Do Foreign Bank Operations Provide a Stabilizing Influence in Korea?

Transcription

1 Do Foreign Bank Operaions Provide a Sabilizing Influence in Korea? Yongil Jeon Deparmen of Economics Cenral Michigan Universiy M Pleasan MI (989) FAX (989) yjeon@mail.cmich.edu Sephen M. Miller Deparmen of Economics Universiy of Nevada, Las Vegas 4505 Maryland Parkway, Box Las Vegas NV (702) FAX (702) sephen.miller@ccmail.nevada.edu Paul A. Nake Deparmen of Economics Cenral Michigan Universiy M Pleasan MI (989) FAX (989) nake1pa@mail.cmich.edu Absrac: This paper examines Korean daa ( ) o deermine if foreign banks behave differenly han domesic banks and if ha behavior provides a sabilizing influence on he Korean banking sysem and economy. Moreover, his paper also considers he effec, if any, of he Asian financial crisis on foreign and domesic bank behavior. Foreign banks financial raios differ from Korean banks wih wo noable excepions: provisions for loan losses and loan growh. Before he Asian financial crisis, all banks loans generally were unresponsive o Korean marke condiions. Once he crisis began, foreign banks reduce oal lending. Foreign banks increase and Korean banks decrease won-denominaed loans when Korean GDP growh increases and when Korean ineres raes increase. Finally, foreign banks lending is sensiive o changes in home-counry condiions. Key Words: Asian financial crisis, Korean commercial banks, sabilizaion, foreign bank operaions JEL Classificaion: E44, G21, O16, O53

2 1. Inroducion Wheher privae credi via foreign banks provides a source of sabiliy or volailiy in emerging markes is an imporan policy issue. Some analyss argue ha foreign credi o counries wih developing financial markes can provide deepening for capial markes, and diversify credi risk hereby reducing he incidence of financial crisis and he severiy of a crisis when i does occur. If foreign bank operaions are less sensiive o hos-marke condiions han he local banking firms, hen foreign banks can offer a source of sabiliy. The operaions of privae foreign banks, however, can increase he volailiy of financial markes in emerging counries. Goldberg (2001) concludes ha large U.S. banks claims on foreign markes are sensiive o U.S. cyclical condiions. In his case, he hos counry enjoys a more diversified supply of credi when foreign banks operae here, bu hese loans can flucuae wihou any change in he hos counry s macroeconomic condiions or governmen policy. 1 This sudy considers lending behavior of foreign and domesic banks during he Asian financial crisis o deermine wheher foreign bank operaions helped o sabilize financial markes in Korea. 2 We ask wheher flucuaions in bank lending reflec changes in economic condiions in boh Korea and he paren firm s counry. We apply basic porfolio heory o bank behavior. Foreign banks expand loans in whichever counry hey expec will yield he higher real rae of reurn adjused for risk. Therefore, he volume of bank loans in Korea depends on he real rae of reurn on invesmens in Korea versus he home-counry marke of he foreign bank. These expeced reurns are funcions of he real ineres raes and real GDP growh raes in Korea versus he home counry. 1 More deailed discussions of he pros and cons of foreign bank operaions wihin a domesic economy occur in Berger, DeYoung, Genay, and Udell (2001), Claessens, Demirgüç-Kun, and Huizinga (2001), Dages, Goldberg, and Kinney (2000), Demirgüç-Kun, Levin, and Min (1998), Jeon and Miller (2001) and Levine (1996). 2

3 2. The Korean Banking Secor and he Asian Financial Crisis The Korean banking sysem includes hree classes of insiuions naionwide, regional, and foreign banks. Naionwide banks were designed o help he cenral governmen o mainain conrol of he financial sysem and arge lending o favored secors of he economy. Regional banks exis in Korea o encourage regional economic developmen and hey can only conduc business in heir own provinces and operae a branch office in Seoul. The firs foreign bank enered he Korean marke in Foreign banks were araced o Korea because hey received more favorable reamen han Korean banks in some aciviies. Deregulaion in he 1980s reduced his preferenial reamen, bu oher resricions on foreign bank operaions were reduced, hus creaing a more level playing field for Korean and foreign banks. Evens in he 1980s fundamenally changed he Korean banking indusry: ownership of he naionwide banks was ransferred from he public o he privae secor, new privae banks were creaed, and deregulaion expanded he range of services offered by privae commercial banks. The governmen, however, remained a poen force because i sill conrolled ineres raes on cerain ypes of loans and deposis and channeled loans o favored indusries and firms in accordance o is plan for economic developmen. 3 Alhough Korea experienced relaively high economic growh and low inflaion in he early 1990s, some weaknesses exised in he financial secor: low inernaional reserves, and poor governmen regulaion and supervision of he banking sysem. When several chaebols (e.g. Kia, Hanbo, Haiai, and Sammi) wen bankrup and oher chaebols defauled on syndicaed loans, foreign 2 Discussions of he Asian financial crisis appear in Io and Krueger (2001) and Peseni and Tille (2000). 3 Bank of Korea (1994, 1998), Gilber and Wilson (1998), and Jeon and Miller (2001) provide more deailed informaion. 3

