banks, the imf, and the asian crisis
DESCRIPTION
journalTRANSCRIPT
Ž .Pacific-Basin Finance Journal 8 2000 177–216www.elsevier.comrlocatereconbase
Banks, the IMF, and the Asian crisis
Bong-Chan Kho a, Rene M. Stulz b,c,)´a Seoul National UniÕersity, College of Business Administration, Seoul, South Korea
b Department of Finance, Fisher College of Business, The Ohio State UniÕersity, 2100 Neil AÕenue,43210-1144, Columbus, OH, USA
c National Bureau of Economic Research, 1050 Massachusetts AÕenue, 02138, Cambridge, MA, USA
Abstract
This paper examines the impact of the Asian crisis on bank stocks. In the second half of1997, Western banks outperformed their stock markets. In contrast, East Asian bank indicesincurred losses in excess of 60% in each of the crisis countries. Most of these poorperformances are explained by stock market movements in the crisis countries. After takinginto account these movements, currency exposures affected banks adversely only inIndonesia and the Philippines. Except for the Korean program, which affected positivelybank stocks in all countries in our sample but one, IMF programs had little effect on bankvalues. q 2000 Elsevier Science B.V. All rights reserved.
JEL classification: F33; F36; F42; G15; G21
Keywords: Banks; IMF; Asian crisis
1. Introduction
For most observers, banks have been at the heart of the Asian crisis. ForŽ .instance, Hamann 1999, p. 9 states that ‘‘the Asian crisis differed from previous
financial crises that created a need for the IMF’s assistance. It was rootedprimarily in financial system vulnerabilities and other structural weaknesses.’’
) Corresponding author. Department of Finance, Fisher College of Business, The Ohio StateUniversity, 2100 Neil Avenue, Columbus, OH 43210-1144, USA. Tel.: q1-614-292-1970; fax:q1-614-292-2359.
Ž .E-mail address: [email protected] R.M. Stulz .
0927-538Xr00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved.Ž .PII: S0927-538X 00 00007-X
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216178
However, the reasons given for the importance of banks in this crisis differ widelyacross observers. For some, currency crises led to banking crises in the affectedcountries. With this view, banks had accumulated large currency exposures basedon the belief that there was little exchange rate risk. When exchange ratescollapsed, they suffered large losses on their currency exposures. For others, bankswere one important contributing factor to the Asian crisis. Asian local banks areaccused of making too many unsound loans and moral hazard is blamed for this
Ž .behavior. Delhaise 1998, p. 35 argues that ‘‘It was generally accepted before thecrisis that most banks would be rescued if they ran into trouble.’’ Western banksare blamed for first lending too much and then for contributing to the credit crunchby lending too little. For instance, Wolf states that the East Asian banking crisiswas ‘‘promoted by overgenerous lending from financial institutions in advancedcountries.’’1 The IMF and governmental bailouts have been blamed for creating
Ž .incentives for banks to take on too much risk, including foreign exchange FXrate risk.2 As one observer puts it, ‘‘These bankers took the opportunity to makevery risky, profitable loans, knowing that if the loans went bad, the IMF or the USgovernment would bail them out.’’3
These various views of the Asian crisis raise important questions: Did bankshareholders get hurt because of the crisis? Did the crisis pose a threat to thebanking systems in Western countries? Can exchange rate changes explain theperformance of Asian banks? Did specific events in the Asian crisis affect bankshareholders? How were banks affected by the announcement of IMF programs?Did IMF programs have systemic benefits or did they help only those banks withexposures in the countries benefiting from the programs? To examine thesequestions, we examine the returns to bank shareholders from January 15, 1997 toJuly 15, 1998. Our examination uses Datastream banking indices for four Western
Ž .countries the US, France, Germany, and the UK and for six Asian countriesŽ .Indonesia, Japan, Korea, Malaysia, Philippines, and Thailand . We also investi-gate the returns of the three US banks that took a lead role in the renegotiations ofKorean debt, namely the Chase Manhattan Bank, Citibank, and JP Morgan.
We find that during our sample period, shareholders of East Asian banksincurred dramatic losses. For instance, an investor who had invested US$1 at thestart of our sample period in Korea’s bank index would be left at the end of oursample period with 14.7 cents. An investor who had invested US$1 in Indonesia’sbank index would be left with 3.3 cents. The story is very different for theWestern banks. An investor who invested US$1 at the start of our sample period in
1 Financial Times, Wednesday, October 21, 1998, p. 14.2 Ž .In a recent paper, Burnside et al. 1999 develop a theoretical model where implicit guarantees
make it advantageous for banks not to hedge foreign currency exposures arising from their financing.3 The quote is from Chung Hoon Lee, president of the Korea America Economic Association, in
Economists blame short-term loans for Asian crisis by Louis Uchitelle, New York Times, January 8,1999.
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216 179
the US bank index would have had US$1.73 at the end of our sample period. Incontrast, an investment of US$1 in the US market index at the start of our sampleperiod would have netted US$1.54 at the end of our sample period. A simpleexplanation for why Western banks were not affected more by the crisis is thattheir exposures were small enough that the impact of the crisis was offset by goodnews in other parts of their businesses. With this explanation, if the East Asiancrisis mattered at all for Western banks, we should find that on days of adverseevents in East Asia, Western bank stocks should have performed poorly.
We then try to understand why the performance of East Asian bank commonstocks during our sample period is so poor. We regress dollar bank excess returnson stock market excess returns, currency excess returns, and interest rate changes.Even though East Asian banks perform poorly, banks in Korea, Indonesia, and
Ž .Thailand do not have abnormal returns ARs over the period from July 1997 tothe end of January 1998 once we account for the performance of their nationalmarkets. An explanation advanced in the assessments of the Asian crisis is thatbanks had large currency exposures because of their use of offshore funding. Thisview has led some to argue that banks should be required to borrow and lend inthe same currency.4 We investigate this explanation by estimating exposures ofbank indices to exchange rate changes. Our estimates show that after taking intoaccount market returns, currency returns do not seem to contribute to the poorperformance of East Asian banks except for Indonesia and the Philippines. It isimportant to understand that these estimates do not mean that exchange ratechanges did not have an adverse impact on banks. We show that decreases in thevalue of East Asian currencies affect stock markets adversely, so that adversecurrency movements affect banks through their impact on stock markets. Ourresults mean, however, that the currency crises did not have an impact on banksbeyond their overall impact on the economy, so that there was nothing uniqueabout the exposure of banks to exchange rates.
Since Western banks performed well over our sample period and since EastAsian banks performed poorly, we try to understand the impact of the crisis byexamining how bank stocks were affected by various events of the crisis. Such anapproach might allow us to find traces of the crisis in the returns of Western banksthat get swamped by positive news over our sample period and to find whichevents can explain the poor performance of Asian banks. We select events over thewhole sample period that were important in the chronology of the Asian crisis andinvestigate whether they are associated with significant ARs for banks. We findlittle evidence of important impacts of Asian crisis events on banks acrosscountries with this approach. There are only five event periods where we canreject the hypothesis that banks have ARs equal to zero across countries with ap-value of 0.05 or better. The only period in 1997 corresponds to the announce-
4 Ž .See Hall’s comments in Furman and Stiglitz 1998, p. 124 .
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216180
ment of the IMF program in Korea. The other periods are in January 1998. One ofthese periods in 1998 coincides with the Peregrine debacle and another with theIMF agreement with Indonesia. The announcement of the IMF program in Koreahad a positive impact on banks across most countries, which is inconsistent withthe view that bailouts are fully anticipated. However, the impact of IMF actionsfor US banks is large only for the banks with the highest exposures. IMF actions,therefore, do not appear to have significant systemic effects on Western banks.Rather, they simply ensure that banks with exposure are more likely to be repaidwithout benefiting banks that do not have exposures.
The paper is organized as follows. In Section 2, we present our data and discussthe returns of banks over our sample period. We explore the exposures of banks toexchange rates and interest rates, as well as the exposure of Western banks to EastAsia. In Section 3, we consider the returns of banks over key events during theAsian crisis. In Section 4, we investigate the returns of individual American banks.We conclude in Section 5.
2. Bank returns over the sample period
We consider a sample period that starts on January 15, 1997 and ends on July15, 1998. This period of exactly 18 months includes all the important events of theAsian crisis. The data we use consist of the historical Datastream retail bankingand market indices for the sample countries.5 The Datastream industry indices areproduced according to the same criteria across countries and are therefore compa-rable. These indices are value-weighted. They are not comprehensive and arecomposed of the larger firms. Throughout the study, we also use data on exchangerates and on interest rates. Table 1 provides a summary of the data we use. Fig. 1shows the evolution of the banking indices in dollars during our sample period.
The lesson from Table 1 is that Western banks did well during our sampleperiod while Asian banks performed poorly. For all Western countries, the averagereturns on the banking indices in excess of the risk-free rate were positive duringour sample period. Furthermore, for these countries, the average returns on thebanking indices exceeded the average returns on the country indices. The bottomline from this is that there is no evidence that the Asian crisis affected the Westernbanks in a way that their shareholders would have suffered. The opposite is thecase for the banks in the East Asian countries. In these countries, the averagereturns of the banking indices were negative and were lower than the returns of the
5 Datastream has two different series of sector indices. One series is recalculated as indexcomponents changes and the other series is not recalculated. The series that are not recalculated usedhere have neither a survival bias nor a backfilling bias over our sample period because they wereobtained before April 1999. In April 1999, Datastream reconstructed the series in a way that created abackfilling bias and a survival bias.
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216 181
market indices. Japan was not a crisis country, but Japanese banks had negativereturns lower than the negative returns of the Japanese market during our sampleperiod. Consequently, the experience of Japanese banks was more similar to theexperience of banks in crisis countries than it was to the experience of banks inWestern countries.
Exposures of developed country banks to Korea, Indonesia, Malaysia andThailand were US$90 billion for European Union banks, US$20 billion for USbanks, and US$85 billion for Japanese banks at the end of 1997.6 Roughly half theexposures were short-term loans in the middle of 1997. The total assets of thebanks from the US, the European Union, and Japan in the middle of 1997 asreported by the IMF were of the order of US$20 trillion. Exposures to the crisiscountries were therefore less than 1% of assets. In terms of capital, Japanese bankshad the highest exposures, which raises the question we attempt to answer later ofwhether their poor performance during the crisis can be explained by the fact thatthey had higher exposures than the Western banks. Some estimates show that theexposure to Asian emerging markets for Japanese banks was in excess of theircapital. In contrast, the exposure of US banks to Asian emerging markets wasabout 30% of capital in the middle of 1997. This was less than for German banksŽ . Ž .60% or French banks 45% . Another way to evaluate these exposures is asfractions of the market value of the equity of the banks that form the Datastreamindices. From this perspective, a complete loss of the loans from US banks toIndonesia, Korea, Malaysia and Thailand existing at the end of June 1997 wouldhave amounted to a loss of less than 4% of the market value of the equity of allthe US banks that belong to the Datastream index of retail banks. For theEuropean Union banks, the loss would have been less than one-fourth of theirequity value. Finally, it would have been less than one-sixth of the equity value ofthe Japanese banks. Obviously, a total loss of the loans was never a possibility, soour exposure estimates provide upper bounds of the losses if the banks’ exposuresarise only from loans. It is important to note that all these exposures are computedusing loans only. For banks, off-balance sheet exposures have grown dramatically.There is no data on the off-balance sheet exposures of banks to the crisis countries.However, some banks made large losses on derivatives whose counterparties werein crisis countries.
The returns of the Asian banking indices were dramatically different from thereturns of Western banking indices. For all Asian countries, including Japan,banking indices had negative daily average returns as shown in Table 1 so thattheir shareholders experienced losses. These losses were particularly substantialfor the crisis countries and exceeded the losses on the market index in Asiancountries. Panel B of Table 1 compares total losses of the indices for the five crisis
6 Ž .See IMF 1998, pp. 134–135 .
()
B.-C
.Kho,R
.M.Stulzr
Pacific-B
asinF
inanceJournal8
2000177
–216
182
Table 1Ž . Ž .Summary statistics in percent of daily equity excess returns and FX excess returns for the period from January 15, 1997 to July 15, 1998 390 days
Equity indices are from the Datastream Global indices, and their excess returns are calculated as logarithmic daily dollar returns in excess of the 1-day returnon the 7-day Eurodollar deposit. FX rates are also from the Datastream quoted by the Reuter, and their excess return is calculated as the 1-day interest rate ofthat currency compounded by the FX rate change relative to the US dollar, minus the 1-day return on the 7-day Eurodollar deposit.
Panel A. Summary statistics of variables used
Variables Mean Standard Minimum Maximumdeviation
UK equity market index excess return 0.080 0.801 y3.062 3.326UK banking industry index excess return 0.090 1.459 y5.102 4.931Germany equity market index excess return 0.112 1.156 y5.453 4.397Germany banking industry index excess return 0.143 1.528 y6.160 7.258France equity market index excess return 0.107 1.038 y3.879 4.954French banking industry index excess return 0.175 1.550 y4.885 5.612US equity market index excess return 0.096 0.981 y6.839 4.284US banking industry index excess return 0.126 1.315 y6.955 4.046Japan equity market index excess return y0.076 1.616 y5.889 7.487Japan banking industry index excess return y0.151 2.497 y8.664 12.066Korea equity market index excess return y0.287 4.394 y21.753 26.896Korea banking industry index excess return y0.507 4.886 y21.806 25.899Indonesia equity market index excess return y0.539 5.682 y39.700 23.098Indonesia banking industry index excess return y0.888 7.127 y36.973 33.026Thailand equity market index excess return y0.406 3.583 y15.379 15.436Thailand banking industry index excess return y0.476 4.309 y14.674 20.985
()
B.-C
.Kho,R
.M.Stulzr
Pacific-B
asinF
inanceJournal8
2000177
–216
183
Malaysia equity market index excess return y0.416 3.351 y13.711 22.440Malaysia banking industry index excess return y0.530 4.166 y13.339 30.178Philippines equity market index excess return y0.290 2.462 y10.186 13.445Philippines banking industry index excess return y0.326 2.424 y10.329 13.773Excess return on BP holding y0.002 0.492 y1.865 2.033Excess return on DM holding y0.037 0.557 y2.064 2.030Excess return on FF holding y0.035 0.547 y2.015 2.027Excess return on JY holding y0.057 0.804 y3.013 4.354Excess return on Korea won holding y0.046 2.392 y12.717 21.968Excess return on Indonesia rupiah holding y0.276 4.224 y21.205 22.828Excess return on Thailand baht holding y0.071 1.557 y6.791 6.503Excess return on Malaysia ringgit holding y0.115 1.425 y6.548 6.713Excess return on Philippines peso holding y0.087 1.262 y11.850 3.867Ž .D Eurodollar 0.009 1.491 y7.847 17.662Ž .D Eurodollary Euroyen 0.004 3.096 y15.626 19.268
Ž .Panel B. Holding period returns % on bank and market indices over the sample period
Period Index UK Germany France US Japan Korea Indonesia Thailand Malaysia Philippines
January 15, 1997–July 15, 1998 Bank 50.84 85.25 109.95 73.22 y41.19 y85.3 y96.67 y83.44 y86.5 y70.2Market 44.76 64.41 61.05 54.48 y21.02 y65.3 y87.03 y78.26 y79.0 y65.8Difference 6.08 20.84 48.90 18.73 y20.17 y19.9 y9.64 y5.18 y7.4 y4.4
July 2, 1997–January 30, 1998 Bank 17.24 25.31 25.88 14.40 y32.50 y61.1 y93.98 y64.75 y82.2 y68.4Market 14.19 5.2 7.78 10.13 y24.34 y51.9 y82.63 y54.39 y68.0 y58.5Difference 3.05 20.11 18.11 4.2 y8.16 y9.1 y11.35 y10.36 y14.2 y9.8
February 2, 1998–April 9, 1998 Bank 15.10 16.96 42.32 20.52 y7.81 y18.4 43.10 19.12 65.08 49.25Market 15.52 19.90 22.90 13.63 y3.46 y7.2 36.55 13.91 34.53 29.57Difference y0.42 y2.9 19.41 6.8 y4.35 y11.1 6.54 5.21 30.55 19.68
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216184
Ž .Fig. 1. Banking industry indices by country for the period from January 15, 1997 sUS$100 to July15, 1998.
countries. In all cases, the losses were devastating. Fig. 2 shows the evolution ofthe bank and market indices for these countries. There is little evidence of a fall inbank stock prices relative to the market index in the first 6 months of 1997 exceptfor Korea. In all other countries, bank stocks collapsed when the market collapsed.
Looking at Fig. 2, it is hard to make the case that there was a bank crisisunfolding before the currency crisis starting from the middle of 1997. Thissuggests that the poor performance of banks should be due to the currencycollapse and to the deterioration of expectations about future economic activitythat accompanied the collapse. There could be at least three reasons for this. First,the profits of banks increase with the level of economic activity so that bankstocks suffer when news indicates that economic growth will be lower. Second,banks were holding marketable securities and this exposure increased before the
Ž .crisis see World Bank, 1998, p. 40 . The value of these marketable securitieswould fall as the stock market falls. Third, banks are exposed to exchange ratechanges because of their activities, so that they make losses if they have net shortpositions in foreign currencies when their country’s currency collapses.
Banks used offshore financing extensively. This offshore financing was gener-ally short-term and dollar-denominated. The banks would then turn around andmake domestic loans. When domestic loans were denominated in local currency,banks would bear currency risk directly. Since local interest rates were generallyhigher than offshore rates, being long in the domestic currency and short in theforeign currency could be highly profitable as long as exchange rates remained
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216 185
Fig. 2. Bank index divided by stock market index for Asian countries from January 15, 1997Ž .sUS$100 to July 15, 1998.
relatively stable. The apparently large exposures resulting from this practice haveŽ .been blamed for the banks’ problems. For instance, the World Bank 1998, p. 35
states that ‘‘The organizational design and special incentives of Thailand’s BangkokŽ .International Banking Facility BIBF contributed, to a great degree, to Thailand’s
crisis.’’ If loans financed in foreign currency were also denominated in foreigncurrency, banks still had an indirect foreign currency exposure despite having longand short positions in foreign currency because large currency changes made itless likely that borrowers would repay their loans.
