Comparison of accountin based bankruptcy prediction models of altman(1968), ohlson(1980) nad zmijewski(1984) using vietnamese listed companies during 2008 2014

UNIVERSITY OF ECONOMICS HO CHIMINHCITY VIETNAMTHE NETHERLANDS INSTITUTE OF SOCIAL STUDIES THE HAGUE VIETNAM – NETHERLANDSPROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS Comparison of accounting-based bankruptcy prediction models of Altman(1968), Ohlson(1980) and Zmijewski(1984)using Vietnamese listed companies during 2008 – 2014 A thesis submitted in partial fulfilment of the requirements for the degree of MASTER OF ARTS IN DEVELOPMENT ECONOMICS ByNGUYEN THI PHUONG TRAM Academic Supervisor: DR.NGUYEN THI THUY LINH HO CHI MINH CITY, December 2015 DECLARATION “I certify that the substance of this thesis has not already been submitted for any degree and have not been currently submitted for any other degree I certify that to the best of my knowledge and help received in preparing this thesis and all sources used have been acknowledged in this thesis.” HCMC, December 2015 Signature NGUYEN THI PHUONG TRAM ACKNOWLEDGEMENTS Actually, this thesis could not have been finished without a collaborative experience involving the helps from many people I would like to acknowledge and express my heartfelt gratitude to those who give me the tremendous support to complete this thesis I greatly express my special thanks to my supervisor Dr Nguyen Thi Thuy Linhwho was abundantly helpful and offered invaluable assistance, support and guidance through finishing this process I express my warm thanks to Mr Pham Khanh Nam for all his academic recommendations I am using this opportunity to express my gratitude to everyone who supported me throughout the course of this MBA project Thank you so much! ABSTRACT Bankrupt prediction is practical research topic and attractsmore attention of debtors, creditors, shareholders and other stakeholders Moreover, it can be applied in a wide variety of situations to business managers make crucial decision The bankruptcy prediction models have been developed based on old original modes ofAltman(1968), Ohlson(1980) and Zmijewski(1984)usingmultivariate discriminate, logit and probit analysis respectively This thesisconducted the comparison of predictive power among these modelsusingVietnameselisted firms during the period from 2008 to 2014 The data is taken from listed companies in HOSE and HNX stock market and characterized the determinants of financial distress in terms of firm accounting andfinancial ratios To compare the predictive power of these models one year, two years and three year beforehand bankruptcy, the analysis of differences of accuracy rates and Receiver Operating Curves is applied Overall, Ohlson’sprobit model (1980) performed most accurate on the Vietnam listed companies within investigation periods This implies that Ohlson’smodel is the best predictor for bankruptcy likelihood for the listed companied in Vietnam However, since the unstable accuracy rates and the highest frequency of Type I and Type II errors, such resultshould be set into perspective and studied cautiously Keywords: bankruptcy prediction;multivariate discriminate analysis; logit model; probit model ABBREVIATIONS MDA: Multivariate discriminate analysis OLS Ordinary Least Squares A: Altman O Ohlson Z Zmijewski ROCReceiver Operating Curves AUROC CRV Area under the ROC curve Credit Rating of Vietnam GDPGross Domestic Product GNP Gross National Product HOSE Ho Chi Minh City Stock Exchange HXN Hanoi City Stock Exchange TABLE OF CONTENTS CHAPTER 1: INTRODUCTION 1.1 The empirical problem 1.2 Problem Statement 10 1.3 Research objective and methodology 11 1.4 Contribution 12 1.5 Outline 13 CHAPTER 2: LITERATURE REVIEW 14 2.1 Concepts 14 2.2 Bankruptcy prediction models 16 2.2.1 Z-score model-Altman (1968) 16 2.2.2 O-score model- Ohlson (1980) 18 2.2.3 Zm-score- Zmijewski (1984) 19 2.3 Empirical studies of comparing bankruptcy prediction models 20 CHAPTER 3: METHODOLOGY 24 3.1 Analytical framework 24 3.2 Research Models 26 3.2.1 Multiple discriminant analysis- Altman (1968) 27 3.2.2 Logit regression - Ohlson (1980) 29 3.2.3 Probit regression - Zmijewski (1984) 31 3.3 Research Methodology 32 3.3.1 Pseudo R2 and