Impact of Credit Rating Announcements: An evidence from Major Asian Markets

8/16/2019 Impact of Credit Rating Announcements: An evidence from Major Asian Markets

1/54

Chapter 1: Introduction1.1 Introduction

Credit rating agencies have compete a very important role for many

years in monetary markets. The foremost responsibility of rating agencies

is to convey their opinions concerning the default risks of sure issuers or

market instruments by assignment credit ratings. The ratings they supply

square measure wide employed by numerous market participants.

Additional specically, lenders will place condence in ratings in decision-

making while not having to have interaction themselves within the pricey

and long operation method. or borrowers, ratings will widen their access

to funding through dissemination of their credit quality info to investors.

!ortfolio managers con"ointly use credit ratings in portfolio management.

#atings from sure reliable agencies are employed by regulator. As an

e$ample, the %&C pro"ected that public rms inform investors what

square measure the ratings of their securities as given by the agencies.

The city Committee on banking direction '(C(%) establishes capital

adequacy needs supported ratings provided by e$ternal credit rating

agencies. The performance of rating agencies has been the topic of intense

dialogue within the past few years. #ating agencies are defendant of

failing to supply reliable and timely ratings. *t can be a con+ict of interest

owing to a relationship with the corporate management. Accusations of

lagged reaction to info have intense particularly when many rising

market monetary crises. #ating agencies square measure criticied for


2/54

passively reacting to the crisis rather than having the ability to predict

the crisis. The disputation over rating agencies leads to a copious

quantity of analysis being carried out on the data worth of ratings.

(eneath the economical markets hypothesis, there would be market

reaction related to rating revision announcement if the event

incorporates valuable and price-relevant info.A large variety of studies have e$amined the impact of rating changes

on stock costs. The bulk agitate learning the orth American nation

market, whereas few analysis studies are done on work the Asian

markets. The dearth of analysis arises from the embedded characteristics

of Asian economies. *n distinction to the orth American nation market,

the Asian markets square measure comparatively tiny and fewer well-

regulated. %pecially, the weaker is that the market regulation and the

larger the market segmentation, the additional seemingly it is to look at

info leak and trading before any natural event.owever, owing to the continuing development and increasing

importance of the Asian markets, additional analysis ought to be

conducted to ascertain if the results obtained from mature markets like

orth American nation square measure similar. #ising markets square

measure characteried by info spatiality and low transparency. There

square measure restricted channels for rms to broadcast info to the

investment public. Credit rating therefore is one among the few out there

signals to convey relevant info in smaller Asian markets. Consequently,

the data discharged by rating agencies may be additional news-worthy in


3/54

these markets. Also, in lightweight of the vital restrictive role of rating in

some mature markets, this study may be additional substantive for

smaller Asian markets wherever the regulation could also be weaker

compared with orth American nation market. This study con"ointly seeks to look at the in+uence of world credit

rating agencies relative to native agencies. Against the background of an

apace growing rising economy, the demand for domestic ratings has

been gaining larger importance over time. The worldwide rating agencies

like /oody0s, normal and !oor0s usually charge abundant higher fees for

rating services than native ones given their larger specialied skills and

knowledge in rendering rating service. Therefore, from the monetary

perspective, the native rater is in a very stronger position to service the

rising market by allocating its resources to rate tiny issuers that will be of

less interest to the worldwide raters. Additionally, the domestic rater is

looked as if it would have a large and straightforward access to native

info, therefore facilitating a much better understanding and insights of

native issuers0 credit goodness. *n contrast, the orth American nation

headquartered international rating agencies might react additional slowly

owing to the inherent geographic disadvantages. *f that1s the case, the

rating announcement by domestic raters may be additional informative

relative to their international counterparts.owever, given the short history of rating record, lack of transparency,

the native agencies might receive very little recognition from investors.

This argument is according to Associate in ursing Asian development


4/54

bank paper that surveys a pool of investors on their opinions of native

raters. *t0s reportable that forty fth of investors surveyed aforesaid the

native raters were completely not timely the least bit. *f this market

perception dominates, the market may be insensitive to the native rating

announcement. Another criticism of native agencies is that the native

rating agencies square measure additional seemingly to use weaker

standards to domestic corporations, that is mentioned a home bias. To

induce an additional comprehensive understanding of however info is

mirrored in rising market, e$ploring the distinction between native and

international news announcement is of important importance.

1.2 Objectives of the study The ob"ect of this study is to e$amine the impact of credit rating

changes on common stock returns.2. 3o rating agencies have superior information and analytical skills

and hence can their rating revisions in+uence e$cess equity returns4

5. 3o the response patterns to rating events vary across di6erentmarkets4

7. 3oes the placement to credit watch list have information value4

8. 3o the unanticipated events carry more information than theanticipated ones4

9. 3oes the Asian crisis have an influence on the market reaction to

news announcement4

:. *s there any di6erence between the market reactions to rating events

announced by international agencies relative to those announced by

local agencies4;.


5/54

1.3 Organization of the study This paper proceeds as follows=

Chapter Two reviews the e$isting literature and provides the

theoretical background of this study.

Chapter Three describes the data used and e$plain how we select the

sample in detail. *n addition, the methodology employed in empirical

analysis is discussed.Chapter our presents the empirical results with corresponding

e$planations and related discussions.Chapter ive outlines the main ndings of this empirical work and

points out the limitations and suggests additional avenues for future

research.


6/54

Chapter 2: Literature Revie

A large body of literature has investigated the informativeness of

credit ratings by operating with the rating-related events. Those events

cowl varied views relating to credit rating. or e$ample, (arron, Clare and

Thomas '2>>;) analye the impact of assignment of latest rating on ?@

capital market and realie no important abnormal come related to it.

Associate degree empirical work by @liger and %arig '5) suggests that

rating data is price-relevant and valuable, citing proof that bond worth

ad"usts to data following /oody0s renement of its rating system.

Additional recently, Aintablian and /ora '59) realie no market

response around /oody0s announcement of eliminating the sovereign

ceiling on company rating. These studies square measure comparatively

rare since the events coated happen less oftentimes. Consequently,

additional and additional works specialie in the upgrade and downgrade

events.2.1 I!pact of rating changes on security price

The argument that there1s a linkage between rating revision and

security value has its principle embedded within the belief that rating

agencies area unit capable of gathering, process and so transfer vital

personal data to the market. At intervals the theoretical framework of

market potency, the knowledge content of rating will be tested by means

that of e$amining the abnormal returns close the announcement of rating

revision. The initial focus of the literature during this space is that the result of

rating changes on bond costs. Those works report con+icting noticing=


7/54

;8), Brierand

@at '2>;:), *ngram, (rooks and Copeland '2>7), >5) do notice vital bond

value reaction. These variations in results will be attributed mostly to the

problems of methodology, sample periods and also the knowledge

frequency 'monthly, weekly or daily). #esearches began to think about

the stock value reaction to bond ratings further.umerous studies have investigated the knowledge worth of credit

rating changes by measure the abnormal stock returns round the

announcement.'BriEn and %anvicente 2>5, Boh and &derington 2>>7,

2>>, 3ichev and !iotroski, 52) though there1s some variation in results

across studies, there1s proof that associate degree imbalance e$ists

available value reactions to credit rating downgrades and upgrades.

