FINANCIAL LIBERALIZATION AND THE EVOLUTION OF BANKING ...

FIN A N C IA L LIBER A LIZA TIO N A N D TH E EV O LU TIO N O F BA N K IN G A N D FIN A N C IA L R ISK S:TH E C A SE O F SO U TH K O R EA § By Panicos O .D em etriades∗ U niversity ofLeicester

Bassam Fattouh CeFiM S,SO A S

K alvinderShields U niversity ofLeicester

A BSTR A C T The paper provides new insights into the role of financial liberalization in the South K orean financial crisis using a num ber of novel approaches. Firstly, prim ary inform ation regarding the relaxation offinancialrestraints,such as interestrate ceilings, capitalcontrols and reserve requirem ents,is collected and sum m arised. Secondly,this inform ation is used to constructsum m ary m easures of financialliberalization. Thirdly, qualitative inform ation on the role of financial liberalization in the financial crisis is presented from a new survey of 44 IM F, W orld Bank and K orean officials w ho had direct exposure to the events surrounding the financial crisis. Fourthly,the effects of financial liberalization on the evolution of banking and financial risks are estim ated utilising a conditional CA PM w ith tim e-varying m arket risk. Finally,qualitative and quantitative findings are juxtaposed,allow ing insights into the extentto w hich financial m arkets recognized the increased banking and financial risks, w hich em anated from financialliberalization. JEL classification: E44,F34,G 28,G 12 K eyw ords:Financialliberalization,financialcrisis,banking and financialrisks January 2001

♣

This paper is partof an ESRC-funded projectentitled “The South K orean FinancialCrisis:Lessons for Financial Liberalization” (G rant R000 22 2773). W e w ould like to thank, w ithout im plicating, the follow ing individuals: D r Jerry Caprio,D irector, Financial Strategy and Policy G roup, the W orld Bank for hosting our visitto W ashington,D .C.in O ctober 1999.D r Jea Chun K im ,D eputy D irector,M onetary Policy D epartm ent,The Bank of K orea,for hosting our visitto Seoulin A pril2000. M r Soon sang O h, Research D epartm ent,Bank of K orea for providing necessary facilities in Seoul.Eun-Ju Jang and JiEun Jung, postgraduate students at Ew ha W om ens U niversity, for assisting w ith the interview s in Seoul. Finally w e are indebted to G iovanniFerri(U niversity of Bari) and Tae-Soo K ang (W orld Bank and Bank ofK orea),forproviding us w ith theirlistofcontacts in Seoul. W e w ould also like to thank participants atM M F2000 conference (South Bank U niversity,London,6-8 Septem ber 2000) and the French-British M acroeconom ics W orkshop (U niversity of Paris 1, 24-25 N ovem ber 2000) for helpfulcom m ents on an earlier version of the paper;w e are specifically indebted to M ark P.Taylor and A bhay A bhyankar for constructive suggestions. N aturally all rem aining errors are ourow n. ∗ Corresponding author. A ddress for correspondence: Prof. P. D em etriades, D epartm ent of Econom ics, U niversity of Leicester,U niversity Road,Leicester,LE1 7RH . Tel:++44 (0) 116 2522835. Fax:++ 44 (0)116 2522908.Em ail:pd28@ le.ac.uk.

1.Introduction Financialliberalization has recently becom e alm ostsynonym ous to financialinstability, especially but not exclusively so in the case of em erging m arket econom ies (Stiglitz, 2000; D em irgüç-Kunt and D etragiache, 1999; D em etriades, 1999; A restis and D em etriades,1999). The m ostpopular explanation for this infam ous association is that financial liberalization usually fuels a lending boom , w hich funds the creation of an asset price bubble (e.g. A llen and G ale 2000). W hen the bubble bursts, collapsing collateral values result in bank insolvencies and a credit crunch, resulting in severe recessions. In the recent financial crisis in East A sia1, the lending-boom explanation accords reasonably w ell w ith the experience of Thailand (D em etriades, 1999). H ow ever,itdoes notappearto fitcom fortably the case ofSouth K orea,w here there w as hardly a detectable lending boom or an obvious assetprice bubble.Instead,the K orean crisis appears to be very m uch a case of inadequately m anaged financial risks. A necdotal evidence suggests that K orean financial interm ediaries borrow ed short i n foreign currencies and acquired low-quality foreign assets w ith longer m aturities. This created m aturity and exchange rate m is-m atches and increased overallcreditrisk,since even w hen exchange risk w as hedged, it w as substituted by increased credit risk (D em etriades and Fattouh,1999). W hile a lotm ore is now know n aboutthe K orean crisis than atthe tim e iterupted,there is very little evidence docum enting the evolution of financial risks before the crisis. Perhaps m ore im portantly, the role of financial liberalization in this process rem ains largely unknow n or even unrecognised. A t best, existing discussions of the role of financialliberalization are based on anecdotalevidence. A tw orst,the role of financial liberalization is neglected or m isunderstood. Y et,if there are any policy lessons to be learned from virtually any financial crisis they are alm ost inevitably related to the tim ing and im plem entation offinancialreform s. The paper provides new insights into the role of financial liberalization in the South K orean financialcrisis using the follow ing novelapproaches. (i)

Episodes of financial liberalization are docum ented by collecting prim ary inform ation from officialpublications on the relaxation of a variety of financial

1

restraints, including capital controls, interest rate ceilings and reserve requirem ents on bank deposits. This inform ation is used to constructsum m ary m easures offinancialliberalization,w hich are used in estim ations. (ii)

N ew

qualitative inform ation on the m echanism s by w hich financial

liberalization led to increased banking and financial risks is presented. This inform ation is obtained from a new survey of 44 IM F,W orld Bank and K orean officials w ho had directexposure to the events surrounding the K orean financial crisis.The survey w as carried outin W ashington,D .C.during O ctober1999 and in Seoul during A pril 2000. The findings from the survey are tabulated and analysed, providing a useful background to the form ulation of an em pirical m odel. (iii)

N ew econom etric evidence on the evolution of financial risks for the period 1987-1997 is presented w hich is aim ed at quantifying the effects of financial liberalization. This evidence is obtained by estim ating a conditionalCA PM in w hich the conditional variance-covariance m atrix of portfolio innovations follow s a m ultivariate G A RCH process. The m odelspecification allow s testing for the effects of financial liberalization on the conditional variance and riskiness ofthe banking and financialsectorportfolios.

(iv)

Q ualitative and quantitative findings are juxtaposed. This allow s insights into the extent to w hich financial m arkets recognized the increased banking and financialrisks,w hich em anated from financialliberalization.

The restofthe paperis structured as follow s.Section 2 provides a conceptualdiscussion of financialliberalization and its association w ith increased risks.Section 3 docum ents the K orean experience. Section 4 sum m arizes the findings from the IM F/W orld Bank survey.Section 5 presents the econom etric evidence on the evolution of banking and financialrisks.Finally,section 6 concludes. 2.FinancialLiberalization and Banking R isks:C onceptualIssues The traditionalapproach tow ards financialliberalization,w hich dates back to the w ork of M cK innon (1973) and Shaw (1973),em phasizes the benefits thatw ould accrue from m arket determ ined interest rates and credit allocation decisions (see also Fry, 1997). 1

Fora recentcom prehensive overview ofthe A sian crisis see H unter,K aufm an and K rueger(1999).

2

The potential benefits of financial liberalization include greater levels of financial savings and investm ent,as w ellas im provem ents in resource allocation,w hich em anate from m ore productive investm ents. It is im portant to note that even though this literature predated the econom ics of inform ation revolution, w hich after all explained the raison d’être for banks and financial institutions, it nevertheless had an enorm ous im pact on econom ic policy through the Bretton W oods institutions, since financial liberalization becam e an im portant elem ent of the set of policies associated w ith the ‘W ashington consensus’. In spite of unsuccessful im plem entation of financial liberalization in Latin A m erica and othercountries in the late seventies and eighties,the core of the financial liberalization thesis has rem ained intact, even though som e peripheralconcessions w ere m ade,including acknow ledging the im portance of policies and institutions that are expected to address m arket failures (see A restis and D em etriades, 1999). These included the appropriate ‘sequencing’ of reform s, in the form of attaining m acroeconom ic stability and adequate prudential regulation of the financialsystem ,priorto financialliberalization,as w ellas a specific orderforfinancial reform s,w ith the liberalization ofshort-term capitalflow s being placed atthe end ofthe reform sequence (see forexam ple M cK innon,1981). W e posit in this section that the im pact of financial liberalization on banking and financial risks is am biguous. W hile financial liberalization typically offers greater opportunities for diversification, by offering banks and other financial institutions a w ider range of assetchoices,w hich in principle should lead to m ore efficientportfolio choices,itm ay w ellexpose them to greaterrisks,due to lack ofexpertise in operating in new m arkets,w eaknesses in prudentialregulation and/or m oralhazard em anating from inform ation problem s. The m odern literature on financial liberalization reflects these tw o opposing forces. The restofthis section draw s on this literature to argue the case. Capitalaccountliberalization m ay in principle be expected to:(i)offerinvestors greater opportunities for risk diversification, achieving m ore effective insurance than purely dom estic arrangem ents w ould allow ,(ii)raise consum erw elfare by allow ing a sm oother consum ption path,(iii) resultin a m ore efficientallocation of resources by channelling the w orld’s savings tow ards the w orld's m ostproductive investm entopportunities,(iv) com plem entdom estic savings,thereby increasing investm entand prom oting econom ic

