unlicensed SME credit rating model in vietnam probability of default assessment

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES HO CHI MINH CITY THE HAGUE VIETNAM THE NETHERLANDS VIETNAM – THE NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS SMEs CREDIT RATING MODEL IN VIETNAM: PROBABILITY OF DEFAULT ASSESSMENT By NGUYEN VIET DUC MASTER OF ARTS IN DEVELOPMENT ECONOMICS Ho Chi Minh City, December, 2016 VIETNAM – THE NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECON SMEs CREDIT RATING MODEL IN VIETNAM: PROBABILITY OF DEFAULT ASSESSMENT by Nguyen Viet Duc A Thesis Submitted in Partial Fulfilment of the Requirements for the Degree of Master of Art in Development Economics Academic Supervisor: Dr Nguyen Thi Thuy Linh Class 21 Ho Chi Minh City, December, 2016 CERTIFICATION “I certify the content of this dissertation has not already been submitted for any degree and is not being currently submitted for any other degrees I certify that, to be the best of my knowledge, any help received in preparing this dissertation and all source used, have been acknowledged in this dissertation.” Signature NGUYEN VIET DUC Date: December 2016 ACKNOWLEDGEMENT I would like to extend my grateful thanks to all people who always encouraged me, supported me, and helped me, by their very own ways to the completion of this thesis First and foremost, in special to my parents, my Chip (NGUYEN Duc Ha Anh), who inspired me to participate this MDE class, firmly supported me, always love me while supported me during hard studying time To my supervisor, Ms NGUYEN Thi Thuy Linh and Mr TRUONG Dang Thuy, who were always willing to, patiently, comprehensively guide me to build up my ideas and structure them in an academic added value paper Last but not the least, I want to extend my thanks to my friends, to all the Risk Management Division members and colleagues from An Binh bank that in some way contributed to my study experiences and made my days happier and more pleasant TABLES AND FIGURES TABLE 1-SAMPLE OF CREDIT RISK ASSESSMENT RANKING BY MOODY IN US 2015 TABLE – VARIABLES TABLE – A SAMPLE OF VARIABLE SELECTION BY ALTMAN TABLE 4-VARIABLES USED BY ALTMAN2 TABLE 5- VARIABLES USED BY MOODY TABLE – DROPPED VARIABLES DUE TO UNFULL FILLED OR MEANINGLESS TABLE 7-APPROPRIATED INDICATORS WITH VALUE TYPE AND TRANSFORMATION TABLE – INDEPENDENT VARIABLES EXPECTED SIGNS IN THE RELATIONSHIP TABLE 9-DEPENDENT VARIABLES DESCRIPTIVE STATISTIC TABLE 10-INDEPENDENT VARIABLES DESCRIPTIVE STATISTIC TABLE 11-DEFAULT STATISTIC FREQUENCY TABLE 12-EXPLANATION OF INDEPENDENT, STATISTICAL SAMPLE TABLE 13-SUMMARY OF FULL MODEL FOR BINOMINAL DEFAULT01 LOGISTIC FUNCTIONS TABLE 14-FINAL MODEL FOR BINOMINAL DEFAULT01 LOGISTIC FUNCTIONS TABLE 15-EXAMPLE OF CHECKING THE POWER OF QUALITATIVE VARIABLES’ CLASSIFYING TABLE 16-EXAMPLE OF QUALITATIVE MARGINAL EFFECTS ON CREDIT RATING TABLE 17-EXAMPLE OF AN AUTOMATIC DEFAULT PREDICTING TOOL TABLE 18-LOAN GROUP DISTRIBUTION FREQUENCY TABLE 19-SUMMARY OF FULL MODEL FOR OLOGIT LOAN GROUP LOGISTIC FUNCTIONS TABLE 20-FINAL MODEL FOR OLOGIT LOAN GROUP LOGISTIC FUNCTIONS TABLE 21-MARGIN TESTING TABLE 22-PREDICTING LOAN GROUP SAMPLE TABLE 23-CLIENT’ PROBABILITY OF FUTURE LOAN GROUP TABLE 24-NUMBER OF DAY IN LATE PAYMENT DISTRIBUTION FREQUENCY TABLE 25-SUMMARY OF FULL MODEL FOR LINEAR DAY OF LATE TABLE 26-FINAL MODEL FOR OLOGIT LOAN GROUP LOGISTIC FUNCTIONS TABLE 27-CLIENT WITH EXPECTED NUMBER OF LATE DAYS IN PAYMENT TABLE 28- CORRELATION OF DUMMY VARIABLES TABLE 29- TEST FOR MULTICOLLINEARITY TABLE 30- TEST FOR HOMOSCEDASTICITY TABLE 31-COMPARE MODEL 1.2 AND 1.3-4 (BOTH IN PREDICTING DEFAULT RISK), BETWEEN RISK LOVER, RISK NEUTRAL AND RISK ADVERSE 16 21 28 29 29 30 33 37 39 40 43 44 FIGURE – LOAN’S SIZE FIGURE - CREDIT RISK MANAGEMENT TECHNIQUES IN BANKING MANAGEMENT FIGURE 3- SAMPLE OF A LOAN’S LIFE FIGURE – CONCEPTUAL FRAMEWORK FIGURE 5: DEPENDENT VARIABLES