%pecically, a downgrade announcement is followed by a ma"or negative

accumulative abnormal stock come back, whereas associate degree

upgrade associate degree announcement is followed by an insignicant

positive accumulative abnormal stock come back. This puling empirical regularity is e$plained from the side of the

incentives of the businesses and also the rating agencies 'Boh and

&derington '2>>)). *t1s argued that rm0s area unit additional probably to

unharness the favorable data rather than the unfavorable data. Fn the

opposite hand, the rating agencies have the sturdy incentive to pay

additional resources in police work deterioration in credit quality instead


8/54

of enhancements as a result of the upper reputational price of failing to

discover credit issues. Those biases create the downgrade data additional

interesting within the market and so the market reaction to the

downgrade is stronger.

There are studies viewing the worth activity before and following the

rating changes.


9/54

whether or not new and valuable data is sent to the market. The rating

revision has detail content as long as the announcement communicates

new and value-relevant information to plug. *f the data is already

anticipated by investors, the market might not answer the news. They use

equity possession dispersion and charge per unit volatility as live of the

number of data o6ered to regulate for the market anticipation of a rating

amendment announcement. They notice no market response to rating

changes for larger companies and vital worth movements for smaller

companies with less data o6ered within the market. Creighton, Bower and

*. A. #ichards '58) additionally o6er proof for di6erential market

response associated with rm sie. Gorion and Hhang '59) recommend that initial worth of the rating is a

vital however neglected consider the reason of cross-sectional distinction.

They notice a way stronger stock worth movement for rating changes

ranging from a lower initial rating. %upporting proof for this argument will

be found within the study by Boh and &derington '2>>>). The magnitude of market response additionally depends on whether or

not the rating amendment crosses the investment to speculative-grade

border. *nstitutional investor are prohibited from holding the problems of

investment grade, resulting in a bigger worth movement once the border

is crossed. This Iinvestment-gradeJ result is mentioned in Gorion and

Hhang '59).2.3 $he in for!ativeness of credit atch procedure

Aside from enormous empirical evidence concerning the upgrades and

downgrades, the credit watch procedure, an issue little studied, has


10/54

stirred interest in this related stream of research. Fn the theoretical side,

(oot, /illbourn and %chmeits '5:) analye the economic role played by

credit rating agencies through the credit watch procedures in the

nancial markets. *n their proposed model, the credit watch procedure

helps to establish an implicit contractual relationship between the

agencies and the rated rms. The mechanism works as follows. %uppose

the agency observes potential deterioration in the firm0s credit quality, it

will ask the firm to deal with the unfavorable situation and place the rm

on the watch list. The rated rm can choose whether or not to undertake

recovery e6ort 'observable to the rating agency but not by the market)

after being put on negative credit watch list. *f the rm succeeds in

restoring the credit quality, the rating may get reconfirmed. Ftherwise, it

will be downgraded ultimately. The subsequent downgrade enables the

market to realie that the downgraded rm did not undertake

recovery e6ort or failed in improving the credit quality. *n contrast, for the

downgrade in absence of a prior credit watch, the market only learns

about the failure of the recovery e6ort over time, thus the price impact

takes more time to materialie. Biven that, they draw an empirical

prediction= a rating change after a credit watch procedure is more

likely to be informative than a rating change without an early warning. *n

addition, they propose that this mechanism only works in the case of

downgrade because the rating agency has less incentive to put a

company on the positive credit watch list. /ore importantly, their model


11/54

is based on the conditionality of investment decisions on the credit

rating, an institutional feature prevailing in the ?% market.*n related empirical work, and, #ichard and Deftwich '2>>5) e$amine

the security price reactions associated with the announcement of

additions to credit watch list, more specifically to the negative watch list

and positive watch list. >8 and 2>>: by (hattacharya, 3aouk, Gorgenson

and @ehr '5) shows no abnormal "ump in stock returns around the

announcement date. The authors then provide 9 possible candidates for

the e$planation of such an interesting nding. They are small sample

sie, ineEcient market, eEcient market but no information value inherent


12/54

in those events, full anticipation by the market and unrestricted insider

trading. urther, they provide evidence favoring the last one. They

conclude that the insider trading causes the prices to fully incorporate

the information before its public release, thus turning an event into a

non-event.As regards the information content of credit rating in developing and

small economies, there appears to be little research addressing this

issue. 3i6erent from many ?% studies that concentrate on company

rating, most emerging market studies are primarily concerned with the

e6ect of sovereign rating. The only study investigating the

informativeness of rating at rm level in emerging markets is by #ichards

and 3eddouche '2>>>). %urprisingly, the market is either insensitive to

rating changes or responsive in an opposite direction from what is

e$pected. owever, the surprising results in this paper may arise from a

non-robust empirical methodology. /ore broadly, the combination of 29

markets with di6ering level of sophistication could lead to misleading

results. ?sing weekly data instead of daily data is not ideal since

using a shorter observation interval is more e6ective in e$ploring daily

trends or security price reactions on specic event days. *n addition,

further investigation on all industries is preferable to the concentration

on only the banking industry. Therefore, there is scope for a more

comprehensive analysis on this issue.

2.& Co!parison of the in'uence of rating agenciesA very limited number of studies have previously e$amined the


13/54

di6erential impact of rating agencies. @ish, @aren and Flson document

no evidence that market values one agency over the other by

comparing the ratings from /oody0s and %K!. urther, %hin and /oore

'57) compare credit ratings assigned to Gapanese rms by two leading

?.%. rating agencies and two leading Gapanese agencies. Their ndings

suggest that the ratings by the ?.%. agencies are systematically lower

than those assigned by Gapanese raters. A follow-up study by Di, %hin and

/oore '5:) conclude that global agencies are more in+uential than

local raters for downgrades in Gapanese conte$t. Also, they nd the

ratings by the two prominent rating agencies are homogenous.

Chapter 3: (ata and )ethodo*ogy3.1 +vent de,nition


14/54

possible di6erent impact. An anticipated event refers to a change where

the rated company was put on the credit watch list prior to the

announcement date, with a direction consistent with the actual

subsequent changes. Credit watch is a valuable predictor of issuer0s

creditworthiness. (y placing rms on the credit watch list, the rating

agencies disseminate the information that an improvement or

deterioration is likely to take place in the short term to market

participants. %uppose the rm is ultimately downgraded or upgraded as

the credit watch list indicates, the market reaction to the rating revision

announcement might be muted since the news has already been

anticipated by the public through the credit watch procedures. Therefore,

we e$pect that the unanticipated events carry more information than the

anticipated ones.Aside from the actual rating changes, we also consider the credit watch

events. The credit watch event deals with the announcement of

additions to the credit watch list. %pecically, Inegative watch eventsJ

and Ipositive watch eventsJ are dened based on the direction of the

likely change indicated by the agency. The e$act definitions and specific e$amples of various rating events

to be employed in the conte$t of this study are outlined in table 7.2.

Table 7.2 about here

3.2

(ata3.2.1 (ata

source


15/54

The data required for the analysis include the credit revision

events and time series stock price data of several ma"or Asian markets.