Bank balance sheets offer a poor measure of the bank’s exchange rate expo-sures.7 They do not tell us whether exchange rates affect banks through fee-gener-ating businesses. They ignore option features in loan contracts that affect currencyexposures, such as protective covenants. Further, banks have many off-balancesheet derivatives positions, some of which are designed to hedge balance sheetexposures. We therefore focus directly on measures of the exchange rate exposuresof banks obtained from equity returns. To obtain such measures, we regress bank
7 Ž .Burnside et al. 1999, p. 7 review the data sources and the evidence on exposure of emergingmarket banks. They point out that ‘‘Given data limitations, it is not possible to precisely measure theextent to which large net foreign asset positions were hedged in the different crisis countries.’’
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216186
Tab
le2
Est
imat
esof
the
SU
Rm
odel
ofba
nkin
gin
dust
ryex
cess
retu
rns
onco
ntem
pora
neou
sin
stru
men
tsŽ
.T
hefo
llow
ing
SU
Rm
odel
sar
ees
tim
ated
ina
syst
emfo
r10
coun
trie
s’ba
nkin
gin
dust
ryin
dice
sov
erth
epe
riod
from
Janu
ary
15,
1997
toJu
ly15
,19
9839
0da
ys:
K
Rs
aq
ad
qb
qb
dR
qg
qg
dX
q´
,Ž
.Ž
.Ž
.p
t0
1t
01
tm
t0
,k1
,kt
kt
tÝ ks
1
whe
reR
isth
elo
gari
thm
icda
ily
doll
arre
turn
onth
eD
atas
trea
mba
nkin
gin
dust
ryin
dice
sfo
rea
chof
the
10co
untr
ies,
and
Ris
the
corr
espo
ndin
gst
ock
mar
ket
inde
xre
turn
.B
oth
retu
rns
are
pt
mt
inex
cess
of1-
day
retu
rnon
the
7-da
yE
urod
olla
rde
posi
t.d
isth
ein
tera
ctiv
edu
mm
yfo
rth
eA
sian
cris
ispe
riod
from
July
2,19
97to
Janu
ary
30,
1998
.X
incl
udes
the
dail
ydo
llar
exce
ssre
turn
st
kt
Ž.
onth
ecu
rren
cyho
ldin
gsB
P,
DM
,F
F,
and
JY,
the
chan
ges
inth
e7-
day
Eur
odol
lar
rate
and
the
spre
adbe
twee
nth
e7-
day
Eur
odol
lar
rate
and
Eur
oyen
rate
,la
gsof
the
Wes
tern
vari
able
sdu
eto
Ž.
the
tim
ezo
nedi
ffer
ence
,an
dda
ily
doll
arex
cess
retu
rns
onea
chof
the
five
Asi
anco
untr
ies’
curr
enci
esK
orea
won
,In
done
sia
rupi
ah,
Tha
ilan
dba
ht,
Mal
aysi
ari
nggi
t,P
hili
ppin
espe
so.
p-va
lues
are
inbr
acke
ts.
F-t
ests
inP
anel
Bar
ete
stin
gfo
rth
esu
mof
the
two
coef
fici
ents
for
each
vari
able
.
Coe
ffic
ient
esti
mat
esU
KG
erm
any
Fra
nce
US
Japa
nK
orea
Indo
nesi
aT
hail
and
Mal
aysi
aP
hili
ppin
es
Pan
elA
.SU
Res
tim
ates
2Ž
.A
djus
ted
RO
LS
0.73
90.
681
0.62
70.
724
0.84
50.
789
0.84
40.
894
0.89
40.
797
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
1a
:In
terc
ept
y0.
042
0.40
y0.
033
0.57
0.01
30.
840.
006
0.90
y0.
087
0.18
y0.
279
0.06
y0.
197
0.29
y0.
041
0.67
0.13
30.
140.
036
0.61
0w
xw
xw
xw
xw
xw
xw
xw
xw
xw
xa
:C
risi
sdu
mm
y0.
018
0.82
0.12
40.
180.
087
0.38
0.01
10.
880.
095
0.36
0.14
60.
54y
0.40
00.
18y
0.06
40.
67y
0.48
80.
00y
0.27
20.
021
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
2b
:O
wn
mar
ket
exce
ssre
turn
s1.
687
0.00
1.02
20.
001.
189
0.00
1.18
80.
001.
689
0.00
1.07
40.
000.
700
0.00
0.96
40.
001.
302
0.00
0.86
20.
000
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
b:
Cri
sis
dum
my
y0.
063
0.55
0.02
20.
79y
0.05
00.
60y
0.10
60.
15y
0.14
30.
10y
0.12
70.
160.
192
0.07
0.30
10.
00y
0.10
80.
09y
0.05
10.
401
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
3g
:E
xces
sre
turn
son
BP
y0.
665
0.00
0.02
50.
86y
0.00
50.
98y
0.02
20.
850.
186
0.27
0.47
70.
220.
286
0.55
0.04
00.
870.
006
0.98
0.11
40.
540,
1w
xw
xw
xw
xw
xw
xw
xw
xw
xw
xg
:C
risi
sdu
mm
y0.
205
0.26
0.17
30.
38y
0.10
50.
63y
0.05
50.
73y
0.06
00.
79y
0.49
50.
35y
0.45
10.
490.
132
0.69
0.13
70.
66y
0.07
60.
771,
1Ž.
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
4g
:E
xces
sre
turn
son
DM
y0.
082
0.90
0.53
60.
500.
093
0.91
y0.
677
0.28
y0.
160
0.86
y1.
629
0.43
0.08
40.
971.
176
0.37
y1.
094
0.38
y0.
914
0.37
0,2
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
g:
Cri
sis
dum
my
0.46
40.
73y
6.46
60.
00y
1.02
40.
55y
0.49
00.
69y
0.43
10.
821.
225
0.77
0.90
30.
87y
5.95
50.
030.
721
0.78
y1.
582
0.45
1,2
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
5g
:E
xces
sre
turn
son
FF
0.18
40.
79y
0.44
20.
58y
0.02
60.
980.
599
0.35
0.29
30.
751.
807
0.40
0.44
60.
87y
1.13
50.
401.
140
0.38
1.04
40.
320,
3w
xw
xw
xw
xw
xw
xw
xw
xw
xw
xg
:C
risi
sdu
mm
yy
0.52
90.
696.
022
0.00
0.79
90.
640.
621
0.62
0.29
10.
88y
1.35
60.
75y
1.10
30.
845.
774
0.03
y0.
891
0.73
1.42
40.
501,
3Ž.
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
6g
:E
xces
sre
turn
son
JY0.
051
0.44
0.05
20.
50y
0.02
50.
770.
006
0.92
y0.
743
0.00
y0.
357
0.09
y0.
389
0.14
y0.
036
0.78
0.06
40.
62y
0.12
30.
240,
4w
xw
xw
xw
xw
xw
xw
xw
xw
xw
xg
:C
risi
sdu
mm
y0.
033
0.77
0.00
01.
000.
096
0.49
y0.
065
0.52
0.27
30.
160.
488
0.15
0.40
30.
350.
126
0.55
y0.
232
0.27
0.58
80.
001,
4Ž.
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
7g
:D
Eur
odol
lar
y0.
001
0.99
0.04
00.
520.
035
0.61
0.04
30.
390.
026
0.73
y0.
238
0.18
y0.
176
0.42
0.06
80.
54y
0.02
60.
800.
033
0.70
0,5
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
g:
Cri
sis
dum
my
0.03
40.
590.
026
0.72
y0.
034
0.67
y0.
033
0.58
y0.
072
0.43
0.32
50.
110.
239
0.35
y0.
148
0.25
y0.
161
0.19
y0.
061
0.54
1,5
Ž.
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
8g
:D
Eur
odol
lary
Eur
oyen
y0.
008
0.79
y0.
008
0.80
0.01
10.
75y
0.06
30.
02y
0.03
70.
36y
0.08
50.
35y
0.02
30.
84y
0.01
10.
840.
038
0.48
0.01
10.
800,
6w
xw
xw
xw
xw
xw
xw
xw
xw
xw
xg
:C
risi
sdu
mm
yy
0.00
30.
93y
0.06
10.
11y
0.02
20.
600.
060
0.05
0.07
20.
12y
0.01
20.
910.
094
0.48
0.05
40.
41y
0.04
70.
460.
009
0.86
1,6
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216 187
Ž.
Ž.
wx
wx
wx
wx
wx
wx
9g
:L
agof
3y
0.07
90.
640.
838
0.03
y0.
459
0.34
0.10
00.
670.
216
0.35
y0.
197
0.29
0,7
wx
wx
wx
wx
wx
wx
g:
Cri
sis
dum
my
0.01
30.
95y
1.24
90.
020.
585
0.36
0.44
40.
17y
0.08
70.
78y
0.02
50.
921,
7Ž
.Ž.
wx
wx
wx
wx
wx
wx
10g
:L
agof
40.
214
0.82
y1.
691
0.42
1.73
50.
511.
284
0.33
y0.
095
0.94
y0.
225
0.82
0,8
wx
wx
wx
wx
wx
wx
g:
Cri
sis
dum
my
1.51
00.
411.
145
0.78
y2.
733
0.60
y1.
384
0.60
y1.
473
0.56
y2.
612
0.19
1,8
.Ž.
wx
wx
wx
wx
wx
wx
11g
:L
agof
50.
162
0.86
1.25
10.
56y
2.48
50.
36y
1.09
60.
420.
175
0.89
0.31
20.
760,
9w
xw
xw
xw
xw
xw
xg
:C
risi
sdu
mm
yy
1.59
30.
39y
0.53
30.
902.
839
0.59
1.00
40.
711.
114
0.66
2.56
80.
211,
9Ž
.Ž.
wx
wx
wx
wx
wx
wx
12g
:L
agof
70.
040
0.60
0.20
50.
230.
004
0.99
0.05
80.
590.
023
0.83
y0.
040
0.63
0,10
wx
wx
wx
wx
wx
wx
g:
Cri
sis
dum
my
y0.
063
0.47
y0.
255
0.21
0.01
10.
97y
0.11
20.
37y
0.17
50.
150.
089
0.36
1,10
Ž.
Ž.
wx
wx
wx
wx
wx
wx
13g
:L
agof
8y
0.00
60.
88y
0.06
20.
48y
0.05
70.
61y
0.01
20.
83y
0.01
10.
830.
030
0.49
0,11
wx
wx
wx
wx
wx
wx
g:
Cri
sis
dum
my
0.05
20.
250.
089
0.39
0.10
70.
410.
036
0.58
0.06
60.
29y
0.03
40.
501,
11Ž
.w
xw
xw
xw
xw
x14
g:
Exc
ess
retu
rns
y0.
092
0.61
0.71
70.
000.
188
0.04
y0.
187
0.09
0.16
80.
160,
12
onow
ncu
rren
cyw
xw
xw
xw
xw
xg
:C
risi
sdu
mm
y0.
168
0.42
y0.
396
0.01
y0.
320
0.01
y0.
087
0.56
y0.
041
0.76
1,12
Pan
elB
.F
-tes
tst
atis
tics
for
the
sum
ofth
etw
oco
effi
cien
tsfo
rea
chÕar
iabl
eFŽ1
,36
60.
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
H:
aq
as
00.
152
0.70
1.64
90.
201.
662
0.20
0.08
80.
770.
008
0.93
0.51
20.
476.
463
0.01
0.77
80.
389.
687
0.00
6.63
30.
010
01
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
H:
bs
188
.420
0.00
0.12
80.
729.
215
0.00
12.3
200.
0010
9.06
80.
001.
320
0.25
14.1
280.
000.
807
0.37
36.7
540.
008.
544
0.00
00
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
H:
bq
bs
168
.179
0.00
0.63
50.
433.
710
0.05
2.68
30.
1094
.414
0.00
0.71
30.
402.
351
0.13
49.8
120.
0023
.472
0.00
23.0
430.
000
01
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
H:
gq
gs
014
.450
0.00
2.21
80.
140.
571
0.45
0.53
70.
460.
675
0.41
0.00
30.
960.
141
0.71
0.61
20.
430.
455
0.50
0.05
10.
820
0,1
1,1
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
H:
gq
gs
00.
112
0.74
20.1
760.
000.
403
0.53
1.21
10.
270.
133
0.72
0.01
20.
910.
046
0.83
4.28
70.
040.
028
0.87
1.89
00.
170
0,2
1,2
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
H:
gq
gs
00.
090
0.76
17.4
760.
000.
271
0.60
1.30
00.
250.
127
0.72
0.01
50.
900.
020
0.89
3.94
60.
050.
012
0.91
1.80
90.
180
0,3
1,3
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
H:
gq
gs
00.
894
0.34
0.25
00.
620.
403
0.53
0.52
50.
479.
605
0.00
0.23
30.
630.
002
0.97
0.28
50.
591.
044
0.31
12.2
600.
000
0,4
1,4
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
H:
gq
gs
01.
065
0.30
3.02
70.
080.
000
0.99
0.10
90.
740.
959
0.33
0.66
10.
420.
226
0.63
1.50
00.
228.
559
0.00
0.29
70.
590
0,5
1,5
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
H:
gq
gs
00.
474
0.49
14.6
440.
000.
263
0.61
0.06
40.
802.
311
0.13
3.33
20.
071.
122
0.29
1.66
00.
200.
073
0.79
0.62
00.
430
0,6
1,6
wx
wx
wx
wx
wx
wx
H:
gq
gs
00.
192
0.66
1.44
20.
230.
087
0.77
6.30
90.
010.
380
0.54
1.78
50.
180
0,7
1,7
wx
wx
wx
wx
wx
wx
H:
gq
gs
01.
205
0.27
0.02
30.
880.
049
0.82
0.00
20.
970.
523
0.47
2.67
80.
100
0,8
1,8
wx
wx
wx
wx
H:
gq
gs
00.
824
0.36
0.04
00.
840.
006
0.94
0.00
20.
970.
351
00,
91,
9w
xw
xw
xw
xw
xw
xH
:g
qg
s0
0.28
70.
590.
221
0.64
0.01
30.
910.
754
0.39
6.15
20.
011.
022
0.31
00,
101,
10w
xw
xw
xw
xw
xw
xH
:g
qg
s0
3.78
80.
050.
242
0.62
0.56
00.
450.
529
0.47
2.83
30.
090.
029
0.87
00,
111,
11w
xw
xw
xw
xw
xH
:g
qg
s0
0.50
80.
4810
.813
0.00
2.05
70.
157.
280
0.01
3.36
80.
070
0,12
1,12
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216188
index returns on returns of foreign currency positions in Table 2. We use dollarreturns for all countries to make our results comparable.
To understand whether there was something unique about bank FX exposures,we have to account for the impact of FX shocks on aggregate economic activity.Since stock prices are forward-looking, we have to take into account changes inexpected future economic activity when the market participants learn about suchchanges. The return to a country’s stock market provides a forward-lookingmeasure of changes in expected future economic activity. We therefore account forshocks to aggregate economic activity through the exposure of banks to theircountry’s stock market and use the dollar excess return of the local stock market
Ž .as a control variable. Given the importance that Eichengreen and Rose 1997 , aswell as others, attribute to changes in interest rates in the US in banking crises, weinclude as our explanatory variables changes in the US interest rate as well aschanges in the US–yen interest rate spread. The interest rates used are Euro rateswith a 7-day maturity obtained from Datastream. A FX rate excess return isdefined as the dollar return on a risk-free investment in that currency over thereturn on a risk-free investment in the US. For the exchange rate excess returns,we use exchange rate returns in dollars for the crisis currencies as well as those ofthe Western countries and of Japan. We include the Western currencies in ourregressions since banks could have exposures to these currencies through theirborrowings from banks in these countries. We assume that a risk-free investmentin a country earns the shortest maturity money market rate in that country that wecan get from Datastream. Because of the time zone differences between Asia andthe Western countries, it is important to use both contemporaneous and laggedvariables for the Western variables. Remember that the Asian markets for daytq1 are already closed when the Western markets open for the same day. Thismeans that changes in these variables that take place during trading hours in theWest cannot be incorporated in the Asian share prices on day tq1. In contrast,changes that took place on day t during trading hours in the West can only beincorporated in Asia during day tq1. The period with the most dramatic changesin stock markets and exchange rates is the period from the Thai baht devaluationon July 2, 1997, through the end of January 1998. We define this period as thecrisis period and allow the exposures to differ from the Thai baht devaluation tothe end of January 1998. The regressions we reproduce are estimated with a
Ž .seemingly unrelated regression SUR specification that allows for contemporane-ous correlation across 10 countries.8 We examined the robustness of our resultsusing other regression specifications. In particular, we estimated the regressionsusing ordinary least squares and forcing the exposures to be constant over the
8 Ž .We use the same data source and same regression specification as Dewenter and Hess 1998 , butwe use daily returns because of our focus on the Asian crisis.
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216 189
whole sample period. Our conclusions are not altered if we use these differentspecifications.
