The Effectiveness of Corporate Governance in Constraining Fraud: Evidence from Listed Manufacturing Firms in Indonesia

Objective – The purpose of this study is to examine the direct effect of good corporate governance mechanisms on financial statement fraud with earnings management as a mediating


Introduction
Financial statement fraud is intentionally doing something that should not be done and not doing something that should be done resulting in a misstatement of financial statements that can be misleading (Tuanakotta, 2014). Financial statement fraud occurs in large multinational companies, an example is British Telecom. The disclosure of the British Telecom case began when there were reports of complaints which were followed by forensic accounting by KPMP in early 2017. It was revealed that the practice of fraud by British Telecom had occurred since 2013 through corrupt cooperation with corporate clients and financial services. The mode was to increase the company's income through the extension of fake contracts and fake transactions with vendors. Additionally, fictitious contracts were entered through cash inflows. The management conducted accrualization by recognizing fictitious trade receivables into company income. Accrualization actions by management are called earnings management.
Financial statement fraud that are not detected can develop into a big scandal that harms many parties. Rahman & Bremer (2016) found a significant negative effect of good corporate governance on financial statement fraud. Salim & Marietza (2017) found no effect of good corporate governance on financial statement fraud. Financial statement fraud begins with earnings management from financial statements that are considered immaterial but eventually grow into fraud and produce annual financial statements that are materially misleading (Ramirez, Martínez, & Mari, 2017).
Earnings management is a controversial and important area in financial account-ing. Earnings management is not always associated with efforts to manipulate accounting data or information, but is more likely to be associated with choosing accounting methods that are deliberately chosen by management for certain purposes (Nawaiseh, 2016;Perols & Lougee, 2011).
To reduce the occurrence of earnings management actions, an effective mechanism is needed in the management of the company, that is the mechanism of good corporate governance. The mechanism of good corporate governance can be interpreted as a rule of the game, procedures, and a clear relationship between the parties that make decisions with those who supervise those decisions (Rahman & Bremer, 2016). A good corporate governance mechanism is a set of mechanisms used to limit the emergence of information asymmetry problems that can encourage earnings management (Dowlatabadi & Filsaraei, 2016).
Besides that, in order to establish effective supervision and control in a company, two parties are needed, namely independent commissioners and audit committees. One of the tasks of independent commissioners is to ensure transparency and disclosure of the company's financial statements. Companies that implement transparency and disclosure of financial statements can reduce the occurrence of asymmetry of information between owners and managers of the company, thereby reducing the possibility for management to conduct earnings management, thereby reducing financial statement fraud (Nekhili, Mehdi, Amar, Chtioui, & Lakhal, 2016).
The audit committee is responsible for ensuring that the company has been run in accordance with applicable regulations, carrying out effective supervision of conflicts of interest and fraud committed by company employees (Patrick, Paulinus, & Nympha, 2015). The existence of an audit committee is very important to support accountability and transparency through quality financial reports so as to minimize the possibility of financial statement fraud by reducing earnings management practices (Owuigbe et al., 2019). Owuigbe et al., (2019) found that independent commissioners and audit committees can minimize earnings management practices thereby reducing financial statement fraud. Nekhili et al., (2016) and Dowlatabadi & Filsaraei (2016) found a significant negative effect on good corporate governance and earnings management. Keshteli & Fathi (2015) did not find any influence on good corporate governance and earnings management.
Although there has been previous research, inconsistencies have still been found in the previous research results. The inconsistency of the results of previous studies prompted researchers to use other variables that can mediate the relationship between good corporate governance on financial statement fraud by combining earnings management as a mediating variable.