4 lenders began o reevaluae heir posiions in Korea. Foreign sources of funds quickly decreased. As financial condiions in banking secor deerioraed, he Korean governmen began searching for poenial merger parners. Did foreign bank operaions promoe sabiliy or volailiy during his crisis? Supporers of foreign bank enry argue ha hese banks provide an imporan channel for foreign capial inflows o finance domesic aciviies. If hese foreign funds complemen raher han subsiue for domesic sources of funds, hen a ne expansion of available funds ha suppors higher economic growh occurs. Foreign banks can also increase compeiion in domesic markes ha can improve he efficiency of domesic bank operaions, lower he cos of providing financial services, reduce ineres raes charged on loans, and increase he ineres paid on deposis hereby simulaing domesic saving and invesmen. Foreign banks can also promoe improvemens in governmen regulaion and supervision of he financial sysem by imporing business pracices forged by more sringen home counry regulaions. Opponens of foreign bank enry sugges ha he presence of foreign banks increases he volailiy of domesic financial markes. The foreign capial channel provided by foreign banks no only encourages an inflow of capial in good imes bu also expedies a rapid ouflow of foreign capial when a financial crisis occurs. In addiion, if foreign banks appear more sable han domesic insiuions, hey may arac he bes domesic borrowers (higher-profi and lower-risk borrowers), puing domesic banks in he more precarious posiion of lending o less credi-worhy borrowers. Also, foreign banks can inroduce unfamiliar business pracices ha domesic regulaors may find difficul o evaluae and supervise. Raher han improving he regulaory or supervisory process, foreign banks can creae complex problems for he domesic governmen. 4

5 The Korean governmen s inervenion ino banking markes, however, did no limi iself o supervision and regulaion. The governmen played a prominen role in allocaing credi o prioriy indusries as par of is economic developmen sraegy. Banks graned policy loans o favored firms because he governmen direced hem o do so. In some cases, he credi exended possessed lile connecion o he financial viabiliy of he underwrien projecs and conribued o loan qualiy problems experienced by Korean banks during he financial crisis. Barh, Caprio, and Levine (2002) provide an ineresing wis o he exiing lieraure. To wi, heir paper, a par of a much larger research agenda using a new cross-counry daabase on bank regulaion and supervision, finds ha banking crises posiively correlae wih limiaions on foreign bank enry ino domesic banking markes. Thus, merely reducing such limiaions and easing he abiliy of foreign banks o ener he domesic banking marke reduces he incidence of banking crises, even if foreign banks do no ener. In sum, heir finding suggess ha poenial enry of foreign banks proves saluary on he sabiliy of he domesic banking marke. 3. Comparing Foreign, Naionwide, and Regional Banks: Financial Raios Annual balance shee and income saemen daa for Korean and foreign banks come from Bank Managemen Saisics published by Financial Supervisory Services in 1999 and The daa comprise 61 foreign banks represening 17 counries, 16 Korean naionwide banks, and 10 Korean regional banks from 1994 o The breakdown of he sample by year and ype appears in Table 1 4 One cavea is ha he banking daa encompasses he usual limiaions encounered in annual balance shee accouns: 1) he moneary measures are aggregaed over he course of one fiscal year and his may be oo long o pick up significan variaions; 2) fiscal year-end figures may no be represenaive due o various window dressing echniques; and 3) firms end heir fiscal years a differen poins during he calendar year. In addiion, supervision by he Korean governmen can affec measures of banking performance. In Korea, loan losses ge charged agains earnings only when he bank and he governmen regulaors are ready o recognize he losses. Regulaors may be relucan o reclassify acive loans ino he defaul caegory: reclassificaions may damage fragile public confidence in financial markes or discourage poenial merger parners for a roubled bank. Therefore, some banks profis and loan porfolios may be less sensiive o changing 5

6 as do he counries of origin for he foreign banks. All moneary figures are denominaed in real won wih 2001 as he base year. Some banks ener and/or exi he indusry over he sample period so an unbalanced panel daa se emerges wih 558 observaions (a panel of 696 observaions wih 138 missing values). In addiion o he banking firms daa, we collec measures of macroeconomic and financial marke condiions for boh Korea and he foreign banks home counries. As a proxy measure of he demand for loans in each counry, we calculae he growh rae of real Gross Domesic Produc (annual percenage rae in consan 1995 U.S. dollars) using daa available from World Bank Developmen Indicaors. Recen financial crises promped researchers o invesigae he role of foreign banks in domesic economies. Crysal, Dages, and Goldberg (2002) compare he performance of foreign and domesic banks in Lain America over he period. They employ wo approaches credi raings and financial raios. This secion uses financial raios o compare and conras he performance of foreign and domesic banks in Korea over he 1994 o 2001 period, a period ha includes he Asian financial crisis. Moreover, we also evaluae he performance of wo classes of domesic banks naionwide and regional banks. Table 1 records he number of banks in each caegory over he sample period. The number of foreign banks sayed in he neighborhood of 50 hrough 1998, dropping o 42 in The number of naionwide Korean banks rose from 14 in 1994 o 16 in 1997 and fell o 9 as a resul of he Asian financial crisis. Similarly, regional Korean banks held seady a 10 unil he Asian financial crisis and hen dropped o 6 in 1999, where i remained hrough Table 2 repors he average values for various financial raios loans o asses (la), liquid marke condiions during periods of financial crisis because of hese regulaory pracices. 6

7 asses 5 o asses (lqaa), he rae of growh of loans (lgr), loans o deposis (ld), deposis o asses (da), equiy o asses (ea), provisions for loan losses o loans (plll), reurn on asses (roa), reurn on equiy (roe), foreign currency loans o loans (fll), foreign currency deposis o deposis (fdd), and foreign currency loans o foreign currency deposis (flfd). The firs nine financial raios focus on he general characerisics of he banks balance shees and income saemens. Crysal, Dages, and Goldberg (2002) employed similar variables in heir sudy of Lain American banks. The remaining hree financial raios focus on he role of foreign currency loans and foreign currency deposis in he balance shee of he banks. The las hree raios deserve special discussion because alhough he loans and deposis ener he balance shee in won, hey acually represen foreign currency values. Thus, he won value changes in response o depreciaions and appreciaions in he won. If, on average, he won appreciaes (depreciaes), hen he raios of foreign currency loans o loans and foreign currency deposis o deposis decrease (increase). Tha is, changes in he won s value can lead o changes in a bank s equiy posiion. If a bank holds more (less) foreign currency loans han foreign currency deposis, hen an appreciaing won causes foreign currency loans o fall more (less) han foreign currency deposis, reducing (increasing) he bank s equiy. So, if he raio of foreign currency loans o foreign currency deposis equals one, hen changes in he value of he won do no affec he banks equiy posiion. When he foreign currency loan o foreign currency deposi raio exceeds (falls below) one, hen an appreciaing won hurs (improves) bank equiy while a depreciaing won 5 We define liquid asses o equal oal asses minus oal loans. Thus, he raios for liquid asses o asses and loans o asses auomaically sum o one. 7

