Default predictors in retail banking – an empirical study in vietnam

HO CHI MINH CITY 2013 MASTER'S THESIS IN DEVELOPMENT ECONOMICS NGUYEN BAO QUOC UNIVERSITY OF ECONOMICS HO CHI MINH CITY VIETNAM INSTITUTE OF SOCIAL STUDIES THE HAGUE THE NETHERLANDS VIETNAM – NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS DEFAULT PREDICTORS IN RETAIL BANKING – AN EMPIRICAL STUDY IN VIETNAM By NGUYEN BAO QUOC MASTER OF ARTS IN DEVELOPMENT ECONOMICS HO CHI MINH CITY, SEPTEMBER 2013 UNIVERSITY OF ECONOMICS HO CHI MINH CITY VIETNAM INSTITUTE OF SOCIAL STUDIES THE HAGUE THE NETHERLANDS VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS DEFAULT PREDICTORS IN RETAIL BANKING – AN EMPIRICAL STUDY IN VIETNAM A thesis submitted in partial fulfillment of the requirements for the degree of MASTER OF ARTS IN DEVELOPMENT ECONOMICS By NGUYEN BAO QUOC Academic supervisor: Dr LE CONG TRU HO CHI MINH CITY, SEPTEMBER 2013 DECLARATION I declare that "Default Predictors in Retail Banking – An Empirical Study in Vietnam" is my own work; it has not been submitted to any degree at other universities I confirm that I have made by effort and applied all knowledge for finishing this thesis in the best way Ho Chi Minh City, September 2013 NGUYEN BAO QUOC i ACKNOWLEDGEMENTS First and foremost I would like to offer my gratitude to my supervisor, Dr Le Cong Tru, for invaluable comments, remarks and engagement through the learning process of the thesis Then I have Mr Le Duc Anh to thank for introducing me to the topic I am also much obliged to Associate Prof Dr Nguyen Trong Hoai, Dr Pham Khanh Nam and Dr Luca Tasciotti for helpful remarks on my TRD as well as keeping me on the right track For the availability of the dataset, I am thankful to MDE Tran Thu Trang from the Head Office of BIDV Last but not least, I am deeply indebted to my parents, my dearly beloved wife, my brothers and sisters for all the understanding and spiritual assistance I will wholeheartedly be grateful forever for your love ii ABSTRACT Due to intense competition, over-lending and economic turmoil, banking system in Vietnam is suffering a huge amount of non-performing loans Given the considerable growth of retail banking market, an exploration of risk predictors becomes crucial more than ever This paper investigates key factors that influence loan repayment performance among individual customers The survey covers a representative sample of personal loans from one of the largest Vietnamese commercial banks A logistic regression technique is employed to evaluate the relationship between delinquency and borrower characteristics and loan features The regression results reveal that borrower characteristics, e.g borrowing history, bankaccount holding and education level, rather than loan factors, such as purposes, duration and credit limit, have stronger effects on the default outcome This suggests that bankers apply appropriate adjustments to borrower characteristics to minimize default risk Key words: retail banking, credit scoring, default, risk, logistic regression, probability iii TABLE OF CONTENTS DECLARATION i ACKNOWLEDGEMENTS ii ABSTRACT iii TABLE OF CONTENTS iv LIST OF TABLES vii LIST OF FIGURES vii LIST OF ABBREVIATIONS .viii Chapter INTRODUCTION 1.1 Background 1.2 Problem statement 1.3 Research objectives 1.4 Research questions 1.5 Justification of the study 1.6 Scope of the study 1.7 Organization of the study Chapter LITERATURE REVIEW 2.1 History of credit scoring 2.2 Concepts of credit scoring 2.3 Reviews of economic theories 10 2.4 Reviews of empirical studies 12 2.4.1 Default predictors in markets for credit cards and instant loans 12 2.4.2 Default predictors in markets for automobiles, mortgages and real property construction 14 2.4.3 Default predictors in markets for individual loans 16 2.5 Chapter summary 17 2.5.1 Empirical literature summary 17 2.5.2 Problems and limitations of previous studies 20 2.5.3 Conceptual framework 21 iv Chapter DATA AND RESEARCH METHODOLOGY 22 3.1 Data collection 22 3.2 Variables measurements 23 3.2.1 Response variable 23 3.2.2 Explanatory variables 24 3.2.2.1 Borrower characteristics 24 3.2.2.2 Loan characteristics 27 3.3 Research methodology 28 3.3.1 Descriptive analysis 28 3.3.2 Econometric model 29 3.3.2.1 Methodologies for CSM 29 3.3.2.2 Logistic regression 30 3.4 Validation 32 3.4.1 Overall evaluations 33 3.4.2 Statistical tests of individual predictors 34 3.4.3 Goodness-of-fit statistics 34 3.4.3.1 Pseudo R-squared statistics 34 3.4.3.1.1 Cox and Snell's R2 34 3.4.3.1.2 Nagelkerke's R2 35 3.4.3.2 Hosmer and Lemeshow test 35 3.4.4 Validations of predicted probabilities 36 3.4.4.1 Classification table 36 3.4.4.2 Area under the ROC curve 37 3.5 Analytical framework 38 3.6 Chapter summary 38 Chapter DATA ANALYSIS AND RESULTS 39 4.1 Descriptive statistics 39 4.1.1 Personal tastes for loans by ages 41 4.1.2 Discretionary incomes and default 42 4.1.3 Nexus between loan amount and loan outcomes 43 4.1.4 Loan duration and loan outcomes 43 4.1.5 Collateral value and loan outcome 44 4.1.6 Differences in variables between defaulted and non-defaulted loans 45 4.1.7 Correlation matrix among independent variables 45 v 4.2 Information value 46 4.3 Empirical results 48 4.3.1 Model estimation 48 4.3.2 Assumption verification 50 4.3.3 Model validation 51 4.3.3.1 Overall evaluations and statistical tests of individual predictors 51 4.3.3.2 Goodness-of-fit statistics 52 4.3.3.3 