Corresponding author: Akbar Pourreza Soltan Ahmadi

Transcription

1 Technical Journal of Engineering and Applied Sciences Available online at TJEAS Journal / ISSN TJEAS The Comparative Study of Explanatory Power of Bankruptcy Prediction s In Accordance With Article 141 of Commercial Law of Companies Listed In Tehran Stock Exchange Akbar Pourreza Soltan Ahmadi 1*, Saeid jabbarzadeh kangar luei 2, Mohammad Baradar Salimi 3 1. Department of Accounting, Salmas Branch, Islamic Azad University, Salmas, Iran 2. Department of Accounting, Urmia Branch, Islamic Azad University, Urmia, Iran 3. Department of Accounting, Salmas Branch, Islamic Azad University, Salmas, Iran Corresponding author: Akbar Pourreza Soltan Ahmadi ABSTRACT: The purpose of this study was to find the answer which a bankruptcy prediction models is more consistent with Article 141 of Iran's commercial law? In this regard, was selected a sample of firms covered by Article 141 of commercial law from Also according to the aim of this research, we can be said that the research method is causal comparison. Too, the research hypotheses were tested using Z woung statistical model and t and f statistics And finally, the findings indicate that, firstly, there are significant differences between explanatory powers of these models (Altman, Shirata, Zemigewski, Springate, C-score, and Fulmer) to predict financial distress and secondly, explanatory power of mentioned models, is significant in accordance with Article 141 and Altman model is more consistent than other models with Article 141 Iran's commercial law. Key Words: Article 141 Iran's commercial law, bankruptcy prediction models, Altman model INTRODUCTION The purpose of development of profit seeking enterprises is earn profit and increase shareholder wealth but such factors as financial and economic crisis, decline in demand and consequently decline in production will cause these enterprises, can not provide various financial resources such as company's cash required. Onder also believes that both public and private companies need cash to sustain their activities. When company unable to provide sufficient cash, management will be forced to borrow or convert assets into cash but when management is not able to pay claims companies will be forced to declare bankruptcy (Onder, 2010). Gitman believes that the first and most important reason of organizations fail is their mismanagement. Despite the bankruptcy causes differ from company to another company, management incompetence, high cost of production, weak financial activity, and inactive board of directors are joint causes of the bankruptcy among firms (Saidi and Agaiee, 2009). In the nineties, researchers examined the reasons for business and financial failure. They used financial ratios and offered models for predicting bankruptcy of companies see (Altman (1968) - Springate (1978) - Ohlsan (1980) - Fulmer (1984) - Zemigewski (1984) - Ca-score (1987) - Shirata (1998) - Farajzadeh dehkordi (2007)). In Iran, according to Article 141 of Commercial Law (CL), if due to losses at least half of capital wasted, board is obligated to immediately invite extraordinary general meeting of shareholders to decide about liquidation or survival of the company. If the company is declared bankrupt in accordance with Article 141, Six months from this date will be given the opportunity to firm to solve the problem of accumulated losses and to act according to the law. After this time, if the company is declared bankrupt and the company does not reduce losses, the company's logo is stopped and for re-enter company's logo to stock market should company re-do the all legal process. If after a specified period of time firm does not work in this area, at this time the company is suspended and finally, firm is removed from the stock panel. So for admission and re-enter the Tehran Stock Exchange, firm should perform the whole process from the beginning (Saidi and aghaiee, 2009). Now here is asked this question that are usable the different models used in predicting bankruptcy? It can be clearly suggests that bankruptcy prediction models which are more consistent with Article 141 of the commercial law of Iran.

