ࡱ> 796'`RW-bjbj ;Z$\\\\@$@$@$8x$t$<\4%p%(%%%$'$'$'A\C\C\C\C\C\C\$V_hag\\$' '$'$'$'g\\\%%T|\((($'\%\%A\($'A\((2EW\\Z%(% Pb@$(jX.[d\0\Yb(.b\Zb\Z($'$'$'g\g\($'$'$'\$'$'$'$'$!!pD,D,,X\\\ MULTIVARIATE STATISTICAL APPROACH TO CREDIT RISK ASSESSMENT AND VALUATION FOR LOANS Professor Ante, ROZGA, Ph.D. University of Split, Faculty of Economics Matice hrvatske 31, 21000 Split, Croatia Tel: ++385 21 430649, fax: ++385 21 430701 E-mail:  HYPERLINK "mailto:rozga@efst.hr" rozga@efst.hr Ivica, KLINAC, Msc. Econ. Hypo-Alpe Adria Bank, Croatia Andrije Hebranga 3, 23000 Zadar Tel: ++385 21 023 230580 E-mail:  HYPERLINK "mailto:iklinac@yahoo.co.uk" iklinac@yahoo.co.uk Roberto, ERCEGOVAC, Ph.D. Splitska banka Societe Generale Group E-mail:  HYPERLINK "mailto:roberto.ercegovac@efst.hr" roberto.ercegovac@efst.hr Tel: ++385 91 219 1022 ABSTRACT Bad approach to credit risk exposure in the United States and some other countries has led to financial crisis and economic depression all over the world, yet without signs of recovery. Very loose conditions for banking loans produced exaggerated financial activity and therefore, as a consequence, impossibility to get loans paid back. In Croatia, banks used much stronger criteria when allowing loans with much better performances then in the U.S., which are the main reason that we do not have financial crisis as many other countries. Still, in this paper we want to get statistical multivariate approach to get it better. Authors analyzed the sample of retail portfolio of Croatian banks classified by the loan structure and borrower selected attributes, and wanted to challenge this classification by statistical methods. Multivariate statistical approach is used to get the most significant variables which contribute to classification of the loans as risky or not (default or not). There were several groups of applicants due to loan repayments, but we have made only two groups: those who repay their loans regularly and those who get some problems (delayed or not paying at all). We used following variables as predictors: economic development of the region where subject live, age of the subjects, percentage of the loans already repaid, interest rate and the currency in which the loans are tied (Croatian kuna, Euro, US dollar and Swiss franc. We have found that probability of bad loans is higher if the subject comes from less developed region, if hi (she) is older, if interest rate is higher and if the loan is tied to Croatian kuna. Logistic regression and discriminant analysis produced similar results. Classification results were satisfactory and showed some difference regarding bank risk assessment. Key words: credit risk, logistic regression, discriminant analysis. JEL Classification: G01, G32 Related field(s): 3 MULTIVARIATE STATISTICAL APPROACH TO CREDIT RISK ASSESSMENT AND VALUATION FOR LOANS VARIABLES DETERMINING CREDIT RISK ASSESSMENT Recent