the CEO. On the other hand, Cunningham (2003) concludes that this conspiracy between high level management and lower level management facilitates the circumvention of the procedures which had to prevent fraud. Thereby, CEO’s connections with other top executives increase the likelihood of fraud because the executives together can easier coordinate corporate fraud throughout the organization (Khanna et al., 2015).
Besides, the pressure of higher management to start the slippery slope is needed. Earnings management literature describes that the responsibility of executives to meet or beat analysts’ expectations or forecasts resulted in manipulations (Bartov et al., 2002). Furthermore, the CEO more often has performance-based rewards, such as variable compensation and stock options (Goldman & Slezak, 2006; Efendi et al., 2007). Moreover, an analysis of Cormier et al. (2016) suggest that financial misreporting is positively related to CEO power because governance mechanisms seem to be less effective. Furthermore, regarding the “tone at the top”, the CEO is considered to set and create the corporate ethical climate (Castellano & Lightle, 2005; Center of Audit Quality, 2010). Therefore, the expectation is that chiefly the CEO influences peers. Taken together, the CEO often exerts pressure, has the opportunity, alone or together, to coordinate fraud through the organization. Therefore, the expectation is that the CEO is the initiator in the fraud process.
Schrand & Zechman (2012) researched that executives overconfidence, intentional or unintentional, can cause an initial misstatement because overconfidence of executives is associated with optimistic biases (Weinstein & Klein, 1996). This endorses the rationalization part of the fraud triangle, since management characteristics are an important indicator of rationalization (Apostolou et al., 2001). In this case, they rationalize their optimistic behaviour
11 based on their norms. When an optimistic bias turns out to be an initial misstatement in the next year, managers have a choice. On the one hand they have the option to reveal the initial misstatement and on the other hand to continue misstating in greater amounts (Schrand & Zechman, 2012). The expectation is that to conceal the optimistic biases the initial misstatement is improper timing of revenue. When the CEO forecasts too optimistic earnings, he or she probably would like to conceal it by accelerating revenue recognition. Following the model of Schrand & Zechman (2012), the optimistic allocation of revenues can gradually merge into fraudulent behavior. Therefore, the expectation is that the initial misstatement is improper timing of revenues.
Moreover, in the process model of collective corruption the expectation is that the CEO initiates the fraud because they are able to make rational cost-benefits calculations and subsequently authorize wrongdoing (Palmer & Maher, 2006). Besides, they have the opportunity and pressure to commit fraud (Cunningham, 2003; Bartov et al., 2002). In the end, the expectation is that the CEO in the organisation initiate the fraud process and improper timing of revenue is used.
H1: In the initiation stage, the CEO is involved, and improper timing of revenue is used.
In the proliferation stage of the process model of collective corruption the formal authority is used to diffuse the misstatement as correct (Palmer & Maher, 2006). The c-level managers are able, alone or together, to coordinate the fraud throughout the organization (Khanna et al, 2015). They have the power to shape the environment by rewarding and punishing (Brief et al., 2001). People who occupy roles have professional relationships, which provides them guidance via communication channels (Palmer & Maher, 2006). Via these channels, professionals can facilitate organizational functioning by providing them the most relevant information. However, this information channel can also facilitate wrongdoing (Palmer & Maher, 2006). Therefore, misstatements can be diffused as rational and correct. Wrongdoers who then embark in wrongful actions might not experience cognitive dissonance because they fail to retrospectively evaluate their behavior from a cost benefit analysis or normative assessment (Palmer, 2008). Therefore, people embarked in wrongful behavior may still believe that they behave according to the norms of the company. However, Aronson (1973) remarks that the cognitive dissonance is depending whether their identity is linked to their best interest or behaving in a socially responsible way.
12 Hartmann & Maas (2010) found that social pressure increases the inclination to create budgetary slack, implying that social pressure influences the decisions of middle managers. Moreover, Davis et al. (2006) researched that management accountants give in to obedience pressure, despite the present awareness of the ethical conflict. Furthermore, Linke (2012) found a significant relation between the involvement of Financial Middle Manager (FMM) and the CFO as well as other C-level managers in financial statement fraud, implying that they collude. No relation was found between the CEO and the FMM. However, the influence of C-level managers on the middle managers is very plausible. Financial middle managers, according to Burns & Baldvinsdottir (2007), follow the role model of CFO’s and assist them often with business-partnering, so the chance that they engage only in the institutionalization stage of Palmer & Maher (2006), is smaller. Therefore, the expectation is that the middle managers engage in the proliferation stage, before the fraudulent behavior is embedded in the organizational structure and routines.