Validations of predicted probabilities 52 4.3.3.3.1 Classification table 52 4.3.3.3.2 Receiver operating characteristic and area under the ROC curve 53 4.3.4 Result interpretation 53 4.3.4.1 Borrower characteristics 54 4.3.4.2 Loan characteristics 56 4.4 Chapter summary 57 Chapter CONCLUSION AND POLICY IMPLICATIONS 58 5.1 Conclusion 58 5.2 Policy implications 59 5.3 Limitations and further studies 62 REFERENCES 63 APPENDIX 67 vi LIST OF TABLES Table 2.1 Credit scoring vs Credit rating Table 2.2 Summary of variables 18 Table 3.1 Overview of variables 31 Table 3.2 Predictive accuracy of CSMs 36 Table 4.1 Variables initially considered for the CSM 40 Table 4.2 Loan type statistics 41 Table 4.3 Loan duration 43 Table 4.4 Differences in variables between the loan outcomes 45 Table 4.5 Correlation coefficients among continuous independent variables 46 Table 4.6 Information values for explanatory variables 47 Table 4.7 Regression results 49 Table 4.8 Classification table 52 Table 4.9 Performance of the models 53 LIST OF FIGURES Figure 2.1 Process of credit scoring Figure 2.2 Conceptual framework 21 Figure 3.1 ROC Curve and AUC 37 Figure 3.2 Steps in binary logistic regression 38 Figure 4.1 Gender and loan sample 39 Figure 4.2 Average loan size vs age and purposes 42 Figure 4.3 Default frequencies among different groups of discretionary incomes 42 Figure 4.4 Default frequencies among different groups of loan amounts 43 Figure 4.5 Default frequencies among different groups of loan duration 44 Figure 4.6 Default frequencies among different ratios of collateral-to-loan 44 Figure 4.7 ROC curves and AUC 53 vii LIST OF ABBREVIATIONS BIDV Joint Stock Commercial Bank for Investment and Development of Vietnam BIS Bank for International Settlements CAPM Capital Asset Pricing Model CSM Credit Scoring Model ECOA Equal Credit Opportunity Act NPL Non-performing Loan SBV State Bank of Vietnam VND Vietnam dong viii Chapter CONCLUSION AND POLICY IMPLICATIONS This is the final section of our research, finishing the study by summarizing the findings and concluding the thesis with managerial implications We will also discuss the limitations as well as reveal suggestions for further research in the area of credit scoring and default predicting at the end of this chapter 5.1 Conclusion The study has accessed personal credit risk in the context of Vietnam's emerging market where bank loans and particularly NPLs have dramatically risen over the past few years Given the growing credit to individuals and increased regulatory requirements of risk management, the development of a powerful and reliable CSM is essential Typically, the objective of such model is to minimize the rate of credit default and the number of misclassified loans Our research has empirically investigated determinants of default behavior with a set of data taken over a period of four years The sample includes unique and sensitive information of clients such as age, marital status, education, income, etc reported by one of the largest banks in Vietnam There are more than 30 variables for each borrower; however, many of them are disregarded because they lack a univariate relation with the response variable or they are better explained by other independent ones Out of 18 variables from the first selection, once again five are eliminated because of their low predictive power In the last stage, all the logistic models are able to survive a wide range of validation tests to make sure that the outputs fully satisfy main requirements of stability and discriminatory power We have managed to build powerful CSMs which can capture a borrower's repayment performance Based on the baseline model, we have constructed three others and compared them in terms of discriminatory power and efficiency Our study has produced results that corroborate the findings of much of the previous work in the field Although most of the factors relevant to credit decision are found to be the same as that in the developed countries, they show some different relationships with the default in our study, where they unveil economic and cultural aspects that are unique to Vietnam as an emerging market economy Strong default outcome is associated with clients who are less educated, young in work experience but have many dependents Especially, those who used to have a bad debt history are most likely to fall into arrears again On the contrary, non-default behavior is connected 58 with applicants who get a high educated level, well-paid job, high savings and long-standing relationship with the bank Particularly, borrowers who have a checking account and use it for daily trading are most probable to manage their debt well Interestingly, long duration and high interest rate not exert a positive influence on the default behavior All of these findings suggest the bank apply appropriate adjustments to predictive characteristics to minimize credit risk Our overall aim and research questions are successfully solved As expected, incomes impact negatively on the insolvency whereas loan size positively does Loan duration, meanwhile, has a negative effect on the default against all expectations There is a tendency for high-income borrowers to get large-size loans; however, in case there is unexpected falling in incomes or rising in expenses, the borrowers may not stand the financial burden of large debts Borrowing history is the most powerful predictor which reveals