financial crisis mainly rose from financial problems inside banks and insurance industry. Risk management policy, standards and measures cracked in recent history. The value of the bank assets was riskier than modern financial models indicated, and banks requested additional capital to avoid liquidity and solvency problems. In Croatian bank industry conservative risk management policy and prudentional regulation measures restricted exposure to the volatile financial market of Croatian commercial banks. In empirical sample ratio of bed debt portfolio is less than 3%, thus showing conservative risk policy of Croatian banks. Even though, not all of them are the same, some have only temporarily problems when repaying bank loans. It is obviously that ratio of bad debt loans will increase in the near future due to worldwide financial crises and its transfer to Croatian financial system and risk management tools and strategies have to be improved. In this paper it is analyzed which variables of the retail loan portfolio inside Croatian banking system are dominant sources of credit risks and important in borrower credit capacity. Retail loans can be analyzed on portfolio level by grouping the loans according the related clients attributes. Retail loans are classified in groups as: cash loans, car loans, credit card loans and housing loans. The variables analyzed were: age in years and also categorized (young, middle-aged and older population), interest rates, the amount of loans, the rest of loans, region where clients live (south-north) and loans currency denomination. UNIVARIATE ANALYSIS OF CREDIT RISK IN CROATIA First, we took a look to particular variables mentioned in the paragraph 1 trying to find whether they are associated with default loans. We have made a cross tabulation for each variable and also Chi-square statistic to determine statistical significance for categorical variables. For age in years we employed Student t-test for independent samples (equal variances assumed). For default loans mean age was 45,46 years and for non-default was 44,53 which means that the older population tends to have defaults loans with the significance level of 0,042 (one-sided test). The difference in years does not have some practical importance, but significant result was due to rather big sample. As far as the interest rate is concerned the authors used same type of test and have found that default loans had average interest rate 10,22 and non-default loans 9,89. The difference was statistically significant at the level of 0,012. Interest rate is connected with some other variables so we need to work out multivariate analysis in later part of this paper. We have also tried to see if the difference in the amount of credit determines classification into good and bad loans. For default loans the average amount of credit was 22468 Euros and for non-default it was 24103 Euros. The