2 Literature Research Importance of bankruptcy prediction Bankruptcy of firms in the economy of any country can be irreversible costs to those countries. Branch is divided bankruptcy costs into four categories: 1 - Costs that directly affects the bankrupt company. 2 - Costs that directly affect the company's creditors. 3 - Losses that the bankrupt company to impose other companies and 4 - Costs that other sectors except bankrupt company or creditors incurred from bankrupt company (Branch, 2002). In this context, policy makers can evaluate and monitor the companies with enacted business laws (Sarbense-Oxely, Basel, and Solvency). On the other hand, many researchers (Altman, Ohlsan, shumway and...) tried to provide indexes for predicting corporate bankruptcy. They were evaluated the financial condition of the bankrupt companies and provide models for predicting financial distress (Nobuyuki, Noriyuki, 2011). bankruptcy prediction models Altman model Z=β0+β1A+β2B+β3C+β4D+β5E, that is; A: working capital to total assets ratio, B: ratio of retained earnings to total assets, C: ratio of earnings before interest and taxes to total assets, D: ratio of market value of stockholders equity to book value of total liabilities, E: ratio of sales to total assets. Shirata model Z=β0+β1A+β2B+β3C+β4D, that is; A: ratio of retained earnings to total assets, B: ratio of debt and stockholders equity in this year to debt and stockholders equity in the previous year, C: ratio of interest expense to the sum of average of loans and debts and bonds and notes receivable discounted, D: ratio of average the sum of accounts payable and notes payable multiplied by 12 divided to sales. Zemigewski model Z=β0+β1A+β2B+β3C, that is; A: ratio of net income to total assets, B: ratio of total debt to total assets, C: ratio of current assets to current liabilities. Springate model Z=β0+β1A+β2B+β3C+β4D, that is; A: ratio of working capital to total assets, B: ratio of net income before interest and taxes to total assets, C: ratio of net income before taxes to current liabilities, D: ratio of sales to total assets. Ca-score model Z=β0+β1A+β2B+β3C, that is; A: ratio of stockholders equity (1) to total assets (1), B: financial expense plus a profit before tax and extraordinary items (1) to total assets (1),: C ratio of sales (2) to total assets (2), number (1) is remaining digits of a prior period and number (2) is remaining digits of the two prior period. Fulmer model Z=β0+β1A+β2B+β3C+β4D+β5E+β6F+β7G+ β8h+β9i, that is; A: ratio of retained earnings to total assets, B: ratio of sales to total assets, C: ratio of profit before tax to stockholders equity, D: ratio of cash flow to total debt, E: ratio of debt to total assets; F: ratio of current liabilities to total assets, G: logarithm of total tangible assets, H: ratio of working capital to total debt, I: ratio of the logarithm of earnings before interest and tax to interest (Godrati and Manavi Moghaddam, 2010). Background Of Research William Beaver was the first scientist who attempted to provide a model for the prediction of bankruptcy of companies. After Beaver scientists such as Altman, Springate, Zemigewski, Fulmer, Shirata, Ohlsan, Farajzadeh, McKee performed lot of research in this area and have attempted to provide corporate financial distress prediction models. Wang and Campbell in 2010 studied the z-score bankruptcy prediction model, and profit before tax (PBT) in China. They concluded that both models have the ability to predict bankruptcy (Campbell and Wang, 2010). Chung et al in 2010 investigated with using Multivariate Discriminant Analysis (MDA) and Artificial Neural Network (ANN) models to financial distress in New Zealand. They found that the financial ratio of distressed companies have major differences with other companies, distressed companies have less profitability and cash ratio and models they used to have high accuracy in predicting financial distress (Chung et al, 2010). Jabbarzadeh Kangarluei and et al examined the relationship between income smoothing and financial distress companies of Tehran Stock Exchange. They were evaluated income smoothing at three levels of gross profit, operating profit and net profit and financial distress in the three levels of latency, cash deficit and insolvency. The findings showed that managers of firms in different stages of financial distress, Attempting to income smoothing to better illustrate their financial situation and to maintain them in capital market 2477