eigenvalue 33 3.3.2 Accuracy rate 34 3.3.3 Receiver Operating Curves 36 3.4 Derivation of Hypotheses 37 3.5Sample Selection and data 39 CHAPTER 4: EMPIRICAL RESULTS 42 4.1 Descriptive statistics 42 4.1.1 Descriptive statistic 42 4.2 Analysis of the Bankruptcy Prediction Models 46 4.2.1 Analysis of Altman´s model (1968) 46 4.2.2 Analysis of Ohlson´s model (1980) 49 4.2.3 Analysis of Zmijewski (1984) 51 4.3 Evaluation of the models 52 4.3.1 Accuracy rate 52 4.3.2 Receiver Operating Curves 55 4.4 Hypothesis 57 CHAPTER 5: CONCLUSIONS 60 5.1 Main results 60 5.2 Limitations 61 5.3 Suggestions for future research 62 TABLE OF TABLES Figure1:Trend in amount of bankrupt companies from 2011-2014………………… Figure 2: Conceptual Framework-The relationship between financial ratio and SMEs bankruptcy prediction…………………………………………………………………24 Figure 3: Receiver Operating Curves…………………………………………………36 Figure 4:Accurate comparison …………………………………………………… 54 Figure 5: ROC curve……………………………………………………………… 56 Table 1: The expected signs of variables used in the research……………………… 25 Table 2: Accuracy rate……………………………………………………………… 34 Table 3: Descriptive Statistic……………………………………………………….…42 Table 4: Correlation Coefficients………………………………………………….….44 Table 5: VIF index…………………………………………………………………….45 Table 6: Regression result of Altman model………………………………………… 46 Table 7: Wilks’ lambda test……………………………………………………… …47 Table 8: Regression result of Ohlson model………………………………………… 48 Table 9: Regression result of Zmijewski model……………………………… …….50 Table 10: Possible classifications of response……………………………… ….……52 Table 11: Result of classifications of response………………………………….….…52 Table 12: AUROC result…………………………………………………….……… 55 Table 13: Summary accuracy rate of three models……………………………………56 Table 14: One-way ANOVA………………………………………………………… 56 Table 15: Turkey-Kramer post-hoc analysis…………………………………….…….57 CHAPTER INTRODUCTION 1.1 The empirical problem In 2008, the financial crisis broke out in the United States and quickly spread around the world, leading to the collapse of many financial institutions and the downward of stock markets During 2011-2014 period, under the impact of the world economic crisis and the internal issues, Vietnam economy faced a series of challenges such slower growth rate fluctuating around 5-7 % per year, reduction of purchasing power and bankruptcy phenomenon According to data of Department of Statistic in 2014, amount of bankrupt companies significantly increases annual Figure 1: Trend in amount of bankrupt companies from 2011 to 2014 Bankruptcy 52739 2011 60767 54261 2012 2013 67823 2014 Source: Author’s summary from The General Statistics Office of Vietnam It is alarmed that there are more than fifty-thousand declaring bankruptcy since 2011 Once issue of bankruptcy surges suddenly, it is not only an economic phenomenon , but also a social concern because of its consequence as rising unemployment , spreading impact and losing faith onstable economy Moreover, the failure of acompany causes spread of crisis to other parts of the financial system and crisis (TonatiuhPena etal.2009), leading business bankruptcy phenomenon has received interest from governance managers and parties involved 1.2 Problem Statement Bhagarva et al., 1998 said “Bankruptcies are devastating”, especiallybusiness failure creates significant high cost and heavy losses In fact, predicting likelihood of business bankruptcy is the main concern because it might reduce future costs as well as minimize risk and loss of business and perhaps even prevent the bankruptcy itself Also, prediction about bankruptcy likeihood will help investors, banks or credit institution to choose the good enterprise performance to cooperate, finance and invest capital.Naturally, a reliable method to predict a possible bankruptcy is interested by not only academic but also private agents and government In order to control and measure the financial distress issues as well, business managers need tools for financial analysis and the prediction of bankruptcy model Prediction of bankruptcy is of increasing importance to corporate governance Moreover, the prediction of corporate bankruptcy has been demonstrated in many empirical studies, such so-called Z-Score