Earnings manipulation behavior in the banking industry of Bangladesh: the strategical implication of Beneish M-score model

Asia Khatun (Department of Accounting and Information Systems, University of Dhaka, Dhaka, Bangladesh)
Ratan Ghosh (Department of Accounting and Information Systems, Bangladesh University of Professionals, Dhaka, Bangladesh)
Sadman Kabir (Department of Business Administration-General, Bangladesh University of Professionals, Dhaka, Bangladesh)

Arab Gulf Journal of Scientific Research

ISSN: 1985-9899

Article publication date: 1 September 2022

Issue publication date: 15 September 2022

3730

Abstract

Purpose

This study aims to determine the number of companies involved in earnings manipulation. Additionally, this study has empirically investigated the common manipulation items among the companies.

Design/methodology/approach

Bangladesh's listed commercial banks are selected as a sample for this study, and financial data from 2009 to 2018 were collected. The likely and nonlikely manipulator Beneish model (1999) divides the sample into two groups. Based on the M-score of the model, the banks are put into two groups. To identify the most influential variables, an independent sample t-test was done with the help of Statistical Package for Social Sciences (SPSS).

Findings

The findings show that banks in Bangladesh have an unstable trend in making manipulated financial reports. Results of the t-test reveal that overstating revenues, increasing intangible assets, lessening cost and accruals are the most appealing items for preparing a fraudulent financial report. The findings of this research work will help the investors take the right decision having the idea of manipulation in the banking sector of Bangladesh.

Originality/value

In the presence of many irregularities in the banking sector Bangladesh, very few studies have been carried out in forensic accounting and fraudulent financial reporting practices. Much research has focused on earnings management techniques. This research specifically focuses on identifying earnings manipulation in financial statements for micro-level variables like accounting accruals, intangible assets, etc. This will help policy-makers and financial statement readers to be proactive while reading financial statements and taking any investment decision.

Keywords

Citation

Khatun, A., Ghosh, R. and Kabir, S. (2022), "Earnings manipulation behavior in the banking industry of Bangladesh: the strategical implication of Beneish M-score model", Arab Gulf Journal of Scientific Research, Vol. 40 No. 3, pp. 302-328. https://doi.org/10.1108/AGJSR-03-2022-0001

Publisher

:

Emerald Publishing Limited

Copyright © 2022, Asia Khatun, Ratan Ghosh and Sadman Kabir

License

Published in Arab Gulf Journal of Scientific Research. Published by Emerald Publishing Limited. This article is published under the Creative Commons Attribution (CC BY 4.0) licence. Anyone may reproduce, distribute, translate and create derivative works of this article (for both commercial and non-commercial purposes), subject to full attribution to the original publication and authors. The full terms of this licence may be seen at http://creativecommons.org/licences/by/4.0/legalcode


1. Introduction

Financial statements are the core information resource for any organization that trades publicly. Publicly traded organizations are listed under any country's Securities and Exchange Commission. According to agency theory, firms are directed by a management body, and the company's owner selects these groups of people. Management people are responsible for preparing the financial report of any firm, screening its operating performance and proficiency. With the help of this disclosed information, investors, creditors and other related parties decide that firm. Therefore, the steadfastness of those reports depends on their clarity and exactness. Moreover, the capital market uses the company's disclosed information to place the price for the listed security, and the users rely on them to establish their decision regarding that firm.

It should be noted that managers occasionally intentionally exploit the financial report to pressure some market competitors. The intentional manner of a manager to manipulate the financial figure is known as earnings management. Nonetheless, in 2014, 40 US public companies reported profitable conditions using creative accounting techniques while in a loss position under traditional accounting methods (Bhasin, 2016). Top management of any firm discloses falsified and misleading financial information to hide the company's actual scenario or financial position. The mindset of unprincipled management people is the main cause of increasing corporate accruals, not the business cycle (Hasan, Rahman, & Hossain, 2014; Hasan, Omar, Rahman, & Hossain, 2016). Users, especially the investors' groups, are defenseless due to this falsification of accounting data. The authentication of accounting systems and the dependence on disclosed reports for creating management and investment decisions are being questioned as the recent list of failed economic and business enterprises is not too short. A few such cases are the disintegration of Enron, WorldCom, Robert Maxwell Pension Funds and the downfall of Arthur Anderson, the “Big Five” accounting firm belonging to the “Big Five” above stated issue. Undoubtedly, these issues have proved that the financial report prepared by the administration of a firm and specialized by the external auditors could not bring the actual picture of a company. Decisions made following those reports became fraudulent and destroyed the belief of stakeholders. The management body misleads the firm's owners and the organization's users by adopting unethical steps through earning management. Transparency has become a vital issue regarding the annual reports provided by the company. The authenticity of the financial reports may endanger a company's inflated profit and expertise-related information. Doing fraud is the intentional decision of a firm's top management (CEO or CFO) to optimize their personal need and uphold the company's image towards the public. However, this is unethical.

Recent scams and financial fraud in Bangladesh's banking sector have urged further scrutinization of those financial institutions' operational details and financial solvency. Hallmark Scam, Bismillah Textile Scandal and AnnonTex fraud have highlighted the loopholes in the regulatory system of the financial institutions of Bangladesh. The process of approving loan and advances are not clear and transparent. Consequently, there has been a rise in the corporate indiscipline and accountability of the banking sector of Bangladesh. Lack of good governance, the management's bad intention and the regulators' ineffective control mechanism are hurting the Bangladeshi economy's most sensitive and vital sector. This also amplifies the opportunity to engage in fraudulent activity. Following the fraud triangle theory, this study analyzes the management body's opportunistic behavior. The fraud triangle theory consists of three components: pressure, opportunity and rationalization. The pressure of making wealth may transform a white-collar employee into a white-collar criminal, as most white-collar crimes are committed by formerly good people (Ghosh, Sen, & Riva, 2020). This tendency to engage in fraudulent activity is materialized when they get a good opportunity. Such kind of opportunity arises in the company due to poor governance, ineffective internal control mechanism and immoral attitude of the management. Fraudsters usually complete their plan by giving loans and advances based on personal connections, political affiliation and personal business interests. Finally, these persuaded investments are reported as nonperforming loans (NPLs) in the financial statements and shown to stakeholders as a loss arising from normal business operations. This can be termed as the rationalization of the fraud triangle theory. The management of One Bank Limited, a listed commercial bank, has manipulated the financial statements to overstate the profit without complying with the regulatory requirements of keeping provisions for loans (Alo, 2021). PK Halder, former Managing Director of a bank and two other financial institutions, had been found guilty of laundering 100 m taka from various financial institutions (Daily Star, 2021). These are the latest scandals that have taken in Bangladesh. The result of these unexpected events hit the firm performance. Reportedly the growing level of scams and NPLs is decreasing the banks' profitability. NPLs ratio is in increasing mood over the year for every kind of bank, and according to theory, it is affecting the profit percentage of the banks in Bangladesh negatively (Financial Express, 2020). However, the profit percentage of listed private commercial banks in Bangladesh is also growing, except in 2020 (due to the COVID-19 pandemic) (Bangladesh Bank, 2022). Moreover, NPLs and profit margins do not stand in a parallel way as they have negative assertions (Financial Express, 2020). Farmers Bank was established in 2013, and by the end of 2018, it showed NPLs of 58%. It disclosed a positive profit margin with a high amount of NPLs (Dhaka Tribune, 2019). The list of these loan scams, defaults and increasing NPLs is not short in Bangladesh, and this will create the problem of authenticity and transparency of disclosed information by this banking and other financial institutions doing business here. Ghosh et al. (2020) highlighted that lack of board independence is one of the major causes of the poor performance of the financial institutions in Bangladesh. As a result, the board of directors feels the pressure to manage earnings using different techniques. Few cases come to the news of general investors, and others remain behind the market scene about which general investors know nothing but put their investment in the market. Recently, procedures like statistical models, financial ratios, mathematical models and data mining have been used to find fraud in financial statements. According to forensic accounting, this detection of fraud needs a long investigation process, and the primary activity of this process should be uncovering fraud. A predictive diagnosis of manipulation is needed to accelerate the fraud investigation procedure to find the fraud in financial statements or discover the intention of fraud. Five types of patterns are mainly used for the falsification of financial statements. These include fabricated revenues, inappropriate timing schemes, understating liabilities, less disclosure and problematic asset valuation procedures (Hasan, Omar, Barnes, & Handley-Schachler, 2017). Investigating each stated variable can be a good start for finding the distortion pattern in the financial statement. On that note, Beneish's M-score model (Beneish, 1999) combines eight ratios that act as a forensic accounting tool for identifying fraud in the financial statement. This research aims to find out the fraud of listed commercial banks in Bangladesh and analyze them in an organized manner with the help of the Beneish M-score (Beneish, 1999). To motivate the study, there are some specific inquiries to construct a view of the present scenario of disclosing fraudulent information in Bangladesh. These identify the number of listed banks in Bangladesh that manipulate financial statements, find the specific pattern or criteria for fraud by the banks and find the governing or leading ratios or variables mostly used in manipulation. From the analysis, it can be said that the manipulating behavior of listed private commercial banks in Bangladesh shows an unstable trend. This means those banks are not engaging themselves with the same manipulating items to falsify their disclosed accounting information. Moreover, interest income and balance with other financial institutions, an unusual growth in intangible assets, accounting accruals and growth indicators are the main items for the sampled banks to manipulate their accounting data.