3 (Jabbarzadeh Kangarluei et al, 2009). Dijkstra and Faure in a legal and economic analysis examined the compensating victims of bankrupted financial institutions. They found that the current compensating victim system in insolvent companies in Netherlands will not provide sufficient incentives for all stakeholders to prevent the failure of a financial institution (Dijkstra and Faure, 2011). In another research Ren evaluated reorganization law in Chinese bankrupt companies and found that Chinese legislators should revise in this law to the unfair distribution of wealth (Ren, 2011). ZHANG, ALTMAN, YEN believing that Z-score model, alone can not be used in China; have attempted to provide a model derived from this model. Their finding suggests that this new model with 80% accuracy according to the China trade law, are able to predict financially distressed firms (ZHANG et al., 2010). Gang and Xiaomao with using Modified Altman Z-score model in listed companies in China's stock market, examined the explanatory power of the model in predicting financial distress. They kept the Altman model three variables (total assets turnover ratio - the ratio of retained earnings to total assets, yield of total assets) and some of variables of cash flows (ability to pay debts - Throughput - profitability and growth ability) were added to the model. They found that against Altman Z-score model, the modified model can better predict financial distress (Gang and Xiaomao, 2009). Research Hypotheses Main Hypothesis 1: There are significant differences between the explanatory powers of financial distress models Altman, Shirata, Ohlsan, Zemigewski, Springate, Ca-score, and Fulmer. Main Hypothesis 2: There are significant differences between the explanatory powers of financial distress models Altman, Shirata, Ohlsan, Zemigewski, Springate, Ca-score, and Fulmer in accordance with Article 141 of the Commercial law of Iran. Population And Sample Populations in this study are all financially distressed companies in accordance with Article 141 of the commercial law in Tehran Stock Exchange for the period , due to the low number, the sample size will be equal to the population size. Research Variables In this study, independent variable in each of bankruptcy prediction models is financial ratios and z scale is the dependent variable that is characterized according to each bankruptcy model. RESEARCH METHODS This study in this sense that can be used numerous groups of users of accounting information is application. METHOD OF DATA ANALYSIS In this study, to evaluate the model due to lack of establish the basic assumptions of regression (variance stabilization, normalization and correlation of residuals) initially with a power transformation ( ) try to normalize the remaining and for assess normality of residuals is used Quantile - Quantile Plot and kolmogorove- smirnov test and also used the statistic Durbin - Watson correlation for the detection of residual. On the other hand, to test the hypothesis i.e. (compares the explanatory power of financial distress prediction models with Article 141 of the Commercial law of Iran) used to test Z woung and t and f statistics to test for significant coefficients and financial distress prediction regression model. Altman model Altman model is as follows: Z: criterion to be covered by Article A: working capital to total assets ratio. B: ratio of retained earnings to total assets. C: ratio of earnings before interest and taxes to total assets. D: ratio of market value of stockholders equity to book value of total liabilities. E: ratio of sales to total assets. : Regression residuals. To evaluate the model due to lack of establish the basic assumptions of regression (variance stabilization, normalization and correlation of residuals) initially with a power transformation ( remaining. The following table reports the results of testing hypotheses. ) try to normalize the 2478

4 Table 1. Table of regression of Altman model fitting Adjusted Mean Square Degrees of Freedom Sum of Squares Standard Deviation Estimation Working capital to total assets ratio Ratio of retained earnings to total assets Ratio of operating income to total assets Ratio of shareholders Equities to liabilities Ratio of sales to total assets about 79% of the dependent variable in companies explained by independent variables and can accept According the above model and the probability (P-value = ) and compared with a significance level (0.05 = α) we can accept that model is statistically significant. Shirata model Shirata model is as follows: Z: criterion to be covered by Article A: ratio of retained earnings to total assets, B: ratio of debt and stockholders equity in this year to debt and stockholders equity in the previous year, C: ratio of interest expense to the sum of average of loans and debts and bonds and notes receivable discounted, D: ratio of average the sum of accounts payable and notes payable multiplied by 12 divided to sales, : Regression residuals. To evaluate the model due to lack of establish the basic assumptions of regression (Variance stabilization, normalization and correlation of residuals) initially with a power transformation ( remaining. The following table reports the results of testing hypotheses. ) try to normalize the Table 2. Table of regression of Shirata model fitting Mean Square Standard Deviation Adjusted Degrees of Freedom Estimation Sum of Squares Ratio of retained earnings to total assets Ratio of debt and stockholders equity in this year to debt and stockholders equity in the previous year Ratio of interest expense to the sum of average of loans and debts and bonds and notes receivable discounted Ratio of average the sum of accounts payable and notes payable multiplied by 12 divided to sales 2479