Literature Review, Theoretical Framework, and Hypothesis Development
Agency theory is the rationale for understanding the concept of corporate governance. According to Jensen & Meckling (1976), an agency relationship is a contract between manager (agent) and investor (principal). Agency relationships occur when one or more people (principals) employ another person (agent) to provide a service and then delegate decision making authority. This theory assumes that each individual prefers his own interests because basically human nature as an individual creature causes conflicting interests between principal and agent. Dowlatabadi & Filsaraei (2016) stated that this relationship between investors and managers can lead to an asymmetrical information condition. This condition occurs because managers have more information about the company as a whole compared to information received by investors so this will encourage the behavior of managers to hide some information from investors. In conditions of asymmetrical information, managers have the oppor-JAROE VOL. 2(3) tunity to commit fraud by manipulating the numbers in the financial statements presented to investors. Owuigbe et al., (2019) explained that the agency theory related to corporate governance can be used as a tool for managers (agents) to convince investors (principals) in ensuring the return of funds they have invested. Corporate governance is expected to be able to overcome conflicts of interest and asymmetrical information between principals and agents to prevent and avoid fraud in financial statements.

The Effect of Good Corporate Governance Mechanisms on the Financial Statement Fraud
Owuigbe et al., (2019) found a relationship between the proportion of the board of commissioners and financial reporting fraud. Companies that commit fraud have a percentage of the external board of commissioners which is significantly lower compared to companies that do not commit fraud. This is in line with the results of Rahman & Bremer's research (2016) which stated that companies that do not commit fraud have a high percentage of external and independent commissioners compared to companies that commit fraud.
Independent commissioners do not experience pressure from the company's internal organization. Therefore, they are more likely to act independently and as separators of interests between principals and agents. With the existence of an independent commissioner, it can guarantee the transparency of financial statements so that shareholders receive higher quality information (Owuigbe et al., 2019). Azwin, Noorul, Ali, Razzaque, & Ahmed (2018) stated that the existence of an audit committee in a company will prevent managers from committing fraud. Owuigbe et al., (2019) stated that having an audit committee reduces the likelihood of fraud. Increasing the number of independent audit members will tend to provide positive performance so as to minimize indications of possible fraud in the company's financial statements. D'onza & Lamboglia (2011) found that independent audit committees were negatively associated with indications of fraud. Based on the description, the hypotheses proposed is: H1: The mechanism of good corporate governance has a significant negative effect on financial statement fraud.

The Effect of Good Corporate Governance Mechanisms on Earnings Management
Dowlatabadi & Filsaraei (2016) stated that board size and characteristics of commissioners from outside the company minimize the occurrence of earnings management practices by company management. The board of commissioners needs an independent commissioner to supervise and control the actions of the directors in connection with their opportunistic behavior. Abbadi, Hijazi, & Al-Rahahleh (2016) found that earnings management practices in companies could be reduced by the existence of an independent board of commissioners who had a higher percentage. Idris, Mohammed, Siam, & Nassar (2018) stated that in preventing earnings management practices, the effectiveness of the board of commissioners in overseeing management can be improved by the inclusion of commissioners from outside the company. Patrick et al., (2015) investigated the relationship of audit committees who were experts in finance and earnings management. The results of this study indicated that audit committee education in the financial field has proven to be effective in reducing earnings management. The existence of an audit committee in the company is expected to increase supervision of the company so as to create transparent company practices in order to minimize earnings management in the company (Abbadi et al., 2016). Nekhili et al., (2016) and Zarza, César, López, & Reguera (2018) showed a negative relationship between the audit committee and earnings management. This shows that the audit committee has succeeded in reducing the company's earnings management practices. Based on the description, the hypotheses proposed is: H2: The mechanism of good corporate governance has a significant negative effect on earnings management.

The Effect of Earnings Management on the Financial Statement Fraud
Companies that use accruals to increase revenues must be prepared with the consequences of their actions when they encounter flexibility in earnings management. One consequence is having to commit fraud to cover up or lighten accrual accounts that have been manipulated in order to achieve their goals (Hasnan & Mahenthiran, 2014). Ramirez et al., (2017) stated that companies that commit fraud have discretionary accruals, which is an accrual component that can be regulated and engineered according to managerial policies, which is greater than companies that do not commit fraud. Hasnan & Mahenthiran (2014) suggested that fraud can be detected through earnings management in the year prior to fraud. Based on the description, the hypotheses proposed is: H3: Earnings management has a significant positive effect on financial statement fraud.