3

grow th w ithout sharp increases in savings rates, and (iv) low er the cost of capital to creditw orthy firm s and sm all and m edium enterprises (see O bstfeld and Rogoff, 1996 and Edw ards,1999). In reality capitalaccountliberalization has presented im portantchallenges and risks for policym akers (see, for exam ple, Stiglitz, 2000). Specifically, the recent East A sian financialcrisis has show n thatcapitalaccountliberalization can m agnify the risks and w eaknesses of the banking system , especially w hen capital inflow s are interm ediated through poorly m anaged and ill-supervised banking system s. The interm ediation of capital inflow s through such banking system s usually leads to an expansion in banks’ lending activity as banks have m ore resources available forlending.This generates w hat is know n as a 'lending boom '. Furtherm ore, dom estic banks can exploit m arket im perfections to generate over-optim istic expectations know ing that in case of default the governm ent w ill be forced to bail out distressed banks and firm s (M cK innon and Pill, 1997). Since entrepreneurs and firm s do not have enough inform ation to assess banks’ signals adequately, they consider these signals as correct and hence base their investm ent decisions on such over-optim istic expectations. Consequently, they bid eagerly for funds to finance their investm ents, further fuelling the lending boom (M cK innon and Pill,op.cit).2 O ne undesirable consequence of a lending boom fuelled by capitalinflow s is thatitcan exacerbate the m aturity and risk m ism atch betw een banks’ assets and liabilities.This is especially true if capital inflow s are short-term and in foreign currency w hile banks' loans are long-term and in dom estic currency. Furtherm ore, unregulated capital flow s m ay be m isallocated tow ards risky projects, speculative activities, the equity m arket, and cyclical sectors such as real estate. In the short run, the expansion of lending activity bids up (inflates) the price of assets in these m arkets generating an assetprice bubble. Such bubbles inevitably lead to deterioration in banks' portfolios as banks increase their holdings of 'inflated' assets and becom e heavily exposed to cyclical sectors.

2

W hat is interesting in M cK innon and Pill’s fram ew ork is that banks finance the lending boom by attracting capitalfrom abroad.The authors referto this process as the “over-borrow ing” syndrom e.

4

D om estic financialliberalization,w hich com prises m ainly of relaxation of controls on interest rates, lifting of restrictions on the asset choices of banks and low ering or abolishing reserve requirem ents,can also significantly increase the risks in the financial sector (Fischer and Chenard, 1997; D em irgüç-K unt and D etragiache, 1998; Stiglitz, 1994). Intense com petition that usually follow s financial liberalization low ers profits for banks, w hich in turn erodes banks’ franchise values and low ers their incentive for m aking good loans.This exacerbates the problem s ofm oralhazard and looting behavior in the banking system (see H ellm an,M urdock,and Stiglitz,2000;A kerlof and Rom er, 1993).These have the effectof increasing the riskiness of banks’ portfolios.A closely related argum ent is that financial liberalization erodes the protection provided by a regulated term structure and stable interm ediation m argin (G oldstein and Turner,1996). This m ay intensify the m oralhazard problem ,encouraging banks to engage in lending to m ore risky borrow ers in orderto increase the returns on theirfunds.Indeed,H ellm an et al (2000) show that certain types of financial restraints,such as ceilings on deposit rates,by keeping profitm argins w ithin certain lim its can reduce reduce the riskiness of banks’ portfolios by lim iting banks’ incentives to invest in assets that facilitate gam bling.Financialliberalization can also change the banks’ custom er base w ith larger and better-know n firm s raising a larger share of funding through the securities m arkets or international m arkets. The resulting effect is generally deterioration in the risk com position of the bank and financial sector’s loan portfolios (Fischer and Chenard, 1997). In principle,how ever,dom estic financialliberalization can generate efficiency gains by rem oving various constraints on banks’feasible risk-return frontier,w hich m ay resultin low er overall banking risks (H ogan and Sharpe, 1984). Furtherm ore, financial liberalization m ay open new profitable opportunities,w hich bankers could exploitand thereby avoid the erosion of their franchise value.3 H ence,w hile a case could be m ade thatthe im pactoffinancialliberalization on the financialsector’s overalllevelofrisk is am biguous atthe theoreticallevel,m oststudies usually associate financialliberalization w ith higherrisks (Fischerand Chenard,1997;H ellm an etal,2000).

3

H ow ever, H ellm an et al (2000) argue that greater investm ent opportunities, w ide ranges of new activities such as derivative trades and foreign currency transactions and greater freedom to allocate assets also increase the potentialscope forgam bling by banks.

5

A n im portant elem ent of dom estic financial liberalization that has a direct im pact on banks’ riskiness is reserve requirem ents on bank deposits. Reserve requirem ents are usually considered as tax on financial interm ediation, w hich low er the profitability of the banking sector and hence m ay affectits attitude tow ards risk.Furtherm ore,required reserve ratios affect the level of liquidity available for banks and hence affect their lending decisions.M itchell(1986)finds thatthe im pactofreserve requirem ents on bank riskiness depends on the w ay bank risk is m easured and on the assum ptions m ade about risk aversion.Specifically,ifbank risk is m easured by the ratio ofbanks'risky assets to total assets, then an increase in the required reserve ratio w ill drive dow n the level of bank risk.O n the other hand,if bank risk is m easured by the probability thatthe banks' profitw illfallbelow zero,a rise in required reserves w illalso drive dow n bank risk if and only ifthere is increasing relative risk aversion.The reverse holds,how ever,ifthere is decreasing relative risk aversion.H ence,atthe theoreticallevel,the im pactofreserve requirem ents on bank riskiness is also am biguous.4 To sum -up,the traditionalfinancial liberalization thesis,as w ellas its m odern version, tends to em phasize its potential benefits, in the form of efficiency gains and opportunities for diversification, w hich in principle should led to m ore efficient portfolio choices; these m ay be reflected in both greater investm ent returns and low er risks,in both the realand financialsectors. H ow ever,skeptics argue thatbecause ofthe endem ic nature of im perfect inform ation and institutional w eaknesses, associated m arket failures, such as m oral hazard, could w ell m ean that financial liberalization instead leads to substantially increased financial risks and low er ex-post investm ent returns. The K orean experience,to w hich w e now turn,reflects both the traditionalist beliefs,w hich resulted in under-estim ation ofbanking and financialrisks,as w ellas the realities ofincreased risks through m arketfailures and institutionalw eaknesses. 3.FinancialLiberalization in South K orea In the last tw o decades or so, the South K orean financial system w itnessed m ajor liberalization efforts, especially on the capital account front. A ccording to the discussion of Section 2, these regulatory changes are likely to be associated w ith a 4

G elles (1991) show s that all the above conclusions hold for any bank w ith reserves and a risk-averse utility function w ith a m ean-standard deviation fram ew ork thatis consistentw ith expected utility.

6

change in the levelof the riskiness of K orean financialinstitutions.Before w e present qualitative and quantitative evidence on the im pactof such liberalization efforts on the riskiness of the K orean financialsystem ,itis usefulfirstto discuss briefly the K orean experience w ith capitalaccountand dom estic financialliberalization.5 Capitalaccountliberalization In the late 1980s, the K orean governm ent accelerated the liberalization process of its capitalaccount.The liberalization of the capitalaccounttook place m ainly by relaxing controls on banks and corporations’fund-raising activity in internationalm arkets and by allow ing foreigners to invest in the K orean stock, bond and m oney m arkets. In D ecem ber 1989, foreign exchange banks w ere allow ed to raise offshore funds by issuing foreign currency denom inated bonds orborrow ing from the offshore accounts of otherdom estic foreign exchange banks.The m ain liberalization step,how ever,occurred in January 1992, w hen non-residents w ere allow ed for the first tim e to invest in any dom estic stock unless specified in som e particularact,even though som e lim its w ere set on the level of total foreign investm ent.6 D uring Septem ber 1992 regulations on the overseas issue offoreign currency denom inated securities w ere greatly eased. The type of securities thatcould be issued abroad by K orean residents,restricted previously only to bonds, convertible bonds, bonds w ith w arrants and stock depository receipts, w ere expanded to include negotiable CD s and com m ercial papers. Furtherm ore, the authorization procedures necessary forthe issue ofsecurities w ere greatly sim plified. D uring 1993-1998,the K orean governm entresum ed the opening ofits financialm arkets to foreign investors. For instance, in July 1994, the governm ent partially opened the dom estic bond m arket allow ing non-residents to purchase non-guaranteed convertible bonds issues by sm alland m edium enterprises (SM Es) subjectto certain lim itations.In M ay 1996,non-residents w ere allow ed to purchase and trade bonds w ith w arrants and to trade the stock index futures on the K orean Stock Exchange.In June 1997,foreign investors w ere granted access to non-guaranteed bonds of SM Es and of conglom erates 5

A m ore detailed discussion can be found in the appendix. The inform ation in this section and the appendix w as obtained from the Bank ofK orea A nnualReports. 6 For instance, a 3% lim it on investm ent by an individual foreign and 10% lim it on total foreign investm ent w ere applied respectively and in the case of public utilities and those com panies in infant industry,the totalforeign investm entlim itis setat8% .

7

and by 1998,allkinds of securities stipulated in the Securities and Exchange A ctw ere m ade available to foreign investors.7 A nother im portant developm ent has been the abolition of ceilings on the purchase of dom estic stocks by foreigners.In parallelw ith these developm ents, controls on foreign borrow ing w ere largely dism antled during 1993-1995.In February 1993,overseas branches of dom estic banks w ere perm itted to supply loans to K orean residents engaged in the trading of the com m odity futures or financial futures. Later in the sam e year,security issuers in foreign m arkets w ere no longer required to obtain perm ission before issuing foreign currency denom inated securities. Furtherm ore, the list of corporations and banks that could issue foreign securities w as considerably w idened. By O ctober 1996, the governm ent dism antled m ostof the restrictions on directforeign borrow ings,enabling even non-m anufacturing SM Es to receive loans from abroad. InterestRate Liberalization U nlike the capital account liberalization process, dom estic financial liberalization occurred gradually over a long period of tim e. A t the heart of dom estic financial liberalization in K orea w as the liberalization ofinterestrates.Since the early 1960s,one of the m ost im portant characteristics of the South K orean credit m arket has been the direct intervention of the state in the pricing of credit, w hich w as m ainly achieved through controls on lending, and deposit interest rates. In Septem ber 1979, the M onetary Board abolished the m axim um interest rate on bank loans. H ow ever, given K orean banks’ inexperience in setting interest rates, the K orean Bankers A ssociation decided to link the interestrate on loans to the Bank of K orea’s rediscountrate w hich seriously lim ited the ability of K orean banks to alter lending rates.In July 1984,banks w ere allow ed to charge different rates according to the creditw orthiness of borrow ers butw ithin a narrow band. It is only in D ecem ber 1988 that banks began to enjoy com plete freedom over interestrate determ ination w hen controls on lending rates from banks and non-bank financial interm ediaries w ere relaxed despite the fact that som e controls on policy loans rem ained in place.In a m ove tow ards furtherliberalization,the interestrate on policy based loans w ere liberalized in July 1995 and,in January 1996,

7

For exam ple,short-term financialproducts such as com m ercialpapers,com m ercialbills,and trade bills and CD s issued by financialinstitutions;and unlisted stocks and bonds.

