(Luận văn) determinants of financial distress, a study of listed companies in vietnam

MINISTRY OF EDUCATION AND TRAINING t to UNIVERSITY OF ECONOMICS HOCHIMINH CITY ng hi ep w n TRẦN THỊ KIM PHƯỢNG lo ad ju y th yi pl ua al DETERMINANTS OF FINANCIAL n DISTRESS: A STUDY OF LISTED n va ll fu COMPANIES IN VIET NAM oi m at nh z ECONOMICS MASTER THESIS z ht vb k jm Major: Business Administration om l.c gm n a Lu n va y te re Ho Chi Minh City - 2012 t to MINISTRY OF EDUCATION AND TRAINING ng UNIVERSITY OF ECONOMICS HO CHI MINH CITY hi ep w n lo TRẦN THỊ KIM PHƯỢNG ad ju y th yi pl DETERMINANTS OF FINANCIAL ua al n DISTRESS: A STUDY OF LISTED n va ll fu COMPANIES IN VIET NAM oi m nh at ECONOMICS MASTER THESIS z z ht vb jm Major: Business Administration k INSTRUCTOR: Võ Xuân Vinh, Ph.D om l.c gm Major Code: 60.34.05 n a Lu n va y te re Ho Chi Minh City - 2012 i t to ABSTRACT ng hi ep This study focuses on researching the relationship between a set w n financial ratios and the probability of failure companies Through the Logistic lo ad regression method, the results show that EPS, Cash per shares and Asset ju y th turnover are the most important financial ratios, which help investors to identify the financial distress of listed companies in Vietnam Stock Exchange yi pl n ua al n va ll fu oi m at nh z z ht vb k jm om l.c gm n a Lu n va y te re ii ACKNOWLEGEMENTS t to ng hi During the time for conducting my thesis, I have been strongly ep supported by many people Through these words, I would like to extend my w sincere to all of them The first one I would like to express my sincere n lo gratitude is my direct supervisor PhD Vo Xuan Vinh, who provides me his ad ju y th great guidance day by day until completing the thesis The next ones I would like to express my special gratitude are my yi pl family, who are always by my side and encourage me if necessary ua al Last but not least, my thesis would be nothing without the enthusiasm n n va and information from my friends ll fu m oi Tran Thi Kim Phuong at nh z z vb ht Ho Chi Minh City, December 2012 k jm om l.c gm n a Lu n va y te re iii STUDENT DECLARATION t to ng hi I hereby declare that the content in this thesis is my own, except for ep special references, quotations and summaries All data, references using in w this research are clearly identified The thesis has not been accepted for any n lo degree until now ad ju y th yi SIGNED: pl DATE: n ua al n va ll fu oi m at nh z z ht vb k jm om l.c gm n a Lu n va y te re iv TABLE OF CONTENT t to ng hi ABSTRACT i ep ACKNOWLEGEMENTS ii w n STUDENT DECLARATION iii lo ad TABLE OF CONTENT iv y th LIST OF TABLES vi ju yi Chapter 1: Introduction of the study pl al Rationale of the study 1.2 Research objectives and questions n ua 1.1 n va Research objectives b) Research questions oi m Structure of the study at nh 1.3 ll fu a) z Chapter 2: Literature Review z ht vb 2.1 Definition of financial distress k jm 2.2 Ratios in designing models gm 2.3 Techniques used in financial distress predictions 12 om l.c 2.4 Hypotheses 15 2.5 Conclusions 16 a Lu Chapter Research Methods 17 n n va 3.1 The model 17 Chapter Data analysis and Findings 23 y 3.3 Data set 20 te re 3.2 Selection of predictor variables 18 v 4.1 Descriptive Statistics 23 t to ng hi 4.2 Correlations 24 4.3 Regression model 25 ep Chapter Conclusions 30 w Summary 30 5.2 Limitation of the research study 31 n 5.1 lo ad ju y th REFERENCES 33 yi APPENDICES 36 pl n ua al n va ll fu oi m at nh z z ht vb k jm om l.c gm n a Lu n va y te re vi LIST OF TABLES t to ng hi Table Summary statistics 23 ep Table Variable correlation 25 w Table The performance of logistic regression for models 26 