difference was not significant (p = 0,423). Analyzing the rest of the loan we have not found any practical difference. Significance level was 0,438. Using Chi-square statistic we tested the association between default and non-default loans and the region where the clients belong. We divided Croatia in roughly two regions: north and south, assuming that north region of Croatia is more developed. In the south of Croatia there were 2,2% of bad loans while in the north region the percentage was only 0,2% (p < 0.001). Also, it has proved very interesting to see the situation with type of the loan. Cash and credit card loans were more risky (2,2% for both of them). For car loans the percentage of default loans was 1,1% and for housing loans 1,3%. The differences were significant (p = 0,033). The situation in Croatia is that almost all loans are tied with the exchange rate of foreign currencies. Only small percent of loans is in Croatian national currency. After financial crisis broke out there are no loans in kunas at all. The majority of our selected loans are tied to Euro (72,18%), the rest in Swiss francs (14,98%) and kunas (12,84%). We excluded the other currencies because their number was very small. Croatian tradition is to save money in German Marks (DEM), and after that, consequently in Euro. Significance level was 0,07. Loans in Swiss francs had default rate of 1,4%, in Croatian kuna 2,3% and in Euro 2,0%. MULTIVARIATE APPROACH TO CREDIT RISK IN CROATIA As the situation with credit risk assessment is rather complex and very serious issue, we used two multivariate methods: discriminant analysis and logistic regression. For both analyses we used same variables as in univariate analysis. Main theoretical assumptions for discriminant analysis are linearity and homoscedasticity. In discriminant analysis the main discriminant function is:  EMBED Equation.3  (1) Probability that particular case with disciminant score D belongs to group i is calculated using Bayes theorem:  EMBED Equation.3  (2) Using discriminant analysis classification results showed sensitivity of 82,4% using following variables: selected currency, type of the loan, region and interest rate. Logistic regression gives some additional results compared with discriminant analysis and becomes more and more popular. Model for logistic regression is:  EMBED Equation.3  (3) From the model above we can calculate odds ratios for particular predictor variables. We used backward elimination (Wald) and obtained the following results:  The reference category for type of the loan is cash loans and for the region is south region. Sensitivity was similar to the results of discriminant analysis. The only variables left after backward elimination were type of the loan and the region. Odds ratio for the region is the highest and this is really embarrassing to see such a big differences between north and south of