Effect of Hexagon Fraud Against Financial Statement Fraud with Company Size as Moderation

This research aims to determine the effect of hexagon fraud with company size moderating its effect on financial statement fraud. Hexagon fraud is measured through six components, namely, stimulus, capability, collusion, opportunity, rationalization and ego. Meanwhile, company size is measured through total assets and financial statement fraud is measured through the Beneish M-Score model. This research is included in descriptive quantitative research. Through purposive sampling, 35 companies were determined as samples for a 5-year observation period from a total population of 113 non-cyclical consumer sector companies as of 2022 listed on the Indonesia Stock Exchange. Data were processed using multiple linear regression. Based on the results of data analysis, it is known that partially hexagon fraud has a significantly positive effect on financial statement fraud, while company size has no effect on financial report fraud. As a moderator, company size can strengthen the influence of hexagon fraud on financial statement fraud. Simultaneous testing shows that hexagon fraud and company size influence financial statement fraud. However, both of them were only able to explain 9% of financial statement fraud with the rest explained by other variables.


INTRODUCTION
Financial reports are a structured presentation of the financial position and financial performance of a company.Financial reports have an important role in the management of every company because they are a means for managers to account for the resources entrusted to them.Statement of Financial Accounting Standards No. 1 states that the presentation of financial reports aims to provide information to the majority of users in making decisions (IAI, 2019).Furthermore, Statement of Financial Accounting Concept No.1 emphasizes that financial reports are addressed primarily to external parties, in this case investors and creditors (FASB, 1978).Therefore, financial reports must be able to correctly present the company's economic phenomena both in numbers and words.
The truth of financial report information is often associated with presentation that is not manipulated because it tends to lead to fraud.Auditing Standard ("SA") 240 defines fraud as an intentional act by one or more individuals in management, those charged with governance, employees or third parties, involving the use of deception to obtain an unfair or unlawful advantage (IAPI , 2021).In more detail, ACFE (2022) states that fraud in financial statements is defined as fraud committed by management in the form of material misstatements that harm investors and creditors where fraud can be financial and non-financial.Thus, manipulation of financial reports is a form of fraud if it is carried out intentionally by violating rules and laws regarding material values for the benefit of management to the detriment of investors or creditors.
Financial statement fraud is one of three types of fraud that is often carried out and is detrimental besides misuse of assets and corruption.The Association of Certified Fraud Examiners in "Occupational Fraud 2022: A Report to Nations" published that there had been 2,110 cases of fraud in 133 countries between January 2020 to September 2021, where financial statement fraud contributed to a total of 9% of all cases.The total losses incurred by these fraud cases reached 3.6 billion dollars with an average loss per case reaching 1.7 million dollars, where losses in 21% of cases were above 1 million dollars.
Based on the description above, financial reports are an important component in company management, both as a form of managerial accountability and as a basis for decision making for external parties.Therefore, financial reports must be free from misleading elements and material misstatements which are part of fraudulent acts.Cases of fraudulent financial reporting also occur in the non-cyclical consumer sector, where this sector is a fairly stable sector in terms of share price developments.In terms of fraud detection, the Beneish model is used because it has been tested and is relatively cheap and simple.At the same time, financial statement fraud cannot be separated from the trigger factors, where hexagon fraud is one of the latest fraud factors from the development of fraud theory.From the research results mentioned above, there are still inconsistencies regarding the significance of the influence of the components that cause fraud on financial statement fraud itself.Thus, the use of a moderating variable, in this case company size, can provide a more diverse dimension and perspective that financial statement fraud can involve various aspects.
Therefore, the author was motivated to conduct research with the title, "The Effect of Fraud Hexagon on Financial Statement Fraud with Company Size as Moderation." Based on the background above, the author identifies that fraud in financial reports is a problem with the explanation as follows: 1. Financial statement fraud is a type of fraud that often occurs and is detrimental and is also found in the non-cyclical consumer sector. 2. The Beneish model is among the tested models that indicate the risk of financial statement fraud.3. Determining factors for fraud, including hexagon fraud, are part of efforts to detect financial statement fraud.4.There are differences between studies regarding the influence of the components in the fraud hexagon on financial statement fraud.5. Company size as a moderation can be used to see the influence of hexagon fraud on financial statement fraud

LITERATURE REVIEW 1. Theoretical basis Agency Theory
Agency theory, hereinafter referred to as agency theory, was put forward by Michael C. Jensen and William H. Meckling in 1976 that a company can be seen as a relationship in a set of contracts (nexus of contracts) between the principal or owner of economic resources and the agent or manager who manages the resources.the.In line with this, Eisenhardt (1989) explains that an agency relationship arises when one or more individuals, in this case the principal, employ one or more other individuals, in this case the agent, to carry out all company operational activities on behalf of the principal, including decision making.Therefore, this agency relationship reflects the separation of functions between the principal and the agent.