n lo ad ju y th yi pl n ua al n va ll fu oi m at nh z z ht vb k jm om l.c gm n a Lu n va y te re vii LIST OF ABBREVIATIONS t to HOSE: Ho Chi Minh City Stock Exchange ng hi MDA: Multiple Discriminant Analysis ep ANN: Artificial Neural Networks w n WOCA: Working capital lo ad GROPROM: Gross profit margin y th EPS: Earnings per share ju yi DEBTTOTAL: Total debt to total assets pl ua al CASPSHARE: Cash flow per share n ATURNOVER: Asset turnover va n SALEPERCA: Sales per cash fu ll SALEPERRE: Sales per receivables oi m at nh z z ht vb k jm om l.c gm n a Lu n va y te re t to ng hi ep w n lo ad ju y th yi pl n ua al n va ll fu oi m at nh z z ht vb k jm om l.c gm n a Lu n va y te re 29 Besides, from the result of the model and model 5, McFadden R- t to squared of the model decreased from 29.25% to 21.74% when replacing ng EPS with CASPSHARE It is clearly that EPS had stronger effect on the hi ep probability of failure stronger in comparison with CASPSHARE As such, analyzing the model and model showed that the influence of w n lo ATURNOVER on the probability of failure was more powerful than the ad effect of SALEPERRE y th ju Moreover, the result of logistic regression also demonstrated that yi pl the impact of CASPSHARE on the probability of failure seemed to be higher n ua al compared to ATURNOVER va Finally, with coefficients of independent variables achieved, it is n reasonable to assume that EPS, Cash per shares and Asset turnover have the fu ll most impact on the probability of the failure oi m at nh z z ht vb k jm om l.c gm n a Lu n va y te re 30 Chapter Conclusions t to ng hi 5.1 Summary ep This study have been conducted to find out the relationship between a w n set of financial ratios and the probability of financial distress for listed lo ad companies in Viet Nam Stock Exchange, especially one year prior to the ju y th failure yi Through applying the logistic regression analysis, the results show that pl ua al six out of eight financial ratios as Earnings per share, Asset turnover, Sale per receivables, Cash per share, Working capital, Gross profit margin are n n va negatively related to the probability of failure Besides, Earnings per share, ll fu Asset turnover, and Cash per share are the most important three financial oi m ratios, which have dramatic effect on the state of firms The two ratios at nh positively correlated to the probability of a firm going into financial distress are Total debt to total assets and Sale per cash By comparison, the findings of z z vb this thesis regarding to the profit ratios and activity ratios is consistent with ht the previous researches, i.e the Z – Score model of Altman (1978) jm k Moreover, the result of logistic regression also indicates that the gm l.c financial distress of listed companies in Viet Nam Stock Exchange derives om from ineffective operation activities leading the loss (EPS) as well as the y te re about company’s health with involved ratios It will help them identify firms n investors in Viet Nam For example, investors should collect information va financial ratios and the state of a company seems to be quite helpful for n The aforementioned conclusions regarding the relationship between a Lu capacity in managing companies’ assets to generate revenue (Asset turnover) 31 in financial distress, and reduce risk as much as possible when investing some t to stocks like that ng hi 5.2 Limitation of the research study ep However, one of some limitations is that independent variables still w not contributed enough to the explanation for the financial distress McFadden n lo ad R-squared of a model fluctuates around 30% One reason is that the data can y th only