Benenish m score model in market returns measurement empirical evidence in vietnam

In Vietnam, the research work "Using the Beneish M-score model to assess the quality of financial statements in Vietnam" by Vo Minh Duong 2016 has demonstrated the feasibility of using t

MINISTRY OF EDUCATION AND TRAINING THE STATE BANK OF VIETNAM

BANKING UNIVERSITY OF HO CHI MINH CITY

*********************

GRADUATION THESIS

DUONG MINH KHANH

BENEISH M-SCORE MODEL IN MARKET RETURNS MEASUREMENT: EMPIRICAL EVIDENCE IN

VIETNAMGRADUATE THESIS

MAJOR: BANKING - FINANCE

CODE: 7340201

Ho Chi Minh City, November 2022

MINISTRY OF EDUCATION AND TRAINING THE STATE BANK OF VIETNAM

BANKING UNIVERSITY OF HO CHI MINH CITY

CODE: 7340201

Supervisor: PhD NGUYEN DUY LINH

Ho Chi Minh City, November 2022

COMMENTS OF THE GRADUATE THESIS ADVISOR

Ho Chi Minh City, …………, … …… 2022

Advisor

PhD Nguyen Duy Linh

COMMENTS OF REVIEW BOARD

Ho Chi Minh City, …………, … …… 2022

Chairman of the Review Board

ABSTRACT

The thesis aims to apply the Beneish M-score model to investigated a potential link between the probability of manipulation in financial statements of 4,336 observations corresponding from 2011-to 2021 From there, classifying potentially manipulation companies according to M-score, the study also shows a significant difference in stock returns between these two groups of companies Companies with a high risk of manipulation (high M-score) will have low stock returns, and vice versa

At the same time, the regression model finding the relationship between the return to scale adjusted for the M-score index and other predictors also shows a

negative correlation with the M-score, a positive correlation with the M-score,

Accruals, Momentum, MVE and BTM

Thereby, the M-score is proven to have a significant impact on the profitability when impacting other forecast plates

Keywords: Beneish M-score, financial statement fraud, stock return

DECLARATION

I hereby declare that this is my research work, and the research results presented

in the thesis are honest and objective The sources of information cited in this thesis are clearly indicated I am responsible for my thesis

Signature

Khanh Duong Minh Khanh

ACKNOWLEDGEMENT

First of all, I am grateful and would like to send my sincerest and special thanks

to my supervisor, PhD Nguyen Duy Linh for helping me to finish this challenging thesis Despite my defects throughout the process of performing the thesis, he always showed his caring and enthusiasm to support me to overcome Without PhD Nguyen Duy Linh con0scientious guidance and support, this thesis cannot be successfully accomplished

Next, I would like to express my deepest gratitude to the teachers who are teaching at the Banking University of Ho Chi Minh City, who have spread the fire, passion and knowledge about economics from the most basic subjects, helping me to have a background in finance - banking

Finally, I would like to thank to my family, friends, and colleagues who have always encouraged and helped me in the process of studying and researching the topic Although I have tried a lot, the thesis is not free of shortcomings I hope to receive the sympathy, guidance, help, and support of my colleagues