Regarding the techniques, improper timing recognition is expected to be primarily used because it allows optimistic forecasts to be met. The expectation is that financial middle managers conceal the misstatement of improper timing of revenue. For example, the FMM is responsible for closing the books and bill sales that are not shipped yet (Linke, 2012). These inaccuracies are more likely to be forced upon them and after that, the sales department engages more mindlessly because it becomes normal to assign transactions to books that had to be closed already. In the end, financial middle management is needed to circumvent internal control procedures. Therefore, the expectation is that these functions engage in the fraud process in the proliferation stage of the process model of Palmer & Maher (2006), because their authority is needed to instruct the lower level employees to engage in wrongdoing.
H2: In the proliferation stage, financial middle managers are involved, and improper timing of revenue is used.
Keeping in mind that formal authority by doing small requests to subordinates is an important influence process (Ashforth & Anand, 2003) it is likely that when the fraud intention is diffused from above, the sales force is engaged in the fraud process. In the example of sales that are billed but not shipped, the middle managers recognize the transaction as wrongly timed, but still approve it. By asking the sales department to enter and start processing the transaction, they often give a rational for entering the false transaction (Ashforth & Anand, 2003). When this practice becomes habit, they become taken-for-granted and enacted mindlessly because
13 the individuals perform the corrupt actions without giving a significant thought to the reasons of this action (Ashforth & Fried, 1988; Brief et al., 2001; Ashforth & Anand, 2003). When this is the case the false transactions are entered mindlessly and the fraud sustain the turnover of generations of employees because the fraudulent behavior is now embedded in the organization structures and culture, which is the socialization and last stage of the process model of collective corruption (Palmer & Maher, 2006). Especially when both high and middle level managers are involved in the fraud process, it is likely that the organizational structures and internal control procedures are shaped in such a way that fraudulent actions by the sales force becomes routine (Ashforth & Anand, 2003; Palmer & Maher, 2006).
When fictitious revenue fraud, part of improper revenue recognition, is committed with dispersed methods throughout the organization there is often a lack of documented fraud files which makes it very hard to detect the fraud (Jans et al., 2011). Therefore, the more people are involved in the fraud process the longer the fraud process and the bigger the loss (ACFE, 2016). When the fraud is embedded in the organization, there is a general absence of supervised data sets on transactional basis, meaning that the internal controls (f.e. segregation of duties) are circumvented (Jans et al, 2011). As mentioned earlier, to circumvent the internal controls the c-level managers need to collude with middle managers to ensure that the sales department is engaged in the fraud process (Palmer & Maher, 2006). In the process the typical revenue related false transactions are carried out by sales personnel. (Linke, 2012). Following the research of Schrand & Zechman (2012), initial misstatements in timing can evolve into bigger misstatements. The expectation is that this will increase the pressure of recognizing fictitious revenue to conceal this increasing amount of improper timing of revenue. Sales personnel, who are responsible for entering invoices, are for circumventing segregation of duties needed to enter false revenues in the books. As mentioned before, with providing rationales to the sales personnel from above wrongdoing can become habits and can be embedded in the organization. (Ashforth & Fried, 1988; Brief et al., 2001; Ashforth & Anand, 2003).
In the end, the expectation is that fictitious revenue recognition is occurring in the institutionalization stage where the sales force is involved because, higher functions have formal power to influence subordinates and shaped the organizational routines and structures and for fictitious revenue the sales force is needed to circumvent internal controls.
H3: In the institutionalization stage, fictitious revenue as a technique is used where the sales force is involved.
14 The CFO is the responsible when it comes to the finance organization and has a notable influence over the company’s reported financial results and information (Ge et al., 2010). The responsibility and function of finance functions consists logically of compiling of actively reviewing the financial data to ensure the correctness of the financial reports, and therefore are responsible for the disclosures. However, Linke (2012) found that the CEO does not always need help of the CFO or financial middle managers, this research suggests that in case of disclosure failures they are needed, because it is their responsibility. So, in the case the optimistic bias resulted in a misstatement the CEO may want to conceal the misstatement by a wrongful disclosure. For example, when there was a wrongful timing of revenue, the CEO does not want to reveal this and therefore would like to provide wrong information to the public. Therefore, he or she need the help of the CFO and FMM, because they are responsible for fair financial statements. In the end, expected is that due to small requests from the CEO and the opportunity to manage disclosures, that later in the fraud process disclosure failures arise and the finance functions involve.
The small, rational, requests from above might indicate the proliferation stage. However, in the institutionalization stage, where fraud becomes stable and embedded in the organization (Palmer & Maher, 2006), the fraud has developed over time throughout the organization. Although acknowledged is that finance functions can be involved earlier in improper revenue recognition, expected is that the disclosure failures arise more often in the institutionalization stage of the fraud, because the pressure to conceal other misstatements with wrong disclosures increase over time. Therefore, the expectation is that finance functions (e.g. CFO and FMM) engage together with disclosure failures in the institutionalization stage.