be collected in the period from 2007 – 2011, when the number of ju financial distress listed companies is still not much Although there have been yi pl more and more listed firms falling into the financial distress in 2012, al n ua gathering financial ratios is impossible due to not having the financial va statements at this time The limited data lead to the result that it is difficult to n find out more ratios affecting the probability of the failure such as market fu ll ratios oi m at nh Besides, financial ratios are computed from financial statements, which have problems inherent in the interpretation of accounting standards This z z together with the fact that the results of the thesis depend much on the vb ht reliability of the financial statements are the another limitation jm k With these limitations mentioned above, they suggest further research gm l.c to supplement what the thesis could not reach The conclusions involved in om the relationship between variables and the probability of failure are used as a a Lu base for the further research However, each financial market with different n characteristic will react in not exactly the same way Thus, a set of financial va ratios continue to be researched as adding more macroeconomic ratios such as n longer period as well as analysis the two or three years prior to the financial y In addition, the future researches need to study the data over a much te re inflation 32 distress time Furthermore, apart from the traditional techniques such as t to Logistic Regression and Multivariate Discriminant Analysis, some new ng methods have proven the advantage of predicting the financial distress, i.e hi ep Neural Networks Thus, these techniques will be taken into consideration in later researches Such studies have a significant contribution to the field of w n lo forecasting the failure in Viet Nam ad ju y th yi pl n ua al n va ll fu oi m at nh z z ht vb k jm om l.c gm n a Lu n va y te re 33 REFERENCES t to ng hi Altman, E.I 1968, 'Financial ratios, discriminant analysis and the prediction of ep corporate failure', Journal of Finance, vol 23, no 4, pp 589-609 w Beaver, W.H 1966, 'Financial ratios as predictors of failure', Journal of n lo Accounting Research, vol 4, no 3, pp 71-111 ad Boritz, K., and Sun 2007, 'The effect of general price level adjustments on the y th ju predictability of financial ratios', Journal of Accounting Research, pp 273-84 yi Bradstreet 1985, 'Classifying Bankrupt Firms with Funds Flow Components', pl ua al Journal of Accounting Research, pp 146-60 n Charalambous, C., Charitou, A & Kaourou, F 2000, 'Comparative analysis of va artificial neural network models: Application in bankruptcy prediction', Annals of n ll fu Operations Research, vol 99, pp 403-25 m Chin, F.a 2002, 'The differential bankruptcy predictive ability of specific oi at pp 228-46 nh price level adjustments: some empirical evidence', Accounting Review, vol 58, no 2, z z Deakin, E.B 1972, 'A Discriminant Analysis of Predictors of Business vb ht Failures', Journal of Accounting Research, vol 10, no 1, pp 167-79 jm k Dougall, G.a 1952, 'Predicting Corporate Failure', Accountants Digest, p gm Foster 1986, 'Current Cost Accounting and the Prediction of Small Company om l.c Performance', Journal of Business Finance and Accounting, vol 13, pp 51-70 Goudie 1987, 'Funds Flow Components, Financial Ratios and Bankruptcy', n a Lu Journal of Business Finance, pp 595-606 of oil and gas accounting', Journal of Accounting and Economics, vol 7, pp 39-65 y Heine 1995, 'Mandated accounting changes and debt covenants: The case te re Distress Prediction Models', Journal of Accounting Research, vol 