Croatia. It means that the client from the south is about 15 times more risky to belong to default group then the client from the north. REFERENCES Dillon, W. R., and M. Goldstein (1984): Multivariate Analysis: Methods and Applications. New York: Wiley. Hair, J. F, R. E. Anderson, R. L. Tatham, and W. C. Black (1998): Multivariate Data Analysis. Fifth Edition. New Yersey: Prentice Hall. Hosmer, D. W., and S. Lemeshow (1989): Applied Logistic Regression. New York: Wiley. Johnson, R. A., and D. W. Wichern (2002): Applied Multivariate Statistical Analysis. Upper Saddle River: Prentice-Hall, Inc.     International Journal of Economic Policy Studies PAGE 2 PAGE 1 STU^_qr   * + - G pdpRdAdp:p hw{=h@ hw{=h@0JCJaJmHsH#jh*$h@CJUaJjh@CJUaJhw{=h@CJaJmHsHhh@;CJaJmH sH hh@CJaJmH sH h@CJaJmH sH h((nh@mH nHsH tHh@h((nh((nCJaJ%huhc)5\mH nHo(sH tH"hw{=hc)5CJ\aJmH sH "hw{=hw{=5CJ\aJmH sH TUr, - G e  ' { gd@$a$gdc)gdw{=$a$gd@gd((n$a$gdc)$a$gdw{=,,V-  / 0 ^ _ ` y z ݨp\@p*p+hw{=h0JCJ\aJmH nHsH tH6jhw{=hCJU\aJmH nHsH tH'hw{=hCJ\aJmH nHsH tH0jhw{=hCJU\aJmH nHsH tH'hw{=h@CJ\aJmH nHsH tHhw{=h@nHtH"hw{=hc)5\mHnHsHtH hw{=h@0JCJaJmHsH#jhw{=h@CJUaJ hw{=h@0JCJaJmHsH!jhw{=h@0JCJUaJz {     0 \]ȵ{ph`h{phhXhhXCJaJhWCJaJh({CJaJhz^h({CJaJh@CJaJhCJaJhz^h@CJaJhc)\mH nHo(sH tHhc)5\mH nHo(sH tH%huhc)5\mH nHo(sH tHhc)5\mH nHsH tH'hw{=hCJ\aJmH nHsH tH'hw{=h@CJ\aJmH nHsH tH   \]%&:;<=$a$gdj$a$gd- $ & Fa$gd-$a$gd-$a$gd({$a$gd@gd@ $ ha$gd@%&.6ɷl[G9)h@5CJ\aJnHo(tHhh@5CJ\aJ'hxh@CJ]aJmH nHsH tH!h@CJ]aJmH nHsH tH*hh@CJ]aJmH nHo(sH tH-hh@5CJ]aJmH nHo(sH tHh@6CJ]aJmH sH h@6CJ\aJmH sH "hch@6CJ\aJmH sH h@5CJ\aJmH sH "hh@5CJ\aJmH sH *hz^h@5CJ\aJmH nHsH tH 689:;< @b@>pswnnnnnnhCJ\aJh.CJ\aJhjCJ\aJh hjCJ\aJh"eh-5\ haa5\h-5CJ\aJhaa5CJ\aJmH sH h(x5CJ\aJmH sH hz^5CJ\aJmH sH h({hz^CJ\aJh@CJ\aJhyg.h@CJ\aJ+@A23NOWXD E [!\!# $h`ha$gd$a$gd&$a$gdJ& $ & Fa$gdaa$a$gd-$a$gdjs123HBMNOO[quònn]nL;Ln!h*QCJ]aJmH nHsH tH!h CJ]aJmH nHsH tH!hTCJ]aJmH nHsH tH!hd<CJ]aJmH nHsH tH!h.CJ]aJmH nHsH tH!h4@CJ]aJmH nHsH tH!hCJ]aJmH nHsH tH!hJ&CJ]aJmH nHsH tHhuqhJ&5CJ\aJh"ehuq5\ haa5\huqCJ\aJhj5CJ\aJhjCJ\aJ 1VW7f Z![!U"["\"e""ݻwfwffUfUDUDU!hU}CJ]aJmH nHsH tH!hfCJ]aJmH nHsH tH!hECJ]aJmH nHsH tH!h_jCJ]aJmH nHsH tH!hfCJ]aJmH nHsH tH!hLCJ]aJmH nHsH tH!hz^CJ]aJmH nHsH tH!hKjCJ]aJmH nHsH tH!h.CJ]aJmH nHsH tH!h4@CJ]aJmH nHsH tH!h@CJ]aJmH nHsH tH"### $ $F$G$R$q$$$$$U%m%%%%%%%̽wj]w]w]RK<j`ОI h6<CJUVaJ hxoh6<jhxoh6<Uh6<]mH nHsH tHhlY]mH nHsH tHh(x]mH nHsH tHh"e]mH nHsH tHhM]mH nHsH tHh"eh"e]mH nHsH tHh"e5]mH nHsH tHhn\5]mH nHsH tH!h"eCJ]aJmH nHsH tH!hfCJ]aJmH nHsH tH!hzNCJ]aJmH nHsH tH## $ $$$U%%%%%Y&Z&&&y'z'(( $h`ha$gd&T $h`ha$gdM $h`ha$gd& $h`ha$gd& $h`ha$gd"e$a$gd&$a$gd"e $ & Fa$gd"e $h`ha$gd%%%%%%4&6&P&W&X&Y&Z&&&&&&&&&&&&''x'۴ߚsfUD!hMCJ]aJmH nHsH tH!h(xCJ]aJmH nHsH tHh(x]mH nHsH tHh@s]mH nHsH tHh+ ]mH nHsH tHhM]mH nHsH tHh6<]mH nHsH tHjvh_Zh6<EHUjҞI h6<CJUVaJjh6<U h6<h6<h*Q h6<6h6<h&h jhxoh6<Ujhxoh6<EHUx'y''''((7(8(K(L(M(N(k(l(m((( ) )޼vrn]I'hOh4@CJ]aJmH