H4: In the institutionalization stage, finance functions are involved, and disclosure failures are used.
When revenues are improper and the disclosures in the annual reports are wrong, it influences the work of the auditors, since they would like to determine whether the financial statements give a true and fair view of the company’s economic situation. Lennox & Pittman (2010) found that the presence of big four auditors is associated with a higher quality of financial statements, which is indicating that the work of the auditor has a significant influence on the quality of financial statements. However, Beasley et al. (2010) report a trend that increasingly external parties are engaged in wrongful behavior of an organization including the auditor. For example, Markelevich & Rosner (2013) found that SEC-sanctioned fraud firms
15 paid, on average, significantly higher total audit fees relative to non-SEC sanctioned control firms. Their explanation is that fees strengthen the economic bonds, implying that auditors can be influenced.
When the executives are engaged in the fraud process, the duration of the fraud scheme is 24 months on average. When they are not engaged in the fraud process, the duration of the fraud is about a half of that (ACFE, 2018). Besides, in the same report it becomes clear that the impact of the fraud is bigger when the owner or executive is one of the perpetrators in wrongdoing (ACFE, 2018). When the auditor has detected the fraudulent behavior, the duration is about 23 months (ACFE, 2018). It is expected when the impact of the fraud is bigger, the auditor takes note of it and therefore lying is more often required. Related to the process model of collective corruption (Palmer, 2008), fraud can be embedded in the organizational structures and can be normalized. It is often very hard to detect fraud when it is embedded with dispersed methods throughout the organization, because there is a lack of documented fraud files (Jans et al., 2011). An explanation might be that firm members provide wrong information to the auditors. Therefore, the expectation is that lying to the auditor starts more often in the institutionalization stage of the fraud.
H5: In the institutionalization stage, lying to the auditor as a fraud technique starts more often
than in the other stages.
In the past, about the 40s, the historical cost was the only proper valuation base for assets on the balance sheet, where the historical purchase amount exactly is the valuation of that asset on the balance sheet (Ratcliffe & Munter, 1980). However, according to Ratcliffe & Munter (1980), during the twentieth century the current replacement costs method was gradually more often used because the FASB (1978) requires that the current replacement costs were disclosed, regardless whether the valuation on the balance sheet was based on historical costs or not. However, since 2016 only the fair value is allowed by the legislators (NBA, 2016). That means that the current fair value is the current purchase price deducted by the cumulative depreciations while the current replacement costs were the replacement costs for an asset in the same condition and lifespan. The current purchase price is based on taxations or similar assets on the market. However, the value of the assets is debatable, because market prices are difficult to trace (Krumwiede, 2008). Therefore, the fair value method is sensitive to manipulations because revaluations are subject to managerial discretion (Chen et al., 2011). In the end the c-level managers are responsible and have the opportunity to override subordinates, meaning that
16 these managers use their discretion to value the assets, or at least have the opportunity to change something in it. In contrast to improper revenue recognition, the lower officials in the organization are not needed, since they do not influence the asset valuation principles. Furthermore, the mean cumulative misstatement of asset valuation fraud is 226 million dollars, while revenue fraud misstatement mean is about 455 million dollars (Beasley, 2010). Since the ACFE (2016) researched that the bigger the amount of fraud, the more people are involved, the expectation is that less people are involved in asset valuation fraud. Therefore, expected is that asset valuation fraud is less embedded, because the lower officials are not needed in asset valuation fraud. In short, the expectation is that in cases where the higher officials are only involved, asset valuation is more often the fraud technique used relative to improper revenue recognition.
H6: When higher officials are only engaged in the fraud process, asset valuation fraud is used as a technique.
IV. Research design
In this section is determined how the proposed research will be executed. The research approach and source will be discussed, as well as the sample and the research and control variables.
4.1. Research approach
A quantitative research approach is appropriate since it is independent from the perpetrators. The advantage of qualitative research, for example performing interviews, is that it will provide information about fraud cases more in depth, which make data less dependent on interpretation and more reliable in that sense. However, fraud is a sensitive topic and therefore own interests of perpetrators in interviews make the data less reliable. Moreover, perpetrators of fraud cases are less available to interview which make the results less generalizable. Regarding financial statement fraud, quantitative data is more reliable and generalizable and therefore more appropriate as research approach for this topic. Therefore, it is not surprising that quantitative data is most common used in financial statement fraud research (Bonner et al., 1998; Beasley et al., 2010; Dunn, 2004).