that bad debt records have a significant and positive effect on the default Notwithstanding, there is no convincing evidence of a link between a good repayment history and exposure to the default Borrowers with business purposes are inclined to have higher chance of making good repayment than those who seek financing for consumption This explains the reason why interest rate for business loans is always lower than that for consumption loans with respect to the same duration Collateral is not sufficient to ensure a non-default but it can induce the customer's willingness to keep up punctual installments Indeed, most large-size loans are not accepted if they are not fully collateralized 5.2 Policy implications The risk management models which have been constructed and validated can serve as a useful tool for the bank and policy makers to assess credit default risk as well as apply appropriate strategies to minimize the risk Thus, this emerges as a key contribution of our study We have estimated four CSMs in which Model serves as the base for the other three Model is recognized as the best of all since it can differentiate between the good and bad clients with fewer predictors and categories This will help to control over-fitting problem as well as unnecessarily lengthy application forms The model can be employed to implement risk-based pricing to manage the bank's loan portfolio From the baseline model, we have identified another specification without the variable Repayment resources Model can also produce reliable results though it has lower discriminatory power because the typically significant predictor is eliminated As a result, this model offers the bank an effective means of checking for misinformation or potential fraud by 59 using two different CSMs: one with the declared repayment resources taken into account and one without If there are any serious differences in the findings, the applicant should need stricter investigation The last model has low discriminatory power compared to the others since the significant variables of loan characteristics are dropped However, it can help the bank forecast performance capacity of its potential customers After the prospective clients have been approached and the bank has realized their need, Models and can be used interchangeably for risk management purposes On closer comparison between the models, we find it necessary to pay more attention to some sensitive factors such as Loan duration, Interest rate, Collateral-to-loan ratio and CIC, i.e credit history profile Regarding loan duration, as it reveals a significantly negative effect on the default, it reflects a client's risk aversion and self-assessment of repayment ability Therefore, the longer the loan term gets, the higher chance for the debtor is to repay well This implies that bankers should take every opportunity to finance funds with long-term contracts For example, the terms for mortgages should be extended to 15–20 years and even more Also with long term portfolio, the outstanding balance would be more stable and unit cost per loan can be kept at a minimal level In addition, longer duration often brings in better relationship between the bank and its clients, which in turn lessens the chance of default Furthermore, there is a wide variety of insurance policies to choose from, e.g insurance for borrowers' health or pledged assets, and even insurance for outstanding loans in case of injury, loss of working capacity or death Either of these measures can be taken to alleviate the risk of bad debt in the long run With respect to interest rate, since it is obvious for those with low creditworthiness to default more, it is practical to differentiate the loan price in accordance with the applicant's classification Accordingly, those with higher credit rating should reasonably pay a lower price for their loans However, the current situation is more problematic than what is expected First on the bank's side, as the loan market is highly competitive, the creditor tends to offer the same loan price for the same purpose even if some borrowers appear to be riskier than the others Second regarding the banking regulations, loan price is subject to the base interest rate, which is decided by the State Bank and considered as a benchmark against any loan, especially loans for business purposes Such a command policy will benefit only badrisk borrowers and induce the bank not to take on business loans in their portfolio All in all, interest rate should reflect the result of negotiations between the bank and its clients on the basis of supply and demand of funds 60 Concerning loan collateral, the changing coefficient signs of Collateral-to-loan ratio between Models and pose two interesting findings In Model 1, collateral is observed to be associated with greater chance of defaulting Instead of ignoring or reducing pledged assets the bank, however, is prompted to ask more security from risky clients, and charge them with higher price for the loan as a consequence While in Model 3, a secured loan is considered to be connected with higher quality of customers As a result, more security is recommended since it can lessen the problem of moral hazard In both cases, collateral minimizes the danger of loan loss in the event of non-reimbursement Even if default does not happen, a loan with collateral helps to save the bank from loan loss provision,33 which is periodically