TABLE OF CONTENTS

ABSTRACT i

DECLARATION ii

ACKNOWLEDGEMENT iii

LIST OF TABLES AND FIGURES vi

CHAPTER I: INTRODUCTION 1

1.1 Rationales 1

1 2 Research objectives and research questions 2

1.2.1 Research objectives 2

1.2.2 Research questions 3

1 3 Subject and scope of the study subject 3

1.3.1 Subject 3

1.3.2 Research scope 3

1.4 Structure of the study 4

CHAPTER 2: THEORETICAL BASIS AND LITERATURE REVIEW 5

2.1 Theoretical overview 5

2.1.1 Definition of financial statement 5

2.1.2 Definition of financial reporting fraud 5

2.1.3 Theory explaining the motives for fraudulent financial statements 5

2.1.4 Acts of performing fraudulent financial statements 7

2.2 Literature review 8

2.2.1 Identify distortions in financial statements 8

2.2.2 Beneish M-score model 8

2.2.3 Research on developing the Beneish M-score model 11

2.2.4 Research on the correlation between financial statement distortion and stock market profitability 12

CONCLUSION CHAPTER 2 15

CHAPTER 3: DATA AND RESEARCH METHODS 16

3.1 Research data 16

3.2 Research methods 17

3.2.1 M – Score model 18

3.2.2 M – score and forecast future expected return 19

3.2.2.1 Calculate adjusted rate of return on a security (BHSAR) 19

3.2.2.2 Comparison of BHSAR (+1) of two groups of companies classified by M-score 20

3.2.2.3 Consider the relationship between M – Score and profitability 20

3.2.2.4 Consider the relationship between profitability and other factors 20

3.2.2.5 Empirical model about the relationship between M-score and return on stock market 21

3.2.2.6 Regression fit method – GMM 22

CONCLUSION CHAPTER 3 23

CHAPTER 4: RESEARCH RESULTS AND DISCUSSION 24

4.1 Financial distortion in Vietnam 24

4.1.1 Financial distortion from the perspective of the exchange 24

4.1.2 Fraud risk according to the capitalization classification 28

4.2 Impact of financial fraud on stock returns 29

4.3 M – Score and the influence of other factors on profitability 30

4.3.1 Correlation matrix between M - score and other influencing factors 30

4.3.2 Combined effect of M-score and other factors on stock return 31

4.3.2.1 Find the defects of the model 31

4.3.2.2 Regression by GMM method 34

4.3.2.3 Interpretation of regression results 35

CONCLUSION CHAPTER 4 36

CHAPTER 5: CONCLUSIONS AND RECOMMENDATIONS 37

5.1 Concluding remarks 37

5.2 Limitations and suggestions for further studies 38

5.3 Proposing solutions to limit fraud in financial statements 39

CONCLUSION CHAPTER 5 41

ENGLISH REFERENCES 42

VIETNAMESE REFERENCES 43

APPENDIX 44

LIST OF TABLES AND FIGURES

Figure 4.1 Statistics on the number of companies at risk of fraud on stock exchanges Figure 4.2 Statistics on the number of companies at risk of fraud over the years

Figure 4.3 Statistics by percentage of companies at risk of fraud over the years

Table 3.1 Statistics of research observations on HOSE and HNX 2011 - 2021

Table 4.1 Average M-score of fraudulent and non-fraudulent firms by exchange

Table 4.2 Fraud risk according to the capitalization classification

Table 4.3 Compare the average BHSAR return (t+1) of the two groups of companies Table 4.4 Correlation matrix between M - score and other influencing factors

Table 4.5 Regression results

Table 4.6 Table of test results for VIF coefficient

Table 4.7 Modified Wald Test Results

Table 4.8 Wooldridge Test Result Table

Table 4.9 Table of regression results according to GMM method

CHAPTER I: INTRODUCTION 1.1 Rationales

Nowadays, financial statements play a central and vital role in making investment decisions in the stock market Financial statements provide information about the assets, sources of capital owned, results of operations, cash flows, and interpretation of material transactions during the business period enterprise Thereby helping users of financial statements check and monitor the current status of business activities in the past period, how enterprises use capital, and assess the ability to raise capital and generate revenue potential, future profits

However, many businesses have committed fraud and manipulated financial statements, and this behavior is increasing day by day For various purposes, a company's management may seek to influence and alter the figures in the reports to make financial results better, but not to reflect the actual situation Typically, the manipulation of the global financial statements was Enron in 2002 – the seventh-largest company in the US The Enron scandal caused a loss of more than $80 billion in capitalization for investors investing in this company Also, it led to the collapse of the fifth auditing firm in the world at that time - Arthur Andersen After Eron was the collapse of Lehman Brothers when fraudulently recording short-term debt financing operations into sales worth up to $50 billion This scandal cost Lehman Brothers' auditing company at the time, EY, up to USD 109 billion to settle this case and compensate Lehman's investors