model (Altman 1968), O-score model (Ohlson 1980) and Zmijewski's model (Zmijewski 1984) These models are used to predict the profitability that a firm will go into bankruptcy within two years and forecast corporate defaults Z-score model being a well-known study of Edward Altman was developed in 1968 The multivariate discriminate analysis (MDA) with more financial ratios was constructed discriminant function inZ-score model Another model to predict company failure is the logit approached by Ohlson (1980) He used the logit model and US firms to develop an estimate of the probability of failure for each firm Ohlson´s work (1980) was then followed by study of Zmijewski (1984) that created another bankruptcy prediction model, the probit model 10 In order to present about the prediction power the accuracy rate of one year later, twoyears later, and three years later from a particular season is computed In the multigroup case, results are illustrated in a classification matrix comprising the results ofoverall accuracy rate, type I and type II Type I errors represent the misclassification of bankrupt firms as non-bankrupt ones Type II errors are the misclassification of non-bankrupt firms as bankrupt firms Table 10: Possible classifications of response Symbol Actual Prediction Description 11 1: bankruptcy 1: bankruptcy Predicting corporation bankruptcy as bankruptcy 10 1: bankruptcy 0:non- bankruptcy Predicting corporation in bankruptcy as non- bankruptcy 01 0:non- bankruptcy 1: bankruptcy Predicting corporation in nonbankruptcy as bankruptcy 0:non- bankruptcy Predicting corporation in nonbankruptcy as non - bankruptcy 00 0:non- bankruptcy in Table 11: Result of classifications of response Predicted year 11 10 Class Class Type I Error Type Ⅱ Error A 43 20 24 41 34% 32% t-1 O 42 21 11 56 32% 43% A 41 22 29 36 32% 28% t-2 O 42 21 58 33% 45% Z 43 20 22 45 33% 35% 16% 16% 19% 8% Z 41 22 19 47 32% 36% 15% 17% 16% 17% 23% 6% 53 A 41 22 29 37 32% 29% t-3 O 43 20 60 33% 46% Z 42 21 29 38 32% 29% 17% 17% 15% 16% 15% 22% 5% 22% Total Accuracy(%) Mean 65.63 75.38 67.69 60.16 67.69% 77.52 68.22 60.47 68.22% 79.23 61.54 61.54% Source: Author’s Calculation Figure 4: Accurate comparison 80% 70% 60% 50% 40% 30% 20% 10% 0% Altman Ohlson Zmijewski t-1 t-2 t-3 Source: Author’s Calculation A comparison of the classification accuracy between the three models affirms thatthe overall predictive ability of Ohlson’s model is highestand the accuracy rate of Altman’s model is lowest in all three years preceding bankruptcy The overall difference between Altman’s model, Ohlson’s model, and Zmijewski (1984) respectively was 65.63%, 73.38% and 67.69% for the first year prior to bankruptcy; 60.16%, 77.52% and 68.22% for the second year prior to bankruptcy and 60.47%, 79.23% and 61.54% for the third year to bankruptcy The mean of overall accuracy rate in one year, two years and three year prior bankruptcy are 67.69%, 68.22% and 61.54% respectively One can conclude that on average t-2 firms are better classified than t-1 and t-3 firms.It is logic to say that predict 54 bankruptcy in about three years is more difficult Nevertheless, t-2 firms better classified than t-1 firms need to be remarked It can be explained that firms when approaching bankruptcy often not provide their financial results with respect to parties, very volatile or even provide very extreme results As mentioned above, type I errors are more costly than type II errors, so the remark concerns is type I error Generally, type I of three models in three investigation period does not represent significant difference that approximate of 16%-17% However, in the Ohlson model, type I and type II are unbalanced that there is an important spread about 10% between their performances It means that the observed bankrupt firms are more misclassified than doing non-bankrupt firms 4.3.2Receiver Operating Curves The recalibrated models of Altman(1968), Ohlson(1980) and Zmijewski(1984)report overall accuracy without correcting the cutoff point The overall accuracy will be higher when all recalibrated models are classified by using an optimal cutoff point which minimizes the sum of the percentages of type I and type II errors Therefore, in order to evaluate