2. Literature review

2.1 Financial statement and fraud

With the help of disclosed audited financial statements of any firm or company, investors can take their investment decision. Sound corporate information is the precondition to maintain the interest of the investor. However, falsification in financial statements caused by fraud is the main concern in the present corporate world. Fraud is the intentional wrong representation of something to achieve an advantage or to deprive someone of the right. In the recent corporate world, fraud in financial statements has become common. Gupta and Gupta (2015) analyzed the concept of fraud and its consequences from an Indian perspective. They concluded that a weak regulatory system, absence of fraud reporting guidelines, inefficiency of financial institutions and ineffectiveness of board members of any firm are the main reasons for corporate fraud in India. Correspondingly, Bhasin (2015) concluded that a lower level of compliance, weak internal control system, inaccurate employment procedures, less training and excessive work pressure are the main reasons for bank fraud in India. Huang, Lin, Chiu, and Yen (2017) claimed that corporate pressure and desire for incentives are the main factors behind fraud in the financial statement. Using the Analytic Hierarchy Process (AHP) model, the authors concluded that lowly performance, external financing necessity, board members' inefficiency, financial anguish and competition provoke fraud. Kizil and Kasbasi (2018) commented that fraudulent financial information reduces the possibility of the right investment decision from the users of financial information.

2.2 Empirical evidence of Beneish M-score model: fraud detective tool

Finding fraudulent financial reporting, the accrual accounting model is mostly used methods initiated by Healy (1985) and advanced by DeAngelo (1988) and Jones (1991). Nonetheless, the Beneish M-score model (Beneish, 1999) procedures a set of dissimilar variables along with the accruals to spot manipulation. The Beneish M-score model can be used as a forensic accounting tool to detect fraud in financial statements as it gives more results than measures of fraud detection tools (Özcan, 2018; Akra & Chaya, 2020). Kamal, Salleh, and Ahmad (2016) sampled 17 listed public companies charged for fraudulent financial reporting in Malaysia. Using the Beneish model to check its authenticity for working as a forensic tool and their conclusion, they claimed that 14 out of 17 (82%) companies were accused of financial misrepresentation before any community broadcast. This model is quite effective in detecting financial irregularities before any announcement. Aghghaleh, Mohamed, and Rahmat (2016) also reported average correctness of 73.17% in detecting fraud in the Malaysian context. Repousis (2016) took 25,468 companies in Greece for 2011 and 2012 and found that 33% of the total companies were engaged in earnings manipulation as their M-score is more than the benchmark (−2.22) of the Beneish model. Several studies showed that revenues, assets (current and fixed), administrative expenses and accounting accruals are indicators of financial manipulation (Repousis, 2016; Tahmina & Naima, 2016; Mamo & Shehu, 2017; Ramírez-Orellana, Martínez-Romero, & Mariño-Garrido, 2017). Tahmina and Naima (2016) pointed out that inflating intangible assets is the key to manipulating earnings in the financial statements in Bangladesh.

Sakib (2019) exposed that textile companies in Bangladesh were engaged in earnings manipulation, and receivables, cash and accruals were the significant way to misappropriate information. Arman and Sharmin (2019) used 105 listed companies on Dhaka Stock Exchange (DSE) to expose their fraud-making percentage with the help of the M-score. Using −1.78 as a benchmark, 25.81% of companies disclosed wrong information, and −2.22 as a benchmark, 54.28% of companies had a fraud-making attitude. A logistic model proved that statistically, some variables like sales, receivables and accruals were the items for fraud in financial statements. Companies engaged in fraudulent financial reporting are expected to have lower returns in the future (Subiyono & Suardi, 2020).

Umar, Partahi and Purba (2020) used fraud diamond analysis along with the Beneish model to find the reason for fraud, and they concluded that financial stability, auditor replacement, nature and rationalization of industry affected fraud in the financial statement. Anning and Adusei (2020) found that most of Ghana's manufacturing and trading companies were involved in financial manipulation and used M-score for this conclusion. However, predicting the fraud in a financial statement by M-score will give better results than by Z-score (Akra & Chaya, 2020). Hołda (2020) showed the efficacy of eight ratios M-score over five ratio M-score for finding the manipulating firms of Warshow stock exchange. In 2021, Valaskova and Fedorko (2021) showed that the Beneish model can predict the fraudulent behavior of companies by detecting the manipulating one. Shakouri, Taherabadi, Ghanbari, and Jamshidinavid (2021) did a regression analysis and found that DSRI, GMI, AQI, SGI, DEPI and TATA of the Beneish model significantly impacted fraudulent financial reporting. Durana, Blazek, Machova, and Krasnan (2022), using the indicator of creative accounting, found that both the parameter (eight ratios and five ratios) of the Beneish model were able to find and predict the fraudulent behavior of financial reporting. Samuel (2022), using the banking sector of East Africa as a sample, proved that the Beneish model result divided the sample group into likely manipulator and nonlikely manipulation with accuracy.

The above literature concludes that the Beneish model is used to predict and detect any firms' fraudulent financial reporting behavior and is also flawless in this global environment. All the variables included in the Beneish M-score model are very much interrelated. Common variables of fraud or earnings manipulation are accruals, intangible assets, noncash expenses and divergence of cash flow and accrual earnings. Variables contained in the Beneish model are explanatory variables of earnings manipulation. However, there is lacking clear-cut conclusions that which of the variables are the indicator of fraud adopted by firms to falsify their disclosed financial information. This study tried to find the specific variable of the Beneish model, mostly used for falsification of information and instigating fraud.

3. Research methodology:

3.1 Research design

The first stage activity is finding the likely and nonlikely manipulator banks using the Beneish model. Messod Daniel Beneish created an eight-variable mathematical model to classify the happening of fraud of financial nature or propensity to involve in earnings manipulation. Eight ratios create a score named M-score that can express the misrepresentation of financial data in financial statements, and this distortion will result in earnings manipulation. Sometimes this score indicates susceptibility to earnings manipulation. When M-score is less than −2.22, the respective firm or organization will be treated as a not likely manipulator, and when it is more than −2.22, the firm will be pickled as a likely manipulator (Beneish, 1999). Through his analysis, Beneish said that this model's weighted or unweighted possibilities of earnings management are significantly connected with the presence of fraud as he could correctly identify 76% of manipulators. Moreover, only 17.5% of nonmanipulators were incorrect with the model. Beneish and Nichols (2005) again found the probability of financial distortion by using five ratios in the previously stated Beneish model.