17 4.2. Data collection and sample
4.2.1. Source
The fraud firms are originating from the United States. US-firms are chosen because the information is public available in complaints and there are more fraud cases in the United States (Beasley et al., 1999; Linke, 2012). The database used in this research is the SEC-database, which consists of, among other things, litigation releases and complaints disclosed by the U.S. Securities Exchange Commission (SEC), which is the regulator of the United States stock exchange. Next to supervision over the stock market, they check whether stock-listed firms comply with financial reporting regulations (Beasley et al., 2010). However, the SEC database has some limitations. The SEC litigation database may be biased due to enforcement strategies conducted by the SEC staff (Kedia & Rajgopal, 2011). Beasley et al. (2010) acknowledge these limitations and point out an additional limitation: the majority of the civil actions are settled by defendants without admitting or denying the SEC allegations. All in all, there are some limitations but many researchers (Bonner et al., 1998; Beasley et al., 2010; Linke 2012; Farber, 2005) acknowledge that this is the best available database to get an objective insight in financial statement fraud.
The SEC initiates civil actions which are, among other things, announced in an Accounting and Auditing Enforcement Release (AAER). Furthermore, investigations of financial statement fraud are documented in a complaint or/and a litigation release. In this research both AAER’s and litigation releases with or without complaints are used. These documents contain also information about settlements of SEC investigations as well as information about sanctions, which are also used by Beasley et al. (2010). Hand-collecting data out of the SEC litigation releases and AAER’s is appropriate since it is independent from the perpetrators and the result of thorough SEC investigations, because the SEC guarantees its completeness (Linke, 2012).
4.2.2. Sample and timing
To decide whether allegations are an appropriate fraud case for this study the earlier mentioned definition of fraud of Beasley et al. (2010, p.7) is used: ‘Financial statement fraud is defined as the intentional material misstatement of financial statements or financial disclosures or the perpetration of an illegal act that has a material direct effect on the financial statements or disclosures’. The AAER’s in the SEC litigation database have to be in line with that definition to involve the data from that complaint in this research. There is a possibility
18 that an AAER violates SEC regulations but cannot be classified as financial statement fraud corresponding with the definition of Beasley et al. (2010). These violations are often more related to asset misappropriation instead of financial statement fraud, including bribery, embezzlement or other self-enrichments techniques. These cases are excluded from the data sample. Furthermore, there may be cases, which are excluded, where the misstatement was, during the whole fraud period, unintentional. That means that an initial error or failure has turned out in a material misstatement. This is not including the optimistic bias which is passed into fraud later in the process (Schrand & Zechman, 2012). Besides, cases where only a company is sentenced are excluded, since this research is focussing on perpetrators.
Regarding the AAER’s the original sample includes 215 fraud cases which occurred between 2000 and 2015. The most recent and available period is chosen. Cases only until 2015 are chosen because SEC investigations take some time. Therefore, it is possible that investigations are still ongoing for fraud cases in the most recent years. Important to mention is that the date of the complaint is irrelevant. That means that the fraud actually took place between 2000 and 2015, where the date of the SEC-charge is irrelevant. Included are all the cases where perpetrators are alleged for commitment of financial statement fraud corresponding with the definition of Beasley et al. (2010).
Data-collection is performed by a group of five researchers, who analyzed each part of the 215 fraud cases. First is determined whether the case is suitable and if so, all the variables that all five researchers need were hand-collected in a data sheet. In advance, clear agreements have been pointed out regarding criteria for the variables. I have thoroughly investigated fifty of the fraud cases out of the 215 in the original sample and collected all the data the five researchers need for their research. In the end, the data was merged, checked for inaccuracies and loaded in SPSS.
4.3.Variables
4.3.1. Research variables
The variables used to test the hypothesis includes the Fraud Process (FrP). The variable is measured in terms of dummy variables. The variable measures the moment of occurrence in the fraud process regarding the involvement of a perpetrator or a technique used in the fraud process. This implies that the contribution of a perpetrator or a technique is present in a specific stage in the process model of collective corruption (Palmer & Maher, 2006). The involvement of a perpetrator and the use of a fraud technique are assigned to one of these stages. Since the
19 socialization stage and the institutionalization stage are interwoven in each other these stages are measured as one stage. This information is hand-collected out of the AAER’s and litigation releases with or without complaints, therefore the collection group has set criteria to demarcate the stages. A quarterly distribution has been chosen to pursue the best possible consistency in data collection among the five researchers. The initiation stage is defined as the first quarter after the first fraudulent action took place. The proliferation stage is defined as the second and third quarter after the first fraudulent action took place. The proliferation stage is defined as two quarters because the next step, embedding fraud in the organization, takes some time because procedures and routines need to be changed. Third, the institutionalization stage, is in principle demarcated as the period after the third quarter of the fraud process. Furthermore, for that stage the data collector has to judge whether the fraud is stable enough to determine the institutionalization stage. In the end, the fraud process will be included in the statistical model as a dummy variable whether a specific stage is applicable to a technique or perpetrator.