In Vietnam, the financial scandal of Cuu Long Pharmaceutical Company in

2014 or the manipulation of Viet Nhat JVC Medical Joint Stock Company in 2015 up

to VND 900 billion are typical examples and many the case of other listed companies

on the stock exchange These behaviors have highly negative impacts on the capital market and the economy in general It leads to financial statements' quality that does not guarantee stakeholder decision-making Therefore, assessing the reliability of

financial statements is always an urgent need when not everyone is capable of analyzing and identifying the company's actual situation

In Vietnam, the research work "Using the Beneish M-score model to assess the quality of financial statements in Vietnam" by Vo Minh Duong (2016) has demonstrated the feasibility of using the Beineish M-score model to examine the quality of financial statements by detecting signs of data distortion of companies in Vietnam However, no research explicitly studies how this distortion affects the return

of the company's stock during the Covid-19 Pandemic Therefore, the range of studies assesses financial statement fraud through the M-score model and extends to understanding the correlation between financial statement fraud and the profitability of stocks in the stock market before and during the Covid-19 epidemic

The study results can apply as a reference measure for investors and credit providers to evaluate the company's transparency and add a supporting tool to Invest in the stock market

1 2 Research objectives and research questions

 Proposing and giving appropriate solutions to limited fraud in financial statements in Vietnam

 Second, what is a suitable model to evaluate the impact of M-Score model

on stock market return?

 Third, correlation of M-score model with the profitability of listed companies on Vietnam stock market

 Fourth, what is solution to limited fraud in financial statements in Vietnam?

1 3 Subject and scope of the study subject

 Scope of space: The companies listed on Ho Chi Minh Stock Exchange and

Hanoi Stock Exchange

 Scope of time: The data are collected from 2011 to 2021

 Research content: In this study, the author focuses on studying the impact

of financial statement distortion on stock profitability Thereby, we clarify the influence of the Beneish M-score model to detect fraudulent financial statements and find out the relationship between it and the profitability of companies on the stock market in Vietnam

1.4 Structure of the study

The content of the study is divided into five specific chapters:

 Chapter 1: Introduction to the topic

 Chapter 2: Theoretical basis and literature review

 Chapter 3: Data and research methods

 Chapter 4: Research result and discussion

 Chapter 5: Conclusion and recommendations

CHAPTER 2: THEORETICAL BASIS AND LITERATURE REVIEW

2.1 Theoretical overview

2.1.1 Definition of financial statement

Financial statements are consolidated reports that show the current financial position of an enterprise and how that business uses capital from shareholders and credit providers

The financial statement includes four main reports: (1) the balance sheet shows

a company's financial position at a point in time, (2) the income statement shows how a company's profits have formed over a period, and (3) cash flow statements show the cash inflows and outflows of a company over some time and the last is the statement of changes in equity

2.1.2 Definition of financial reporting fraud

Financial statements fraud is the company's use of techniques to change information on financial statements, making the numbers no longer truthful and objective to achieve specific purposes These could be reducing taxes, matching reporting with analysts' forecasts, securing debt contracts, and managers achieving short-term compensation However, fraud in financial statements causes information distortion, affects the market and decisions of investors and credit providers, and reduces the quality of the company's financial statements

2.1.3 Theory explaining the motives for fraudulent financial statements

 Fraud Triangle by Donald R Cressey (1919 - 1987)

Cressey focuses on analyzing fraud from the perspective of embezzlement and embezzlement by surveying about 200 cases of economic crimes to find out the causes

of the above violations He came up with a model: The Fraud Triangle consisting of Motivation - Attitude - Opportunity

 Managers gain short-term compensation

Healy (1985) argues that short-term bonuses, dividends, or incentives motivate managers to distort financial statements His subsequent study et al (1999) further extended that senior managers have a high proportion of stock or stock options in compensation, while lower-level managers have cash account for a high percentage As

a result, low-level managers tend to focus on maximizing short-term bonuses Haulthausen, Lacker, and Sloan (1995) and Guidry, Leone, and Rock (1999) give similar results Cheng and Warfield (2005) argue that managers who hold many common stocks will sell their shares in the future, so they have an incentive to distort financial statements to give an excellent signal to the market Research results also show that such managers will report profits that meet or exceed analyst expectations