nHsH tH!h&TCJ]aJmH nHsH tHh(ShOjhxoh(SEHU#jGeI hxoh(SCJUVaJ hxoh(Sjhxoh(SUh&!hOCJ]aJmH nHsH tH!h(SCJ]aJmH nHsH tH!hU}CJ]aJmH nHsH tH!h+ CJ]aJmH nHsH tHh(xhz^]mH nHsH tH(l(m((( )))o)q)++#+$+++,,m,n,gd&gd6: $h`ha$gd  $h`ha$gd 7$8$H$gdO 7$8$H$gd&T$a$gd& $h`ha$gd&T $h`ha$gd& )))))))C)I)M)Q)m)n)o)p)q)èq_q_qLq07hOh6<CJOJPJQJ\^JaJmHnHsHtH%hOCJPJ\aJmHnHsHtH#h*QhOCJPJ\aJnHtH+hOhOCJPJ\aJmHnHsHtHhOCJPJ\aJnHtH#hOhOCJPJ\aJnHtH4hO5CJOJPJQJ\^JaJmHnHsHtHCj h# hO5CJOJPJQJU\^JaJmHnHsHtH4h6<5CJOJPJQJ\^JaJmHnHsHtHq) *u***++++"+#+$++++,̻̪xiXD5Xh6:5]mH nHsH tH'h6:h6:CJ]aJmH nHsH tH!h6:CJ]aJmH nHsH tHh 5]mH nHsH tHh*Q5]mH nHsH tH"h h 5]mH nHsH tH!h*QCJ]aJmH nHsH tH!hlYCJ]aJmH nHsH tH!h*n1CJ]aJmH nHsH tH!h(SCJ]aJmH nHsH tH!h]CJ]aJmH nHsH tH!h&TCJ]aJmH nHsH tH,,m,,,,,,,,,,,,,-------̸vive]Y]Y]Y]YGY"hx{ h6CJ]aJmHsHhjhUhc)hc)mH nHo(sH tHhc)mH nHsH tH'hhc)CJ]aJmH nHsH tH!hlYCJ]aJmH nHsH tH!h*QCJ]aJmH nHsH tH'h h_jCJ]aJmH nHsH tH!h_jCJ]aJmH nHsH tH!h6:CJ]aJmH nHsH tH!h CJ]aJmH nHsH tHn,,,,,,,,,,,,,,,,,,-------gdc)$a$gdc) $h`ha$gd $h`ha$gdgd&-8-9-;-<-B-C-D-E-F-H-I-O-P-Q-R-S-U-V-W-ýòýýòýͮhc)mH nHsH tHhc)hw{=0JmHnHu h0Jjh0JUh1h<h6CJOJQJ]mHnHo(sHtH*h6CJOJQJ]mHnHo(tHu-9-:-;-F-G-H-S-T-U-V-W-$a$gdc)$hh]h`ha$gdc) hh]h`hgdc) &`gdc)$&dPa$gdc) $4$a$gdc) 901hP:pc). A!"#$% DyK rozga@efst.hryK Bmailto:rozga@efst.hryX;H,]ą'cDyK iklinac@yahoo.co.ukyK Nmailto:iklinac@yahoo.co.ukyX;H,]ą'cDyK roberto.ercegovac@efst.hryK Zmailto:roberto.ercegovac@efst.hryX;H,]ą'cDd hb  c $A? ?3"`?2~h3H?8I`!~h3H?8I@&|xcdd``ab``baV d,FYzP1n:&B@?b 5= UXRY7&meabM-VK-WMcرsA V0Z~[4 P.P16z(-ҙ] 㗋Bg>#UdBD%͋W@1.1I9 "\|O;qnO0 7/ȉ*Î*U~8l'?DfxNG $XxrC`pH  @F,<|eplPיA|=.h3n``s&#RpeqIj.?: @> 13Xtv2Dd  b  c $A? ?3"`?2| e|ͩͤX`!P e|ͩͤv@ ~ xڥTkA~]vޞUTbGD "&A&lg!iw]D[QD %(Λ[gb,y7 $~ ˋa "4BJDFMܑΕ2^Ɇ'섄 cF WZJ9e835y2xCb; ,1FQio#E"q?S1qdqU۝z}| O/*M^똳d](rXFzΝH%]ɿS[63ڬOKnzVeױ!~8 g?qPus%ߒ}]~n2_#vf3!Fu횙oV☭ju~@s:?1T|'}I[@yBQe!Ѿxn݁ED^XLX^`doaGDd b  c $A? ?3"`?2O{v^Ckzm`!eO{v^Ckz```/3xcdd``$d@9`,&FF(`T%ANRcgbR  PT obIFHeA*/&*de-b Yn wfRQ9B&br<K>b98!f22b`^( ҙ] !7#|&?Y-2!7AC` ;F, ~ &F'fU>ķ5AP&T _Os~T w6pbKL,HZŌ;\1Ӄ8%}Xʟ c_=,~9暈D-l C5| pJԀe@|C=X;Š~ z "'=2PEAFP'xQ\ A,W`g1@*y#RpeqIj.'] @ ]` "z(3ODd $ 0  # A2aWl_Y}{At3 `!laWl_Y}{AhY:xڵoEgni()JEh Bbxlj95 hZ@ !$Lu"۵]ɁR%.֒!q[^ iRقمHΛ{3oV8c$gYA%N z6ywNT'rE?pEaL;Fx/HQ1=­Vk\] R@ѣ#'N?[2^oH;)/pW^{8~GG?v'ZrtMv9";yDAeBlB[Q,;B<}<>x .5  0CKADVЇwZ6h'="o%hhx !'AC -AC4X ew*hh.hh=Ĥ5$; uAC4쁆ĂlW аG1-By aO:TCuC< ](C]P/%` lLPjHBwx L(D"YPB\2 YPCIP9  !"#$%&'()*+,-/012345A8;<=>?@CDBEFHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxRoot Entry F0 Pb:Data . WordDocument;ZObjectPoolqzPb0 Pb_1235144800 FqzPb|PbOle CompObjfObjInfo  !"$%&'()*+,-./0123457 FMicrosoft Equation 3.0 DS Equation Equation.39q D= 0 + 1 "X 1 + 2 "X 2 +.......+ k "X kEquation Native _1235145450 F|Pb|PbOle CompObj f FMicrosoft Equation 3.0 DS Equation Equation.39q P(G i /D)=P(D/G i )"P(G i )P(D/G i )"P(G i ) i=1k ObjInfo Equation Native  _1231395655F|PbXPbOle " FMicrosoft Equation 3.0 DS Equation Equation.39q lnP(A)1"P(A)()= 0 + 1 "X 1 + 2CompObjfObjInfoEquation Native 1TableGc "X 2 +....