The model is as follows:

M= 4.84+(0.920×DSRI)+(0.528×GMI)+(0.404×AQI)+(0.892×SGI)+(0.115×DEPI)(0.172×SGAI)+(4.679×TATA)(0.327×LEVI)

where

This eight ratios or variables model of M-score is proficient in uncovering the accounting falsification and poor eminence of reporting. This study occupies the commonly used yardstick of a −2.22 score (Beneish, 1999) for categorizing the banks into two likely and non-likely manipulators for a year.

3.2 Population and sample selection

This study aims to determine the banks engaged in fraudulent financial reports and disclosing materially misstated information in Bangladesh. In total, 61 scheduled banks in Bangladesh function under the complete governance and administration of the Bangladesh Bank. Therefore, we must take all scheduled banks to portray the overall banking scenario. However, the analysis takes ten years (2009–2018), and many scheduled banks were established after 2008. Data were taken from 2008 for calculating the ratios for M-score; previous year information is needed. Availability of annual reports is the second concern for collecting data or information for concerned banks. Considering both the issues of time and availability of resources for collecting data, 30 listed commercial banks are selected as samples for this study, covering around 50% of the total population of banks in Bangladesh.

3.3 Data analysis technique

For this study, data were analyzed in two stages. At first, the eight ratios for calculating M-score were developed using M.S. excel. With the help of the previously mentioned model (Beneish, 1999), banks are divided into the likely manipulator group and the nonlikely manipulator group. The next stage of data analysis of this research combined the statistical test to uncover the utmost substantial ratios directed to such differentiation of banks. Statistical Package for Social Sciences (SPSS) will be used here to analyze the collected data and answer the research question; an independent t-test was done between the two banks to find the ratios that are statistically responsible for categorizing the banks into two groups. Moreover, this test tells the dominating ratios and concern variables used mostly for manipulating financial data.

4. Analysis and findings:

4.1 Classification of banks based on M-score

All collected data from 30 commercial listed banks over the period 2009–2018 are tested using the Beneish model to find the M-score. Compared with the benchmark value of −2.22, banks are divided into groups, and summarized results are presented in Table 1. These results also help determine the pattern of banks disclosing misleading information (if any).

Table 1 shows the number of likely and nonlikely manipulator banks in Bangladesh from 2009 to 2018. Unfortunately, the result did not provide any increasing or decreasing trend of manipulation. However, the numbers of likely manipulators are more than nonlikely manipulators. In 2009, 63.33% of banks were tested to appear to be likely manipulators, and 36.67% were not-likely manipulators. Next year, the number of likely manipulators reached 80%, and the rest are not-likely manipulators. There was again a decrease in 2011 in the number of a manipulator; it was 53.33%, and 46.67% tested as not being a manipulator. Among ten years of calculation, the number of expected manipulator banks was lower in 2012 and 36.67%. However, the number of probable manipulators increased again in 2013 with 60% of the total sample. Next year, the nonprobable manipulator was 23.33%, which increased the probable manipulator. In 2015, the number of expected manipulators decreased slightly, but in 2016, it reached its peak (83.33%). The expected manipulator number is somewhat decreasing in 2017 (70.0%) compared to 2016. However, in 2018, the increasing trend of expected manipulators was again in the picture; that year, it was 73.33%. In short, the number of the expected and nonexpected manipulators did not have any increasing or decreasing trend. Rather the rate was fluctuating in nature.

4.2 Findings of most significant ratios

This part of the findings deals with the result of an independent t-test to find the most significant ratios among the eight ratios stated in the Beneish model (Beneish, 1999). The banks are divided into two groups based on the M-score, and these two groups are likely manipulators and non-likely manipulators by name. An independent sample t-test was done with the help of statistical analysis software SPSS, and analysis was done with the two groups using data yearly. The result is portrayed in Tables 2–11.

The result will be discussed with the help of the value of t and the value available in the significance column of the table. When the significance column value is less than or equal to 0.05, it denotes that the variability of the variable is not identical, and the difference between the first and second groups is statistically significant. Moreover, if the value is greater than 0.05, the difference between the two groups is not significant. Table 2 recapitulates the results of the year 2009. The stated results of Table 2 denote that the groups do not differ significantly among the eight ratios GMI, AQI, SGI, DI, LEVI and TATA.

Additionally, only DSRI in 2009 portrays significant differences between the group as the p-value is less than 0.05. In 2010, only SGI significantly differed between the groups due to manipulation. The other seven ratios in the result table do not differ significantly as the p-value is greater than 0.05. Table 4 depicts the statistical test result of 2011; here, only the ratio DSRI is significant. DSRI, AQI and DI are exposed significantly in terms of differences between the groups in 2012.

Other ratios like GMI, SGI, LEVI, SGAI and TATA are not differentiated. In 2013, which results showed that SGI, DI, LEVI, SGAI and TATA have no significant difference. Moreover, DSRI, GMI and AQI significantly vary between the groups. Table 7 states the results of the year 2014, and it displays that DSRI, GMI and SGI significantly vary between groups. The results of 2015 tell that only AQI is significant between groups. In Table 9, the results ensure that DSRI and TATA are significant in terms of statistical tests. DSRI has a p-value of .000, indicating the significant variation between the groups in Table 10. Results of AQI and DSRI in 2018 portray significant variation, and other ratios do not differ between them. According to the statistical test, the difference between the groups of the same ratio denotes the financial manipulations using those variables.

4.3 Interpretation of findings

This research tries to determine the number of listed banks involved in manipulation. From the calculated result, the intention to falsify the financial information does not have any constant increasing or decreasing trend. The pattern of manipulation in the banks of Bangladesh is fluctuating. However, the average rate of likely manipulator banks is high around 68.88% over the ten years.

The next part of the analysis portrays the result of an independent sample t-test to figure out the governing variables for misstating information. Data are divided into two groups based on M-score over ten years yearly. However, 2009–2018 shows that DSRI, SGI, AQI, DI, GMI and TATA variables have significant results. The significance level is not the same over time. In 2009, DSRI gave a significant result, but in 2010, SGI depicted significant variation between the groups. From the ten-year data, DSRI gives eight times the significant result, AQI gives four times, SGI and GMI provide two times, and DI and TATA ensure one-time significant results. Banks use DSRI variables that mean interest income and balance with other financial institutions as their main manipulation item. Inflated revenues and disproportionate balance with others denote a higher increase in DSRI (Warshavsky, 2012). As the results of this study ensure some balances are available with others, this may be the reason for inflated revenues over time.

The second manipulating item is SGI, which measures the sales growth index. However, sales growth is good for the company and should be consistent with the operating cash flow for a certain time. The next AQI that deals with the asset quality and increase in AQI expects to weaken the quality of assets portrayed in the financial statement. The high asset index denotes an increase in intangible assets without proper justification of recognition, and sometimes, firms use the increase of cost deferral to increase this asset quality index (Warshavsky, 2012).

Depreciation is one of the easiest ways to falsify information as it is a noncash item. DI has a direct connection with the asset quality of any firm. GMI is related to the revenues and direct cost of earning those revenues. It indicates the actual growth of a company, and research says that high-growth companies engage themselves in earnings manipulation. TATA is the common form of making or providing misleading information. It is concluded that a higher amount of accruals signs accounting manipulation.

4.4 Discussion

Investigating the quality of earnings is an important element of the company's financial statement. This importance is increasing due to the fall of some big companies worldwide for manipulating their earnings. The Beneish model can find the fraudulent report-making firm through its yardstick score and the eight variables (Beneish, 1999). This research tries to identify the number of manipulators and nonmanipulation-listed banks in Bangladesh from 2009 to 2018. Data are collected for 11 years to calculate the ten-year ratio as the ratio needs previous year information. The first phase of calculation shows that the numbers of manipulator banks are high in percentage. However, the manipulations in banks do not have any constant trend but rather have an unstable increasing and decreasing trend. In addition, on average, 68.88% of listed banks are engaged in earnings manipulation in Bangladesh.