8 improves (hurs) bank equiy. 6 The various raios in Table 2 are average values across he foreign, naionwide, and regional banks for each year 1994 o 2001 and overall averages across 1994 o We also perform -ess o see if he averages differ significanly from each oher (no shown in he paper). We adop he 5- percen level of significance in our discussion ha follows. We firs compare he financial raios for naionwide and regional Korean banks. Tha is, we consider in wha ways hey mirror each oher and in wha ways hey differ from each oher. In a number of areas, naionwide and regional banks possess similar financial raios. Where hey do differ, he regional banks generally ake, on average, a more conservaive approach o heir operaion. For example, deposis o asses (da), foreign currency loans o oal loans (fll), and foreign currency deposis o oal deposis (fdd) differ significanly beween naionwide and regional banks. Regional banks show higher deposis o asses (da), bu lower foreign currency loans o oal loans (fll) and foreign currency deposis o oal deposis (fdd). Furhermore, loans o deposis (ld), excep from 1998 o 2001, and foreign currency loans o foreign currency deposis (flfd), excep in 1999 and 2000, also differ significanly. Regional banks show lower raios for boh of hese financial raios. Bu significan differences beween hose raios disappear afer he Asian financial crisis. The remaining financial raios generally do no differ significanly beween naionwide and regional banks. For example, reurn on asses and reurn on equiy only differ significanly beween naionwide and regional banks for he reurn on asses in Also provisions for loan losses o loans (plll) only differ significanly in 1994 and Liquid asses o asses (lqaa) and loans o 6 Table 2 illusraes ha, on average, Korean regional banks exhibi a raio less han one, indicaing ha a depreciaing (appreciaing) won will reduce (increase) equiy. Korean naionwide banks exhibi, on average, a raio greaer han one, indicaing ha an appreciaing (depreciaing) won reduces (increases) bank equiy. Foreign banks, on average, exhibi a 8

9 asses, mirror images of each oher, differ significanly from each oher in Furher, equiy o asses differs significanly only in 1995, 1996, and Finally, he loan growh rae differs significanly in 1997, 1999 and In sum, Korean naionwide and regional banks exhibi similar characerisics in a number of financial raios. When hey differ, moreover, he regional banks exhibi more conservaive, pruden operaion. The foregoing discussion merely ses he sage for he discussion of he similariies and differences beween he financial raios of foreign and domesic banks in Korea. Foreign banks diverge significanly from domesic Korean banks in mos of he financial raios. 7 Foreign banks differ significanly from domesic banks in every year for he following financial raios: loans o asses (la), liquid asses o asses (lqaa), deposis o asses (da), equiy o asses (ea), and foreign currency loans o loans (fll). Foreign banks also vary from domesic banks in he following raios: loans o deposis excep in 2000 and 2001, reurn on asses excep in 2001, and foreign currency loans o foreign currency deposis excep in Foreign banks fund a much smaller porion of heir asses wih deposis han domesic banks. Moreover, foreign banks fund a much larger fracion of heir asses wih equiy (i.e. a much higher equiy o asse raio han domesic banks). Noneheless, foreign banks use sources of funding oher han deposis and equiy more inensively han domesic banks. Since foreign banks have a smaller raio of deposis o asses, hey also exhibi a higher raio of loans o deposis. Foreign banks also show a higher (lower) reliance on liquid asses (loans) o asses han domesic banks. Finally, foreign banks exhibi more willingness o raio much larger han one, indicaing ha a depreciaing (appreciaing) won increases (decreases) bank equiy. 7 The discussion in he ex generally compares he foreign raio o he raio of he domesic Korean banks naionwide and regional. We do discuss differences beween foreign banks and naionwide or regional banks where ha informaion provides some addiional insigh. Generally speaking, comparing he foreign banks o he domesic banks proves sufficien o our ask. 9

10 use foreign currency loans and deposis, possessing higher raios of foreign currency loans o loans (fll) and foreign currency loans o foreign currency deposis (flfd). The foreign currency deposi o deposis raio differs significanly beween foreign and domesic banks in 1995 and 1998 o Over he sample period, foreign banks exhibi a higher reurn on asses and equiy han domesic banks. Tha is, foreign banks earned a higher rae of reurn han domesic banks before, during, and immediaely afer he Asian financial crisis. Bu a couple of financial raios do no differ significanly beween foreign and domesic banks. The provisions for loan losses o loans and he growh rae of loans do no differ significanly beween foreign and domesic banks, excep for he growh rae of loans in So far, we compare he differences in means beween foreign, Korean naionwide, and Korean regional banks on a variable-by-variable basis. Cluser analysis allows us o examine differences in a mulivariae framework. Moreover, he grouping of banks emanaing from he cluser need no resric groups o only Koran naionwide banks, Korean regional banks, or foreign banks. 9 Raher, he cluser procedure will group banks irrespecive of heir legal designaion based enirely on he balance shee variables included in he analysis. We implemen a hierarchical agglomeraive weighed-average linkage cluser analysis using he Euclidean disance measure. 10 We apply he cluser analysis for 1994 and 1996 wo years ha precede he Asian financial crisis and 1998 and 2000 wo years ha follow ha crisis. We canno include all balance shee variables repored in Table 3, since ha will inroduce 8 Table 2 repors ha he provisions for loan losses o loans for foreign banks average in Tha observaion incorporaes an oulier, Credi Suisse Firs Boson, which enered he marke in This bank began operaions wih a provisions for loan losses raio of 20. Eliminaing ha observaion lowers he average o Ursacki and Verinsky (1991) perform a cluser analysis on foreign banks in Korea during he mid-1980s. 10 All saisical analyses use STATA 8. 10