The credit rating change data are collected from the (loomberg

database. The information provided by (loomberg consist of the

announcement date, rating type, rating agency, current rating, last rating

and industry type of the rated issuers. The ratings selected for analysis

are from /oody0s, %tandard and !oor0s and itch, the three most

recognied rating agencies in the world. The raw sample collected

covers changes in credit ratings between Ganuary 2>>; and 3ecember

59. %i$ Asian ma"or markets e$cluding Gapan are selected= *ndonesia,

ong@ong, %outh @orea, /alaysia, Taiwan and Thailand. *n addition to t he

rating information from the international raters, we also e$amine the

rating revision announcement by local raters to obtain a more

comprehensive analysis. Those data are obtained from (loomberg. Aside

from the actual rating changes information, (loomberg also provides the

credit watch indicators, which forms a part of our analysis. As for the

stock return of the rated companies, we use the daily closing price from

the data stream. These are supplemented with information from

(loomberg as well as from the local stock e$change. To measure the

abnormal return, we also collect the daily local stock indices data

denominated in local currencies of the : markets from data stream. The

sample of credit events is reduced substantially because of the data

problems.


16/54

3.2.3 -a!p*e

se*ection To avoid a loss of power in our tests resulting from other events with

potentially confounding e6ect on stock prices, we eliminate the

observation if another news announcement concerning earnings and

mergers occurred during the event window of 82 days. There are also

some cases where the rating is changed consecutively by multiple

agencies. The temporal clustering of rating revisions for the same

company might be potentially problematic for empirical analysis since it

may induce a biased estimate of stock price movement around a

particular event. or the two consecutive events within a2 day window, we include the event if both are upgrades or both are

downgrades. *n this case, the earlier not the later event is included. The

resulting sample used for analysis is summaried in table 7.5.


3.3 +vent study !ethodo*ogy The information content of rating is e$amined by means of an event

study. The length of the event window and the estimation window are

illustrated below=

tL-25 tL-5 tL tL5&stimation window &vent window

The choice of the event window is based on an attempt to obtain a

comprehensive picture of how the market responds to rating-related

events.


17/54

5 days on both sides of this day are chosen. The event window is split

into three sub-windows. To capture the likely information leakage, we

dene the period '-5,-2) as the pre-announcement period. Although the

information from (loomberg includes the announcement day of each

event, we cannot identify the e$act time of announcement and thus

cannot determine whether the announcement occurs during the

trading hours. Therefore, we choose a two-day period ', 2) as the

announcement window to account for this uncertainty. *n addition, we

also e$amine another period '5, 5) following the announcement window

to take the possible lagged reaction into consideration. This paper uses daily stock price data to compute two di6erent

measures of returns for the rated companies. The rst measure is based

on the market model as follows= Rit = α I + β iRmt + ε it (1)

Where

Rit = return on firm i for day t Rmt = return on market index for day t

ε it = error for day t

The coefficients are estimated for a given firm over a 100day estimation !indo! starting 1"0 days

#rior to the announced event and ending "0 days $efore the event day% The a$norma& return is defined

as the difference $et!een actua& returns and the returns #redicted $y the a$ove mode&%

'nother measure com#utes the excess return version of the market mode&

Rit = R ft + β i ( Rmt R ft ) + ε it (")

Where the risk free rate R ft is co&&ected from data stream% The resu&ts of this mode& are not

materia&&y different from the other mode&* and hence !e on&y re#ort the resu&ts of euation 1%


18/54

,ext* the average a$norma& return* ''-* is ca&cu&ated $y averaging the n events for a s#ecific event day

t

1

,

AAR t = . A R it

(/ )

N i= 1

Then* the ''-s are aggregated in order to dra! overa&& inferences for a certain #eriod

t "

CAARt 1* t " =

. AARt

t =t 1

The test statistic for the significance of ''- eua&s to the ratio of ''- to its standard

deviation% The standard deviation is estimated $y the data o$tained !ithin the estimation !indo!%

urther* the standard deviation of cumu&ative average a$norma& return is given $y standard

deviation of ''- mu&ti#&ied $y the suare root of the num$er of days in the #eriod%

t =

AARt

s( AARt )

1

)"0

()

s( AARt ) = . ( AARt AAR)"

100 1"0

(2

)


19/54

Chapter %: +!pirica* Resu*ts This chapter discusses the empirical analysis by e$amining the

information value of credit rating in four di6erent aspects. *n each

case, the results for average abnormal returns on ve specic event

days are presented. (esides, cumulative average abnormal returns over

the pre-announcement, announcement and post-announcement windows

are reported respectively. The discussion in this chapter centers on the

price movement over the 7 windows in order to give a more

comprehensive picture of the price ad"ustment.%.1 Is there any !ar#et reaction to actua* rating changes and

credit atch re*ated events The rst section e$amines whether stock returns are signicantly

in+uenced by actual rating changes announcement and by credit watch

placement announcement. The inclusion of credit watch event is based

on an e$pectation that the placement of a rm on the credit watch list

may convey some new price-relevant information to the market because

the credit watch list serves as an early warning system for likely rating

changes in the short run. The results are reported by market in table 8.2.


ong @ong and %outh @orea display a uniform response pattern to the

events investigated. /ore specically, price ad"ustment surrounding

downgrades are signicantly negative, reinforcing the evidence of

information value of downgrades observed in other markets. or

upgrades, there is little evidence of signicant price movement. *t is


20/54

apparent that the two markets do not treat upgrades as an informational

event. These ndings suggest that %outh @orea and ong @ong appear to

behave in a manner consistent with other mature markets. or the credit

watch events, only the negative watch placement has an impact on

market and the magnitude of reaction is smaller than that of downgrades

for both markets, indicating that a possibility of being downgraded is

less informative than an actual downgrade. This evidence is consistent

with prior research 'and, #ichard and Deftwich '2>>5)). As regards the

di6erence between the two markets, the price ad"ustment observed in

%outh @orea is more drastic than that in ong @ong in terms of

magnitude and signicance level.*n the case of *ndonesia, there is no market reaction associated with

upgrades and credit watch events. As regards downgrades, there appears

to be a full information leakage which translates into a strong negative

cumulative average abnormal return of -29.> over the pre-

announcement window. The information is found to be incorporated into

prices prior to downgrade announcements, suggesting a degree of

ineEciency in the *ndonesia market.As shown in table 8.2, the /alaysia market reacts to both negative

watch placement and downgrades announcements. An interesting nding

is that a signicant lagged price ad"ustment appears for upgrades over

the post-announcement window, indicating that upgrade constitutes an

informational event in the conte$t of /alaysia. This detected response is

contrary to the usual result found in previous literature that upgrades do


21/54

not have an impact on market.or the Thailand market, it appears that the good news