The t-test helped determine the most persuasive ratios or how variable banks manipulated each year in these ten years took time. In the first year (2009), most banks used inflated revenues and embezzled balances with others (receivables) as their earnings mismatch. Moreover, in these ten years, banks mostly used revenues and receivables to deflate the actual scenario of financial reports. Asset quality index is the other element of manipulation. When the index of asset quality increases, this denotes misleading asset valuation as the increase intangible asset so frequently is not a sign of a trustworthy report. Cost deferral is also related to asset quality and is an easy item for misstating financial information. Sales growth index and gross margin are both related to revenues. Sales growth is a positive sign for banks, and excessive growth also raises questions.

Moreover, when the growth rate is excessively high than the competitors, it denotes the unsound mentality of the concerned organization. Gross margin growth denotes the increase in revenue in high percentage and decreases in cost in high percentage. Understating costs can be an option for inflating the amount of revenue. Lastly, total accruals are the recognized medium to manipulate. However, banks useless this item for their manipulation, which is found in the t-test result. Only in one year the value of p of TATA is institute significant, indicating significant variation between the groups.

This study has several policy implications. First, as the use of accruals is one of the predominant reasons of earnings manipulation in the banking industry, the Bangladesh Bank, the Central Bank of Bangladesh and the regulator of the money market, should be stricter on loan rescheduling and recognition of poor investment as an interest income. Second, investments in intangible assets should be shown in the cost price in the financial statement, which will help to avoid the overstatement of intangible assets. Finally, the regulators should impose more regulations on credit assessment and credit follow-up to reduce the NPLs in the banking sector of Bangladesh.

5. Conclusion

This research aims to identify the likely and not-likely manipulator banks in Bangladesh. Moreover, to determine the most influential ratios or variables among the eight ratios of the Beneish model through an independent sample t-test using SPSS. The management body is the key personnel to decide on the organization. Their intention of using earnings management for giving misled information will emerge when they feel deprived. Optimizing management needs is one of the main reasons for making materially misstated data as management has some discretionary power to control the organization. The code of corporate governance will act as a solution here. Different corporate appliances of the corporate governance code create accountability issues in management activities (Tassadaq & Malik, 2015). Management accountability will increase when the company ensures auditors' independence and increases the number of outside directors.

Moreover, the strong ethical and moral values of people involved in preparing and disclosing financial information is essential to lessen the manipulation of financial numbers; thus, the quality of financial report will accelerate. The present study only took banks as their sample. Therefore, overall, the financial sector is not included here. The next study may incorporate the nonbank financial institutions to formalize the manipulation behavior of the financial sector of Bangladesh. No corporate governance indicator impact is not considered here, and further study may try to find the mediation or moderator effect of governance variable in making misleading information.

Proportion of likely manipulator firms to nonlikely manipulator firms

YearLikely manipulators (M-Score > −2.22)Nonlikely manipulators (M-Score < −2.22)
200919 (63.33%)11 (36.67%)
201024 (80%)6 (20%)
201116 (53.33%)14 (46.67%)
201211 (36.67%)19 (63.33%)
201318 (60%)12 (40%)
201423 (76.67%)7 (23.33%)
201522 (73.33%)8 (26.67%)
201625 (83.33%)5 (16.67%)
201721 (70%)9 (30%)
201822 (73.33%)8 (26.67%)

Result of t-test for the year 2009

Independent sample test
Levene's test for equality of variancest-test for equality of means
FSig.tdfSig. (2-tailed)Mean differenceStd. error difference95% confidence interval of the difference
LowerUpper
DSRIEqual variances assumed9.1590.005−3.339280.037−0.177010.52154−1.245330.89132
Equal variances not assumed −2.43122.360.027−0.177010.41047−1.027460.67345
GMIEqual variances assumed0.1440.707−0.780280.442−0.071880.09210−0.260550.11679
Equal variances not assumed −0.80222.810.431−0.071880.08960−0.257320.11356
AQIEqual variances assumed2.0340.165−0.699280.490−0.073170.10465−0.287540.14120
Equal variances not assumed −0.92318.290.368−0.073170.07926−0.239490.09315
SGIEqual variances assumed2.0900.159−0.434280.667−0.016190.03725−0.092500.06012
Equal variances not assumed −0.48927.660.629−0.016190.03312−0.084060.05169
DIEqual variances assumed2.2220.1470.776280.4441.248831.60843−2.045884.54354
Equal variances not assumed 1.02718.090.3181.248831.21580−1.304533.80219
LEVIEqual variances assumed1.8880.1801.156280.2580.017840.01543−0.013780.04945
Equal variances not assumed 1.45923.050.1580.017840.01223−0.007460.04314
SGAIEqual variances assumed0.1660.686−0.769280.448−0.031800.04135−0.116500.05291
Equal variances not assumed −0.76020.270.456−0.031800.04184−0.119000.05540
TATAEqual variances assumed0.0050.945−0.958280.346−0.031370.03275−0.098460.03573
Equal variances not assumed −1.0627.240.297−0.031370.02949−0.091850.02911

Result of t-test for the year 2010

Independent sample test
Levene's test for equality of variancest-test for equality of means
FSig.tdfSig. (2-tailed)Mean differenceStd. error difference95% confidence interval of the difference
LowerUpper
DSRIEqual variances assumed1.0240.3200.473280.64022.9942748.60872−76.5761122.5647
Equal variances not assumed 0.95823.020.34822.9942723.99040−26.631172.61970
GMIEqual variances assumed0.4300.5170.686280.4980.045840.06679−0.090970.18265
Equal variances not assumed 0.5976.680.5700.045840.07680−0.137550.22922
AQIEqual variances assumed0.8520.364−0.373280.712−0.093810.25149−0.608960.42134
Equal variances not assumed −0.75223.530.460−0.093810.12479−0.351640.16402
SGIEqual variances assumed3.6210.0672.655280.0180.032940.05032−0.070140.13602
Equal variances not assumed 2.96216.370.0500.032940.03424−0.039510.10539
DIEqual variances assumed0.0750.7860.865280.3950.206650.23901−0.282930.69624
Equal variances not assumed 0.9088.200.3900.206650.22759−0.315920.72923
LEVIEqual variances assumed1.0330.3180.483280.6330.484361.00326−1.570732.53944
Equal variances not assumed 0.97823.010.3380.484360.49509−0.539801.50851
SGAIEqual variances assumed0.4140.525−0.578280.568−0.054400.09404−0.247020.13823
Equal variances not assumed −0.70510.440.496−0.054400.07716−0.225330.11653
TATAEqual variances assumed0.4590.5030.185280.8540.015660.08461−0.157650.18898
Equal variances not assumed 0.33627.820.7390.015660.04660−0.079830.11

Result of t-test for the year 2011

Independent sample test
Levene's test for equality of variancest-test for equality of means
FSig.tdfSig. (2-tailed)Mean differenceStd. error difference95% confidence interval of the difference
LowerUpper
DSRIEqual variances assumed9.2870.0053.527280.0260.300840.57080−0.868391.47007
Equal variances not assumed 2.56017.420.0580.300840.53768−0.8314910.43317
GMIEqual variances assumed1.1020.3031.663280.1070.171280.10298−0.039660.38222
Equal variances not assumed 1.69127.700.1020.171280.10130−0.036330.37889
AQIEqual variances assumed4.8760.0361.315280.1990.012500.00951−0.006980.03197
Equal variances not assumed 1.37621.530.1830.012500.00908−0.006370.03136
SGIEqual variances assumed5.8710.022−1.942280.062−0.131820.06787−0.270830.00720
Equal variances not assumed −1.85217.120.081−0.131820.07119−0.281920.01829
DIEqual variances assumed8.7130.0060.003280.9970.000430.12578−0.257220.25808
Equal variances not assumed 0.00315.870.9970.000430.13272−0.281100.28196
LEVIEqual variances assumed0.2520.6201.282280.2100.010930.00853−0.006540.02840
Equal variances not assumed 1.29227.970.2070.010930.00846−0.006410.02827
SGAIEqual variances assumed3.0470.0921.007280.3220.639500.63475−0.660721.93972
Equal variances not assumed 1.07815.220.2980.639500.59297−0.622751.90175
TATAEqual variances assumed3.5950.068−1.511280.142−0.091210.06036−0.214860.03244
Equal variances not assumed −1.60816.720.127−0.091210.05672−0.211020.02860