Signal Theory
Signaling theory, hereinafter called signal theory, was put forward by Michael Spence in 1973 in his paper entitled "Job Market Signaling".It was further explained that the sender or owner of the information gives a signal or signal that describes the company's condition to the recipient or potential investor.Meanwhile, the theory signals from a financial and investment perspective, as Ross (1977) states that top management with more complete information is encouraged to convey this information to potential investors so that the company's shares increase.Meanwhile, Brigham and Houston (2018) state that the company's future prospects are a consideration for management to provide signals to potential investors.Therefore, this signal theory is related to the initiative of management to convey information that emphasizes the decisions of parties outside the company.

Fraud Concept Definition of Fraud
Fraud, hereinafter referred to as cheating, basically has a broad meaning including false statements, dishonesty and deception.In terms of fraud that occurs in companies, Dyck et al (2023) state that fraud is a false statement by making untrue statements regarding material facts, not stating material facts including not stating how the statement was made.Fraud in the form of fraud is equated with false statements as Zhu et al (2021) state that fraud refers to the misrepresentation or truth or concealment of material facts to encourage other people to act to their detriment.Meanwhile, Sanchez et al (2021) state that fraud is fraud to gain illegal profits.

Fraud Hexagon
In Auditing Standards ("SA") 240 concerning Auditor Responsibilities related to Fraud in an Audit of Financial Statements, it is stated that fraud is an intentional act that causes material errors in the presentation of financial statements.In this context, fraud in financial reports ignores legal aspects (IAPI, 2021).Meanwhile, ACFE (2022) states that fraud in financial reports is defined as fraud committed by management in the form of material misstatements of financial reports that are detrimental to investors and creditors, where fraud can be financial and non-financial fraud.According to Statement of Auditing Standards (SAS) No. 99 financial statement fraud is an intentional act that causes material misstatement in the audited financial statements in the following forms: a. Manipulating, engineering, falsifying or changing supporting documents as well as manipulating financial records in the preparation of financial reports.b.Omission, intentional omission of accounts, events, transactions, or information that form the source of financial statements.c.Deliberately incorrect application of accounting principles where they are used to disclose, present or measure economic events.

Beneish Model
The Beneish model was proposed by Messod Daniel Beneish in 1999 through his examination of 74 companies that were proven and indicated to have carried out earnings manipulation and of 2,332 other United States public companies from 1982 to 1992.These companies' financial reports were analyzed using eight indices designed to capture there are deviations in the form of profit manipulation where each index uses data comparisons between one period and the previous period (Beneish, 1999).

Company Size
The main goal of the company is to maximize the wealth or value of the company.Maximizing the value of a company is very important for a company because by maximizing the value of the company, one also maximizes shareholder wealth.Company size can generally be interpreted as a company scale where the size of the company can be classified according to various aspects, including total assets, average total assets, share market value, total sales/income, average sales, total profit, number of employees and others.In line with this, Irawan et al (2022) state that the term company size refers to a quantity that is specifically stated to measure business scale, such as total assets, total sales, market capitalization, total income and total sales.Furthermore, company size can be measured by transforming the total assets owned by the company into a natural logarithm.Meanwhile, Karlsson (2021) states that company size is an important element in viewing company performance because it reflects more complex policies.Therefore, company size can be used as an element in various studies, including in relation to financial statement fraud.