5 about 24% of the dependent variable in companies explained by independent variables and can accept According the above model and the probability (P-value = ) and compared with a significance level (0.05 = α) we can accept that model is statistically significant. Zemigewski model Zemigewski model is as follows: Z: criterion to be covered by Article A: ratio of net income to total assets, B: ratio of total debt to total assets, C: ratio of current assets to current liabilities, : Regression residuals. To evaluate the model due to lack of establish the basic assumptions of regression (Variance stabilization, normalization and correlation of residuals) initially with a power transformation ( remaining. The following table reports the results of testing hypotheses. ) try to normalize the Table 3. Table of regression of Zemigewski model fitting Mean Square Degrees of Freedom Standard Deviation Estimation Adjusted Sum of Squares Ratio of net income to total assets Debt ratio Current ratio about 38% of the dependent variable in companies explained by independent variables and can accept According the above model and the probability (P-value = ) and compared with a significance level (0.05 = α) we can accept that model is statistically significant. Springate model Springate model is as follows: Z: criterion to be covered by Article A: ratio of working capital to total assets, B: ratio of net income before interest and taxes to total assets, C: ratio of net income before taxes to current liabilities, D: ratio of sales to total assets : Regression residuals. To evaluate the model due to lack of establish the basic assumptions of regression (Variance stabilization, normalization and correlation of residuals) initially with a power transformation ( remaining. The following table reports the results of testing hypotheses. ) try to normalize the 2480

6 Table 4. Table of regression of Springate model fitting Adjusted Mean Square Degrees of Freedom Sum of Squares Standard Deviation Estimation Ratio of working capital to total assets Ratio of net income before interest and taxes to total assets Ratio of net income before taxes to current liabilities Ratio of sales to total assets about 47% of the dependent variable in companies explained by independent variables and can accept According the above model and the probability (P-value = ) and compared with a significance level (0.05 = α) we can accept that model is statistically significant. Ca-score model Ca-score model is as follows: Z: criterion to be covered by Article A: ratio of stockholders equity to total assets, B: financial expense plus a profit before tax and extraordinary items to total assets, C: ratio of sales to total assets, : Regression residuals. To evaluate the model due to lack of establish the basic assumptions of regression (Variance stabilization, normalization and correlation of residuals) initially with a power transformation ( remaining. The following table reports the results of testing hypotheses. ) try to normalize the Table 5. Table of regression of Ca-score model fitting Adjusted Mean Square Degrees of Freedom Sum of Squares Standard Deviation Estimation Ratio of sales to total assets Ratio of stockholders equity to total assets Financial expense plus a profit before tax and Extraordinary Items to total assets about 60% of the dependent variable in companies explained by independent variables and can accept According the above model and the probability (P-value = ) and compared with a significance level (0.05 = α) we can accept that model is statistically significant. 2481

7 Fulmer model Fulmer model is as follows: Z: criterion to be covered by Article A: ratio of retained earnings to total assets, B: ratio of sales to total assets, C: ratio of profit before tax to stockholders equity, D: ratio of cash flow to total debt, E: ratio of debt to total assets; F: ratio of current liabilities to total assets, G: logarithm of total tangible assets, H: ratio of working capital to total debt, I: ratio of the logarithm of earnings before interest and tax to interest : Regression residuals. To evaluate the model due to lack of establish the basic assumptions of regression (Variance stabilization, normalization and correlation of residuals) initially with a power transformation ( normalize the remaining. The following table reports the results of testing hypotheses. ) try to Table 6. Table of regression of Fulmer model fitting Adjusted Mean Square Degrees of Freedom Sum of Squares Standard Deviation Estimation Ratio of retained earnings to total assets Rratio of sales to total assets Ratio of debt to total assets Ratio of cash flow to total debt Ratio of working capital to total debt Ratio of current liabilities to total assets Ratio of profit before tax to -2 stockholders equity Logarithm of total tangible assets Ratio of the logarithm of earnings before interest and tax to interest about 53% of the dependent variable in companies explained by independent variables and can accept According the above model and the probability (P-value = ) and compared with a significance level (0.05 = α) we can accept that model is statistically significant. Research Hypotheses Deduced The main hypothesis of this study is as follows: There are significant differences between the explanatory powers of financial distress models Altman, Shirata, Ohlsan, Zemigewski, Springate, Ca-score, and Fulmer. Comparison of absolute magnitude errors of estimated To compare the explanatory power of the prediction models should be compared to the estimated distress rate with derived distress of according to law 141.For do this survey, can be used residual absolute magnitude of models (distance of estimated values to actual values). To evaluate the remaining equity in models due to the existence of several residual for each firm in each year for different models (repeated measures) and too normality of residuals, can used analysis of variance for repeated measures And the following statistical hypotheses can be raised. 2482