announcements contain new information that surprises investors. /ore

specifically, positive watch placement contributes to a highly signicant

cumulative average abnormal return of 5.:: over the two-day

announcement window. *n addition, positive price ad"ustment is observed

for upgrades announcement. *n terms of bad news, the market

apparently does not treat downgrades and negative watch placement as

informational events, thus no evident market reaction being detected. *t

is obvious that the Thailand market behaves in a totally di6erent way

from its more developed counterparties. The informative upgrades

may arise from the limited access to information in developing

economies, an e$planation can be provided to account for the similar

market reaction to upgrades observed in /alaysia. or the lack of

reaction to bad news, it is diEcult to identify the e$act reason. As

discussed by (hattacharya, 3aouk, Gorgenson and @ehr '5), five

possibilities are suggested. *t is possible that the market is too ineEcient

to re+ect new information. Alternatively, there might be no price-relevant

information in downgrades. Fr the market has fully anticipated the event,

thus no abnormal "umps in stock return are being detected around

announcement period. Also, unrestricted insider trading, which can

cause prices to fully incorporate the information prior to public

announcement, may contribute to the observed insensitivity. Aside from

those possibilities, another likely reason is that this result is only unique


22/54

in this specic period which includes the Asian crisis. *n our case, the

rst reason mentioned above is unlikely because there is evidence of

response to upgrades in Thailand. *n terms of other possibilities, it is

too premature to conclude which one is the most relevant candidate and

further discussion will follow.As regards Taiwan, the results are to a great e$tent surprising. either

upgrades nor downgrades have an e6ect on security prices. The

discussions regarding Thailand can be applied to the case of Taiwan. o

sign of information leakage is observed, indicating that insider trading

seems unlikely in this case. *n the subsequent sections, we0ll seek to

e$plore why Taiwan behaves in such an interesting way.(ased on findings in this section, it is apparent that a variety of

di6erent reactions e$ist across the Asian markets. The di6erent market

behaviors observed are in accord with prior e$pectation that markets

with di6ering level of sophistication may e$hibit unique patterns.

%.2 (o the unanticipated events carry !ore infor!ation than

the anticipated onesaving determined that the si$ Asian markets di6er in reaction to

rating related events, we now attempt to reveal more about the nature

of rating changes. This section hence e$tends the main results by

segmenting the rating changes into anticipated and unanticipated ones

according to whether the rerated rm was placed on credit watch list

with the same direction as the announced rating change before the

announcement. Table 8.5 provides the results from the si$ investigated

markets.


23/54


our markets present a similar pattern that the unanticipated

events have a more drastic market impact. /ore specically, in the case

of ong @ong, %outh @orea and /alaysia, the information value of

downgrades observed in section 8.2 almost entirely comes from the

unanticipated downgrades. or the anticipated downgrades, no

informational content is found, as shown by the insignicant market

reaction over the event window. Also, the di6erential impact of

anticipated and unanticipated downgrades holds true for the upgrades in

/alaysia.


24/54

e6ect over the announcement window. ?nlike other markets, the

di6erence between anticipated and unanticipated events lies in the

response time instead of the magnitude. This di6erence appears to be

driven by the differing incentives to possible leakage of information. *n

particular, for the unanticipated downgrades, insiders tend to trade on

relevant information because it is more likely to be a shock to market,

leading to a negative CA# of -2.857 preceding the announcement. or

anticipated downgrades, however, the same incentives may not e$ist

because the information regarding the downgrades was already

conveyed to market when the rms in question were added to the

negative credit watch list. *n addition, the signicant announcement

e6ect of anticipated downgrades contrasts with what was observed in

other markets, suggesting that although investors may receive prior

warning of downgrades in the form of credit watch placement, there is

still uncertainty regarding the timing and magnitude of the possible

downgrades. This uncertainty is reflected in the negatively significant

market reaction, which is suggestive of a component of surprise in the

news.


25/54

As with Taiwan, the market is still insensitive to any kind of rating

revisions when the market anticipation is taken into consideration.

Therefore, it is tempting to conclude that the market anticipation may not

e$plain the curious ndings of Taiwan in section 8.2.%.3 (oes the /sian

crisis have an i!pact on the !ar#et reaction to nes

announce!ents*n addition to the di6erences in the response induced by market

anticipation, the market reaction may di6er in other ways. The selected

period in previous discussion incorporates the Asian nancial crisis which

leads to a spike in downgrades. As discussed in #eisen '55), the rating

agencies were accused for reacting to events rather than anticipating

them in the /e$ican crisis of 2>>8-2>>9. This opinion is in line with the

argument of #eisen and von /altan '2>>>) that the performance of

rating agencies is poor during the Asian crisis. The discussion with regard

to agency performance in emerging markets is mainly concerned with

the sovereign rating. >; to 3ecember, 2>>. %ince almost all the

upgrades cluster in the non-crisis period, this analysis is only carried out

for downgrades. %ample sie and empirical results are presented in table

8.7 and 8.8 respectively.


26/54



!anel A in table 8.8 looks at Thailand, ong @ong and *ndonesia.

Among the three markets, the rst two markets share some common

features. *n particular, no e6ect is encountered for downgrades in crisis

period, whereas strong impact e$ists for those during the non-crisis

period. This phenomenon can be attributed to the likely e$cess

pessimism due to the financial crisis in which investors tend to have a

negative anticipation towards the financial status of firms in the market.

This psychological factor may turn a piece of negative news into a non-

surprise. *n this way, downgrades in a bad time would not have any

impact on market. ow we can safely conclude that the interesting

nding for downgrades in Thailand detected in section 8.2 can be

attributed to the time period chosen. The result of overall sample is

neutralied by the lack of response in the crisis period. The *ndonesia market appears to behave in a slightly di6erent way.

The negative e6ect of downgrades during non-crisis period is much

stronger than that of the other two markets.


27/54

insider trading causes the market to fully incorporate relevant

information into prices before publicly announced. Taken together, the

results in panel A suggest that the rating changes during non-crisis

period e$ert a much stronger announcement e6ect on market.urther evidence from %outh @orea and Taiwan can be found in panel

(. The result of %outh @orea contrasts with that of markets in panel A.


28/54

for the lack of market reaction in Taiwan.*n a practical sense, there is something to get e$cited about. #eferring

to the third column in panel (, while the CA#s over the announcement

and post-announcement window are not statistically signicant, the

di6erence between the two gures is economically signicant. An

opportunity for prot might arise supposing the investors buy stocks of

downgraded rms during the announcement period and then hold them

for one month 'about 5 trading days).


29/54

55-59 'T#*%). *n order to make the results more comparable and

meaningful, the sample periods for global raters are confined to be

consistent with their respective local counterparts. *n addition, caution

and prudence will be e$ercised in comparing the di6erent markets due to

the di6ering sample period across markets.


To get a sense of the distinct rating activities across agencies, table

8.9 outlines the coverage frequency of global and local raters using the

raw sample. or all the markets, local agencies rate the largest number

of rms, >7, 5; and 57; for T#C, T#*% and @*% respectively. owever,

among the rated rms, only a small number of firms are covered by both

local and global agencies. This common feature can be attributed to the

di6erent client groups they have been serving. Benerally, firms with large

capitaliation are more willing to request rating service from international

agencies since their ob"ective is to step into the global market and attract

more international investors. The financially disadvantaged small issuers,

in contrast, tend to choose local agencies who charge less prohibitive

fees. Dooking at the number of events announced by local and global

raters, %outh @orea and Thailand share a resemblance that the ma"ority

of the upgrades and downgrades are by global raters and local raters

respectively. This can be e$plained by the fact that large firms rated by

global raters are more likely to be upgraded due to their relatively sound

nancial status, whereas there is a higher possibility for small and


30/54

medium enterprises '%/&) to be downgraded because they are in more

volatile and vulnerable nancial position. Among the three markets,

%outh @orea has the largest number of rating revision events and the

largest coverage of rms rated. This is possibly due to its relatively well-

established and sophisticated nancial market. This pattern still remains

after its longer sample period has been taken into consideration.