Result of t-test for the year 2012

Independent sample test
Levene's test for equality of variancest-test for equality of means
FSig.tdfSig. (2-tailed)Mean differenceStd. error difference95% confidence interval of the difference
LowerUpper
DSRIEqual variances assumed4.0870.053−3.75280.001−0.735740.19579−1.13680−0.33469
Equal variances not assumed −4.2927.960.000−0.735740.17124−1.08653−0.38496
GMIEqual variances assumed0.0770.7840.618280.5410.056590.09150−0.130840.24402
Equal variances not assumed 0.63422.600.5330.056590.08929−0.128300.24148
AQIEqual variances assumed4.4780.0433.525280.0010.030360.008610.012720.04800
Equal variances not assumed 2.99913.170.0100.030360.010120.008520.05220
SGIEqual variances assumed2.9380.098−1.68280.103−0.083110.04930−0.184100.01788
Equal variances not assumed −1.8827.460.070−0.083110.04412−0.173570.00734
DIEqual variances assumed7.3870.011−2.78280.045−0.224230.12550−0.481300.03284
Equal variances not assumed −2.2821.820.033−0.224230.09825−0.42808−0.02038
LEVIEqual variances assumed1.9720.1710.943280.3540.007050.00748−0.008270.02236
Equal variances not assumed 0.76611.780.4590.007050.00920−0.013040.02714
SGAIEqual variances assumed0.3120.5810.902280.3750.108550.12034−0.137960.35505
Equal variances not assumed 0.78313.890.4470.108550.13857−0.188880.40597
TATAEqual variances assumed2.0040.168−1.56280.128−0.043670.02783−0.100680.01335
Equal variances not assumed −1.2511.450.233−0.043670.03469−0.119650.03231

Result of t-test for the year 2013

Independent sample test
Levene's test for equality of variancest-test for equality of means
FSig.tdfSig. (2-tailed)Mean differenceStd. error difference95% confidence interval of the difference
LowerUpper
DSRIEqual variances assumed2.5960.118−3.95280.000−0.677590.17122−1.02831−0.32687
Equal variances not assumed −3.4914.770.003−0.677590.19389−1.09141−0.26377
GMIEqual variances assumed15.5980.000−2.91280.007−0.539490.18488−0.91819−0.16079
Equal variances not assumed −2.4111.720.033−0.539490.22385−1.02849−0.05049
AQIEqual variances assumed0.1820.6732.426280.0220.016900.006970.002630.03117
Equal variances not assumed 2.43824.130.0230.016900.006930.002590.03120
SGIEqual variances assumed0.3030.587−0.968280.341−0.041400.04276−0.128990.04620
Equal variances not assumed −0.95022.200.352−0.041400.04357−0.131700.04891
DIEqual variances assumed4.0050.055−1.02280.316−0.302640.29628−0.909550.30427
Equal variances not assumed −0.84011.550.418−0.302640.36036−1.091160.48588
LEVIEqual variances assumed0.7200.4030.311280.7580.001490.00480−0.008330.01132
Equal variances not assumed 0.34527.620.7330.001490.00433−0.007380.01037
SGAIEqual variances assumed1.7320.1991.031280.3110.556720.54005−0.549531.66297
Equal variances not assumed 1.25518.450.2250.556720.44363−0.373681.48711
TATAEqual variances assumed0.5580.461−2.25280.032−0.055620.02469−0.10620−0.00505
Equal variances not assumed −2.6223.82.015−0.055620.02121−0.09941−0.01183

Result of t-test for the year 2014

Independent sample test
Levene's test for equality of variancest-test for equality of means
FSig.tdfSig. (2-tailed)Mean differenceStd. error difference95% confidence interval of the difference
LowerUpper
DSRIEqual variances assumed0.0840.775−2.49280.046−1.000170.66891−2.370370.37004
Equal variances not assumed −2.1823.310.039−1.000170.45747−1.94580−0.05453
GMIEqual variances assumed0.6000.445−2.44280.016−0.227530.15784−0.550850.09579
Equal variances not assumed −2.2326.510.034−0.227530.10178−0.43654−0.01852
AQIEqual variances assumed0.2410.6271.249280.2220.028790.02306−0.018440.07602
Equal variances not assumed 1.69619.140.1060.028790.01698−0.006730.06431
SGIEqual variances assumed1.6410.2112.141280.0410.091210.042610.003920.17850
Equal variances not assumed 2.7277.7150.0120.091210.05281−0.031360.21378
DIEqual variances assumed1.7220.2000.430280.6710.143750.33441−0.541260.82877
Equal variances not assumed 0.75425.760.4580.143750.19059−0.248170.53568
LEVIEqual variances assumed0.3430.5631.420280.1670.004710.00332−0.002080.01150
Equal variances not assumed 1.0947.4110.3080.004710.00430−0.005350.01477
SGAIEqual variances assumed0.8750.3580.459280.6500.353370.77053−1.225001.93174
Equal variances not assumed 0.83322.990.4140.353370.42435−0.524491.23123
TATAEqual variances assumed0.5660.458−0.723280.475−0.023280.03218−0.089190.04264
Equal variances not assumed −1.0120.940.322−0.023280.02293−0.070960.02441

Result of t-test for the year 2015

Independent sample test
Levene's test for equality of variancest-test for equality of means
FSig.tdfSig. (2-tailed)Mean differenceStd. error difference95% confidence interval of the difference
LowerUpper
DSRIEqual variances assumed4.2960.0480.114280.9100.111820.98205−1.899812.12345
Equal variances not assumed 0.18225.020.8570.111820.61334−1.151321.37497
GMIEqual variances assumed0.4600.5030.162280.8730.077640.47952−0.904621.05989
Equal variances not assumed 0.25126.940.8040.077640.30909−0.556640.71191
AQIEqual variances assumed1.6150.2143.267280.0210.031490.02486−0.019440.08242
Equal variances not assumed 2.00425.840.0560.031490.01571−0.000820.06380
SGIEqual variances assumed1.3820.2500.507280.6160.017770.03507−0.054060.08961
Equal variances not assumed 0.45510.420.6590.017770.03908−0.068830.10437
DIEqual variances assumed0.1690.6841.566280.1290.098680.06301−0.030390.22774
Equal variances not assumed 1.92719.940.0680.098680.05120−0.008150.20551
LEVIEqual variances assumed1.2510.273−0.533280.598−0.032380.06075−0.156820.09207
Equal variances not assumed −0.89021.350.383−0.032380.03636−0.107910.04316
SGAIEqual variances assumed0.2170.6450.381280.7060.019640.05149−0.085840.12512
Equal variances not assumed 0.37612.110.7140.019640.05230−0.094180.13346
TATAEqual variances assumed0.2740.605−1.00280.322−0.037450.03712−0.113470.03858
Equal variances not assumed −1.4026.350.173−0.037450.02674−0.092370.01747

Result of t-test for the year 2016

Independent sample test
Levene's test for equality of variancest-test for equality of means
FSig.tdfSig. (2-tailed)Mean differenceStd. error difference95% confidence interval of the difference
LowerUpper
DSRIEqual variances assumed0.0360.850−3.30280.003−1.270170.38477−2.05834−0.48201
Equal variances not assumed −3.766.5970.008−1.270170.33757−2.07839−0.46195
GMIEqual variances assumed1.0520.3140.875280.3890.122360.13980−0.164010.40873
Equal variances not assumed 0.6734.7640.5320.122360.18189−0.352240.59696
AQIEqual variances assumed0.2790.6020.181280.8580.013490.07457−0.139260.16624
Equal variances not assumed 0.36426.640.7190.013490.03710−0.062680.08966
SGIEqual variances assumed0.0210.886−1.34280.189−0.042780.03174−0.107800.02224
Equal variances not assumed −1.365.7980.222−0.042780.03130−0.120010.03445
DIEqual variances assumed0.8120.3750.422280.6761.344573.18704−5.183797.87293
Equal variances not assumed 0.95724.020.3481.344571.40568−1.556464.24560
LEVIEqual variances assumed0.8370.3680.458280.6500.153290.33436−0.531620.83819
Equal variances not assumed 1.04024.000.3090.153290.14744−0.151020.45759
SGAIEqual variances assumed0.7760.3861.004280.3240.074360.07404−0.077310.22602
Equal variances not assumed 1.61913.260.1290.074360.04593−0.024660.17338
TATAEqual variances assumed0.1280.723−2.29280.028−0.060240.04670−0.155900.03541
Equal variances not assumed −2.2316.750.039−0.060240.02693−0.11711−0.00337