8 That is Average of residual absolute magnitude of i model. This model can be rewritten as follows: H0: There are not significant differences between the explanatory powers of financial distress models Altman, Shirata, Ohlsan, Zemigewski, Springate, Ca-score, and Fulmer H1: There are significant differences between the explanatory powers of financial distress models Altman, Shirata, Ohlsan, Zemigewski, Springate, Ca-score, and Fulmer. Statistical tests results are in the next table: Table 7. Comparison of average of absolute magnitude errors of estimated Standard Deviation Average of errors of estimated Altman Shirata Zemigewski Springate Ca-score Fulmer As seen in the above table average of deviations absolute magnitude is not the same and Shirata model is the highest and the lowest is the Altman model (compare the coefficients determination of the models shows the same results). Significance of these differences is presented in the following table: Table 8. Analysis of variance with repeated measures Mean Square Degrees of Freedom Sum of Squares Factors According to the table ANOVA for repeated measures and compare the amount of probability with significance level of test, we can see in significantly at 5% level, average of absolute error is different and therefore, the null hypothesis or this assumption that " there are not significant differences between the explanatory powers of financial distress models Altman, Shirata, Ohlsan, Zemigewski, Springate, Ca-score, and Fulmer" is rejected. Comparison of coefficient of determination For comparison coefficient of determination individual coefficient of determination values are compared using the Z woung test. Accordingly, we have: is coefficient of determination of i model and in other words: H0: Between explanatory powers of i model and predict financial distress j, there is no significant difference. H1: Between explanatory powers of i model and predict financial distress j, there is significant difference. Z woung statistic values are listed below. Fulmer P Z Table 9. Table Z woung test to compare coefficient of determination values Ca-score Springate Zemigewski Shirata Altman P Z P Z P Z P Z P Z coefficient of determination Altman Shirata Zemigewski Springate Ca-score Fulmer According to the probability values and compare them with the significant level of test we have: Altman model respect to Shirata, Ohlsan, Zemigewski, Springate, Ca-score, and Fulmer in 5% significant level have significant difference and is better than the other models. Ca-score model respect to Springate and Fulmer in 5% significant level have not significant difference but have less predictive power rather than Altman model and have more than respect to Shirata and Zemigewski model. 2483