The remaining sample after dropping events with insu6icient data and

those contaminated by other price-relevant news is summaried in table

8.:. e$t, table 8.; provides the event study results for the selected

period.

Table 8.: about here

Table 8.; about here

or %outh @orea in panel A, there is no big di6erence in the global

upgrades and local upgrades. As with downgrades by global raters, a

statistically significant negative abnormal return of -5.28;7 is detected.

Compared to the price behavior of rated rms by global raters over the

event period, the results for local downgrades show di6erent

characteristics. There is some indication that most downgrades are

foreseen by the market. A signicant negative CA# is observed over the

'-5, -2) window.


31/54

surprise in the news announcement. *n addition, a prolonged market

reaction is observed over the subsequent period '5, 5). Therefore, the

%outh @orea market, on average, reacts strongly to local downgrades, a

supporting evidence for the higher influence of local agency.*n the case of Taiwan, upgrades, regardless of being rated by global

agencies or by local ones, are statistically insignicant during the event

window. The downgrades, however, e$hibit di6erent characteristics.

There is no evidence that the market is responsive to downgrades by

global raters, consistent with the earlier nding using an alternative

period from 2>>;-59. '%ee panel ( in table 8.2). %urprisingly, stock

price responds signicantly to downgrades by local agency in the

e$pected direction. This result contrasts greatly with that of global

agencies, providing evidence that local agency provides better quality

information than its international counterparts at least in the Taiwan

conte$t.As noted in the previous discussion regarding the non-response of

Taiwan market to both upgrades and downgrades by global raters, si$

possibilities are potentially relevant= full market anticipation, insider

trading, sample period chosen, small sample, ine6icient market and

value-irrelevant rating information. The fourth e$planation fails to hold

because the sample sie of informative local downgrades is only slightly

larger than that of global downgrades. %imilarly, if Taiwan market is

informational ine6icient, we should not observe any signicant price

impact for the local downgrades. Biven that the first three alternative


32/54

possibilities have been ruled out in the earlier section, we, therefore, can

conclude that the investors in Taiwan do not value the information as

much from global rating agencies. *n other words, there is little

information value inherent in global rating changes.#esults in panel C of table 8.; are for rated firms in Thailand. As

can be seen, the news by global raters has a CA# signicant at 2M for

the two-day announcement window, a result consistent with the nding

in section 8.2 where a longer sample period is involved. *t appears that

the upgrades by global agencies contain new information that surprises

investors and the market interprets it as good news. onetheless, this

fails to hold true for upgrades by local agency and no signicant e6ect is

produced during the event period. There appears to be a greater

average price impact of global rating relative to local rating. *n the case

of downgrades, comparison is not available because all the downgrades

by local raters happened before 55. The result of local news shows no

market reaction associated with downgrade announcements.aving investigated the above three markets individually, we now

seek to e$plore the reason why the relative importance of local raters

vary across markets. To gain further insight into the three local rating

agencies, we present their characteristics in table 8..

Table 8. about here

*n terms of available services provided by local raters, T#*% has been

o6ering the most e$tensive services to its clients, followed by @*%. *n

contrast, T#C do not e$pand its services as its peers do. *nstead, it


33/54

chooses to concentrate e$clusively on the rating services. Dooking at the

various shareholding structures outlined in panel (, the I*nternational

C#A-3omestic C#AJ link-ups seem to be a common practice. or instance,

/oody has a 9M stake in @*% and %K! holds a 9M equity stake in T#C.

evertheless, T#*% is an e$ception with no "oint partnership with

international C#As. The di6erent background shown above may contribute to e$plaining

the fairly divergent market reaction patterns observed. Among the

analyed local agencies, only T#*% has no impact on security price. *t may

be true that the possible political dependence, as shown by a

government ownership of 2>M, may deplete the credibility of T#*%. The

stock price, therefore, is not responsive due to the market

perception of political intervention. Aside from that, a more likely

candidate might be the lack of "oint partnership with global raters. The

market may anticipate that the local raters are able to receive technical

support and skill transfer from their partner and thus enhance their

credibility. This raises the possibility that the responsiveness observed in

the case of T#C and @*% may be partly the result of international rating

agencies0 participation in ownership.Another interesting nding in this section is the strong di6erential

e6ects of downgrades detected for global versus local agencies 'T#C). *t

is possible that the di6erence arises from T#C0s concentration on rating

service, as indicated by panel A in table 8..


34/54

Chapter &: -u!!ary and Conc*usion

&.1 -u!!ary of the resu*tsCredit ratings are quite prevalent in nancial markets and a large

body of literature has documented that the security price will be

significantly a6ected when rating agencies revise the ratings they issued.

owever, there has been a lack of studies on Asian markets. Credit rating

has been regarded as an e6ective channel to reduce the degree of

information asymmetry. ence the role played by rating agencies is more

meaningful to smaller markets where transparency is weaker. Biven that,

this study aims to discuss the information value of credit ratings in

smaller Asian markets e$cluding Gapan. This issue is investigated by

measuring the pre-announcement and post-announcement as well as the

contemporaneous responses of si$ Asian nancial markets to rating

related events.

This study begins with an e$amination of market response to rating

related events announced by /oody0s, %K! and itch. The evidences

from the si$ Asian markets are mi$ed, suggesting the manner in which

markets react to credit rating di6ers across markets and relates to the

distinct characteristics of specic markets. *n particular, ong @ong and

%outh @orea, more advanced markets are found to behave in a similar

way to their developed counterparties= immediate reaction to bad news

but no response to good news. %trong evidence of information leakage is

observed for *ndonesia, indicating a degree of ineEciency. The results


35/54

from the other three markets= /alaysia, Thailand and Taiwan are quite

interesting. /alaysia and Thailand markets respond to upgrades. This

nding can be attributed to the lesser availability of opportunities by

rms to communicate good news in the two markets. Thailand,

however, shows non-response to downgrades. /ore surprisingly and

interestingly, there is no impact associated with any kind of rating

events announced by the international rating agencies in Taiwan.e$t, the main results from above are further investigated by the

partition of rating changes. The segmentation is based on the di6ering

market anticipation incorporated in rating revisions. The results suggest

that, on average, the unanticipated events carry more information value

than the anticipated ones. *n the case of Taiwan, no di6erence is

detected because the market remains unresponsive after the partition of

the overall sample.


36/54

analysis, three markets are investigated. %outh @orea and Taiwan seem

to be more sensitive to local raters. The di6erential impact of local and

global rater is e$tremely apparent in Taiwan, indicating that the local

rater is much more credible in the domestic market. or the Thailand

market, the global rater is more influential in terms of both market

coverage and market impact. To sum up, all the ndings in this study show that generally the credit

ratings in emerging markets do contain new information. owever,

the market reaction to the information di6ers in term of market

characteristics, rating agencies, markets conditions and investors

anticipation.