 Similar to analysts' forecasts

Graham, Harvey, and Rajpogal (2005) interviewed more than 400 CFOs and found that 73.5% of respondents agreed or strongly agreed that the analyst consensus is for earnings per share The current quarter is the key criterion when reporting quarterly earnings, and they tend to guide financial statements by these analyses Most managers want to avoid projects with a positive NPV but cause earnings to fall in the current quarter They want to trade economic value for a stable return because it reduces uncertainty about returns

 Achieve high prices on initial public offerings (IPOs) or additional issues

Companies want to manipulate financial statements to increase the amount of money from IPOs or additional issues of shares Ducharme, Malatesta, and Sefcik (2002) suggest that pre-IPO extraordinary accruals have a positive relationship with the initial value of the firm However, this will reduce investors' returns in subsequent years, an average of 3 years, according to the research of Teoh, Welch, and Wong (1998)

 Negotiate better contract terms and avoid violations a clause in the loan contract

According to Bowen, Ducharme, and Shores (1995), a company can get better contract conditions from suppliers and related parties if it can report stable profits

Debt contracts often include clauses about the company's earnings Thus, managers may adopt a policy of increasing reported earnings or other financial statement items in order to avoid a breach or proximity to a breach of such provisions

In addition, beautifying financial statements can increase the willingness of lenders or suppliers to obtain short-term credit

2.1.4 Acts of performing fraudulent financial statements

Distortion of financial statements often has the following mechanism (Schweser, 2015)

 Aggressive revenue recognition includes the activity of accelerating the supply

of goods to distribution channels beyond their ability to sell (channel stuffing), recognizing revenue when the goods do not yet reach consumers (bill and hold sales), or pretending to sell and then re-enter (outright fake sales);

 The company uses a finance lease for outsourced assets to exclude rent from expenses to increase profits;

 Record revenue/non-operating income into operating revenue/income and recognize operating expenses as operating expenses;

 Record gains in net income and losses in OCI (Other comprehensive income);

 Inappropriate selection of depreciation estimation models;

 Recognition of current items as long-term items;

 Recognition of provisions that are higher or lower than they are needed;

 Under recognition of tangible assets and high recognition of intangible assets for M&A related purpose;

 Performing transactions affecting cash flow from business activities;

 Record cash flow from operating activities into investment cash flow

2.2 Literature review

2.2.1 Identify distortions in financial statements

There are many methods to identify companies with distorted financial statements Burcu and Guray (2010) classified into three main technical groups:

The first is the discretionary accruals technique, which is represented by the research of De Angelo (1986) These studies focus on evaluating accruals to detect distorted accruals

The second is applying an artificial neural network model with representatives such as Green & Choi (1997) and Fanning & Gogger (1998) Green & Choi (1997) presented an artificial neural network model using endogenous financial data; Fanning

& Gogger (1998) used an artificial neural network model to detect administrative fraud They found a model of 8 variables with high fraud detection ability

The third is the statistical techniques developed by Beneish (1997, 1999) and Hansen (1996) Hansen (1996) uses a powerful generalized qualitative response model (EGB2) to predict financial reporting distortions based on data from globally listed accounting firms The EGB2 model consists of a probit model and a logit model with a tight combination of a trade-off between Type I and Type II errors

2.2.2 Beneish M-score model

Research by Messod D.Beneish (2003) has built an M - Score model to determine whether companies have distorted financial statements The variables in the model are designed to identify inaccuracies in the financial statements stemming from the motives and conditions that motivate the company to commit fraud The results show a systematic relationship between the likelihood of deviations and the variables

in the financial statements

Beneish model includes eight factors: Collection period index (DSRI), gross profit index (GMI), asset quality index (AQI), sales growth index (SGI), index depreciation (DEPI), SG&A expense index (SGAI), accrual value index (TATA), financial leverage index (LVGI)

Table 2.1 The formula of eight factors in Beneish model DSRI = (Receivable / Sales)t / (Receivable / Sales)(t-1)

GMI = Gross margin (t-1) / Gross margin t

[ 1- (Current Assets + Fixed Assets) / Total Assets]t / [ 1- (Current Assets + Fixed Assets) / Total Assets](t-1)

SGI = Sales t / Sales (t-1)