11 several singulariies. Moreover, we do no wan he groups o form based on some elemens of he banks balance shee ha do no represen imporan balance shee informaion. Thus, we resric he cluser analysis o hree imporan balance shee variables loans o asses, deposis o asses, and equiy o asses. Neverheless, we observe good clusering wih he use of hese hree balance shee raios. We employ he Caliński and Harabasz (1974) pseudo-f and he Duda and Har (1973) psuedo--squared values o deermine he number of groups. Those wo saisics generally, bu no always, give he same answer for he number of groups. We examine 7, 6, 6, and 8 groups for 1994, 1996, 1998, and 2000, respecively. The cluser analysis provides one ineresing conclusion for our purposes. In each year excep one, one of he clusers involved only Korean banks naionwide and regional. The only excepion occurred in 1998 when Ciibank appears in he group wih all he Korean banks. Oher groups in each year included differen combinaions of he foreign banks, bu never included Korean banks. 11 In general, we observe many more differences beween foreign and domesic banks in Korea over our sample period han beween naionwide and regional domesic banks. Wha explains he difference in performance beween foreign and domesic banks? We now urn o address his quesion wih our regression analysis in he nex secion. 4. The Empirical Model While a review of he sample daa reveals differences among Korean and foreign bank operaions, i does no address how lending was affeced by changes in financial and macroeconomic condiions. In his secion, economeric analysis deermines he sensiiviy of bank lending o marke condiions 11 Ciibank enered Korea in 1967 wih corporae finance services and opened is firs branch office in Pusan in I began consumer finance services and enered privae banking services for he firs ime in Korea in I currenly operaes 12 branch offices in major ciies including Seoul and Pusan. Finally, i was seleced as he bes foreign bank in 11

12 and wheher differences exis in behavior across differen ownership groups -- Korean naionwide, Korean regional, and foreign banks -- as well as across he pre- and pos-asian financial crisis. The heoreical foundaion for he regression model employs basic porfolio heory. Lending by foreign banks in Korea depends on he expeced real rae of reurn on invesmens in Korea relaive o he expeced real rae of reurn in he home counry of he paren bank. The expeced rae of reurn in a counry depends, we assume, on he real rae of ineres and he real GDP growh rae. A higher real ineres rae increases he expeced reurn, holding risk consan, and higher GDP growh decreases he risk of defaul and hereby increases expeced reurns, given he real ineres rae. The basic model for foreign banks is given as follows: Loans i = α 0 + α 2 r K + α 2 r H + α 3 dlny K + α 4 dlny H + A i + ε, (1) where Loans i equal he quaniy of loans, eiher oal loans, won loans, or foreign currency loans, for bank i in period ; r K and r H equal he Korean and home-counry ineres rae spread over he LIBOR (hereafer, spread); dlny K and dlny H equal he Korean and home-counry growh rae of real GDP (where d represens a differencing operaor); and A i equals he size (measured in oal asses) of bank i in period. 12 The consan erm is allowed o vary across bank and ime period. For Korean banks, he model is given as follows: Loans i = β 0 + β 1 r K + β 2 dlny K + A i + ε. (2) Tables 3 o 6 presen he esimaed equaions using oal loans, won loans, and foreign currency loans as he dependen variables. Moreover, we also inroduce an Asian financial crisis Korea in Since he mached pairs of Korean ineres rae o LIBOR and home-counry ineres rae o LIBOR each include he expeced inflaion premium, he spreads reflec boh nominal and real ineres rae differences of he Korean and homecounry ineres raes over LIBOR. 12

13 dummy variable, aking a value of zero from 1994 o 1996 and a value of one from 1997 o The examinaion of he daa s mean values (see Table 2) suggess subsanial differences in behavior across Korean naionwide banks, Korean regional banks, and foreign banks as well as across he preand pos-asian financial crisis. Bank-specific characerisics (e.g. managemen philosophy, foreign vs. domesic ownership, U.S. vs. Japanese ownership, locaion of branch offices) affec bank-lending behavior. Under hese condiions, ordinary leas squares esimaion may yield biased and inconsisen coefficien esimaes. We employ he fixed- and random-effecs model wih he bank panel daa o produce unbiased and consisen parameer esimaes. For each regression, we run fixed- and random-effecs specificaions and perform a Hausman es o selec he mos-appropriae specificaion. The resuls discussed below use he chosen specificaion fixed or random effecs. 5. The Empirical Resuls Foreign Banks Table 3 reveals ha foreign banks oal loan aciviy significanly responds o he scale variable (oal asses) used in he model, bu nohing else, prior o he Asian financial crisis. This behavior changes once he crisis begins. Toal lending decreases overall, given he significanly negaive crisis dummy variable. Foreign banks lend more in Korea as he rae of income growh rises in Korea and he home counry, as he Korean ineres rae premium increases, and he home-counry ineres rae premium decreases. The response o ineres rae changes and he Korean growh rae adhere o he predicions of porfolio allocaion heory, bu he home-counry growh rae does no. Rising income growh in he home counry may influence foreign bank oal lending in wo conrary ways: 1) decreasing he risk of repaymen in he home counry causes foreign banks o lend more in he home counries and less in Korea; and 2) increasing deposis in he home counry may increase 13