DISTRIBUTION FIGURE 6: DISTRIBUTION COMPARING (AT 10% CUT VALUE) FIGURE 7: OLOGIT BACK TEST FIGURE 8: LINEAR BACK TEST 10 14 20 28 74 75 75 45 46 49 51 52 53 54 55 60 60 61 62 63 64 68 70 71 72 72 ABSTRACT The largest part in Asset of any bank in its balance sheet is loans, which accounted over 70% of total bank’s asset Therefore loan becomes the biggest factor affecting bank’s profit/loss (PnL) and managing loan becomes the main point in banking management For the reason that credit portfolio plays important role in bank’s PnL, it is required that banks need to issue and implement policies and techniques to manage risk in every stage of granting loans process Bank Vietinbank Eximbank Vietcombank Figure – Loan’s Size Audited Public Data End of 2014 End of 2015 Loans Total Loans Total assets (bill % assets (bill (bill (bill dong) dong) dong) dong) 661,242 542,674 82.07% 779,483 676,688 160,145 97,956 61.17% 124,850 96,188 576,996 323,349 56.04% 674,395 387,103 % 86,81% 77,04% 57,40% (Source: Banks audited financial reports) In the past, credit rating decisions were made by bank’s individual experts (or a team) It seems to be not an efficient way for banks to so Together with the growth of banking industry, many statistical methods for credit rating were developed and introduced as an important tool in finance and banking area Credit models are become an effective way to evaluate the credit risk of clients Many applications of the statistical techniques with more precisions and powers in predicting credit risk create benefits for financial institutions by helping them establish an appropriated strategy for risk mitigation This thesis presents the approach and results of an attempt at using logistic regression to develop a probability of default (PD) predicting model, a linear regression which is also supported by literatures of relevant factors by an ologit model for predicting the future loan group of any applicant Both logistic and linear regression are applied to find out the fit models for commercial banks By choosing suitable models and deeply data analysis from Vietnamese commercial banks, the paper address almost big concerns in credit risk management and client credit worthiness assessment: determine suitable models for Vietnamese SMEs market for both predicting probability of default (PD), number of late payment days (ELG); specify factors that could cause a loan’s potential downgrade (PDL), important information that contributes in creditworthiness of an individual SMEs, the role of cut-off points in implementing banks’ risk appetite and suitable data treatment approaches Key words: credit rating, logistic regression, binominal, multinomial, linear regression, prediction, risk assessment TABLE OF CONTENTS CHAPTER - INTRODUCTION 1.1 Problem statement 1.2 Research objectives and research questions 1.3 Scope of the study 1.4 Contributions and Implications 1.5 Organization of the thesis 11 CHAPTER – LITTERATURE REVIEW 12 2.1 Concepts of credit rating 12 2.2 Worldwide approaches for credit rating 15 2.3 Empirical studies on credit rating 17 CHAPTER – RESEARCH METHODOLOGY 20 3.1 Analytical framework and hypotheses 20 3.2 Estimation methods 22 3.2.1 Binominal logistic regression 23 3.2.2 Multinomial logistic regressions 23 3.2.3 Linear regression quick reviews 24 3.3 Model specification 25 3.4 Data sources and data treatment 26 3.4.1 The Data Set 26 3.4.2 Data treatments 26 3.4.3 Variable Selection 27 3.4.3.1 Dependent variables 27 3.4.3.2 Independent variables 28 3.4.3.2.1 Independent variables selecion 30 3.4.3.2.2-Independent variables transformation 33 CHAPTER – EMPERICAL RESULTS 39 4.1 Descriptive statistics 39 4.2 Regression results 43 4.2.1 The first model with Dependent variable is Default01 43 4.2.1.1 The author’s observation after run a “full model” 44 4.2.1.2 Final binominal regression models 46 4.2.1.3 Interesting results for Model 1.3-4 