8

the Bank ofK orea lifted the restriction on the size ofprem ium a bank could charge over itsprim e lending rate. The relaxation ofcontrols on depositrates in K orea w as very gradual.A lthough in 1979 the M onetary Board abolished the m axim um interestrate on personalchecking deposits, it is not until D ecem ber 1988, w hen the M onetary Board liberalized interest rates on certain tim e and saving deposits, that banks started enjoying som e freedom in determ ining deposit rates.8 In N ovem ber 1991, the scope of initial liberalization w as extended to cover rates on long-term deposits w ith a m aturity of 3 years offered by banks, m utual credit facilities, and credit unions. It is only as recently as N ovem ber 1995 thatthe Bank ofK orea freed up the rem aining regulated interestrates on bank and non-bank tim e deposits w ith m aturity ofless than six m onths. The Bank of K orea concentrated its efforts in developing m oney m arkets by relaxing controls on the issue and sale ofexisting instrum ents and introducing new ones.In June 1982,the callrate,w hich had been subjectto an upperlim itof16% w as deregulated.In M arch 1986, the rates on negotiable CD s, introduced only in June 1984, w ere also liberalized.Further liberalization m easures took place in D ecem ber 1988 w hen interest rate on repurchase agreem ents (RPs), com m ercial papers of certain m aturities (CPs), financial debentures and corporate bonds w ere fully liberalized. The m ajor change how ever cam e in O ctober 1989, w hen the governm ent m erged the call m arkets, previously segm ented into an inter-bank m arketm ainly for banks and over the counter m arket betw een non-bank financial interm ediaries and liberalized the interbank rate. Furtherliberalization w as carried in the 1990s w here the M onetary Board liberalized the rates on governm ent and public bonds,shortened the m aturity of RPs,CD s and other financialinstrum ents,and significantly deregulated the bond m arketin N ovem ber1991. D uring the 1992-1995 period,the bank low ered gradually the m inim um denom inations of CD S and shortened the m aturities of the RPs. In fact, by 1995 the K orean m oney m arkets had becom e highly liberalized. It em erges from this brief overview that in the last decade or so, K orean financial institutions w itnessed m ajorregulatory changes thatm ay have increased the riskiness of 8

Specifically,only interestrate on tim e deposits of m aturity greater than 2 years atbanks,postalsavings and creditunions and on tim e and savings deposits of m aturity greater than 1 year atm utualsavings and finance com panies w ere liberalised.

9

these financial institutions. In w hat follow s, w e exam ine this i ssue both qualitatively and quantitatively. 4. The evolution ofbanking and financialrisks:qualitative evidence This section presents the results of tw o sets of interview s carried outin (i) W ashington, D .C. during the autum n of 1999 and (ii) Seoul during A pril 2000. The interview s follow ed a sem i-structured questionnaire,w hich contained 21 questions relating to the factors thatcaused the crisis. The respondents in W ashington w ere 15 officials of the InternationalM onetary Fund and the W orld Bank who had directexposure to the events surrounding the K orean financial crisis. The respondents in South K orea w ere 29 private and public sector econom ists w ith directexperience of the financialcrisis.They included senior officials of the Bank of K orea,the M inistry of Finance and Econom ics, the K orean D evelopm ent Institute, the K orean Institute of Finance, private research institutes (funded by K orean chaebols), com m ercial banks (both K orean and international) and other financial institutions. Tables 1 and 2 present the sum m ary responses to seven questions thatfocus on the evolution of banking and financialrisks and the effects of financial liberalization, as perceived by the respondents after the crisis. The answ ers to these questions from both sets of interview s seem to support the view thatfinancialliberalization increased the riskiness ofthe K orean financialsector.A llthe respondents in W ashington and 72% of the respondents in Seoulthoughtthatfinancial liberalization (defined as the rem ovalof interestrate restraints and capitalcontrols) on balance - taking into account the responses of financial institutions and regulators, increased the risks faced by K orean financial institutions. The survey also reveals another interesting observation: it show s that 73% of the respondents in W ashington and 97% of the respondents in Seoul thought that the institutional fram ew ork of prudential regulation and supervision w as not w ell developed to deal w ith the risks associated w ith substantialvolum es of capitalflow s. A llbutone respondent(i.e.93% ) in W ashington and 86% of the respondents in Seoul thought that K orean financial institutions did not have in place adequate risk m anagem ent system s. 87% of respondents in W ashington and 79% of respondents in Seoul thought that K orean financialinstitutions did nothave the hum an capitalorthe expertise to m anage the risks

10

associated w ith the interm ediation of large am ounts of foreign capital. These results suggest that financial liberalization m ay have resulted in increased banking and financialrisks due to inadequate risk m anagem entby financialinstitutions and expertise and due to w eaknesses in prudentialregulation. Table 1:Interview responsesofIM F and W orld Bank officials A nsw er Q uestion M any econom ists believe thatfinancialliberalization (i.e. rem oval of interest rate restraints and capital controls) leads to higher investm entreturns.D o you agree? G iven thatm any econom ists do believe thatfinancial liberalization increases investm ent returns, do you think that it m ay have played som e role in creating over-optim istic expectations about investm ent payoffs? Som e econom ists believe thatfinancialliberalization leads to increased risks in the financialsystem ,in the form of exchange risk, credit risk and interest rate risk.D o you agree? D id K orean financial institutions have in place the risk m anagem ent system s required to m anage the new risks that financial liberalization m ay bring about? D o you think thatK orean financialinstitutions w ere equipped w ith the hum an capital and expertise to adequately m anage the risks associated w ith the interm ediation oflarge am ounts offoreign capital? Do you think that the institutional fram ew ork of prudentialregulation and supervision w as sufficiently w elldeveloped to dealw ith the risks associated w ith substantialvolum es offoreign capital? Taking into accountthe new types ofrisks as w ellas the responses of financialinstitutions and regulators to these risks, w ould you say that on balance financial liberalization increased the risks faced by K orean financialinstitutions?

11

Yes

No

M aybe

Don’tknow

10

2

2

1

8

4

3

0

14

0

1

0

0

14

1

0

1

13

1

0

0

11

2

2

15

0

0

0

Table 2:Interview responsesofSouth K orean officials A nsw er Q uestion

Yes

No

M aybe

Don’tknow

M any econom ists believe thatfinancialliberalization (i.e. rem oval of interest rate restraints and capital controls) leads to higher investm entreturns.D o you agree?

18

1

10

0

G iven thatm any econom ists do believe thatfinancial liberalization increases investm ent returns, do you think that it m ay have played som e role in creating over-optim istic expectations about investm ent payoffs?

15

6

8

0

Som e econom ists believe thatfinancialliberalization leads to increased risks in the financialsystem ,in the form of exchange risk, credit risk and interest rate risk.D o you agree?

20

3

5

1

D id K orean financial institutions have in place the risk m anagem ent system s required to m anage the new risks that financial liberalization m ay bring about?

2

25

2

0

D o you think thatK orean financialinstitutions w ere equipped w ith the hum an capital and expertise to adequately m anage the risks associated w ith the interm ediation oflarge am ounts offoreign capital?

1

23

5

0

D o you think that the institutional fram ew ork of prudentialregulation and supervision w as sufficiently w elldeveloped to dealw ith the risks associated w ith substantialvolum es offoreign capital?

0

28

1

0

Taking into accountthe new types ofrisks as w ellas the responses of financialinstitutions and regulators to these risks, w ould you say that on balance financial liberalization increased the risks faced by K orean financialinstitutions?