&.2 Li!itation of the study There are some limitations with respect to the analysis and data that

may a6ect the accuracy of the results.2. The sample sie for the covered markets is small due to the

di6iculty in the availability of data going farther back in time. This may

bring bias into research. ?sing small sample sie may increase the

possibility that the nal sample "ust happen to behave in the observed

manner. The small number of events in this study might be a caveat, we

therefore e$ercise prudence and caution when interpreting the empirical

results to ensure a more reliable conclusion.

5. The sample of credit events is reduced substantially because of the

data problems. /any rated companies do not issue stocks and the data

of some listed companies is not available due to inactive trading.

&.3 (irection for future research There are several additional avenues that could be e$plored, all of


37/54

which would help in understanding this issue better.An e$tension to this study would be to consider the cases of listing

abroad. The globaliation of nancial market has catalyed the growing

migration of nancing activities abroad and the number of companies

seeking a foreign listing has increased over the last few decades. This

is particularly the case in emerging markets where the minority

shareholder protection is weaker and hence the rms have incentives

to signal the willingness to protect minority shareholder right through

cross listing.


38/54

implications from the policy perspective. The regulators in emerging

markets can rely on ratings to monitor the banking industry, "ustifying

the incorporation of e$ternal credit ratings into the capital adequacy

requirements proposed by (asel Committee.


39/54

References

Abad-#omero, !., K #obles-ernande, /. 3. '5:). #isk and #eturn around (ond

#ating Changes= ew &vidence from the %panish %tock /arket Gournal of

(usiness inance K Accounting. Journal of Business Finance & Accounting, 33

'9-:), 9->.

(arren, /. G., Clare, A. 3., K Thomas, %. . '2>>;). The e6ect of bond rating

changes and new ratings on ?@ stock returns. Journal of Business Finance &

Accounting, 24 '7), 8>;-9>.

(erk, G., K 3e/aro, !. '5;). Corporate Finance. !earson &ducation.

Cantor, #. '525). The Role of Ratings in the Financial Syste.


40/54

ama, &. . '2>>2). &Ecient Capital /arkets= **. The Journal of Finance, 4+ '9), 29;9-

2:2;.

Boh, G. C., K &derington, D. . '2>>>). Cross-sectional variation in the stock market

reaction to bond rating changes. The 0uarterly Re%ie1 of /conoics an"

Finance, 3 '2), 22-225.

Boh, G. C., K &derington, D. . '2>>7). *s a (ond #ating 3owngrade (ad ews, Bood

ews, or o ews for %tockholders4 The Journal of Finance, 4*-, 52-

5.

Bonale, ., aas, ., Gohannes, #., !ersson, /., Toledo, D, Nioli, #., et al. '58).

/arket dynamics associated with credit ratings= A literature review. /CB

ccasional aper '2:), 2-8.

BriEn, !. A., K %anvicente, A. H. '2>5). Common %tock #eturns and #ating

Changes= A /ethodological Comparison. The Journal of Finance, 7;'2), 27-

22>.

Bropp, #., K #ichards, A. G. '52). #ating Agency Actions and the !ricing of 3ebt

and &quity of &uropean (anks=


41/54

and, G. #., olthausen, #. .

sueh, D. !., K Diu, O. A. '2>>5). /arket anticipation and the e6ect of bond rating

changes on common stock prices. Journal of Business Research, 24 '7), 559-

57>.

Gorion, !., K Hhang, B. '5;). *nformation &6ects of (ond #ating Changes= The

#ole of the #ating !rior to the Announcement. The Journal of Fi6e" $ncoe,

2:'8), 89-9>.

@liger, 3., K %arig, F. '5). The *nformation Nalue of (ond #atings. The Journal of

Finance, ** ':), 5;>-5>5.

Dinciano, . '58). The #eaction of %tock !rices to #ating Changes. 7oring

paper.

/ac@inlay, A. C. '2>>;). &vent studies in economics and nance. Journal of

/conoic 8iterature, 3* '2), 27-7>.

/atolcsy, H. !., K Dianto, T. '2>>9). The incremental information content of bond

rating revisions= The Australian evidence. Journal of Baning & Finance, )

'9), >2->5.

/oody1s *nvestors %ervices. '55). Role of the Cre"it Rating Agencies in Securities

!arets. /oody1s *nvestors %ervice.


42/54

orden, D, K -88.

!oon, -

;>;.

%chweiter, #., %ewcyk, %. ., K Narma, #. '2>>5). (ond rating agencies and their

role in bank market discipline. Journal of Financial Ser%ices Research, + '7),

58>-5:7.

%teiner, /., K einke, N. B. '52). &vent study concerning international bond price

e6ects of credit rating actions. $nternational Journal of Finance & /conoics,

+ '5), 27>-29;.

%tickel, %. &. '2>:). The e6ect of preferred stock rating changes on preferred and

common sloc. Journal of Accounting an" /conoics, 3-, 2>;-52:.

%ylla, #. '52). A istorical !rimer on the (usiness of Credit #atings. The Role of

Cre"it Reporting Systes in the $nternational /conoy 'pp. 2-7).


43/54

The ;). (ond rating agencies and the capital markets.

7oring aper.

-79.

, pp. 82-:7).

ew Oork= %pringer ?%.

Haima, G. @., K /cCarthy, G. '2>). The impact of bond rating changes on common

stocks and bonds= Tests of the wealth redistribution hypothesis. Financial

Re%ie1, 23 '8), 87-8>.


44/54

$ab*e 3.1: +vents de,nition

+vent type (e,nition +1a!p*es

?pgrade The announcement of a positive change in rating ((( 3 AAAPAAA

((( QR 3 AAAPAAAQRPAAAQ-

3owngrade The announcement of a negative change in rating AAA3 (((P (((QRP

AAAQ- 3(((P (((QRP(((Q-

Anticipated The announcement of an upgrade preceded by a positive ((( QR 3 AAAPAAAQRPAAAQ-?pgrade

?nanticipate The announcement of an upgrade not preceded by a prior ((( 3 AAAPAAAQRPAAAQ-?pgrade

Anticipated The announcement of an downgrade preceded by a negative AAAQ- 3 (((P (((QRP(((Q-downgrade

?nanticipate The announcement of a downgrade not preceded by a prior AAA 3 (((P (((QRP(((Q-downgrade

!ositivewatch

The announcement of additions to the positive watch list AAA3 AAA QR

egativewatch

The announcement of additions to the negative watch list AAA3 AAA Q-

ote= 2. QR = positive credit watch indicatorS Q- = negative credit watch

indicatorS Cases involving developing credit watch 'e.g. (((Q) are

e$cluded5. There are only a few cases in which the change direction is inconsistent

with the credit watch indicators. These are omitted. 'e.g. (((Q-

AAAPAAAQRPAAAQ-, AAAQR (((P (((QRP (((Q-)7. There are only a few cases where rms are removed from the credit

watch list without subsequent rating change. 'e.g. AAAQR AAA ,

AAAQ- AAA) These are omitted.

8. There are only a few cases involving changes in the status on credit

watch list, but without changes in rating classification 'e.g. AAAQR

AAAQ- , AAAQ- AAAQR). These are not considered.


