Information Asymmetry with Emphasis on the Role of Financial and Managerial Criteria Based on Fuzzy Logic and Artificial Neural Networks
محورهای موضوعی : Financial Accounting
Mohammad Amir Golshani
1
( Department of Accounting, Kermanshah Branch, Islamic Azad University, Kermanshah, Iran)
Mehrdad Ghanbari
2
(Department of Accounting, Kermanshah Branch, Islamic Azad University, Kermanshah, Iran.)
Babak Jamshidi Navid
3
(Department of Accounting, Kermanshah Branch, Islamic Azad University ,Kermanshah ,Iran)
Forouzan Mohammadi Yarijani
4
( Department of Accounting, Kermanshah Branch, Islamic Azad University, Kermanshah, Iran)
کلید واژه: Information Asymmetry, Corporate Profit Forecasting, Corporate Governance, Capital Market, Capital Return,
چکیده مقاله :
This paper addresses the absence of a suitable criterion for measuring information asymmetry between managers forecasting earnings and analysts forecasting earnings through statistical methods. Besides, this paper aims to provide a model of information asymmetry, emphasizing the role of financial and managerial criteria. This is applied qualitative and quantitative research (mixed method). The library method is used to prepare and formulate theoretical bases. In addition, the field method is used for collecting data to measure and identify indices and modeling. Factor analysis was used to analyze the data, following identifying the dimensions and variables of financial and managerial criteria of information symmetry to eliminate extraneous factors and classify. The following five main dimensions were determined, including corporate profit forecast, corporate governance, capital market, capital return, and management characteristics of the company. Then, the modeling was done using fuzzy mathematics through triangular numbers, Mamdani implication, and center of gravity methods. The final results of the study of the company listed on the Tehran Stock Exchange show that the level of information symmetry in the range of zero to 100 equals 55.1, to predict the company's profit is 48.54; corporate governance is 56.95; the capital market is 1/59; capital return is 61.07, and managerial characteristics of the company are 67.84. Finally, we examined the factors affecting the information asymmetry obtained from fuzzy neural networks. The findings show a higher prediction accuracy of fuzzy neural network methods than other related prediction methods.
This paper addresses the absence of a suitable criterion for measuring information asymmetry between managers forecasting earnings and analysts forecasting earnings through statistical methods. Besides, this paper aims to provide a model of information asymmetry, emphasizing the role of financial and managerial criteria. This is applied qualitative and quantitative research (mixed method). The library method is used to prepare and formulate theoretical bases. In addition, the field method is used for collecting data to measure and identify indices and modeling. Factor analysis was used to analyze the data, following identifying the dimensions and variables of financial and managerial criteria of information symmetry to eliminate extraneous factors and classify. The following five main dimensions were determined, including corporate profit forecast, corporate governance, capital market, capital return, and management characteristics of the company. Then, the modeling was done using fuzzy mathematics through triangular numbers, Mamdani implication, and center of gravity methods. The final results of the study of the company listed on the Tehran Stock Exchange show that the level of information symmetry in the range of zero to 100 equals 55.1, to predict the company's profit is 48.54; corporate governance is 56.95; the capital market is 1/59; capital return is 61.07, and managerial characteristics of the company are 67.84. Finally, we examined the factors affecting the information asymmetry obtained from fuzzy neural networks. The findings show a higher prediction accuracy of fuzzy neural network methods than other related prediction methods.
[1] Mashayekh, S., Arbabi, Z., Rahimi Far, M., Study of Earnings Management Incentives, Journal of Ac-counting Research, 2013; 3(2):53-70. (in Persian) Doi. 10.22051/IJAR.2014.460
[2] Shari Anaqiz, S., Rahimian, N., Salehi Sedghiani, J., Khorasani, A., Investigate and apply the accuracy of the results obtained from the revised models based on the Iranian economic environment in detecting and exposing fraudulent financial reporting, Financial Management Perspectives Quarterly, 1817) 2017; 105-123. (in Persian). Doi, 10.52547/JFMP.11.36.9.
[3] Moradi, M., Designing a profit quality model in the Tehran Stock Exchange with emphasis on the role of accruals, Journal of Accounting and Auditing Research, 2015; (25), 76-99. (in Persian). Doi. 10.22034/IAAR.2015.103918.
[4] Barton, J., Hansen, T., Pownall, G., Which performance measures do investors around the world value the most—and why? The Accounting Review, 2010; 85, 753–789. Doi:10.2308/accr.2010.85.3.753
[5] Kordestani, G., Tatli, R., Identification the Efficient and Opportunistic Earnings Management Ap-proaches in the Earnings Quality Levels. Accounting and Auditing Review, 2014; 21(3), 293–312. (in Per-sian) Doi:10.22059/ACCTGREV.2016.57021
[6] Beneish, M. D., The Detection of Earnings Manipulation, Financial Analysts Journal, 1999; 55(5), 24-36. Doi:10.2469/faj.v55.n5.2296.