DEPI = [Depreciation / (Fixed Assets + Depreciation)] (t-1)

/ [Depreciation / (Fixed Assets + Depreciation)]t

SGAI = (SG&A / Sales)t / (SG&A / Sales)(t-1)

TATA = [(Net income – CFO) / Total Assets ]t

LVGI = (Liabilities / Total Assets)t / (Liabilities /Total Assets)(t-1)

Then, the author uses a weighted exogenous sample maximum likelihood (WESML) probit model and an unweighted probit unit probability model to get the

results M-Score model:

The author calculates the threshold value of the model as -1.78, companies with

M - Score higher than -1.78 will be marked as having manipulated financial statements,

and the rest will have no sign of manipulation of financial statements main

M – score = -4.84 + 0.0920(DSRI) + 0.528(GMI) + 0.404(AQI) + 0.892(SGI) +

0.115(DEPI) – 0.172(SGAI) + 4.679(TATA) – 0.327(LVGI)

The meaning of the independent variables in the model:

 DSRI: Measures the ratio of receivables to total sales next year compared to the

ratio of receivables to total sales the previous year If this ratio is greater than one, the company has recorded more revenue but has not yet collected money from customers It may be the company's policy to increase competition but at the same time increase disproportionately This ratio is also an indication of revenue manipulation

 GMI: Measure the gross profit margin of the previous year compared to the

following year If this ratio is greater than one, the company's gross profit relative to sales has decreased next year, which creates a negative sign for the company's ability to operate in the future Therefore, companies with lousy potential for future growth are more likely to manipulate financial statements

 AQI: Asset quality is measured by the ratio of non-fixed long-term assets to total

assets A ratio is greater than one means that the company is more likely to have

an increased capitalize and defer costs, leading to manipulate financial statements

 SGI: Growth does not imply distortion of financial statements However, the

board of directors of companies with higher growth this year than last year will

be under more pressure to maintain their financial position and maintain their financial position Such as fulfilling commitments, thereby creating many incentives to distort financial statements

 DEPI: This indicator uses the depreciation rate of the previous year compared to

the current year The indicator has a value greater than one that conveys the message that the company is reducing its depreciation record, which may be due

to the manager increasing estimates of the useful lives of assets or adopting measures to increase profits

 SGAI: This indicator shows that if the ratio of SG&A expenses to sales has

increased in the following year relative to the previous year, there are likely to

appear signs that the company will operate in the future Less effective performance in the future, making it more likely that management has an incentive to manipulate the financial statements

 TATA: Total Accruals is the change in working capital accounts excluding cash

and depreciation Total accrual represents the ability of management to make independent decisions to change earnings This variable also represents the difference between operating profit and cash flow from operating activities The larger the accrual value, the more likely the company has intentionally manipulated its financial statements

 LVGI: A ratio greater than one means the company uses more debt in its capital

structure in the following year than in the previous year When a company uses more debt, it means that the company will accept more terms in debt contracts (which often include many different requirements to ensure the ability to repay the debt), leading to its management It distorts financial statements to show creditors the company's financial strength partly

2.2.3 Research on developing the Beneish M-score model

Roychowdhury (2006) develops an experimental method to detect distortion of actual performance Roychowdhury looks at operating cash flow, manufacturing costs, and optional expenses, variables that can impact actual performance is better than accrual accounting The author then uses this measure to detect performance manipulation around the zero return threshold The results of this study are consistent with Hayn (1995), Burgstahler and Dichev (1997), Dechow et al et al (2003), Beaver

et al (2004), Durtschi and Easton (2005)

Dechow et al (2009) considers 2190 companies listed in the period 1982 - 2005

on AAERs (Accounting Auditing Enforcement Releases) and builds a model that gets

an F-score with three levels with a threshold for determining the risk of financial fraud Three levels 1,2,3 has an accuracy of 65.9%, 65.78%, and 63.36%

Marinakis (2011) also reconstructed the Beneish M-score model for Great Britain with more relevant coefficients The author compares two models: the original