THEORETICAL FRAMEWORK
Fraud hexagon with financial stability as a pressure factor because management is considered to be trying to present financial reports with asset values that are experiencing growth.So, it is possible to obtain growth in asset value through presentations that involve fraud.Next, the change of directors is a capability factor due to a stress period so that the new directors with their interests are deemed capable of exploiting loopholes to commit fraud or the new directors are incompetent in detecting fraud.Next, market performance as a collusion factor because the value of a company's shares on the market is lower than its book value is considered to be a conspiracy in determining the share price.So, with a lower share price it is possible to obtain it through fraudulent deals.Next, supervision is less effective as an opportunity factor because the higher composition of independent commissioners in the board of commissioners is considered by outsiders to carry out better supervision so as to reduce fraud.Next, changing auditors is a justification factor because a new auditor is considered to reduce the possibility of detecting fraud.Next, the dual position of CEO is an ego factor because it reflects superiority and dominance which prioritizes personal interests so that they are not optimal in their duties.So, companies with multiple positions allow fraud to occur.
The size of the company can strengthen the factors in hexagon fraud because with a larger size, the pressure on management is also greater, the capabilities of directors are required to be higher, share price determination becomes more flexible, supervision becomes more complex, changing auditors becomes more reasonable, and holding multiple positions.cannot be suppressed, which ultimately increases the risk of fraud in the financial statements themselves.Therefore, the relationships between the variables described above are described in the model as follows:

Hypothesis Development Hexagon Fraud on Financial Statement Fraud
Asset growth may be generated through fraudulent actions due to the drive felt by management to demonstrate that financial conditions appear stable.This statement is in line with research by Achmad T (2022), Aviantara (2021) and Rizkiawan & Subagio (2022) which states that financial stability has a significant positive effect on fraudulent financial statements.Furthermore, changing directors will put the company in a stress period which is seen as a gap, so that perpetrators with their abilities can take advantage of this gap to commit fraud.This statement is in line with research by Aviantara (2021), Dewi & Yuliati (2022) and Rizkiawan & Subagio (2022) which states that changing directors has a significant positive effect on financial statement fraud.Furthermore, when the market has a high level of concentration where companies control a percentage of the market share, it is possible for collusion to occur through price fixing in the market.This statement is in line with research by Dewi & Yuliati (2022), Kusumosari & Solikhah (2021) which states that collusion practices have a significant positive influence on fraudulent financial statements.Furthermore, opportunity is a condition that supports the occurrence of fraud, including opportunities obtained from certain positions or authority within the company.This statement is in line with research by Riyandi & Trisanti (2021) which states that effective supervision (the higher the composition of independent commissioners) has a significant negative effect on financial report fraud.In line with this, Lastanti et al (2022) and Lindasari (2019) through their research stated that the composition of independent commissioners compared to the total number of members of the board of commissioners is lower (less effective supervision) has a significant positive effect on financial report fraud.Furthermore, the existence of maximum limits on the use of audit services shows that there is a tendency for companies not to change auditors frequently.On the other hand, with frequent changes in auditors, it is possible for the company that these actions include efforts to reduce the detection of fraud or eliminate traces of fraud.This statement is in line with research by Rizkiawan & Subagio (2022) Wijaya & Witjaksono (2023) which states that changing auditors has a significant positive effect on financial statement fraud.Furthermore, it is considered that the CEO's dual position is not optimal in carrying out its duties, thereby triggering fraud.This statement is in line with research by Kusumosari & Solikhah (2021) and Rahayuningsih & Sukirman (2021) which states that having multiple CEO positions has a positive effect on financial report fraud.This means that the more CEOs hold multiple positions, the more fraudulent financial reports will increase.Therefore, the proposed hypothesis is as follows: H1: Hexagon fraud has a significant positive effect on financial statement fraud

Company Size on Financial Statement Fraud
Company size can generally be interpreted as a reference in looking at the company's business scale so that it can classify the company as large or small.Companies with larger assets tend to have more transactions, more complex operations and broader interests, which can trigger fraudulent financial statements.This statement is in line with research by Rahayu & Purnamasari (2023) which states that company size has a significant positive effect on financial statement fraud.Therefore, the proposed hypothesis is as follows: H2: Company size has a significant positive effect on financial statement fraud.

Company Size Moderates the Fraud Hexagon on Financial Statement Fraud
Company size can strengthen the influence of the fraud hexagon on financial statement fraud.With a larger company size, the pressure becomes stronger, the competence of directors is increasingly required, share price determination can be controlled, supervision becomes more complex, auditor changes become more frequent and multiple positions weaken supervision which ultimately increases fraud in financial reports.The use of company size as a moderating variable refers to research by Barus et al (2021) with the natural logarithm of total assets as the measure.This is in line with research by Lindasari (2019) which states that company size can moderate the influence of financial stability and ineffective supervision on financial statement fraud.Therefore, a hypothesis is proposed which is as follows: H3: Company size strengthens the influence of hexagon fraud on financial statement fraud.