The results in Table 2 are striking. The intercept estimates are insignificantacross all 10 countries, and similarly for their crisis dummies except for Malaysiaand the Philippines. This indicates, together with the high adjusted R2, that ourexplanatory variables describe the bank excess returns quite well. However, theexplanatory variable that is the most important in explaining bank excess returns isthe excess return on the stock market index of the country in which banks arelocated. The coefficient on this variable, b , is significant in all countries. It0
exceeds one for all Western bank indices.9 This means that, in general, a 1%return on the market index implies a return on the bank index in excess of 1%.Since Western stock markets performed well during the period we consider, thehigh b coefficient implies that Western banks should have outperformed their0
respective stock markets. Japanese banks also have an exposure to the local marketthat exceeds one. In that case, however, the market did poorly, so Japanese bankswould be expected to perform worse than the Japanese stock market, which theydid. Note, however, that during the crisis period, the intercept for Japanese banksis higher but not significantly so. It does not seem, therefore, that the crisiscontributed to the poor performance of Japanese banks directly. Turning next tothe crisis countries, all countries have significant market exposures. We allow forthe market exposure to differ during the crisis period, so that b qb provides us0 1
with an estimate of the relation between the market return and the bank indexreturn during the crisis period. The F-tests for the sum of b and b in Panel B0 1
of Table 2 show that the total coefficient on the market during the crisis periodsignificantly exceeds one in Thailand and Malaysia, is about one in Korea andIndonesia, and is significantly lower than one in the Philippines. Based on thesecoefficients, one would expect banks to perform poorly in all the crisis countries,but to underperform the market only in Thailand and Malaysia.
There is a large literature that examines the currency exposure of firms inWestern countries. The surprising result of this literature is that exchange rateexposures are small and often insignificant after taking into account the exposureof firms to the stock market.10 We call currency exposures estimated taking intoaccount aggregate stock market returns ‘‘net-of-market currency exposures’’ in thefollowing. As discussed above, however, there are good reasons to think thatnet-of-market currency exposures for Asian banks might have been quite large and
9 Ž .Dewenter and Hess 1998 estimate regressions using the same Datastream indices for banks in anumber of countries using monthly data from January 1984 to March 1996. They find coefficients onthe domestic market that are comparable to ours.
10 Ž .See Griffin and Stulz 1999 for a study of exchange rate exposures for developed countries thatuses the Datastream indices and provides further references to the literature.
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216190
could have played an important role in their problems.11 The returns in ourregressions are measured in dollars. The coefficient that measures the net-of-marketexposures of the dollar return of banks to their own currency is g . Our0,12
regressions include the dollar return on the stock market index of the bank’scountry. Consequently, a fall in the market index in dollars caused by a fall in thedollar value of the local currency affects the bank return in dollars through thelocal stock market exposure of the bank. The net-of-market currency exposurestherefore measure the impact on dollar bank returns of changes in the exchangerate that cannot be explained by the impact of changes in the exchange rate on thedollar return of their local market. As a result, if a change in the dollar value of thelocal currency has no impact on the local currency value of the stock market indexand no impact on the local currency value of the bank index, the net-of-marketexposure coefficient would be zero. A positive net-of-market currency exposurecoefficient means that an appreciation of the local currency has a positive effect onthe dollar return of a bank in addition to its impact through its effect on the dollarreturn of the stock market in the bank’s country.
For Western banks, the net-of-market exposures to the exchange rates ofŽ . Ž .Western countries, as measured by coefficients in rows 3 – 5 of Table 2, are
generally insignificant. There are only two exceptions. First, British banks are hurtby an appreciation of the pound for the whole sample period, in that theirnet-of-market exposure coefficient on the excess return of the pound in dollars,g , is significantly negative. German banks appear to have a large negative0,1
coefficient for the DM excess return, g , during the crisis period. Turning to the1,2
crisis countries, we find that the net-of-market own-currency exposures as mea-sured by g are insignificant for Korea and the Philippines. They are signifi-0,12
cantly positive for Indonesia and Thailand, but significantly negative for Malaysia.The net-of-market exposure coefficient for Indonesia is large, 0.717, while theother net-of-market exposure coefficients are less than 0.2 in absolute value. Theway to understand the coefficient on the own-currency return for Indonesian banksis that a 1% depreciation of the Indonesian rupiah leads to a decrease in the valueof the Indonesian bank stock index relative to the Indonesian stock market index
Ž .of 0.717%. Griffin and Stulz 1999 report evidence on net-of-market exchangerate exposure coefficients for more than 300 industries in developed economiesand never find coefficients that large in absolute value. However, we cannotconclude from this that FX depreciation explains the poor performance of banks in
11 Using dollar returns does not change the interpretation of the regressions for the exposurecoefficient. Suppose that a bank finances itself in dollars and has domestic currency assets. The dollarvalue of its liabilities is not affected by changes in the exchange rate, but the value of its assets is. Inthis case, with risk-free assets in local currency, the impact of an unexpected change in the value of the
Ž .exchange rate on equity is the product of value of assetsrvalue of equity and the change in theexchange rate, which would be quite large because of the leverage. As the bank finances itself more indomestic currency, the coefficient on the exchange rate falls because domestic currency liabilitieshedge assets denominated in domestic currency.
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216 191
Indonesia and Thailand. Remember that we allow the exposures to differ duringthe crisis period from July to the end of January. Hence, to find out the impact ofthe depreciation on banks after taking into account general stock market move-ments, we have to add the two net-of-market exposure coefficients, g and0,12
g . For Indonesia, we have a positive exposure for the whole period and during1,12
the crisis period. Hence, a depreciation of the Indonesian rupiah has an adverseeffect on Indonesian banks controlling for the return of the Indonesian market indollars. Paradoxically, however, the net-of-market currency exposure coefficient issignificantly lower during the crisis period, so that during the crisis, a 1%depreciation of the rupiah leads to an AR of 0.321% for Indonesian banks, wherewe define an AR as a return that could not be predicted based on general stockmarket movements of the country. The net-of-market exposure of Philippine banksduring the crisis is also significantly positive, but the coefficient is 0.127, so that a1% depreciation of the Philippine peso leads to an AR of Philippine banks of0.127%. For Thailand, the net-of-market exposure to the Thai baht of the banks isnegative during the crisis period. In other words, net-of-market exchange rateexposure helped rather than hurt Thai banks during the crisis period, in that itmade them perform better relative to the Thai stock market. This result issurprising in that Thai banks are generally viewed as a key example of banksusing short-term foreign currency financing. In the case of Malaysia, the net-of-market exposure is negative as well. This is consistent with the view expressed by
Ž .Furman and Stiglitz 1998 that FX exposures of banks in Malaysia were lowbecause of government policies.12 It follows from our analysis that the onlycountries where FX movements can help explain why banks performed worse thantheir local market are Indonesia and the Philippines.
To understand why we might reach a conclusion about bank exposures thatdiffers from conventional wisdom, it is important to remember that balance sheetexposures are only part of the story when one is evaluating currency exposures ofbanks. Short of knowing all the derivatives positions of a bank as well as itsbalance sheet exposure, one cannot accurately assess that bank’s currency expo-sure. In addition, it is possible that the currency collapses created large losses forbanks, but the market expected these losses to be made up through bailouts. In thiscase, currency exposures would create accounting losses but not equity valuelosses. Without accounting data for the banks in our sample, we cannot explorethis possibility.
Some have argued that we do not find much of a role for currency changes onbanks beyond their impact on aggregate economic activity as measured by thestock market because the capital markets were inefficient and did not correctlyincorporate the impact of currency changes in bank values. For instance, one viewis that investors could not possibly know what currency exposures were on aday-to-day basis. We address this issue in several ways. Using daily returns as we
12 Ž .See Furman and Stiglitz 1998, p. 97 .
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216192
do in the regressions reported here opens the possibility that pricing mistakes orslow adjustment could explain the low net-of-market currency exposures. Tocheck this, we estimated regressions with leads and lags of exchange rate returnsas well as regressions using weekly returns. We also estimated regressionsallowing for non-linear effects of exchange rate changes and regressions where theonly explanatory variables were market returns and exchange rate changes. Noneof these additional regressions changes our conclusions.
A final difficulty with our estimates is that we had changes in exchange rateregimes for the crisis countries. Up to some date, the exchange rate was pegged.After that date, the country has a floating exchange rate. As a result, we could do apoor job of capturing exchange rate effects because our sample period incorporatesa period where the exchange rate fluctuates little. If this were an explanation forour results, then the estimates of net-of-market own-currency exposures forThailand, the Philippines, and Malaysia during the crisis period should be esti-mated precisely because for these countries, the crisis period corresponds to afloating exchange rate period. However, for Thailand and Malaysia, we do notfind significant positive net-of-market own-currency exposures during the crisisperiod. In Section 3, we also find that measuring net-of-market exposures duringdays of sharp exchange rate changes does not alter our conclusions either.
Our approach assumes that the market return captures common effects ofexchange rate shocks across industries. Since banks are part of the market index,our procedure would underestimate the impact of exchange rate shocks on banks ifexchange rate shocks do not have a pervasive effect across industries but mostlyhave an effect on banks and if banks have a non-trivial weight in the market index.The reason for this is that part of the bank-specific effect of exchange rates wouldbe captured by the market return because banks are part of the market. Theweights of banks in the Datastream country indices at the start of our sampleperiod go from 3.26% for Indonesia to 35.28% for Thailand. This means that if theexchange rate shocks had no impact outside of banks, we would understate theimpact of exchange rate shocks on banks substantially in the case of Thailand.However, in the case of Thailand, we find that banks benefited from depreciationof the local currency so that the bias discussed here would make our results evenmore surprising. In the case of Korea, the weight of banks in the index is 12.47%.This weight is 14.45% in the case of Malaysia and 23.23% in the case of thePhilippines. For these countries, therefore, the bias in our estimates would besmall in the unlikely event where exchange rate shocks have no common effectsacross industries and would not be large enough to alter our conclusions.
In Table 2, we also allow for an impact of interest rate changes on the equity ofbanks. The coefficients on changes in the Eurodollar rate, g and g are0,5 1,5
insignificant for banks in all countries. The coefficients on the dollar–yen interestrate spread, g and g , are insignificant in all countries except for US banks.0,6 1,6
This suggests that the banking crisis in East Asia is atypical, in that it cannot beattributed to interest rate changes in developed countries in contrast to the
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216 193
Ž .evidence reported by Eichengreen and Rose 1997 who argue that high interestrates in developed countries are generally strongly associated with the onset ofbanking crises.
From our analysis, there is no support for the view that currency movementswere consistently important determinants of the performance of banks in the crisiscountries once one takes into account the stock market returns in these countries.Given the many statements made about the importance of the currency exposuresfor the East Asian banks, our results are surprising. However, our results do notmean that currency movements were not important for banks through their impacton stock markets. What our estimates of net-of-market exposures show is thatthere was nothing special about bank FX exposures during the crisis period in thatbank performance is explained by the market’s performance and that currencyreturns add little, if anything, to that explanation. In Table 3, we estimate howexchange rate movements are related to the performance of East Asian stockmarkets. We use the same estimation approach as in Table 2, but now thedependent variable is the dollar excess return of stock markets as opposed to thedollar excess return of bank indices. Except for the stock market, the explanatoryvariables are the same as in the regressions of Table 2. Since the dependentvariable is a dollar excess return, a 1% own-currency depreciation that has noimpact on the local stock market decreases the dollar value of the index by 1%.This means that for a currency depreciation to be associated with losses in thelocal stock market in local currency, it has to be that the 1% own-currencydepreciation decreases the dollar value of the stock market index by more than1%. Consequently, the coefficient on the own-currency excess return has to exceedone for the correlation between local currency stock returns and local currencyvalue to be negative. We therefore test the hypothesis that the own-currencyexposure of East Asian stock markets exceeds one. We find that for all East Asiancountries during the crisis, there is a negative relation between stock marketreturns and currency returns in local currency. For instance, a 1% unexpecteddepreciation of the Korean won relative to the dollar is associated with a 1.419%unexpected drop in the dollar value of the Korean stock market. This means that a1% unexpected depreciation of the Korean won is associated with approximately a0.4% decrease in the local value of the stock market. Our regressions cannotestablish causation but these results are consistent with the view that currencycrises affected stock markets negatively in the East Asian countries. To reiterateour main result, however, if currency crises affected banks adversely, there is nosystematic evidence that they did so more for banks than for stock markets ingeneral.
3. The returns of bank indices around crisis events
In this section, we evaluate the returns of bank indices around crisis events. Todo that, we estimate the regressions of Table 2 without the crisis dummy for the
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216194
Tab
le3
Est
imat
esof
the
SU
Rm
odel
ofm
arke
tex
cess
retu
rns
onva
riou
sin
stru
men
tsŽ
.T
hefo
llow
ing
SU
Rm
odel
sar
ees
tim
ated
ina
syst
emfo
r10
coun
trie
s’m
arke
tex
cess
retu
rns
over
the
peri
odfr
omJa
nuar
y15
,19
97to
July
15,
1998
390
days
:K
Rs
aq
ad
qg
qg
dX
q´
,Ž
.Ž
.m
t0
1t
0,k
1,k
tk
tt
Ý ks1
whe
reR
isth
elo
gari
thm
icda
ily
doll
arre
turn
onth
eD
atas
trea
mst
ock
mar
ket
inde
xre
turn
sfo
rea
chof
the
10co
untr
ies,
inex
cess
of1-
day
retu
rnon
the
7-da
yE
urod
olla
rde
posi
t.d
isth
em
tt
Žin
tera
ctiv
edu
mm
yfo
rth
eA
sian
cris
ispe
riod
from
July
2,19
97to
Janu
ary
30,
1998
.X
incl
udes
the
dail
ydo
llar
exce
ssre
turn
son
the
curr
ency
hold
ings
BP
,D
M,
FF
,JY
,an
dth
ecu
rren
cyfo
rk
tw
x.ea
chof
the
five
Asi
anco
untr
ies
Kor
eaw
on,
Indo
nesi
aru
piah
,T
hail
and
baht
,M
alay
sia
ring
git,
Phi
lipp
ines
peso
,th
ech
ange
sin
the
7-da
yE
urod
olla
rra
tean
dth
esp
read
betw
een
the
7-da
yE
urod
olla
rra
tean
dE
uroy
enra
te,
and
lags
ofth
eW
este
rnva
riab
les
due
toth
eti
me
zone
diff
eren
ce.
p-va
lues
are
inbr
acke
ts.
F-t
ests
inP
anel
Bar
ete
stin
gfo
rth
esu
mof
the
two
coef
fici
ents
for
each
vari
able
.
Coe
ffic
ient
esti
mat
esU
KG
erm
any
Fra
nce
US
Japa
nK
orea
Indo
nesi
aT
hail
and
Mal
aysi
aP
hili
ppin
es
Pan
elA
.SU
Res
tim
ates
2Ž
.A
djus
ted
RO
LS
0.17
00.
119
0.05
90.
057
0.45
40.
649
0.76
20.
455
0.55
30.
430
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
1a
:In
terc
ept
0.08
70.
070.
203
0.00
0.18
20.
010.
119
0.06
0.04
90.
53y
0.30
60.
08y
0.14
00.
44y
0.51
80.
00y
0.21
30.
15y
0.12
40.
310
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
a:
Cri
sis
dum
my
y0.
024
0.76
y0.
165
0.15
y0.
166
0.12
y0.
086
0.39
y0.
133
0.29
0.15
60.
57y
0.11
30.
700.
628
0.03
y0.
028
0.91
y0.
194
0.32
1Ž.
wx
wx
wx
wx
wx
wx
wx
wx
wx
2g
:E
xces
s0.
039
0.83
y0.
257
0.13
y0.
170
0.29
0.12
20.
55y
0.31
00.
49y
0.18
00.
700.
113
0.80
0.03
20.
93y
0.04
10.
900,
1
retu
rns
onB
Pw
xw
xw
xw
xw
xw
xw
xw
xw
xg
:C
risi
sdu
mm
yy
0.32
70.
190.
117
0.61
0.21
10.
33y
0.34
90.
21y
0.72
20.
230.
335
0.60
y0.
242
0.69
y0.
060
0.91
0.02
10.
961,
1Ž.
wx
wx
wx
wx
wx
wx
wx
wx
wx
3g
:E
xces
s0.
687
0.29
y1.
927
0.03
y0.
463
0.59
0.58
60.
59y
1.14
50.
631.
523
0.55
y2.
976
0.22
1.37
80.
511.
326
0.44
0,2
retu
rns
onD
Mw
xw
xw
xw
xw
xw
xw
xw
xw
xg
:C
risi
sdu
mm
yy
1.16
40.
37y
1.29
10.
47y
1.12
80.
511.
122
0.62
0.92
70.
85y
1.62
60.
762.
218
0.66
y5.
403
0.20
y4.
947
0.16
1,2
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
wx
4g
:E
xces
sy
0.68
70.
31y
1.73
70.
080.
296
0.74
y0.
694
0.54
0.51
10.
84y
1.59
00.
553.
274
0.19
y1.
607
0.45
y1.
384
0.43
0,3
retu
rns
onF
Fw
xw
xw
xw
xw
xw
xw
xw
xw
xg
:C
risi
sdu
mm
y0.
589
0.65
1.79
00.
360.
740
0.67
y1.
392
0.54
y0.
809
0.87
0.88
70.
87y
3.10
10.
545.
149
0.23
4.23
20.
241,
3Ž.
wx
wx
wx
wx
wx
wx
wx
wx
wx
5g
:E
xces
s0.
086
0.18
0.11
10.
250.
148
0.10
0.11
70.
170.
719
0.00
0.00
90.
97y
0.09
80.
690.
103
0.63
0.03
70.
830,
4
retu
rns
onJY
wx
wx
wx
wx
wx
wx
wx
wx
wx
g:
Cri
sis
dum
my
0.05
30.
620.
299
0.06
y0.
073
0.62
y0.
080
0.57
y0.
872
0.03
0.88
10.
040.
157
0.70
y0.
004
0.99
0.16
20.
571,
4Ž.
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
6g
:D
Eur
odol
lar
0.01
10.
830.
081
0.30
0.09
80.
180.
039
0.57
0.08
30.
370.
174
0.39
y0.
053
0.81
0.01
70.
94y
0.02
20.
90y
0.16
70.
250,
5w
xw
xw
xw
xw
xw
xw
xw
xw
xw
xg
:C
risi
sdu
mm
yy
0.01
30.
830.
000
1.00
y0.
123
0.15
y0.
158
0.05
y0.
080
0.46
y0.
360
0.13
0.22
40.
370.
042
0.86
0.03
70.