Result of t-test for the year 2017

Independent sample test
Levene's test for equality of variancest-test for equality of means
FSig.tdfSig. (2-tailed)Mean differenceStd. error difference95% confidence interval of the difference
LowerUpper
DSRIEqual variances assumed2.5800.119−4.15280.000−0.855170.20604−1.27722−0.43312
Equal variances not assumed −3.3710.420.007−0.855170.25364−1.41723−0.29311
GMIEqual variances assumed1.5760.220−0.710280.483−0.271530.38231−1.054650.51159
Equal variances not assumed −1.0722.250.296−0.271530.25374−0.797410.25436
AQIEqual variances assumed12.2220.002−1.44280.159−0.415330.28717−1.003570.17292
Equal variances not assumed −0.9248.0030.382−0.415330.44929−1.451320.62067
SGIEqual variances assumed0.0000.989−0.700280.490−0.023860.03409−0.093680.04597
Equal variances not assumed −0.74417.560.467−0.023860.03206−0.091330.04362
DIEqual variances assumed3.6150.0680.545280.5900.046040.08447−0.126980.21906
Equal variances not assumed 0.3918.9180.7050.046040.11786−0.220940.31303
LEVIEqual variances assumed0.7750.386−0.210280.835−0.000760.00360−0.008140.00662
Equal variances not assumed −0.26325.970.795−0.000760.00288−0.006670.00516
SGAIEqual variances assumed6.4980.0170.475280.6390.020420.04302−0.067700.10853
Equal variances not assumed 0.61727.450.5420.020420.03307−0.047390.08822
TATAEqual variances assumed0.3150.579−0.937280.357−0.031290.03341−0.099720.03713
Equal variances not assumed −1.2627.990.215−0.031290.02469−0.081870.01928

Result of t-test for the year 2018

Independent sample test
Levene's test for equality of variancest-test for equality of means
FSig.tdfSig. (2-tailed)Mean differenceStd. error difference95% confidence interval of the difference
LowerUpper
DSRIEqual variances assumed1.5160.229−6.21280.000−0.874520.14063−1.16259−0.58645
Equal variances not assumed −5.249.5800.000−0.874520.16683−1.24848−0.50057
GMIEqual variances assumed2.3160.139−0.779280.443−0.123570.15865−0.448550.20140
Equal variances not assumed −1.2524.530.221−0.123570.09845−0.326520.07938
AQIEqual variances assumed0.2900.5952.767280.0280.014960.00847−0.002390.03230
Equal variances not assumed 2.06217.450.0540.014960.00725−0.000310.03023
SGIEqual variances assumed0.7560.392−0.363280.719−0.016730.04607−0.111110.07764
Equal variances not assumed −0.33911.050.741−0.016730.04940−0.125390.09192
DIEqual variances assumed1.5610.222−0.173280.864−0.014740.08496−0.188770.15930
Equal variances not assumed −0.27924.630.782−0.014740.05279−0.123540.09407
LEVIEqual variances assumed0.1140.738−0.469280.643−0.002010.00429−0.010800.00677
Equal variances not assumed −0.58520.640.565−0.002010.00344−0.009170.00515
SGAIEqual variances assumed0.1280.723−0.294280.771−0.011100.03771−0.088340.06615
Equal variances not assumed −0.28311.640.782−0.011100.03919−0.096770.07458
TATAEqual variances assumed2.0930.159−0.817280.421−0.054560.06681−0.191410.08228
Equal variances not assumed −1.3422.940.193−0.054560.04068−0.138720.02960

Descriptive statistics of M-score sample wise

SerialNMinimumMaximumMeanStd. deviation
110−3.04−2.06−2.58140.35423
210−2.79−1.58−2.25100.43086
310−3.456.13−1.62612.83577
410−2.851.40−1.89051.26803
510−3.52−0.92−2.29690.84367
610−3.24−1.14−2.41440.60587
710−3.411.29−2.01521.32040
810−2.99−1.45−2.30350.53440
910−3.510.41−2.07471.27746
1010−3.420.77−2.18881.19835
1110−6.910.48−4.07112.03871
1210−3.580.36−2.24601.04476
1310−7.25−1.87−2.85821.57963
1410−4.07−1.70−2.84860.75177
1510−3.431.11−2.19071.38028
1610−3.60−1.31−2.45710.76081
1710−3.29−0.56−2.29620.82311
1810−3.260.94−1.88201.47741
1910−3.20−1.04−2.29850.67837
2010−3.271.40−1.88921.43707
2110−3.21−1.52−2.45290.50180
2210−4.270.63−2.56121.25942
2310−4.17526.5750.3891167.31541
2410−5.621.52−1.91021.91271
2510−3.30−1.00−2.30950.62478
2610−3.18−1.59−2.34980.51187
2710−3.42−0.74−2.20470.83566
2810−3.440.74−1.87731.29217
2910−3.465.11−1.78042.46611
3010−3.493.98−1.11982.77909

Group statistics for the year 2009

Group statistics
GroupNMeanStd. deviationStd. error mean
DSRILikely manipulator191.46517311.681525470.38576840
Nonlikely manipulator111.64217910.465152070.14024863
GMILikely manipulator190.99252500.251388640.05767251
Nonlikely manipulator111.06440430.227430110.06857276
AQILikely manipulator190.92602450.344054000.07893140
Nonlikely manipulator110.99919520.023881580.00720057
SGILikely manipulator191.18319990.110863300.02543379
Nonlikely manipulator111.19938550.070344270.02120959
DILikely manipulator192.21879495.292668451.21422142
Nonlikely manipulator110.96996500.205632350.06200049
LEVILikely manipulator191.00676280.049576340.01137359
Nonlikely manipulator110.98892340.014916800.00449758
SGAILikely manipulator191.09351060.107457520.02465245
Nonlikely manipulator111.12530900.112118040.03380486
TATALikely manipulator19−0.00975540.096465910.02213080
Nonlikely manipulator110.02161180.064635680.01948839

Group statistics for the year 2010

Group statistics
GroupNMeanStd. deviationStd. error mean
DSRILikely manipulator2424.7467117.5018023.98496
Nonlikely manipulator61.75241.251830.51106
GMILikely manipulator240.81640.139390.02845
Nonlikely manipulator60.77050.174740.07134
AQILikely manipulator240.90870.607730.12405
Nonlikely manipulator61.00250.033260.01358
SGILikely manipulator241.20170.118450.02418
Nonlikely manipulator61.16870.059380.02424
DILikely manipulator241.27010.530610.10831
Nonlikely manipulator61.06350.490310.20017
LEVILikely manipulator241.47362.425190.49504
Nonlikely manipulator60.98920.017900.00731
SGAILikely manipulator241.20090.215480.04398
Nonlikely manipulator61.25520.155290.06339
TATALikely manipulator24−0.00450.203080.04145
Nonlikely manipulator6−0.02020.052160.02129

Group statistics for the year 2011

Group statistics
GroupNMeanStd. deviationStd. error mean
DSRILikely manipulator161.61232.067790.51695
Nonlikely manipulator141.31140.553320.14788
GMILikely manipulator161.33600.310460.07762
Nonlikely manipulator141.16470.243580.06510
AQILikely manipulator160.98250.032680.00817
Nonlikely manipulator140.97000.014860.00397
SGILikely manipulator161.39120.105860.02646
Nonlikely manipulator141.52300.247270.06608
DILikely manipulator161.05590.167700.04192
Nonlikely manipulator141.05540.471150.12592
LEVILikely manipulator161.00890.024450.00611
Nonlikely manipulator140.99800.021910.00585
SGAILikely manipulator161.47602.363010.59075
Nonlikely manipulator140.83650.191460.05117
TATALikely manipulator16−0.10890.220550.05514
Nonlikely manipulator14−0.01770.049720.01329