14 he supply of home-counry currency o a bank s foreign subsidiaries in Korea. Some suppor emerges for he supply argumen in he foreign currency loan regressions. Before he crisis, foreign banks increase foreign currency lending in Korea, when home-counry income growh increases, bu afer he crisis home-counry growh does no influence lending. Perhaps he greaer expeced risk of lending in Korea causes foreign banks o shif loans away from Korea and oward less risky areas. Foreign banks won loans before he crisis significanly reac o income growh in he home counry (negaive) and o oal asses (posiive). During and afer he crisis, home-counry growh does no significanly affec won loan aciviy. Insead, won loans reac o changes in he Korean markes: increases in Korean income growh and Korean ineres raes associae wih an expansion of won loans by foreign banks. This suggess ha foreign banks acquire won deposis and issue won loans o service he local needs of heir cusomers and hese loans respond o Korean marke condiions. Foreign currency loans by foreign banks prior o he Asian crisis posiively correlae wih income growh in he home counry and negaively correlae wih he ineres rae in he home counry and he Korean rae of income growh. The Korean ineres rae and oal asses are insignifican. Foreign banks change behavior afer he crisis. They issue fewer loans, holding all oher variables consan (a negaive adjusmen o he consan). Foreign currency loans now respond posiively o Korean marke condiions (boh income growh and ineres rae variables) and negaively o he ineres rae premium in he home counry. The home-counry income growh variable is now insignifican. Pos-crisis behavior for foreign currency loans follows quie closely he predicions of porfolio heory: a higher expeced risk-adjused rae of reurn in Korea and a lower expeced rae of reurn in he home counry drive an increase in foreign currency lending by 14

15 subsidiaries in Korea. The Chow ess indicae ha across all hree loan caegories foreign banks lending behavior changes once he Asian financial crisis sars. In general, foreign banks loan less, holding all oher variables consan, and loan volume reacs more o home-counry condiions. Korean Banks In Table 4, Korean banks (boh naionwide and regional) display some of he same lending paerns ha foreign banks show. Like foreign banks, oal loans significanly respond only o oal asses prior o he Asian crisis. Korean banks lending, however, proves more sensiive o changes in oal asses, since he coefficiens for oal asses are much larger for Korean banks (e.g versus 0.22). Afer he crisis begins, Korean banks oal loans increase when Korean income growh and Korean ineres raes increase. Unlike foreign banks, no overall reducion in oal lending aciviy occurs (an insignifican dummy variable for he consan erm). Looking a oal loans, however, hides some imporan differences in Korean banks lending aciviy. Won loans, like oal loans, do no reac o changes in Korean marke condiions prior o he Asian crisis. Korean banks, however, increase oal won lending (posiive dummy variable for he consan erm) afer he crisis begins and won lending increases in response o decreases in he domesic income growh rae and he domesic ineres rae. These resuls sugges ha Korean banks provide a couner cyclical influence during and afer he financial crisis: hey provide more loans when domesic economic condiions worsen and fewer loans when economic condiions improve. Informaion problems in loan markes can explain he negaive effec of he Korean ineres rae on won lending. As ineres raes rise, paricularly during a financial crisis, lenders recognize an adverse selecion problem among heir loan applicans: hose mos likely o borrow a high real raes 15

16 are also he mos likely o defaul. Thus, lending aciviy falls as he ineres rae rises. Foreign currency loans, however, respond differenly a Korean banks han a foreign banks o he same simuli. Like he oher loan caegories, only oal asses influence foreign currency loans prior o he crisis. Afer he crisis and conrary o won loans, foreign currency loans increase wih increases in Korean income growh and ineres raes, which mach he predicions of porfolio heory. Also in conras o he experience of won loans, foreign currency lending decreases during he financial crisis, holding all oher influences consan (negaive dummy variable for he consan erm). Korean banks may face difficuly acquiring foreign currency for lending purposes as financial capial leave Korea and he won s value falls relaive o widely-raded inernaional currencies. The Chow ess indicae ha Korean banks lending behavior for oal loans and foreign currency loans changes once he Asian financial crisis sars, bu no difference exiss for won loans. Regional Versus Naionwide Banks Tables 5 and 6 presen regression resuls for Korean naionwide and regional banks. Before he crisis, oal loans respond o oal asses for boh naionwide and regional banks, alhough he coefficien for naionwide banks is wice as large. In addiion, regional banks oal lending significanly reacs o Korean income growh (negaive) and he Korean ineres rae (negaive). Afer he crisis sars, naionwide banks increase lending when Korean income growh increases and Korean ineres raes rise, while regional banks reduce oal loans in response o he same marke condiions. Before he crisis, naionwide and regional banks won loans respond only o oal asses. Afer he crisis begins, naionwide banks increase won lending, holding all oher variables consan, while regional banks do no. Boh naionwide and regional banks issue fewer won loans as domesic 16

17 income growh and ineres raes increase afer he crisis begins, indicaing ha boh ypes of banks play a role in sabilizing he Korean economy. Naionwide and regional banks also exhibi similar behavior wih regard o foreign currency loans: each ype of bank reduces oal lending afer he crisis begins, alhough naionwide banks reduce overall lending by a much larger amoun (coefficiens of versus 1,661). Each ype of domesic bank also increases foreign currency lending in response o higher domesic income growh and ineres raes. The Chow ess indicae ha Korean banks lending behavior for oal loans and foreign currency loans changes once he Asian financial crisis sars, bu no difference exiss for won loans. In general, naionwide banks gran more won loans and fewer foreign currency loans, holding all oher variables consan, while regional banks simply reduce foreign currency loan volume. Comparison wih Goldberg s Resuls Goldberg s (2001) sudy of foreign lending by U.S. banks offers a useful comparison. In general, she finds ha local marke condiions (i.e., he ineres rae and GDP growh) produce lile effec on bank lending worldwide; bu when limiing her analysis o lending in Asian counries, she discovers ha local GDP growh generaes a negaive effec on lending. Our sudy also finds ha foreign bank lending (foreign currency loans) decreases when an increase in Korean income growh occurs before he Asian financial crisis. Afer he crisis, however, foreign banks lend more in Korea across all hree loan caegories, as he Korean income growh rae increases. The Korean ineres rae spread does no significanly influence foreign banks foreign currency lending before he crisis, bu generaes a posiive influence aferward. Goldberg finds no significan effec for he local ineres rae in he Asian sample. Goldberg s resuls for oal lending by U.S. banks operaing in Asia reveal ha he home- 17