Regarding the perpetrators, finance functions are defined as the CFO and/or the financial middle managers, where the FMM include business controllers. The sales force is defined as administration employees where it appears that they are busy with administrative tasks regarding sales. The remaining perpetrators are reasonably straight forward. Regarding the techniques, improper timing of revenues includes holding books open after the close, improper recognition of third party transactions such as: bill and hold sales, consignment sales, side letters or other techniques that indicates improper timing of revenues (Linke, 2012). Fictitious revenue is defined as falsification of sales documents, not recorded side agreements with customers, top-side adjustments, round-trip transactions, unbilled receivables or other techniques that indicates fictitious revenues (Linke, 2012). Disclosure failures are demarcated as disclosure failures regarding personal benefits and related party transactions and false press releases (Linke, 2012). Lying to the auditor is defined as providing wrong information to the auditor about information that is relevant for the annual report. Lastly, asset valuation fraud is defined as improper asset valuation or classification (Linke, 2012). For the hypothesis, the values the variables can assume are shown in the table.
Research variables for additional analysis also consists of a combination of perpetrator and technique, which implies that a value 1 is assigned when a combination of perpetrator and technique is present and value 0 is assigned when the combination of perpetrator and technique is not present in the fraud process. Besides, a variable is used for additional analysis to determine whether the CEO is initiator in the fraud process. The variable “PerpCEO” is used, which implies that value 1 is assigned if the CEO is involved in the fraud process and value 0
20 is assigned if the CEO is not involved in the fraud process. This variable is tested with “CEO initiator”. This variable is different from “CEO in the initiation stage”, where the latter means that the CEO is actively involved in the stage and the former one means that the CEO also can only exert pressure on others to commit fraud.
4.3.2. Variable scheme
VARIABLE ABBREVIATION HOW TO MEASURE? TYPE OF
VARIABLE
HYPOTHESIS 1 Pearson-Chi-Square test
Initiation stage / CEO IniCEO Is the CEO involved in the initiation stage of the fraud process? (yes = 1, no = 0)
Dummy
Initiation stage / Improper timing of revenue
IniITR Is improper timing of revenue
used in the initiation stage of the fraud process? (yes = 1, no = 0)
Dummy
HYPOTHESIS 2 Pearson-Chi-Square test
Proliferation / FMM ProFMM Is the Financial Middle
Manager (FMM) involved in the proliferation stage of the fraud? (yes = 1, no = 0)
Dummy
Proliferation stage / Improper timing of revenue
ProITR Is improper timing of revenue
used in the proliferation stage of the fraud process? (yes = 1, no = 0)
Dummy
HYPOTHESIS 3 Pearson-Chi-Square test
Institutionalization stage / Sales force
InstSF Is the sales force involved in the institutionalization stage of the fraud process? (yes = 1, no = 0)
Dummy
Institutionalization stage / Fictitious revenue
InstFICT Is fictitious revenue used in the institutionalization stage of the fraud process? (yes = 1, no = 0)
Dummy
HYPOTHESIS 4 Pearson-Chi-Square test
Institutionalization stage / Finance functions
InstFF Are the finance functions
(e.g. CFO and/or FMM) involved in the
institutionalization stage of the fraud process? (yes = 1, no = 0)
Dummy
Institutionalization stage / Disclosure failures
InstDF Are disclosure failures used
in the institutionalization stage of the fraud process? (yes = 1, no = 0)
Dummy
HYPOTHESIS 5 Non-parametric Chi-Square
test Technique/ Lying to the
auditor
TechLyA Is the technique lying to the auditor used? (yes = 1, no = 0)
21 Fraud Process / Start lying
to the auditor
FrpLyA In which stage is lying to the auditor as technique first used in the fraud process?
Initiation stage = value 1 Proliferation stage = value 2 Institutionalization & Socialization stage = value 3
Categorical
HYPOTHESIS 6 Pearson-Chi-Square test
Fraud process/ No development
FrpND Are only the c-level functions
perpetrator in the fraud case? (yes = 1, no = 0)
Dummy
Technique / Asset Valuation
TechAV Is the technique asset
valuation fraud used? (yes = 1, no = 0)
Dummy
CONTROLVARIABLES Logistic regression
Firm Size (LOG) FSize What is the total value of the
assets? (log transformation)
Numeric
Financial Crisis FinCri Is the fraud period including
2008/2009? (yes = 1, no = 0)
Dummy
Big four auditor Big4A Big 4 auditor? (yes = 1, no =
0)
Dummy
Industry dummy IndDum What industry is the company
part of, based on the SIC-code? (first two digits)
Dummies
SOX-era SOXe Is the fraud period before
2002 or in or after 2002? (before = 0, in and after = 1)
Dummy
4.3.3. Control variables
The first control variable is whether one of the Big Four accountancy firms was auditing the company when the fraud starts. Empirical studies on audit quality provide extensive evidence that Big Five1 public accounting firms are associated with higher quality of financial
statements (Francis, 2004). Big Four accountants have motives to monitor more strictly because the last years the regulation has become stricter and supervision is sharpened to increase audit quality (Knechel, 2016). Moreover, Lennox & Pitmann (2010) found that the incidence of fraudulent financial reporting is significantly lower for Big Four clients and that the external monitoring is significantly better when a Big Four accountancy firm is auditing the client. Expected is that the better monitoring influences the development of fraud, so Big Four auditor is included as a control variable.