the model without depending on a specific cut-off point, it is possible to use the areas under Receiver Operating Curves Because of not distinguishing between false positive rates and false negative rates, models with a larger area under ROC are better Moreover, according tostudy of (Fischer, Bachmann, &Jaeschke, 2003), models with an area under ROC smaller than or equal to 0.5, are useless, the accuracy of models with are under ROC between 0.5 and 0.7 is low, the accuracy of models with area under ROC between 0.7 and 0.9 is moderate, and when the area under ROC is greater than 0.9 the accuracy is high Figure plots the ROC curve of all classifies models It can seethat the five models are close to each other Hence, to get further information looking at the area of ROC(AUROC) in table 12 evaluated atdifferent fitting and prediction periods 55 The results of performance power using AUROC is pretty homologous as using overall accuracy rate.The Ohlson(1980) models have greater areas under ROC than the Altman(1968) and Zmijewski(1984) model in all three years preceding bankruptcy while Altman(1968) andZmijewski(1984) perform equally well.Based on evaluation standard stated by Fischer, Bachmann, &Jaeschke, 2003, the accuracy of Ohlson model is moderate due to area under ROC about 0.8 that between 0.7 and 0.9 and the predictive power of Altman and Zmijewsk model is low Figure 5:ROC curve Ohlson(1980) Zmijewski(1984) 0.50 0.25 Sensitivity 0.75 1.00 Altman(1968) 0.00 t-1 0.00 0.25 0.50 - Specificity 0.75 1.00 0.50 0.25 Sensitivity 0.75 1.00 Area under ROC curve = 0.6818 0.00 t-2 0.00 0.25 Area under ROC curve = 0.7256 56 0.50 - Specificity 0.75 1.00 1.00 0.75 0.50 0.25 Sensitivity 0.00 t-3 0.00 0.25 0.50 - Specificity 0.75 1.00 Area under ROC curve = 0.6579 Source: Author’s Calculation Table 12: AUROC result Predicted year Models AUCROC t-1 A 687 O 804 t-2 Z 0.6818 A 669 O 864 t-3 Z 726 A 638 O 840 Z 658 Source: Author’s Calculation 4.4 Hypothesis After the results above illustrated the accuracy of each model, in this section will present what the effect of these results are on the hypotheses The following hypotheses were formulated and will be tested: Hypothesis (null hypothesis) H0: There is no difference in the accuracy rate between accounting-based bankruptcy prediction models of Altman(1968), Ohlson(1980) and Zmijewski(1984)in the context of Vietnam Table 13: Summary accuracy rate of three models Model Count Sum Average Variance 57 Altman 186.26 62.08 9.44 Ohlson 232.13 77.38 3.72 Zmijewski 197.32 65.77 13.68 Source: Author’s Calculation A one-way ANOVA is calculated in table 14 This table shows the differences between the models and within the models In the table ‘ANOVA’ the variation (Sum Of Squares), the degrees of freedom (df), and the variance (Mean Square) are given for the within and the between groups, as well as the F value (F) and the significance of the F (Sig.) P-value indicates whether the null hypothesis has to be rejected or not Table 14: One-way ANOVA Source of Variation SS Df MS F P-value F-crit Between Groups 382.01 191 21.4 0.0019 5.14 Within Groups Total 53.68 435.69 8.95 Source: Author’s Calculation There is much difference between the two Mean Squares (191 and 8.95), resultingin a significant difference (F =21.35; Sig = 0,00187) It can conclude that this hypothesis can be rejected and that there is a difference in the accuracy rate between accountingbased bankruptcy prediction models of Altman(1968), Ohlson(1980) and Zmijewski(1984)regarding the listed companies in Vietnam Because the models statistically significant differ from each other, it is interesting to perform a Turkey-Kramer post-hoc analysis to find out how the models differ from each other The results of these multi comparisons analysis is shown at table 15 58 Table 15: Turkey-Kramer post-hoc analysis Comparison Altman to Ohlson Altman to Zmijewski Ohlson to Zmijewski Absolute Difference 15.29 3.69 11.6 Critical Results Range 7.49 Means significantly different 7.49 Not Significantly Different 7.49 Means significantly different Source: Author’s Calculation When we look at the Tukey’s Post-hoc table, we see that the post hoc tests are consistent with what we observed with the means