FFR

Fraudulent financial reporting (FFR), also known as financial statement fraud or management fraud, is a type of financial fraud that adversely affects stakeholders through misleading financial reports (Elliot and Willingham, 1980). FFR involves the intentional misstatement or omission of material information from an organization’s financial reports (Beasley et al., 1999). FFR, although with the lowest frequency, casts a severe financial impact (Association of Certified Fraud Examiners, ACFE 2008).

FFR can lead not only to significant risks for stockholders and creditors, but also financial crises for the capital market. According to the ACFE (2008), financial

misstatements are the most costly form of occupational fraud, with median losses of $2 million per scheme. FFR, or financial statement fraud, is known as “cooking the books” that often has severe economic consequences and makes front page headlines (Beasley et al., 1999). While ACFE (1998) reported that fraud has become more prevalent and costly, the detection of fraud has been badly lagging. The KPMG (1998) survey found that over one third of fraud cases were discovered by accident and that only 4 percent of cases were detected by independent auditor. When the auditor makes inquiries about fraud-related transactions, he or she is likely to be deceived with false or incomplete information (Weisenborn and Norris, 1997). Though the ability to identify fraudulent behavior is desirable, humans are only slightly better than chance at detecting deception (Bond and DePaulo, 2006) or identifying fraudulent behaviors beforehand. Therefore, there is an imperative need for decision aids of identifying FFR.

More reliable methods are needed to assist auditors and enforcement officers in maintaining trust and integrity in publicly owned corporation.

Most prior FFR-related research focused on the nature or the prediction of FFR.

The nature-related FFR research often uses the case study approach and provides a descriptive analysis of the characteristics of FFR and techniques commonly used. For example, the Committee of Sponsoring Organizations (COSO) and the Association of Certified Fraud Examiners (ACFE) regularly publish their own analysis on fraudulent financial reporting of U.S. companies. Based on the FFR samples, COSO examines and summarizes certain key company and management characteristics. ACFE analyzes the nature of occupational fraud schemes and provides suggestions to create adequate internal control mechanisms. As shown in Table 1, nature-related FFR research often uses case study, statistic or data mining approach to archival data and identifies significant variables that help predict the occurrence of fraudulent financial reporting.

Other nature-related FFR studies focus on the audit assessment and planning (Bell and Carcello, 2000; Newman et al., 2001; Carcello and Nagy, 2004; Gillett and Uddin, 2005).

Table 1. Research methodology and findings in nature-related FFR studies.

Research Methodology Findings

Beasley et al.

(1999)

• Case study

• Descriptive statistics • Nature of companies involved

– Companies committing financial statement fraud were relatively small.

– Companies committing the fraud were inclined to experience net losses or close to break-even positions in periods before the fraud.

• Nature of the control environment

– Top senior executives were frequently involved.

– Most audit committees only met about once a year or the company had no audit committee.

• Nature of the frauds

– Cumulative amounts of fraud were relatively large in light of the relatively small sizes of the

companies involved.

– Most frauds were not isolated to a single fiscal period.

– Typical financial statement fraud techniques involved the overstatement of revenues and assets.

• Consequences for the company and individuals involved

– Severe consequences awaited companies committing fraud.

– Consequences associated with financial statement fraud were severe for individuals allegedly involved.

ACFE (2008)

• Case study

• Descriptive statistics • Occupational fraud schemes tend to be extremely costly. The median loss was $175,000. More than one-quarter of the frauds involved losses of at least $1 million.

• Occupational fraud schemes frequently continue for years, two years in typical, before they are detected.

• There are 11 distinct categories of occupational fraud.

Financial statement fraud was the most costly category with a median loss of $2 million for the cases

examined.

• The industries most commonly victimized by fraud in our study were banking and financial services (15% of cases), government (12%) and healthcare (8%).

• Fraud perpetrators often display behavioral traits that serve as indicators of possible illegal behavior. In financial statement fraud cases, which tend to be the most costly, excessive organizational pressure to perform was a particularly strong warning sign.