21

4

4

0

Finally, the survey results show that roughly tw o thirds of the respondents in both W ashington and Seoul believed that financial liberalization norm ally leads to higher investm ent returns. Just over half the respondents in both W ashington and Seoul thought that financial liberalization played a role in creating over-optim istic expectations about investm ent payoffs in K orea. These findings suggest that the traditionalbeliefs concerning financialliberalization w ere atleastpartly responsible for the financial crisis for at least tw o reasons. Firstly, by em phasising efficiency gains through enhancing the quality of investm ent, they seem ed to have contributed to 12

creating over-optim istic expectations concerning investm ent payoffs.9 In fact, m any have argued thatthe creation ofover-optim istic expectations is the m ain im petus behind the lending boom and the assetprice bubble (M cK innon and Pill,1997).Secondly,by failing to em phasize the greater risk-taking opportunities that accom pany financial liberalization they generated com placency in relation to the recognition ofrisks and risk m anagem ent both by the financial institutions them selves and the supervisory authorities.10 5.The Evolution ofBanking and FinancialR isks:Econom etric Evidence In this section,w e investigate the extentto w hich the survey results,w hich indicate an increase in risks em anating from financial liberalization, w ere reflected in financial m arkets.To this end,w e use an approach w hich exploits the inform ation contained in the price index of securities issued by banks and other financial firm s such as investm entbanks,m erchantbanks and securities com panies.Specifically,w e exam ine the changes in the prices ofthe banking and financialsectorportfolios in orderto obtain inform ation on the m arket’s assessm entofthe evolution ofrisks ofthe K orean financial system .In orderto m easure the riskiness ofthe banking and financialsector,w e use the conditional Capital A sset Pricing M odel (CA PM ) in w hich the conditional variancecovariance m atrix of portfolio innovations is assum ed to follow a m ultivariate G eneralized A utoregressive Conditional H eteroskedastic (G A RCH ) process. G A RCH m odels provide a flexible m ethod for m odeling tim e-varying conditionalvariances and co-variances and m ore im portantly capture the em pirical regularities found in stock returns (N g,1991).W e investigate the im pactoffinancialliberalization on the volatility 9

Edw ards (1997) reaches a sim ilar conclusion in the contextof the M exican crisis w here he argues that the “financial m edia, academ ic analysts, W all Street experts,and m ultilateral institutions invented the M exican m iracle” w hich created a w ave ofover-optim ism notbased on realeconom ic perform ance. 10 There is little doubtthatanother im portantfactor responsible for inadequate risk m anagem entw as the m oral hazard em anating from the history of the socialization of risks in South K orea. H ow ever, the im plicitprovision ofsafety nets by the governm entw as notnew . Itw as presentin South K orea since the early 1960s,yetno m ajor financialcrisis w as observed before 1997. G overnm entprovided safety nets go som e w ay in explaining disincentives in m anaging risk,butthey do notexplain the increase in risk taking opportunities thataccom panies financialliberalization. Indeed,up to the early 1990s the socialization of risks w as an im portant factor in ensuring the large investm ents undertaken by chaebols,m ost of w hich w ere responsible fortransform ing K orea into a highly industrialized country. Itis conceivable thatw ith a different set of beliefs, nam ely one w hich acknow ledged the substantially increased risk taking opportunities associated w ith financial liberalization, necessary im provem ents in prudential regulation, risk recognition and m anagem ent w ould have taken place, even in the presence of im plicit safety nets. Indeed,m any such im provem ents have now taken place in K orea itself,w here the risks em anating from financialliberalization,are now w idely acknow ledged.

13

of bank and financialsector stock returns by focusing on tw o areas of reform s,nam ely, dom estic financial liberalization and capital account liberalization. Consequently, w e exam ine the effects of (i) interest rate liberalization and relaxations of reserve requirem ents on dom estic deposits,w hich form the centerpiece of various attem pts of dom estic financial liberalization (ii) the relaxation of controls on capital flow s, on banking and financial risks. To this end, w e augm ent the conditional covariance equations of the G A RCH m odel w ith indices of interest rate liberalization, reserve requirem ents and capitalaccountliberalization. The rest of this section describes our m odeling fram ew ork,presents the data and data sources, including the construction of the relevant liberalization indices, and presents the m ain em piricalfindings. TheM odelling Fram ework The CA PM due to Sharpe (1964),Linter (1965) and Black (1972) explains the risk of a particularassetorportfolio using the excess return on the m arketportfolio.Specifically, the conditionalCA PM m odelforan assetorportfolio ican be stated as follow s: E(rit/W t-1)= bitE(rm t/W t-1)= d H tw t-1

(1)

W here rit is the return on portfolio iin excess of the return on the risk-free asset,rm t is the return on the m arketportfolio in excess of the return on the risk-free asset,d is the aggregate m easure ofrelative risk,H t is the conditionalcovariance m atrix w ith elem ents ⎨hijt⎬, w t-1 is the vector of assets w eights in the m arket portfolio m , and bit is the conditionalbeta ofportfolio iw ith the m arketportfolio and represents the dependence

on m arketportfolio risk.bit can be defined as follow s:

bit= him t/hm m t

(2)

w herehim tis the conditionalcovariance betw een the return on portfolio iand the m arket portfolio and hm m t is the conditionalvariance of the m arketportfolio.In this version of the CA PM ,allm om ents are m ade conditionalon the inform ation available attim e t–1 as given by the inform ation setW t-1.

14

For estim ation purposes, it is useful to decom pose the actual return on the different portfolios into forecastable and unforecastable parts: rit = E(rit/W t-1)+ uit

(3)

rm t= E(rm t/W t-1) + um t

(4)

w here rit and rm t are actualorrealized returns and uit and um t denote the colum n vectors of the differences betw een realized excess returns and expected excess returns. Substituting (1)into (3)and using the definition ofbit,w e obtain the follow ing: rit= (him t/hm m t)πm t+ uit

i=1,2

(5)

w here π m t = E(rm t/W t-1).For the purposes of this study,w e include,in addition to the m arket portfolio, the banking sector and financial sector portfolios, hence i=1,2, respectively.11 W e stack the innovations from the banking sector, the financial sector and the m arket portfolio into the vectoretw here

⎛ uit et/W t-1 = ⎜⎜ ⎝ um t

⎞ ⎟⎟ ~ N(0,H t) ⎠

(6)

and the conditionalvariance-covariance m atrix ofassetinnovations in (6)is assum ed to follow a m ultivariate G A RCH process (Bollerslev,1990).Follow ing Bollerslev,Engle and W ooldridge (1988), w e assum e that the innovation vector follow s a sim ple G A RCH (1,1) process. The sim plest generalization of the G A RCH (1,1) m odel can be stated as: (et/W t-1 ) ~ N(0,H t) Vech(H t)= w + y Vech(H t-1)+ Λ Vech(e t-1 e’t-1)

(7)

w hereVech(.)denotes the colum n-stacking operatorofthe low erportion ofa sym m etric m atrix,etisan (N ·1)vectorofinnovations,w is a (½ N(N+1)·1)param etervector,and

y and Λ are(½ N(N+1)· ½ N(N+1))m atrices ofconstantparam eters.The specification in (8) has (½ N 2(N+1) 2 + ½ N(N+1)) param eters in the conditional variance and covariances, w hich m akes estim ation of the system

of equations practically

unm anageable. In our sim ple three-portfolio m ultivariate G A RCH (1,1) m odel, the num ber of param eters to be estim ated w ould be 78.In order to achieve tractability,w e 11

In principle, w e could use data on stock prices of individual com m ercial banks, investm entbanks, securities com panies, etc. H ow ever, the inclusion of a w ide list of stocks entails the estim ation of too m any param eters.

15

need to im pose som e reasonable restrictions on the variance-covariance m atrix. Bollerslev,Engle and W ooldridge (1988) suggestthatthe covariance m atrix is w ritten as a set of univariate G A RCH m odels w here the conditional covariance of each portfolio is assum ed to depend only on its ow n lagged covariance and the cross product of past forecast errors.12 This can be obtained by m aking the m atrices y and Λ in (8) diagonal.In this restricted m odel,the num berofparam eters w ould be 3N(N+1)/2,hence for our three-portfolio m odel,the num ber of param eters to be estim ated w ould be 18. Based on this specification,the elem ent(i,j)of H t is given by: hijt= w ij+ a ijhijt-1 + b ijuit-1 ujt-1

(8)

W e augm entthe conditionalvariance and covariance equations to incorporate m easures of interest rate liberalization, reserve requirem ents on dom estic dem and deposits and capitalaccountliberalization.A s postulated in section 2,allthese indices are likely to have an im pact on the volatility of bank stock returns.In addition,the excess m arket return equation (4) incorporates indices on interest rate and capital account liberalization.13 It is often argued in the literature that a shift from a ‘financially repressed’ econom y to a ‘financially liberalized’ econom y is likely to result in m ore efficient allocation of resources, w hich has the im pact of increasing the return on investm ent.14 H ow ever,itis now w idely recognized thatin the presence of inform ation asym m etries and contract enforcem ent problem s, it is not necessarily true that the banking system w ill allocate resources to projects or firm s w ith the highest return. Furtherm ore, in the presence of inadequate regulation and bank supervision, capital account liberalization m ay, in fact, have an adverse im pact on productivity. For instance,in M cK innon and Pill’s (1997)fram ew ork,dom estic banks can exploitm arket im perfections and generate ‘over-optim istic’ expectations. A s a result, entrepreneurs and firm s w ill bid eagerly for funds to finance their investm ents, creating a lending boom and an assetprice bubble.Price distortions and resource m isallocations of these types have an adverse im pact on the productivity of capital. G iven these com peting explanations,the im pactoffinancialliberalization on m arketreturns is am biguous.

12

See also N g (1991)and Engle and K roner(1993). There does notseem be any strong theoreticaljustification for reserve requirem ents to have an im pact on the m arketreturn;hence this variable is notincluded in the m arketreturn equation. 13

16

The com plete system of equations of our three-portfolio m odel using the diagonal representation is given by: rm t= a0 + a1 INT t+ a2 CAP t+ um t

(9.1)

hm m t= w 01 + a 11 hm m t-1 + b11 um 2 t-1 + d11 INT t + q11 RD t + g11 CAP t

(9.2)

h11t= w 02 + a22 h11t-1 + b22 u12t-1 + d22 INT t + q22 RD t + g22 CAP t

(9.3)

h22t= w 03 + a33 h22t-1 + b33 u22t-1 + d33 INT t + q33 RD t + g33 CAP t

(9.4)

h1m t= w 04 + a44 h1m t-1 + b44 u1t-1 um t-1 + d44INT t + q44 RD t + g44 CAP t

(9.5)

h2m t= w 05 + a55 h2m t-1 + b55 u2t-1 um t-1 + d55 INT t + q55 RD t + g55 CAP t

(9.6)

h12t= w 06 + a66 h12t-1 + b66 u1t-1 u2t-1 + d66 INT t + q66 RD t + g66 CAP t

(9.7)

r1t= (h1m t/hm m t)πm t+ u1t

(9.8)

r2t= (h2m t/hm m t)πm t+ u2t

(9.9)

w here INT,RD,and CAP are the m easures relating to interestrate liberalization,reserve requirem ents on dom estic dem and deposits and capital account liberalization, respectively.This system of equations can be estim ated using the m ethod of m axim um likelihood assum ing the conditional norm ality of the forecast errors,15 w here the loglikelihood function is as follow s: Ln L(f)= const– ½