18 counry ineres rae and GDP growh rae possess a significan and negaive effec on lending. Our resuls for foreign currency lending by foreign banks confirm Goldberg s findings for he homecounry ineres rae spread, which is significanly negaive boh before and afer he Asian financial crisis. We ge differing resuls for he home-counry income growh rae: i exers a posiive influence before he Asian financial crisis and no effec afer he crisis. Using oal lending by foreign banks as he dependen variable produces resuls for Korea fundamenally differen from Goldberg s findings for Asia. Our home-counry spread possesses a significanly negaive effec afer he Asian financial crisis, bu an insignifican effec before (Goldberg finds a significanly negaive effec). The home-counry income growh possesses a significanly posiive effec afer he Asian financial crisis, bu an insignifican effec before (Goldberg finds a significanly negaive effec). Korean income growh possesses a significanly posiive effec afer he Asian financial crisis, bu an insignifican effec before (Goldberg finds an insignifican effec). Finally, he Korean spread possesses a significanly posiive effec afer he Asian financial crisis, bu an insignifican effec before (Goldberg finds an insignifican effec) Conclusions An analysis of financial raios suggess ha foreign bank operaions differ significanly from Korean-owned banks. Foreign banks rely more heavily on non-deposi sources of funds, are more likely o make foreign currency loans, and earn higher reurns on asses and equiy. Higher raes of reurn may sugges ha foreign banks operae more efficienly han domesic banks, reflecing 13 When Goldberg limis he sample o large banks only (oal asses greaer han $250 million), no independen variables prove significan for he Asian counry regressions. Readers should use cauion when comparing hese resuls because he sudies sar wih differen daa. Goldberg s sample only includes U.S. banks, nearly half of hese are small banks (oal asses below $50 million), and Korean marke lending was excluded from he daa used o esimae he Asian marke regression equaions. Our sudy includes banks from 17 foreign counries, mos are large banks in heir home counry, and we examine only lending in Korean markes. 18

19 lower-cos operaions or improved invesmen and lending pracices ha improve earnings. Korean banks, facing relaively new bank regulaions and less governmen direcion in lending, experience more difficuly prospering in he liberalized financial sysem. Some evidence emerges ha foreign banks provide sabiliy o he Korean banking sysem. Compared o Korean banks, foreign banks hold more equiy and liquid asses, indicaing ha foreign banks possess a lower probabiliy o fail when Korean economic and financial marke condiions worsen. Foreign banks also rely more heavily on foreign currency deposis as a source of funds. This allows foreign banks o expand loans when Korean condiions worsen, hereby providing a sabilizing influence in loan markes. Alernaive evidence indicaes, however, ha foreign banks do no use hese exernal sources of funds o expand loans, since heir loan growh equals ha of Korean banks. Foreign banks do no differ from Korean banks in anoher imporan measure of sabiliy: provisions for loan losses. Do foreign banks sabilize he Korean economy? Our regression resuls indicae ha foreign banks reduce oal lending once he financial crisis begins, holding all oher variables consan. This reducion in oal lending reflecs a decrease in foreign currency loans, since he consan erm for won-denominaed loans does no change. A comparison of consan erms shows ha Korean banks do no reduce oal loans in he pos-crisis period, bu Korean banks do change he composiion of heir loan porfolio: increasing won-denominaed loans while reducing foreign currency loans. The Korean ineres rae spread does no significanly affec any caegory of loan for any ype of bank prior o he Asian financial crisis. In addiion, he Korean income growh rae generally does no influence bank lending (income growh is significan in only wo of welve loan equaions). Thus, before he crisis, all banks generally do no respond o Korean economic condiions. 19

20 A differen behavioral paern emerges afer he financial crisis: boh he Korean ineres rae and growh rae significanly influence won-denominaed lending. In addiion, changing Korean economic condiions elici differing responses from domesic and foreign banks. When Korean income growh rises, foreign banks increase won-denominaed loans while Korean banks decrease hem. When he Korean ineres rae rises relaive o he LIBOR rae, foreign banks increase wondenominaed loans while Korean banks decrease hem. Those resuls sugges ha afer he financial crisis begins, foreign bank lending in he local currency proves pro-cyclical (decreasing lending when growh decreases and ineres raes fall) while Korean bank lending proves couner-cyclical. Foreign bank lending also reacs o changing condiions in he home-counry economy. The regression resuls for Korea sugges ha foreign currency lending responds o exernal markes. The home-counry ineres rae spread exers a negaive influence on foreign currency lending by foreign banks before, during, and afer he Asian financial crisis. The home-counry income growh rae only affecs lending (posiively) in foreign currency prior o he financial crisis. Shifing focus o oal loans afer he crisis began, foreign banks lend more in Korea when he home-counry ineres rae decreases and home-counry income growh rises. Lending by foreign banks responds o he opporuniy cos of lending in Korea (as measured by he home-counry ineres rae spread). This may reflec increasing financial marke inegraion across counries and Korea s pursui of financial liberalizaion policies. Wheher foreign bank operaions would conribue o financial marke sabiliy in Korea depends on he degree of financial marke inegraion beween Korea and he home counry, i.e. how closely correlaed are raes of reurn in he respecive counries financial markes. Income growh in he home counry leads o increased lending (foreign currency) before he 20