850.
255
0.13
1,5
Ž.
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
x7
g:
DE
urod
olla
ry0.
012
0.67
y0.
048
0.24
y0.
062
0.10
y0.
048
0.19
0.03
60.
450.
093
0.38
y0.
002
0.99
0.00
60.
95y
0.07
70.
390.
011
0.89
0,6
.E
uroy
enw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xg
:C
risi
sdu
mm
yy
0.00
10.
990.
040
0.40
0.05
80.
190.
052
0.21
y0.
046
0.41
y0.
077
0.53
0.05
10.
69y
0.10
80.
380.
036
0.73
y0.
040
0.64
1,6
Ž.
wx
wx
wx
wx
wx
wx
8g
:L
agof
exce
ssy
0.05
20.
790.
411
0.35
0.24
50.
60y
0.04
60.
92y
0.02
30.
95y
0.30
50.
320,
7
retu
rns
onB
Pw
xw
xw
xw
xw
xw
xg
:C
risi
sdu
mm
y0.
257
0.32
y0.
220
0.71
y0.
026
0.97
y0.
075
0.90
0.12
80.
800.
456
0.27
1,7
Ž.
wx
wx
wx
wx
wx
wx
9g
:L
agof
exce
ss1.
062
0.32
1.50
30.
530.
191
0.94
y2.
561
0.29
1.03
10.
620.
075
0.96
0,8
retu
rns
onD
Mw
xw
xw
xw
xw
xw
xg
:C
risi
sdu
mm
yy
1.41
30.
50y
3.17
30.
504.
558
0.36
12.7
400.
01y
1.05
70.
802.
671
0.42
1,8
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216 195
Ž.
wx
wx
wx
wx
wx
wx
10g
:L
agof
exce
ssy
1.13
70.
30y
1.55
80.
530.
071
0.98
2.82
90.
26y
0.99
30.
640.
053
0.98
0,9
retu
rns
onF
Fw
xw
xw
xw
xw
xw
xg
:C
risi
sdu
mm
y1.
142
0.59
3.32
90.
49y
5.47
20.
28y
13.9
090.
000.
866
0.83
y3.
055
0.37
1,9
Ž.
wx
wx
wx
wx
wx
wx
11g
:L
agof
D0.
034
0.70
0.17
20.
380.
012
0.95
0.00
01.
000.
012
0.95
y0.
106
0.44
0,10
Ž.
Eur
odol
lar
wx
wx
wx
wx
wx
wx
g:
Cri
sis
dum
my
y0.
063
0.53
y0.
482
0.04
y0.
085
0.72
y0.
208
0.37
y0.
262
0.18
0.12
90.
421,
10Ž
.w
xw
xw
xw
xw
xw
x12
g:
Lag
ofD
y0.
034
0.44
y0.
020
0.84
0.04
30.
680.
053
0.61
y0.
087
0.31
0.02
50.
720,
11.
Eur
odol
lary
Eur
oyen
wx
wx
wx
wx
wx
wx
g:
Cri
sis
dum
my
y0.
008
0.88
0.03
00.
80y
0.09
40.
45y
0.11
20.
350.
149
0.14
y0.
097
0.25
1,11
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
wx
13g
:E
xces
sre
turn
s0.
753
0.00
2.43
60.
012.
467
0.01
1.25
70.
001.
552
0.00
0.97
40.
001.
353
0.00
1.55
80.
001.
307
0.00
0,12
onow
ncu
rren
cyw
xw
xw
xw
xw
xw
xw
xw
xw
xg
:C
risi
sdu
mm
yy
0.20
50.
22y
2.81
40.
140.
774
0.67
0.45
00.
01y
0.13
30.
450.
238
0.00
0.23
00.
170.
193
0.23
y0.
164
0.33
1,12
Pan
elB
.F
-tes
tst
atis
tics
for
the
sum
ofth
etw
oco
effi
cien
tsfo
rea
chÕar
iabl
eFŽ1
,36
80.
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
H:
aq
as
01.
135
0.29
0.18
60.
670.
040
0.84
0.18
80.
660.
739
0.39
0.49
00.
481.
217
0.27
0.24
60.
621.
649
0.20
4.28
90.
040
01
wx
wx
wx
wx
wx
wx
wx
wx
wx
H:
gq
gs
03.
044
0.08
0.83
30.
360.
080
0.78
1.50
00.
226.
459
0.01
0.12
80.
720.
097
0.76
0.00
60.
940.
005
0.95
00,
11,
1w
xw
xw
xw
xw
xw
xw
xw
xw
xH
:g
qg
s0
0.18
40.
674.
385
0.04
1.19
50.
270.
770
0.38
0.00
30.
960.
001
0.98
0.03
10.
861.
189
0.28
1.40
20.
240
0,2
1,2
wx
wx
wx
wx
wx
wx
wx
wx
wx
H:
gq
gs
00.
008
0.93
0.00
10.
970.
495
0.48
1.12
10.
290.
005
0.95
0.02
30.
880.
002
0.97
0.90
10.
340.
847
0.36
00,
31,
3w
xw
xw
xw
xw
xw
xw
xw
xw
xH
:g
qg
s0
2.65
80.
1010
.281
0.00
0.40
00.
530.
107
0.74
0.24
00.
627.
124
0.01
0.03
50.
850.
133
0.72
0.78
70.
380
0,4
1,4
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
H:
gq
gs
00.
004
0.95
2.88
60.
090.
318
0.57
8.08
50.
000.
003
0.96
2.32
10.
131.
723
0.19
0.22
40.
640.
022
0.88
1.00
60.
320
0,5
1,5
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
H:
gq
gs
00.
539
0.46
0.14
30.
710.
047
0.83
0.05
20.
820.
124
0.72
0.06
50.
800.
564
0.45
2.64
70.
100.
600
0.44
0.45
80.
500
0,6
1,6
wx
wx
wx
wx
wx
wx
H:
gq
gs
01.
395
0.24
0.23
60.
630.
278
0.60
0.09
10.
760.
095
0.76
0.29
70.
590
0,7
1,7
wx
wx
wx
wx
wx
wx
H:
gq
gs
00.
038
0.85
0.16
70.
681.
205
0.27
5.97
10.
010.
000
0.99
0.90
90.
340
0,8
1,8
wx
wx
wx
wx
wx
wx
H:
gq
gs
00.
000
1.00
0.18
60.
671.
546
0.21
7.01
40.
010.
001
0.97
1.07
70.
300
0,9
1,9
wx
wx
wx
wx
wx
wx
H:
gq
gs
00.
331
0.56
6.80
30.
010.
364
0.55
3.23
10.
076.
477
0.01
0.08
70.
770
0,10
1,10
wx
wx
wx
wx
wx
wx
H:
gq
gs
02.
388
0.12
0.02
90.
870.
617
0.43
0.90
60.
341.
341
0.25
2.72
60.
100
0,11
1,11
wx
wx
wx
wx
wx
wx
wx
wx
wx
H:
gq
gs
024
.386
0.00
0.05
20.
824.
351
0.04
142.
329
0.00
556.
829
0.00
940.
205
0.00
184.
348
0.00
260.
926
0.00
192.
974
0.00
00,
121,
12w
xw
xw
xw
xw
xw
xw
xw
xw
xH
:g
qg
s1
16.5
630.
000.
692
0.41
2.08
00.
1524
.418
0.00
48.5
610.
0028
.799
0.00
24.9
980.
0047
.995
0.00
3.01
30.
080
0,12
1,12
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216196
Tab
le4
Est
imat
esof
the
CA
Rs
ofba
nkin
gin
dust
ryex
cess
retu
rns
arou
ndne
ws
anno
unce
men
tsŽ
.T
hefo
llow
ing
SU
Rm
odel
sar
ees
tim
ated
ina
syst
emfo
r10
coun
trie
s’ba
nkin
gin
dust
ryin
dice
sov
erth
epe
riod
from
Janu
ary
15,
1997
toJu
ly15
,19
9839
0da
ys:
KJ
Rs
aq
bR
qg
Xq
Df
q´
,p
t0
0m
t0
,kk
tj
jtt
ÝÝ
ks1
js1
whe
reR
isth
elo
gari
thm
icda
ily
doll
arre
turn
onth
eD
atas
trea
mba
nkin
gin
dust
ryin
dice
sfo
rea
chof
the
10co
untr
ies,
and
Ris
the
corr
espo
ndin
gst
ock
mar
ket
inde
xre
turn
.B
oth
retu
rns
are
pt
mt
Ž.
inex
cess
of1-
day
retu
rnon
the
7-da
yE
urod
olla
rde
posi
t.X
incl
udes
the
dail
ydo
llar
exce
ssre
turn
son
the
curr
ency
hold
ings
BP
,D
M,
FF
,an
dJY
,th
ech
ange
sin
the
7-da
yE
urod
olla
rra
tean
dk
t
the
spre
adbe
twee
nth
e7-
day
Eur
odol
lar
rate
and
Eur
oyen
rate
,la
gsof
the
Wes
tern
vari
able
sdu
eto
the
tim
ezo
nedi
ffer
ence
,an
dda
ily
doll
arex
cess
retu
rns
onea
chof
the
five
Asi
anco
untr
ies’
Ž.
curr
enci
esK
orea
won
,In
done
sia
rupi
ah,
Tha
ilan
dba
ht,
Mal
aysi
ari
nggi
t,P
hili
ppin
espe
so.
fis
adu
mm
yva
riab
leta
king
ava
lue
of1r
nfo
rth
ejt
hev
ent
days
show
nbe
low
,or
zero
,ot
herw
ise.
jt
Thu
s,th
epa
ram
eter
esti
mat
eD
repr
esen
tsa
CA
Rfo
rth
ejt
hev
ent
days
.W
eon
lyre
prod
uce
the
esti
mat
esof
Dsi
nce
the
para
met
ers
esti
mat
esar
esi
mil
arto
thos
eof
Tab
le2.
p-va
lues
are
inj
j
brac
kets
.F
-tes
tsin
the
last
colu
mn
are
join
tte
sts
acro
ss10
coun
trie
s.
Est
imat
esof
CA
Rs
UK
Ger
man
yF
ranc
eU
SJa
pan
Kor
eaIn
done
sia
Tha
ilan
dM
alay
sia
Phi
lipp
ines
FŽ1
0,34
45.
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9705
14–
y0.
112
0.93
1.31
90.
390.
322
0.85
y0.
572
0.64
y1.
920
0.27
y0.
089
0.98
y2.
155
0.65
0.63
50.
801.
339
0.58
y1.
794
0.35
0.36
80.
961
.97
0516
:n
s3
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9705
22–
1.23
70.
35y
0.53
20.
730.
747
0.65
0.42
10.
731.
299
0.46
5.98
30.
142.
351
0.63
y0.
001
1.00
y3.
546
0.15
2.49
90.
200.
842
0.59
2.
9705
26:
ns
3Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
0618
–2.
661
0.04
y2.
478
0.10
y2.
167
0.19
0.49
80.
681.
407
0.42
y2.
831
0.48
4.36
00.
36y
4.28
60.
09y
0.40
50.
87y
1.87
30.
331.
582
0.11
3.
9706
20:
ns
3Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
0626
–y
0.46
80.
72y
1.44
60.
34y
1.84
20.
26y
2.04
70.
091.
101
0.53
1.37
70.
73y
1.21
40.
80y
3.58
40.
161.
202
0.62
2.73
30.
160.
946
0.49
4.
9706
30:
ns
3Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
0701
–y
0.52
30.
69y
0.68
90.
65y
0.58
70.
721.
597
0.19
0.17
20.
92y
0.16
00.
97y
1.56
50.
74y
0.47
60.
860.
036
0.99
0.28
10.
880.
240
0.99
5.
9707
03:
ns
3Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
0711
–0.
619
0.64
1.23
60.
42y
0.10
90.
95y
1.57
10.
202.
173
0.22
y0.
724
0.86
2.13
10.
660.
668
0.80
2.82
90.
252.
724
0.18
0.70
40.
726
.97
0715
:n
s3
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9707
23–
y0.
085
0.96
y0.
704
0.72
5.50
40.
012.
117
0.18
y0.
936
0.68
5.13
40.
33y
0.12
50.
98y
2.53
50.
45y
0.37
50.
911.
838
0.47
1.15
10.
327
.97
0729
:n
s5
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9708
04–
y0.
352
0.79
y2.
392
0.12
1.26
30.
45y
0.39
10.
754.
735
0.01
1.85
30.
65y
3.11
00.
52y
0.53
60.
84y
0.11
90.
96y
1.24
60.
521.
197
0.29
8.
9708
06:
ns
3Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
0808
–y
1.39
70.
292.
818
0.07
0.50
50.
760.
180
0.88
1.55
10.
38y
0.28
20.
942.
235
0.64
y0.
770
0.77
0.59
80.
81y
0.05
60.
980.
593
0.82
9.
9708
12:
ns
3Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
0813
–0.
120
0.93
0.82
40.
59y
0.25
10.
881.
743
0.16
y1.
062
0.55
y0.
204
0.96
3.74
70.
442.
199
0.40
0.05
10.
981.
100
0.57
0.43
40.
9310
.97
0815
:n
s3
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9708
19–
y0.
447
0.73
2.23
30.
14y
0.37
10.
82y
0.73
10.
552.
533
0.15
0.15
80.
97y
11.6
930.
020.
901
0.73
0.31
20.
90y
1.79
90.
351.
157
0.32
11.
9708
21:
ns
3Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
0919
–y
0.96
70.
361.
488
0.23
0.85
10.
53y
0.26
50.
79y
0.52
80.
711.
152
0.73
0.78
30.
84y
0.19
40.
93y
1.49
30.
46y
1.57
60.
320.
456
0.92
12.
9709
22:
ns
2Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
1007
–y
1.63
30.
21y
0.49
30.
751.
695
0.30
y0.
503
0.68
y1.
420
0.41
2.87
30.
47y
2.16
40.
65y
3.89
90.
13y
2.04
50.
413.
709
0.05
1.25
70.
2513
.97
1009
:n
s3
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216 197
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9710
30–
y0.
582
0.65
y1.
317
0.39
y1.
649
0.31
y1.
485
0.22
0.73
30.
67y
0.28
70.
94y
3.29
90.
496.
962
0.01
y3.
027
0.22
2.38
50.
221.
504
0.13
14.
9711
03:
ns
3Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
1104
–y
0.21
40.
87y
0.27
60.
86y
1.70
20.
300.
441
0.72
y3.
151
0.07
y1.
983
0.63
y0.
432
0.93
0.41
80.
870.
325
0.89
y0.
102
0.96
0.49
20.
9015
.97
1106
:n
s3
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9711
14–
y0.
787
0.55
y0.
567
0.71
0.05
20.
970.
122
0.92
y0.
020
0.99
y0.
572
0.89
y4.
842
0.31
3.37
40.
192.
179
0.37
y0.
819
0.67
0.42
20.
9416
.97
1118
:n
s3
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9711
19–
y2.
350
0.12
y0.
241
0.89
2.60
50.
170.
719
0.61
0.41
50.
843.
127
0.50
y0.
344
0.95
y2.
780
0.35
y3.
869
0.17
1.05
60.
630.
881
0.55
17.
9711
24:
ns
4Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
1128
–1.
934
0.14
y1.
412
0.36
y1.
293
0.43
2.03
60.
100.
814
0.64
y1.
640
0.69
y2.
045
0.67
y3.
492
0.18
y2.
589
0.29
y3.
844
0.05
1.33
00.
2118
.97
1202
:n
s3
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9712
03–
1.68
00.
274.
968
0.01
4.44
80.
021.
881
0.19
2.56
80.
2012
.503
0.01
5.94
20.
292.
848
0.34
y0.
154
0.96
y4.
503
0.05
3.13
40.
0019
.97
1208
:n
s4
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9712
12–
1.61
70.
222.
054
0.18
y0.
311
0.85
0.08
10.
952.
470
0.16
6.28
90.
13y
9.87
90.
04y
1.15
90.
660.
768
0.75
1.76
60.
381.
489
0.14
20.
9712
16:
ns
3Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
1217
–y
0.38
00.
782.
303
0.14
y2.
177
0.20
0.14
60.
91y
1.28
60.
480.
159
0.97
y9.
762
0.05
y3.
153
0.24
y6.
655
0.01
y1.
218
0.54
1.61
00.
1021
.97
1219
:n
s3
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9712
23–
y0.
214
0.89
y1.
674
0.37
y3.
072
0.13
y0.
636
0.67
y1.
091
0.65
0.15
60.
983.
747
0.57
0.17
90.
964.
301
0.20
1.84
30.
490.
538
0.86
22.
9712
26:
ns
4Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
1229
–0.
087
0.95
y2.
199
0.19
y1.
092
0.54
y0.
449
0.74
0.03
40.
992.
282
0.66
y6.
059
0.32
y1.
874
0.56
6.36
30.
040.
470
0.85
0.84
90.
5823
.97
1231
:n
s3
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9801
01–
y0.
473
0.72
y2.
474
0.11
0.15
60.
92y
0.06
30.
960.
583
0.74
y0.
765
0.85
1.43
90.
77y
1.61
30.
53y
4.08
00.
10y
3.17
50.
100.
871
0.56
24.
9801
05:
ns
3Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D98
0108
–y
1.50
00.
25y
3.16
20.
04y
3.72
70.
02y
2.16
30.
080.
399
0.82
y7.
191
0.08
8.76
20.
07y
7.85
50.
00y
1.39
80.
570.
137
0.94
3.12
40.
0025
.98
0112
:n
s3
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9801
13–
0.82
60.
440.
189
0.88
1.00
00.
461.
460
0.15
1.22
60.
39y
4.00
10.
23y
14.5
570.
00y
0.61
10.
77y
2.42
60.
23y
1.42
50.
371.
939
0.04
26.
9801
14:
ns
2Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D98
0115
–0.
824
0.44
y0.
175
0.89
0.66
30.
62y
0.00
90.
990.
199
0.89
y2.