51 4.2.1.3.1-Qualitative marginal effects 51 4.2.1.3.2-Initiative for automatic tool 52 4.2.2 The second model with Dependent variable is F2-loan group 53 4.2.2.1 The author’s observation after run a “full model” 53 4.2.2.2 Final ologit regression models 55 4.2.2.3 Some more interesting results 59 4.2.3 The third model with Dependent variable is F2n-day of late payment 62 4.2.3.1 The author’s observation after run a “full model” 62 4.2.3.2 Final linear regression models 64 4.2.3.3 Some more interesting results 67 4.3 Test for any other limitation 70 4.3.1 Checking for Multicollinearity 70 4.3.2 Checking for Homoscedasticity 72 4.4 Comparing results of some models 72 4.4.1 Logit functions back test 72 4.4.2 Ologit functions back test 75 4.4.3 Linear functions back test 75 CHAPTER – CONCLUSION AND IMPLICATIONS 77 5.1 Main findings 77 5.2 Limitations of the study 79 5.3 Implications 80 5.4 Suggestion for further studies 82 References 84 Appendix 1-Variable Descriptive Statistic 87 Appendix 2-Full model for Logit functions 90 Appendix 3-Full model for Ologit functions 94 Appendix 4-Full model for Linear functions 96 Appendix 5-Testing of Multicollinearity 98 Appendix 6-Testing of Homoscedasticity 100 Appendix 7-Terminations 101 CHAPTER - INTRODUCTION 1.1 Problem statement Credit is a large industry in the worldwide economy and plays an important role in the growth of firms and countries In one hand, many enterprises have utilized credit lines to make profit or spur their sales In another hand, credit injects capital to the economy, allow production and expansion to bring development to firms in particular and country in general Bank credit can be categorized into four primary types which are loans, discounts, finance leasing and warranties The granting of credit plays a crucial role in the economic development because the fact that not every businesses have enough money or use all their money to finance their projects However, credit institutions not grant credits on its own to all applicants but it should come through a procedure in which they decide whether or not to provide credit to a particular applicant The reason for that is they need to avoid the risk of accepting bad loan (high probability of default) or rejecting good loans (profitable loans) Therefore credit risk management holds an important role in banking industry The banks manage credit risk exposure through a credit risk policies system in which evaluate the credit risk of applicant, so to minimize the default risk together with maximize profit The process of identify credit risk includes collecting previous borrowers’ information, classifying, analyzing multi elements and variables to assess the ability of clients’ repayment The increasing demand of credit and industrial intense competition force banks to implement new schemes to refined statistical methods to facilitate the procedure of making decisions that is also the standard required by Basel Committee on Banking Supervision (BCBS) through their sophisticated documents and papers leading the banking global to a new credit rating/grading generation The credit models make process systematically and also shorten time spent in the loan granting process, thus reduce the cost of banks in approving, minimizing objectiveness and inaccurate decisions by using statistical techniques and Independent variables-Margin Independent variables-Liquidity Independent variables-Payable Independent variables-Jurisdiction Independent variables-Organization and procedures 10 Independent variables-Planning 11 Independent variables-Historical relationship with banks 12 Independent variables-Competitive factors Appendix 2-Full model for Logit functions Model 1.1a Model 1.1b Model 1.2 Model 1.3-4 Appendix 3-Full model for Ologit functions Model 2.1a Model 2.1b Appendix 4-Full model for Linear functions Model 3.1a Model 3.1b IN D Appendix 5-Testing of Multicollinearity