METHODOLOGY
This research uses a quantitative approach as Sugiyono (2018:13) states that quantitative research methods can be interpreted as research methods based on the philosophy of positivism, used to research certain populations or samples, data analysis is quantitative or numbers in which there are statistical calculations.This type of research is included in explanatory research because it aims to explain the relationship or influence between variables based on a stated hypothesis (Sugiyono, 2018:223).This research explains hypothesis testing in terms of the influence of hexagon fraud as an independent variable on financial statement fraud as the dependent variable with company size as a moderating variable.

RESULT AND DISCUSSION 1. Data Analysis Classic Assumption Test
The classical assumption is a prerequisite test that must be carried out first before carrying out data analysis.The classical assumption test aims to determine the condition of the data that will be used in the research.The classical assumption test consists of normality, multicollinearity, autocorrelation and heteroscedasticity tests.The following are the results of each classical assumption test in this research.

Normality Test
The normality test is a test carried out to see whether the distribution of data from the research sample is normally distributed or not.The data normality test process is carried out using the Jarque-Bera (J-B) test, with decision making guidelines: H0: Statistical value > 0.05, then the distribution is normal H1: Statistical value <0.05 means the distribution is not normal Based on the image above, it is known that the Jarque-Bera statistical value is 0.204192, which is greater than the significance level, namely 0.05, so H0 is accepted, meaning that the normality assumption is met.The normality test for hexagon fraud with its six components in the principal component factor analysis produces a Jarque-Beras statistical value of 0.004775 which is smaller than the significance level, namely 0.05, so H0 is rejected that the normality assumption is not met.Based on the picture above, the normality test is not fulfilled.This result is possible because the data has a relatively large number of observations so that the distribution of the data is very diverse.Apart from that, by using a common effect model with an OLS (Ordinary Least Square) approach, normality testing is not an absolute requirement to fulfill.

Multicollinearity Test
The multicollinearity test aims to find out whether in the regression model there is a correlation between the independent variables.A good regression model should have no correlation between the independent variables.Multicollinearity can be seen from the tolerance value and VIF (Variance Inflation Factor).If the tolerance value is greater than 0.1 and the VIF value is smaller than 10, then the regression is free from multicollinearity.H0: The tolerance value is > 0.1 and the VIF value is < 10, so multicollinearity does not occur H1: Tolerance value < 0.1 and VIF value > 10 means multicollinearity occurs Source: Eviews Data Processing Based on the table above, the correlation value between variables is known to be less than 0.90, so H0 is accepted that there are no symptoms of multicollinearity between independent variables.As for the multicollinearity test of hexagon fraud with its six components together with company size as an independent variable, it is known that the correlation value between the variables is less than 0.90.

Source: Eviews Data Processing
The test above is a multicollinearity test after factor analysis is carried out where the result is less than 0.90, so H0 is accepted that there are no symptoms of multicollinearity between the independent variables.

Heteroscedasticity Test
The heteroscedasticity test aims to test whether in the regression model there is an inequality of variance from the residuals of one observation to another.If the variance from the residual from one observation to another observation is constant, it is called homoscedasticity.On the other hand, if it is different, it is called heteroscedasticity.A good regression model is one that is homoscedastic or does not have heteroscedasticity.Testing whether there is heteroscedasticity in this research can be done using the Breusch-Pagan-Godfrey test by looking at the probability numbers with decision making guidelines: H0: Statistical value > 10% means heteroscedasticity does not occur H1: Statistical value < 10% means heteroscedasticity occurs Testing is not available because the panel data has been weighted

Auto Correlation Test
The autocotellation test is often known as serial correlation and is often found on time series data.The autocorrelation test aims to see whether in the regression model there is a correlation between the confounding error in a certain period (t) and the error in the previous period (t-1).A good regression model is a regression that is free from autocorrelation.The autocorrelation test in this research was carried out using the Durbin Watson test method with a significance level of 0.05 which requires the existence of a constant in the regression model and no further variables among the independent variables with decision making guidelines: H0: There is no autocorrelation H1: There is autocorrelation Autoceleration detection in this way begins by calculating du and d1 using Durbin Watson results.Provision: du < d < 4 -du = no autocorrelation d < d1 = no positive autocorrelation d > 4 -d1 = no negative autocorrelation d1 < d < du = no decision about autocorrelation -du < d < 4 -d1 = no decision about autocorrelation