463
0.45
18.1
660.
001.
575
0.45
y1.
614
0.42
y0.
142
0.93
2.37
60.
0127
.98
0116
:n
s2
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9801
28–
2.28
70.
081.
779
0.24
y0.
055
0.97
y0.
696
0.57
y0.
139
0.94
0.80
40.
84y
0.30
20.
954.
138
0.12
y0.
034
0.99
y6.
425
0.00
1.97
90.
0328
.98
0130
:n
s3
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9802
13–
2.20
90.
14y
1.57
60.
371.
045
0.58
y0.
005
1.00
y1.
025
0.61
y4.
192
0.36
y1.
738
0.75
y4.
690
0.11
y1.
428
0.61
y0.
166
0.94
0.80
20.
6329
.98
0218
:n
s4
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9803
06–
y0.
035
0.98
y2.
409
0.11
1.10
50.
50y
0.15
00.
901.
350
0.43
y1.
713
0.67
1.49
30.
750.
154
0.95
y1.
057
0.66
y1.
353
0.48
0.52
00.
8830
.98
0310
:n
s3
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9803
20–
y1.
570
0.14
0.48
20.
70y
1.79
10.
18y
1.61
50.
10y
0.71
30.
61y
0.32
40.
92y
6.11
80.
12y
1.46
70.
481.
303
0.51
0.05
60.
970.
989
0.45
31.
9803
23:
ns
2Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D98
0320
–y
1.57
00.
140.
482
0.70
y1.
791
0.18
y1.
615
0.10
y0.
713
0.61
y0.
324
0.92
y6.
118
0.12
y1.
467
0.48
1.30
30.
510.
056
0.97
0.98
90.
4531
.98
0323
:n
s2
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9803
25–
0.52
40.
692.
927
0.05
2.93
00.
07y
0.83
60.
49y
2.12
40.
220.
051
0.99
6.63
20.
172.
420
0.34
y1.
195
0.62
2.82
60.
141.
311
0.22
32.
9803
27:
ns
3Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D98
0407
–1.
278
0.33
0.72
70.
64y
0.41
40.
80y
0.32
60.
79y
1.08
90.
53y
6.49
40.
11y
2.36
00.
62y
1.66
60.
520.
473
0.85
0.78
70.
680.
590
0.82
33.
9804
09:
ns
3
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216198
intercept but with dummy variables that capture the returns of bank indices aroundcrisis events that cannot be explained by the independent variables used in Table2. This means that we estimate the impact of crisis events on the return of bankindices net of the impact of FX changes, interest rate changes, and domesticmarket returns. Table 4 shows the estimates for the various events of the crisis,which are again obtained using the SUR specification. The traditional event studyapproach in finance uses only the market return as the independent variable.However, here we wanted to make sure that we understood the extent to whichevents affect returns in addition to the change in other variables. Admittedly, thereis some arbitrariness in how one selects event windows in a study like this one. Adifficulty with defining event windows is that it is not always clear when an eventtakes place. For instance, an event taking place on day t in the US getsincorporated in stock prices in Korea on day tq1. Also, an event could take placeat date t in Korea, but only be announced after the market closes so that it showsup in stock returns at date t in the US and date tq1 in Korea. We thereforeextend our event windows to make sure that they include enough days to allow fordifferences in time zones. The event windows we use in our estimation aredescribed in Appendix A. We also investigate ARs for individual days.
Our approach consists of estimating the ARs of bank indices during periodscorresponding to important events in the Asian crisis. We perform this estimationfor 10 bank indices. The problem with such an approach is that during the periodswe focus on, other events could affect bank indices besides those of the Asiancrisis. Hence, we could find significant ARs that have nothing to do with events ofthe Asian crisis. Further, it could be that most of the information about the crisisare too small to have much of an impact on stock prices, even though the
Ž .cumulative impact of events might be significant. Cornell and Shapiro 1986studied the impact of the debt crisis of the 1980s on stock prices and found thatthey could not identify days that had significant impacts on stock prices. Yet,when they looked at yearly returns, they found that the performance of banks wasrelated to their exposure. We also fail to find significant ARs because ofinsufficient power in our tests. Table 1 shows that the volatility of stock returnsdiffers across countries and is large among the crisis countries. This means thatARs of same size could be significant in one country but not in another and thatARs that are economically significant might not be statistically significant.
Another study investigates returns on specific days for the Asian crisis.Ž .Kaminsky and Schmukler 1998 look at days with large market returns and then
investigate whether these large returns can be explained by contemporaneousevents. They find that agreements with international organizations are generallycontemporaneous with large returns. However, their study focuses on marketreturns. In contrast, our study focuses on ARs of banks. Remember that we defineARs as returns that cannot be predicted based on the evolution of the stock market.Hence, if bank values fall only because the stock market falls, the AR is zero inthe following analysis irrespective of why the stock market fell. In this case, a fall
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216 199
in the stock market brought about by a currency depreciation could lead to a fall inbank values, but there would be nothing ‘‘bank-specific’’ about the impact of thecurrency depreciation on bank values. Consequently, our analysis is designed toestimate the impact of currency changes that is unique to banks as opposed to theimpact of those changes that affect the whole economy. Further, we investigateARs on important dates of the Asian crisis, whereas they select the dates based onmarket returns. Their approach is different because they seek the answer to thequestion of whether markets were acting rationally, whereas we try to understandwhether specific events had an economic impact.
Rather than considering a chronology of events, we discuss first the impact onbanks of the initial event in each country where the central bank gives up ondefending the existing exchange rate regime and lets the exchange rate dropsubstantially. We then turn to events associated with IMF programs.
3.1. The currency deÕaluations
For Thailand, the initial devaluation was announced on July 2. We thereforeconsider a window from July 1 to 3. Table 4 shows that the ARs of banks in theUK, France, Germany, and Japan were all less than 1% in absolute value andinsignificant. For the US, the AR was 1.60%, but also insignificant. For the crisiscountries, the ARs of banks were less than for the US banks in absolute value andwere insignificant.
The Philippines abandoned their peg on July 11, and on July 14, the Malaysiancentral bank abandoned the defense of the ringgit. We use an event window fromJuly 11 to 15. The ARs of the Western banks were trivial and insignificant, withthe US banks having the highest AR in absolute value at y1.57% with a p-valueof 0.20. All Asian countries, except for Korea, experienced positive ARs. Thehighest ARs were for Malaysia and the Philippines. They were slightly above2.7% for both countries, but insignificant. The problem is, however, that thecurrency events took place at the same time that the IMF extended a program tohelp the Philippines on July 14, so that it could be that the positive ARs were dueto the IMF program. Table 5 provides estimates of ARs on these individual days.If letting the currency fall has a positive effect on banks, we should observe apositive AR on banks in the Philippines on July 11 and a positive AR on banks inMalaysia on July 14. The AR for the bank index in the Philippines on July 11 was1.81% with a p-value of 0.18, but it was y2.84% the preceding day with ap-value of 0.01. The AR for the bank index in Malaysia on July 14 was a trivial0.85%, but it was 2.32% on the following day with a p-value of 0.10. Based onthis evidence, there is no reason to suspect that these devaluations had strongadverse impacts on bank values.
The Indonesian rupiah plunged on August 14 after Indonesia gave up managingthe exchange rate. The story for the banks of developed countries is similar towhat happened with Thailand. For the UK, France, Germany, and Japan, the AR
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216200
Tab
le5
Est
imat
esof
the
AR
sof
bank
ing
indu
stry
exce
ssre
turn
son
each
day
ofth
ecu
rren
cyde
valu
atio
nsŽ
.T
hefo
llow
ing
SU
Rm
odel
sar
ees
tim
ated
ina
syst
emfo
r10
coun
trie
s’ba
nkin
gin
dust
ryin
dice
sov
erth
epe
riod
from
Janu
ary
15,
1997
toJu
ly15
,19
9839
0da
ys:
KJ
Rs
aq
bR
qg
Xq
Df
q´
,p
t0
0m
t0
,kk
tj
jtt
ÝÝ
ks1
js1
whe
reR
isth
elo
gari
thm
icda
ily
doll
arre
turn
onth
eD
atas
trea
mba
nkin
gin
dust
ryin
dice
sfo
rea
chof
the
10co
untr
ies,
and
Ris
the
corr
espo
ndin
gst
ock
mar
ket
inde
xre
turn
.B
oth
retu
rns
are
pt
mt
Ž.
inex
cess
of1-
day
retu
rnon
the
7-da
yE
urod
olla
rde
posi
t.X
incl
udes
the
dail
ydo
llar
exce
ssre
turn
son
the
curr
ency
hold
ings
BP
,D
M,
FF
,an
dJY
,th
ech
ange
sin
the
7-da
yE
urod
olla
rra
tean
dk
t
the
spre
adbe
twee
nth
e7-
day
Eur
odol
lar
rate
and
the
Eur
oyen
rate
,la
gsof
the
Wes
tern
vari
able
sdu
eto
the
tim
ezo
nedi
ffer
ence
,an
dda
ily
doll
arex
cess
retu
rns
onea
chof
the
five
Asi
anŽ
.co
untr
ies’
curr
enci
esK
orea
won
,In
done
sia
rupi
ah,
Tha
ilan
dba
ht,
Mal
aysi
ari
nggi
t,P
hili
ppin
espe
so.
fis
adu
mm
yva
riab
leta
king
ava
lue
ofon
efo
rth
ejt
hev
ent
day
show
nbe
low
,or
zero
,jt
othe
rwis
e.T
hus,
the
para
met
eres
tim
ate
Dre
pres
ents
anA
Rfo
rth
ejt
hev
ent
day.
We
only
repr
oduc
eth
ees
tim
ates
ofD
sinc
eth
epa
ram
eter
ses
tim
ates
are
sim
ilar
toth
ose
ofT
able
2.p-
valu
esj
j
are
inbr
acke
ts.
F-t
ests
inth
ela
stco
lum
nar
ejo
int
test
sac
ross
10co
untr
ies.
Est
imat
esof
AR
sU
KG
erm
any
Fra
nce
US
Japa
nK
orea
Indo
nesi
aT
hail
and
Mal
aysi
aP
hili
ppin
esFŽ1
0,36
45.
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9707
01:
0.55
50.
46y
0.32
50.
72y
1.07
60.
260.
869
0.21
0.15
20.
88y
0.29
00.
90y
0.56
30.
840.
306
0.84
y0.
314
0.82
0.65
30.
560.
476
0.91
1.
Tha
ilan
dy
1Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
0702
:y
0.87
20.
241.
001
0.26
0.99
30.
300.
420
0.55
y0.
406
0.68
0.09
40.
97y
0.25
40.
930.
138
0.93
y0.
021
0.99
0.71
30.
520.
466
0.91
2.
Tha
ilan
d0
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9707
03:
y0.
416
0.59
y1.
401
0.13
y0.
503
0.61
0.30
40.
670.
435
0.67
0.15
70.
95y
0.48
30.
87y
0.79
50.
620.
649
0.65
y1.
073
0.35
0.44
60.
923
.T
hail
and
q1
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9707
10:
y0.
573
0.45
y0.
194
0.83
0.16
00.
870.
254
0.72
y1.
103
0.27
0.37
50.
87y
3.11
40.
28y
0.32
70.
831.
474
0.30
y2.
841
0.01
1.11
50.
354
.P
hili
ppin
esy
1Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
0711
:y
0.31
70.
671.
710
0.06
y0.
569
0.56
y0.
116
0.87
y0.
099
0.92
0.99
10.
671.
912
0.51
y0.
603
0.69
y0.
142
0.92
1.81
40.
180.
825
0.60
5.
Phi
lipp
ines
0Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
0714
:y
0.51
70.
49y
0.38
10.
670.
068
0.94
y0.
498
0.48
1.90
30.
06y
0.51
50.
83y
2.80
60.
331.
158
0.45
0.84
50.
561.
224
0.28
0.83
10.
606
.M
alay
sia
0Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
0715
:1.
347
0.07
y0.
048
0.96
0.37
20.
70y
0.99
20.
160.
228
0.82
y1.
324
0.57
3.46
80.
230.
276
0.85
2.32
10.
100.
668
0.56
1.00
70.
447
.M
alay
sia
q1
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9708
13:
0.65
80.
39y
0.25
30.
780.
544
0.58
0.66
10.
350.
284
0.78
y1.
754
0.46
y0.
543
0.85
1.52
80.
320.
622
0.66
y0.
208
0.86
0.36
80.
968
.In
done
sia
y1
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9708
14:
0.35
40.
640.
804
0.38
y0.
481
0.62
0.08
10.
91y
1.48
10.
152.
653
0.26
1.70
60.
56y
0.79
50.
60y
0.12
60.
930.
415
0.72
0.56
20.
859
.In
done
sia
0Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
0815
:y
0.85
30.
260.
378
0.68
y0.
145
0.88
1.08
10.
130.
349
0.73
y0.
920
0.70
1.84
10.
531.
805
0.24
0.04
90.
970.
562
0.62
0.65
50.
7710
.In
done
sia
q1
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9711
14:
0.05
50.
94y
1.28
10.
15y
0.01
00.
99y
1.07
30.
12y
0.27
50.
781.
037
0.65
y1.
782
0.53
1.79
00.
230.
483
0.73
y0.
419
0.71
0.81
50.
6111
.K
orea
y1
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9711
17:
y0.
711
0.35
y0.
027
0.98
y0.
268
0.78
0.76
40.
280.
966
0.35
y1.
543
0.51
1.81
80.
530.
163
0.91
0.10
80.
94y
0.37
80.
740.
414
0.94
12.
Kor
ea0
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9711
18:
y0.
219
0.77
0.72
00.
420.
248
0.80
0.43
20.
54y
0.71
30.
47y
0.28
70.
90y
4.90
50.
091.
740
0.25
1.79
40.
210.
339
0.76
0.72
60.
7013
.K
orea
q1
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216 201
of banks was less than 1% in absolute value. For the US banks, the AR was 1.74%with a p-value of 0.16 in Table 4. Banks in Indonesia had an insignificant AR of3.75%. The returns of the banks of the other crisis countries were positive, exceptfor Korea where banks have a trivial negative AR. None of these ARs issignificant. From this, no case can be made that this event had an adverse effect onbanks in crisis countries once the impact on the market and the exchange rate istaken into account.
The last currency to abandon its peg was the won on November 17. Weestimate the ARs for a window from November 14 to 18. Over that window, nobank index in developed countries experienced an AR greater than 1% in absolutevalue. Korean banks had an insignificant AR of y0.57%. The Indonesian bankslost 4.84%, but the Thai banks gained 3.37% and the Malaysian banks gained2.18%. None of the ARs is significant.
The last period of interest with respect to currency movements is the period ofthe currency meltdown of late July that led Mahathir to blame ‘‘rogue speculators.’’We use an event window from July 23 to 29. This window starts with the ringgitreaching a 38-month low on July 24 and ends when Thailand called the IMF onJuly 28. Malaysian banks did not have much in the way of ARs during this period.While French, US and Korean banks experienced economically large ARs, onlythe French bank index had a significant positive AR. Thai banks had a negativeAR of y2.53%, but this AR is not significant.
The bottom line from our analysis is that there is no evidence that banksperformed poorly during periods of large depreciations. One could even argue thatthese periods were reasonably good for banks. This is especially clear for the US
Ž .banks that have a cumulative abnormal return CAR of 4.01% over the fiveevents discussed above using the data in Table 4. Further, all countries, except forIndonesia and the UK, also have positive CARs over these windows. Theconclusions from our analysis do not change if we exclude exchange rate changesas independent variables.
3.2. IMF programs
Except for Malaysia, all crisis countries received IMF programs. The firstprogram was the one for the Philippines. On July 14, the IMF offered almostUS$1.1 billion of financial support to the Philippines. As discussed earlier, weestimated a window from July 11 to 15 in Table 4. While the ARs for Westernbanks are mixed, but small in absolute value over that period, no bank index inAsia except for Korea had a negative AR during that period. Because this periodwas also one with exchange rate events, we investigate the ARs on July 14 inTable 6, which corresponds to the day of the IMF program agreement for thePhilippines. The announcement seems to leave no trace on Western banks.However, Asian banks experienced positive ARs, except for those of Korea andIndonesia. The only significant AR was the one for Japanese banks, which was1.91% with a p-value of 0.06.
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216202
Tab
le6
Est
imat
esof
the
AR
sof
bank
ing
indu
stry
exce
ssre
turn
son
each
day
ofth
eIM
Fpr
ogra
man
noun
cem
ents
Ž.
The
foll
owin
gS
UR
mod
els
are
esti
mat
edin
asy
stem
for
10co
untr
ies’
bank
ing
indu
stry
indi
ces
over
the
peri
odfr
omJa
nuar
y15
,19
97to
July
15,
1998
390
days
:K
J
Rs
aq
bR
qg
Xq
Df
q´
,p
t0
0m
t0
,kk
,tj
jtt
ÝÝ
ks1
js1
whe
reR
isth
elo
gari
thm
icda
ily
doll
arre
turn
onth
eD
atas
trea
mba
nkin
gin
dust
ryin
dice
sfo
rea
chof
the
10co
untr
ies,
and
Ris
the
corr
espo
ndin
gst
ock
mar
ket
inde
xp
tm
tŽ
.re
turn
.B
oth
retu
rns
are
inex
cess
of1-
day
retu
rnon
the
7-da
yE
urod
olla
rde
posi
t.X
incl
udes
the
dail
ydo
llar
exce
ssre
turn
son
the
curr
ency
hold
ings
BP
,D
M,
FF
,an
dJY
,th
ech
ange
sin
the
kt
7-da
yE
urod
olla
rra
tean
dth
esp
read
betw
een
the
7-da
yE
urod
olla
rra
tean
dE
uroy
enra
te,
lags
ofth
eW
este
rnva
riab
les
due
toth
eti
me
zone
diff
eren
ce,
and
dail
ydo
llar
exce
ssre
turn
son
each
ofŽ
.th
efi
veA
sian
coun
trie
s’cu
rren
cies
Kor
eaw
on,
Indo
nesi
aru
piah
,T
hail
and
baht
,M
alay
sia
ring
git,
Phi
lipp
ines
peso
.f
isa
dum
my
vari
able
taki
nga
valu
eof
one
for
the
jth
even
tda
ysh
own
jt
belo
w,
orze
ro,
othe
rwis
e.T
hus,
the
para
met
eres
tim
ate
Dre
pres
ents
anA
Rfo
rth
ejt
hev
ent
day.