22, pp 59 - 82 n va Hamer 1983, 'Methodological Issues Related to the Estimation of Financial 34 Ingram, G.a 2001, 'Motives underlying the method of payment by UK t to acquirers: the influence of goodwill‟', Accounting and Business Research, vol 30, no ng 3, pp 227- 40 hi ep Jones 1987, 'The Prediction of Small Company Failure: Some Behavioural Evidence for the UK”', Accounting and Business Research vol 65, pp 49-58 w n Lam 1994, 'Identifying failing companies: a re-evaluation of the logit, probit lo ad and DA approaches', Journal of Economics and Business, vol 51, pp 347–64 ju y th Lau 1987, 'A Five-State Financial Distress Prediction Model', Journal of Accounting Research, vol 25, no 1, pp 127-38 yi normality pl Lee, H.a 1985, 'Multivariate and forecasting of business al n ua bankruptcy', Journal of Business Finance and Accounting, vol 14, no 4, pp 573-93 va Marais, P.W 1984, 'Failing company discriminant analysis', Journal of n Accounting Research, vol 12, no 1, pp 1-25 ll fu oi m Muller, S.-B., and Hamman 2009, 'The Non-Submission of Accounts and Small Company Failure Prediction', Accounting and Business Research, vol 73, no at nh 1, pp 47-54 z z Newton 1975, 'The Probability of Bankruptcy: A Comparison of Empirical vb Predictions and Theoretical Models', Journal of Banking and Finance, pp 317 - 44 ht jm Ohlson, J.A 1980, ' Financial ratios and the probabilistic prediction of k gm bankruptcy', Journal of Accounting Research, vol 18, no 1, pp 109-31 l.c Pinches, M.C 1973, 'Non-Financial Symptoms and the Prediction of Small n va Accounting Studies, vol 9, pp 5-34 n Porporato, S.a 2007, 'Assessing the probability of bankruptcy', Review of a Lu and Accounting vol 14, pp 335-54 om Company Failure: A Test of the Argenti Hypotheses', Journal of Business Finance y Usefulness', British Journal of Management, vol 2, pp 89 - 102 te re Prakash, K 1987, 'Financial Distress Prediction Models: A Review of Their 35 Sharda, W 1994, 'Predicting Corporate Bankruptcy and Financial Distress: t to Information Value Added by Multinomial Logit Models', Journal of Economics and ng Business, pp 269-86 hi ep Somerville 1989, 'Failing company discriminant analysis', Journal of Accounting Research, vol 12, no 1, pp 1-25 w n Taffler 1985 ' Industrial classification in UK capital markets: a test of lo ad economic homogeneity', Journal: Applied Economics, vol 17, no 2, pp 291-308 y th Udo 1993, 'An Empirical Comparison of Bankruptcy Models', The Financial ju Review, vol 33, pp 35 – 54 yi pl Zavgren 1983, 'Predicting bankruptcy resolution', Journal of Business Finance al n ua & Accounting, vol 29, no 3, pp 497-520 n va ll fu oi m at nh z z ht vb k jm om l.c gm n a Lu n va y te re 36 APPENDICES t to ng hi Appendix 1: The model ep Dependent Variable: Y Method: ML - Binary Logit (Quadratic hill climbing) Date: 12/20/12 Time: 20:51 Sample: 982 Included observations: 982 Convergence achieved after iterations Covariance matrix computed using second derivatives w n lo ad ju y th 1.905946 0.323463 -0.316477 345.0324 1.428423 -0.002377 n ua n va fu Mean dependent var S.E of regression Sum squared resid Log likelihood Restr log likelihood Avg log likelihood Prob 5.636473 0.414111 -0.639835 5.939647 3.374964 -1.579853 oi m at nh z z vb 982 ht Total obs 0.6788 0.5223 0.0007 0.1141 0.96334 0.172039 28.88697 -108.205 -154.351 -0.11019 ll k jm 0.298966 0.188021 0.232597 0.262472 0.243961 92.29115 36 946 z-Statistic 0.338145 0.781103 0.494623 58.08971 0.423241 0.001504 al McFadden R-squared S.D dependent var Akaike info criterion Schwarz criterion Hannan-Quinn criter LR statistic Prob(LR statistic) Obs with Dep=0 Obs with Dep=1 Std Error pl C