calculated based on the difference between the outstanding loan and the loan value that has collateral as a backup To conclude, it is of importance that bankers demand as much collateral as possible In connection with a customer's credit history profile, the study reveals no empirical evidence that a great borrowing history will guarantee the applicant's good performance in the future On the contrary, the findings demonstrate that clients with bad debt histories will be most likely to miss their loan repayments again This means the lender could benefit from being reluctant to business with those applicants and in some cases even turning them down However, the negative impact of the default tends to reduce over time and may be outweighed by latest positive repayment records Therefore, if the bank is about to dealing with them, such credit applications should be exposed to stricter examinations Once they have passed the credit assessment, high price, i.e reasonable interest rate, and/or high-quality liquid assets are strongly recommended Concerning the SBV's credit information activities, a credit report is inherent in each customer's identification and permanent residential place which is managed in terms of a household registration book in accordance with current regulations As a result, if an applicant has another ID number due to moving to a new place, for instance, the credit report retrieving may be unavailable This is a defect of the credit reference agency as each Vietnamese has not yet been granted a lifelong ID number Besides, since a default stays on the credit-data-sharing net for many years there should be room for defaulters to improve their status by explaining on the credit reports and updating notices of correction such as recent good repayment information Accordingly, they are able to find competitive deals that suit their current circumstances Overall, this study only provides a single estimation of CSMs Due to unceasing development of the credit market, especially in such a developing country where economic changes are often more pronounced than that in developed countries, the risk management 33 Decision No 18/2007/QĐ-NHNN dated April 25, 2007 61 models should be updated regularly In addition, the cut-off value can be raised to a higher level so that more customers are to be classified into the refused group and then the quality of loan portfolios will be improved 5.3 Limitations and further studies The most important limitation lies in the fact that our data sample is collected from the portfolio which has already passed the credit risk assessment; hence, the regression may not produce best results if it is applied to the selection of new loans This means reject inference can constitute some challenges Though it is common to be ignored in literature, the problem of bias selection can be improved by taking into account applicants whose loans are rejected However, a major problem with keeping the profile of denied applicants is that no guarantee can be given whether a refused loan would turn into a good or bad outcome As a result, advances in solving such defect are needed to strengthen the desired discriminatory power So far we have carried out the research on the dichotomous outcome of default and non-default as we not have any specific information on the timing of default However, it is necessary for the bank not only to know whether an applicant will default but also when he or she will Therefore, a further investigation which takes into account the survival of loan is substantially significant for the provision making and thus the bank's expected earnings Finally, our default predicting, which has just concentrated on account borrower characteristics and loan features, may lead to some limited results In reality, defaults can vary greatly due to macroeconomic factors or regional effects, especially when the sluggish economy in Vietnam can continue for the coming years Hence, it would be highly recommended for macroeconomic factors to be incorporated into the risk management models In this way, we can compare the results of our estimated models in various stages of economic development As a consequence, we expect other studies to explore more linkages between a wider range of characteristics that determine loan default To put it in a nutshell, the above arguments present the extent to which our study does not cover As a consequence, more research is needed to improve the predictive value of our econometric models By means of a better specification, the default rates could be controlled, which would in turn make it possible for bankers to boost loan portfolios and translate considerably into future profits 62 REFERENCES Agarwal, S., Ambrose, B., & Chomsisengphet, S (2008) Determinants of automobile loan default and prepayment Economic Perspectives, 32(3), 17-28 Agarwal, S., Chomsisengphet, S., & Liu, C (2011) Consumer bankruptcy and default: The role of individual social capital Journal of Economic Psychology, 32(4), 632-650 doi: 10.1016/j.joep.2010.11.007 Allen, L., DeLong, G., & Saunders, A (2004) Issues in the credit risk modeling of retail markets Journal of Banking & Finance, 28(4), 727-752 doi: 10.1016/j.jbankfin.2003.10.004 Altman, E I (1980) Commercial bank lending: Process, credit scoring, and costs of errors in lending The Journal of Financial and Quantitative Analysis, 15(4), 813-832 Altman, E I., & Sabato, G (2007) Modelling credit risk for SMEs: Evidence from the U.S market Abacus, 43(3), 332-357 doi: 10.1111/j.1467-6281.2007.00234.x Altman, E I., & Saunders, A (1997) Credit risk measurement: Developments over the last 20 years Journal of Banking & Finance, 21(11), 1721-1742 Anderson, R (2007) The Credit Scoring Toolkit: Theory and Practice for Retail Credit Risk Management and Decision Automation New York: Oxford University Press Andy, F (2000) Discovering Statistics Using SPSS for Windows: Advanced Techniques for Beginners (1 ed.) California: Sage Publications, Inc Autio, M., Wilska, T.