We
only
repr
oduc
eth
ees
tim
ates
ofD
sinc
eth
epa
ram
eter
ses
tim
ates
are
sim
ilar
toth
ose
ofj
j
Tab
le2.
p-va
lues
are
inbr
acke
ts.
F-t
ests
inth
ela
stco
lum
nar
ejo
int
test
sac
ross
10co
untr
ies.
Est
imat
esof
AR
sU
KG
erm
any
Fra
nce
US
Japa
nK
orea
Indo
nesi
aT
hail
and
Mal
aysi
aP
hili
ppin
esFŽ1
0,34
65.
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9707
11:
y0.
266
0.72
1.72
70.
05y
0.53
20.
57y
0.05
30.
940.
012
0.99
0.79
90.
721.
895
0.49
y0.
786
0.56
y0.
165
0.90
1.75
70.
190.
808
0.62
1.
Phi
lipp
ines
y1
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9707
14:
y0.
492
0.51
y0.
371
0.68
0.06
10.
95y
0.45
20.
511.
912
0.06
y0.
616
0.78
y3.
044
0.27
1.07
40.
430.
960
0.48
1.16
30.
300.
860
0.57
2.
Phi
lipp
ines
0Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
0715
:1.
328
0.08
y0.
075
0.93
0.37
60.
69y
0.95
70.
170.
264
0.79
y1.
177
0.60
3.53
30.
200.
186
0.89
2.24
20.
100.
607
0.59
1.02
20.
423
.P
hili
ppin
esq
1Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
0804
:0.
411
0.58
y1.
066
0.23
0.54
10.
56y
0.28
70.
681.
153
0.24
1.14
90.
60y
0.13
90.
96y
0.35
30.
790.
800
0.55
y0.
080
0.94
0.47
20.
914
.T
hail
and
y1
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9708
05:
y0.
735
0.33
y0.
562
0.53
y0.
448
0.63
y0.
635
0.36
1.95
10.
05y
0.26
60.
90y
0.80
90.
770.
132
0.92
0.60
00.
66y
0.56
90.
610.
664
0.76
5.
Tha
ilan
d0
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9708
06:
y0.
109
0.89
y0.
825
0.36
0.92
40.
330.
487
0.49
1.35
00.
180.
695
0.76
y1.
424
0.61
y0.
577
0.67
y1.
420
0.30
y0.
491
0.67
0.60
10.
816
.T
hail
and
q1
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xiD
9708
08:
y1.
218
0.11
1.84
50.
040.
282
0.77
y0.
650
0.36
0.64
90.
520.
547
0.81
y1.
135
0.68
y0.
366
0.79
y0.
458
0.74
y1.
592
0.17
0.98
20.
467
.T
hail
and
y1
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9708
11:
0.67
40.
370.
179
0.84
0.42
10.
651.
140
0.10
1.69
80.
09y
0.93
10.
683.
174
0.25
y0.
081
0.95
y0.
638
0.64
0.50
00.
660.
887
0.54
8.
Tha
ilan
d0
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9708
12:
y0.
957
0.20
0.70
20.
43y
0.30
20.
75y
0.28
00.
69y
0.77
10.
44y
0.09
70.
970.
206
0.94
y0.
281
0.83
1.26
00.
350.
500
0.66
0.44
60.
929
.T
hail
and
q1
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9710
30:
y0.
168
0.82
y0.
464
0.60
y1.
648
0.08
y0.
929
0.18
y0.
199
0.84
3.13
10.
160.
911
0.74
3.59
20.
010.
373
0.78
1.30
80.
241.
368
0.19
10.
Indo
nesi
ay
1Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
1031
:y
1.03
60.
17y
0.11
00.
90y
0.37
10.
70y
0.19
00.
790.
751
0.45
2.43
40.
28y
1.04
60.
711.
059
0.44
y3.
716
0.01
0.10
60.
931.
300
0.22
11.
Indo
nesi
a0
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9711
03:
0.57
10.
44y
0.67
50.
450.
407
0.66
y0.
336
0.63
0.28
10.
78y
6.04
30.
01y
2.95
10.
292.
571
0.06
0.20
50.
880.
899
0.43
1.52
80.
1212
.In
done
sia
q1
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9711
19:
0.63
00.
401.
144
0.20
1.25
60.
181.
012
0.14
y0.
178
0.86
y0.
852
0.70
y0.
088
0.97
y1.
254
0.35
y1.
752
0.19
1.47
10.
191.
101
0.36
13.
Kor
eay
2
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216 203
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9711
20:
y1.
553
0.04
y0.
518
0.56
0.22
20.
810.
000
1.00
y1.
622
0.10
4.39
40.
05y
1.09
90.
691.
272
0.34
3.75
20.
01y
0.49
50.
662.
022
0.03
14.
Kor
eay
1Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
1121
:y
0.58
00.
44y
0.38
10.
670.
237
0.80
y0.
422
0.54
2.01
50.
041.
054
0.64
y0.
380
0.89
y0.
566
0.67
y0.
974
0.48
0.57
00.
610.
651
0.77
15.
Kor
ea0
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9711
24:
y0.
915
0.23
y0.
486
0.59
0.88
80.
350.
164
0.81
0.27
90.
78y
1.34
70.
551.
597
0.56
y2.
506
0.06
y4.
429
0.00
y0.
494
0.66
1.60
80.
1016
.K
orea
q1
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9711
28:
1.20
30.
11y
0.26
70.
77y
1.43
40.
13y
0.28
20.
691.
469
0.15
y4.
020
0.08
0.04
70.
990.
813
0.55
y0.
594
0.67
0.00
80.
991.
217
0.27
17.
Kor
eay
1Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
1201
:y
0.25
50.
740.
059
0.95
0.34
90.
712.
027
0.00
1.00
30.
32y
3.45
30.
130.
810
0.77
y1.
543
0.26
0.34
70.
80y
1.17
00.
301.
315
0.22
18.
Kor
ea0
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9712
02:
1.11
30.
15y
1.19
40.
19y
0.07
90.
930.
217
0.76
y1.
560
0.12
6.14
30.
01y
2.22
90.
42y
2.21
00.
11y
1.84
30.
18y
2.37
20.
042.
266
0.01
19.
Kor
eaq
1Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
1203
:y
0.28
60.
710.
471
0.61
0.70
20.
470.
533
0.46
0.68
90.
514.
201
0.07
1.85
10.
521.
277
0.36
0.28
40.
84y
0.25
70.
830.
713
0.71
20.
Kor
eay
1Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
1204
:0.
338
0.65
0.17
00.
850.
852
0.36
0.92
30.
18y
0.11
70.
91y
0.06
50.
98y
0.48
50.
865.
528
0.00
y2.
078
0.14
y1.
580
0.17
2.50
90.
0121
.K
orea
0Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
1205
:0.
442
0.55
2.12
00.
020.
460
0.62
0.07
70.
910.
475
0.63
0.46
10.
840.
208
0.94
0.05
30.
97y
2.28
00.
10y
2.21
60.
051.
224
0.27
22.
Kor
eaq
1Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D97
1208
:1.
048
0.16
2.32
40.
012.
220
0.02
0.35
60.
611.
267
0.21
7.31
40.
005.
053
0.07
y4.
051
0.00
2.87
00.
04y
0.31
20.
784.
435
0.00
23.
Kor
eaq
2Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D98
0107
:y
0.60
80.
42y
0.30
90.
73y
3.62
40.
00y
1.15
70.
10y
1.16
30.
24y
2.15
70.
333.
501
0.20
y3.
165
0.02
y0.
388
0.78
y1.
356
0.23
3.26
90.
0024
.IM
FN
ego
y1
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9801
08:
y0.
879
0.24
y1.
011
0.26
y2.
775
0.00
y1.
320
0.06
0.86
70.
38y
4.10
60.
077.
200
0.02
y3.
245
0.02
y1.
236
0.37
y2.
203
0.05
3.81
50.
0025
.IM
FN
ego
0Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D98
0109
:y
0.86
10.
25y
1.39
00.
12y
0.33
10.
720.
595
0.40
0.52
40.
60y
3.73
20.
091.
204
0.66
y2.
179
0.10
y0.
733
0.59
0.54
10.
641.
063
0.39
26.
IMF
Neg
oq
1Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D98
0112
:0.
243
0.75
y0.
815
0.36
y0.
750
0.43
y1.
477
0.03
y1.
096
0.27
0.37
20.
870.
974
0.72
y2.
476
0.07
0.63
00.
641.
144
0.31
1.22
80.
2727
.N
ego
q2
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9801
13:
0.55
10.
460.
609
0.50
0.60
50.
520.
253
0.72
0.15
90.
87y
2.67
60.
23y
7.97
60.
001.
812
0.18
2.12
10.
12y
0.06
10.
961.
505
0.13
28.
IMF
Neg
oq
3Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D98
0114
:0.
222
0.77
y0.
459
0.60
0.47
40.
611.
248
0.07
1.16
40.
24y
1.35
70.
54y
6.44
90.
02y
2.02
00.
14y
4.51
50.
00y
1.04
40.
362.
266
0.01
29.
IMF
Neg
oq
4Ž
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
wx
D98
0115
:y
0.20
90.
78y
0.08
50.
920.
486
0.60
0.05
70.
930.
154
0.88
0.44
70.
841.
186
0.67
y5.
739
0.00
y0.
743
0.58
y1.
847
0.10
2.16
10.
0230
.IM
FN
ego
q5
Žw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xw
xD
9801
16:
1.00
60.
18y
0.20
90.
820.
190
0.84
y0.
068
0.92
0.18
00.
86y
2.74
30.
2217
.214
0.00
7.41
90.
00y
0.66
20.
631.
681
0.13
7.39
40.
0031
.IM
FN
ego
q6
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216204
The next program was the one for Thailand. We have two different windows.The first window is from August 4 to 6. It captures the announcement of theausterity measures in Thailand responding to the suggestions of the IMF. ARs forthat event window are small and insignificant at the 10% level for all bank indicesexcept for a 1.95% AR for Japanese banks with a p-value of 0.05. Thai banksearned an insignificant 0.13%. The second event window covers the announce-ment of the program on August 11, so that it goes from August 8 to 12. US andJapanese banks had significant positive ARs on August 11. The banks in crisiscountries had ARs of less than 1% in absolute value and all are insignificant.
A program for Indonesia was announced on October 31. We use a windowfrom October 30 to November 3. During that window, all banking indices, exceptthose for Thailand, the Philippines, and Japan, have negative insignificant ARs.Indonesian and Malaysian banks experienced the biggest losses during that periodand they slightly exceed 3%. Thai banks experienced a large gain of 7.22%.French, German, and American banks lost between 1% and 2%. Japanese banksthat had an exposure in excess of US$20 billion in Indonesia had a smallinsignificant positive AR on the announcement day. American banks had a smallnegative AR. Indonesian banks lost an insignificant 1.05%. The hypothesis, thatthe ARs across countries were equal to zero on October 31, is not rejected.
Following the announcement of the IMF program on October 31, Indonesia hada number of negotiations with the IMF leading to further agreements andclarifications. In particular, on January 15, 1998, Suharto signed an agreementwith the IMF. We look at three different windows in Table 4: from January 8 to12, from January 13 to 14, and finally, from January 15 to 16. For all thesewindows, we can reject the hypothesis that bank ARs are equal to zero acrosscountries. During the first window, there were substantial concerns about reachingan agreement between developed country banks and Indonesian corporations. OnJanuary 9, the New York Times had a negative headline on Indonesia. ThePeregrine bank collapse on January 12 in Hong Kong was unrelated to the IMFprogram in Indonesia. An equally weighted portfolio of Western banks lost 2.83%
Ž .with a p-value of less than 0.01 during the first window not reported in the tableand made small but positive gains during the next two windows. US banks lost2.16% during the first period, which is significant at the 10% level. Korean bankslost a dramatic 7.19%. Thai banks lost 7.86%. In contrast, Indonesian banksgained 8.76%. During the window corresponding to the IMF agreement announce-ment, the returns to Western banks were trivial, but Indonesian banks earned18.17% on January 15 and 16 after having lost 14.56% on January 13 and 14.During March 1998, issues surrounding the IMF agreement with Indonesia wereprominent. A meeting between the IMF and Indonesia took place on March 21after Indonesia gave up a plan of using a currency board on March 20. US bankslost a significant 1.62% during that window and Indonesian banks lost 6.12% witha p-value of 0.12 from March 20 to 23. A new agreement between the IMF andIndonesia took place on April 8. No bank index had a significant AR for an event
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216 205
window from April 7 to 9. All Asian banks except for Malaysia and thePhilippines had negative ARs during this window, with Korean banks losing6.49% and Indonesian banks losing 2.36%.
Korea announced that it would seek help from the IMF on November 21. WeŽconsider an event window from November 19 to 24 November 22 and 23
.correspond to a weekend . Table 4 shows that this window has positive ARs infive countries and negative ARs for the rest. Banks in Korea earned an insignifi-cant 3.13% over that window. Banks in the UK earned an AR of y2.35% with ap-value of 0.12. Looking at daily ARs in Table 6 gives a clearer picture. OnNovember 21, all Western banks had insignificant ARs of less than 0.6% inabsolute value. Japanese banks had a large significant AR of 2.02% with ap-value of 0.04. Korean banks made a significant gain of 4.39% on November 20and an insignificant gain of 1.05% on November 21. For November 24, we canreject the hypothesis that ARs of banks across the world were equal to zero with ap-value of 0.10. Malaysian and Thai banks all had significant negative ARs on thatday at the 10% level.
The IMF program for Korea was accepted at the beginning of December. OnDecember 1, an agreement was announced by Korea, but this collapsed the nextday. A more definitive agreement was announced on December 4. We thereforeuse two windows in Table 4. The first one is from November 28 to December 2and the second one is from December 3 to 8. All countries, except for Malaysiaand the Philippines, had positive ARs for the second window, which is the onecorresponding to the acceptance of the program. These ARs are significant withp-values of 0.05 or better in three countries: Germany had an AR of 4.97% with ap-value of 0.01, France had an AR of 4.45% with a p-value of 0.02, and Koreahad an AR of 12.50% with a p-value of less than 0.01. US banks had an AR of1.88% with a p-value of 0.19. No country, except for the Philippines andMalaysia, had an AR of less than 1.6%. The results for the first window are moremixed. The reason for this turns out to be straightforward. This window includesboth good news and bad news for the IMF program. A good approach to obtain anestimate for the US is to look at New York Times headlines. On December 1, theNew York Times had a front page headline stating that Korea had agreed to a hugebailout. The AR for the US banks on that day is shown on Table 6 and was 2.03%with a p-value of less than 0.01.
4. The experience of US banks with high exposures
The general conclusion that emerges from the event study is that devaluationscannot explain the poor performance of banks in Asia and generally, when IMFprogram announcements convey information, that information is positive. We sawevidence that Western banks were affected by events in Asia, especially theannouncements related to the IMF program in Korea. If we are correct in
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216206
Tab
le7
Est
imat
esof
the
CA
Rs
ofth
ree
US
bank
s’ex
cess
retu
rns
arou
ndne
ws
anno
unce
men
tsŽ
.Ž
.T
hefo
llow
ing
SU
Rm
odel
sar
ees
tim
ated
ina
syst
emfo
rth
ree
US
bank
s’ex
cess
retu
rns
Cha
se,
Cit
icor
p,an
dJP
Mor
gan
over
the
peri
odfr
omJa
nuar
y15
,19
97to
July
15,
1998
390
days
:K
J
Rs
aq
bR
qg
Xq
Df
q´
,p
t0
0m
t0
,kk
tj
jtt
ÝÝ
ks1
js1
Ž.
whe
reR
isth
elo
gari
thm
icda
ily
doll
arre
turn
onon
eof
the
US
bank
stoc
ksfr
omD
atas
trea
m,
and
Ris
the
corr
espo
ndin
gU
Sst
ock
mar
ket
inde
xre
turn
.B
oth
pt
mt
Ž.
retu
rns
are
inex
cess
of1-
day
retu
rnon
the
7-da
yE
urod
olla
rde
posi
t.K
incl
udes
the
dail
ydo
llar
exce
ssre
turn
son
the
curr
ency
hold
ings
BP
,D
M,
FF
,an
dJY
,an
dth
ech
ange
sin
the
7-da
yk
t
Eur
odol
lar
rate
and
the
spre
adbe
twee
nth
e7-
day
Eur
odol
lar
rate
and
Eur
oyen
rate
.f
isa
dum
my
vari
able
taki
nga
valu
eof
1rn
for
the
jth
even
tda
yssh
own
belo
w,
orze
ro,
othe
rwis
e.T
hus,
jt
the
para
met
eres
tim
ate
Dre
pres
ents
aC
AR
for
the
jth
even
tda
ys.
We
only
repr
oduc
eth
ees
tim
ates
ofD
sinc
eth
epa
ram
eter
ses
tim
ates
are
sim
ilar
toth
ose
ofT
able
2.p-
valu
esar
ein
brac
kets
.j
j
F-t
ests
next
toth
ela
stco
lum
nar
ejo
int
test
sac
ross
thre
eba
nks.
The
last
four
colu
mns
show
the
resu
lts
for
the
thre
eeq
ual-
,va
lue-
wei
ghte
dth
ree-
US
bank
port
foli
os,
US
bank
inde
xex
clud
ing
the
thre
eba
nks,
and
the
diff
eren
ce.