Effect of Good Corporate Governance Mechanisms on Financial Statement Fraud through Earnings Management
Theoretically, the practice of good corporate governance can improve the quality of corporate financial reporting, reduce the risk that may be carried out by the board with decisions that support corporate governance and generally companies can increase investors' confidence in investing their capital. The influence of the board of commissioners in a company is more emphasized on the monitoring function of the implementation of directors' policies.
The role of the commissioner is expected to minimize agency issues that arise between the board of directors and shareholders. Therefore, the board of commissioners should be able to oversee the performance of directors produced in accordance with the interests of shareholders. The existence of an independent commissioner and audit committee can provide better oversight to management so as not to commit fraud in the financial statements through reduced earnings management practices. Based on the description, the hypotheses proposed is: H4: Good corporate governance mechanism has a significant negative effect on financial statement fraud through earnings management.

Research Method
The following research model is used in this study: The population of this research was manufacturing companies listed on the Indonesia Stock Exchange in the period 2013 to 2017, namely 131 companies x 5 years = 655 observational data. The data in this study was analyzed using partial least squares and data processed using WarpPLS 5.0 software. Research data collection techniques using nonprobability sampling using purposive sampling methods such as manufacturing companies on the Indonesia Stock Exchange. This study uses secondary data or data derived from financial and non-financial information published on the Indonesia Stock Exchange website (www.idx.co.id) or company websites.
Model analysis and hypothesis testing. The method of data analysis in this research was partial least square (PLS) carried out in two stages, namely the evaluation of structural models (outer models) and evaluation of measurement models (inner models). Outer models were evaluated with three types of tests, namely, first, convergent validity: the outer model meets the convergent validity requirements to build reflective when loading value > 0.70 and significant p-value <0.05. Indicators with loading values <0.40 were removed from the model and loading values between 0.40 and 0.70 were analyzed for the effect of removing the indicators on average variance expected (AVE) and composite reliability, where AVE is a minimum of 0.50 and composite reliability is 0.70 (Sholihin & Ratmono, 2013:66). Second, discriminant validity: to test discriminant validity, which is by looking at the value of cross loading for each variable must be> 0.70. Another way is to compare the square root of AVE for each construct with the construct value in the model (Ghozali & Latan, 2017:87). The square root of AVE must be greater than the value of the other constructs. Third, composite reliability: measures the true value of reliability in a construct. Rule of thumb, alpha value or composite reliability must be greater than 0.7 although the value of 0.6 is still acceptable (Abdillah & Hartono, 2016:62).
After evaluating the outer model, the inner model is evaluated using R2. R2 values 0.75, 0.45 and 0.25 can be concluded that the model is strong, moderate and weak (Ghozali & Latan, 2017:78). How well observational values are generated by the model and parameter estimation in the study used predictive relevance (Q2). Q2 value> 0 indicates that the model has predictive relevance (Ghozali & Latan, 2017:94).
In addition, an evaluation of model fit is carried out to find a model that fits the original data and measures the quality of the model. This study uses four measures of fit namely the average path coefficient (APC), average R2 (ARS), average adjusted R2(AARS) and average block variance factor (AVIF). APC, ARS and AARS measure the average path coefficient values, R2 and adjusted R2 produced in the model. The cut-off P-value for APC, ARS and AARS recommended as an indication of model fit is ≤ 0.05. AVIF is used to test the collinearity problem in the PLS model and the recommended value must be ≤ 3.3, assuming most constructs/variables in the model are measured by two or more indicators (Ghozali & Latan, 2017:96).
Hypothesis testing is based on the research objective, which is to assess the effect of independent variables separately. The significance (two-tailed) in this study is 1% or 0.01, 5% or 0.05, and 10% or 0.10. If the p-value is greater than α (0.01, 0.05, 0.10) then Ho is accepted and Ha is rejected. If the p-value is smaller or equal to α (0.01, 0.05, 0.10) then Ho is rejected and Ha is accepted.
There are two types of tests: the first test, the direct effect that compares the pvalue (the measurement results of the evaluation model) with the level of significance or R2. When the p-value generated from the PLS process is greater than the level of significance used, then there is a significant effect and vice versa. Second, the indirect effect carried out simultaneously estimates the indirect effect with the PLS SEM triangle model. The conclusions about mediation are as follows (Sholihin & Ratmono, 2013:57), if the path coefficient from independent to dependent after the mediation variable remains significant and does not change then the mediation hypothesis is not supported; if the path coefficient from independent to dependent after the mediation variable has gone down but remains significant then the form of mediation is partial Effectiveness, Corporate Governance, Constraining Fraud www/http/jurnal.unsyiah.ac.id/JAROE 237 mediation; if the path coefficient from independent to dependent after the mediation variable drops in value and becomes insignificant then the form of mediation is full mediation.