[7] Jones, J.J., Earnings Management during Import Relief Investigations, Journal of Accounting Research, 1991 ;29(2), 193-228. Doi: 10.2307/2491047
[8] Healy, P., The effect of bonus schemes on accounting decisions, Journal of Accounting and Econom-ics, 1985; 7, 85-107. Doi. 10.1016/0165-4101(85)90029-1
[9] Watts, R., Zimmerman, J., Positive Accounting Theory, Prentice-Hall, Englewood Cliffs, New Jersy. 1986; Doi: 10.1016/0361-3682(88)90037-2.
[10] Najafizadeh, B., Kayhan, M., Investigating of the Relationship between Earnings Management and Information Asymmetry in Environmental Uncertainty in Companies Listed in Tehran Stock Exchange (TSE), 4th 2016
[11] Piri, P., Ghorbani, M., Assess the relationship between the type of independent auditor's opinion and the quality of the profit, Accounting and Auditing Reviews, 2017; 24 (4): 483-502. (in Persian. NDMCONFT04_257
[12] Salehi, M., Farrokhi Pilehroud, L., Predicting earnings management using neural network and deci-sion tree, Quarterly Journal of Financial Accounting and Auditing Research, 2018; 10(37):1-24. (in Per-sian)
[13] Tarjoa, N. H., Application of Beneish M-Score Models and Data Mining to Detect Financial Fraud. Social and Behavioral Sciences, 2015; 211, 924 -930. Doi: 10.1016/j.sbspro.2015.11.122
[14] Li, T., & Zaiats, N., Information environment and earnings management of dual class firms around the world. Journal of Banking & Finance, 2017 74, 1-23. Doi: 10.1016/j.jbankfin.2016.09.009
[15] Ramírez Orellana, A., Martinez Romero, M. J., Mariño Garrido, T., Measuring fraud and earnings management by a case of study: Evidence from an international family business, European Journal of Family Business, 2017; 7(1-2), 41-53. Doi: 10.1016/j.ejfb.2017.10.001
[16] Ajina, A., Habib, A., Examining the relationship between Earning management and market liquidi-ty,Research in International Business and Finance, 2017; 42, 1164-1172. Doi: 10.1016/j.ribaf.2017.07.054
[17] Rahul, K., Seth, N., Dinesh Kumar, U., Spotting Earnings Manipulation: Using Machine Learning for Financial Fraud Detection. In: Bramer M., Petridis M. (Eds) Artificial Intelligence. SGAI. Lecture Notes in Computer Science, Springer, 2018;343-356. Doi: 10.1007/978-3-030-04191-5_29
[18] Lazzem, S., Jilani, F., The impact of leverage on accrual-based earnings management: The case of listed French firms, Research in International Business and Finance, 2018; 44, 350-358.
Doi: 10.1016/j.ribaf.2017.07.103
[19] Mirjalili, S., Mirjalili, S. M., Lewis, A., Let a biogeography-based optimizer train your Multi-Layer Perceptron. Information Sciences, 2014; 269: 188–209. Doi: 10.1016/j.ins.2014.01.038
Adv. Math. Fin. App., 2024, 9(3), P. 806-816 | |
| Advances in Mathematical Finance & Applications www.amfa.iau-arak.ac.ir Print ISSN: 2538-5569 Online ISSN: 2645-4610
|
Case Study
Information Asymmetry with Emphasis on the Role of Financial and Managerial Criteria Based on Fuzzy Logic and Artificial Neural Networks
| ||
Mohammad Amir Golshani, Mehrdad Ghanbari*, Babak Jamshidi Navid, Forouzan Mohammadi Yarijani
| ||
Department of Accounting, Kermanshah Branch, Islamic Azad University, Kermanshah, Iran
| ||
Article Info Article history: Received 2022-04-30 Accepted 2023-02-23
Keywords: Information Asymmetry Corporate Profit Forecasting Corporate Governance Capital Market Capital Return
|
| Abstract |
This paper addresses the absence of a suitable criterion for measuring information asymmetry between managers forecasting earnings and analysts forecasting earnings through statistical methods. Besides, this paper aims to provide a model of information asymmetry, emphasizing the role of financial and managerial criteria. This is applied qualitative and quantitative research (mixed method). The library method is used to prepare and formulate theoretical bases. In addition, the field method is used for collecting data to measure and identify indices and modeling. Factor analysis was used to analyze the data, following identifying the dimensions and variables of financial and managerial criteria of information symmetry to eliminate extraneous factors and classify. The following five main dimensions were determined, including corporate profit forecast, corporate governance, capital market, capital return, and management characteristics of the company. Then, the modeling was done using fuzzy mathematics through triangular numbers, Mamdani implication, and center of gravity methods. The final results of the study of the company listed on the Tehran Stock Exchange show that the level of information symmetry in the range of zero to 100 equals 55.1, to predict the company's profit is 48.54; corporate governance is 56.95; the capital market is 1/59; capital return is 61.07, and managerial characteristics of the company are 67.84. Finally, we examined the factors affecting the information asymmetry obtained from fuzzy neural networks. The findings show a higher prediction accuracy of fuzzy neural network methods than other related prediction methods. | ||