21 financial crisis, bu no afer he crisis. This suggess ha foreign bank lending can sabilize he Korean financial markes only if growh raes of income in he wo economies are negaively correlaed or uncorrelaed. I also suggess ha foreign banks supply more foreign currency loans in pre-crisis Korea for a given home-counry growh rae, han in pos-crisis Korea. This may provide evidence ha foreign banks promoe capial ouflows (or a leas reduce capial inflows) during periods of financial crisis. References: Bank of Korea, Bank Resrucuring in Korea. Seoul: The Bank of Korea Bank of Korea, Financial Liberalizaion and Inernaionalizaion in Korea. Seoul: The Bank of Korea Barh, J. R., G. Caprio, Jr., and R. Levine, Bank Regulaion and Supervision: Wha Works Bes. Naional Bureau of Economic Research Working Paper No Berger, A. N., R. DeYoung, H. Genay, and G. F. Udell, Globalizaion of Financial Insiuions: Evidence from Cross-Border Banking Performance. In R. E. Lian and A. Sanomero (Eds.), Brookings-Wharon Papers of Financial Services, 3. Caliński T., and J. Harabasz, A Dendrie Mehod for Cluser Analysis. Communicaions in Saisics 3, Claessens, S., A. Demirgüç-Kun, and H. Huizinga, How Does Foreign Enry Affec he Domesic Banking Marke? Journal of Banking and Finance 25, Crysal, Jennifer S., B. Gerard Dages, and Linda S. Goldberg, Has Foreign Bank Enry Led o Sounder Banks in Lain America? Curren Issues in Economics and Finance, 8(1), Federal Reserve Bank of New York. Dages, Gerard, Linda Goldberg and Daniel Kinney, Foreign and Domesic Bank Paricipaion in Emerging Markes: Lessons from Mexico and Argenina. Economic Policy Review. Federal Reserve Bank of New York. 6 (Sepember): Demirguc-Kun, Asli, Ross Levine and Hong-Ghi Min, Opening o Foreign Banks: Issues of Sabiliy, Efficiency, and Growh. In Seongae Lee, ed., The Implicaions of Globalizaion of World Financial Markes. Seoul:Bank of Korea. 21

22 Duda, R. O., and P. E. Har, Paern Classificaion and Scene Analysis. New York: John Wiley & Sons. Gilber, R.A. and P.W. Wilson, Effecs of Deregulaion on he Produciviy of Korean Banks. Journal of Economics and Business. 50 : Goldberg, Linda S, When is U.S. Bank Lending o Emerging Markes Volaile? Federal Reserve Bank of New York, Saff Repor 119. (March). Io, T., and A. O. Krueger, Regional and Global Capial Flows: Macroeconomic Causes and Consequences, Eas Asia Seminar on Economics, Volume 10. Jeon, Yongil and Sephen M. Miller, Domesic and Foreign Bank Performance in Korea during he Asian Financial Crisis. Paper presened a he Allied Social Science Associaion/Korea- American Economic Associaion meeings. Levine, Ross, Foreign Banks, Financial Developmen and Economic Growh. In Claude E Barfield, ed. Inernaional Financial Markes: Harmonizaion versus Compeiion. Washingon, D.C.: AEI Press. Peseni, Paolo and Cedric Tille, The Economics of Currency Crises and Conagion: An Inroducion. Economic Policy Review. Federal Reserve Bank of New York. 6 (Sepember): Ursacki, T., and I. Verinsky, Marke Posiioning and Performance of Foreign Banks in Korea. The Inernaional Journal of Bank Markeing 9,

23 Table 1: Number of Banks by Type and Counry of Origin Number of Foreign and Domesic Banks by Bank Type Year Foreign Banks Korean Naionwide Banks Korean Regional Banks Number of Foreign Banks by Counry of Origin Counry of Origin Number of Banks Counry of Origin Number of Banks Unied Saes 15 Swizerland 2 Japan 14 Germany 1 France 6 Hong Kong 1 Canada 4 India 1 Singapore 4 Iran 1 Unied Kingdom 3 Jordan 1 Ausralia 2 Pakisan 1 China 2 Philippines 1 Neherlands 2 Noe: The idenificaion of foreign banks by counry of origin includes all banks over he enire sample for a oal of 61 foreign banks. A number of foreign banks enered or exied from he Korean marke over he sample period. 23

24 Table 2: Financial Raios by Bank Type and Year Foreign Naionwide Regional Foreign Naionwide Regional Foreign Naionwide Regional la da ROE Mean lqaa ea fll Mean N/A N/A N/A lgr plll fdd Mean Ld ROA flfd Mean Noe: The numbers are averages for he various financial raios in each year or for he whole sample (Mean). The financial raios are as follows: loans o asses (la), liquid asses o asses (lqaa), he rae of growh of loans (lgr), loans o deposis (ld), deposis o asses (da), equiy o asses (ea), provisions for loan losses o loans (plll), reurn on asses (roa), reurn on equiy (roe), foreign currency loans o loans (fll), foreign currency deposis o deposis (fdd), and foreign currency loans o foreign currency deposis (flfd). We also perform -ess on he equaliy of means (no shown in his able). These ess are available upon reques.. 24