45/54

$ab*e 3.2: -u!!ary -tatistics of ,na* rating sa!p*e 14"255&6

7pgrades (ongrades Credit atch

Anticipated ?nanticipated

Anticipated?nanticipated

egative !ositive Total Total

ong@ong

22 58 79 25 29 5; 58 25

*ndonesia 2 7 8 5 78 98 25

/alaysia 2> 5; 8: 2 5 5> ; 55

Taiwan 2 5 52 9 59 59 8

Thailand 2> 9> :> 9> :; ; 58

%outh@orea

7 7 77 29 8 99 22 5>

ote= umber of negativePpositive watch e$ceeds number of

anticipated downgradePupgrade in some markets because some

negativePpositive watch rms subsequently became e$cluded from

the sample. %ome also remained as negativePpositive watch at end of

sample period.


46/54

42

$ab*e %.1 )ar#et reaction to actua* rating changes and credit

atch events 14"255&6

8ane* /: 9ood nes

3ays Average abnormal returns

relativeto

event

ong @ong %outh @orea *ndonesia /alaysia Thailand Taiwan

!ositive .99 Q .582 .:9> .:> .;2: .292

.58: .222 .:88 .8:9 -.52 .:2 -.292 .5:: 5.5> .;52 QQ -2. -.859

2 .:> .; .7:> .:>; .> -.2; .875 .9: .7;2 .57:2 -.;5> .5;7

5 .972 -.85 -.7> .58 -2.; -.>2 -.22: -.2 -.9> .2:; -.5 -.52;

Cumulative average abnormal returns

-5 to -5.5;: .798 -.5 -.975 2.; -7.255 2.25 .;> 8.5 -2.78 9.>78 .82

to 2 .729 .2>> 2.27 2.2:7 .;> .828 .52 .755 5.:: .>9; QQ -2.2; -.295

5 to 5 -.7 .9: -2.;9; -5.5 -9.>99 -5.:88 .75 5.97 QQ -2.8>:; -.:77 -:.:72 -7.28

8ane* : ad nes


relativetoevent

ong @ong %outh @orea *ndonesia /alaysia Thailand Taiwan

egative8 .758

-.> .2:7 -5.87; QQ-.::5 2.27 .7;7 -.;88 -.>2 Q 5.2: -.857 -.52 -.2>;

2 -2.5>5 QQ-2.:>7 QQQ -.2: -2.8: QQQ -2.28 .2>; -.;7 -.8>2 -.> -.2:> -2.2>5 .298

5 -2.2>: QQ.:5 -2.87 Q -.7:5 -2.;2 -.97 -.;:; -.; -.>2: -.58: .9 -.9;


-5 to -7.79; 8.5;: -5.:5 .8: -;.55 -29.>QQQ .;: 2.>> 7.89 5.58 -2.7:7 -5.599

to 2 -2.7> Q -2.97 QQ -5.9>;Q -5.28;QQQ -.25 .9; -2.95; Q -2.82 Q 2.2;> -.9>5 -2.7>5 -.85

5 to 5 -.;7: 5.92> 9.52 5.: 2.98> 2.; -2.75>9 5.5: 9.9>: .295 5.72 2.:>>

otes=

1. %ince the e$cess return results are almost identical to the market

model results, only thelatter model is reported.5. The sample sie is listed in table 7.5.


47/54

7. Q indicates signicance at the 2M level, QQ at the 9M level and QQQ at the

2M level$ab*e %.2 )ar#et reaction to anticipated and unanticipated

rating change

8ane* /: /nticipated events


relativetoevent

ong @ong %outh @orea /alaysia Thailand *ndonesia Taiwan

?pgrade3owngrade

?pgrade3owngrade

?pgrade3owngrade

?pgrade3owngrade

?pgrade3owngrade

?pgrade3owngrade

-5 .78 .88 -.9: .:: -.92 .>9 .9>7 .82 .5;2 2.:7 .52 .82;

-2 .92 .59: -.58 2.:5 .527 .52 .85 -.2:8 -2.2; 2.598 2.8 -2.972

-.22 .>: .97> 2.98 .7>9 .22 2.2>5 Q .:> 7.285 -7.7 QQQ -.7; .89:

2 -.5 -2.>8 QQ .;: -.>58 .7: .77 .> .9 -2.;; -7.89 QQQ -.2:: -5.5:

5 -.5: -.559 .5> .>; -.77: -.;> .> -.5; -.887 -2.799 .:; -.72


-5 to .:99 2.779 -2.8;> -.>22 2.;87 -2.;72 -.:>9 7.582 -:.>9 -5.; -.:7> -8.2 to 2 -.5:7 -.>:5 2.589 .:2: .872 .728 2.2>7 2.2> 2.7;59 -:.9:; QQQ -.55 -2.;9

5 to 5 2.75 -5.8 -7.;7 9.8; 2.7:: 8.8> .85; -8.992 -9.292 2.9;2 7.58 .5;8

8ane* : 7nanticipated events


relativetoevent

ong @ong %outh @orea /alaysia Thailand *ndonesia Taiwan

?pgrade3owngrade

?pgrade3owngrade

?pgrade3owngrade

?pgrade3owngrade

?pgrade3owngrade

?pgrade3owngrade

-5 .852 -.98> -.:8 .:9 -.55 -.5;; .2: -.8:8 -.5; -5.925 QQQ .;;2 -.5>7

-2 .22; .:27 -.2;8 -2.: QQ .5:2 -.98 -.9 .2 -.89 -5.::8 QQQ .28 .;

-.2>7 -.8>: -.5; -2.8:9QQ .2;8 -.>87 Q .985 Q -.9:7 -.: .9:: -.889 -.7:

2 -.>; -2.8>QQ .: -5.>;9 QQQ .; -.:2; .759 -.5;2 .57 .78 .5>9 .;88

5 -.88 .5>2 2.>99 -2.9:; QQ -.; -.89 .555 -.585 .2 .785 -.5:5 -.;9


-5 to 2.55 8.5> .>8 -.8>9 .22> 5.272 -2.9>: 5.:7 -5.25> -2.857 .8:7 -2.2:

to 2 -.5> -2.>9QQ .77 -8.889 .588 -2.9: QQ .:; QQ -.78 .2:5 .>9 -.29 .78

5 to 5 2.825 8.;;; .855 2.889 5.97; QQ 2.5;> -2.79 .;> -2.>8 -2.899 -7.75; 5.9:

otes=2. %ince the e$cess return results are almost identical to the market model results,

only the latter model is reported.5. The sample sie is listed in table 7.57. The sample period is 2>>;-598. Q indicates signicance at the 2M level, QQ at the 9M level and QQQ at the 2M

level


48/54

45

$ab*e %.3: -a!p*e size of dongrades in crisis versus non"crisis period

$hai*and

;ong>; to 3ecember 72, 2>>.5. The rating changes are by /oody0s, %K! and itch.7. All the downgrades in /alaysia are announced in the crisis period,

thus this market is not included in this section8. %ince almost all the upgrades were announced in the non-crisis period,

only downgrades are investigated.