Another type of FFR research often uses the empirical approach to archival data and identifies significant variables that help predict the occurrence of FFR. This line of research also inputs these significant variables into fraud prediction models. Such research emphasizes the predictability of the model being used. For example, logistic regression and neural network techniques are used in this line of research (e.g., Persons, 1995; Fanning and Cogger, 1998; Bell and Carcello, 2000; Virdhagriswaran, 2006;

Kirkos et al., 2007). The matched-sample design is typical for traditional FFR empirical studies. That is, a set of samples with fraudulent financial statements confirmed by the Department of Justice is matched with a set of samples without any allegations of fraudulent reporting.

Table 2 summarizes the research methodology and findings of the FFR empirical studies most relevant to our study. The research methodology has shown a trend with an emphasis on the classification mechanization which is used as the decision support information for future risk identification (Basens et al., 2003).

Table 2. Research methodology and findings in FFR empirical studies.

Author Methodology Variable Sample Findings Dechow

• To attract external

financing at low cost was found an important motivation for earnings manipulation

• Firms manipulating

earnings are more likely to have:

- insiders dominated boards

- Chief Executive Officer simultaneously serves as Chairman of the Board Persons

The study found four significant indicators:

• Neural network is more effective

• Financial ratios such as debt to equity, ratios of accounts receivable to sales, trend variables are significant indicators fraud samples, but were equally performed for

• Training dataset: neural network is the most accurate

• Validation dataset:

Bayesian belief network is

network samples the most accurate

Integrated pattern had a wider coverage for suspected fraud companies while it remained lower false classification rate for non-fraud ones Source: (Hsu, 2008; Huang et al., 2011).

As shown in Table 2, Persons (1995) used Stepwise logistic model to found significant indicators relate to FFR. Dechow et al. (1996) used Logistic regression in FFR detection. Bell and Carcello (2000) developed a Logistic regression model useful in predicting the existence of fraudulent financial reporting, and found that the proposed model outperformed auditors for fraud samples, but were equally performed for non-fraud samples.

Green and Choi (1997) applied Back-propagation neural network to FFR detection.

The model used five ratios and three accounts as input. The results showed that Back-propagation neural network had significant capabilities when used as a fraud detection tool. Fanning and Cogger (1998) proposed a generalized adaptive neural network algorithm, named AutoNet, to FFR detection. The input vector consisted of financial ratios and qualitative variables. They compared the performance of their model with linear and quadratic discriminant analysis, as well as logistic regression, and claimed that AutoNet is more effective at detecting fraud than standard statistical methods. Kirkos et al. (2007) compared Decision tree, Back-propagation neural network, and Bayesian belief network in FFR detection and found that Back-propagation neural network is the most accurate method in training dataset, Bayesian belief network is the most accurate method in validation dataset. Hoogs et al.

(2007) applied Genetic Algorithm (GA) in FFR detection, and the performance of GA concluded that the integrated pattern had a wider coverage for suspected fraud companies while it remained lower false classification rate for non-fraud ones.

Humpherys et al. (2010) developed a linguistic methodology for detecting fraudulent financial statements. The results demonstrate that linguistic models of deception are potentially useful in discriminating deception and managerial fraud in

financial statements. Their findings provide critical knowledge about how deceivers craft fraudulent financial statements and expand the usefulness of deception models beyond a low-stakes, laboratory setting into a high-stakes, real-world environment where large fines and incarceration are the consequences of deception. In literature of financial fraud detection (FFD), Ngai et al. (2010) have done a complete classification framework and an academic review of literature which used data mining techniques for FFD. They showed that the main data mining techniques used for FFD are logistic models, neural networks, the Bayesian belief network, and decision trees, all of which provide primary solutions to the problems inherent in the detection and classification of fraudulent data. Huang et al. (2011) used the GHSOM to help capital providers examine the integrity of financial statement. They applied the GHSOM to analysis financial data and demonstrate an alternative way to help capital providers such as lenders to evaluate the integrity of financial statements, a basis for further analysis to reach prudent decisions. Huang and Tsaih (2012) evolved the GHSOM into a prediction model for detecting the FFR. They proposed the initial concept of a dual approach for examining whether there is a certain spatial relationship among fraud and non-fraud samples, identifying the fraud counterpart of a non-fraud subgroup, and detecting fraud samples.

The relevant literatures show that the neural network families have been widely used in many financial applications, such as the FFR detection, credit ratings, economic forecasting, risk management, or other FFD related issues.