The difference between accuracy rate mean of Altman and Ohlson is 15.29 which higher than the critical difference of 7.49 Consequently, there is significant different between the average value of accuracy rate of Altman and Ohlson Similarly, it also performs significantly different between Zmijewski’s model and Ohlson’s model because their absolute difference of 11.6 is greater than critical range of 7.49 Indeed, when comparing Ohlson’s model to the other models, one can conclude that Ohlson’s model outperforms the other models Nevertheless based on the above mentioned evidence, the difference between accuracy rate mean of Altman and Zmijewski is 3.69 which lees than the critical difference of 7.49 showing that two models did not differ significantly in their average value of accuracy rate 59 CHAPTER CONCLUSIONS This study investigated the predictive power of the bankruptcy predicting models of Altman(1968)–multiple discriminant analysis, Ohlson(1980)-logit regression and Zmijewski(1984)-probit regression when it is applied to Vietnamese listed firms in the timeframe 2008 to 2014 These bankruptcy prediction models use accounting ratios to predict bankruptcy The differences between these models are displayed through statistical techniquesand answeringthe underlying question of this Master Thesis whether or not there is a difference between accuracy rates of Altman (1968), Ohlson (1980) and Zmijewski (1984) 5.1 Main results Generally, the result of accuracy rate of three replicated modes is lower than those of original models Moreover, the results display clearly that in the period from 2008 2014 Ohlson (1980) logit model performed most accurate for the one year, two years and three years prior to bankruptcy with an overall predictive accuracy of 77.38% It implies that the selected financial ratios of the Ohlson model (1980) are most accurate in predicting bankruptcy likelihood MDA model has a performance near to probit model but in lower level Indeed, in order to identify bankrupt companies, logit model 60 has better performance It is also noticed that average accuracy rate for two years prior to bankruptcy is higher than of one and three years of investigation period However, unlike expectation, the accuracy rates of model namely Ohlson (1980) decline towards the year of bankruptcy That can be explained by external factors like national economic environments, political affairs, industry.Furthermore, the result of Ohlson model is a highest frequency of Type I and Type II errors when comparing to models of Altman and Zmijewski Therefore, it should be studied carefully about the results of performance and accuracy rate of the three accounting-based bankruptcy prediction models In more depth, thehypothesis was tested in order to assess the predictive power of these bankruptcy prediction models When the original statistical techniques are used, the results demonstrated that there is a significantdifference between the predictive powers of the prediction models The outcome also show that accuracy power of Ohlson model to identify bankruptcy firm is significantly different other models In conclusion, the proposal model among the existing financial bankrupt prediction models is Ohlson model using the binary logistic regression technique because it has slightly better prediction performance However, the existing models chosen could not achieve a satisfactory level of prediction performance Therefore, practitioners should use the bankruptcy prediction models of Altman (1968), Ohlson (1980), and Zmijewski (1984) cautiously when applying them in Vietnam for listed firms As George Box’s Quote“All models are approximations Essentially, all models are wrong, but some are useful However, the approximate nature of the model must always be borne in mind.” Since Vietnam is still an inefficient market with insufficient transparency in the information-disclosure system, the data employed in regressing models is just relatively reliable Nevertheless, it is hoped that this thesis can partly contribute to the bankrupt evaluating ability of Vietnamese enterprises Better evaluation of financial 61 distress helpsmitigate potential harmful effects caused by this economic catastrophe