Group statistics for the year 2012

Group statistics
GroupNMeanStd. deviationStd. error mean
DSRILikely manipulator110.73650.347750.10485
Nonlikely manipulator191.47230.590120.13538
GMILikely manipulator111.14470.227720.06866
Nonlikely manipulator191.08810.248840.05709
AQILikely manipulator111.00940.031240.00942
Nonlikely manipulator190.97910.016180.00371
SGILikely manipulator111.27020.095310.02874
Nonlikely manipulator191.35330.145920.03348
DILikely manipulator110.93460.104220.03142
Nonlikely manipulator191.15890.405760.09309
LEVILikely manipulator111.01980.029260.00882
Nonlikely manipulator191.01280.011400.00262
SGAILikely manipulator110.95090.421230.12701
Nonlikely manipulator190.84240.241590.05543
TATALikely manipulator11−0.05730.111120.03350
Nonlikely manipulator19−0.01360.039180.00899

Group statistics for the year 2013

Group statistics
GroupNMeanStd. deviationStd. error mean
DSRILikely manipulator180.87400.313140.07381
Nonlikely manipulator121.55160.621070.17929
GMILikely manipulator181.15000.169060.03985
Nonlikely manipulator121.68950.763050.22027
AQILikely manipulator181.00790.018870.00445
Nonlikely manipulator120.99100.018420.00532
SGILikely Manipulator181.06500.110540.02605
Nonlikely manipulator121.10640.120960.03492
DILikely manipulator180.93500.239710.05650
Nonlikely manipulator121.23761.232900.35591
LEVILikely manipulator181.00200.014970.00353
Nonlikely manipulator121.00050.008690.00251
SGAILikely manipulator181.53611.842580.43430
Nonlikely manipulator120.97940.313550.09051
TATALikely manipulator18−0.06360.081060.01911
Nonlikely manipulator12−0.00800.031900.00921

Group statistics for the year 2014

Group statistics
GroupNMeanStd. deviationStd. error mean
DSRILikely manipulator231.27321.702090.35491
Nonlikely manipulator72.27340.763690.28865
GMILikely manipulator230.74620.404980.08444
Nonlikely manipulator70.97370.150310.05681
AQILikely Manipulator231.01390.057960.01208
Nonlikely manipulator70.98510.031550.01192
SGILikely manipulator231.05730.087860.01832
Nonlikely manipulator70.96610.131040.04953
DILikely manipulator231.14370.870260.18146
Nonlikely manipulator71.00000.154150.05826
LEVILikely manipulator231.00240.006580.00137
Nonlikely manipulator70.99770.010790.00408
SGAILikely manipulator231.51802.011830.41950
Nonlikely manipulator71.16460.169340.06400
TATALikely manipulator23−0.04600.081350.01696
Nonlikely manipulator7−0.02270.040800.01542

Group statistics for the year 2015

Group statistics
GroupNMeanStd. DeviationStd. Error mean
DSRILikely manipulator221.99912.728160.58165
Nonlikely manipulator81.88730.550490.19463
GMILikely manipulator221.21801.325450.28259
Nonlikely manipulator81.14040.354220.12524
AQILikely manipulator221.02900.068940.01470
Nonlikely manipulator80.99750.015710.00555
SGILikely manipulator221.00320.079450.01694
Nonlikely manipulator80.98540.099620.03522
DILikely manipulator221.00500.165480.03528
Nonlikely manipulator80.90630.104960.03711
LEVILikely manipulator220.96790.169820.03621
Nonlikely manipulator81.00030.009420.00333
SGAILikely manipulator221.12080.123700.02637
Nonlikely manipulator81.10120.127730.04516
TATALikely manipulator22−0.06040.100460.02142
Nonlikely manipulator8−0.02290.045260.01600

Group statistics for the year 2016

Group statistics
GroupNMeanStd. deviationStd. error mean
DSRILikely manipulator251.20040.803900.16078
Nonlikely manipulator52.47060.663710.29682
GMILikely manipulator250.85640.264130.05283
Nonlikely manipulator50.73410.389180.17405
AQILikely manipulator250.98410.163630.03273
Nonlikely manipulator50.97060.039070.01747
SGILikely manipulator250.99130.064980.01300
Nonlikely manipulator51.03410.063660.02847
DILikely manipulator252.33477.026711.40534
Nonlikely manipulator50.99010.068880.03080
LEVILikely manipulator251.15590.737190.14744
Nonlikely manipulator51.00260.002100.00094
SGAILikely manipulator251.16140.160470.03209
Nonlikely manipulator51.08700.073460.03285
TATALikely manipulator25−0.05360.101690.02034
Nonlikely manipulator50.00670.039460.01765

Group statistics for the year 2017

Group statistics
GroupNMeanStd. deviationStd. error mean
DSRILikely manipulator210.91270.415190.09060
Nonlikely manipulator91.76790.710720.23691
GMILikely manipulator210.66861.129640.24651
Nonlikely manipulator90.94020.180510.06017
AQILikely manipulator211.01560.028450.00621
Nonlikely manipulator91.43091.347740.44925
SGILikely manipulator211.08510.088900.01940
Nonlikely manipulator91.10890.076580.02553
DILikely manipulator210.96540.124870.02725
Nonlikely manipulator90.91940.343990.11466
LEVILikely manipulator211.00670.010110.00221
Nonlikely manipulator91.00740.005540.00185
SGAILikely manipulator211.02140.122250.02668
Nonlikely manipulator91.00100.058640.01955
TATALikely manipulator21−0.05840.096070.02096
Nonlikely manipulator9−0.02710.039120.01304

Group statistics for the year 2018

Group statistics
GroupNMeanStd. deviationStd. error mean
DSRILikely manipulator220.83990.302620.06452
Nonlikely manipulator81.71440.435170.15385
GMILikely manipulator220.96090.441170.09406
Nonlikely manipulator81.08450.082210.02907
AQILikely manipulator220.99590.021880.00466
Nonlikely manipulator80.98090.015710.00555
SGILikely manipulator221.25340.107190.02285
Nonlikely manipulator81.27010.123870.04380
DILikely manipulator220.94570.236220.05036
Nonlikely manipulator80.96040.044750.01582
LEVILikely manipulator221.00290.011310.00241
Nonlikely manipulator81.00490.006930.00245
SGAILikely manipulator220.86770.089440.01907
Nonlikely manipulator80.87880.096840.03424
TATALikely manipulator22−0.09080.186290.03972
Nonlikely manipulator8−0.03620.024840.00878

DSRI=DaysSalesinReceivalest÷DaysSalesinReceivalest1
GMI=GrossMarginIndext1÷GrossMarginIndext
AQI=(1CurrentAsset+PropertyPlant & EquipmentTotalAsset)t÷(1CurrentAsset+PropertyPlant & EquipmentTotalAsset)t1
SGI=Salest÷Salest1
DI=(DepreciationDepreciation+PropertyPlant & Equipment)t1÷(DepreciationDepreciation+PropertyPlant & Equipment)t
SGAI=(Sales,General&AdministrativeExpreseSales)t÷(Sales,General&AdministrativeExpreseSales)t1
TATA=(TotalAccrualsTotalAsset)t
LEVI=(TotalLiabilitiesTotalAsset)t÷(TotalLiabilitiesTotalAsset)t1

References

Aghghaleh, S. F., Mohamed, Z. M., & Rahmat, M. M. (2016). Detecting financial statement frauds in Malaysia: Comparing the abilities of Beneish and Dechow models. Asian Journal of Accounting and Governance, 7, 5765. doi: 10.17576/AJAG-2016-07-05.

Akra, R. M., & Chaya, J. K. (2020). Testing the effectiveness of Altman and Beneish models in detecting financial fraud and financial manipulation: Case study Kuwaiti stock market. International Journal of Business and Management, 15(10), 7081. doi: 10.5539/ijbm.v15n10p70.