$ab*e %.%: $he i!pact of dongrades in crisis period and non"crisis

period

8ane* /: ;ong Q

-2 .>7>9 .52 -.2::9 .982:; -2.7>;2 -.:2:8

A .5922 .7>7 .78; -2.789 .8;7 -.89QQ

2 -2.8;:2: -2.>5: -.5>2 -2.5728Q .:98; -9.;7>QQQ

5 -.8>27 .2> -.7>5 .;:9; -.:22 -2.2>:9


-5 to -2 9.;:7 5.:;;8 5.7:;7 5.77>> -27.>8QQQ -:.;;8


49/54

46

to 2 -2.2>29 -2.>9 .9;; -5.5:9> QQ .>95 -27.>2>QQQ

5 to 5 8.:29> .59>>> -5.:98 .858: -2.5>5>> 8.:8:2

8ane* : -outh 89: 5.25:;8 -.28;55> -.2959;

-2 -2.9;85; QQQ .>2;99> 2.98225 -.79>>9

-2.288> QQQ 2.92; .92>9857 -.9>>5>

2 -2.29982 QQQ -5.>;2 Q 2.:2>555 -.:;;

5 -.>5:;: Q 5.;55 -2.:5>9> .5:9


-5 to -2 .>;:5: -5.29>22 .:72>8> -7.:7 to 2 -5.5>>>7 QQQ -2.8:>2 5.27;:87 -2.5:>7;

5 to 5 5.59>>;2: 9.:9;:; .59;2;; 5.:8;2;

otes= 2.The crisis period refers to Guly 5, 2>>; to 3ecember 72, 2>>.5. The rating changes are by /oody0s, %K! and itch.

7. All the downgrades in /alaysia are announced in the crisis period,

thus this market is not included in this section.8. Q indicates signicance at the 2M level, QQ at the 9M level and QQQ at the

2M level.


50/54

$ab*e %.& Coverage fre>uency of g*oba* and *oca* rater ra sa!p*e6

$aian $hai*and -outh .8>M 8; 79.78M 5> 8;.M Total 2;; 2.M 277 2.M :78 2.M

umber of downgrades announced by global82 92.>M 7 5.M 58 85.98M

#ater

umber of downgrades announced by local 7 8.2M 25 .M 779 9;.8:M

Total >9 2.M 29 2.M 97 2.M

umber of rms covered by global rater 92 79.85M 57 8:.M ; 55.M

umber of rms covered by local rater >7 :8.9M 5; 98.M 57; ;;.5M

Total 288 2.M 9 2.M 7; 2.M

umber of rms covered by both global and

9 .A 5 .A 2 .A#ater

ote= The sample periods for comparison are= 2>>;-59 '%outh @orea), 5-

59 'Taiwan)and 55-59 'Thailand).


51/54

$ab*e %.?: -u!!ary -tatistics of ,na* rating sa!p*e

7pgrades (ongrades

/arkets Blobal raters Docal Blobal raters Docal

Taiwan 2: 98 59 5;

Thailand 9 58 ;%outh 77 295 99 2;

otes= 2.The sample periods for comparison are= 2>>;-59 '%outh @orea),

5-59 'Taiwan) and 55-59 'Thailand).5. The global raters refer to /oody0s, %K! and itch for all markets. The

respective local raters are= @*% '%outh @orea), T#C 'Taiwan) and T#*% 'Thailand).7. *n the case of Thailand, no downgrade is announced in the clean sample

by global raters for the period 55-59.


52/54

$ab*e %.4 / co!parison of !ar#et reaction to rating changes

announced by *oca* raters and g*oba* raters

8ane* /: -outh 29278 QQQ

-2 -.529 .9522 -2.::95 QQ -.2>::;

.8:979 .;8:: -.::292 -.7885:

2 .:>;8: .885 -2.89; QQQ -2.58289 QQQ

5 .57; -.9:7> -.7:52: .7>>77


-5 to -2 -.972 .:7872 .8:2>> -8.888;:95 QQQ

to 2 2.2:52 .279>8 -5.28;7 QQQ -2.95955; QQQ

5 to 5 -5.5;; -2.;>88 5.:5> -8.89:2:; QQQ

8ane* : $aianAverage abnormal returns

3ays relative to ?pgrades 3owngrades

&vent Blobal ratersDocalraters

Blobal raters Docal raters

-5 -.2959; -.58>>: -.29:9 .>:72

-2 -.79>>9 .78;>22 .75887; .5957

-.9>>5> -.89 -.2>:92 -.9852>

2 -.:;; -.7958 .298;7 -.;2>>2 Q

5 .5:9 .5228 -.9:>;; .7;:2


-5 to -2 -7.:7 -2.5>78 -5.5992; -5.2;:

to 2 -2.5:>7; -.79:> -.8588 -2.5:52 QQ

5 to 5 5.:8;2; 2.25 2.:>>72 -2.:29

8ane* C: $hai*and

Average abnormal returns

3ays relative to ?pgrades 3owngrades

&vent Blobal ratersDocalraters

Blobal raters Docal raters

-5 .59:8 -.7; .A -.8;>7>

-2 .:595 .2>;2; .A -2.7:29

.579: QQQ .95:>8> .A -2.5>59>;

2 .52:>;; .88;; .A .2>8;:

5 .8852 .5227;7 .A .;>;;>:


-5 to -2 -.8788 -2.7>9 .A -7.:8::7

to 2 2.897; QQQ .>72:9: .A -2.2598>8

5 to 5 .87282 -2.8:2 .A 7.729998

otes= 2.The sample periods for comparison are= 2>>;-59 '%outh @orea),

5-59 'Taiwan) and 55-59 'Thailand).5. The global raters refer to /oody0s, %K! and itch for all markets. The


53/54

respective local raters are= @*% '%outh @orea), T#C 'Taiwan) and T#*% 'Thailand).7. *n the case of Thailand, no downgrade is announced in the nal sample

by global raters for the period 55-59.8. Q indicates signicance at the 2M level, QQ at the 9M level and QQQ at

the 2M level


54/54

$ab*e %.@ -u!!ary of do!estic rating agencies

8ane* /: range of services

Oear

/arket Agency&stablishe

d

/a"or services Additional services

-#ating services -!ro"ect nancing

%outh @*% 2>9 -Advisory services rating

@orea -Training

Taiwan T#C 2>>; -#ating services o additional services

-#ating services -Buarantor rating

-Bovernance rating services -Bovernment rating

Thailand T#*% 2>>7-!erformance evaluationservices

8ane* : shareho*ding structures

Current share holding

/arket Agency !ublic listedDocal

rmsPbanksBovt *ntl C#A Fther

%outh @orea @*% 9M M M 9M M

Taiwan T#C .A .A .A 9M .A

Thailand T#*% M ;:M 2>M M 9M

otes= 2. 3ata source= U3evelopment of #egional %tandards for Asian Credit

#atingAgencies V !rogress K Changes0 prepared for A3( '59)5. .A denotes data is not available

7. %K! initially took a 9M ownership stake in T#C and announced that it

had signed an agreement to raise its stake in T#C from 9M to 92M on 5;

%eptember 59.

Impact of Credit Rating Announcements: An evidence from Major Asian Markets

Documents

Transcript of Impact of Credit Rating Announcements: An evidence from Major Asian Markets