Moreover, this research will add a reference to the topic regarding the forecast of firms’ bankruptcy, which is still a fresh and insufficient-covered topic in Vietnam Moreover, the parties involved will have clearer views to forecast a firm’s financial health and the suitable predict model will bring benefits to users 5.2 Limitations Neither model is sufficient statistic for failure prediction since all of them imply advantages and disadvantages, it should be aware that there are several limitations in this study Firstly, time and size in the study is not big to give a clearer and more defined picture.Actually, the sample includes only 68 companies listed on two stock markets: HOSE and HSX Therefore, the database of financial companies in the sample is so small that be able to estimate the coefficients of independent variables in the models such Altman (1968), Ohlson (1980) and Zmijewski (1984) Secondly, this study focuses only on accounting variables based on historical information and influenced by future trends Accounting variables can be distorted if using different depreciation method and available on yearly basis At last, the study replicated the original models that applied very old data and the variables that are chosen depend on this data It may be better to exclude variables that are insignificant in the recalibrated models, and it is might also be true that there are variables that are not explained in this thesis, but that might have a significant influence on the probability of bankruptcy 5.3 Suggestions for future research 62 First, this study can be further improved in future research by including more nonfinancial variables and market factors in order to significantly improve the prediction accuracies of bankruptcy models Besides, it also suggests that investigate the prediction of bankruptcies during the financial crisis Although the nature and causes of every crisis are different and hardly predict, it is interesting to estimate models which had predicted the bankruptcies during the crisis correctly That is useful “early warning” of potential financial difficulties.It is likely to consider other control variables such the corporate strategy and the competition in the industry Second, the further research should extend area of study; it can be done not only listed companies but also on relatively small size private or non-listed companies Furthermore, a study with multiple statistical techniques would allow to a more detailed analysis about the effects of these statistical techniques 63 REFERENCE Abdullah, N A H., & Ahmad, A H (2008) Predicting corporate failure of Malaysians listed companies: Comparing multiple discriminant analysis, logistic regression and the hazard model.International Research Journal of Finance and Economics, (15), 201-217 Adnan, A.,&Humayon A Dar (2006).Predicting corporate bankruptcy: where we stand? Corporate Governance Vol 6.no.01,pp 18-33 Ali,E,K.,& Nader, N (2011) Comparing Bankruptcy Prediction Models in Iran.Business Intelligence Journal Vol.4 No.2,p.335-343 Altman, E I (2000) Predicting financial distress of companies: revisiting the Z-score and ZETA models Stern School of Business, New York University, 9-12 Altman, E I., Haldeman, R G., & Narayanan, P (1977) ZETA TM analysis A new model to identify bankruptcy risk of corporations Journal of Banking & Finance, 1(1), 29-54 Altman, E.I., Hotchkiss, E (2006) Corporate Financial Distress and Bankruptcy: Predict and Avoid Bankruptcy Analyze and Invest in Distress Debt, Vol.3, John Wiley & Sons, Inc., USA Beaver, W H (1966).Financial ratios as predictors of failure.Journal of Accounting Research, 71- 111 64 Beaver, W.H (1966).Financial ratios as predictors of failure.Journal of Accounting Research, Vol 6, No 2, pp 179 - 192 Beaver, W.H., McNichols, M.F &Rhie, J (2005) Have financial statements become less informative? Evidence from the ability of financial ratios to predict bankruptcy Review of Accounting Studies, 10, 93-122 Chellapilla, K., Larson, K., Simard, P., & Czerwinski, M (2005, April) Designing human friendly human interaction proofs (HIPs) In Proceedings of the SIGCHI conference on Human factors in computing systems (pp 711-720) ACM Edward I Altman (1968) Financial ratios, discriminant analysis and the prediction of corporate bankruptcy.The Journal of Finance, 23, 589-609 Godfrey, K (1985) Comparing the means of several groups.The New England journal of medicine Gordon, M.J (1971) Towards a Theory of Financial Distress.Journal of Finance, pp 347-356 IrajHashi.