9 Fulmer model respect to Springate and Ca-score in 5% significant level have not significant difference but have less predictive power rather than Altman model and have more than respect to Shirata and Zemigewski model. Springate model respect to Zemigewski, Fulmer and Ca-score in 5% significant level have not significant difference but have less predictive power rather than Altman model and have more than respect to Shirata model. Zemigewski model respect to Springate and Shirata in 5% significant level have not significant difference but have less predictive power rather than Altman, Ca-score and Fulmer model. Shirata model respect to Zemigewski model in 5% significant level has not significant difference but have less predictive power rather than Altman, Ca-score, Springate and Fulmer model. RESULTS OF FIRST HYPOTHESES The first research hypothesis is as follows: There are significant differences between the explanatory powers of financial distress models Altman, Shirata, Ohlsan, Zemigewski, Springate, Ca-score, and Fulmer. Research findings have confirmed this assumption and each of models have predicting corporate bankruptcy based on the following table. Table 10. coefficient of determination of bankruptcy models coefficient of determination model Altman Shirata Zemigewski Springate Ca-score Fulmer RESULTS OF SECOND HYPOTHESES The second research hypothesis is as follows: There are significant differences between the explanatory powers of financial distress models Altman, Shirata, Ohlsan, Zemigewski, Springate, Ca-score, and Fulmer in accordance with Article 141 of the Commercial law of Iran. Based on research findings this assumption was confirmed and models were compared based on Z woung test. CONCLUSION The main objective of this study is representing explanatory power of predicting bankruptcy models in conformity with Article 141 of the Commercial law of Iran. In this regard, companies covered by Article 141 of trade law in years of were selected as sample. First, the findings suggest that there are significant differences in explanatory power of different models to predict bankruptcy and secondly, in accordance with Article 141 of trade law is also significant and Altman model have most consistent with coefficient of determination of and then Ca-score model with coefficient of determination of and Fulmer model with coefficient of determination of and Springate model with coefficient of determination of and Zemigewski model with coefficient of determination of and finally Shirata model with coefficient of determination of In general, it can be inferred that Altman model has more consistent with Article 141 of Commercial law of Iran. Suggestions Resulting From Findings ` Given that investors have reasonable knowledge about financial statement information and Altman model derived from these financial ratios therefore to all users, especially major investment of firms is suggested that to analyze the status of investment companies considering to working capital to total assets ratio, ratio of retained earnings to total assets, ratio of earnings before interest and taxes to total assets, ratio of market value of stockholders equity to book value of total liabilities, ratio of sales to total assets. To the Tehran Stock Exchange's managers, corporate executives, professionals, law enforcement and... suggested that use of Altman model to analyze the situation of companies. 2484

10 ACKNOWLEDGMENT This article is extracted from a research project entitled "The Comparative Study of Explanatory Power of Bankruptcy Prediction s in accordance with Article 141 of Commercial Law of Companies listed in Tehran Stock Exchange" which financing from research budget of Islamic Azad University of Salmas Branch. REFERENCES Branch B The costs of bankruptcy a review, Int Rev Finan Anal 11: Campbell M, Wang Y Do Bankruptcy s Really Have Predictive Ability? Evidence using China Publicly Listed Companies. Int Manag Rev 6( 2): Chung KC, Tan SS, Holdsworth DK Insolvency Prediction Using Multivariate Discriminant Analysis and Artificial Neural Network for the Finance Industry in New Zealand. Int J Bus Manag 3(1): Dijkstra RJ, Faure MG Compensating victims of bankrupted financial institutions: a law and economic analysis, J.F.R & C 19( 2): Gang HU, Xiaomao Z Study on Improving Z-score Based on the Logistic : Evidence from Listed Companies in China, International Conference on Information Management, Innovation Management and Industrial Engineering, Godrati H, Manavi MAH Accuracy of bankruptcy prediction models (models Altman Shirata, Ohlsan, Zemigewski, Springate, Cascore, Fulmer, Farajzadeh genetic, McKee genetic) on the Stock Exchange Tehran, Acc Res, 7:128. Jabbarzadeh KS, Khodayar YS, Pourreza SAA The relationship between income smoothing and financial distress companies in Tehran Stock Exchange, J Finan Acc, 1(2): Nobuyuki H, Noriyuki K Bankruptcy dynamics in Japan, Japan and the World Economy, JAPWOR. 735:10. Onder C Bankruptcy Prediction with Support Vector Machines: An Application for German Companies, C.A.S.E Ren Y.2011.Wealth Distribution in Chinese Bankruptcy Reorganization Law and Practice, Int Ins Rev 20: Saidi A, Agaie A Prediction of financial distress of listed companies in Tehran Stock Exchange Using Bayes networks, J Acc & Audit 16(56): ZHANG L, ALTMAN EI, YEN J.2010.corporate financial distress diagnosis model and application in credit rating for listing firms in China, Front. Compute. Sci 4(2):