Beneish model and the enhanced British model The advanced model includes 11 variables, of which eight are similar to the Beneish model In addition, he added three other variables, including EFAXI - Effective tax rate index, DIRAI - remuneration index for directors to total assets, AUDI - Audit remuneration index to a total asset After performing the tests, the author finds that the enhanced model has a 10% higher rate of profit distortion for firms than the original Beneish model The author also defines the threshold value for his model as -1.31

2.2.4 Research on the correlation between financial statement distortion and stock market profitability

Richard G Sloan (1996) studies the relationship between accruals and cash flows of current income reflected in stock prices Sloan (1996) uses data from financial statements and stock prices of 40,679 observed samples of companies by year on the NYSE and AMEX exchanges for 30 years (1962-1991); With three main variables are income, accrual accounting, and cash flow from operations The author builds the following hypothesis:

 H1: Current income persistence will decrease for the size of accruals and increase for the cash flow factor

 H2(i): The expected return expressed in stock prices fails to adequately reflect the higher earnings attributed to the cash flow factor and the lower the attributable accruals

 H2(ii): A long-term buying strategy of stocks of companies associated with low accruals and a short-selling strategy involving high accruals would be associated with positive abnormal returns

 H2(iii): The extraordinary returns anticipated in H2(ii) compounded at the future earnings announcement date

After testing the hypothesis, the results show that firms with higher accruals generate lower returns At the same time, the predictability of accruals comes from the fact that cash-flow-based earnings are more persistent than accrual-based earnings

Bergstresser, Desai, and Rauh (2006) for managers to use financial statement distortion tools to increase their stock prices by introducing retirement assets into the capital market and altering investment decisions to adjust and take advantage of it

Managers increase their assumed returns when they prepare to acquire other companies and exercise stock options The decision on the assumed rate of return affects the asset allocation in the pension plan The variable results show a 0.25% increase in the corresponding hypothetical return with a 5% increase in stock allocation and guide that financial reporting manipulation influences managers' investment decisions

David Hirshleifer et al (2009) studied the extended effects of accruals and cash flows This study finds that overall accruals will positively predict future returns while cash flow is a predictor of future returns In addition, improving accruals is also negatively correlated with overall profitability, while improving cash flow is positively correlated These findings indicate that improvements in accruals and cash flows imply changes in the discount rate or that the company manages earnings to under-report to the market school

Beneish, Lee, and Nichols (2013) show that firms with higher earnings distortions (high M-score) earn lower returns in each category sorted by size, book value-to-market price, momentum, and accruals These returns are statistically significant

When conducting research about accruals and M-score to predict profitability, the authors found that the dominance of M-score over accruals is proven, even when controls are arranged independently or in association with each other

When these two variables are independent, the M-score is particularly effective

in predicting profitability for firms with low accrual accounting For the lowest accrual accounting quintile, the difference between size-adjusted returns between firms with high M-scores and low M-scores is -19.8% for the following year For companies in the second-lowest quintile, the difference is -10.5% a year In general, after controlling

for M-score, we find that accrual accounts exhibit limited predictive power that focuses mainly on the middle quintile of the M-score

Research by Vo Minh Duong (2016) has also applied the M-score index to classify two groups of fraudulent or non-fraudulent companies Companies with data distortion warnings will be segregated to review stock returns at 30 days, 90 days, 180 days, and 360 days from the publication financial statements The author finds a significant difference in the rate of return at 12 months from the publication of financial statements between the two groups of companies Specifically, the companies without distortion have an average return of +5.89%, while the companies in the other group are 3.91%

Research by Ngo Thi Dieu Huong (2020) studies the relationship between the M-score and the profitability ratio of two groups of fraudulent and non-fraudulent companies Specific results: the proportion of companies at risk of financial fraud between the two exchanges is equal to 30%

CONCLUSION CHAPTER 2

Chapter 2 presents the define of financial statement and financial statements fraud Theory explaining the motives for fraudulent financial statements and the acts of performing fraudulent financial statements

In addition, the study also inherits from previous authors, especially Beneish's M-score model with 8 variables to detect fraud risk of enterprises Research on the M-score model has gradually become more popular in Vietnam, especially research in listed companies