WOCA GROPROM EPS ATURNOVER SALEPERCA Coefficient yi Variable om l.c gm n a Lu n va y te re 37 Appendix 2: The model t to ng hi ep w Dependent Variable: Y Method: ML - Binary Logit (Quadratic hill climbing) Date: 12/20/12 Time: 20:35 Sample: 982 Included observations: 982 Convergence achieved after iterations Covariance matrix computed using second derivatives n lo Coefficient ad Variable ju y th C GROPROM EPS DEBTTOTAL ATURNOVER SALEPERCA Std Error yi pl 0.525369 0.503576 56.9017 0.858377 0.422483 0.001454 Prob 3.623257 0.736234 6.23799 0.121682 3.418864 1.697041 0.0003 0.4616 0.9032 0.0006 0.0897 n ua al 1.903546 -0.37075 354.9522 0.104449 1.444411 -0.002467 z-Statistic va 0.298455 0.188021 0.232757 0.262632 0.244122 92.13349 n Mean dependent var S.E of regression Sum squared resid Log likelihood Restr log likelihood Avg log likelihood 0.96334 0.172443 29.02307 -108.284 -154.351 -0.11027 ll fu oi m at nh McFadden R-squared S.D dependent var Akaike info criterion Schwarz criterion Hannan-Quinn criter LR statistic Prob(LR statistic) z Total obs 982 ht vb 36 946 z Obs with Dep=0 Obs with Dep=1 k jm om l.c gm n a Lu n va y te re 38 Appendix 3: The model t to ng hi Dependent Variable: Y Method: ML - Binary Logit (Quadratic hill climbing) Date: 12/20/12 Time: 20:49 Sample: 982 Included observations: 982 Convergence achieved after iterations Covariance matrix computed using second derivatives ep w n lo ju y th Std Error 1.828481 1.437497 -0.029329 324.4049 -0.002101 0.198604 z-Statistic 0.372062 0.85208 0.569854 62.24051 0.001421 0.069535 yi pl Prob 4.914455 1.687044 0.051467 5.212119 1.477884 2.856167 0.0916 0.959 0.1394 0.0043 n ua al C WOCA GROPROM EPS SALEPERCA SALEPERRE Coefficient ad Variable Mean dependent var S.E of regression Sum squared resid Log likelihood Restr log likelihood Avg log likelihood 0.96334 0.172391 29.0055 -109.2 -154.351 -0.1112 n ll fu oi m at nh 0.292518 0.188021 0.234624 0.264499 0.245988 90.30054 va McFadden R-squared S.D dependent var Akaike info criterion Schwarz criterion Hannan-Quinn criter LR statistic Prob(LR statistic) z Total obs 982 ht vb 36 946 z Obs with Dep=0 Obs with Dep=1 k jm om l.c gm n a Lu n va y te re 39 Appendix 4: The model t to ng hi ep w Dependent Variable: Y Method: ML - Binary Logit (Quadratic hill climbing) Date: 12/20/12 Time: 20:49 Sample: 982 Included observations: 982 Convergence achieved after iterations Covariance matrix computed using second derivatives n lo 2.233043 -0.187938 360.2532 -0.105499 -0.002043 0.171723 y th Coefficient ju ad Variable yi pl 0.525691 0.563743 59.97996 0.873924 0.001346 0.065661 n ua al zStatistic Prob 4.247825 0.333374 6.006227 0.120718 1.518495 2.615304 0.7389 0.9039 0.1289 0.0089 n va C GROPROM EPS DEBTTOTAL SALEPERCA SALEPERRE Std Error 0.96334 0.174029 29.55923 -110.647 -154.351 -0.11268 m at nh z z 982 k jm Total obs ht vb 36 946 Mean dependent var S.E of regression Sum squared resid Log likelihood Restr log likelihood Avg log likelihood oi 0.283148 0.188021 0.237569 0.267444 0.248934 87.40812 ll Obs with Dep=0 Obs with Dep=1 fu McFadden R-squared S.D dependent var Akaike info criterion Schwarz criterion Hannan-Quinn criter LR statistic Prob(LR statistic) om l.c gm n a Lu n va y te re 40 Appendix 5: The model t to ng hi Dependent Variable: Y Method: ML - Binary Logit (Quadratic hill climbing) Date: 12/20/12 Time: 20:55 Sample: 982 Included observations: 982 Convergence achieved after iterations Covariance matrix computed using second derivatives ep w n lo ad Coefficient 1.085779 3.487755 1.253483 77.79161 -0.002601 0.191384 Std Error z-Statistic 0.328923 0.789957 0.542987 19.78064 0.001315 0.063976 ju yi pl n ua al C WOCA