∑ ln|H t|– ½ ∑ t

t

(e't−1 H t−1et−1 ),

(10)

and f contains the unknow n param eters in rm t,etand H t. D ata The K orean stock price index (K O SPI) is used as a proxy for the m arket portfolio. W eekly data on K O SPI,the bank and financialsectorindices forthe period 7/1/1987 to 29/7/1997 w ere obtained from D ataStream .16 The three indices are expressed in local currency. The rate of return on the portfolio is defined as the first difference of the 14

See M cK innon (1973)and Shaw (1973)fora sem inalcontribution. The quasi-m axim um likelihood m ethod,w hich provides consistentestim ates provided thatthe firstand second m om ents of the standardised distribution can be specified, can also be used if there are sm all departures from norm ality (see Bollerslev and W ooldridge (1992)).H ow ever,in this em piricalw ork,this assum ption cannotbe rejected and w e use the m ethod ofm axim um likelihood. 16 N ote that during this period, the K orean governm ent rem oved m ost controls on interest rates and em barked on a program of capital account liberalization. H ence, this sam ple allow s us to exam ine w hether the relaxation of various controls had an im pact on the riskiness of banks and other financial institutions.G iven that the East A sian crisis m ust have generated pow erful shocks to the return on the various portfolios,w e exclude the lastquarterof1997 from ourestim ation sam ple. 15

17

logarithm ic stock price index and excess returns are com puted i n local currency in excess of the overnightcallrate (calculated on a w eekly basis),w hich acts as a proxy forthe risk-free interestrate. Figures 1a-1c in the appendix plotthe m arketexcess return series and the tw o-portfolio excess returns series.These figures show thatexcess returns on the various indices are, on average,zero over the period (in factthe m ean return on the three differentindices are insignificantly different from zero during the period under study).The graphs also show periods of clusters of high and low volatility, suggesting the presence of autoregressive conditional heteroskedasticity (A RCH ) effects. The presence of these effects cannotbe rejected (using LM and portm anteau Ljung-Box tests) and the use of the G A RCH m odeling fram ew ork described earliertherefore appears w arranted. The construction of the indices utilises the detailed inform ation aboutfinancialreform s sum m arized in appendix II and obtained from the annualreports of the Bank of K orea. Specifically, the m easures of interest rate and capital account liberalization are constructed using inform ation and data obtained from the Bank of K orea, A nnual A ccounts. They are assum ed to take a value of one prior to any relaxations, and decrease in value w henever financial restraints are relaxed or rem oved; they are therefore increasing w ith the severity offinancialrestraints,and decreasing as financial liberalization progresses. Specifically, for the construction of the interest rate liberalization index,w e use inform ation on controls on depositrates,lending rates and m oney m arketrates. Strong positive correlation betw een the lending rate,depositrate and m oney m arketindices allow s us to average them outinto a single m easure,w hich w e call‘the interestrate liberalization index’.Forthe construction ofthe capitalaccount liberalization index w e use detailed inform ation on the relaxation of controls on banks and corporations’ fund-raising activity in international m arkets and relaxation of restrictions on foreign investm entin the K orean stock,bond and m oney m arkets. Figure 2a show s the m ovem entof the interestrate liberalization index.A s can be seen, itreflects the changes in the underlying policy variables reasonably w ell.The relaxation oflending and depositrate controls in D ecem ber1988 is reflected in a sharp drop ofthe m easure for that m onth. The m easure then drops sharply during the second w ave of

18

reform (1992-1995 period). D uring that period, m ost of the rem aining controls on deposit,lending and m oney m arketrates w ere abolished.Figure 2b plots the m ovem ent ofthe capitalaccountliberalization index.The figure reveals thatthe index also reflects the underlying m easures quite accurately.Itshow s a sharp decline in the beginning of 1992 w hen the K orean stock m arketw as open to foreign investors and dom estic banks w ere allow ed to raise funds in internationalfinancialm arkets.In subsequentyears,m ost controls on capitalinflow s w ere gradually rem oved and this is reflected in the gradual decline of the capital account index. Figures 2a and 2b also show a high correlation (0.97) betw een these tw o indices.Clearly,this poses problem s for estim ation purposes, and w e address this issue using Principal Com ponent A nalysis (see Theil, 1971) in order to sum m arise both liberalization indices in a m eaningful w ay; w e retain one principalcom ponentw ith an eigenvalue greaterthan one.17 The m easure of reserve requirem ents on dom estic dem and deposits is constructed using data on reserve requirem entratios.18 The index,graphed in Figure 2c show s thatreserve requirem ents on dem and deposits increased significantly during the 1987-1989 period and rem ained relatively high untilthe m id 1990s,to decline to very low levels in 1996 and 1997. Em piricalResults The follow ing system of equations is estim ated by m axim ising equation (10) using the BH H H algorithm : rm t= a0 + a1 LIB t + um t hm m t= w 01 + a 11 hm m t-1 + b11 um

(11.1) 2

t-1 +

d11 LIB t + q11 RD t

(11.2)

h11t= w 02 + a22 h11t-1 + b22 u12t-1 + d22 LIB t + q22 RD t

(11.3)

h22t= w 03 + a33 h22t-1 + b33 u22t-1 + d33 LIB t + q33 RD t

(11.4)

h1m t= w 04 + a44 h1m t-1 + b44 u1t-1 um t-1 + d44 LIB t + q44 RD t

(11.5)

17

See also D em etriades and Luintel(1997) or A restis and D em etriades (1997),w ho also advocate using (principalcom ponent)sum m ary m easures offinancialrepression/liberalization. 18 Luarens and Cordoso (1998) argue thatindices based only on the reserve requirem entratio and thatdo nottake into accountthe continued changes in the tax base cannotcapture accurately the restrictiveness of reserve requirem ents. This argum ent applies to the Chilean case w here authorities have continuously changed the tax base to close loopholes and m ake the controls m ore restrictive. In the case of K orea, how ever,there have been no attem pts to change the tax base and as such the index w e use in this paper rem ains valid.

19

h2m t= w 05 + a55 h2m t-1 + b55 u2t-1 um t-1 + d55 LIB t + q55 RD t

(11.6)

h12t= w 06 + a66 h12t-1 + b66 u1t-1 u2t-1 + d66 LIB t + q66 RD t

(11.7)

r1t= (h1m t/hm m t)πm t+ u1t

(11.8)

r2t= (h2m t/hm m t)πm t+ u2t

(11.9)

w here LIB represents the principal com ponent of the financial liberalization policy variables and RD is the index ofreserve requirem ents on dom estic deposits. The estim ated coefficients of this m odel are reported in Table 3 below , w here the figures in parentheses denote the m arginal significance levels.Table 3 show s that the m odel perform s quite w ell in explaining the conditional variances of the bank and financialsectorstock returns,as w ellas ofthe m arketreturns.A llthe coefficients on the lagged conditional variances and lagged squared residuals are significantly different from zero at the 1% level and are w ithin a reasonable range. This suggests that the G A RCH (1,1) conditional variance-covariance m atrix is a good description of the behavior of the bank, financial and m arket sector stock returns.The table also reports results of the diagnostic tests perform ed on the residuals to provide an indication of the adequacy of the m odel.The Ljung-Box Q and Q-squared statistics on the standardized residuals (uˆt /hˆt) and the squared residuals (uˆ2t /hˆ2t),respectively,indicate thatthere is little evidence for residual serial correlation and heteroskedaticity for each of the conditionalvariance equations.W e also carry outdiagnostic tests as a sim ple indication for the presence of m odelm isspecification.In particular,w e exam ine the sign bias test statistic and the negative and positive size bias teststatistics proposed in Engle and N g (1993);the sign bias testinvestigates the im pactof positive and negative excess return shocks on volatility w hich w ere notpredicted,and the positive (negative) size bias test focuses on the effects of large and sm all positive (negative) excess return shocks not predicted by the m odel.W e find no evidence ofm isspecification,and although allthese diagnostic tests are m erely indicative, again, there does seem to be support for the G A RCH (1,1)characterization.19 Som e interesting observations can be m ade from Table 3. The conditional m ean equation for the m arket portfolio (equation 11.1) provides good evidence thatabnorm al

20

profits cannotbe m ade on the m arket,on average,and therefore provides a reasonable basis for the use of the CA PM m odel.20 H ow ever, w e also find that the financial liberalization index (reflecting dom estic i.e. interest rate liberalization, and external account i.e. capital account liberalization) is positive and significantly different from zero (w ith a p-value equalto 0.03).This suggests thatfinancialliberalization,through these policy instrum ents,had a negative effecton the (excess) m arketreturn,w hich is consistentw ith the view thata liberalized banking system m ay notnecessarily allocate investm entfunds to projects w ith the highestreturns. Exam ination of the conditional variance equations gives consistent results. The estim ated coefficients on the financial liberalization index in the conditional variance (and covariance) equations are all positive, im plying that increasing financial liberalization overthis period served to reduce conditionalvolatility and hence riskiness in the banking and financial sectors, in addition to the m arket sector. In four of these equations the coefficients are significant at the 10% level, and in the other tw o equations the p-values take values of 0.187 and 0.221. W ith regard to the estim ated coefficients on the reserve requirem ent index,the results are again interesting; all the coefficients are negative and significantly differentfrom zero atthe 8% levelexceptfor the m arket equation w here the corresponding coefficient has a m arginal significance level of approxim ately 16% . H ence, this suggests that, in general, the reductions in reserve requirem ents w ere associated w ith increases in conditionalvolatility (and hence riskiness) of (especially) the banking and financial sector stock returns. This m ay suggest that reserve requirem ents played a prudential role, preventing large shifts tow ards greater holding of risky assets in bank portfolios, thereby decreasing their riskiness.21 Table 3 also show s that the liberalization index and reserve requirem ents on dem and enter significantly in the conditional co-variances of the banking and financial sector portfolios w ith the m arketportfolio i.e.equations (11.5) and (11.6) respectively.These findings suggestthatfinancialliberalization also affected the (non-diversifiable)m arket 19