CHAPTER 3: DATA AND RESEARCH METHODS

This thesis applies the Beneish M-score model based on the research paper

"Earning Manipulation and Expected Return" by Beneish, Lee, and Nichols (2013) And I inherit the method research from "Beneish M-score model and the impact on the profitability on the Vietnamese financial stock market" by author Ngo Thi Dieu Huong (2020) to experiment on the Vietnamese stock market in the next stage – during the Covid-19 pandemic

Examine the financial statement fraud of the companies for the period 2011 -

2021 Moreover, find the relationship between the predictability of the M-score and the stock return in the relationship correlation with other predictors

In chapter 3, the author gives the sequence of steps to be performed in carrying out this research with the application of Excel and the software STATA 15 The author's data source is taken from reputable information sources such as the website company, Fiinpro, Refinitive

3.1 Research data

To accomplish the research objectives of this topic, the author collects audited financial statements of the listed companies from 2011 to 2021 Sources of financial statements are taken from websites of listed companies and reputable websites that provide information on securities investment, such as the website company, Fiinpro, Refinitive

This data will continue to be filtered to exclude:

 Financial companies, including banks, securities companies, and insurance companies, as they are not subject to the study because the revenue recognition method is somewhat different as well as missing some critical variables of the model

 Companies lack financial reporting data

 Companies delisted during the study period or did not have sufficient stock price data

In addition, to evaluate the correlation between financial statement fraud and stock return on the stock market, the author collects the closing prices 30 days after the date of publication of the financial statements (i.e., April 30) each year to calculate the stock's rate of return for that financial year

Table 3.1 Statistics of research observations on HOSE and HNX 2011 - 2021

Research Process includes:

 Step 1: Measure the possibility of distorting financial statements

 Step 2: Allocate portfolio by capitalization

 Step 3: Calculate the return by portfolio on the stock market

 Step 4: Modeling factors affecting return on the stock market

3.2.1 M – Score model

This paper uses a newer version adapted from the Beneish (1999) model used in the “Earning Manipulation and Expected Return” study (Beneish, M.C Lee and Nichols, 2013) with data from 2011 to 2021 in the Vietnamese stock market to identify companies that distort earnings or not The author estimates the coefficient M - score

according to the following equation:

After calculating the Mscore, we will compare it with the cutoff value of 1.78 The value minimizes the cost of classifying type I and type II errors

If the calculated M-score is greater than -1.78, the author marks the company as

having income distortion Otherwise, marking no income distortion Thereby, the author statistics the number of companies with distorted financial statements

M – score = -4.84 + 0.0920(DSRI) + 0.528(GMI) + 0.404(AQI) + 0.892(SGI) +

0.115(DEPI) – 0.172(SGAI) + 4.679(TATA) – 0.327(LVGI)

3.2.2 M – score and forecast future expected return

3.2.2.1 Calculate adjusted rate of return on a security (BHSAR)

 r it is the one-year return (r) of stock i in year t calculated by the following formula:

Where:

 r it: the stock return in year t

 P it: the closing price of the stock in year i on April 30, year t

 P i(t-1): the closing price of the stock in year i on April 30, year (t-1)

 Stock portfolio ratio

Where:

 R Pt is the rate of return of the portfolio year t

 w it is the proportion of investment in stock i in the portfolio

Based on the market capitalization of securities, this study classified stocks into ten categories The division of shares of similar company size into different investment portfolios is to recalculate the adjusted rate of return according to the portfolio size, thereby eliminating the influence of differences in size capitalizes

In this study, the author uses market capitalization at the time 30 days after the publication of financial statements to measure company size Specifically, depending

on the number of companies each year, the author divides into ten portfolios in order of capitalization from large to small, with equal investment proportions for each stock

BHSARit = rit - RPt

rit = 𝐏𝐢𝐭 − 𝐏𝐢(𝐭−𝟏)

𝐏𝐢(𝐭−𝟏)

RPt = 𝑾𝐢𝐭𝒏𝟏 x rit

3.2.2.2 Comparison of BHSAR (+1) of two groups of companies classified by score

M-The author using the results of the M-score calculated for the observations to classify two groups of fraudulent and non-fraudulent companies, then calculate the average return of the two groups based on the results of the two groups BHSAR (symbol BHSAR - buy and hold size-adjusted return) earlier Then calculate the profit margin difference between the two companies in each year from 2011 to 2021

3.2.2.3 Consider the relationship between M – Score and profitability

Compare annual adjusted returns for each year of the 2011-2021 sample period

In each period, the author will also consider and separate the sample into two groups: fraudulent companies and non-fraudulent companies The primary purpose is to compare the differences in returns across groups to determine whether investing in financial distortion companies yields a higher rate of return?