GROPROM CASPSHARE SALEPERCA SALEPERRE y th Variable Prob 3.301014 4.415119 2.308494 3.932715 -1.977925 2.99149 0.001 0.021 0.0001 0.0479 0.0028 Mean dependent var S.E of regression Sum squared resid Log likelihood Restr log likelihood Avg log likelihood 0.96334 0.177531 30.761 -120.783 -154.351 -0.123 ll fu oi m at nh z 0.217475 0.188021 0.258214 0.28809 0.269579 67.13468 n va McFadden R-squared S.D dependent var Akaike info criterion Schwarz criterion Hannan-Quinn criter LR statistic Prob(LR statistic) z Total obs 982 ht k jm 36 946 vb Obs with Dep=0 Obs with Dep=1 om l.c gm n a Lu n va y te re 41 Appendix 6: The model t to ng hi ep w Dependent Variable: Y Method: ML - Binary Logit (Quadratic hill climbing) Date: 12/20/12 Time: 20:57 Sample: 982 Included observations: 982 Convergence achieved after iterations Covariance matrix computed using second derivatives n lo Coefficient ad Variable ju y th yi pl 1.164864 2.444658 0.677108 96.75102 1.163216 -0.002794 0.317413 0.720135 0.430495 20.99639 0.361959 0.001373 n ua al C WOCA GROPROM CASPSHARE ATURNOVER SALEPERCA Std Error z-Statistic Prob 3.669864 3.394724 1.572858 4.607984 3.213669 -2.035174 0.0002 0.0007 0.1158 0.0013 0.0418 n va ll 0.96334 0.179527 31.45642 121.7928 154.3505 0.124025 oi m Mean dependent var S.E of regression Sum squared resid Log likelihood Restr log likelihood Avg log likelihood at nh z 0.210934 0.188021 0.26027 0.290146 0.271635 65.1154 fu McFadden R-squared S.D dependent var Akaike info criterion Schwarz criterion Hannan-Quinn criter LR statistic Prob(LR statistic) z Total obs 982 ht k jm 36 946 vb Obs with Dep=0 Obs with Dep=1 om l.c gm n a Lu n va y te re 42 Appendix 7: The model t to ng hi Sample: 982 Included observations: 982 Convergence achieved after iterations Covariance matrix computed using second derivatives ep Coefficient w Variable n lo ad ju y th yi 2.530663 0.847624 -2.013271 92.87835 1.257175 -0.003141 z-Statistic 0.533118 0.436244 0.824441 20.66037 0.364339 0.001238 pl ua al C GROPROM DEBTTOTAL CASPSHARE ATURNOVER SALEPERCA Std Error 0.193098 0.188021 0.265877 Schwarz criterion 0.295752 Log likelihood Hannan-Quinn criter 0.277242 Restr log likelihood LR statistic Prob(LR statistic) 59.60965 Avg log likelihood 0.052 0.0146 0.0006 0.0112 0.96334 0.182324 32.44427 124.5457 154.3505 0.126829 ll fu oi m at nh z z 982 ht Total obs vb k jm 36 946 Mean dependent var S.E of regression Sum squared resid n va Obs with Dep=0 Obs with Dep=1 n McFadden R-squared S.D dependent var Akaike info criterion 4.746907 1.943003 -2.441983 4.495483 3.450562 -2.53649 Prob om l.c gm n a Lu n va y te re 43 t to Appendix 8: The model ng hi ep w Dependent Variable: Y Method: ML - Binary Logit (Quadratic hill climbing) Date: 12/20/12 Time: 20:32 Sample: 982 Included observations: 982 Convergence achieved after iterations Covariance matrix computed using second derivatives n lo ad y th Coefficient ju Variable yi pl 2.747885 1.340299 -1.722875 78.46636 -0.002334 0.132492 z-Statistic 0.529652 0.561107 0.819993 19.76388 0.001215 0.056802 n ua al n va C GROPROM DEBTTOTAL CASPSHARE SALEPERCA SALEPERRE Std Error Prob 5.188094 2.388671 2.101085 3.97019 -1.92054 2.332535 0.0169 0.0356 0.0001 0.0548 0.0197 0.96334 0.184105 33.08126 128.5445 154.3505 0.130901 oi at nh z z ht vb Total obs jm 982 k om l.c gm 36 946 Mean dependent var S.E of regression Sum squared resid Log likelihood Restr log likelihood Avg log likelihood m Obs with Dep=0 Obs with Dep=1 0.167191 0.188021 0.274021 0.303897 0.285386 51.61202 ll fu McFadden R-squared S.D dependent var Akaike info criterion Schwarz criterion Hannan-Quinn criter LR statistic Prob(LR statistic) n a Lu n va y te re