Alo, J. (2021). ONE Bank cooks the books to siphon cash to owners. available from: https://www.tbsnews.net/economy/banking/one-bank-cooks-books-siphon-cash-owners-282673 (accessed 7 August 2022).

Anning, A. A., & Adusei, M. (2020). An analysis of financial statement manipulation among listed manufacturing and trading firms in Ghana. Journal of African Business. doi: 10.1080/15228916.2020.1826856.

Arman, M., & Sharmin, S. (2019). Likelihood of a company's manipulation of its financial statement: An empirical analysis using Beneish M-score model. International Conference on Management and Information Systems September (Vol. 29, pp. 30).

Bangladesh Bank (2022). Economic data. available from: https://www.bb.org.bd/en/index.php/econdata/index.

Beneish, M. (1999). The detection of earning manipulation. Financial Analyst Journal, 55(5), 2436.

Beneish, M. D., & Nichols, C. (2005). Earnings quality and future returns: The relation between accruals and the probability of earnings manipulation. available from: SSRN 725162.

Bhasin, M. L. (2015). An empirical study of frauds in the banks. European Journal of Business and Social Sciences, 4(7), 112.

Bhasin, M. L. (2016). Survey of creative accounting practices: An empirical study. Wulfenia, 23(1), 143162.

Daily Star (2021). Money laundering: Interpol issues red notice for PK halder. available from: https://www.thedailystar.net/frontpage/news/money-laundering-interpol-issues-red-alert-pk-halder-2024545 (accessed 3 August 2022).

DeAngelo, L. (1988). Discussion of evidence of earnings management from the provision for bad debts. Journal of Accounting Research, 26, 3240.

Dhaka Tribune (2019). Loan scams, soaring NPLs, bank owners increasing clout hurt banking sector. available from: https://archive.dhakatribune.com/business/banks/2019/01/02/loan-scams-soaring-npls-bank-owners-increasing-clout-hurt-banking-sector (accessed 3 August 2022).

Durana, P., Blazek, R., Machova, V., & Krasnan, M. (2022). The use of Beneish M-scores to reveal creative accounting: Evidence from Slovakia. Equilibrium. Quarterly Journal of Economics and Economic Policy, 17(2), 481510. doi: 10.24136/eq.2022.017.

Financial Express (2020). NPL and its impact on the banking sector of Bangladesh. available from: https://thefinancialexpress.com.bd/views/npl-and-its-impact-on-the-banking-sector-of-bangladesh-1595343188 (accessed 3 August 2022).

Ghosh, R., Sen, K. K., & Riva, F. (2020). Behavioral determinants of non-performing loans in Bangladesh. Asian Journal of Accounting Research, 5(2), 327340. doi: 10.1108/AJAR-03-2020-0018.

Gupta, P. K., & Gupta, S. (2015). Corporate frauds in India – perceptions and emerging issues. Journal of Financial Crime, 22(1), 79103. doi: 10.1108/JFC-07-2013-0045.

Hasan, M. S., Omar, N., Barnes, P., & Handley-Schachler, M. (2017). A cross-country study on manipulations in financial statements of listed companies: Evidence from Asia. Journal of Financial Crime, 24(4), 656677. doi: 10.1108/JFC-07-2016-0047.

Hasan, M. S., Omar, N., Rahman, R. A., & Hossain, S. Z. (2016). Corporate attributes and corporate accruals. Aestimatio: The IEB International Journal of Finance, (12), 2447.

Hasan, S., Rahman, R. A., & Hossain, S. Z. (2014). Corporate accruals practices of listed companies in Bangladesh. Journal of Corporate Governance, Insurance and Risk Management, 1(1), 1243.

Healy, P. M. (1985). The effect of bonus schemes on accounting decisions. Journal of accounting and economics, 7(1–3), 85107.

Hołda, A. (2020). Using the Beneish M-score model: Evidence from non-financial companies listed on the warsaw stock Exchange. Investment Management and Financial Innovations, 17(4), 389401. doi: 10.21511/imfi.17(4).2020.33.

Huang, S. Y., Lin, C. C., Chiu, A. A., & Yen, D. C. (2017). Fraud detection using fraud triangle risk factors. Information Systems Frontiers, 19(6), 13431356. doi: 10.1007/s10796-016-9647-9.

Jones, J. J. (1991). Earnings management during import relief investigations. Journal of Accounting Research, 29(2), 193228. doi: 10.2307/2491047.

Kamal, M. E. M., Salleh, M. F. M., & Ahmad, A. (2016). Detecting financial statement fraud by Malaysian public listed companies: The reliability of the Beneish M-Score model. Jurnal Pengurusan, 46, 2332.

Kizil, C., & Kaşbaşı, B. (2018). Accounting scandals and eye-catching frauds: USA-Japan comparison by considering the role of auditing. Journal of Asian Research, 2(3). doi: 10.22158/jar.v2n3p123.

Mamo, J., & Shehu, I. (2017). An estimating model for accounting manipulation phenomena. Case of Albania. Interdisciplinary Journal of Research and Development, 4(2), 13044.

Özcan, A. (2018). The use of Beneish model in forensic accounting: Evidence from Turkey. Journal of Applied Economics and Business Research, 8(1), 5767.

Ramírez-Orellana, A., Martínez-Romero, M. J., & Mariño-Garrido, T. (2017). Measuring fraud and earnings management by a case of study: Evidence from an international family business. European Journal of Family Business, 7(1-2), 4153. doi: 10.1016/j.ejfb.2017.10.001.

Repousis, S. (2016). Using Beneish model to detect corporate financial statement fraud in Greece. Journal of Financial Crime, 23(4), 10631073. doi: 10.1108/JFC-11-2014-0055.

Sakib, I. A. (2019). Detection of earnings manipulation practices in Bangladesh. International Journal of Management, Accounting and Economics, 6(8), 616631.

Samuel, N. (2022). Probable earning manipulation and fraud in banking sector. Empirical study from East Africa. Cogent Economics and Finance, 10(1), 2083477. doi: 10.1080/23322039.2022.2083477.

Shakouri, M. M., Taherabadi, A., Ghanbari, M., & Jamshidinavid, B. (2021). Explaining the Beneish model and providing a comprehensive model of fraudulent financial reporting (FFR). International Journal of Nonlinear Analysis and Applications, 12, 3948. doi: 10.22075/IJNAA.2021.4793.

Subiyono, A. K., & Suardi, I. (2020). The Effect of fraud detection using M-Score on expected returns in Indonesian public listed companies. PalArch's Journal of Archaeology of Egypt/Egyptology, 17(7), 27552763.

Tahmina, A., & Naima, J. (2016). Detection and analysis of probable earnings manipulation by firms in a developing country. Asian Journal of Business and Accounting, 9(1), 5982.

Tassadaq, F., & Malik, Q. A. (2015). Creative accounting & financial reporting: model development & empirical testing. International Journal of Economics and Financial Issues, 5(2), 544551.

Umar, H., Partahi, D., & Purba, R. B. (2020). Fraud diamond analysis in detecting fraudulent financial report. International Journal of Scientific and Technology Research, 9(3), 66386646.

Valaskova, K., & Fedorko, R. (2021). Beneish M-score: A measure of fraudulent financial transactions in global environment? SHS Web of Conferences (Vol. 92, pp. 02064). EDP Sciences.

Warshavsky, M. (2012). Analyzing earnings quality as a financial forensic tool. Financial Valuation and Litigation Expert Journal, 39, 1620.

Further reading

Abbas, A. (2017). Earnings fraud and financial stability. Asia Pacific Fraud Journal, 2(1), 117134. doi: 10.21532/apfj.001.17.02.01.010.

Razali, W.A.A.W.M., & Arshad, R. (2014). Disclosure of corporate governance structure and the likelihood of fraudulent financial reporting. Procedia-Social and Behavioral Sciences, 145, 243253. doi: 10.1016/j.sbspro.2014.06.032.

Corresponding author

Ratan Ghosh can be contacted at: ratanghosh18@gmail.com

Related articles