1 Introduction
This paper addresses the absence of a suitable criterion for measuring information asymmetry between managers forecasting earnings and analysts forecasting earnings through statistical methods. The purpose of explaining the factors affecting the information asymmetry of budgeted managers' profit with other profit forecasting methods is to identify the degree of information asymmetry between people inside the organization and people outside the organization. This paper is based on agency theory and the theory of information asymmetry. According to the information asymmetry theory, company managers have the advantage of confidential information from within the company. Investors pay special attention to the company's profit figure to examine the company's situation and make investment decisions. Therefore, profit management is one of the most controversial and attractive topics in accounting research [1]. As the number of companies listed on the stock exchange and the process of privatization and capital growth has increased, and the public release of financial statements has become mandatory, measuring the forecast of the probability of fraud in the financial statements in the published financial information in methods other than conventional auditing methods seems necessary to disclose and prevent fraudulent reporting methods [2]. Much research has been done on detecting fraud and its possible manipulation. Beneish [6] proposed a model with eight accounting variables to detect profit manipulation. His model showed that the likelihood of profit manipulation also increased with unusual increases in receivables, the decline in gross profit margins, the reduction in asset quality, sales growth, and increased accruals. The Beneish studies model has been based on a survey of companies from the United States; however, studies conducted in other countries showed that the Beneish model could not have the same performance in all societies and capital markets [3]. Applications of accounting figures work differently in different markets, so it is not possible to identify the nature of these figures just by looking at a specific market. As a result, in a world of multiple countries and changing institutions and structures, it is essential to understand the applications of accounting numbers in other countries as much as possible [4]. Subsequent research based on the Beneish model for detecting profit manipulation showed that the Beneish model does not have the same function in predicting manipulation and profit management, so localized variables or other variables should be used [5]. Although effective in different societies, any fraud detection model may not necessarily be very accurate and should be localized according to the economic situation of each country. In the primary model of Beneish, the financial condition of Iran is not taken into account. Therefore, it is less accurate than the adjusted Beneish model, which has been localized for the economic environment of Iran. Thus, the fraud detection models should be localized based on the economic structure of that country before implementing in any country. [6] predicted profit manipulation methods using financial ratios and accruals. He used three sources to select his model variables. The first source is variables related to the future of the company, for it is assumed that the possibility of profit manipulation is when the company is in a weak position in the future. The second source is the variables based on cash flow and liabilities based on the models of [7] and [8]. Finally, he used the contract hypothesis based on the positive theory of [9].
Information asymmetry occurs when one party to a contract or transaction is aware of more information, provided that it uses that information effectively when communicating with the other party. This awareness is more likely to create economic benefits for the information party In fact, these benefits are more about having more information. The phenomenon of information asymmetry in stock market markets arises from the fact that there is a conflict of interest between information users and information providers. In such a way that increasing information asymmetry reduces market efficiency and decreases it increases market efficiency. Decreasing market efficiency can both threaten the interests of investors and shareholders and can threaten the interests of companies and stakeholders. One of the most effective ways to combat this negative phenomenon is to provide additional information, but providing additional information is time consuming and costly, which in itself makes the provision of such information unwelcome.