3.2.2.4 Consider the relationship between profitability and other factors

According to Beneish's study (2013) and previous research papers, the group used 4 factors that were correlated with the following rate of return:

(1) Accruals

Accruals, or the difference between profit in the income statement and cash flow

in the statement of cash flows, has a negative relationship with the profit margin (Sloan, 1996) In other words, the predictive signature of accruals comes from the fact that earnings based on the cash flow component will be more sustaining than those

(2) Price momentum

The stocks with good price performance in the last 3 to 12 months tend to continue their growth momentum in the next 3 to 12 months (Jegadeesh and Titman, 1993) Similarly, stocks with high returns outperform stocks with low earnings momentum In this study, the Momentum variable is calculated from when the investor holds the stock within one year: from April 30 of the first year to April 30 of the following year

(3) MVE (Firm size is represented by market capitalization of equity)

(4) BTM (Book value-to-Price market)

3.2.2.5 Empirical model about the relationship between M-score and return on stock market

Inheriting the work of Beneish, Lee, Nichols (2013), we have the following model:

Where:

 BHSAR: Buy and hold size adjusted return

 M – Score: The index predicts the possibility of financial statement fraud

according to the Beneish model

 α 0: Block coefficient

 α 1 - 5: Regression coefficients of the independent variable

 e t+1: Random error

To find out the correlation relationship between the variables affecting the

return on BHSAR, the author conducts a Pearson correlation analysis between the

dependent and independent variables in the equation The values of the variables

BHSAR, M-score, Accruals, Momentum, and BTM were preserved according to the

initial calculation results However, because the value of the MVE variable is

substantial compared to the other variables, the author uses the base natural logarithm

of market capitalization Ln(LVE) instead to measure the firm size

3.2.2.6 Regression fit method – GMM

The study examines the defects of the regression model with 4,336 observations

in the ten years 2011-2021 At the same time, during the verification process, the study also proposes solutions and chooses an appropriate regression method to handle these defects The process of method testing and selection starts from the regression by the least-squares method and ends with selecting the GMM method

CONCLUSION CHAPTER 3

Chapter 3 presents a detailed description of the research methodology The specific research steps are as follows: The first step is to determine the data collection sample which is the financial statements for the period from 2011 to 2021, and at the same time, exclude enterprises with special reports and lack of data Next, Beneish M-Score (1999) model is used to classify the group of companies suspected of fraud

Then, the author calculates the rate of return of stocks using the BHSAR index according to the factors M - Score, Accruals, Momentum, MVE, BTM and runs the

model to evaluate the level of their influence The author's model chosen is the GMM model

CHAPTER 4: RESEARCH RESULTS AND DISCUSSION

In this chapter, the research author analyzes and evaluates the results obtained from the data set processing based on the research methods mentioned in Chapter 3 At the same time, the results are discussed and compared Compared with the practice of previous studies to comprehensively assess the impact of the independent variables on the dependent variable

4.1 Financial distortion in Vietnam

4.1.1 Financial distortion from the perspective of the exchange

Based on Beneish's (2013) research formula, the author collects the necessary accounting data and calculates the M - score coefficient to review, in general, the distortion of companies' financial statements in Vietnam Using a cut-off value of –1.78, the author marks firms with M-scores greater than –1.78 as potentially fraudulent (there is a possibility of fraud in financial statements in the future for a current year or the next several years), otherwise will be marked as no fraud potential The research results showed that the number of enterprises with the risk of fraudulent financial statements on the stock market accounted for 30.2% Specifically, 1261/4336 firm-year observations on the stock market are at risk of fraudulent financial statements, of which HOSE recorded 756/2599 and HNX 505/1737