-A., Kaartinen, R., & Lähteenmaa, J (2009) The use of small instant loans among young adults – a gateway to a consumer insolvency? International Journal of Consumer Studies, 33(4), 407-415 doi: 10.1111/j.1470-6431.2009.00789.x Backé, P., & Wójcik, C (2008) Credit booms, monetary integration and the new neoclassical synthesis Journal of Banking & Finance, 32(3), 458-470 doi: 10.1016/j.jbankfin.2007.07.005 Baesens, B., Gestel, T V., Viaene, S., Stepanova, M., Suykens, J., & Vanthienen, J (2003) Benchmarking State-of-the-Art Classification Algorithms for Credit Scoring The Journal of the Operational Research Society, 54(6), 627-635 doi: 10.2307/4101754 Banasik, J., Crook, J., & Thomas, L (2003) Sample Selection Bias in Credit Scoring Models The Journal of the Operational Research Society, 54(8), 822-832 doi: 10.2307/4101652 Beaver, W H (1966) Financial ratios as predictors of failure Journal of Accounting Research, 4, 71-111 Bester, H (1985) Screening vs Rationing in Credit Markets with Imperfect Information The American Economic Review, 75(4), 850-855 Bewick, V., Cheek, L., & Ball, J (2005) Statistics review 14: Logistic regression Critical Care, 9(1), 1-7 doi: 10.1186/cc3045 BIDV (2012) Annual report 2011 Retrieved from http://bidv.com.vn/Nha-dau-tu/Bao-caotai-chinh/Bao-cao-thuong-nien/BIDV-Annual-Report-2011.aspx BIS (2006) International Convergence of Capital Measurement and Capital Standards: A Revised Framework, Comprehensive Version (B C o B Supervision Ed.) Switzerland: Bank for International Settlements Boot, A., Thakor, A V., & Udell, G (1991) Secured Lending and Default Risk: Equilibrium Analysis, Policy Implications and Empirical Results The Economic Journal, 101(406), 458-472 Brill, J (1998) The importance of credit scoring models in improving cash flow and collections Business Credit, 100(1), 16-17 63 Brown, S., Taylor, K., & Wheatley Price, S (2005) Debt and distress: Evaluating the psychological cost of credit Journal of Economic Psychology, 26(5), 642-663 doi: 10.1016/j.joep.2005.01.002 Burns, R P., & Burns, R (2008) Logistic regression (Chapter 24, Extra advanced chapters) Business research methods and statistics using SPSS London: SAGE Publications Ltd Chen, M.-C., & Huang, S.-H (2003) Credit scoring and rejected instances reassigning through evolutionary computation techniques Expert Systems with Applications, 24(4), 433-441 doi: 10.1016/s0957-4174(02)00191-4 Chen, X., Ender, P., Mitchell, M., & Wells, C (2003) Regression with Stata Retrieved from http://www.ats.ucla.edu/stat/stata/webbooks/reg/default.htm Claessens, S., Krahnen, J., & Lang, W (2005) The Basel II reform and retail credit markets Journal of Financial Services Research, 28(1), 5-13 doi: 10.1007/s10693-005-4354-6 Cohen, J., Cohen, P., West, S G., & Aiken, L S (2003) Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences (3 ed.) New Jersey: Lawrence Erlbaum Associates Crook, J., & Banasik, J (2004) Does reject inference really improve the performance of application scoring models? Journal of Banking & Finance, 28(4), 857-874 doi: http://dx.doi.org/10.1016/j.jbankfin.2003.10.010 Dinh, T H T., & Kleimeier, S (2007) A credit scoring model for Vietnam's retail banking market International Review of Financial Analysis, 16(5), 471-495 doi: 10.1016/j.irfa.2007.06.001 Duesenberry, J S (1949) Income, saving and the theory of consumer behavior (Vol 87) Cambridge: Harvard University Press Durand, D (1941) Risk elements in consumer installment lending Nev York: National Bureau of Economic Research Durkin, T A (2000) Credit cards: Use and consumer attitudes, 1970-2000 Federal Reserve Bulletin, 86, 623-634 Equal Credit Opportunity Act (ECOA) § 202.1 (1974) Fed (2007) Report to the Congress on Credit Scoring and Its Effects on the Availability and Affordability of Credit: Washington, DC: US Federal Reserve Fishelson-Holstine, H (2004) The role of credit scoring in increasing homeownership for underserved populations Harvard Joint Center for Housing Studies Greene, W (1998) Sample selection in credit-scoring models Japan and the World Economy, 10(3), 299-316 doi: 10.1016/s0922-1425(98)00030-9 Gross, D B., & Souleles, N S (2002) Do Liquidity Constraints and Interest Rates Matter for Consumer Behavior? Evidence from Credit Card Data The Quarterly Journal of Economics, 117(1), 149-185 doi: 10.1162/003355302753399472 Hand, D J., & Henley, W E (1993) Can reject inference ever work? IMA Journal of Management Mathematics, 5(1), 45-55 doi: 10.1093/imaman/5.1.45 Hand, D J., & Henley, W E (1997) Statistical Classification Methods in Consumer Credit Scoring: A Review Journal of the Royal Statistical Society Series A (Statistics in Society), 160(3), 523-541 doi: 10.2307/2983268 Hosmer, D W., & Lemeshow, S (2000) Applied logistic regression (2nd ed.) New York: John Wiley & Sons Jacobson, T., & Roszbach, K (2003) Bank lending policy, credit scoring and value-at-risk Journal of Banking & Finance, 27(4), 615-633 doi: 10.1016/s0378-4266(01)00254-0 64 Jappelli, T., Pischke, J.