2 Theoretical Foundations and Empirical Background
Najafizadeh and Kayhan [10] showed that earnings management and information asymmetry are directly related to each other. Also, with increasing environmental uncertainty, the relationship between earnings management and information asymmetry is weakened. Kurdistani and Tatli [5] concluded that in the Iranian economic environment, the initial model of Benish, compared to the adjusted Beneish model, does not have good power to identify levels of profit manipulation. Adjusted Beneish model and models developed with differential analysis and logit approach can identify manipulating and non-manipulating companies with a total accuracy of 72%, 75%, and 81%, respectively. Based on their evidence, accounting information is also helpful in predicting earnings manipulation.Shari Anaghiz et al [2] examined 100 companies listed on the stock exchange based on the two main adjusted Beneish models in the economic situation of Iran to determine which of these two models works better in detecting fraud. In this paper, fraudulent and non-fraudulent companies are divided into two groups: bankrupt and healthy. Their results showed an adjusted Beneish model with an overall accuracy of 66.2% and a total error of 33.8%. The original Beneish model, with an overall accuracy of 61% and an absolute error of 39%, can better show the fraud in companies' financial statements. Business owners turn to an independent auditor to oversee the agent, prevent opportunistic behaviors and information asymmetries and reduce agency costs. This audit role is significant for investigating financial statement manipulation and, consequently, for reporting earnings managed by managers. Piri and Ghorbani [11] examined the relationship between the independent auditor's comments and earnings quality metrics, indicating a significant relationship between the type of independent auditor's report and earnings quality indices. They also identified two sets of information in their decisions. The accrued earnings management is a potential factor to distort this feature. To evaluate the comparability feature, they used the output of the accounting system and examined the relationship between profits and returns of companies in the same industry. Their research findings show no significant relationship between the comparability of accounting figures and the management of actual and accrued earnings. Salehi and Farrokhi Pilehroud [12] showed that the neural network and decision tree method in predicting profit management is more accurate than linear methods with a lower level of error. In the absence of a model for detecting fraud in Iran, the Beneish model is commonly used in Iran for fraud detection. Tarjoa [13] examined the ability of the Beneish model to detect financial fraud. He showed that the general Beneish model could detect financial fraud. In this study, gross margin index, depreciation index, sales index, administrative cost index, and total accruals were significant in identifying financial fraud, and sales index, asset quality index, and financial leverage index were not statistically significant in detecting financial fraud. Li and Zeits [14] also focused on the corporate information environment, examining the extent of profit manipulation by companies operating in a poor information environment. They showed that there is a good platform for profit manipulation in companies with high information asymmetry and poor information environment. Managers of these companies have more motivation to manage profits. Orlana, Romero, and Grido [15] tested the Beneish model to assess the likelihood of fraud and profit management. In other words, they measured fraud through the Beneish model. According to their findings, fraud and daring accounting practices occur before financial problems are disclosed. Companies have taken bold accounting measures by manipulating the days of credit sales in the debt ratio and the sum of accruals to total assets. Ajina and Habib [17] found that companies manage profit intending to manage liquidity, so, based on experimental results, less liquid companies do more profit management.
3 Research Methods
This is an applied descriptive survey. The relevant literature was reviewed to prepare and formulate theoretical foundations. Also, the field method was used to collect information to measure and identify indices and evaluate information asymmetry. This research will be done in two phases. In the first phase, we examine and identify the components of information asymmetry by emphasizing the role of financial and managerial criteria in providing a model. Then, the modeling was done using fuzzy mathematics through triangular numbers, Mamdani implication, and center of gravity methods. In the second phase, the artificial neural networks were used to predict information asymmetry, emphasizing the role of financial and managerial criteria. These analyzes were performed using MATLAB and SPSS software. Below we give a complete description of the mentioned cases.
Fig. 1. Quantitative Research Model Design Process
4 Research Findings
We have researched and identified all the variables to finalize the model in the qualitative section, cover the various dimensions of the problem, and examine the main variables. In the qualitative part of the present study, after in-depth interviews with experts, the collected information was analyzed in three stages open coding, axial coding, and selective coding. Then, the conceptual research model was summarized. The interviewees answered questions under the framework of open-ended interview questions. The researchers classified the answers in each session, taking into account the challenges created by the researcher in each interview, his knowledge and the audit committee expertise, and the number of respondents per company. Thus, the answers were categorized based on their type and content under a research category. They were defined and determined as a criterion for evaluating that category. Finally, according to the content of the answers provided, each of the concepts' subcategories was placed in related categories. It is noteworthy that in the qualitative research process, based on a sampling of managers and experts in the field, a sample of 15 subjects was selected during in-depth interviews. In the continuation of the chapter, we will present the process and results of qualitative research to finally reach the final research framework.