Source: Eviews Data Processing
Based on the table above, the Durbin Watson statistical value is known to be 2.173959 with a dL value of 1.6703 (4-dL is 2.3297) and dU is 1.8364 (4-dU is 2.1636) so the position is dU < D-W < 4-dL (1.83640 < 2.173959 < 2.3297), then H0 is accepted that it does not experience autocorrelation.The auto correlation test for hexagon fraud with its six components in principal component factor analysis produces a Durbin Watson statistical value of 1.688054 where the dL value is 1.7296 (4-dL is 2.2704) and dU is 1.7758 (4-dU is (2.22242) so that dU < D-W is not met while D-W < 4-dL is met.Source: Eviews Data Processing Based on the picture above, the auto correlation test does not meet this result, this is possible because the data being tested is panel data which has a continuous and sequential nature and cannot be changed, while in auto correlation testing tends to get different results if the data sequence changes.

Selection of Panel Data Regression Models
Panel data regression can be carried out by testing three analysis models, namely common, fixed and random effects with the following chronology:

Test Chow
The Chow test is carried out by choosing the fixed effect model as the basic model for testing based on decisions, namely as follows: H0: Chi-square cross section value > 0.05, then choose the common effect model then carry out the Lagrange Multiplier (L-M) test.H1: Cross section Chi-square value < 0.05, then choose the fixed effect model then carry out the Hausman test.Based on the table above, it is known that the statistical value is 0.00430, which is smaller than 0.05, so H0 is rejected, so the fixed effect model is selected, followed by the Hausman test.
Based on the table above, it is known that the statistical value is 0.0589, where if it is greater than 0.05, then H0 is accepted, so the common effect model is the most appropriate to use.Source: Eviews Data Processing

Multiple Regression
The model determined for multiple linear regression is based on the results of selecting the regression model, namely the common effect model.Multiple linear regression itself functions to measure the influence of more than one independent variable on the dependent variable projected by the model, namely as follows: The first regression model equation where the constant symbolized by C indicates that the dependent variable will have a value of -0.033606 if the independent variable has a value of 0 or does not change.Meanwhile, the regression coefficient for the hexagon fraud variable has a value of 0.066536 which is in the same direction as the financial report fraud variable.So that a 1% increase in this variable means that the financial report fraud variable will increase by 0.066536 (other variables are assumed to be constant).The regression coefficient for the company size variable has a value of 0.014920 which is in the opposite direction to the financial statement fraud variable.So that a 1% increase in this variable means that the financial statement fraud variable will decrease by 0.014920 (other variables are assumed to be constant).
The second regression where the constant symbolized by C indicates that the dependent variable will have a value of -0.212165 if the independent variable has a value of 0 or does not change.Meanwhile, the regression coefficient for the hexagon fraud variable has a value of 0.788018 which is in the opposite direction to the financial statement fraud variable.So that a 1% increase in this variable means that the financial statement fraud variable will decrease by 0.788018 (other variables are assumed to be constant).The regression coefficient for the company size variable has a value of 0.008731 which is in the opposite direction to the financial statement fraud variable.So that a 1% increase in this variable means that the financial statement fraud variable will decrease by 0.008731 (other variables are assumed to be constant).The regression coefficient for the hexagon fraud variable with company size as moderation has a value of 0.028414 which is in the same direction as the financial report fraud variable.So that a 1% increase in this variable means that the financial statement fraud variable will increase by 0.028414 (other variables are assumed to be constant).

Hypothesis Test Partial Test (t-Test)
The t-test is a test used to determine whether partial independent variables have a real effect on the dependent variable or not, the degree of significance used is 0.05.If the significance value is smaller than the degree of confidence then we accept the alternative hypothesis which states that an independent variable partially influences the dependent variable.To find the ttable using the sample size (n) = 175, the number of variables (k) = 3, the significance level α = 0.05 is with the formula df = n -k = 175 -3 = 172, then the t-table value is obtained amounting to 1.97385.