Est
imat
esof
CA
Rs
Cha
seC
itic
orp
JPM
orga
nF
EW
thre
e-ba
nkV
Wth
ree-
bank
Ban
kin
dex
VW
thre
eba
nksy
Ž3,
1047
.po
rtfo
lio
port
foli
ow
itho
utth
ree
bank
sin
dex
wit
hout
thre
e
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
D97
0514
–97
0516
:n
s3
y3.
647
0.14
y0.
668
0.85
y0.
004
1.00
0.98
50.
40y
1.43
90.
53y
1.62
70.
53y
0.21
50.
83y
1.41
20.
541Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9705
22–
9705
26:
ns
31.
229
0.62
0.16
80.
96y
0.06
00.
980.
117
0.95
0.44
60.
850.
426
0.87
0.20
30.
840.
223
0.92
2Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9706
18–
9706
20:
ns
3y
0.14
90.
951.
590
0.64
y0.
124
0.96
0.12
80.
940.
439
0.85
0.61
30.
810.
529
0.60
0.08
40.
973Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9706
26–
9706
30:
ns
3y
2.26
30.
36y
0.79
90.
82y
3.35
20.
190.
749
0.52
y2.
138
0.35
y1.
804
0.48
y1.
890
0.06
0.08
60.
974Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9707
01–
9707
03:
ns
32.
792
0.26
2.69
40.
441.
643
0.52
0.45
00.
722.
377
0.30
2.62
00.
311.
161
0.25
1.45
90.
535Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9707
11–
9707
15:
ns
3y
0.48
60.
84y
4.22
50.
22y
1.53
30.
550.
593
0.62
y2.
081
0.36
y2.
491
0.33
y1.
612
0.11
y0.
880
0.70
6Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9707
23–
9707
29:
ns
51.
902
0.55
1.52
60.
734.
422
0.19
0.62
50.
602.
617
0.38
2.09
60.
531.
846
0.16
0.25
00.
937Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9708
04–
9708
06:
ns
3y
2.21
90.
371.
731
0.62
y1.
306
0.61
0.77
10.
51y
0.59
80.
79y
0.18
90.
940.
627
0.53
y0.
816
0.72
8Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9708
08–
9708
12:
ns
32.
724
0.27
1.78
60.
610.
347
0.89
0.46
30.
711.
619
0.48
1.81
90.
48y
0.63
70.
532.
456
0.29
9Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9708
13–
9708
15:
ns
33.
775
0.13
2.83
90.
421.
361
0.60
0.78
40.
502.
658
0.25
2.85
50.
271.
475
0.15
1.38
00.
5510
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
D97
0819
–97
0821
:n
s3
1.65
30.
50y
0.62
00.
86y
0.64
40.
800.
344
0.79
0.13
00.
950.
172
0.95
y0.
443
0.66
0.61
50.
7911
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
D97
0919
–97
0922
:n
s2
y0.
615
0.76
y0.
447
0.87
2.40
90.
250.
790
0.50
0.44
90.
81y
0.10
30.
96y
0.96
80.
240.
865
0.64
12
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216 207
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
D97
1007
–97
1009
:n
s3
0.46
50.
850.
907
0.79
y1.
409
0.58
0.24
20.
87y
0.01
21.
000.
294
0.91
y0.
871
0.38
1.16
50.
6113
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
D97
1030
–97
1103
:n
s3
y3.
678
0.13
y2.
814
0.41
y2.
373
0.35
0.78
50.
50y
2.95
50.
19y
3.04
40.
23y
0.66
20.
51y
2.38
20.
3014
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
D97
1104
–97
1106
:n
s3
1.37
50.
582.
323
0.50
1.99
30.
440.
247
0.86
1.89
70.
411.
833
0.47
0.62
70.
531.
206
0.60
15Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9711
14–
9711
18:
ns
3y
2.89
40.
24y
0.11
30.
97y
1.95
10.
450.
701
0.55
y1.
653
0.47
y1.
343
0.60
0.21
40.
83y
1.55
60.
4916
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
D97
1119
–97
1124
:n
s4
1.64
00.
56y
0.89
40.
820.
608
0.84
0.24
40.
870.
451
0.86
0.31
60.
910.
780
0.50
y0.
465
0.86
17Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9711
28–
9712
02:
ns
32.
857
0.25
4.83
50.
162.
904
0.26
0.79
60.
503.
532
0.12
3.93
50.
131.
124
0.26
2.81
10.
2218
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
D97
1203
–97
1208
:n
s4
1.17
90.
688.
380
0.04
0.54
00.
861.
852
0.14
3.36
60.
204.
447
0.13
2.53
60.
031.
911
0.47
19Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9712
12–
9712
16:
ns
3y
1.04
80.
67y
0.71
60.
831.
290
0.61
0.28
00.
84y
0.15
80.
94y
0.50
30.
840.
620
0.53
y1.
123
0.62
20Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9712
17–
9712
19:
ns
3y
0.04
80.
992.
180
0.54
1.09
90.
680.
200
0.90
1.07
70.
651.
143
0.66
y0.
447
0.66
1.59
10.
5021
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
D97
1223
–97
1226
:n
s4
1.30
30.
67y
1.64
70.
70y
0.61
10.
850.
247
0.86
y0.
318
0.91
y0.
467
0.88
y0.
559
0.65
0.09
20.
9722
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
D97
1229
–97
1231
:n
s3
y1.
516
0.57
y1.
763
0.64
y4.
438
0.11
0.87
60.
45y
2.57
30.
30y
2.09
50.
450.
675
0.53
y2.
770
0.27
23Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9801
01–
9801
05:
ns
32.
789
0.26
2.74
50.
422.
482
0.33
0.52
50.
672.
672
0.24
2.68
60.
29y
0.65
50.
513.
341
0.14
24Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9801
08–
9801
12:
ns
3y
1.99
00.
42y
3.76
70.
27y
4.84
00.
061.
211
0.30
y3.
532
0.12
y3.
256
0.20
y3.
372
0.00
0.11
60.
9625
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
D98
0113
–98
0114
:n
s2
y0.
928
0.64
2.41
10.
391.
288
0.54
0.71
60.
540.
923
0.62
1.00
70.
632.
062
0.01
y1.
055
0.57
26Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9801
15–
9801
16:
ns
20.
384
0.85
0.05
70.
98y
0.59
90.
780.
070
0.98
y0.
053
0.98
0.05
40.
980.
206
0.80
y0.
152
0.94
27Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9801
28–
9801
30:
ns
30.
122
0.96
y0.
141
0.97
y2.
714
0.29
0.51
50.
67y
0.91
10.
69y
0.48
20.
85y
0.51
10.
610.
029
0.99
28Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9802
13–
9802
18:
ns
41.
327
0.64
y0.
680
0.86
0.12
00.
970.
152
0.93
0.25
60.
920.
152
0.96
y0.
240
0.83
0.39
20.
8829
Ž.
wx
wx
wx
wx
wx
wx
wx
wx
D98
0306
–98
0310
:n
s3
y2.
397
0.33
y1.
815
0.60
y2.
592
0.31
0.47
10.
70y
2.26
80.
32y
2.19
00.
390.
731
0.46
y2.
921
0.20
30Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9803
20–
9803
23:
ns
2y
2.34
90.
24y
3.14
90.
26y
0.54
10.
800.
675
0.57
y2.
013
0.28
y2.
415
0.24
y1.
013
0.21
y1.
402
0.45
31Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9803
25–
9803
27:
ns
30.
842
0.73
y0.
471
0.89
y0.
438
0.86
0.10
70.
96y
0.02
30.
99y
0.02
50.
99y
0.67
00.
500.
645
0.78
32Ž
.w
xw
xw
xw
xw
xw
xw
xw
xD
9804
07–
9804
09:
ns
3y
3.06
10.
21y
7.74
30.
03y
2.15
50.
401.
709
0.16
y4.
319
0.06
y4.
967
0.05
y0.
233
0.82
y4.
734
0.04
33
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216208
attributing the positive ARs to the announcement of IMF programs, it should bethe case that banks with greater exposure should be affected more by these events.We therefore investigate the experience of three US banks: Chase, JP Morgan, andCiticorp. We form an equally weighted portfolio of these three banks and estimatethe returns for the event-windows discussed in Section 3. We also investigateChase separately because it had the largest exposures. In addition to the events ofSection 3, we also consider the earnings announcement events to evaluate theability of the market to understand the implications of the Asian events for thebanks. If the market underestimated the impact of the crisis on banks, earningsannouncements would be informative.
4.1. Currency deÕaluations
Table 7 shows that Chase earned 2.79% over the window corresponding to theThai baht devaluation on July 2 with a p-value of 0.26. The equally weightedportfolio of Chase, Citicorp and JP Morgan earned 2.38% with a p-value of 0.30.The devaluation of the peso decreased the equally weighted portfolio by slightlymore than 2%. Interestingly, however, the loss was only 0.02% on the day of thedevaluation. Chase then earned 3.78% when the rupiah devalued on August 14with a p-value of 0.13. For that event, the equally weighted portfolio earned2.66% with a p-value of 0.25. When the won devalued on November 17, Chaselost 2.89% with a p-value of 0.24, while the equally weighted portfolio fell by
Ž .1.65%. During the period of the dramatic fall in the ringgit July 23–29 , theequally weighted portfolio earned 2.62% with a p-value of 0.38. We estimated theCAR associated with the days when it was announced that the five currencieswould no longer be defended. For Chase, these 5 days had a CAR of 5.09% fromTable 7. For Citicorp, the AR was 2.72%. Finally, for JP Morgan, the AR was3.94%. This evidence suggests, therefore, that the events corresponding to theabandonment of the defense of parities and massive devaluations of the fivecurrencies were positive events for the three US banks.
4.2. IMF announcements
Table 8 shows that the Philippines program announced on July 14 had noinformation for the equally weighted portfolio. The banks had insignificantpositive ARs at the announcement of the Thai program on August 11 andinsignificant negative ARs at the announcement of the Indonesian program at theend of October. On December 1, Chase earned 3.65% with a p-value of 0.01. Itlost 0.96% on December 2 when it looked like the IMF program was in trouble. Itthen earned 0.38% the next day and 1.83% on December 4. To put things inperspective, the AR of Chase on December 1 was the highest AR of Chase for themonth of December 1997 and January 1998. No other crisis event during thatperiod had such an effect on Chase. The value-weighted portfolio earned a highly
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216 209
significant 4.55% on December 1, lost 0.81% the next day, and gained 0.36% onDecember 3 and 3.27% on December 4 with a p-value of 0.02. In total, thisamounted to more than 7% of the market value of these three banks. Again, noother day during December or January and no other event during our wholesample period has a higher AR for the equally weighted portfolio than the AR onDecember 1. One might be concerned that other news explains the ARs corre-sponding to the days discussed in this paragraph. However, news published innewspapers cannot explain these ARs. The Wall Street Journal index had no newsabout Citicorp and Chase in early December. It had news about JP Morgan,namely the suspension of traders for possible price manipulation on December 2and a realignment of duties of high aides on December 5.
An important question is whether the announcement of the Korean IMFprograms had a general positive impact on banks because of a reduction insystemic risk or just affected positively the banks with exposures in Korea. Table8 makes it possible to answer that question. In that table, we compute ARs for twopieces of the Datastream US retail banking index. The first piece is the AR on avalue-weighted portfolio of the three banks discussed in this section. The secondpiece is the value-weighted portfolio of the other banks in the Datastream index.The AR on the banks that did not have much of an exposure to Korea wasinsignificant and less than 1% for each announcement. Hence, there is a dramaticdifference in the impact of the announcements related to Korea’s IMF programbetween the banks with the highest exposure and the other banks in the Datas-tream index. This suggests that the benefit of the program for banks is restricted tothe banks with exposure. It is inconsistent with the view that such programs arejustified because of their benefit in reducing systemic risk.
4.3. Earnings announcements
An important question is whether the market correctly interpreted the implica-tions of the Asian crisis for the banks discussed in this section. On December 10,1997, JP Morgan warned that its earnings were hurt by the behavior of themarkets. This pre-announcement of earnings had a strong effect on the three banksdiscussed in this section. JP Morgan had a negative AR of y3.18% on December
Ž .10 with a p-value of 0.04 not reported . Chase had a negative AR of y3.49% onDecember 10 with a p-value of 0.02. Citicorp had an AR of y4.62% with ap-value of 0.03 on December 10. The banks’ earnings were affected by marketturmoil that decreased their trading income. When earnings were subsequentlyannounced, however, they had no significant effects on the stock price of thesebanks. The earnings pre-announcement of JP Morgan therefore conveyed substan-tial information about the banks that was not available to investors beforehand.Strikingly, however, this earnings pre-announcement did not have a significantimpact on the Datastream bank index for the US. On December 10, the AR on thatindex was y0.97% and insignificant.
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216210
Tab
le8
Est
imat
esof
the
AR
sof
thre
eU
Sba
nks’
exce
ssre
turn
son
each
day
ofth
eIM
Fpr
ogra
man
noun
cem
ents
Ž.
Ž.
The
foll
owin
gS
UR
mod
els
are
esti
mat
edin
asy
stem
for
thre
eU
Sba
nks’
exce
ssre
turn
sC
hase
,C
itic
orp,
and
JPM
orga
nov
erth
epe
riod
from
Janu
ary
15,
1997
toJu
ly15
,19
9839
0da
ys:
KJ
Rs
aq
bR
qg
Xq
Df
q´
,p
t0
0m
t0
,kk
tj
jtt
ÝÝ
ks1
js1
Ž.
whe
reR
isth
elo
gari
thm
icda
ily
doll
arre
turn
onon
eof
the
US
bank
stoc
ksfr
omD
atas
trea
m,
and
Ris
the
corr
espo
ndin
gU
Sst
ock
mar
ket
inde
xre
turn
.B
oth
pt
mt
Ž.
retu
rns
are
inex
cess
of1-
day
retu
rnon
the
7-da
yE
urod
olla
rde
posi
t.X
incl
udes
the
dail
ydo
llar
exce
ssre
turn
son
the
curr
ency
hold
ings
BP
,D
M,
FF
,an
dJY
,an
dth
ech
ange
sin
the
7-da
yk
t
Eur
odol
lar
rate
and
the
spre
adbe
twee
nth
e7-
day
Eur
odol
lar
rate
and
Eur
oyen
rate
.f
isa
dum
my
vari
able
taki
nga
valu
eof
one
for
the
jth
even
tda
ysh
own
belo
w,
orze
ro,
othe
rwis
e.T
hus,
the
jt
para
met
eres
tim
ate
Dre
pres
ents
anA
Rfo
rth
ejt
hev
ent
day.
We
only
repr
oduc
eth
ees
tim
ates
ofD
sinc
eth
epa
ram
eter
ses
tim
ates
are
sim
ilar
toth
ose
ofT
able
2.p-
valu
esar
ein
brac
kets
.j
j
F-t
ests
next
toth
ela
stco
lum
nar
ejo
int
test
sac
ross
thre
eba
nks.
The
last
four
colu
mns
show
the
resu
lts
for
the
thre
eeq
ual-
,va
lue-
wei
ghte
dU
Sba
nkpo
rtfo
lios
,U
Sba
nkin
dex
excl
udin
gth
eth
ree
bank
s,an
dth
edi
ffer
ence
.
Est
imat
esof
AR
sC
hase
Cit
icor
pJP
Mor
gan
FE
Wth
ree-
bank
VW
thre
e-ba
nkB
ank
inde
xV
Wth
ree
bank
syŽ3
,10
53.
port
foli
opo
rtfo
lio
wit
hout
thre
eba
nks
inde
xw
itho
utth
ree
Žw
xw
xw
xw
xw
xw
xw
xw
xD
9707
11:
0.30
10.
83y
0.72
40.
710.
364
0.80
0.16
30.
92y
0.02
00.
99y
0.20
10.
890.
163
0.77
y0.
365
0.78
1.
Phi
lipp
ines
y1
Žw
xw
xw
xw
xw
xw
xw
xw
xD
9707
14:
y0.
270
0.85
0.09
60.
96y
1.15
80.
420.
288
0.83
y0.
444
0.73
y0.
255
0.86
y0.
813
0.15
0.55
80.
662
.P
hili
ppin
es0
Žw
xw
xw
xw
xw
xw
xw
xw
xD
9707
15:
y0.
373
0.79
y3.
435
0.08
y0.
550
0.70
1.24
10.
29y
1.45
30.
26y
1.86
70.
20y
0.91
30.
11y
0.95
40.
463
.P
hili
ppin
esq
1Ž
wx
wx
wx
wx
wx
wx
wx
wx
D97
0804
:y
1.52
00.
27y
0.33
60.
86y
0.79
20.
580.
487
0.69
y0.
883
0.49
y0.
858
0.55
y0.
228
0.68
y0.
629
0.62
4.
Tha
ilan
dy
1Ž
wx
wx
wx
wx
wx
wx
wx
wx
D97
0805
:y
1.49
50.
28y
0.29
30.
88y
0.10
00.
940.
487
0.69
y0.
630
0.62
y0.
712
0.62
0.32
40.
57y
1.03
70.
425
.T
hail
and
0Ž
wx
wx
wx
wx
wx
wx
wx
wx
D97
0806
:0.
892
0.53
2.52
30.
20y
0.43
80.
760.
896
0.44
0.99
20.
451.
498
0.31
0.41
00.
471.
087
0.40
6.
Tha
ilan
dq
1Ž
wx
wx
wx
wx
wx
wx
wx
wx
D97
0808
:0.
349
0.81
0.46
50.
82y
0.60
70.
680.
162
0.92
0.06
90.
960.
204
0.89
y1.
013
0.08
1.21
70.
367
.T
hail
and
y1
Žw
xw
xw
xw
xw
xw
xw
xw
xD
9708
11:
1.07
50.
441.
388
0.48
2.41
60.
090.
941
0.42
1.62
60.
211.
431
0.32
0.58
50.
300.
846
0.51
8.
Tha
ilan
d0
Žw
xw
xw
xw
xw
xw
xw
xw
xD
9708
12:
1.40
50.
310.