Table 1: Axial Coding of Exploratory Study
No. | Concept | Corresponding codes | Repetition |
1 | Economic uncertainty | A1B1 A11B6 A4B10 A12B7 A4B10 A15B17 A14B10 A2B7 A2B14 | 9 |
2 | Management behavior in profit forecasting | A3B7 A1B13 A8B4 A4B2 A9B15 A10B6 A14B18 A10B17 | 8 |
3 | Sales returns | A1B17 A15B7 A8B1 | 3 |
4 | Number and expertise of the audit committee | A7B9 A8B7 A12B7 A8B2 | 4 |
5 | Institutional ownership | A13B17 A3B18 A13B9 A10B8 A1B17 | 5 |
6 | Presence of internal audit | A2B1 A13B1 A3B13 A14B16 | 4 |
7 | Quality of profit | A4B11 A4B14 A14B15 A15B18 A12B15 A3B4 A13B10 | 7 |
8 | Systematic risk | A1B14 A1B15 A2B15 A3B15 A9B4 A9B16 | 6 |
9 | Company competitiveness | A1B3 A12B2 A9B17
| 3 |
Table 1: Continue | |||
10 | Company growth opportunities | A10B2 A7B1 A7B16 A9B4 A9B17 | 5 |
11 | Return on assets | A3B16 A7B13 A6B18 A4B9 | 4 |
12 | Number and expertise of board members | A1B11 A8B12 A4B5 A9B6 A6B1 | 5 |
13 | Independence of the Audit Committee | A2B12 A4B4 A5B4 A7B10 | 4 |
14 | The dual role of the CEO | A1B4 A10B5 A11B1 | 3 |
15 | Change of CEO | A2B1 A13B2 A2B3 A8B15 | 4 |
The coding Table depicts the concepts per category and their frequency. Accordingly, 18 identified concepts were classified into 5 categories, shown in the Table below.
Table 2. Selective Coding Table of Qualitative-Exploratory Study
Frequency | Category | No. |
3 | Company profit forecast | 1 |
3 | Corporate governance | 2 |
5 | Capital Market | 3 |
2 | Stock returns | 4 |
5 | Company management characteristics | 5 |
Information asymmetry variables with emphasis on the role of financial and managerial metrics, include company profit forecast, corporate governance, capital market, stock return, and management characteristics of the company. The variables are measured by the Likert scale, so that the items represent the measurement of different aspects of each variable. The fuzzy mathematics method (method of triangular averages), was used and then these means were defuzzificated using the center of gravity method and entered in MATLAB software and in the model as inputs, and finally the model output was obtained as the amount of micro and asymmetry information with emphasis on the role of financial and managerial criteria of companies listed on the Tehran Stock Exchange.The general formula for calculating the average of triangular numbers is:
| (1) |
The mean score given to each person is shown according to the mean of the triangular numbers and also the calculated fuzzy mean of each index (or variable) is at the bottom of the Table.
For example, the average of a triangle of person 3 is as follows:
| (2)
|
Now that the fuzzy mean of each of the indices has been determined, it is time to enter the fuzzy means into the domestic model designed in the software and obtain the output, which is the information asymmetry amount.
After giving these inputs to the system, the presented result of the information asymmetry amount with emphasis on the role of financial and managerial criteria is 55.1. In other words, the score of information asymmetry of companies listed on the Tehran Stock Exchange in the range of [100 0] is 55.1, in other words, the rate of applying information asymmetry in companies listed on the Tehran Stock Exchange is 55.1%. The following figures present the output of the fuzzy model. Membership Degree is the amount of information asymmetry at all three levels:
| (3) |
That is, information asymmetry in companies listed on the Tehran Stock Exchange is 89.8% at the medium level and 10.2% at the high level.
Fig 2. Degree of membership of information asymmetry with emphasis on the role of financial and managerial criteria
Also, using the mentioned formula for each of the dimensions, we de-fuzzificate the results, that will be:
First index
Second index
Third index
Fourth index
Fifth index
| (4) |
Table 3: Statistical information of MSE, RMSE and coefficient of determination
Prediction method | MSE | RMSE | Adjusted R-Squared |
Multiple linear regression method with data panel method | 7.4 | 2.72 | 0.7646 |
Artificial network method | 1.206 | 1.098 | 0.84 |
Fuzzy neural network method | 0.136964 | 0.37 | 0.894 |
The low error of fuzzy neural network prediction method in comparison with two methods of multiple linear regression - data panel and artificial neural network method is confirmed. Also, the multiple linear regression method has the highest mean and the square mean of the errors.
Table 4: Mean, SD, and variance of earnings error per share by fuzzy neural network method
Statistical information | Mean | SD | Variance |
Earnings error per managed budgeted share | 66.74 | 319.202 | 101890 |
Profit error per share predicted by neural network method | 78.672 | 402.5 | 107800 |
Given the assumption of inequality of variance between the two statistical populations, the value of t-test statistics is:
| (5) |
© 2024. All rights reserved. |