25 Table 3: Foreign Banks Loans Won Loans Foreign Currency Loans Model Random effecs Fixed effecs Fixed effecs D A -3,280** -2,150-1,550** (-2.07) (-1.42) (-2.42) dlny K ** (-0.17) (0.44) (-2.36) D A *dlny K * (1.22) (0.40) (2.87) dlny H (-0.87) (-1.89) (1.80) D A *dlny H 118** 150* -47.1** (2.13) (2.88) (-2.15) r K (0.46) (0.76) (-0.84) D A * r K (0.83) (0.09) (1.92) r H (1.05) (0.46) (-1.72) D A * r H -150* ** (-3.15) (-1.58) (-2.49) A i 0.22* 0.14* (6.34) (3.81) (-0.71) D A * A i * 0.03 Consan (1.76) 32.0 (0.02) Chow Tes a 54.0 [0.00] 2.77 Hausman Tes b [0.986] RE (2.62) -633 (-0.43) 9.22 [0.00] 49.6 FE (1.86) 1460** (2.38) 5.42 [0.00] 29.8 [0.001] FE Significance Tes [H 0 : Coefficien of Variable + Coefficien of (D A *Variable) = 0] c dlny K dlny H r K r H A i [0.064] [0.025] 1, [0.001] 1.88 [0.172] [0.531] [0.021] 0.33 [0.566]

26 Table 3: Foreign Banks (coninued) Noes: The variables are defined as follows: r K and r H equal he Korean and home-counry ineres rae spread over he LIBOR, dlny K and dlny H equal he Korean and home-counry growh rae of real GDP, and A i equals he size (measured in oal asses) of bank i in period. T- values are in parenheses for individual coefficiens. Significance levels appear in brackes for he Chow es, he Hausman, es, and all he significance ess a he end of he able. * coefficien significanly differen from zero a he 1-percen level. ** coefficien significanly differen from zero a he 5-percen level. coefficien significanly differen from zero a he 10-percen level. a b c The Chow es deermines wheher he coefficiens of all independen variables aken as a group differ beween he pre-asian crisis period and he Asian crisis period. F-values appear for fixed-effec models and chi-squared saisics, for he random-effecs models. The Hausman es deermines wheher he fixed- or random-effecs models are he appropriae specificaion. The saisic is disribued as a chi-squared. The es oucome is repored as FE (fixed effecs) or RE (random effecs). Only he appropriae specificaion is repored in he able. The significance ess a he end of he able deermine wheher he coefficiens during he Asian financial crisis differ significanly from zero. F-saisics are repored for he fixedeffec models and chi-squared saisics, for he random-effecs models. 26

27 Table 4: Korean Domesic Banks Loans Won Loans Foreign Currency Loans Model Random effecs Random effecs Random effecs D A -16,800 64,000** -61,700** (-0.61) (2.15) (-2.38) dlny K 887 2, (0.48) (1.19) (-0.43) D A *dlny K 174 (0.09) -3,870 (-1.88) 3,080 (1.72) r K 1,940 5,460-1,470 (0.37) (0.98) (-0.30) D A * r K 1,660-10,100 9,070 (0.31) (-1.77) (1.82) A i 0.57* 0.46* 0.14* (30.29) (16.47) (6.51) D A * A i ** Consan (-1.49) -14,400 (-0.55) Chow Tes a 25.1 [0.00] 4.77 Hausman Tes b [0.445] RE (-0.87) -39,700 (-1.40) 6.06 [0.19] 3.10 [0.685] RE (-2.38) 9,940 (0.40) 29.3 [0.00] 8.09 [0.151] RE Significance Tes [H 0 : Coefficien of Variable + Coefficien of (D A *Variable) = 0] c dlny K 4.37 [0.037] r K 8.57 [0.003] A i 4,890 Noes: See Noes for Table [0.007] 11.6 [0.001] 1,

28 Table 5: Korean Domesic Naionwide Banks Loans Won Loans Foreign Currency Loans Model Fixed effecs Random effecs Random effecs D A -23,800 92, ,000* (-0.58) (1.86) (-3.14) dlny K 1,630 3,730-2,070 (0.61) (1.16) (-0.85) D A *dlny K 1.30 (0.00) -5,840 (-1.76) 5,880** (2.32) r K 4,310 8,790-4,730 (0.58) (0.98) (-0.69) D A * r K 1,300-15,300 17,000** (0.17) (-1.65) (2.40) A i 0.60* 0.46* 0.13* (13.93) (11.34) (4.11) D A * A i (-1.03) (-0.24) (-0.76) Consan -30,100-64,400 35,200 (-0.76) (-1.37) (0.98) Chow Tes a [0.00] [0.19] [0.00] [0.901] [0.670] RE RE Significance Tes [H 0 : Coefficien of Variable + Coefficien of (D A *Variable) = 0] c 21.2 Hausman Tes b [0.001] FE dlny K 4.66 [0.034] r K 8.12 [0.006] A i 772 Noes: See Noes for Table [0.019] 7.94 [0.005]

29 Table 6: Korean Domesic Regional Banks Loans Won Loans Foreign Currency Loans Model Fixed effecs Fixed effecs Random effecs D A -6, ,560* (-1.06) (0.08) (-3.16) dlny K -7, (-1.77) (-0.93) (-1.59) D A *dlny K 474 (1.17) 91.4 (0.23) 233** (2.40) r K -2,040-1, (-1.83) (-1.08) (-1.43) D A * r K 1, * (1.29) (0.21) (2.74) A i 0.30* 0.32* 0.01 (5.27) (5.66) (0.88) D A * A i (0.85) (0.38) (0.48) Consan 20,600** 13,300 2,900 (2.65) (1.71) (1.84) Chow Tes a [0.07] [0.137] [0.011] [0.457] FE RE Significance Tes [H 0 : Coefficien of Variable + Coefficien of (D A *Variable) = 0] c 280 Hausman Tes b FE dlny K 7.14 [0.010] r K 5.99 [0.018] A i 109 Noes: See Noes for Table [0.006] 11.7 [0.001] [0.002] [0.027] 29