A fullsetofdiagnostic results is available from the authors on request. Furtherm ore,in a prelim inary analysis,coefficients on lags ofthe excess returns on the m arketportfolio w ere found to be insignificant. 21 This evidence is also consistentw ith G elles’s (1986) theoreticalfram ew ork in w hich an increase in the required reserve ratio decreases the ratio of banks' risky assets to total assets (a m easure of bank riskiness). 20

21

Table 3:C onditionalC A PM w ith M ultivariate G A R C H (1,1) C onditionalV ariance C ovariance M atrix Estim ated Coefficients ofthe M arketPortfolio

Conditional M ean

a0

a1

-0.0019 [0.1447] w 01

0.0021 [0.0306] a11

Conditional 0.0005 0.5192 V ariance (0.0008) (0.0000) Ljung-Box (6)forlevels = 7.333 (0.291) Ljung-Box (6)forsquares = 2.999 (0.809)

b11

d11

q11

0.0908 (0.0000)

0.0028 (0.2207)

-0.0019 (0.1576)

Estim ated Coefficients ofthe Bank ConditionalV ariance Equation

w 02

a22

Conditional 0.0018 0.4839 V ariance (0.0000) (0.0000) Ljung-Box (6)forlevels = 4.594 (0.597) Ljung-Box (6)forsquares = 0.319 (0.999)

b22

d22

q22

0.1467 (0.0000)

0.0071 (0.0321)

-0.0110 (0.0004)

Estim ated Coefficients ofthe FinancialSectorConditionalV ariance Equation

w 03

a33

Conditional 0.0012 0.5648 V ariance (0.0000) (0.0000) Ljung-Box (6)forlevels = 3.968 (0.681) Ljung-Box (6)forsquares = 0.747 (0.993)

b33

d33

q33

0.1005 (0.0000)

0.0043 (0.1875)

-0.0061 (0.0091)

Estim ated Coefficients ofthe ConditionalCovariance Equations

w 04

a44

b44

d44

q44

H 1m

0.0006 (0.0001) w 05

0.6185 (0.0000) a55

0.0744 (0.0000) b55

0.0058 (0.0035) d55

-0.0027 (0.0388) q55

H 2m

0.0006 (0.0002) w 06

0.5986 (0.0000) a66

0.0827 (0.0000) b66

0.0046 (0.0450) d66

-0.0024 (0.0826) q66

H 12

0.0013 0.5946 0.1099 0.0050 -0.0061 (0.0000) (0.0000) (0.0000) (0.1070) (0.0091) N otes:The estim ated coefficients refer to the system of equations (11.1)-(11.9) and the figures in parentheses denote m arginalsignificance levels.

risk of the K orean banking and financial sector. This is because (i) m arket risk is defined as the ratio of the conditional co-variance of the banking and financial sector portfolios w ith the m arketportfolio (i.e.equations (11.5) and (11.6) respectively) to the conditional variance of the m arket portfolio (i.e. equation 11.2); (ii) both the

22

liberalization index and reserve requirem ents enter significantly in both of these equations. Further evidence on this issue is presented in Figures 3a and 3b, w hich respectively plot the tim e-varying betas of the banking and financial sectors against tim e,during the 1987:1-1997:6 period. These figures reveal that w ith tw o exceptions the banking sectorand the financialsectordid notincrease during the sam ple period.If anything,the figures show a steady decline in the betas after 1988.The only exceptions are M arch 1994, w hen beta increased slightly and becam e highly volatile, and the period from February 1997 onw ards w hen the betas forthe banking and financialsector started to increase sharply.Itis im portantto note thatduring 1997 there w as no change in our policy indices and hence the increase in betas in the latter case cannot be attributed to changes in financial policies. Instead the increase in bank and financial riskiness m ustbe attributed to ‘bad new s’,both from the region and K orea itself – the collapse of som e of the largest chaebols such as K IA M otors - w hich increased substantially the volatility ofthe stock m arket. 6.A nalysis and C oncluding R em arks O ur em pirical findings suggest that financial liberalization reduced banking and financialrisks,as im plied by the significance of the coefficients of the policy m easures in the conditional variance and co-variance equations. The em pirical analysis also suggests thatfinancialliberalization,w ith tw o exceptions,reduced the non-diversifiable m arketrisk ofthe banking and financialsector. In fact,ourfindings suggestthatm arket risk only began to increase in early 1997,w hich coincides w ith ‘bad new s’in the period prior to the crisis. Thus, the econom etric findings contrast sharply w ith the ex-post qualitative survey findings,w hich dem onstrate thatfinancialinstitutions in factbecam e exposed to greater risks, through a com bination of inadequate risk m anagem ent system s,lack ofexpertise and w eaknesses in prudentialregulation. The tw o sets of contrasting findings can be reconciled, in that the survey findings are clearly ex-post,having the benefitofhindsightw hich included an expertanatom y ofthe crisis, w hile the econom etric findings to a large extent reflect the ex-ante view s of m arketparticipants,based on available inform ation atthattim e as w ellas their beliefs concerning the effects financial liberalization. In this sense, the econom etric findings indicate thatfinancialm arketparticipants had traditionalview s,w hich over-em phasize

23

the benefits of financial liberalization and under-em phasize the pitfalls. A dditionally, they indicate thatinform ation flow s from financialinstitutions to financialm arkets w ere too slow or even inaccurate,as a resultof (now ) w ellknow n w eaknesses in corporate governance,bad accounting practices and com plex com pany linkages. Thus,itis likely that traditional beliefs w ould have been shattered m uch earlier had ‘bad new s’ concerning poorrisk m anagem entpractices hitthe m arkets sooner. Im portantly our surveys also revealthatexpertopinion – atleastin Korea,the IM F and the W orld Bank – has now shifted, acknow ledging that, even though financial liberalization m ay in principle offer potential benefits (such as greater investm ent returns and opportunities for diversification), its practical im plem entation results in greatly increased risks because of w eaknesses in risk m anagem ent and prudential regulation. A dditional inform ation from our surveys suggests that the safety nets that have historically been provided by successive K orean governm ents to banks and industry m ay w ell have been responsible for holding back necessary im provem ents in risk m anagem entand prudentialregulation. Im plicitorexplicitsafety nets clearly actas disincentives in m anaging risks,representing a certain type of m oralhazard,albeitof a m ilder form than the one postulated by M cK innon and Pill (w hich posits that banks m islead investors in order to deliberately take advantage of safety nets). This form of m oral hazard w as critical in creating vulnerabilities in the banking system , including currency and m aturity m ism atches,w hich broughtthe K orean econom y to a stage w here even sm allshocks could triggera full-blow n financialcrisis. A conjecture that em erges from our analysis is that traditional beliefs concerning financial l iberalization, w hich over-em phasize efficiency gains and under-em phasize risks,m ay w ellhave been responsible for the thesis’ failure,by holding back necessary im provem ents in both the m anagem ent of financial risks by financial institutions and prudential regulation. W hile this m ay, for som e, be itself a som ew hat speculative conjecture,itis certainly one thatopens up fruitfulavenues forfurtherresearch.

R eferences:

A kerlof, G . and P. Rom er (1993), “Looting: The Econom ics U nderw orld of Bankruptcy for Profit”,Brooking Paperson Econom ic Activity,Septem ber,1-60.

24

A llen,F.and D .G ale (2000),“Bubbles and Crises”,The Econom ic Journal,Vol.110,236 -255. A restis, P., and P.O . D em etriades (1997), “Financial D evelopm ent and Econom ic G row th: A ssessing the Evidence”, Econom ic Journal,V ol.107,N o.442 (M ay),783-799,reprinted in H .D .D ixon (ed.)Controversies in M acroeconom ics,O xford:Blackw ellPublishers,2000. A restis, P., and P.O . D em etriades (1999), “Financial Liberalization: The Experience of D eveloping Econom ies”,Eastern Econom ic Journal,V ol.25,N o.4 (Fall),441-457. Black, F. (1972), " Capital M arket Equilibrium w ith Restricted Borrow ing", Journal of Business,V ol.45,3-18. Bollerslev, T., R.F.Engle, and J. W ooldridge (1988), " A Capital Asset Pricing M odel w ith Tim e V arying Covariances",JournalofPoliticalEconom y,Vol.96,116-131. Bollerslev,T.and J.W ooldridge (1992),"Q uasi-m axim um Likelihood Estim ation and Inference in D ynam ic M odels w ith Tim e V arying Covariances”,Econom etric Review,Vol.11,143-172. Bonom o,M .and R.G arcia,(1997)," Tests ofConditionalA ssetPricing m odels in the Brazilian Stock M arket",CIRA N O Scientific Series,M ontreal:CIRA N O . Caprio, G ., and L. Sum m ers (1994) “Financial Liberalization: Beyond Laissez-Faire”,W orld Bank Policy Research W orking Papers. D em etriades,P.O .(1999) “FinancialLiberalization and Credit-assetBoom s and Busts in East A sia”, presented at Conference "The Credit Crunch in East A sia", W ashington D .C., W orld Bank,N ovem ber(http://w orldbank.org/research/interest/confs/past/papersnov30-d). D em etriades,P.O .,and B.Fattouh (1999b),“The South K orean Financial Crisis: Com peting Explanations and Policy Lessons for Financial Liberalization”, International Affairs,Vol.75, N o.4 (O ctober),779-792. D em etriades, P.O . and K . B. Luintel (1997), “The D irect Costs of Financial Repression: Evidence from India”,Review ofEconom icsand Statistics,Vol.79,No.2 (M ay),311-320. Dem irgüç-K unt, A . and E. D etragiache (1998), “Financial Liberalization and Financial Fragility”,InternationalM onetary Fund W orking Paper,98/83. Edw ards,S.(1997),“ The M exican Peso Crisis? H ow M uch D id W e K now ? W hen D id W e K now It? The W orld Econom y, Edw ards,S.(1999),“ H ow Effective are CapitalControls?” JournalofEconom ic Perspectives, Fall,V ol.13,65-84. Engle,R.and K .K roner (1993)," M ultivariate Sim ultaneous G eneralised A RCH ",D iscussion PaperN o.89,D epartm entofEconom ics:U niversity ofCalifornia,San D iego. Fischer, K .P. and M . Chenard (1997), “ Financial Liberalization Causes Banking System s Fragility.” Centre de Recherché en Econom ie etFinance A ppliqués,U niversity Laval(Canada), W orking PaperN o.97-12,June. Fry,M .J.(1997),“In FavourofFinancialLiberalization”,Econom ic Journal,V ol.107,N o.442 (M ay),754-770. G elles,G .(1991)," Som e Regulatory D eterm inants ofBank Risk Behavior:Com m ent",Journal ofM oney,Creditand Banking,Vol.23,120-127. H ellm an,T.F.,K .M urdock,and J.E.Stiglitz (2000),“ Liberalization,M oralH azard in Banking, and Prudential Regulation: A re Capital Requirem ents Enough?”, The Am erican Econom ic Review ,Vol.90,147-165. H ogan,W .P.and I.G .Sharpe (1984)," Regulation,Risk,and the Pricing of A ustralian Bank Shares",Econom ic Record,Vol.60,34-44.