-S., & Souleles, N S (1998) Testing for Liquidity Constraints in Euler Equations with Complementary Data Sources Review of Economics and Statistics, 80(2), 251-262 doi: 10.1162/003465398557492 Keynes, J M (1936) The General Theory of Employment, Interest and Money London: Macmillan Kočenda, E., & Vojtek, M (2011) Default predictors in retail credit scoring: Evidence from Czech banking data Emerging Markets Finance and Trade, 47(6), 80-98 doi: 10.2753/REE1540-496X470605 Lawrence, D., & Solomon, A (2002) Managing a Consumer Lending Business: 1st Impression Publishing Liu, Y (2001) New issues in credit scoring application Work report No, 16 Lusardi, A (2006) Comment on: “A portfolio view of consumer credit” Journal of Monetary Economics, 53(1), 85-88 doi: 10.1016/j.jmoneco.2005.10.006 Menard, S (2000) Coefficients of Determination for Multiple Logistic Regression Analysis The American Statistician, 54(1), 17-24 doi: 10.2307/2685605 Menard, S (2002) Applied logistic regression analysis (2nd ed.) Thousand Oaks, CA: SAGE Publications, Incorporated Mester, L J (1997) What’s the point of credit scoring? Federal Reserve Bank of Philadelphia Business Review (September–October), 3, 3-16 MGI (2012) Sustaining Vietnam's growth: The productivity challenge McKinsey Global Institute Musto, D K., & Souleles, N S (2006) A portfolio view of consumer credit Journal of Monetary Economics, 53(1), 59-84 doi: 10.1016/j.jmoneco.2005.10.009 Myers, J H., & Forgy, E W (1963) The Development of Numerical Credit Evaluation Systems Journal of the American Statistical Association, 58(303), 799-806 doi: 10.1080/01621459.1963.10500889 Nagelkerke, N J D (1991) A Note on a General Definition of the Coefficient of Determination Biometrika, 78(3), 691-692 doi: 10.2307/2337038 Nurkse, R (1957) Problems of capital formation in underdeveloped countries (5th ed.) Oxford: Blackwell Ohlson, J A (1980) Financial Ratios and the Probabilistic Prediction of Bankruptcy Journal of Accounting Research, 18(1), 109-131 Özdemir, Ö., & Boran, L (2004) An empirical investigation on consumer credit default risk Turkish Economic Association Working Paper(2004/20) Peng, C.-Y J., & So, T.-S H (2002) Logistic Regression Analysis and Reporting: A Primer Understanding Statistics, 1(1), 31-70 doi: 10.1207/s15328031us0101_04 Peter, V., & Peter, R (2011) Risk Management Model: an Empirical Assessment of the Risk of Default Journal of Risk and Diversification(1), 1-18 Plaut, S E (1985) The theory of collateral Journal of Banking & Finance, 9(3), 401-419 doi: 10.1016/0378-4266(85)90041-x Ritter, D (2012) Do We Still Need the Equal Credit Opportunity Act? FRB of Philadelphia Payment Cards Center Discussion Paper(12-03) doi: 10.2139/ssrn.2154865 Rothschild, M., & Stiglitz, J (1976) Equilibrium in Competitive Insurance Markets: An Essay on the Economics of Imperfect Information The Quarterly Journal of Economics, 90(4), 629-649 State Bank of Vietnam (2007) Decision No 18/2007/QĐ-NHNN dated April 25, 2007 of the State Bank on amendment of and addition to a number of articles of the Regulations on classification of debts, and establishment and use of reserves to deal with credit risks in banking operations by credit institutions, issued with Decision 493/2005/QĐ65 NHNN of the State Bank dated April 22, 2005 (Vietnamese title: "Quyết định số 18/2007/QĐ-NHNN ngày 25/4/2007 Về việc sửa đổi, bổ sung số điều Quy định phân loại nợ, trích lập sử dụng dự phòng để xử lý rủi ro tín dụng hoạt động ngân hàng tổ chức tín dụng ban hành theo Quyết định số 493/2005/QĐNHNN ngày 22/4/2005 Thống đốc Ngân hàng Nhà nước") State Bank of Vietnam (2007) Decision No 51/2007/QĐ-NHNN dated December 31, 2007 of the State Bank on the issuance of the Regulation on credit information activities (Vietnamese title: "Quyết định số 51/2007/QĐ-NHNN ngày 31/12/2007 Thống đốc Ngân hàng Nhà nước Về việc ban hành Quy chế hoạt động thơng tin tín dụng") State Bank of Vietnam (2013) Circular No 03/2013/TT-NHNN dated January 28, 2013 of the State Bank on the credit information activities of the State Bank of Vietnam (Vietnamese title: "Thông tư số 03/2013/TT-NHNN ngày 28/1/2013 Thống đốc Quy định hoạt động thơng tin tín dụng Ngân hàng Nhà nước Việt Nam") Siddiqi, N (2006) Credit Risk Scorecards: Developing and Implementing Intelligent Credit Scoring (Vol 3): Wiley Stevens, J (2009) Applied multivariate statistics for the social sciences (5th ed.) New York: Taylor & Francis Group Stiglitz, J E., & Weiss, A (1981) Credit rationing in markets with imperfect information The American Economic Review, 71(3), 393-410 Straka, J W (2000) A shift in the mortgage landscape: The 1990s move to automated credit evaluations Journal of Housing Research, 11(2), 207-232 Summers, B., & Wilson, N (2000) Trade Credit Management and the Decision to Use Factoring: An Empirical Study Journal of Business Finance & Accounting, 27(1-2), 37-68 doi: 10.1111/1468-5957.00305 Tabachnick, B G., & Fidell, L S (2007) Using multivariate statistics (5th ed.) Boston, MA: Pearson Tasche, D (2005) Rating and probability of default validation In Basel Committee on Banking Supervision (ed.), Working Paper No.14, Rating and Probability of Default Validation Bank of International Settlements, Basel Thomas, L C (2000) A survey of credit