The second control variable is the size of the fraud company. Beasley et al. (1999) found that smaller companies, defined as under $50 million of revenues, have a weaker level of monitoring of the audit committee and with regard to the control environment, the internal
1 Francis (2004) refers to Big Five auditors because their sample period is between 1981 and 2001, which precedes the demise of Arthur Andersen, a former part of the Big accountancy firms.
22 controls are weaker, the segregation of duties are less and the financial expertise available is less to. When the segregation of duties are weaker, it will influence the fraud process in terms of perpetrators fraud opportunities. Therefore, firm size is included as control variable. Firm size is measured as a log transformation of the total assets to standardize the values. Dang (2015) found that firm size in terms of total assets are less sensitive to market conditions than revenues. The log of total assets is even used before by Schrand & Zechman (2012), which research was a motivation to start this research. Furthermore, since the research includes fraud firms, the total assets value before the start year of the fraud is taken to prevent the effect of manipulation.
The third control variable is derived from a subsequent study of Beasley et al. (2000), they researched industry differences regarding fraud methods. They found that fraud methods substantially vary among industries. Technology companies inflate more often the revenues and in the financial industry asset valuation is more difficult and therefore a bigger risk for fraud. The report of the ACFE (2018) endorses Beasley’s (2000) study with stating that the amount of fraud differs among industries as well as the methods used. They found, in line with Beasley (2000), that in the banking and financial services industry improper revenue recognition is less likely than in the manufacturing industry. In short, industry is included as a control variable to control for this effect on the fraud process. In the logistic regression, the industries are categorized in main categories with dummies. The division can be found in the results section.
Fourth, the financial crisis is added as a control variable. Since the sample period is 2000–2015 it includes the financial crisis. Huang et al. (2012) found that the companies in financial distress feel the pressure to commit fraud and therefore can influence the start of the fraud process and the pressure to engage in fraud. Logically, in the financial crisis the chance to get into financial distress is bigger. Therefore, including the financial crisis as control variable smooth out this effect. The financial crisis is defined as the years 2008 and 2009, where the economic decline was the greatest in the U.S. (Business Insider, 2015).
Lastly, the Sarbanes-Oxley legislation, introduced in 2002, influences the fraud process. For example, the legislation introduced the CEO certification requirement to increase the responsibility and accountability of the CEO for the financial statements (SOX, section 302). Besides, other studies review the usefulness of the SOX legislation as fraud prevention measure and analyse the impact in the pre and post SOX era (Romano, 2005; Feldmann and Read, 2010). Moreover, Beasley et al. (2010) mentioned that their post-SOX sample was to small but acknowledge that it influences fraud techniques. Since the sample includes more years after
23 2002, a control variable for the SOX-legislation is included. The measurements are indicated in the variable scheme in 4.3.2.
4.3.4. Empirical method
However, since chi-square tests with only two categorical variables are used there is not determined a variable to be dependent or independent. With the chi-square tests is determined whether there is a relationship between the two variables and by interpreting the crosstab SPSS provides, the conclusions can be formulated.
The conditions to apply a chi-square test are first, that at most 20 percent of the expected counts are less than 5 and second, the minimum expected count is 1. The control variables are tested with a logistic regression, since the control variables are both nominal and numeric. For hypothesis 5, the conditions are not fulfilled, since lying to the auditor cannot be used in a stage when it is not used at all. That implies than one of the four counts are zero. Therefore, a non-parametric chi-square test is conducted, where only the cases where lying to the auditor used are included, implying that this variable is constant.
V. Results
5.1. Descriptive statistics
24 In the table above the frequencies of the involvement of perpetrators specified by stage are depicted. The total amount of sufficient fraud cases out of the complaints is 186. In respectively 102 and 120 of the fraud cases the CEO and CFO are involved in the fraud process at any time. The other perpetrators are involved in around the 50 cases out of the 186, implying that the CEO and CFO are the main perpetrators in fraud. When you have a look at the specification per stage, the CEO and CFO are mostly involved in the initiation stage, respectively 98 out of 102 and 109 out of 120 times. There are only 4 CEO’s and 11 CFO’s who join the scheme in the proliferation or institutionalization stage. However, in the subsequent stages the number is decreasing for both the CEO and CFO. This is caused due to the fact that some fraud schemes end earlier when it only covers a few quarters or that a perpetrator drop off the scheme. This is the reason the maximum indicated in the second column is not reached. The other C-levels are mostly involved in the proliferation stage, however, the differences in stages for this group is little. For the Financial Middle Manager (FMM) an ascending number of times involved is shown in the table, implying that during the fraud they get involved. The same applies for the Vice Presidents (VP’s). For the sales force the differences in the ascending number of times involved is bigger than for the FMM and VP’s. This suggests that the sales force get involved later on in the fraud process more often. The Other category consists of among other things general presidents, auditors, vendors and council members. Because this research does not focus on these people they are not specified in this table.