25

H unter, W .C., G .G.K aufm an and T.H .K rueger (eds.) The Asian FinancialCrisis: O rigins, Im plicationsand Solutions,Boston:K luw erA cadem ic Publishers. Laurens, B and J. Cardoso, “Capital Flow s-Lessons from the Experience of Chile”, IM F W orking Paper W P/98/168,W ashington D .C.:InternationalM onetary Fund. Linter,J.(1965)," The valuation ofRisky A ssets and the Selection ofRisky Investm ents in the Portfoliosand CapitalBudgets",Review ofEconom icsand Statistics,Vol.47,13-37. M cK innon, R.I. (1973), M oney and Capital in Econom ic D evelopm ent. W ashington D .C.: Brooking Institutions. M cK innon, R.I. (1981), "Financial Repression and the Liberalization Problem in Less D eveloped Countries",in S.G rassm an and E.Lundberg (eds),The Pastand Prospects of the W orld Econom ic O rder,M acm illan,356-386. M cK innon, R. and H . Pill, (1997) “Credible Econom ic Liberalization and O verborrow ing”, Am erican Econom ic Review ,Vol.87,189-193. M itchell,D .(1986),"Som e Regulatory D eterm inants ofBank Risk Behavior:A N ote",Journal ofM oney,Creditand Banking,V ol.18,Issue3,374-382 Ng, L. (1991), " Tests of CAPM with Tim e-V arying Covariances: A M ultivariate G A RCH A pproach",JournalofFinance,Vol.46,1507-1521. O bstfeld,M .and K .Rogoff(1996),FoundationsofInternationalM acroeconom ics.Cam bridge, M assachusetts:The M IT Press. Scheicher, M . (1996), " A sset Pricing w ith Tim e V arying Covariances: Evidence from the G erm an Stock M arket", Paper presented at the 1996 European M eeting of the Econom etric Society,Istanbul. Sharpe,W .(1964)," CapitalA ssetPrices:A Theory of M arketEquilibrium U nder Conditions ofRisk",JournalofFinance,Vol.19,425-442. Shaw , E.S., (1973), Financial D eepening in Econom ic D evelopm ent, N ew Y ork: O xford U niversity Press. Stiglitz, J.E. (1994),"The Role of the State in Financial M arkets",Proceedings ofthe W orld Bank AnnualConference on D evelopm entEconom ics 1993,19-52. Stiglitz,J.E.(2000),"CapitalM arketLiberalization,Econom ic G row th,and Instability",W orld D evelopm ent,V ol.28,N o.6,(2000)pp.1075-1086.

26

19 87 19 :01:1 87 4 19 :06:2 87 4 19 :12:0 88 2 19 :05: 88 11 19 :10:1 89 9 19 :03: 89 29 19 :09:0 90 6 19 :02: 90 14 19 :07:2 91 5 19 :01: 91 02 19 :06:1 91 2 19 :11:2 92 0 19 :04:2 92 9 19 :10:0 93 7 19 :03:1 93 7 19 :08:2 94 5 19 :02: 94 02 19 :07:1 94 3 19 :12: 95 21 19 :05:3 95 1 19 :11: 96 08 19 :04:1 96 7 19 :09: 97 25 :03 :05

19 87 :0 19 1:14 87 :0 19 6:24 87 :1 19 2:02 88 :0 19 5:11 88 :10 19 :19 89 :0 19 3:29 89 :09 19 :06 90 :0 19 2:14 90 :0 19 7:25 91 :0 19 1:02 91 :06 19 :12 91 :1 19 1:20 92 :0 19 4:29 92 :1 19 0:07 93 :0 19 3:17 93 :0 19 8:25 94 :02 19 :02 94 :0 19 7:13 94 :12 19 :21 95 :0 19 5:31 95 :1 19 1:08 96 :0 19 4:17 96 :0 19 9:25 97 :03 :05

19 87 19 :01:1 87 4 19 :06:2 87 4 19 :12:0 88 2 19 :05:1 88 1 19 :10:1 89 9 19 :03: 89 29 19 :09:0 90 6 19 :02:1 90 4 19 :07:2 91 5 19 :01:0 91 2 19 :06:1 91 2 19 :11:2 92 0 19 :04: 92 29 19 :10:0 93 7 19 :03:1 93 7 19 :08:2 94 5 19 :02:0 94 2 19 :07: 94 13 19 :12:2 95 1 19 :05:3 95 1 19 :11:0 96 8 19 :04:1 96 7 19 :09:2 97 5 :03 :05

Figure 1a:M arketExcess R eturn (localcurrency)

0.2

0.15

0.1

0.05

0

-0.05

-0.1

-0.15

Figure 1b: B anking Index E xcess R eturn (local currency)

0.3

0.25

0.2

0.15

0.1

0.05

0

-0.05

-0.1

-0.15

-0.2

Figure 1c:FinancialSector Index Excess R eturn (localcurrency)

0.3

0.25

0.2

0.15

0.1

0.05

0

-0.05

-0.1

-0.15

27

R eserve R equirem ent R atios

0

28

1997.03.18

1996.09.03

1996.02.21

1995.08.08

1995.01.24

1994.07.12

1993.12.28

1993.06.15

1992.12.01

1992.05.19

1991.11.05

1991.04.23

1990.10.10

1990.03.27

1989.09.12

1989.02.28

1988.08.16

1996:12:17

1996:04:16

1995:08:15

1994:04:12 1994:12:13

1993:08:10

1992:12:08

1992:04:07

1991:08:06

1990:12:04

1989:08:01 1990:04:03

1988:11:29

1988:03:29

1987:07:28

1986:11:25

1.2

1988.02.02

1987.07.21

1986:03:25

1995:05:09 1995:11:21 1996:06:04 1996:12:17 1997:07:01

1989:06:13 1989:12:26 1990:07:10 1991:01:22 1991:08:06 1992:02:18 1992:09:01 1993:03:16 1993:09:28 1994:04:12 1994:10:25