260
0.89
y1.
426
0.33
1.16
30.
320.
080
0.95
0.37
70.
79y
0.26
10.
640.
638
0.62
9.
Tha
ilan
dq
1Ž
wx
wx
wx
wx
wx
wx
wx
wx
D97
1030
:y
1.93
40.
17y
1.05
50.
59y
1.51
20.
300.
762
0.52
y1.
500
0.25
y1.
447
0.32
y0.
224
0.69
y1.
223
0.34
10.
Indo
nesi
ay
1Ž
wx
wx
wx
wx
wx
wx
wx
wx
D97
1031
:y
1.75
30.
22y
1.91
80.
33y
1.44
60.
320.
620
0.60
y1.
706
0.19
y1.
804
0.22
y0.
097
0.87
y1.
708
0.19
11.
Indo
nesi
a0
Žw
xw
xw
xw
xw
xw
xw
xw
xD
9711
03:
0.04
60.
970.
129
0.95
0.70
00.
630.
094
0.96
0.29
10.
820.
233
0.87
y0.
317
0.58
0.54
90.
6712
.In
done
sia
q1
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216 211
Žw
xw
xw
xw
xw
xw
xw
xw
xD
9711
19:
3.11
70.
031.
757
0.37
2.77
90.
052.
198
0.09
2.55
10.
052.
468
0.09
0.61
20.
281.
856
0.15
13.
Kor
eay
2Ž
wx
wx
wx
wx
wx
wx
wx
wx
D97
1120
:y
0.88
30.
53y
0.68
20.
73y
0.96
10.
510.
202
0.90
y0.
842
0.51
y0.
813
0.57
0.39
80.
48y
1.21
10.
3414
.K
orea
y1
Žw
xw
xw
xw
xw
xw
xw
xw
xD
9711
21:
0.33
80.
81y
0.53
80.
78y
0.79
00.
590.
213
0.89
y0.
330
0.80
y0.
266
0.85
y0.
660
0.24
0.39
40.
7615
.K
orea
0Ž
wx
wx
wx
wx
wx
wx
wx
wx
D97
1124
:y
0.90
00.
52y
1.36
80.
49y
0.34
60.
810.
211
0.89
y0.
871
0.50
y1.
012
0.49
0.44
00.
44y
1.45
10.
2616
.K
orea
q1
Žw
xw
xw
xw
xw
xw
xw
xw
xD
9711
28:
y0.
118
0.93
0.27
10.
89y
0.90
30.
540.
201
0.90
y0.
250
0.85
y0.
094
0.95
y0.
104
0.86
0.01
00.
9917
.K
orea
y1
Žw
xw
xw
xw
xw
xw
xw
xw
xD
9712
01:
3.64
60.
015.
139
0.01
3.67
90.
013.
418
0.02
4.15
50.
004.
549
0.00
0.77
90.
173.
770
0.00
18.
Kor
ea0
Žw
xw
xw
xw
xw
xw
xw
xw
xD
9712
02:
y0.
955
0.50
y0.
903
0.65
y0.
098
0.95
0.18
80.
90y
0.65
20.
62y
0.81
30.
580.
494
0.39
y1.
307
0.32
19.
Kor
eaq
1Ž
wx
wx
wx
wx
wx
wx
wx
wx
D97
1203
:0.
383
0.79
0.76
00.
71y
0.66
10.
660.
232
0.87
0.16
00.
900.
361
0.81
0.93
20.
11y
0.57
10.
6720
.K
orea
y1
Žw
xw
xw
xw
xw
xw
xw
xw
xD
9712
04:
1.83
30.
194.
701
0.02
2.29
50.
112.
054
0.10
2.94
30.
023.
269
0.02
0.89
70.
112.
372
0.06
21.
Kor
ea0
Žw
xw
xw
xw
xw
xw
xw
xw
xD
9712
05:
y0.
903
0.52
1.65
80.
400.
679
0.64
0.82
20.
480.
478
0.71
0.58
20.
690.
089
0.88
0.49
30.
7022
.K
orea
q1
Žw
xw
xw
xw
xw
xw
xw
xw
xD
9712
08:
0.02
30.
991.
450
0.46
y1.
665
0.25
1.11
50.
34y
0.06
40.
960.
411
0.78
0.51
00.
37y
0.10
00.
9423
.K
orea
q2
Žw
xw
xw
xw
xw
xw
xw
xw
xD
9801
07:
1.50
20.
28y
0.21
40.
911.
183
0.41
0.72
30.
540.
823
0.52
0.63
20.
66y
2.14
30.
002.
776
0.03
24.
IMF
Neg
oy
1Ž
wx
wx
wx
wx
wx
wx
wx
wx
D98
0108
:y
3.18
30.
02y
2.57
00.
19y
2.83
60.
052.
141
0.09
y2.
863
0.03
y2.
810
0.05
y1.
310
0.02
y1.
501
0.24
25.
IMF
Neg
o0
Žw
xw
xw
xw
xw
xw
xw
xw
xD
9801
09:
1.32
90.
352.
065
0.30
1.35
00.
360.
493
0.69
1.58
10.
221.
646
0.26
y0.
736
0.20
2.38
20.
0726
.IM
FN
ego
q1
Žw
xw
xw
xw
xw
xw
xw
xw
xD
9801
12:
y0.
228
0.87
y3.
276
0.09
y3.
408
0.02
2.56
80.
05y
2.30
40.
07y
2.13
90.
14y
1.38
40.
01y
0.75
50.
5627
.IM
FN
ego
q2
Žw
xw
xw
xw
xw
xw
xw
xw
xD
9801
13:
y0.
692
0.62
y0.
887
0.65
2.09
60.
151.
444
0.23
0.17
30.
89y
0.32
30.
820.
935
0.10
y1.
258
0.33
28.
IMF
Neg
oq
3Ž
wx
wx
wx
wx
wx
wx
wx
wx
D98
0114
:y
0.13
20.
923.
377
0.08
y0.
685
0.63
1.90
50.
130.
854
0.50
1.42
80.
321.
164
0.04
0.26
40.
8429
.IM
FN
ego
q4
Žw
xw
xw
xw
xw
xw
xw
xw
xD
9801
15:
y0.
309
0.82
0.10
50.
96y
0.44
30.
760.
061
0.98
y0.
216
0.87
y0.
164
0.91
y0.
331
0.56
0.16
60.
9030
.IM
FN
ego
q5
Žw
xw
xw
xw
xw
xw
xw
xw
xD
9801
16:
0.78
00.
580.
059
0.98
y0.
082
0.95
0.15
40.
930.
253
0.84
0.30
70.
830.
522
0.36
y0.
215
0.87
31.
IMF
Neg
oq
6
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216212
The events in East Asia left little trace on the three banks when they announcedtheir earnings for the last quarter of 1997. On January 20, 1998, JP Morganannounced that it was designating more than 10% of its exposures to South Korea,Malaysia and Thailand as non-performing. At that time, analysts were interpretingthis decision as a political decision made to influence the bargaining process.13
The non-performing positions were primarily swaps rather than loans. On JanuaryŽ .26, Standard and Poor’s S&P placed JP Morgan on CreditWatch ‘‘to reflect the
possibility of a modest downgrade.’’ The S&P analyst stated that ‘‘There aremany reasons for this. The southeast Asian situation, on its own, would not havejustified this action.’’ Asked whether other banks would be downgraded, sheadded that ‘‘As long as the southeast Asian crisis remains isolated to thosecountries which have been given bailouts by the International Monetary Fund,there will be no need to downgrade US banks’’.14
5. Conclusion
In this paper, we examined the impact of the Asian crisis on the shareholderwealth of Western and East Asian banks. Our main conclusions are as follows.
Ž .1 Equity investors in shares in East Asian banks made dramatic losses.Ž .2 The impact of the crisis on Western banks, in general, was small and was
not enough to prevent these banks from outperforming their respective markets.Ž .3 Although much attention has been paid to the dramatic loss in value of the
East Asian currencies and many have argued that this loss played a major role inthe difficulties of the banks in these countries, our evidence shows that changes inexchange rates have no additional explanatory power for the performance of banksbeyond their impact on general market movements in Korea, Thailand, andMalaysia. There was a direct impact of currency exposures on bank performancein the Philippines and Indonesia.
Ž .4 We can reject the hypothesis that the initial currency collapses across EastAsian countries hurt US banks. For instance, Chase Manhattan earned an AR of5.09% across the 5 days when the East Asian countries announced that they wouldno longer defend their peg.
Ž .5 In total, the IMF program announcements increased bank shareholderwealth. Of all the events we consider, the IMF program announcement with Koreastands out. It is one of five events where we find evidence that banks weresignificantly affected across countries. Banks with the greatest exposure in the USexperienced significant ARs in excess of 7% from this announcement.
13 See J. Authers, JP Morgan redesignates Asia loans, Financial Times, London edn., January 21,1998, p. 28.
14 See J. Authers, S&P places JP Morgan on CreditWatch, Financial Times, USA edn., January 27,1998, p. 26.
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216 213
Ž .6 The IMF program in Korea benefited only those US banks with largeexposures in Korea. No case can therefore be made that the IMF programs havethe positive effect of somehow reducing systemic risk.
Further research should investigate more closely why the currency collapses didnot hurt bank shareholder equity more directly. There are three possible explana-tions worth considering. First, it could be that banks were hedged more thancommentators believed they were. Second, it could be that the market expectedcurrency losses to be offset by bailouts. Third, the market might have beeninefficient in incorporating information about exchange rate changes. We exploredthe third hypothesis to some extent and found no support for it. Detailedaccounting data would be required to investigate the first two hypotheses.
Acknowledgements
Bong-Chan Kho and Rene M. Stulz, respectively, are Lecturer, Seoul National´University, and Reese Chair of Banking and Monetary Economics, The Ohio StateUniversity, and Research Associate, NBER. We thank Dong Lee for researchassistance. We are grateful for comments from Bhagwan Chowdhry, Jan Jindra,Andrew Karolyi, Olivier Jeanne, Rick Mishkin, an anonymous referee, and toseminar participants at the IMF, Federal Reserve Board, Columbia University, theOhio State University, and the University of St. Gallen.
Appendix A. Brief description of news announcements on selected event days(From the website ‘Asia’s Currency and Economic Crisis’ by Noriel Roubini,
)http:rrrrrrrrrrwww.stern.nyu.edurrrrr;nroubinirrrrrasiarrrrrAsiaHomepage.html
May 14–15, 1997 Thailand’s currency is hit by a massive attack by speculators.May 23 Moves to save Finance One, Thailand’s largest finance
company, fails.June 19 Amnuay Viravan, staunchly against devaluing the baht, re -
signs as Thailand’s finance minister.June 27 The Thai central bank suspends operations of 16 cash-strapped
finance companies.June 30 Thai Prime Minister Chavalit Yonchaiyudh assures that there
will be no devaluation of the baht.July 2 The Bank of Thailand announces a managed float of the baht
and calls on the IMF for technical assistance. The announce-ment effectively devalues the baht by about 15–20% to 28.80per dollar. This is a trigger for the East Asian crisis.
July 14 The IMF offers the Philippines almost US$1.1 billion infinancial support. The Malaysian central bank abandons thedefense of the ringgit.
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216214
July 24 The ringgit hits 38-month low of 2.6530 per dollar. MalaysianPrime Minister Mahathir Mohamad launches bitter attack on‘‘rogue speculators’’.
July 26 Malaysian Prime Minister Mahathir names hedge fund man-ager, George Soros, as the man responsible for the attack onthe ringgit. He later brands Soros a ‘‘moron’’.
July 28 Thailand calls in the IMF.August 5 Thailand unveils austerity plan and completes revamp of
finance sector as part of an IMF rescue package.August 11 The IMF unveils a rescue package for Thailand including
loans totaling US$16 billion.August 14 Indonesia abolishes its system of managing the exchange rate
through a band and allows it to float. The rupiah plunges to2,755 per dollar. Bank Indonesia tries mopping up liquiditywith high interest rates.
August 20 IMF approves a US$3.9 billion credit for Thailand. Thepackage now totals US$16.7 billion.
September 20 Mahathir tells delegates to the IMFrWorld Bank annualconference in Hong Kong that currency trading is immoraland should be stopped.
September 21 Soros says, ‘‘Dr. Mahathir is a menace to his own country’’.October 8 Indonesia says it will ask the IMF for financial assistance.October 31 IMF gives Indonesia a US$23 billion financial support pack-
age.November 17 South Korea abandoned its defense of the battered won.November 20 Dashing any early hope for controlling its financial turmoil,
South Korea’s currency fell 10% in trading.November 21 South Korea said it would seek a rescue package from the
IMF.December 1 South Korea and the IMF resumed talks on a rescue package
after an initial deal floundered.December 4 A record loan package led by the IMF to bail out South
Korea helped calm jitters in most regional markets.December 5 South Korea agreed to lower its economic growth to 3% in
1998 from a projected 6% under the terms of its IMF rescuepackage.
December 15 The IMF board meeting in Washington considers a Koreanrequest to speed up delivery of a portion of the US$60 billioninternational bailout announced on December 3.
December 18 Fed up with their economy’s freefall, voters in South Koreaelected longtime dissident Dae-Jung Kim as president, leav-ing some concerned that the country’s financial markets willbe further battered.
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216 215
December 24 The IMF said that it would make US$2 billion available toSouth Korea on December 30 from the US$21 billion setaside for the financially troubled country. The IMF plans todole out another US$2 billion to Seoul on January 8.
Ž .December 30 The world’s major banks key US and German banksprepared to join an effort to roll over a mountain of short-termdebt of South Korea due on December 31. The effort isexpected to help Korea manage its estimated US$100 billionin short-term debt, of which US$15 billion comes due byDecember 31.
January 2, 1998 While major US and European banks announced that theywould allow South Korean customers more time to pay offthe US$15 billion short-term debt, several smaller banks areunwilling to do the same.
January 9 Concern over Indonesia hit Asian stocks, but currencies wonsome support on hopes of an imminent deal between USbanks and heavily indebted Indonesian companies. A pro-posal for the South Korean government to issue about US$25billion in bonds won increased support at a meeting ofinternational banks. But several major banks were still hesi-tant about endorsing the plan, and the Korean governmentindicated it needs another week to make a decision.
January 13 The IMF and Indonesia appear to be near an agreement overthe IMF bailout.
January 15 Indonesian President Suharto announced wide-ranging eco -nomic reforms that, if carried out, would overturn the coun-try’s entrenched ways of doing business and curb its eco-nomic growth.
January 29 South Korean government and 13 leading international banksagreed that Korean banks can exchange their short-termnon-trade credits for new loans guaranteed by the Republic ofKorea with maturities of 1, 2 or 3 years, and with a floatingrate of 2.25%, 2.5%, and 2.75% over the 6-month LIBOR.
February 16 The rupiah dived through 10,000 in early trade in the wake ofweekend reports that the IMF had threatened to withdrawassistance to Indonesia if it adopts a currency board.
February 17 President Suharto fired Indonesia’s central bank governorwho have opposed government plans to create a fixed ex-change rate system for the rupiah through a currency board.The IMF, the United States, Germany and Australia had allcome out in opposition to such a board. The IMF hasthreatened to withhold further money under a US$43 billionbailout package. Separately, the IMF released a further US$2
( )B.-C. Kho, R.M. StulzrPacific-Basin Finance Journal 8 2000 177–216216
billion to South Korea, bringing total IMF lending to aboutUS$15 billion so far out of its US$21 billion rescue packageagreed last December.
March 9 A simmering dispute between the IMF and Indonesia cast ashadow over Pacific Rim markets, sending some regionalmarkets down and limiting gains in others. Over the week-end, news broke that the IMF would delay the disbursementof funds to Indonesia. The rupiah plunged to as low as 12,250per dollar.
March 21 The IMF and the Indonesian government have made ‘‘con -siderable progress’’ toward a new deal to counter the country’sgrave economic crisis.
March 26 Indonesia said that it is close to a comprehensive package ofmeasures to lift the country out of its worst economic crisis inthree decades, which Indonesia has agreed to in exchange fora US$40 billion bailout
April 8 Indonesia said that it had reached agreement with the IMF ona new package of economic reforms and targets, which theIMF would watch closely to ensure compliance.
References
Burnside, C., Eichenbaum, M., Rebelo, S., 1999. Hedging and financial fragility in fixed exchange rateregimes. Unpublished working paper. Northwestern University, Evanston, IL.
Cornell, B., Shapiro, A.C., 1986. The reaction of bank stock prices to the international debt crisis.Journal of Banking and Finance 10, 55–73.
Ž .Delhaise, P.F., 1998. Asia in Crisis: The Implosion of the Banking and Finance Systems. Wiley Asia ,Singapore.
Dewenter, K.L., Hess, A.C., 1998. An international comparison of banks’ equity returns. Journal ofMoney, Credit and Banking 30, 472–492.
Eichengreen, B., Rose, A., 1997. Staying afloat when the wind shifts: External factors and emerging-market banking crises. Unpublished working paper. University of California at Berkeley, Berkeley,CA.
Furman, J., Stiglitz, J.E., 1998. Economic crises: Evidence and insights from East Asia. BrookingsŽ .Papers on Economic Activity 2 , 1–136.
Griffin, J.M., Stulz, R.M., 1999. International competition and exchange rate shocks: A cross-countryindustry analysis of stock returns. Dice Center working paper. Ohio State University, Columbus,OH.
Hamann, J., 1999. Background to the crisis, in IMF-supported programs in Indonesia, Korea, andThailand: A preliminary assessment, Occasional paper 178, IMF, Washington, DC.
IMF, 1998. International Capital Markets: Developments, Prospects, and Key Policy Issues. IMF,Washington, DC.
Kaminsky, G.L., Schmukler, S.L., 1998. What triggers market jitters? A chronicle of the Asian crisis.Unpublished working paper. The World Bank, Washington, DC.
World Bank, 1998. East Asia: The Road To Recovery. World Bank, Washington, DC.