and behavioural scoring: Forecasting financial risk of lending to consumers International Journal of Forecasting, 16(2), 149-172 doi: http://dx.doi.org/10.1016/S0169-2070(00)00034-0 Van Gestel, T., Baesens, B., Van Dijcke, P., Garcia, J., Suykens, J A K., & Vanthienen, J (2006) A process model to develop an internal rating system: Sovereign credit ratings Decision Support Systems, 42(2), 1131-1151 doi: http://dx.doi.org/10.1016/j.dss.2005.10.001 Van Gestel, T., Baesens, B., Van Dijcke, P., Suykens, J., Garcia, J., & Alderweireld, T (2005) Linear and nonlinear credit scoring by combining logistic regression and support vector machines Journal of Credit Risk, 1(4), 31-60 Van Gool, J., Verbeke, W., Sercu, P., & Baesens, B (2012) Credit scoring for microfinance: Is it worth it? International Journal of Finance & Economics, 17(2), 103-123 doi: 10.1002/ijfe.444 Vuong, Q H., Dao, G H., Nguyen, V H., Tran, M N., & Le, H P (2006) Phương pháp thống kê xây dựng mơ hình định mức tín nhiệm khách hàng thể nhân (English title: "Statistical method for individual credit scoring") VN Journal of Mathematical Applications, 4(2), 1-16 Yang, Y., Nie, G., & Zhang, L (2009) Retail Exposures Credit Scoring Models for Chinese Commercial Banks Paper presented at the Proceedings of the 9th International Conference on Computational Science, Baton Rouge, LA, USA 66 APPENDIX Table A.1 List of branches for the survey Order I 10 11 12 Branch Northern Active Area No.1 Center Transaction Ha Noi Quang Trung Cau Giay East Ha Noi South Ha Noi Quang Ninh Southern West Quang Ninh Hung Yen Ha Tay Bac Ninh Son Tay Chi nhánh Khu vực Trọng điểm Phía Bắc Sở Giao dịch Hà Nội Quang Trung Cầu Giấy Đông Hà Nội Nam Hà Nội Quảng Ninh Tây Nam Quảng Ninh Hưng Yên Hà Tây Bắc Ninh Sơn Tây II 13 III 14 15 Song Hong River Delta Area Ninh Binh Northern Mountainous Area Phu Tho Thai Nguyen Khu vực Đồng Sơng Hồng Ninh Bình Khu vực Miền núi Phía Bắc Phú Thọ Thái Nguyên IV 16 17 Northern Central Area Phu Quy Quang Binh Khu vực Bắc Trung Bộ Phủ Quỳ Quảng Bình V 18 19 Southern Central Area Quang Ngai Ninh Thuan Khu vực Nam Trung Bộ Quảng Ngãi Ninh Thuận VI 20 21 Highland Area Gia Lai Đaklak Khu vực Tây Nguyên Gia Lai Đaklak VII 22 23 24 25 26 27 28 29 Southern Active Area No.2 Center Transaction North Sai Gon Gia Dinh Tien Giang Long An Phú Mỹ Ba Ria Tay Ninh Khu vực Trọng điểm Phía Nam Sở Giao dịch Bắc Sài Gòn Gia Định Tiền Giang Long An Phú Mỹ Bà Rịa Tây Ninh VIII 30 31 32 Cuu Long River Delta Area Vi Thanh Kien Giang Dong Thap Khu vực Đồng Sông Cửu Long Vị Thanh Kiên Giang Đồng Tháp 67 Table A.2 Information values for variables Variable Value Age a = 56 41 78 119 0.1038 0.0551 1.8830 0.0308 395 1,415 1,810 1 Female 110 572 682 0.2785 0.4042 0.6889 0.0469 Male 285 843 1,128 0.7215 0.5958 1.2111 0.0241 395 1,415 1,810 1 257 1,050 1,307 0.6506 0.7420 0.8768 0.0120 138 365 503 0.3494 0.2580 1.3544 0.0277 395 1,415 1,810 1 88 154 242 0.2228 0.1088 2.0470 0.0816 b High school 194 485 679 0.4911 0.3428 1.4329 0.0534 c Tertiary and above 113 776 889 0.2861 0.5484 0.5216 0.1707 395 1,415 1,810 1 a Non-high school Total a Rent 0.0710 0.0398 0.3057 50 96 146 0.1266 0.0678 1.8658 0.0366 107 306 413 0.2709 0.2163 1.2526 0.0123 c Home owner 238 1,013 1,251 0.6025 0.7159 0.8416 0.0195 395 1,415 1,810 1 59 442 501 0.1494 0.3124 0.4782 0.1203 b 85 470 555 0.2152 0.3322 0.6479 0.0508 c 140 395 535 0.3544 0.2792 1.2697 0.0180 d >= 111 108 219 0.2810 0.0763 3.6818 0.2668 395 1,415 1,810 1 11 0.0177 0.0028 6.2690 0.0273 b Office staff 85 605 690 0.2152 0.4276 0.5033 0.1458 c Skilled worker/ Manager 45 306 351 0.1139 0.2163 0.5268 0.0656 196 339 535 0.4962 0.2396 2.0712 0.1869 62 161 223 0.1570 0.1138 1.3795 0.0139 a Total a Unemployed/Retired d Self-employed/ Entrepreneur e Other Total 0.0685 0.4558 395 1,415 1,810 1 a = 21 0.4395 56 279 335 0.1418 0.1972 395 1,415 1,810 1 64 249 313 0.1620 0.1760 0.9207 0.0012 50 200 250 0.1266 0.1413 0.8956 0.0016 45 266 311 0.1139 0.1880 0.6060 0.0371 181 280 461 0.4582 0.1979 2.3157 0.2186 55 420 475 0.1392 0.2968 0.4691 0.1193 395 1,415 1,810 1 No 310 614 924 0.7848 0.4339 1.8086 0.2079 Yes 85 801 886 0.2152 0.5661 0.3801 0.3394 395 1,415 1,810 1 Total Discretionary a 30 Total Current account 0.1156 b Living with parents Total Work experience Information value 35 Total Occupation Odds 611 Single Number of dependents Non-default rate Total Housing Default rate 125 Marital status Married Education Total b 26 - 35 Total Gender Nondefault Total 68 0.1476 0.3778 0.5473 Table A.2 Information values for variables (continued) Default Nondefault Total Default rate Non-default rate Odds Information value a < 124 177 301 0.3139 0.1251 2.5096 0.1738 b - 173 331 504 0.4380 0.2339 1.8723 0.1280 c - 85 314 399 0.2152 0.2219 0.9697 0.0002 d - 10 355 365 0.0253 0.2509 0.1009 0.5173 e >= 238 241 0.0076 0.1682 0.0452 0.4975 Variable Value Length of relationship Total CIC 395 1,415 1,810 1 a No information available 78 413 491 0.1975 0.2919 0.6766 0.0369 b Bad debt history 74 23 97 0.1873 0.0163 11.5256 0.4182 c Good performance or no outstanding loan 243 979 1,222 0.6152 0.6919 0.8892 0.0090 Total 395 1,415 1,810 1 80 498 578 0.2025 0.3519 0.5755 0.0826 b 200 - 500 44 258 302 0.1114 0.1823 0.6109 0.0350 c 500 - 1000 124 293 417 0.3139 0.2071 1.5160 0.0445 d 1,000 - 2,000 128 319 447 0.3241 0.2254 1.4374 0.0358 1.4482 0.0055 Loan amount a 2,000 Total Loan purpose No (Business) Yes 1.3168 0.4641 19 47 66 