+ k "X kOh+'0 ,8H Xd    TMultivariate Statistical Approach to Credit Risk Assessment and Valuation for LoansMitsuo SasakiNormalrd`녇7\CRYQADIQ9ң"J y " "*>( "* R O$"-Mhɿ8oks;9CVMed5ߋs#tVSy[EaS]^ti8 [G:i5ollT~ 3%S}(9k+i@}[ʤde+E޽'bBK|qSlF&vz/J %![IJrLQV0ni9o#  '>z3pM1kϝUO6e0;I7[#]\ё<)jpn^UP95K_7dakzP2-D˪.mܖ]1Pt v@Nۣ]]z?wi0Kw!BA{[e&wܖշ&D)L}]ܓ7^]j#G/ýX]䐰-({﷍K˪)(ej6FN?Roz}wy#?lh&iCg挍˚[nbc+^W)|h̝ѻf:2IYut߀][ksv#oЫL'2$JX7rn8}9j\θ}]@%shNi 2v`=S80Uq wOaM)>d~lv_eSummaryInformation(DocumentSummaryInformation8#CompObj6qrozga47Microsoft Office Word@h %,@b@RL:1@J .bu՜.+,D՜.+,dx  DC$' TMultivariate Statistical Approach to Credit Risk Assessment and Valuation for Loans Title X2:F _PID_LINKBASE _PID_HLINKS_AdHocReviewCycleID_EmailSubject _AuthorEmail_AuthorEmailDisplayName_PreviousAdHocReviewCycleID _ReviewingToolsShownOnceAAH9Q!mailto:roberto.ercegovac@efst.hr1S2mailto:iklinac@yahoo.co.uk1xAmailto:rozga@efst.hr13 ekte gnd.can.aktan@yasar.edu.trCokun Can Aktanxv  FMicrosoft Office Word Document MSWordDocWord.Document.89q666666666vvvvvvvvv666666>6666666666666666666666666666666666666666666666666hH666666666666666666666666666666666666666666666666666666666666666666@@@ ^0NormalCJ_HaJmH sH tH H@H ^0 Heading 1$$@&a$5\mH sH B@B ^0 Heading 2$@&5\mH sH F@F ^0 Heading 3 $@&5\^JaJDA@D ^0Default Paragraph FontViV ^0 Table Normal :V 44 la (k@(^0No List @&@ ^0Footnote ReferenceH*6U@6 ^0 Hyperlink >*B*ph>> ^0 Footnote TextCJaJ.)@!. ^0 Page Number4 @24 ^0Footer  p#DB@BD ^0 Body Text$a$5CJmH sH 4@R4 ^0Header  p#TC@bT ^0Body Text Indent^`mH sH FV@qF ^0FollowedHyperlink >*B* ph,@ Table of Authorities,$  WDXDxYDx^`a$CJPJaJmHnHsHtHjOj kBa_l1k"$dx^`a$"5CJOJQJaJhmHsHtH0O0 @ eudoraheaderW%W%Z !"& "&!"&!"&= !W%TTUr,-Ge'{\ ]  % & : ; < =   @A23NOWXDE[\  UYZyz  l m !!!o!q!####$###$$m$n$$$$$$$$$$$$$%%9%;%F%H%S%X%n#n#n#n#sn#sn#sn#sn#sn#n#sn#sn#sn#sn#n#n#sn#sn#sn#n#n#n#n#sn#sn#sn#s n#sn#sn#sn#sn#sn#sn#sn#sn#n#n#n#n#sn#sn#n#sn#sn#sn#sn#sn#sn#sn#sn#sn#sn#n#sn#sn#sn#sn#sn#sn#sn#sn#sn#sn#sn#sn#sn#sn#sn#n#n#n#n#n#n#n#|n#n#n#n#~ n#n# n#sn#sn#sn#n#n#sn#sn#sn#Yn#n#sn#Yn#sn#sn#n#n#sn#n#sn#sn#sn#sn#sn#sn#sn#n#n#n#n#n#n#n#n#n#n#v:n#@9n#/Zyn#yn#TUr,-Ge'{\ ]  % & : ; < =   @A23NOWXDE[\  UYZyz  l m !!!o!q!####$###$$m$n$$$$$$$$$$$$$$$$$$%%%%%%%9%:%;%F%G%H%S%T%U%X%0000U0U0U0U0U00-0-0-0-0-00000000000000000000000000 00000000000 0000000000000000 00V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V0V000V0V0V0V0V0V0V0V000000000000000000000T,-Ge3 yz  !!!o!X%Z00Z00Z00R00Z00@0fnn#@0Z00 OZ00 Z0 0 جZ0 0Z0 00000000 AN[[[^ z 6s"%x' )q),-W- !"$%'()+ #(n,-W-#&*,V-*/_y7 K M W%XXX:::AHJNUW^!!`  C 7""7""@H 0(  0(  B S  ?W%*-:<?AFG   *.  + 7 $$$$X%+Gde tw$$X%:: $$$$X%$$X%F?+jf x, x8E Ws VPq6؁w{BjOr{D6t/!FZL4d$$ 3q ?2SMultivariate Statistical Approach to Credit Risk Assessment and Valuation for Loans Mitsuo Sasakirozga\