Source: Eviews Data Processing
Based on the table above, the hexagon fraud variable has a probability value of 0.004302 which is smaller than 0.05, so H0 is accepted and the t-count value is 2.893513 which is greater than the t-table value of 1.973850 so it can be concluded that financial stability has a significant positive effect on financial statement fraud.Meanwhile, the company size variable has a probability value of 0.328051 which is greater than 0.05, so H0 is rejected so that it can be concluded that company size has no effect on financial statement fraud.
The t-test in the context of Moderated Regression Analysis is a test to determine the linear relationship between two or more independent variables, including independent variables with moderation on the dependent variable, which can be concluded based on the hypothesis, namely as follows: H0: If the probability value is <0.05 then variable M can moderate the influence of variable X on variable Y H1: If the probability value is > 0.05 then variable M cannot moderate the influence of variable X on Y Based on the table above, hexagon fraud with company size as the moderator has a probability value of 0.022691 which is smaller than 0.05, so H0 is accepted and the t-count value is 2.299410 which is greater than the t-table value of 1.973850 so it can be concluded that hexagon fraud with company size as the moderator has a significant effect positive towards financial statement fraud where company size strengthens the influence of the hexagon fraud on financial statement fraud.

Simultaneous Test (f-Test)
The f-test is a test to find out whether all independent variables together (simultaneously) influence the dependent variable.Source: Eviews Data Processing Based on the results of the regression test on the first equation, it is known that the f-count value is 5.361557 with a probability value of 0.005511 or less than 0.05, which means that the independent variables jointly influence financial statement fraud.

Source: Eviews Data Processing
Based on the results of the regression test in the second equation, it is known that the f-count value is 6.037341 with a probability value of 0.000623 or less than 0.05, which means that the independent variables together have an influence on financial statement fraud.

Analysis of the Coefficient of Determination
It is a value (proportion value) that measures how far the ability of the independent variables used in the regression equation is to explain variations in the dependent variable.The coefficient of determination value is between zero and one.A small value of the Adjusted R Square coefficient of determination (close to zero) means that the ability of the independent variables simultaneously to explain variations in the dependent variable is very limited.An Adjusted R Square coefficient of determination value that is close to one means that the independent variables provide almost all the information needed to predict variations in the dependent variable.Source: Eviews Data Processing Based on the table above, it is known that the R-squared value is 0.095774, meaning that the independent variables are able to jointly influence the dependent variable, in this case 9% of financial statement fraud and conversely, the remaining 91% is explained by other variables not touched upon in this research.

Interpretation of Research Results
Based on the above, it is known that hexagon fraud has a significant positive effect on financial statement fraud.Thus, the first hypothesis is accepted where an increase in the hexagon will also increase fraud in financial statements.From the test results is also known that company size has no effect on financial statement fraud.Thus, the second hypothesis is rejected, where the larger the company size has no effect on financial statement fraud.Meanwhile, in terms of moderation, it is known that company size is able to moderate by strengthening the influence of hexagon fraud on financial statement fraud.Thus, the third hypothesis is accepted, where the influence in the same direction shows that the increase in hexagon fraud in larger companies will increase financial statement fraud.These results are in line with fraud theory itself, where the components contained in hexagon fraud can trigger financial statement fraud.Apart from that, the size of the company can strengthen this influence in relation to the theory of fraud where the company with a larger size, the pressure felt by management is also greater, there are more agreements in management, capabilities are required to be more qualified, aspects of supervision are more complex and justifications become more diverse and the egos that are accommodated are also getting stronger which ultimately has an impact on financial statement fraud through the components contained in hexagon fraud.

Figure
Figure 1.Theoretical Framework

Figure
Figure 2. Normality Test Source: Eviews Data Processing

Figure 3 .
Figure 3. Normality Test of Main Component Factors Source: Eviews Data Processing

Table 2 .
Multicollinearity Test on Main Component Factors

Table 3 .
Auto Correlation Test

Table 4 .
Auto Correlation Test on Main Component Factors

Table 8 .
Selected Common Effect Models

Table 9 .
Multiple Regression Test

Table 11 .
Moderated Regression Analysis t-test

Table 13 .
Moderated Regression Analysis f-test

Table 14 .
Coefficient of Determination