Table 3: Frequencies techniques in stages.
In the table above frequencies of the techniques used specified by stage are depicted. The most used techniques at any time in the fraud process are respectively disclosure failures and lying to the auditor with 113 and 102 times out of the 186 frauds. However, since improper revenue recognition is specified into four subcategories, improper revenue recognition as a main category is the most used technique because in total 163 times a subcategory of improper revenue recognition is used. Regarding the specification per stage, improper timing of revenues
25 is mostly used in the proliferation stage, however the differences are very small. The same applies for improper expense recognition techniques. Fictitious revenues are about equally divided over the fraud process. The maximum in depicted in the second column is not reached. First, for example regarding fictitious revenue, 64 times of the 79 times the technique is first used in the initiation stage. In the subsequent stages there are 15 times that the technique is used for the first time. However, it is possible that the fraud scheme ends earlier than the second or third quarter. Moreover, it is possible that use of a technique ends, while the fraud scheme continues.
For the techniques disclosure failures and lying to the auditor you see a strong ascending amount of times used over the stages, implying that these techniques are used as an extra technique later on in the fraud process. Here, the ending fraud period or drop off of a technique is not an issue because the number is still ascending, which endorses the conclusion that these techniques are used later in the fraud process anyway. For asset valuation fraud you see an ascending amount of times used to, however, these differences are very small. In the end, improper timing of revenue and fictitious revenue are more initiating techniques, while disclosure failures and lying to the auditor are more used later in the fraud process.
Table 4: Frequencies of a perpetrator who is involved in a technique
First, from perspective of the perpetrators, the CEO and CFO are about equally involved in improper timing of revenues, fictitious revenues and asset valuation fraud. However, the CFO is more often than the CEO involved in improper expense recognition, disclosure failures and lying to the auditor. The involvement of the Financial Middle Managers is quite equal over all fraud techniques, except from asset valuation fraud. Relative to FMM, the VP is more involved in fictitious revenue, disclosure failures and lying to the auditor, but less in improper expense recognition. The sales force is mostly involved in fictitious revenues, which is logical since they are responsible for booking invoices.
Second, from technique perspective, all techniques except from fictitious revenue recognition are mostly actively used by the CEO and CFO, followed by the FMM and VP.
26 When fictitious revenues are recognized the VP and the sales force are the most involved people after the CEO and CFO.
Concerning table 4 the major conclusion is that for most perpetrators the contribution to a technique is in line with the frequency a technique or a perpetrator is present in the fraud process. Notable is that the finance functions (e.g. CFO and FMM) are relatively more involved in improper expense recognition, and the CFO is relatively more involved in disclosure failures and Lying to the auditor. Furthermore, the sales force is relatively more involved in fictitious revenue and c-levels are involved in most cases where asset valuation fraud is used.
Table 5.1: Descriptive statistics control variables
Table 5.2: Frequency statistics SIC-code industry
Table 5.3: Frequency statistics start year
27 In the tables above, the descriptive statistics of the control variables are provided. The start year of the fraud period varies from 2000 till 2014, where a high percentage of valid fraud schemes start in the first years of the sample period. The sample period includes the introduction of the SOX-legislation, which may influence the amount of fraud cases in the years before 2002. The SOX-era is included as a control variable, where in 30 cases the fraud scheme is started and ended before the SOX-legislation was applicable. The sample period also includes the financial crisis. In 35 of the 186 valid cases, the fraud period includes the financial crisis in the years 2008 and 2009. Regarding the big four auditor variable, 110 out of the 180 valid cases had a big four auditor during the fraud process, which represent 61,1 percent of the cases. In 6 cases the data whether the firm had a Big 4 auditor during the fraud was not available. Third, the industry SIC-codes represent an industry. Most fraud cases are related to the manufacturing industry (SIC codes 20-39, (Siccode.com, 2018)) and specifically SIC-code 73, which represent Business Service providers (Siccode.com, 2018). Lastly, the mean of the total assets value of the fraud firms equal $2.587.009.845 with a standard deviation of $8.389.531.176. The biggest fraud company in this sample is Enron, with an asset value of 67,6 billion.