(1995) The Economics of Bankruptcy, Re-organization and Liquidation.Center for Social & Economic Research Jame A, Ohlson (1980) Financial Ratio and the probabilities prediction of bankruptcy.Journal of Accounting Research, Vol 18 Jo & Han.(1996) Integration of case-based forecasting, neural network and discriminant analysis for bankruptcy prediction.Expert System with application 11(4):415-422 Karels, G., Prakash, A (1987) Multivariate normality and forecasting of business bankruptcy.Journal of Business Finance Accounting, Vol 14, No 4, p 573 – 593 65 KhaliliAraghi, M &Makvandi, S (2012) Evaluating Predictive power of Data Envelopment Analysis Technique Compared with Logit and Probit Models in Predicting Corporate Bankruptcy.Australian Journal of Business and Management Research Vol.2 No.09 [38-46] Kleinert, M K (2014) Comparison of bankruptcy prediction models of Altman (1969), Ohlson (1980) and Zmijewski (1984) on German and Belgian listed companies between 2008-2013 University of Twente, 1-61 Kovatchev, B., Anderson, S., Heinemann, L., & Clarke, W (2008) Comparison of the numerical and clinical accuracy of four continuous glucose monitors.Diabetes Care, 31(6), 1160-1164 Maryam (2013) Comparing logit, probit and Multiple discriminant analysis models in predicting bankruptcy of companies Asia Journal of Finance & Accounting, Vol.5, No.1 McFadden, D (1974) Conditional logit analysis of qualitative choice behavior.In P Zarembka (Ed.), Frontiers in econometrics, pp 104-142.New York: Academic Press McKee, T E (2003) Rough sets bankruptcy prediction models versus auditor signaling rates Journal of Forecasting, 22, 569–586 Philippe & Partners (2002) Bankruptcy and a fresh start: Stigma on failure and legal consequences of bankruptcy Brussel: Deloitte &Touche Corporate finance Platt, H.D & Platt, M.B (2002).Predicting corporate financial distress: reflections on choice-based sample bias.Journal of Economics and Finance, 26, 184-199 Rada, D.,Claudia,C &Daniel.B (2007) Bankruptcy prediction in Norway: a comparison study Statistical Research Report No ISSN 0806–3842 Radha, G, K., &Kishore, K.(2012) A comparison of bankruptcy model International Journal of Marketing, Financial service and Management research 66 Ugurlu, M., Aksoy, H.(2006) Prediction of Corporate Financial Distress in an Emerging Market: the Case of Turkey Cross Cultural Management: An International Journal (4), 277-295 Varadraj,B.&Abhay,N.(2014) Comparison of Bankruptcy Prediction Models: Evidence from India Accounting and Finance Research Vinh (2015) Using Accounting Ratios in Predicting Financial Distress: An Empirical Investigation in the Vietnam Stock Market Journal of Economics and Development, Vol.17, No.1, April 2015, pp 41-49 Zheng &Agresti (2000) Summarizing the predictive power of a generalized linear model.Statistics in medicine Zhou, X.-H., Obuchowski, N., &McClish, D (2011).Statistical methods in diagnostic medicine.Hoboken, New Jersey: John Wiley & Sons, Inc Zmijewski, M E (1984) Methodological issues related to the estimation of financial distress prediction models Journal of Accounting Research, 22, 59-82 Zweig & Campbell, 1993 Receiver-Operating Characteristic (ROC) Plots: A Fundamental Evaluation Tool Clinical Medicine Clinical chemistry, Vol 39, No.4,1, p 561-577 www.vietstock.vn 67 ... This thesisconducted the comparison of predictive power among these modelsusingVietnameselisted firms during the period from 2008 to 2014 The data is taken from listed companies in HOSE and HNX... variety of situations to business managers make crucial decision The bankruptcy prediction models have been developed based on old original modes ofAltman(1968), Ohlson(1980) and Zmijewski(1984)usingmultivariate... financial analysis and the prediction of bankruptcy model Prediction of bankruptcy is of increasing importance to corporate governance Moreover, the prediction of corporate bankruptcy has been demonstrated