Appendix 6-Testing of Homoscedasticity Appendix 7-Terminations Credit: a contractual agreement in which a borrower receives something of value now and agrees to repay the lender at some date in the future, generally with interest The term also refers to the borrowing capacity of an individual or company (http://www.investopedia.com/terms/c/credit.asp) Credit Risks include type of risk: Default Risk, Credit Spread Risk, and Downgrade Risk In this paper, we will examine the Default risk, which is more relevant in credit issuing area and is used for bank in client rating and credit issuing process The Credit Spread risk and Downgrade risk are almost be used in bond market Default risk: The failure to promptly pay interest or principal when due Default occurs when a debtor is unable to meet the legal obligation of debt repayment Borrowers may default when they are unable to make the required payment or are unwilling to honor the debt (http://www.investopedia.com/terms/d/default2.asp) Market value view: 3.1 Probability of default (PD) : the degree of likelihood that the borrower of a loan or debt will not be able to make 3.2 The necessary scheduled repayments Should the borrower be unable to pay, they are then said to be in default of the debt, at which point the lenders of the debt have legal avenues to attempt obtaining at least partial repayment Generally speaking, the higher the default probability a lender estimates a borrower to have, the higher the interest rate the lender will charge the borrower (as compensation for bearing higher default risk) (http://www.investopedia.com/terms/d/defaultprobability.asp) 3.3 Credit default and bank’s P&L: Expected Loss = PD x LGD x EAD Where: Expected Loss is what can cause to bank P&L; PD: Probability of default; LGD: loss at given default (calculated = % of loosed to credited amount); EAD: exposure at default (credited amount) Accounting value view: Loans group and provisions (affect P&L): A loan loss provision is an expense that is reserved for defaulted credits It is an amount set aside in the event that the loan defaults To establish the loan loss provision amounts, bank regulators require regular screening of bank loan portfolios, ranking each loan) by market conditions, collateral condition, and other business risk factors Each loans group has to make a corresponding rate of provision Basel II, BCBS, National central bank- PD and capital adequacy requirements: Capital requirement (also known as regulatory capital or capital adequacy) is the amount of capital a bank or other financial institution has to hold as required by its financial regulator This is usually expressed as a capital adequacy ratio of equity that must be held as a percentage of risk-weighted assets The bank has to keep the ratio follows the requirement The capital is affected directly by loans rating and scoring under these requirements To measure credit default – concept of Credit rating: A statistical analysis performed by lenders and financial institutions to access a person's credit worthiness Lenders use Credit rating, among other things, to arrive at a decision on whether to extend credit A person's credit score is a number between 300 and 850, 850 being the highest credit rating possible (http://www.investopedia.com/terms/c/credit_scoring.asp) ... it in discriminating bad applicants from others For solving this problem, one way for banks and financial institutions is to start using credit rating Credit rating of client’s creditworthiness.. .VIETNAM – THE NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECON SMEs CREDIT RATING MODEL IN VIETNAM: PROBABILITY OF DEFAULT ASSESSMENT by Nguyen Viet Duc A Thesis Submitted in Partial... operating environment of the client Client Information is expected to be put in a black box of credit rating modeling to get the output of client discriminating, where clients are classified into