0.0481 0.0332 395 1,415 1,810 1 221 1,115 1,336 0.5595 0.7880 0.7100 0.0782 2.0777 0.1671 0.2033 174 300 474 0.4405 0.2120 395 1,415 1,810 1 Loan duration a = 10 22 217 239 0.0557 0.1534 0.3632 0.0989 395 1,415 1,810 1 a 20% Total 0.2453 0.6153 395 1,415 1,810 1 Collateral type a No collateral 74 413 487 0.1873 0.2919 0.6419 0.0463 b Mobile asset 13 66 79 0.0329 0.0466 0.7056 0.0048 1.1788 0.0195 0.3064 c Real estate 308 936 1,244 0.7797 0.6615 Total 395 1,415 1,810 1 74 413 487 0.1873 0.2919 0.6419 0.0463 10 14 0.0101 0.0071 1.4329 0.0011 c 100% - 200% 227 580 807 0.5747 0.4099 1.4020 0.0557 d 200% - 350% 75 272 347 0.1899 0.1922 0.9878 0.0000 e > 350% 15 140 155 0.0380 0.0989 0.3838 0.0584 395 1,415 1,810 1 Collateral-to- a No collateral loan ratio b Partial collateral Total 69 0.0706 0.1615 Table A.3 Categorical variable coding Variable Value Occupation Education CIC Current account Loan purpose (Business) Parameter coding (1) (2) (3) (4) a Unemployed/Retired 0 0 b Office staff 0 c Skilled worker/Manager 0 d Self-employed/Entrepreneur 0 e Other 0 a Non-high school graduate 0 b High school graduate c Tertiary education and above a No information available 0 b Bad debt history c Good performance or no outstanding loan No Yes No Yes 70 Table A.4 Regression results for Models Trial and Default Model Trial Coefficient Odds (Std Err.) (Margins) Model Coefficient Odds (Std Err.) (Margins) -0.9738 (0.2899)*** -1.3556 (0.2954)*** 0.4534 (0.0844)*** 0.3777 (-0.0847) 0.2578 (-0.1135) 1.5737 (0.0339) -1.6729 (0.3867)*** -2.4957 (0.4079)*** 0.5903 (0.1081)*** 0.1877 (-0.0959) 0.0824 (-0.1352) 1.8045 (0.028) -2.1447 (1.2865)* -1.8327 (1.2876) -0.1077 (1.4071) -1.7278 (1.2955) -0.0374 (0.0123)*** -0.2128 (0.0177)*** -1.6033 (0.231)*** -0.5878 (0.0621)*** 0.1171 (-0.1853) 0.1600 (-0.1647) 0.8979 (-0.0116) 0.1777 (-0.1573) 0.9633 (-0.0028) 0.8083 (-0.0159) 0.2012 (-0.1197) 0.5555 (-0.0439) -3.8098 (2.0073)* -3.2526 (2.001) -0.8675 (2.147) -3.4960 (2.0127)* -0.0578 (0.0165)*** -0.4183 (0.0337)*** -2.7280 (0.3307)*** -0.7260 (0.0824)*** 0.0222 (-0.2224) 0.0387 (-0.1992) 0.4200 (-0.0627) 0.0303 (-0.2097) 0.9438 (-0.0027) 0.6581 (-0.0198) 0.0653 (-0.1292) 0.4838 (-0.0344) 3.6783 (0.4953)*** 0.4793 (0.2719)* 0.0040 (0.0003)*** -1.6363 (0.6844)** -0.0228 (0.0044)*** -5.7995 (3.9401) -0.0226 (0.0626) 5.9265 (1.5022)*** 39.5794 (0.3694) 1.6150 (0.0371) 1.0040 (0.0003) 0.1947 (-0.1222) 0.9774 (-0.0017) 0.0030 (-0.4331) 0.9777 (-0.0017) 374.8408 5.9087 (0.7017)*** 0.5655 (0.357) 0.0079 (0.0006)*** -2.7320 (0.9516)*** -0.0562 (0.007)*** -12.8952 (5.1616)** 0.0651 (0.0587) 11.9710 (2.351)*** 368.2220 (0.394) 1.7603 (0.028) 1.0079 (0.0004) 0.0651 (-0.1294) 0.9453 (-0.0027) 0.0000 (-0.6108) 1.0673 (0.0031) 158103.5744 Education Non-high school(R.C) High school Tertiary and above Dependents Occupation Unemployed/Retired(R.C) Office staff Skilled worker/Manager Self-employed/Entrepreneur Other Work experience Repayment sources (or Discretionary incomes) Current account Length of relationship CIC No information(R.C) Bad debt history Good performance/No loans Loan amount Purpose (Businesses) Duration Interest rate Collateral-to-loan ratio Intercept Degrees of freedom Chi-square Deviance (-2LL) Cox & Snell's R2 Nagelkerke's R2 H-L test, significance level 18 1016.387*** 882.886 0.430 0.661 0.000 18 1249.130*** 550.590 0.506 0.793 0.781 Note: (R.C) means reference category; *, ** and *** mean significance level at 10%, 5% and 1% respectively 71 Table A.5 Outlier cases in Model Trial Selected Observed Statusa Default Case Predicted Predicted Group Temporary Variable Resid ZResid 16 S 0** 8983 -.8983 -2.9727 55 57 S S 0** 0** 9499 9928 1 -.9499 -.9928 -4.3560 -11.7579 69 S 1** 1007 8993 2.9879 276 S 1** 0901 9099 3.1769 277 S 1** 1021 8979 2.9654 312 S 0** 9060 -.9060 -3.1043 337 S 1** 1026 8974 2.9568 338 S 1** 1296 8704 2.5917 399 S 1** 0187 9813 7.2432 417 S 1** 0240 9760 6.3797 512 S 0** 9642 -.9642 -5.1918 947 S 1** 0622 9378 3.8841 952 953 S S 1** 1** 1175 0608 0 8825 9392 2.7408 3.9292 974 S 1** 1223 8777 2.6783 988 S 1** 1072 8928 2.8859 1010 S 0** 9105 -.9105 -3.1895 1015 S 1** 0681 9319 3.6981 1064 S 1** 0495 9505 4.3842 1105 S 1** 0006 9994 42.2998 1126 S 1** 0377 9623 5.0539 1137 S 1** 0615 9385 3.9050 1149 S 1** 0644 9356 3.8121 1157 1167 S S 1** 1** 0037 0321 0 9963 9679 16.4042 5.4877 Note: a S = Selected, U = Unselected cases, and ** = Misclassified cases It can be seen from the above tables that the quality of Model is far better than that of the Model Trial due to the latter's excessive outliers Particularly, the Hosmer and Lemeshow's goodness-of-fit test for the Trial produces a χ2 value which is high with a significance outcome, suggesting we reject the null hypothesis that there is no difference between the observed and model-predicted values In short, the Trial does not fit well 72 ... MINH CITY VIETNAM INSTITUTE OF SOCIAL STUDIES THE HAGUE THE NETHERLANDS VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS DEFAULT PREDICTORS IN RETAIL BANKING – AN EMPIRICAL STUDY. .. to intense competition, over-lending and economic turmoil, banking system in Vietnam is suffering a huge amount of non-performing loans Given the considerable growth of retail banking market, an. .. wholesale banking activities cannot be enough for growth and prosperity, and thus not guarantee its leading role in Vietnam' s banking sector Consequently, recent changes and developments in the