1986:03:25 1986:10:07 1987:04:21 1987:11:03 1988:05:17 1988:11:29

0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

1987.01.06

Figure 2a:InterestRate Liberalisation Index

Figure 2b:CapitalAccountLiberalisation Index

1

0.8

0.6

0.4

0.2

0

Figure 2c:Reserve Requirem ents on Dem and Deposits

20

15

10

5

0.8

29 1997:03:26

1996:09:11

1996:02:28

1995:08:16

1995:02:01

1994:07:20

1994:01:05

1993:06:23

1992:12:09

1992:05:27

1991:11:13

1991:05:01

1990:10:17

1990:04:04

1989:09:20

1989:03:08

1988:08:24

1988:02:10

1987:07:29

1997:05:21

1996:11:13

1996:05:08

1995:11:01

1995:04:26

1994:10:19

1994:04:13

1993:10:06

1993:03:31

1992:09:23

1992:03:18

1991:09:11

1991:03:06

1990:08:29

1990:02:21

1989:08:16

1989:02:08

1988:08:03

1988:01:27

1987:07:22

1987:01:14

0.4

1987:01:14

Beta

Beta

Figure 3a:B eta for the B anking Sector

1.8

1.6

1.4

1.2

1

0.8

0.6

Figure 3b:B eta for the FinancialSector

1.6

1.4

1.2

1

A ppendix:A Sum m ary ofthe M ain FinancialR eform s in K orea 1.InterestRates Septem ber 6, 1979: The M onetary Board abolished the existing m axim um interest rate on bank loans to m ake it possible for banks to alter their interestrate on loans.H ow ever,the Bankers association of K orea,considering that banks them selves are not used to determ ining interest rates, decided to link interest rates on loans to the central bank’s rediscountrate. Septem ber 6,1979: The m onetary board abolished the m axim um interestrate on free installm entsavings deposits and the m axim um interestrate on personalchecking deposits. M ay 17,1984:The Board allow ed seven nationw ide com m ercialbanks,localbanks and the K orea Exchange Bank to engage in the negotiable certificate ofdeposit(CD s)from 1 June. 23 July,1984:A narrow band for loan rates w as introduced so thatbanks could charge differentrates according to the creditw orthinessofthe borrow ers. D ecem ber 5,1988:Interestrate on loans from banks and non-bank financialinterm ediaries w ere fully liberalized. D ecem ber 5,1988: Interestrate on tim e deposits of m aturity greater than 2 years atbanks,postalsavings and credit unions w ere liberalized. D ecem ber 5,1988: Interestrate on tim e and savings deposits of m aturity greater than 1 year atm utualsavings and finance com panies w ere liberalized. N ovem ber 21, 1991: Lending rates liberalized further. Lending rates liberalized consisted of those on bank overdrafts;on the discountof com m ercialbills by banks,m utualsavings and finance com panies;on the discountof com m ercialand trade bills by investm entand finance com panies;on the purchase of firm s’ guaranteed papers by banks’trustaccounts;and those on overdue loans by allfinancialinstitutions. N ovem ber 21,1991:The liberalization ofdepositrates applied to those on shortterm ,large denom ination m arketable instrum ents such as CD s,the sale oflarge denom ination trade bills,com m ercialpapers and RPs. N ovem ber 21, 1991: The scope of initial liberalization w as extended to cover rates on long-term deposits w ith a m aturity of 3 years offered by banks, m utual credit facilities, and credit unions and m utual tim e deposits w ith a m aturity of2 years and m ore offered by m utualsavings and finance com panies. N ovem ber 21,1991:The issue rates ofcorporate bonds w ith a m aturity of2 years and m ore w ere deregulated. N ovem ber 1, 1993: A ll lending rates (apart from those financed by the governm ent and the bank of K orea’s rediscounts)w ere liberalized. N ovem ber 1,1993:Rates on long-term depositsw ith a m aturity ofatleasttw o yearsw ere com pletely liberalized. N ovem ber 1,1993:Interestrate on debentures and corporate bonds w ith a m aturity < 2 years w ere liberalized. D ecem ber 1,1994:Interestrate on bank and non-bank tim e deposits w ith a m aturity ofone yearorm ore butless than 2 years w ere liberalized. D ecem ber 1, 1994: Banks w ere perm itted to set freely the interest rates on policy loans financed through the aggregate creditceilings system w ithin theirrespective prim e rates. July 24, 1995: Interest rate on policy-based loans through the aggregate credit ceilings system of BO K w ere liberalized. N ovem ber 20,1995:The Bank and governm entfreed up the rem aining regulated interestrates on bank and non-bank tim e deposits w ith a m aturity less than six m onths and on theirinstallm entdeposits w ith m aturity lessthan one year. January 19,1996: The Bank of K orea lifted the restriction on the size of the prem ium a bank could charge over its prim e-lending rate,w hich had been originally im posed in orderto preventa sharp run-up in bank lending rates in the course ofinterestrate deregulation.

30

2.D evelopm ents in M oney M arkets: M arch 7, 1986: The M onetary Board liberalized the rates on negotiable CD s, secured corporate bonds, and bank debentures. February 13,1987:The M B reduced the denom ination ofCD s from 100 m illion w on to 50 m illion w on. D ecem ber 5,1988:Interestrate on repurchase agreem ents,com m ercialpapers of face value greater than 30 m illion and m aturity m ore than 91 days),financialdebentures and corporate bonds w ere fully liberalized. D ecem ber 5,1988:N ew com m ercialpaperand conventionalcom m ercialpaperw ere m erged into one. O ctober 4,1989: The Bank and the governm entm erged the callm arkets,previously segm ented into an inter-bank m arketm ainly forbanks and overthe counterm arketbetw een N BFIs,w hich expanded the size ofthe m oney m arket (callm arkets,CPs,CD s,RPs,TBs,Bankers’A cceptance).A fterthe m erger,the interbank rate w as fully liberalized. O ctober 19,1989:The BO K adjusted the m aturity period of CD s issued by banks to other banking institutions from betw een 91 days and 180 days to betw een 30 days and 180 days. N ovem ber 21,1991:The liberalization ofdepositrates applied to those on shortterm ,large denom ination m arketable instrum ents such as CD s,the sale oflarge denom ination trade bills,com m ercialpapers and RPs. N ovem ber 21, 1991: The issue rates of corporate bonds w ith a m aturity of 2 years and m ore w ere com pletely deregulated. D ecem ber 19,1992:The Bank extended the m axim um m aturities ofCD s from 180 days to 270 days. N ovem ber 1,1993: Interestrate on financialdebentures and those corporate bonds w ith a m aturity of less than 2 years w ere liberalized.G overnm entand public bonds and M SB w ere also to be issued atprevailing m arketrates. Septem ber 3,1993:The Bank low ered the m inim um denom ination ofCD s from 50 m illion to 30 m illion. July 18,1994:The m inim um m aturities ofCD s,high denom ination RPs w ere shortened from 91 days to 60 days. July 24,1995: The m inim um m aturities of shortterm financialinstrum ents including CD s,high value RPs and high value CPs,w ere shortened from 60 to 30 days.

3.Portfolio Inflows D ecem ber 1, 1989: Foreign exchange banks w ere allow ed to raise offshore funds by issuing foreign currency denom inated bonds orborrow ing from the offshore accounts ofotherdom estic foreign exchange banks. M arch 1,1991: N on-residentK orean w ere allow ed to sellforeign currencies exceeding U S$ 50 thousand to entrust its proceeds to developm enttrusts w ith a m aturity of2 years.Effective 15 July,the lim itw as raised in July 15 to 100 U S$ 100 thousands. Effective from M arch 8,1991:The governm entperm itted the issuance offoreign currency denom inated securities to finance the im portofproduction facilities and equipm entforw hich no dom estic substitute isavailable. January 3,1992: N on-residents w ere allow ed to investin any dom estic stocks,unless specified in som e particular act. A 3% lim it on investm ent by an individual foreign and 10% lim it on total foreign investm ent are applied respectively.In case ofpublic utilities and com panies in infantindustry,totalforeign investm entlim itis setat8% . Septem ber 1, 1992: Regulations on overseas issue of foreign currency denom inated securities w ere greatly eased. Type ofsecurities m ay be issued abroad by K orean residents w ere expanded to include negotiable CD s and CPs. Septem ber 1, 1992: Funds raised by the issue of foreign currency denom inated securities w ere perm itted to be deposited eitherin a residentaccountoran accountw ith overseas branch ofa dom estic exchange bank. Feb 1,1993: O verseas branches of dom estic banks w ere in principle prohibited from supplying loans to residents of K orea.Butfrom the above date,the governm entperm itted them to extend loans to residents ofK orea engaged in the trading ofthe com m odity futures orfinancialfutures.

31

April1,1993: The governm entchanged the system w hereby perm ission has to be obtained from foreign currency denom inated deposits could be issued to a reporting system .Instead of obtaining perm ission before issuing foreign currency denom inated securities,now itonly needed to be reported. April1,1993:Previously restricted to enterprises thathad recorded a netprofits in each of previous 3 years,issuers w ere w idened to include those thathad recorded a netprofiton an accum ulative basis overthe preceding 3 years. July 1,1994:The governm entpartially opened the dom estic bond m arketallow ing non-residents to purchase nonguaranteed convertible bonds issues by SM Es subjectto 30% lim iton totalforeign investm entper issue and a 5% lim itperissue on investm entby individualfirm s. D ecem ber 1,1994: The ceiling on overallforeign investm entin a listed com pany’s outstanding stocks w ere raised from 10% to 12% . M ay 3,1995: Firm s w ere perm itted to undertake foreign borrow ings directly for the redem ption of foreign debts on onerous term forthe im portfacilities w here the firm s w ere sm alland m edium sized m anufactures. June 20,1995:The governm entperm itted the overseas issuance ofexchangeable bonds. July 1,1995:The generalceiling on totalforeign investm entin a listed com pany’s outstanding stock w as raised from 12% to 15% and thatforthose ofpublic corporations from 8% to 10% . April1,1996: The aggregate ceiling w as raised from 15% to 18% of the outstanding stocks issued by a listed firm and from 10% to 12% for those issued by public corporation. The ceiling on holding of individual investors w as raised from 3% to 4% ofa firm ’s outstanding stocks. M ay 1,1996:N on-residents w ere allow ed to purchase and trade bonds w ith w arrants. M ay 3,1996:N on-residents w ere allow ed to trade stock index futures on the K SE. O ctober 1, 1996: The general ceiling w as raised again to 20% for a firm and 15% for a public corporation. The individualceiling w as atthe sam e tim e increased to 5% . O ctober 1, 1996: The governm ent dism antled m ost restrictions on direct foreign borrow ings, enabling nonm anufacturing SM Es to receive loans from abroad forthe im portofproduction facilities. M ay 1,1997:The lim iton foreign ow nership ofK orean equities w as raised to 23% . June 1,1997:Foreign investors w ere granted access to non-guaranteed bonds ofSM Es and ofconglom erates.

4.Reserve Requirem ents on D em and D eposits Effective N ovem ber 23,1987:The M onetary Board raised the m inim um reserve requirem entfrom 4.5% to 7.0% . April20,1989:A m arginalreserve requirem entratio of30% on the average increm entofdem and deposits and tim e and saving deposits has been im posed.The m arginalreserve requirem ents w ere abolished in February 1990. 15 February 1990: The Bank raised reserve requirem entratios on tim e deposits,instalm entsavings deposits w ith m aturity of 2 years or m ore and H ousehold instalm ent saving deposit from 7.0% to 8.0% . O n all other deposits, reserve requirem entratio increased from 10% to 11.5% . February 8,1991:The Bank introduced reserve requirem ents againstm utualinstalm entdeposits. April23,1996:The reserve requirem enton tim e and savings deposits of m ore than 2 years w as broughtdow n from 8% to 6% . April23,1996:The reserve requirem enton checking deposits,pass book deposits,saving deposits,Tim e and saving depositsw ith m aturity oflessthan tw o yearsw as low ered from 11.5% to 9.0% . N ovem ber 8,1996: The reserve requirem enton tim e and savings deposits of m ore than 2 years w as broughtdow n from 6% to 4% . N ovem ber 8,1996: The reserve requirem enton checking deposits,pass book deposits,saving deposits,Tim e and saving depositsw ith m aturity oflessthan tw o yearsw aslow ered from 9.0% to 7.0% .

32