Result and Discussion
Outer models are evaluated with three types of tests, they are convergent validity, discriminant validity, and composite reliability: 1) Convergent validity: from the results of combined loadings and crossloadings output, it can be seen that the reliability indicators of the constructing items of the GCG, EM, and FRAUD dimensions are valid all with the resulting loading factor value> 0.7 and the P value <0.001 ( 3) Composite reliability: the resulting dimensions of each construct are also very good, that is > 0.7 so that it meets internal consistency reliability 0.795 (GCG), 1.000 (EM), 1.000 (FRAUD).
The inner model is evaluated with R2 and Q2 for each endogenous variable and can be seen in Table 5. For H4, earnings management and financial statement fraud variables have R2 of 0.375 and 0.312. This means that the earnings management variable can be explained 38% by the mechanism of good corporate governance and the financial statement variable fraud can be explained 31% by the mechanism of good corporate governance and earnings management. Based on the calculation of predictive validity (Q2), the Q2 value for the financial statement variable fraud on H4 is 0.300 that meets the good predictive validity criteria, that is Q2> 0.  Table 6 shows the results of evaluating the fit model to measure the quality of the model. APC, ARS, AARS on H4 have p-value ≤ 0.05 so that it meets the criteria of goodness of fit model. AVIF on H4 has a p-value ≤ 3.3 so that there is no collinearity problem between constructs/variables in the model. Testing to analyze whether the mediating variables (earnings management) affect the financial statement of fraud shows that the value of the EM-FRAUD path coefficient is 0.060 and significant at 0.10. That is, earnings management has a positive effect on financial statement fraud. The procedure of testing the earnings management hypothesis as a mediating variable of the relationship between the mechanism of good corporate governance and financial statement fraud is in two steps as follows (table 7).
First, estimating the direct effect, which is a mechanism of good corporate governance on financial statement fraud without including mediating variables. This direct effect must be significant. This stage was carried out when a significant H1 test at <0.001 and a significant H2 at 0.043 thus indicating that there was a significant direct effect. Second, to estimate indirect effects simultaneously with the PLS SEM triangle model, namely GCG-FRAUD, GCG-EM, and EM-FRAUD. The mediation effect requirements that must be met are that the GCG-EM and EM-FRAUD pathways must be significant. Based on these provisions it can be concluded that earnings management is a partial mediation in the relationship between the mechanism of good corporate governance and financial statement fraud so that H4 is supported.
For testing the indirect effect can be seen in table 7, hypothesis H4 (GCG-EM-FRAUD path) is significant (p <0.05) which states the mechanism of good corporate governance indirectly influences the financial statement of fraud with earnings management as a mediating variable proved significant. Good corporate governance mechanism has a significant negative effect on financial statement fraud. The results of this study are consistent with research by Owuigbe et al., (2019), Rahman & Bremer (2016) which stated that independent commissioners had a significant negative effect on financial statement fraud. The greater the proportion of independent commissioners indicates that the existence of independent commissioners in a company needs to be more than non-independent commissioners. The existence of an objective oversight by the commissioners can reduce management opportunities to commit fraud.
The results of this study are also consistent with the research of Azwin et al., (2018), D'onza & Lamboglia (2011 which stated that the audit committee has a significant negative effect on financial statement fraud. These results indicate that the audit committee can reduce and prevent financial statement fraud. The audit committee assists the function of the board of commissioners in overseeing the company's performance related to the presentation and disclosure of the company's financial statements.