28 5.2. Hypothesis testing
In this section is pointed out whether the hypothesis are accepted or rejected. For each hypothesis a crosstab is provided together with a chi-square test to determine whether the results are significant.
5.2.1. Hypothesis 1
Hypothesis 1 reads as follows: In the initiation stage, the CEO is involved, and improper
timing of revenue is used.
This hypothesis is tested with two variables, CEO involvement in the initiation stage and whether improper timing of revenue is used in the initiation stage.
a. 0 cells (0.0%) have expected count less than 5. The minimum expected count is 23,66.
Table 6.1: Tests hypothesis 1
From the crosstab appears that in 98 of the total amount of cases the CEO is involved in the initiation stage and in 50 of the 186 cases improper timing of revenue is used in the initiation stage. However, only 25 times both the CEO are initiator and improper timing of revenues is used. Together with the Chi-Square test, which is insignificant at 0.1 level, the first hypothesis is rejected. The asymptotic significance is applicable, since the two conditions pointed out in 4.3.4 are met. Although the hypothesis is rejected additional tests are conducted because in 73 cases the CEO is present in the initiation stage and improper timing of revenue is not used. Therefore, it is interesting to investigate whether there is another technique which is significantly together present in the initiation stage.
First, because the CEO is active in the initiation stage 98 times is tested if the CEO is a significant initiator. The table below reports the results:
29 a. 0 cells (0.0%) have expected count less than 5. The minimum expected count is 39,74.
Table 6.2: Tests hypothesis 1
As depicted in the crosstab above, the CEO is initiator in 98 of the cases, which is significant at a 0,00 level. This implies that the hypothesis is not rejected because the CEO is not the initiator but due to the fact that improper timing of revenues is not significantly used in the initiation stage in combination with the CEO as initiator. Since one of the values in the table is 0, the reliability has to be checked. However, there are no cells with an expected count less than 5 and the minimum expected count is 39,74. Therefore, the test is deemed to be reliable.
To further analyse the insignificant result the same tests are conducted for fictitious revenue and improper expense recognition instead of improper timing of revenue. Techniques that are tested are improper expense recognition and fictitious revenue because these techniques are, appearing from table 3, often used in the first stage of the fraud. The tables 6.3 and 6.4 can be found in Appendix A.
Even though an equal insignificant result is found for fictitious revenues in the initiation stage the result for improper expense recognition in the initiation stage is significant at a 0.1 level. However, there are less cases in which improper expense recognition and the CEO are together present than that they are present separately. Therefore, it cannot be assumed that in the initiation stage the CEO is involved and improper expense recognition is used. In the end, the CEO is a significant initiator in the fraud process, but the technique differs, so the CEO and a specific technique cannot be linked in the initiation stage.
Furthermore, an additional test is conducted with the variable CEO in improper timing of revenues, which is chosen because it is interesting to investigate that when you assume that the CEO is involved in improper timing of revenue, then he would execute it significant in the initiation stage. This variable is tested in two tests with respectively CEO in initiation stage and improper timing of revenue in initiation stage. The results are shown in tables 6.5 and 6.6 in Appendix A. The chi-square tests indicate that both are significant and reliable. This implies for the first table, table 6.5, that out of the 37 times the CEO engages in improper timing of
30 revenue, 35 times the CEO was in the initiation stage. The second table, table 6.6, implies that out of the 37 times the CEO engages in improper timing of revenue, 26 times the technique is used in the initiation stage and 11 times it is not. These tests together suggest that when you assume that the CEO engage in improper timing of revenues, then he or she will conduct it significantly in the initiation stage.
5.2.2. Hypothesis 2
Hypothesis 2 reads as follows: In the proliferation stage, financial middle managers are
involved, and improper timing of revenue is used.
The hypothesis is tested with two variables, FMM involvement in the proliferation stage and whether improper timing of revenue is used in the proliferation stage.
a. 0 cells (0.0%) have expected count less than 5. The minimum expected count is 13,90.
Table 7.1: Tests hypothesis 2
From the crosstab appears that only in 47 times the financial middle manager is involved in the proliferation stage. Out of these 47 times only 18 times the financial middle manager is involved, and the technique improper timing of revenue is used. Besides, the chi-square test is insignificant at 0.1 level, so the second hypothesis is rejected. The asymptotic significance is applicable, since the two conditions pointed out in 4.3.4 are met.
To further analyse the insignificant result some additional tests are conducted to determine if another combination of FMM and technique is present in this stage or another stage. For both variables the stage is changed into institutionalization stage (table 7.3), because appearing from table 2 the FMM is more active later in the fraud process, and the technique is changed from improper timing of revenue into fictitious revenues (table 7.2), where the FMM is active in, appearing from table 4. The tables can be found in Appendix A. However, there is no specific combination of FMM and technique which is found significantly together present in the proliferation stage. This endorses the descriptive table of perpetrators linked to