The Effect of Good Corporate Governance Mechanisms on Earning Management
Good corporate governance mechanisms have a significant negative effect on earnings management. The results of this study are consistent with research by Dowlatabadi & Filsaraei (2016), Abbadi et al. (2016), Idris et al. (2018) stated that independent commissioners have a significant negative effect on earnings management. Idris et al. (2018) stated that independent commissioners can act as mediators in disputes that occur between managers and overseer management policies and provide management advice. Independent commissioners are in the best position to carry out monitoring functions in order to achieve good corporate governance.
The results of this study are also consistent with the research by Patrick et al. (2015), Nekhili et al. (2016) and Zarza et al. (2018) which stated that the audit committee has a significant negative effect on earnings management. Companies with formed audit committees report earnings with smaller discretionary accruals compared to companies that do not form audit committees. If the audit committee is more effective, the greater the decline in earnings management.

The Effect of Earnings Management on the Financial Statement Fraud
Earnings management has a significant positive effect on financial statement fraud. The results of this study are consistent with the research by Hasnan & Mahenthiran 2014), Ramirez et al. (2017) which stated that managers will use the excess information they have, for example by hiding or manipulating some of the information in order to meet the interests of managers who may one day or in some cases conflict with the interests of external parties who have less valid information in the form of earnings management. In earnings management there are two factors that influence, the agency factor and information asymmetry. The existence of these two factors is a very supportive condition for the practice of fraud (Ramirez et al., 2017).

Effect of Good Corporate Governance Mechanisms on Financial Statement Fraud through Earnings Management
The mechanism of good corporate governance has a significant negative effect on financial statement fraud if it has been manipulated earnings management. The results also prove H4 that earnings management can mediate the effects of good corporate governance mechanisms on financial statement fraud. Mahesarani & Chariri (2016) found empirical evidence that the mechanism of good corporate governance has a significant negative effect on financial statement fraud which is proxied by earnings management.
Earnings management based on agency theory states that each individual tends to maximize its utility. Management that has certain interests will tend to prepare earnings reports in accordance with their own goals and not in the interests of the principal. Earnings management actions taken by management, if left unnoticed by the owner, will eventually develop into a materially deceptive financial statement of fraud.
In such conditions, the control mechanism that can align the difference in interests between agent and principal is the mechanism of good corporate governance. The mechanism of good corporate governance is able to provide confidence to the principal as an owner and investor to receive a return on the capital they invest. Corporate governance is able to overcome conflicts of interest and asymmetrical information between principals and agents to prevent the occurrence of financial statement fraud through reduced earnings management practices undertaken by management.

Conclusions
Based on the analysis and testing, results show that the mechanism of good corporate governance has a significant effect on financial statement fraud and earnings management. In addition, earnings management has a positive effect on financial statement fraud and is proven significant. Earnings management can reduce the effect of good corporate governance mechanisms on financial statement fraud.
This study indicates that it is important for a company to uphold the importance of good corporate governance to avoid earnings management practices and financial statement fraud. From this research, investors should seek out companies who apply good corporate governance in their company activities and systems.